TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings (The Minerals, Metals & Materials Series) 3030923800, 9783030923808

Citation preview

The Minerals, Metals & Materials Series

The Minerals, Metals & Materials Society Editor

TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings

Editor The Minerals, Metals & Materials Society Pittsburgh, PA, USA

ISSN 2367-1181 ISSN 2367-1696 (electronic) The Minerals, Metals & Materials Series ISBN 978-3-030-92380-8 ISBN 978-3-030-92381-5 (eBook) https://doi.org/10.1007/978-3-030-92381-5 © The Minerals, Metals & Materials Society 2022, corrected publication 2022 This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer imprint is published by the registered company Springer Nature Switzerland AG The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

This volume is a collection of papers from the TMS 2022 Annual Meeting & Exhibition, held February 27–March 3 in Anaheim, California, USA. The contributions represent 71 symposia from the meeting. This volume, along with the other proceedings volumes published for the meeting, and TMS archival journals represent the available written record of the 103 symposia held at TMS2022.

Contents

Part I

Materials Processing Fundamentals

Comprehensive Recovery of Oxygen Pressure Acid Leaching Residue of Zinc Sulfide Concentrate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Guiqing Liu, Bangsheng Zhang, Zhonglin Dong, Fan Zhang, Fang Wang, Tao Jiang, and Bin Xu

3

Effect of Laser Heat Treatment and Nitrogen Content in Shielding Gas on Precipitation of Widmanstätten Austenite in Lap Laser Welds of Duplex Stainless Steels . . . . . . . . . . . . . . . . . . . . . . . Yunxing Xia, Kenshiro Amatsu, Fumikazu Miyasaka, and Hiroaki Mori

15

Influence of M-EMS Parameters on Flow Characteristics in a Bloom Mold . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Xiang-Lan Yang, Ming-Tao Xuan, Shan Wang, and Min Chen

24

Numerical and Physical Simulations of Bottom Blowing Process Optimization of 120t Refining Ladle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Shan Wang, Min Chen, Ming-Tao Xuan, and Xiang-Lan Yang

34

Parametric Study of Mold Electromagnetic Stirring: Effects of Load Condition and Copper Resistivity . . . . . . . . . . . . . . . . . . . . . . . . . . Qilan Li, Lifeng Zhang, and Jing Zhang

42

Phase and Microstructural Analysis of In-Situ Derived Alumina-TiB2 Composites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Evangelos Daskalakis, Animesh Jha, Andrew Scott, and Ali Hassanpour

49

vii

viii

Contents

Phase Equilibria in the Ag–Ge–Bi–Te System and Thermodynamic Properties of the nGeTe•mBi2 Te3 (n, m = 1–4) Layered Compounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mykola Moroz, Fiseha Tesfaye, Pavlo Demchenko, Myroslava Prokhorenko, Orest Pereviznyk, Bohdan Rudyk, Lyudmyla Soliak, Daniel Lindberg, Oleksandr Reshetnyak, and Leena Hupa Potentiostatic Electrodeposition of Ti–Al Alloy with 40% Titanium from the Lewis Acidic 1-Butyl-3-Methylimidazolium Chloride-Aluminum Chloride Ionic Liquid Electrolyte . . . . . . . . . . . . . . . Pravin S. Shinde, Yuxiang Peng, and Ramana G. Reddy

60

74

Prediction of Distribution of Composition of Inclusion in Continuous Casting Bloom of the Heavy Rail Steel Coupling Element Segregation, Heat Transfer, and Kinetics . . . . . . . . . . . . . . . . . . . Yuexin Zhang, Wei Chen, Jujin Wang, Yadong Wang, Wen Yang, and Ying Ren

87

Strategies for the Upgrade of a TBZC Product (Tetra Basic Zinc Chloride) by Selective Removal of the Impurity Chlorine . . . . . . . . . . . . . L. Höber, R. Ahmed, T. Hofbauer, and S. Steinlechner

95

Study of γ -F4 N Annealing Process Through Molecular Dynamics Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Jianxin Zhu, Guannan Guo, and Jian-Ping Wang

109

Part II

2D Materials—Preparation, Properties and Applications

Raman and Transport Characterization of Semiconducting and Superconducting Selenide-Based Transition Metal Dichalcogenides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Kishan Jayanand and Anupama B. Kaul Synthesis and Characterization of MnCo2 O4 -GQDs Nano-composites for Supercapacitor Electrodes . . . . . . . . . . . . . . . . . . . . . Poonam R. Kharangarh

121

129

Part III Additive Manufacturing Fatigue and Fracture: Developing Predictive Capabilities Effect of Post Heat Treatment on Fatigue Strength of AlSi10Mg Produced by Laser Powder Bed Fusion Process . . . . . . . . . . . . . . . . . . . . . Wei-Jen Lai, Avinesh Ojha, and Ziang Li

141

Effect of Surface Roughness on Fatigue Behavior of 316L Stainless Steel Produced by Binder Jetting Process . . . . . . . . . . . . . . . . . . Wei-Jen Lai, Avinesh Ojha, and Zhenxuan Luo

164

Contents

ix

High Strain Rate Deformation of EBM-Ti–6Al–4V: Microstructure, Texture, Mechanical Properties, Fracture Surface, and Deformation Mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Reza Alaghmandfard and Mohsen Mohammadi Machining Versus Heat Treatment in Additive Manufacturing of Ti6Al4V Alloy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Alireza Dareh Baghi, Shahrooz Nafisi, Reza Hashemi, Heike Ebendorff-Heidepriem, and Reza Ghomashchi The Study on Microstructural Evolution During Post-processing of Additively Manufactured Ti64 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bryan Naab, Denis P. Dowling, and Mert Celikin

177

187

198

Part IV Additive Manufacturing for Energy Applications IV 3D Printing Energetics for Gun Propulsion Technology . . . . . . . . . . . . . . David Bird, Elbert Caravaca, Joseph Laquidara, Nathan Peabody, Christopher Houthuysen, and Nuggehalli M. Ravindra Part V

Additive Manufacturing of Large-Scale Metallic Components

Residual Stress, Microstructure, and Characterization of Self-Mated Additive Repair of Inconel 718 Alloy Using Cold Spray Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hariharan Sundaram and Prasad Raghupatruni Reverse Engineering of Aerospace Components Utilizing Additive Manufacturing Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Balakrishnan Subeshan, Abdulaziz Abdulaziz, Zeeshan Khan, Md. Nizam Uddin, Muhammad Mustafizur Rahman, and Eylem Asmatulu Part VI

211

225

238

Additive Manufacturing of Refractory Metallic Materials

Laser Metal Deposition of Nickel Silicide on S355 Structural Steel . . . . Mohammad Ibrahim, Tor Oskar Sætre, and Ragnhild E. Aune

249

Refractory Metals—Some Historical Observations . . . . . . . . . . . . . . . . . . Jeffrey Wadsworth

259

Part VII

Additive Manufacturing: Beyond the Beam III

Microstructure Evolution and Mechanical Properties of Friction Stir Metal Deposited SS304 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nikhil Gotawala, Neeraj Kumar Mishra, and Amber Shrivastava

269

x

Contents

Part VIII Additive Manufacturing: Materials Design and Alloy Development IV: Rapid Development A 3D Multiple-Slip Crystal-Plasticity Model for Precipitate Hardening in Additively Manufactured High Strength Steels . . . . . . . . . Moustafa M. AbdelHamid and Tarek M. Hatem

281

Development of Al-Cu-Mg and Al-Mg-Si-Zr Alloys with Improved L-PBF Processability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . F. Belelli, R. Casati, and M. Vedani

289

Effect of Hot Isostatic Pressing on the Microstructure of Directionally Solidified Nickel Alloy After SLM . . . . . . . . . . . . . . . . . . . Evgenii Borisov, Anna Gracheva, Vera Popovich, and Anatoly Popovich Functionally Graded Alloys from 316 Stainless Steel to Inconel 718 by Powder-Based Laser Direct Energy Deposition . . . . . . . . . . . . . . . Kun Li, Jianbin Zhan, Peng Jin, Qian Tang, David Z. Zhang, Wei Xiong, and Huajun Cao In-Situ LENS Fabricated Ti–Al–Si Alloy Phase Transformation and Microstructural Evolution After Isothermal Annealing Heat Treatments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sadiq Abiola Raji, Abimbola Patricia Idowu Popoola, Sisa Leslie Pityana, Olawale Muhammed Popoola, Nasirudeen Kolawole Raji, and Monnamme Tlotleng

298

304

313

Part IX Advanced Functional and Structural Thin Films and Coatings In-Air Polymerization and Crosslinking of Monomers During Electrospray Deposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Catherine J. Nachtigal, Michael J. Grzenda, and Jonathan P. Singer

327

Incorporation of Metallic Nanoparticles Into Alkyd Resin: A Review of Their Coating Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . I. H. Ifijen, M. Maliki, S. O. Omorogbe, and S. D. Ibrahim

338

Materials for Antireflection Coatings in Photovoltaics—An Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Vishal Mehta, Cory Conkel, Andrew Cochran, and N. M. Ravindra

350

Nanostructured Graphene Thin Films: A Brief Review of Their Fabrication Techniques and Corrosion Protective Performance . . . . . . . Ikhazuagbe H. Ifijen, Oscar N. Aghedo, Ifeanyi J. Odiachi, Stanley O. Omorogbe, Ekebafe L. Olu, and Innocent C. Onuguh

366

Contents

xi

Numerical Study of Intrinsic Stresses in Perovskite-on-Si Solar Cells with Intermetallic Bonding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Seif Tarek and Tarek M. Hatem Part X

378

Advanced Magnetic Materials for Sensors, Power, and Multifunctional Applications

Custom-Designed Miniature-Coil Winding/Wrapping Machine . . . . . . . Balraj S. Mani, Bilal Adra, and Nuggehalli M. Ravindra

387

Effect of Hot Band Annealing and Final Annealing Temperatures on the Texture, Grain Size, and Magnetic Properties of 1.2 wt% Si Non-oriented Electrical Steel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Youliang He, Mehdi Mehdi, Tihe Zhou, Chad Cathcart, Peter Badgley, and Afsaneh Edrisy

397

Magneto-Mechanical Properties and Magnetocaloric Behaviour of Rapidly Solidified Melt-Spun Ni50 Mn28 Ga22 Heusler Alloy . . . . . . . . . D. K. Satapathy, P. D. Babu, I. A. Al-Omari, and S. Aich

408

Part XI

Advanced Materials for Energy Conversion and Storage 2022

Characterization of AlCl3 -Urea Electrolyte for Speciation, Conductivity, and Electrochemical Stability and Its Application in Al-Ion Batteries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Monu Malik, Kok Long Ng, and Gisele Azimi Multi-layered Thin-Film Metal Contacts for New Generation Solar Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I. Kruhlov, A. Orlov, V. Zakiev, I. Zakiev, S. Prikhodko, and S. Voloshko On Recent Development in Two-Dimensional Transition Metal Dichalcolgenides for Applications in Hydrogen Evolution Reaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chukwudike Ukeje Photoabsorbers with Hybrid Organic–Inorganic Structures for Optoelectronics and Solar Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mohin Sharma, Mritunjaya Parashar, and Anupama B. Kaul Simulating Microstructure Evolution in Ni-YSZ Electrodes of Solid Oxide Cells Under Operating Conditions . . . . . . . . . . . . . . . . . . . . Yinkai Lei, William Epting, Jerry Mason, Tian-Le Cheng, Harry Abernathy, Gregory Hackett, and Youhai Wen

419

431

442

448

457

xii

Contents

Study on a Short Process Method for Preparation of 3.5 Valence Vanadium Electrolyte . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Zhengtuan Li, Chunjing Wu, Heli Wan, and Lanjie Li Thermoelectric Generators System Made with Low-Cost Thermoelectric Modules for Low Temperature Waste Heat Recovery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Manuela Castañeda, Andrés A. Amell, and Henry A. Colorado Part XII

470

479

Advanced Real Time Imaging

In Situ Observation and Investigation of the Wetting Behaviors of Mold Flux on Steel Substrate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Lejun Zhou, Yang Yang, Wanlin Wang, Hao Luo, and Houfa Hu Investigation of Echo Source and Signal Deterioration in Ultrasound Measurement of Metal Melt . . . . . . . . . . . . . . . . . . . . . . . . . . Bitong Wang, Andrew Caldwell, Antoine Allanore, and Douglas H. Kelley Real-Time Quantification of Nickel, Cobalt, and Manganese Concentration Using Ultraviolet– Visible Spectroscopy—A Feasibility Study . . . . . . . . . . . . . . . . . . . . . . . . . . Monu Malik, Ka Ho Chan, and Gisele Azimi

489

495

507

Part XIII Advances and Discoveries in Non-equilibrium Driven Nanomaterials and Thin Films Salt-Assisted Chemical Vapor Deposition Synthesis of 2D WSe2 and Its Integration in High Performance Field-Effect Transistors . . . . . Anupama B. Kaul and Avra S. Bandyopadhyay

521

Part XIV Advances in Biomaterials for 3D Printing of Scaffolds and Tissues Additive Manufacturing of Natural Materials as a Multidisciplinary Approach in Engineering Education . . . . . . . . . . . Henry A. Colorado, Elkin I. Gutierrez, and Mery Gomez-Marroquin Part XV

533

Advances in Multi-principal Elements Alloys X

Development of a High Entropy Alloy AlX (CoCrCuFeNi)1-X for Diverse Security Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D. Butcher, J. C. T. Cullen, N. Barron, S. Mehraban, M. Calvo-Dahlborg, S. G. R. Brown, and N. P. Lavery

543

Contents

Part XVI

xiii

Advances in Powder and Ceramic Materials Science

Catalytic Pyrolysis of Polyethylene and Polypropylene Over Y Zeolite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Xunrui Wang, Chengdong Wang, Xiang Wang, and Jinhong Li Cold Sintering of Iron Powdered Metal Compacts and Their Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Linsea Paradis, Ramakrishnan Rajagopalan, Austin Fairman, Kyle Robertson, Daudi R. Waryoba, and Clive Randall Design of New High Entropy Ceramics in the Pseudo-Binary System RGaO3 -R2 Ti2 O7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Victor Emmanuel Alvarez-Montano, Francisco Brown, Jorge Mata Ramírez, Subhash Sharma, Ofelia Hernández Negrete, Javier Hernández Paredes, and Alejandro Durán Determination of Structuralers of Type Solid Solutions Ba1–3X Gd2X Ti1–3X Eu4X O3. (X = 0.1, 0.15, 0.3, and 0.6% by Weight) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ricardo Martínez López, Miguel Pérez Labra, Francisco Raúl Barrientos Hernández, Martín Reyes Pérez, Julio Cesar Juárez Tapia, Aislinn Michelle Teja Ruiz, Víctor Esteban Reyes Cruz, and José Ángel Cobos Murcia Preparation of V2 AlC Phase Material by Aluminum, Graphite, and V2 O5 Precursor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Youzhi Gao, Xuyang Liu, Liangxiao Wei, and Ning Hu Synthesis and Optimization of BiFeO3 and La-Doped BiFeO3 Prepared by the Solid State Reaction Method . . . . . . . . . . . . . . . . . . . . . . . Subhash Sharma, V. E. Alvarez-Montaño, Eunice Vargas Viveros, Rosario I. Yocupicio-Gaxiola, J. M. Siqueiros, and Oscar Raymond Herrera The Effect of Particle Size on the Morphology of Polyester and Epoxy-Based Auto-Hybrid Composites . . . . . . . . . . . . . . . . . . . . . . . . . Kator Jeff Jomboh, Adele Dzikwi Garkida, and Vershima Cephas Alkali Thermodynamic Analysis of CaO in the Preparation of Fe/FeAl2 O4 by Fe2 O3 -Al2 O3 Electrolysis . . . . . . . . . . . . . . . . . . . . . . . . . Zhenwei Jing, Hongyan Yan, Yanke Xu, Hui Li, and Jinglong Liang

557

564

571

579

587

593

599

609

xiv

Part XVII

Contents

Advances in Surface Engineering IV

An Electrochemical Study of Ferrous and Nonferrous Materials in an Engine Coolant Environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Gaurav Argade, Anusha Chilukuri, Justin Perry, Monica Gehrich, Erica Raisor, and Corey Trobaugh An Investigation of the Microstructure and Oil Retention of Electrolyte Jet Plasma Oxidation (EJPO) Coating . . . . . . . . . . . . . . . . . Nasim Bahramian, Sina Kianfar, Joshua Stroh, Dimitry Sediako, and Jimi Tjong Electrochemical Corrosion Tests of Aluminum 1100 Alloy Coupons in Acid Condensate Environment . . . . . . . . . . . . . . . . . . . . . . . . . . Vasundhara Shinde, Gaurav Argade, Anusha Chilukuri, Monica Gehrich, and Chirag Parikh Electrochemical Study of Stainless Steels in Diesel Exhaust Fluid (DEF) and Simulated Exhaust Acid Condensate Environments . . . . . . . Anusha Chilukuri, Michael Warwick, and Gaurav Argade Part XVIII

621

629

640

649

Advances in Titanium Technology

A Review on Impact Resistance of Partially Filled 3D Printed Titanium Matrix Composite Designed Aircraft Turbine Engine Fan Blade . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Shade Rouxzeta Van Der Merwe, Daniel Ogochukwu Okanigbe, Dawood Ahmed Desai, and Glen Snedden

659

Microstructural Evolution and Mechanical Properties of Additively Manufactured Commercially-Pure Grade 2 Titanium After Post-process Heat Treatment . . . . . . . . . . . . . . . . . . . . . . . . Ralf D. Fischer, Greyson Harvill, Hossein Talebinezhad, and Barton C. Prorok

672

Preparation of TiAl Alloy by Magnesium Aluminum Synergistic Reduction of TiO2 in Molten Salt Medium . . . . . . . . . . . . . . . . . . . . . . . . . . Jialong Kang, Zhenyun Tian, Guibao Qiu, and Yaoran Cui

681

Part XIX AI/Data Informatics: Computational Model Development, Validation, and Uncertainty Quantification Investigating the Suitability of Tableau Dashboards and Decision Trees for Particulate Materials Science and Engineering Data Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bryer C. Sousa, Richard Valente, Aaron Krueger, Eric Schmid, Danielle L. Cote, and Rodica Neamtu

691

Contents

Part XX

xv

Algorithm Development in Materials Science and Engineering

A Finite Difference Analysis of the Effect of Graphene Additions on the Electrical Conductivity of Polycrystalline Copper . . . . . . . . . . . . . William Frazier, Bharat Gwalani, Julian Escobar, Joshua Silverstein, and Keerti. S. Kappagantula Clustering Algorithms for Nanomechanical Property Mapping and Resultant Microstructural Constituent and Phase Quantification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bryer C. Sousa, Chris Viera, Rodica Neamtu, and Danielle L. Cote Part XXI

705

713

Biological Materials Science

Biodegradable Superabsorbent Polymers . . . . . . . . . . . . . . . . . . . . . . . . . . . Kaylon Draney and Jeffrey Bates Characterization of Multi-walled Carbon Nanotube Reinforced into Poly(3-Hydroxybutyrate-Co-3-Hydroxyvalerate) (PHBV)-Epoxidized Natural Rubber 50 (ENR50) Biofilms . . . . . . . . . . . . A. Turner, S. Zainuddin, D. Kodali, and S. Jeelani Deep Learning and Finite Element Method Towards the Application of Microfracture Analysis for Prevention of Fatigue Fractures in Bones . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Gerardo Presbítero-Espinosa, José Quiroga-Arias, Inés Hernández-Ferruzca, Bibiana González-Pérez, Carlos Mora-Núñez, Eduardo Macías-Ávila, Álvaro Gómez-Ovalle, Christian Mendoza-Buenrostro, and Marco A. L. Hernandez-Rodriguez Effect of Suppressing Pressure on the Properties of AZ91 Foamed Magnesium Alloy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hanghang Zhou, Guibao Qiu, Zhenyun Tian, and Qingjuan Li Fabrication and Characterisation of Two-Layered Synthetic Titanium-Chitosan Bone Scaffolds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . L. Yildizbakan, N. Iqbal, D. Abdulaziz, V. Panagiotopoulou, E. Jones, N. T. Do, P. V. Giannoudis, and A. Jha Investigation of Effect of the Urea Content on the Pore Morphology, Porosity, and Mechanical Behavior of Porous Ti . . . . . . . . . Ding Yang, Yaoran Cui, Guibao Qiu, and Tengfei Lu Polymer Interfaces with Small-Scale Biological Systems and the Impact on Sperm Viability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Jeffrey Bates, Kenneth Aston, Benjamin Emery, Ashwin Velraj, Abhishek Pachauri, Parker Toews, and Meredith Humphreys

727

736

748

759

768

776

783

xvi

Contents

Effect of Holding Time on the Properties of AZ91 Foamed Magnesium Alloy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Zhenyun Tian, Guibao Qiu, Yaoran Cui, and Qingjuan Li Part XXII

796

Composite Materials for Sustainable Eco-Friendly Applications

Cadmium (II) Removal from Aqueous Solution by Magnetic Biochar Composite Produced from KOH-Modified Poplar Sawdust Biochar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Lei Zhang, Yongsheng Zhang, Yanfang Huang, Guihong Han, Hafiza Sana, and Shengpeng Su Characterization on the Electrochemical Property of the Ion Flotation Sludges After Thermal Treatment . . . . . . . . . . . . . . . . . . . . . . . . . Guihong Han, Jingwen Wang, Bingbing Liu, Ze Yang, and Yanfang Huang Effect of Thermal Conductivity on the Mechanical Behavior of Marginal Construction Waste as a Structural Material for Recycled Pavements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Liliana Carolina Hernández García and Henry A. Colorado L. Efficient Removal of Molybdenum from Ultra-Low Concentration Solutions via Fe(III) Chelating Precipitation: Precipitation Sludge for MoFe Alloy Production via the Metallothermic Reduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bei Zhang, Bingbing Liu, Yuanfang Huang, Guihong Han, Yubi Wang, and Shengpeng Su Performance for the Treatment of SO2 and NO in Sintering Flue Gas of the Novel Adsorbent Prepared Under Microwave Field . . . . . . . . Qing Guo, Min Chen, and Jun-hong Zhang Removal of Fluoride from Aqueous Solution by NH2 -MIL-101(Al) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Xinhui Liu, Wenjuan Wang, Guihong Han, Yanfang Huang, Bingbing Liu, and Shengpeng Su Study on the Application of Modified MOFs to the Treatment of Simulated Metallurgical Wastewater . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Junpeng Zuo, Guihong Han, Wenjuan Wang, Yanfang Huang, Bingbing Liu, and Shengpeng Su The Formation of Schwertmannite and Its Influence on Mine Environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Xiong Yao, Min Gan, Peng He, Dongli Huang, Jinye Liang, Miao Cai, and Chunyao Gu

807

817

826

835

844

855

863

872

Contents

Part XXIII

xvii

Composites for Energy Applications: Materials for Renewable Energy Applications 2022

Preparation of Coal Liquefaction Residue-PAN Composite Carbon Nanofibers by Electrostatic Spinning . . . . . . . . . . . . . . . . . . . . . . . Xiaoyan Zhang, Tongxin Qiao, and Peng Li Part XXIV

881

Computational Thermodynamics and Kinetics

3D Construction Based on Nephogram Slices of Simulated Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Xiao Han, Jiwu Wang, and Jinwu Kang A CFD Simulation Investigation of Microbubble Generation and Movement During the Mineral Flotation Process . . . . . . . . . . . . . . . . Guihong Han, Hao Wu, Yanfang Huang, Shengpeng Su, and Bingbing Liu A Tailor-Made Experimental Setup for Thermogravimetric Analysis of the Hydrogen- and Carbon Monoxide-Based Reduction of Iron (III) Oxide (Fe2 O3 ) and Zinc Ferrite (ZnOFe2 O3 ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ulrich Brandner, Juergen Antrekowitsch, Felix Hoffelner, and Manuel Leuchtenmueller Grain Precipitation and Growth Model of TiN Inclusions in 22MnB5 Steel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Haohao Zhang, Jialu Wu, Wei Guo, Songyuan Ai, Mujun Long, Dengfu Chen, and Huamei Duan

895

905

917

927

Multi-Physics Modelling of Additively Manufactured Cellular Structures Using Selective Laser Melting . . . . . . . . . . . . . . . . . . . . . . . . . . . Mahmoud A. Elsadek and Tarek M. Hatem

941

Simulation of Steam Film Motion Process on the Surface of Zirconium Alloy Rod . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ju Yi Pu, Xiao Ping Lang, Shuang Liang, and Bai Feng Luan

949

Thermodynamic Assessment of the SiO2 -Y2 O3 System . . . . . . . . . . . . . . . Wenke Zhi, Fei Wang, Xiaoyi Chen, Bin Yang, Yongnian Dai, and Yang Tian Part XXV

959

Defects and Properties of Cast Metals IV

Characterization of Second Phase Particles in Twin-Roll Cast Aluminum Alloy AA 8011 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sooraj Patel and Jyoti Mukhopadhyay

971

xviii

Contents

Effect of Build Height on Micro-cracking of Additively Manufactured Superalloy RENÉ 108 Thin-Wall Components . . . . . . . . . Apratim Chakraborty, Reza Tangestani, Trevor Sabiston, Nicholas Krutz, Lang Yuan, and Étienne Martin Elucidating the Relationship Between Arc Behavior and Solidification Defects During Vacuum Arc Remelting of Superalloys . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Daniel McCulley, Joshua Motley, Matthew Cibula, and Paul King

985

994

Evolution of Microstructure and Mechanical Properties of the As-Cast 1030B Al Sheet During Ultrasound-Assisted Continuous Casting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1004 Ripeng Jiang, Wenhao Zhao, Li Zhang, Xiaoqian Li, and Shaokang Guan Evolution of Physicochemical Properties of Tundish Covering Flux in Continuous Casting Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1012 Hao Hu, Xiangyang Wang, Chenghui Wu, Xin Xie, Mujun Long, Dengfu Chen, Xiaodong Yang, Shuang Liu, and Huamei Duan Experimental Research on Dephosphorization of Mn-Si Alloys Based on CaO-CaF2 Slag System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1023 Chong-yuan Zhang, Ying-dong Wang, Can Sun, Jing-tao Fan, and Zi-zong Zhu Hybrid Additive Manufacturing of Island Grain Bicrystals . . . . . . . . . . . 1033 L. G. Ware, B. S. Herstein, Y. Zhang, H. Z. Bilheux, and Z. C. Cordero Research Progress of Composite Preparation Technology of Bimetallic Wire . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1044 Chenglin Li, Ting’an Zhang, and Yan Liu Uncertainty Quantification of Model Predictions Due to Fluid Flow in Laser Powder Bed Fusion of IN625 . . . . . . . . . . . . . . . . . . . . . . . . . 1054 Scott Wells and Matthew John M. Krane Part XXVI

Deformation and Damage Mechanisms of High Temperature Alloys

Microstructure and Mechanical Properties of Rotary Friction Welded IN-600 and SS316L with Copper Interlayer . . . . . . . . . . . . . . . . . 1067 Neeraj K. Mishra and Amber Shrivastava Part XXVII

Electronic Packaging and Interconnections

Role of the Aging Treatment in the Microstructure Evolution and Mechanical Properties of Cu/Sn-Bi-Ag-In/Cu Joint . . . . . . . . . . . . . . 1079 Zhen Li, Guanzhi Wu, Kai Ding, and Yulai Gao

Contents

xix

Wire Bonding Novel 3D Air-Metal Dielectric Structures with ISIG Passivation: Process Development and Reliability . . . . . . . . . . 1088 Yipin Wu, Pichaya Sommai, Joyce Christiansen-Salameh, Jim Clatterbaugh, and Leyla Hashemi-Sadraei Part XXVIII

Environmental Degradation of Multiple Principal Component Materials

Development of Aluminum-Based Dissolvable Alloys for Hydraulic Fracturing Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1101 Ezz Ahmed, Hani Henein, Ahmed Qureshi, and Jing Liu Part XXIX

Environmentally Assisted Cracking: Theory and Practice

Atomistic Study on Diffusion and Trapping of Hydrogen in Nanocrystalline Steel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1117 Denver Seely, Bradley Huddleston, Sungkwang Mun, Anh Vo, Nayeon Lee, Doyl Dickel, and Krista Limmer Hydrogen-Induced Cracking of Pure Titanium in Hydrochloric and Sulfuric Acid Solutions Using Constant Load Method . . . . . . . . . . . . 1127 Osama M. Alyousif Stress Corrosion Cracking Study of Fe39 Mn20 Co20 Cr15 Si5 Al1 (at.%) Compositionally Complex Alloy in 3.5 wt% NaCl Salt Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1138 P. Varshney and N. Kumar Part XXX

Failure, and a Career That Is Anything But: An LMD Symposium Honoring J. Wayne Jones

Tear Resistance of AA7075-T6 Sheet at Room Temperature and 200 °C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1149 Daniel E. Nikolai and Eric M. Taleff Part XXXI

Functional Nanomaterials: Functional Low-Dimensional (0D, 1D, 2D) Materials 2022

MoS2 Thermoelectrics for Sustainable Energy . . . . . . . . . . . . . . . . . . . . . . 1163 A. A. Ramanathan Nanostructured Materials: A Review on Its Application in Water Treatment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1172 Ikhazuagbe H. Ifijen, Esther U. Ikhuoria, Muniratu Maliki, Godfrey O. Otabor, and Areguamen I. Aigbodion

xx

Contents

Photoabsorbers with 2D Layered Perovskites for Bendable Optoelectronics and Solar Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1181 Anupama B. Kaul and Mohin Sharma Part XXXII

High Performance Steels

Non-metallic Precipitates Evolution Mechanism of Fe-3.0wt%Si Steel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1193 Huilan Sun, Zimo Bi, Di Zhang, Zhihong Guo, and Bo Wang Small-Scale Rapid Alloy Prototyping of Extra-Low Carbon Steel to Investigate the Effects of Cu and Cr Residuals . . . . . . . . . . . . . . . 1202 Mazher Ahmed Yar, Caroline Norrish, Jonathan C. T. Cullen, Lintao Zhang, Stephen Brown, Richard Underhill, and Nicholas Lavery Part XXXIII

Materials and Chemistry for Molten Salt Systems

Thermodynamic Analysis of Fe Reduction by LiFePO4 in Melts Electrolysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1217 Rui Zhang, Jinglong Liang, and Hui Li Thermodynamic Analysis of FeSi/Fe3 Si Intermetallic Prepared from Copper Slag by Electrochemical Method . . . . . . . . . . . . . . . . . . . . . . 1227 Chaolong Xue, Hui Li, and Jinglong Liang Thermodynamics of Fe Reduction in Melts Electrolysis . . . . . . . . . . . . . . 1235 Xianhe Lv, Hui Li, and Jinglong Liang Part XXXIV

Materials Design and Processing Optimization for Advanced Manufacturing: From Fundamentals to Application

Effect of Ultra-High Deformation Process on Micro-Structure and Properties of Cu-0.7 wt.% Ag Wires . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1245 Wang Shusen, Zhang Yuanwang, and Yao Dawei Effect of Welding Current on Liquid Metal Embrittlement of the Resistance Spot Welded Galvanized QP980 Advanced High-Strength Steel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1254 Wufeng Dong, Kai Ding, Hua Pan, Ming Lei, and Yulai Gao Experimental Determination of Forming Limit Diagram for AISI 304 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1262 M. Krishnamraju, Abhishek Kumar, Sushil Mishra, and K. Narasimhan

Contents

xxi

Effect of Nozzle Structure Parameters on Liquid Steel Flow Behavior in Slab Mold . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1271 Si-kun Peng, Ming-mei Zhu, Kun-chi Jiang, and Jie Luo Influence of Single Fold and Double Fold on the Stress and Strain of AMOLED Module . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1281 Qiujun Wang, Weiwei Su, Di Zhang, Weijin Ji, and Bo Wang Mg–Al-Layered Double Hydroxide Doped with Phosphate Radical for Preparation of Slow-Release Phosphate Fertilizers . . . . . . . . 1290 Yanyu Wang, Cuihong Hou, Shiqiang Guan, Shouyu Gu, and Haobin Wang Numerical Simulation of Flow Distribution System in Molten Pool of Twin Roll Strip . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1302 Jie Luo, Ming-mei Zhu, Si-kun Peng, Ai-ping Zhang, and Yong Zhong Preparation of Polydopamine-Mesoporous Silica Nano-Fertilizer Loaded with Multiple Nutrients for Crops Planting . . . . . . . . . . . . . . . . . . 1312 Jianmeng Wu, Cuihong Hou, Shouyu Gu, Haobin Wang, Yanyu Wang, Shiqiang Guan, Jie Wang, and Tingting Wang Part XXXV

Materials in Sport

Measuring Cool Touch of Key Sports Performance Apparel T-Shirt Materials Using a Modified Transient Plane Source (MTPS) Sensor to Inform Future Technology Development . . . . . . . . . . . 1327 Susan L. Sokolowski, Emily Karolidis, Arya Hakimian, and Sarah Ackermann “Stuck on You”: Functional Friction Measurements of Doctored Baseballs Coated with “Sticky Substances” . . . . . . . . . . . . . . . . . . . . . . . . . 1338 Brian J. Love Part XXXVI

Mechanical Behavior and Degradation of Advanced Nuclear Fuel and Structural Materials

Impact Fretting Wear Behavior of Cr-Alloy Coating Layer for Accident-Tolerant Fuel Cladding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1345 Y. H. Lee, D. J. Park, Y. I. Jung, S. C. Yoo, and H. G. Kim Study of Microstructure, Hydrogen Solubility, and Corrosion of Ta-Modified Zr–1Nb Alloys for Nuclear Applications . . . . . . . . . . . . . . 1352 P. A. Ferreirós, E. C. Savoy Polack, L. A. Lanzani, P. R. Alonso, D. P. Quirós, J. I. Mieza, E. Zelaya, A. J. Knowles, and G. H. Rubiolo

xxii

Part XXXVII

Contents

Mechanical Behavior at the Nanoscale VI

Molecular Dynamics Simulations on Nanosuspension Droplet Impact . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1365 Baiou Shi and Siddharth Ravi Part XXXVIII

Phase Stability, Phase Transformations, and Reactive Phase Formation in Electronic Materials XXI

Solution-Processed Perovskite Photoabsorbers with Mixed Cations for Improved Stability in Solar Cells . . . . . . . . . . . . . . . . . . . . . . . . 1377 Mritunjaya Parashar, Mohin Sharma, and Anupama B. Kaul Investigation of Thermal Properties and Thermal Reliability of Ga-based Low Melting Temperature Alloys as Thermal Interface Materials (TIMs) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1385 Yifan Wu, Rajath Kantharaj, Albraa Alsaati, Amy Marconnet, and Carol Handwerker Part XXXIX

Phase Transformations and Microstructural Evolution

Dissolution of Carbides in HAZ During NG-TIG Welding for Alloy 617/9%Cr Dissimilar Welded Joint . . . . . . . . . . . . . . . . . . . . . . . . 1399 Kai Ding, Guanzhi Wu, Wufeng Dong, and Yulai Gao Effect of Changes in Phase and Grain Interface on Physical Properties During Aging of Ultra-Deformation Cu-Ag Alloy Wires . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1409 Zhang Yuan-wang, Wang Shu-sen, and Yao Da-wei Effect of Electron Spin Fluctuation on the Magnetism and Elastic Properties of the Slab Matrix Phase . . . . . . . . . . . . . . . . . . . . . 1417 Songyuan Ai, Chenxi Yang, Mujun Long, Haohao Zhang, Dengfu Chen, and Huamei Duan In Situ Observation of Coupled Growth Morphologies in Organic Peritectics Under Pure Diffusion Conditions . . . . . . . . . . . . . . 1429 Johann Mogeritsch, Wim Sillekens, and Andreas Ludwig Microstructure Evolution of HP40 and HK40 Steels After Isothermal Aging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1442 Victor M. Lopez-Hirata, Maribel L. Saucedo-Muñoz, Hector J. Dorantes-Rosales, Carlos Ferreira Palma, Eduardo Pérez-Badillo, and Diego I. Rivas-Lopez

Contents

xxiii

Precipitation Process During Isothermal Aging of an Austenitic Stainless Fe-12Cr-10Mn-12Ni-5Mo-0.24 N-0.03C Steel and Its Effect on the Mechanical Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1451 Maribel L. Saucedo-Muñoz, Victor M. Lopez-Hirata, Erika O. Avila-Davila, Felipe Hernandez-Santiago, and Jose D. Villegas-Cardenas Part XL

Powder Materials Processing and Fundamental Understanding

Characterization of Gas Atomized Nickel Silicide Powder for Additive Manufacturing with Varying Silicon Content . . . . . . . . . . . . 1463 Mohammad Ibrahim, Tor Oskar Sætre, and Ragnhild E. Aune Analysis of Additive Manufacturing Powders’ Behaviors Using Discrete Element Method-Based Simulation . . . . . . . . . . . . . . . . . . . . . . . . . 1473 Safwat M. Shenouda, Sun Yi, Paul Akangah, and Taher Abu-Lebdeh Preparation and Characterization of Spherical Nickel Silicide Powder by Inductively Coupled Plasma Spheroidization for Additive Manufacturing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1488 Foysal Kabir Tareq, Ragnhild E. Aune, Geir Grasmo, Naureen Akhtar, and Tor Oskar Sætre Thermodynamic Analysis and Synthesis of V2 AlC Phase Material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1497 Liangxiao Wei, Xuyang Liu, Youzhi Gao, and Ning Hu Part XLI

Recent Advances in Printed Electronics and Additive Manufacturing: 2D/3D Functional Materials, Fabrication Processes, and Emerging Applications

A Comparative Study on Supercapacitors Formed with Different Graphene-Based Hybrid Nanostructured Materials . . . . . . . . . . . . . . . . . . 1507 Tasnim Mahjabin and Md. Abdullah Al Amin Effect of Dichloroethane on the Electronic Transport Behavior in Semiconducting MoS2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1516 Ravindra Mehta and Anupama B. Kaul Part XLII

Refractory Metals

Platinum-Based Superalloys: Combating High Temperatures and Aggressive Environments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1527 L. A. Cornish

xxiv

Part XLIII

Contents

Seeing Is Believing—Understanding Environmental Degradation and Mechanical Response Using Advanced Characterization Techniques: An SMD Symposium in Honor of Ian M. Robertson

Effect of Hydrogen on Creep Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1541 Masanobu Kubota, Daiskuke Takazaki, Ryosuke Komoda, Kentrao Wada, Toshihiro Tsuchiyama, Mohsen Dadfarnia, Brian P. Somerday, and Petros Sofronis Part XLIV

Structural Metamaterials

High-Stiffness Metamaterial Composite Structure with Plate-Reinforced Strut-Microlattice . . . . . . . . . . . . . . . . . . . . . . . . . . . 1551 Manash Jyoti Baishya, Bikram Jyoti Sahariah, Nelson Muthu, and Prasenjit Khanikar Part XLV

Ultrafine-Grained and Heterostructured Materials (UFGH XII)

Enhancing Mechanical Performance of a Commercial Al Alloy by Tailoring Their Microstructural Heterogeneity . . . . . . . . . . . . . . . . . . . 1571 Ahmed M. Mohamed, Mohamed Aldlemy, and Khaled F. Adam The Role of the Heterogenous Structure on the Mechanical Properties of Additively Manufactured AlSi10Mg Alloys . . . . . . . . . . . . . 1577 Haoxiu Chen, Sagar Patel, Mihaela Vlasea, and Yu Zou Correction to: Design of New High Entropy Ceramics in the Pseudo-Binary System RGaO3 -R2 Ti2 O7 . . . . . . . . . . . . . . . . . . . . . . Victor Emmanuel Alvarez-Montano, Francisco Brown, Jorge Mata Ramírez, Subhash Sharma, Ofelia Hernández Negrete, Javier Hernández Paredes, and Alejandro Durán

C1

Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1585 Subject Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1591

Part I

Materials Processing Fundamentals

Editors: Samuel Wagstaff Oculatus Marietta, GA, USA Alexandra Anderson Gopher Resource Tampa, FL, USA Jonghyun Lee Iowa State University Ames, IA, USA Adrian S. Sabau Oak Ridge National Laboratory Oak Ridge, TN, USA Fiseha Tesfaye Åbo Akademi University Turku, Finland

Comprehensive Recovery of Oxygen Pressure Acid Leaching Residue of Zinc Sulfide Concentrate Guiqing Liu, Bangsheng Zhang, Zhonglin Dong, Fan Zhang, Fang Wang, Tao Jiang, and Bin Xu

Abstract Oxygen pressure acid leaching residue of zinc sulfide concentrate contains abundant elemental sulfur that has excellent natural hydrophobicity, and thus, it is considered an important secondary resource for recovering simple substance with froth flotation technique. However, the separation of simple substance from sulfide minerals in the residue is difficult mainly because of their similar hydrophobicity. In this research, we proposed an effective flotation process for selective recovery of simple substance from an oxygen pressure acid leaching residue of zinc sulfide concentrate. After one-time blank rougher, two-time agent-added roughers, and twotime cleaners separately using Z-200 as the collector and combined agent Na2 SO3 + ZnSO4 + Na2 S as the inhibitor, 99.2% of the elemental sulfur was recovered and the purity of concentrate product achieved 83.25%. Most of the lead, zinc, and silver went into the tailing that can be used as the raw material of pyro-metallurgical lead smelting for recovering these valuable metals. Keywords Oxygen pressure acid leaching residue of zinc sulfide concentrate · Elemental sulfur · Selective flotation · Z-200 collector · Na2 SO3 + ZnSO4 + Na2 S inhibitor

Introduction Zinc mainly exists in the earth as sulfides, among which sphalerite is one typical zinccontaining mineral [1]. Thus, sphalerite has become a vial raw material for extracting G. Liu School of Metallurgy, Northeastern University, Shenyang 110819, Liaoning, China G. Liu · B. Zhang · F. Zhang · F. Wang Jiangsu BGRIMM Metal Recycling Science & Technology Co. Ltd, Xuzhou 221121, Jiangsu, China Z. Dong (B) · T. Jiang · B. Xu Peace Building, School of Minerals Processing and Bioengineering, Central South University, Changsha 410000, Hunan, China e-mail: [email protected] © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_1

3

4

G. Liu et al.

zinc metal. Sphalerite can be treated by pyro-metallurgical and hydrometallurgical processes [2–6]. However, the pyro-metallurgical process suffers from several evident disadvantages, such as high energy consumption and potential atmospheric pollution [7–10]. In order to overcome the preceding defects of pyro-metallurgical process, hydrometallurgical process has been developed [11]. Among the alternatives, oxygen pressure acid leaching is regarded to be a feasible route [12, 13]. After leaching, zinc sulfide is converted into zinc sulfate and elemental sulfur is generated. Elemental sulfur is remained in the residue and its content can achieve 40–60% [14]. Thus, the leached residue is an important secondary resource for recovering elemental sulfur. Elemental sulfur possesses good natural floatability, and therefore, it can be effectively recovered by froth flotation, a beneficiation method that can realize the selective separation of target minerals from gangue minerals by using their floatability difference [15]. However, some sulfides (e.g. chalcopyrite, pyrite, and galena) generally exist in the residue because of their incomplete oxidation during oxygen pressure acid leaching [16]. These sulfides also have good natural floatability and inevitably also float upward with elemental sulfur. As a result of this, the obtained flotation concentrate usually has a low elemental sulfur grade, which is unbeneficial to its subsequent processing and application [17]. To solve the problem, the combined use of collector with high selectivity and inhibitor that can prevent sulfides flotation may be a feasible countermeasure. In this paper, an effective flotation process was proposed for selective recovery of elemental sulfur from an oxygen pressure acid leaching residue of zinc sulfide concentrate. First, a systematic flotation conditional experiment was performed to ascertain the optimal grinding fineness, pulp concentration, collector and inhibitor types, and their dosages. Then, flotation flowchart experiment was carried out to investigate the optimum numbers of rougher flotation and cleaner flotation. At last, closed-circuit flotation experiment was conducted to verify the feasibility of the proposed flotation scheme.

Materials and Methods Materials and Reagents The oxygen pressure acid leaching residue of zinc sulfide concentrate utilized in this study was supplied by Hulun Buir Chihong Mining Industry Co., Inner Mongolia, China. The contents of main elements in the residue are shown in Table 1. The sulfur content (i.e. weight ratio of the sulfur element to the residue) was as high as 46.21%. In addition, the contents of nonferrous metals including Zn, Pb, and Ag also arrived at 4.31%, 1.92%, and 220 g/t, respectively. So, the residue had a high economic value.

Comprehensive Recovery of Oxygen Pressure Acid …

5

Table 1 Contents of main elements in the oxygen pressure leaching residue Composition

S

Zn

Pb

Cu

Aga

Fe

SiO2

MgO

CaO

Al2 O3

Content (wt. %)

46.21

4.31

1.92

0.20

220

15.4

6.91

1.38

1.60

0.51

a Unit

g/t

80

(a)

100

81.97

80

40

Content Distribution 37.88

20 0

Simple substance sulfur

Sulfide

Sulfur phase

2.77

73.65

Mass Sulfur grade Sulfur distribution 72.07

66

43.91

40 20

12.03 5.56

83.18

60

%

%

60

(b)

36.53

14.58 8.06

6

Sulfae

30.54

22.77

0

+74 μm

3.17

4.98

-74+44 μm -44+37 μm

+37 μm

Size fraction

Fig. 1 Distributions of sulfur phase (a) and particle size (b) of the oxygen pressure leaching residue

Sulfur phase analysis was conducted for the residue, and the result is displayed in Fig. 1a. Sulfur mainly existed in the form of elemental sulfur whose proportion achieved 81.97% of the total sulfur. The proportion of sulfides also accounted for 12.03% of the total sulfur. The particle size distribution of the oxygen pressure leaching residue is shown in Fig. 1b. The sulfur grade was the highest (i.e. 73.65%) in the size fraction of −74 + 44 um, but sulfur distribution in this fraction was only 14.58%. The sulfur distribution in the fraction of −74 um was the largest (i.e. 43.91%), but its sulfur grade (i.e. 30.54%) was lower than that of size fraction of −74 + 44 um. In comparison, both the sulfur grade and distribution were high in the size fraction of +74 um. Therefore, the particle size distribution of the leaching residue was uneven. All the collectors and inhibitors used in this study were of analytical grade and were supplied by Zhuzhou Flotation Reagents Factory, and tap water was used through all the experiments in order to simulate the industrial flotation practice.

Flotation Experiment Self-aeration XFD-63 flotation machine was used as the equipment of test. When the experiment was begun, the residue and tap water were put into the flotation cell and the pH value of the formed pulp was adjusted to around 8.0 with lime. Afterwards, the inhibitor (if used), collector (if used), and foaming agent (methyl isobutyl carbinol, MIBC) were consecutively added into the pulp which was agitated at 1650 rpm for

6

G. Liu et al.

2 min when each reagent was added. The flotation was conducted for 5 min, and the obtained concentrate and tailing fractions were dried and weighed for sulfur content detection to calculate its recovery.

Analytical Methods The elemental contents in the oxygen pressure acid leaching residue were analyzed with acid digestion and atomic absorption spectrometer (AAS). The sulfur phase distribution was analyzed with a chemical selective dissolution method which was performed according to the basic principle that each sulfur phase has different dissolution behaviors in various solvents. The particle size distribution was ascertained using a wet screen analysis.

Results and Discussion Flotation Conditional Experiment Effect of Collector Type Four potential collectors including O-isopropyl-N-ethyl thionocarbamate (Z-200), ammonium dibutyl dithiophosphate (ADDTP), ethyl thiocarbamate (ETCM), and ethyl xanthate (EX) were selected, and its flotation performances for the leaching residue were compared. The flow sheet is displayed in Fig. 2, and the flotation result Pressure acid leaching residue

Blank flotation

5' 2'

Collector 20 g/t

2'

MIBC 20 g/t

Rougher flotation

Concentrate

Fig. 2 Flow sheet of collector selection

5'

Tailing

Comprehensive Recovery of Oxygen Pressure Acid …

7

Table 2 Effect of collector type on the flotation of the leaching residue Collector

Product

Yield (%)

Sulfur grade (%)

Z-200

Concentrate

60.08

71.44

Tailing

39.92

Feed

100

45.94

Concentrate

62.31

66.78

Tailing

37.69

Feed

100

45.24

Concentrate

59.27

71.04

Tailing

40.73

Feed

100

45.99

Concentrate

59.65

70.73

Tailing

40.35

Feed

100

ADDTP

ETCM

EX

7.56

9.64

9.54

7.46 45.20

Sulfur recovery (%) 93.43 6.57 100 91.97 8.03 100 91.55 8.45 100 93.34 6.66 100

is presented in Table 2. From the table, the sulfur recoveries were separately 93.43, 91.97, 91.55, and 93.34% for Z-200, ADDTP, ETCM, and EX. Thus, the collecting abilities of ADDTP and ETCM were relatively weaker, while those of Z-200 and EX are close. In comparison, the sulfur grade achieved 71.44% with Z-200, while the grade was only 70.73% with EX. Therefore, among the four collectors, Z-200 possessed the optimal collecting performance for the sulfur in the leaching residue.

Effect of Collector Dosage The flow sheet of effect of Z-200 dosage on the sulfur grade and recovery of flotation concentrate is indicated in Fig. 3, and the flotation result is shown in Table 3. When the dosage of Z-200 was 20 g/t, the sulfur grade and recovery of the concentrate were relatively high, which were 71.35% and 93.49%, respectively. If the dosage was too low or too high, either the grade or the recovery was unsatisfying. Therefore, the most suitable dosage of Z-200 was 20 g/t.

Effect of Inhibitor Type In order to improve the separation of elemental sulfur from sulfides, three potential inhibitors including Na2 SO3 , ZnSO4 , and Na2 S were used. The flow sheet is displayed in Fig. 4, and the flotation result is presented in Table 4. It can be seen that the use of the three inhibitors was beneficial to improve the sulfur grade of the concentrate. In comparison, when their combination (i.e. Na2 S + ZnSO4 + Na2 SO3 : 350 + 350 +

8

G. Liu et al.

350) was adopted, the sulfur grade of the concentrate reached the maximum 81.05% and the sulfur recovery also arrived at 90.36%. Pressure acid leaching residue

Blank flotation

5' 2'

Z-200

2'

MIBC 20 g/t

Rougher flotation

Concentrate

5'

Tailing

Fig. 3 Flow sheet of effect of Z-200 dosage on the flotation of the leaching residue

Table 3 Effect Z-200 dosage on the flotation of the leaching residue Z-200 dosage (g/t)

Product

5

Concentrate

57.53

70.45

Tailing

43.47

10.80

10.38

101.00

45.23

100.00

Concentrate

60.61

69.61

92.99

Tailing

39.39

8.07

7.01

100.00

45.37

100.00

Concentrate

60.23

71.35

93.49

Tailing

39.77

7.53

6.51

100.00

45.97

100.00

Concentrate

59.62

71.45

93.23

Tailing

40.38

7.66

6.77

100.00

45.69

100.00

Concentrate

59.05

70.86

92.39

Tailing

40.95

8.42

7.61

100.00

45.29

100.00

Feed 10

Feed 20

Feed 30

Feed 40

Feed

Yield (%)

Sulfur grade (%)

Sulfur recovery (%) 89.62

Comprehensive Recovery of Oxygen Pressure Acid …

9

Pressure acid leaching residue

Fig. 4 Flow sheet of inhibitor selection

Blank flotation

5'

3'

Inhibitor

2'

Z-200 20 g/t

2'

MIBC 20 g/t

Rougher flotation

Concentrate

5'

Tailing

Table 4 Effect inhibitor type on the flotation of the leaching residue Inhibitor (g/t)

Product

Na2 SO3 :500

Concentrate Tailing Feed

ZnSO4 :1000

Concentrate Tailing Feed

Na2 S:1000

Concentrate Tailing Feed

Na2 S + ZnSO4 + Na2 SO3 : 350 + 350 + 350

Concentrate Tailing Feed

Yield (%) 52.33

Sulfur grade (%) 78.62

Sulfur recovery (%) 90.45

47.67

9.11

9.55

100.00

45.48

100.00

52.87

78.36

90.12

47.13

9.64

9.88

100.00

45.98

100.00

52.11

79.16

90.03

47.89

9.54

9.97

100.00

45.82

100.00

51.19

81.05

90.36

48.81

9.07

9.64

100.00

45.92

100.00

Effect of Inhibitor Dosage The effect of Na2 SO3 + ZnSO4 + Na2 S dosage on the flotation of the oxygen pressure leaching residue was studied. The flow sheet is presented in Fig. 5, and the flotation result is indicated in Table 5. With the increase of dosage of three inhibitors, the sulfur grade of the concentrate gradually increased and reached its maximum 82.05% at the dosage of 300 g/t Na2 SO3 + 300 g/t ZnSO4 + 300 g/t Na2 S. After that, the grade started to decline. So, the optimal inhibitor dosage was 300 g/t Na2 SO3 + 300 g/t ZnSO4 + 300 g/t Na2 S.

10

G. Liu et al.

Fig. 5 Flow sheet of inhibitor selection

Pressure acid leaching residue

Blank flotation

5'

3'

Na 2SO 3+ZnSO 4+Na 2S

2'

Z-200 20 g/t

2'

MIBC 20 g/t

Rougher flotation

Concentrate

5'

Tailing

Table 5 Effect Na2 SO3 + ZnSO4 + Na2 S dosage on the flotation of the leaching residue Inhibitor dosage (g/t)

Product

Na2 SO3 + ZnSO4 + Na2 S: 100 + 100 + 100

Concentrate Tailing Feed

Na2 SO3 + ZnSO4 + Na2 S: 200 + 200 + 200

Concentrate Tailing Feed

Na2 SO3 + ZnSO4 + Na2 S: 300 + 300 + 300

Concentrate Tailing Feed

Na2 SO3 + ZnSO4 + Na2 S: 400 + 400 + 400

Concentrate Tailing Feed

Yield (%) 52.59

Sulfur grade (%) 78.62

Sulfur recovery (%) 89.91

47.41

9.79

10.09

100.00

45.99

100.00

52.35

79.33

90.42

47.65

9.24

9.58

100.00

45.93

100.00

50.38

82.05

90.28

49.62

8.97

9.72

100.00

45.79

100.00

50.30

81.95

90.08

49.70

9.13

9.92

100.00

45.76

100.00

Flotation Flowchart Experiment Based on the above results, the sulfur grade and recovery separately achieved 82.05% and 90.28% after one-time blank rougher flotation and one-time agent-added rougher flotation. In order to further enhance the sulfur grade and recovery, the effects of the numbers of agent-added rougher flotation and cleaner flotation on the flotation of the residue were investigated, and the obtained results are shown in Table 6 and Table 7, respectively. As indicated in Table 6, when the number of agent-added rougher flotation increased to two, the sulfur recovery rose to 94.16%. Further increase of the number to three, no obvious increase of the sulfur recovery was observed. Therefore, the optimal agent-added rougher flotation number was two.

Comprehensive Recovery of Oxygen Pressure Acid …

11

Table 6 Effect agent-added rougher flotation number on the flotation of the leaching residue Rougher flotation number

Product

1

Concentrate Tailing Feed

2

Concentrate Tailing Feed

3

Concentrate Tailing Feed

Yield (%) 50.38

Sulfur grade (%) 82.05

Sulfur recovery (%) 90.28

49.62

8.97

9.72

100.00

45.79

100.00

62.94

68.73

94.16

37.06

7.24

5.84

100.00

45.94

100.00

63.30

67.56

94.01

36.70

7.42

5.99

100.00

45.49

100.00

Table 7 Effect cleaner flotation number on the flotation of the leaching residue Cleaner flotation number

Product

1

Concentrate

48.26

85.16

89.37

Middling 1

17.43

13.39

5.07

Tailing

34.31

7.44

5.55

100.00

45.98

100.00

Concentrate

44.32

91.88

88.53

Middling 1

12.75

14.45

4.01

Middling 2

8.80

9.8

1.88

Feed 2

Tailing

Sulfur grade (%)

Sulfur recovery (%)

34.13

7.53

5.59

100.00

45.99

100.00

Concentrate

44.33

92.15

89.21

Middling 1

11.70

14.05

3.59

Middling 2

7.44

8.98

1.46 0.28

Feed 3

Yield (%)

Middling 3 Tailing Feed

3.11

4.05

33.41

7.49

5.47

100.00

45.79

100.00

As presented in Table 7, when the number of cleaner flotation was one, the sulfur grade rose to 82.05%. As the number of cleaner flotation was increased to two, the sulfur grade further was enhanced to 91.88%. After that, the increase of the sulfur grade was not obvious, and thus, the optimal number of cleaner flotation was two.

12

G. Liu et al.

Closed-Circuit Flotation Experiment In order to study the effect of middling return on sulfur flotation, a closed-circuit flotation experiment was performed. The flow sheet consisted of one-time blank rougher, two-time agent-added roughers, and two-time cleaners. The dosages of collector, inhibitor, and foaming agent in the second agent-added rougher were half of the dosages used in the first agent-added rougher. The tailing of the first cleaner (middling 1) was returned to the blank rougher, and the tailing of the second cleaner (middling 2) went back to the first cleaner. The result is displayed in Table 8. A sulfur concentrate was obtained whose sulfur grade could reach 91.46% and sulfur recovery was also as high as 90.88%. The sulfur phase distribution of the concentrate is shown in Fig. 6. 83.25% of the sulfur in the concentrate occurred in the form of elemental sulfur, i.e., the purity of elemental sulfur concentrate product arrived at 83.25%. The recovery of elemental sulfur could be calculated as Eq. (1). The chemical composition analysis for the tailing showed the contents of Zn, Pb, and Ag separately reached 7.23%, 3.28%, and 368 g/t. Therefore, these valuable metals were enriched in the tailing which can be mixed with lead concentrate as the raw material of pyro-metallurgical lead smelting, and thus Zn, Pb, and Ag can be recovered. Table 8 Result of closed-circuit flotation of the oxygen pressure leaching residue Product

Yield (%)

Sulfur grade (%)

Sulfur recovery (%)

Concentrate

45.14

91.46

90.88

Tailing

54.86

7.55

9.12

100.00

45.43

100.00

Feed

100 90

91 83.25

80 70

Content Distribution

%

60 50 40 30 20

8.09

10 0

8.85 0.12

Simple substance sulfur

Sulfide

0.15

Sulfae

Sulfur phase Fig. 6 Distribution of sulfur phase of the obtained concentrate product

Comprehensive Recovery of Oxygen Pressure Acid …

Recovery = (45.14% × 83.25%)/(37.88%) = 99.2%

13

(1)

Conclusions An effective flotation process was proposed for selective recovery of elemental sulfur from an oxygen pressure acid leaching residue of zinc sulfide concentrate. The optimal closed flotation process is: one-time blank rougher, two-time agent-added roughers, and two-time cleaners, and the selected collector and inhibitor were Z-200 and Na2 SO3 + ZnSO4 + Na2 S. Under the optimum flotation condition, 99.2% of the elemental sulfur in the residue was recovered and the purity of concentrate product reached 83.25%. The tailing contained 7.23% of Zn, 3.28% of Pb, and 368 g/t of Ag, and thus, it can be used as an ingredient in conjunction with conventional lead concentrate to be the raw material of pyro-metallurgical lead smelting for recovering these valuable metals. Acknowledgements This research was funded by National Key Research and Development Program of China (Nos. 2018YFC1902005 and 2018YFC1902006).

References 1. Dong ZL, Jiang T, Xu B, Yang YB, Li Q (2019) An eco-friendly and efficient process of low potential thiosulfate leaching-resin adsorption recovery for extracting gold from a roasted gold concentrate. J Clean Prod 229:387–398 2. Dong ZL, Jiang T, Xu B, Yang JK, Chen YZ, Yang YB, Li Q (2020) Comprehensive recoveries of selenium, copper, gold, silver and lead from a copper anode slime with a clean and economical hydrometallurgical process. Chem Eng J 393 3. Gu Y, Zhang TA, Liu Y, Mu WZ, Zhang WG, Dou ZH, Jiang XL (2010) Pressure acid leaching of zinc sulfide concentrate. Trans Nonferrous Met Soc China 20:s136–s140 4. Li HL, Wu XY, Wang MX, Wang J, Wu SK, Yao XL, Li LQ (2014) Separation of elemental sulfur from zinc concentrate direct leaching residue by vacuum distillation. Sep Purif Technol 138:41–46 5. Padilla R, Vega D, Ruiz MC (2010) Pressure leaching of sulfidized chalcopyrite in sulfuric acid–oxygen media. Hydrometallurgy 86:80–88 6. Liu FP, Wang JL, Peng C, Liu ZH, Wilson BP, Lundström M (2019) Recovery and separation of silver and mercury from hazardous zinc refinery residues produced by zinc oxygen pressure leaching. Hydrometallurgy 185:38–45 7. Qin SC, Jiang KX, Wang HB, Zhang BS, Wang YF, Zhang XD (2020) Research on behavior of iron in the zinc sulfide pressure leaching process. Minerals 10:224–239 8. Rao S, Wang DX, Liu ZQ, Zhang KF, Cao HY, Tao JZ (2019) Selective extraction of zinc, gallium, and germanium from zinc refinery residue using two stage acid and alkaline leaching. Hydrometallurgy 183:38–44 9. Dong ZL, Jiang T, Xu B, Yang YB, Li Q (2017) Recovery of gold from pregnant thiosulfate solutions by the resin adsorption technique. Metals 7:555–572

14

G. Liu et al.

10. Ozberk E, Jankola WA, Vecchiarelli M, Krysa BD (1995) Commercial operations of the Sherritt zinc pressure leach process. Hydrometallurgy 39:49–52 11. Fan YY, Liu Y, Niu LP, Jing TL, Zhang TA (2019) Separation and purification of elemental sulfur from sphalerite concentrate direct leaching residue by liquid paraffin. Hydrometallurgy 186:162–169 12. Wang ZY, Cai XL, Zhang ZB, Zhang LB, Wang SX, Peng JH (2015) Separation and enrichment of elemental sulfur and mercury from hydrometallurgical zinc residue using sodium sulfide. Trans Nonferrous Met Soc China 25:640–646 13. Huang ZQ, Zhong H, Wang S, Xia LY, Zou WB, Liu GY (2014) Investigations on reverse cationic flotation of iron ore by using a Gemini surfactant: Ethane-1,2-bis (dimethyl-dodecylammonium bromide). Chem Eng J 257:218–228 14. Wang Z, Xu LH, Wang JM, Wang L, Xiao JH (2017) A comparison study of adsorption of benzohydroxamic acid and amyl xanthate on smithsonite with dodecylamine as co-collector. Appl Surf Sci 426:1141–1147 15. Xing P, Ma BZ, Wang CY, Wang L, Chen YQ (2018) A simple and effective process for recycling zinc-rich paint residue. Waste Manag 76:234–241 16. Xu B, Chen YZ, Dong ZL, Jiang T, Zhang BS, Liu GQ, Yang JK, Li Q, Yang YB (2021) Eco-friendly and efficient extraction of valuable elements from copper anode mud using an integrated pyro-hydrometallurgical process. Resour Conserv Recycl 164 17. Halfyard JE, Hawboldt K (2011) Separation of elemental sulfur from hydrometallurgical residue: a review. Hydrometallurgy 109:80–89

Effect of Laser Heat Treatment and Nitrogen Content in Shielding Gas on Precipitation of Widmanstätten Austenite in Lap Laser Welds of Duplex Stainless Steels Yunxing Xia, Kenshiro Amatsu, Fumikazu Miyasaka, and Hiroaki Mori Abstract Duplex stainless steel attracts attention because of its excellent mechanical properties and corrosion resistance, which characterize both the ferrite and the austenite phases. However, it’s well known that the welds are easily affected by the amount of heat input and that the performance may be deteriorated due to the imbalance of the phase. In particular, when laser welding is used, this phenomenon becomes more pronounced due to the effects of rapid solidification. This study investigated the efficiency of laser post weld heat treatment by laser beam, focusing on the austenite stabilizing effect of nitrogen, adding different nitrogen contents to the shielding gas for various lap laser welding. The effects of the two methods of promoting the precipitation of austenite and the influence on the precipitation morphology are compared. In addition, the tensile test was used to summarize the relationship between the amount of austenite precipitated and mechanical strength. After that, the Widmanstätten austenite in the laser beam welds was observed stereoscopically by etching and continuous sectioning. Keywords Laser beam welding · Lap welding · Precipitation behavior · Phase ratio · Nitrogen shielding gas · Duplex stainless steels · Post weld heat treatment

Y. Xia (B) · K. Amatsu · F. Miyasaka Graduate School of Engineering, Osaka University, 2-1 Yamadaoka, Suita 565-0871, Osaka, Japan e-mail: [email protected] Y. Xia New Business Promotion Department, Corporate Planning Division, Hisaka Works, Ltd., 2-12-7, Sonezaki, Kita-ku, Osaka City, Osaka 530-0057, Japan H. Mori JX Nippon Mining & Metals Joint Research Chair for Circular Economy Promotion, Osaka University, Suita, Japan © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_2

15

16

Y. Xia et al.

Introduction Duplex stainless steel contains the same amount of both ferrite and austenite phases. In recent years, in order to further reduce the cost, a new duplex stainless steel was being developed. The new duplex stainless steel is designed to promote the austenite precipitation, by reducing the Ni content and increasing the N content. This type is called lean duplex stainless steel and it has already attracted much attention because it is considered a substitute for ordinary austenitic stainless steel SUS316. The lean duplex stainless steel exhibits its yield strength as twice high as that of ordinary austenitic stainless steels, which makes it an attractive material for some precision equipment and medical fields. In addition, the lean duplex stainless steel shows superior resistance to pitting, crevice corrosion, stress corrosion, or corrosion fatigue under harsh environments such as seawater, high chloride ion content compared to ordinary stainless steel. However, during the welding process, it is difficult to maintain the superior mechanical and corrosion properties of duplex stainless steel because the thermal history changes the amount of ferrite and austenite in the welds. Such phenomena become more significant, when laser welding is used with high nitrogen content. The solidification of duplex stainless steel is fully ferritic, followed by the diffusioncontrolled solid-state ferrite to austenite transformation. When using low-energy input processes such as laser welding, rapid cooling restricts the austenite formation and disturbs the optimum phase balance in duplex stainless steel [1]. This condition also increases the risk of nitride formation due to the supersaturation of nitrogen in ferrite. In the meantime, the nitrogen content is important in promoting austenite formation, especially during autogenous welding, as it has a high diffusion rate and is a strong austenite former. Nitrogen loss will, therefore, limit the austenite formation during the welding of duplex stainless steel. In this study, under lap laser welding, two methods to promote the precipitation of austenite, post weld heat treatment and nitrogen mixing in shielding gas, were applied, and their effect, as well as changes in weld shape and microstructure, were compared. The precipitation behavior of the Widmanstädter austenite in the laser welds of the lean duplex stainless steel was investigated, and it was observed by immersion of penetration corrosion and continuous slicing polishing [2].

Experimental Material In this experiment, the duplex stainless steel (S82031) was subjected to a lap welding by laser beam welding. The specification of the sheet was 100 mm (Length) × 100 mm (Width) × 0.6 mm (Thickness), and the chemical composition thereof is shown in Table 1.

Effect of Laser Heat Treatment and Nitrogen Content in Shielding …

17

Table 1 Chemical compositions of materials used (mass%) Material

C

Mn

Ni

Cr

Mo

N

S82031

≤0.05

≤2.50

2.0–4.0

19.0–22.0

0.60–1.40

0.14–0.24

Table 2 Processing conditions of the laser beam welding Laser beam weld

Laser output

Speed (mm/min)

Focus distance

Shielding gas

1750 W

2000

31.4 mm (J.F.)

100%Ar Ar-10%N2 Ar-20%N2

4000 6000

Laser Beam Welding Condition and Shielding Gas Process parameters for laser beam welding are shown in Table 2. The IPG Photonics YLR-2000 fiber laser was used (wavelength: 1070 ± 10 nm, max power: 2000 W). The laser output is fixed, and the welding speed selects 3 conditions that can penetrate the weld. In addition, as another critical factor to control weld performance, research on TIG welding [3] has shown that nitrogen shielding gas can promote the precipitation of austenite in the welds of duplex stainless steel. To clarify the relationship between the proportion of nitrogen in the shielding gas and the ratio of precipitation of austenite under laser beam welding, the adjustment of the nitrogen content in shielding gas was added to the laser welding conditions.

Post Weld Heat Treatment Condition Post weld heat treatment is a means to reduce various adverse changes caused by welding and restore the usage performance close to that of the base metal. It is also reported that by adjusting the conditions of post weld heat treatment, the austenite phase in the laser welds of duplex stainless steel could be effectively promoted [4, 5]. The post weld heat treatment conditions in this experiment are shown in Table 3. Table 3 Post weld heat treatment conditions Laser output Post weld heat treatment

1000 W

Speed (mm/min) 800 1000 1200

Focus distance

Shielding gas

31.4 mm (J.F.) + 100 mm (D.F.)

100%Ar

18

Y. Xia et al.

Results and Discussion Welds Profile Figure 1 shows the microstructure of the duplex stainless steel laser-welded under different welding speeds and different shielding gases. Due to the influence of heat input, as the welding speed increases, the welds become narrower, and austenite precipitation also decreases. Adding 10% nitrogen (Fig. 1d–f) to the argon shielding gas (Fig. 1a–c) does not noticeably affect the shape and microstructure of the weld.

Microstructural in Welds As-Welded Figure 2 shows the overall microstructure of the weld under three different shielding gases when the welding speed is 2000 mm/min. It can be observed that by using the argon-nitrogen mixture as the shielding gas, the fraction of austenite (shown in white) increases with the rise of the fraction of nitrogen in the mixture. In contrast, it can be found that different shielding gases does not result in significant difference in the morphology of the phases, such as the precipitation morphology of intergranular (grain boundary), Widmannstätten and intragranular austenite particles. Under the shielding gas with nitrogen added, only the precipitation amount of intragranular austenite particles increased slightly. However, it should be noted that the absorption capacity of the nitrogen obtained from the shielding gas in the weld is far less than

Fig. 1 The effect of laser beam welding velocities and shielding gas on the microstructure of S82031 in welds (a As-welded (Ar) 2000 mm/min, b As-welded (Ar) 4000 mm/min, c As-welded (Ar) 6000 mm/min, d As-welded (Ar-10%N2) 2000 mm/min, e As-welded (Ar-10%N2) 4000 mm/min, f As-welded (Ar-10%N2) 6000 mm/min)

Effect of Laser Heat Treatment and Nitrogen Content in Shielding …

19

Fig. 2 Effect of laser beam welding velocities and shielding gas on the microstructure of S82031 in welds (a As-welded (Ar) 2000 mm/min, b As-welded (Ar-10%N2) 2000 mm/min, c As-welded (Ar-20%N2) 2000 mm/min)

that from the base material because it comes from the outside. Just as the influence of the partial pressure of gas and the nitrogen content in the mixed shielding gas on the amount of austenite precipitation mentioned in the research of arc welding by the Kokawa group [3], it can be inferred that similar parameters exist when laser welding is used. In practical applications, the nitrogen content in the shielding gas needs to be adjusted according to different material thicknesses, welding methods, and conditions. The enlarged microstructure of Fig. 2b also shows the precipitation of some nitrides. In terms of thermodynamics and kinetics, higher nitrogen content may help to form more austenite. However, when the austenite fraction is low, some nitrogen may be trapped at a certain distance from the austenite. As a result, this part of the nitrogen becomes saturated in the ferrite and finally combines with chromium to form precipitation of chromium nitrides. Therefore, it is important to seek the most suitable welding conditions to obtain a higher austenite fraction. It can restore the balance of the austenite and ferrite fractions in welds and promote the precipitation of austenite by absorbing nitrogen while reducing the precipitation of impurities in welds and reducing its impact on the performance of welds.

Microstructure in Welds After Post Weld Heat Treatment To better observe the effect of post weld heat treatment, the first pass of lap laser welding used a welding speed of 6000 mm/min with the least austenite precipitation. On the other hand, the purpose of the second pass is to reheat without melting. Therefore, the plan lets the laser heat a wider area defocused manner so that the

20

Y. Xia et al.

Fig. 3 The microstructure of S82031 in welds after post weld heat treatment (a PWHT (Ar) 800 mm/min, b PWHT (Ar) 1000 mm/min, c PWHT (Ar) 1200 mm/min)

welds does not melt. Unfortunately, due to the imperfect parameter settings, the surface of the welds was melted by the laser under certain conditions. But this does not affect the investigation of the effectiveness of post weld heat treatment in this experiment. Figure 3a1 and b1 are the microstructures of the upper half of the lap weld remelted by post weld heat treatment. Although Fig. 3c1 obtains the unmelted welds, the range of austenite fraction affected is minimal. Figure 3b shows the enlarged microstructure of the unmelted part under various conditions. Compared with the microstructure aswelded under the same lap laser welding condition in Fig. 1c, the precipitation of austenite has significantly been promoted. Although its morphology is still dominated by the Widmannstätten structure of the grain boundary and the granular shape in the ferrite grains [6], it is close to the 50% ferrite and 50% austenite. And it can also be found that when the austenite fraction is significantly increased, the precipitation of chromium nitrides is indeed suppressed.

Tensile Test Tensile tests were carried out on test pieces with different shielding gas conditions and post weld heat treatment. The test pieces were produced regarding the Section IX of the ASME Boiler and Pressure Vessel Code. Figure 4 shows the results of the test. Under these four welding conditions, the strength of As-welded (Ar-10%N2) is the strongest and the most stable, which can be considered the effect of nitrogen in the shielding gas. However, when the nitrogen is increased to 20%, the strength of Aswelded (Ar-20%N2) was decreased. Because of the increased quantity of nitrogen under the same welding conditions, excessive nitrogen in the ferrite matrix might

Effect of Laser Heat Treatment and Nitrogen Content in Shielding …

21

Fig. 4 Tensile strength results for laser welding pieces under different shielding gases pieces and post weld heat treatment pieces

accumulate to nitrides, then deterioration of strength. The strength of the specimen after post weld heat treatment is the weakest. This may occur due to the re-melting of the surface of the welds, which causes the upper half of the welds to become wider, but even so, it is basically consistent with the strength of the base metal. Therefore, it can be considered that the weld seam of laser welding is very narrow, the binding force produced extremely improves the tensile strength, compared with the precipitation of microstructure in welds, the shape of laser welds has a more significant influence on their tensile strength.

Widmanstätten Austenite The size and shape of Widmanstätten austenite under laser welding, including the thickness and density of the plate, are far inferior to traditional arc welding. After obtaining nitrogen from the argon-nitrogen mixed gas, although the austenite fraction of the laser welds is slightly increased, as shown in Fig. 5 the shape of the austenite of the Widmanstätten structure does not change much. This is because the precipitation of the Widmanstätten structure is basically at the ferrite grain boundary and then grows into the ferrite grain along a specific direction. Therefore, the size and shape of ferrite grains will also affect the growth of Widmanstätten austenite. Thus, the adjustment of nitrogen alone has almost no effect on the morphology of Widmanstätten structure under the same welding method. Moreover, the currently widely used observation means is carried out through the cross section of the weld. For structures with preferential growth directions such as Widmanstätten structure, there are large errors in identifying the actual size and shape. These errors are caused by the selected position of the cross section. Figure 6

22

Y. Xia et al.

Fig. 5 Microscope observation of Widmanstätten austenite in the laser welds of duplex stainless steel

Fig. 6 The microstructure in welds at the same position and different depths of section

is a microstructure diagram of three cross sections of the same sample piece selected with a polishing interval of 25 μm. It can be found that the Widmanstätten structure changes in size and shape with the migration of ferrite grains. Therefore, it is difficult to obtain the correct size of Widmanstatten austenite and determine its complete morphology. In later research, we will combine hundreds of cross-sectional figures with different polishing intervals to reshape a three-dimensional microstructure as Fig. 7 of the welds to observe it more effectively.

Fig. 7 The three-dimensional image what constructed by multiple captured images obtained by repeating polishing and observation with an optical microscope

Effect of Laser Heat Treatment and Nitrogen Content in Shielding …

23

Conclusions The influence of laser heat treatment and the nitrogen content in the shielding gas on the precipitation of austenite in the lap laser welds of duplex stainless steel are summarized as follows: (1)

(2)

(3)

(4)

During the laser beam welding of duplex stainless steel, when argon-nitrogen shielding gas is used, the shape of the weld is basically the same as the case when argon shielding gas is used. When the nitrogen fraction in the mixed gas increases, the austenite fraction increases slightly. Laser heat treatment can effectively promote the formation of austenite. Still, the setting of processing conditions is relatively strict, and the thickness of the material and the structure of the welds need to be adjusted. Nevertheless, the effect is more prominent than replacing the shielding gas. The tensile test results show that the strength in laser welds is most affected by the restraint characteristics. This means that laser beam welding is a very suitable welding method for duplex stainless steel. Adding nitrogen to the shielding gas also can effectively improve the tensile strength and stability of the weld. Still, it needs to match the welding conditions to reasonably control the diffusion of nitrogen. Nitrides in ferrite may deteriorate the tensile strength of the welds. The formation of austenite in the Widmanstätten structure is related to the content of nitrogen. Still, its growth is more limited by the heat history and the size and shape of the primary phase ferrite.

References 1. Omura T, Kushida T, Komizo Y (1999) Nitrogen distribution on rapid solidification in laser welded duplex stainless steels. Q J Jpn Weld Soc 17(3):448–455 2. Sato N, Adachi Y, Kawata H, Kaneko K (2012) Topological approach to ferrite/martensite dual-phase microstructures. ISIJ Int 52(7):1362–1365 3. Kokawa H, Okada J, Kuwana T (1993) Nitrogen absorption and microstructure of duplex stainless steel weld metal. Weld Int 7(5):384–389 4. Fukui F (1981) Weldability of duplux stainless steel. J Jpn Weld Soc 50(3):235–240 5. Nakade K, Ohe K, Kuroda T (2001) Precipitation behavior of σ phase for reheated duplex stainless steel weld metal. Q J Jpn Weld Soc 19(1):92–99 6. Lagerberg G, Wolff EG (1958) The effects of grain size and texture on the internal friction in α-iron due to interstitial solutes. Acta Metall 6(2):136–137

Influence of M-EMS Parameters on Flow Characteristics in a Bloom Mold Xiang-Lan Yang, Ming-Tao Xuan, Shan Wang, and Min Chen

Abstract For the purpose to improve mold electromagnetic stirring performance and bloom quality for a mold section of 250 mm × 300 mm, the influence of casting speed, structural parameters of the four-side-port submerged-entry nozzle as well as its submergence depth, and M-EMS pattern on the flow characteristics in the bloom mold were studied. The results showed that the electromagnetic stirring pattern had the most significant effect on level fluctuation among the above factors, and when intermittent reversal electromagnetic stirring pattern was loaded with a current intensity of 500 A, the turbulent kinetic energy and velocity of the free surface was 0.0016 m2 s−2 and 0.32 m s−1 , respectively. In addition, to control the level activity and meniscus shape, the current continuous loading time should not be less than 15 s at 500 A. The practical application showed that the quality of bearing steel was much improved by the application of the M-EMS under the above-recommended patterns. Keywords Continuous casting · Bloom · Mold · Flow characteristics · Electromagnetic stirring

Introduction Presently, overcapacity of China’s crude steels is still a problem that needs to be solved urgently, and improving the quality of steel is the main research direction [1]. The mold plays a key role in the continuous casting production process, and the flow state of molten steel in the mold has an important impact on the removal of inclusions, the prevention of mold slag involvement, and the formation of uniform solidified shells, thereby affecting the surface quality and internal quality of the blooms [2]. As is well known, mold electromagnetic stirring (M-EMS) is a significant technical means to improve the quality of blooms, and it is widely used for the production of various alloy steel products. Its working principle is to strengthen the movement of molten steel in the liquid phase cavity through the generated electromagnetic force, X.-L. Yang · M.-T. Xuan · S. Wang · M. Chen (B) School of Metallurgy, Northeastern University, Shenyang 110819, People’s Republic of China e-mail: [email protected] © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_3

24

Influence of M-EMS Parameters on Flow Characteristics …

25

thereby strengthening the convection, heat transfer, and mass transfer process of the molten steel, improving the elimination of the superheat of the molten steel in the mold and increasing the equiaxed crystal ratio of the bloom. In this way, the purpose of improving the quality of the cast bloom is achieved. The parameters of electromagnetic stirring vary depending on the cross-section and the steel grade, and proper electromagnetic stirring parameters can make the molten steel flow in the mold better to improve the quality of the bloom. For the analysis of the electromagnetic stirring process in continuous casting, it is difficult to obtain intuitive results from a single theoretical analysis. With the development of science and technology, the use of computers for simulation calculations has become an important method for studying electromagnetic stirring in continuous casting, and many studies have been carried out concerning the influence of EMS patterns on the electromagnetic field and flow field as well as the solidified shell [3–7]. These studies not only promote understanding the coupling behavior of the electromagnetic field and flow fields but also providing a valuable basis for optimizing the operating parameters. Although much progress has been made in the numerical simulation of M-EMS, few people have studied the effects of intermittent reversal in electromagnetic stirring on the flow field in the bloom mold [8–10]. For specific types of blooms, relevant research must be conducted to guide production and optimize process parameters. In this paper, a coupled mathematical model of the electromagnetic field and flow field in a 250 mm × 300 mm bloom mold using M-EMS is established. Previous studies have compared the influence of factors such as submergence depth, casting speed, side port inclination, side port height, and current intensity on the flow field in the mold, and these studies have found that EMS pattern has a greater effect. Therefore, the present work focuses on the analysis of the current intensity and intermittent reversal on the flow field in the mold, which provides a reliable theoretical basis for optimizing M-EMS parameters.

Mathematical Model Model Assumptions and Governing Equations In order to simplify the model and highlight the influence of electromagnetic stirring on the flow of molten steel, the following assumptions need to be made: (1) (2) (3) (4)

The influence of displacement current and molten steel motion on the electromagnetic field is ignored in the calculation; The molten steel is considered to be an incompressible Newtonian fluid at steady state; It is considered that the mold is full of high-temperature molten steel; The taper and vibration of the mold are ignored, and the influence of the protective slag on the surface of the molten steel is not considered.

26

X.-L. Yang et al.

The electromagnetic field generated by EMS can be calculated by solving Maxwell’s equations. Fluid flow is described by the law of conservation of mass and momentum. The fluid flow inside the bloom can be categorized as a turbulent flow in the liquid core and an interdendritic flow in the mushy zone. The turbulence effects of fluid flow inside bloom are modeled by the standard k − ε model. The following are continuity equation and momentum equation: ∂(ρu i ) =0 ∂ xi

(1)

where ρ is the fluid density, kg/m3 , and u i (i = 1, 2, 3) is the velocity vector, m/s.   ∂ ρu j u i =− ∂x j ∂x

∂ i ∂x j

∂p   (μl + μt ) ∂∂ux ij +

∂u j ∂ xi



(2) Fi

where μl is the dynamic viscosity, Pa s; p is pressure, Pa; μt is the turbulent viscosity.

Boundary Conditions and Property Parameters The length of the mold is 800 mm, to avoid the backflow at the mold outlet, the calculation length of the model has been extended to 1600 mm, and Fig. 1 shows the finite element model of no-load electromagnetic stirrer of the mold. In the model calculation, it is considered that both copper plate and magnetic yoke are isotropic materials, and their relative permeability is constant; Table 1 shows the property parameters of different materials and the specific boundary conditions are given as follows:

Fig. 1 Finite element model of M-EMS without load

Influence of M-EMS Parameters on Flow Characteristics … Table 1 Material property parameters used in calculation

(1) (2) (3) (4) (5)

27

Parameter

Value

Relative permeability of mold copper, H/m plate, H/m

1.0

Relative permeability of molten steel, H/m

1.0

Relative permeability of air, H/m

1.0

Relative permeability of yoke, H/m

1000

Conductivity of mold copper plate, S/m

5.51 × 107

Conductivity of molten steel, S/m

7.14 × 105

Set as the flow velocity perpendicular to the inlet, and the inlet velocity is determined by the casting speed and cross-sectional area; It is assumed that the flow at the outlet of the model is fully developed, namely, the normal derivative of each physical quantity along this section is zero; The top surface of the model is set as the free surface, and the free surface is defined as the adiabatic surface with zero-shear force; The wall of the mold and the nozzle are considered to be nonslip and adiabatic; Set the magnetic field lines to be parallel to the outer surface of the air surrounding the stirrer.

The SIMPLE algorithm is adopted for model calculation. When the convergence residual is less than 10–5 , the solution is finished.

Results and Discussion Effect of Current Intensity Figures 3 and 4 show the influence of current intensity on the behavior of the free surface, and the position of free surface is shown in Fig. 2. It can be seen that the flow field of the free surface presents an axisymmetric form with the submergedentry nozzle as the axis, and the velocity of the free surface in the thickness direction is higher than that in the width direction. In addition, due to the difference in the flow field, the swirling flow generated by electromagnetic stirring cannot reach the free surface, resulting in the velocity of the free surface and turbulent kinetic energy being only 0.00016 m2 s−2 and 0.11 m s−1 , which are similar to the situation without electromagnetic stirring. The increase of current intensity can greatly increase the fluctuation of the free surface. The turbulent kinetic energy and velocity of the free surface can be increased from 0.0012 m2 s−2 and 0.27 m s−1 at 400 A to 0.0016 m2 s−2 and 0.32 m s−1 at 500 A and 0.0026 m2 s−2 and 0.40 m s−1 at 600 A, respectively. Compared with the increase from 400 to 500 A, the increase from 500 to 600 A makes the velocity and turbulent kinetic energy of the free surface increase more, so it is not suitable to continue to increase the current intensity.

28

X.-L. Yang et al.

640

Mold

800

Free surface

255

EMS

200

300

Fig. 2 Schematic diagram of mold

(a) 300A

(b) 400A

(c) 500A

(d) 600A

Fig. 3 Effect of current intensity on turbulent kinetic energy of free surface

Influence of M-EMS Parameters on Flow Characteristics …

(a) 300A

(c) 500A

29

(b) 400A

(d) 600A

Fig. 4 Effect of current intensity on free surface velocity

Effect of Intermittent Inversion Figure 5 shows the free surface characteristics at 400 A. During continuous stirring, the free surface velocity and the maximum difference of liquid level can reach about 0.18 m s−1 and 8 mm, respectively. In contrast, intermittent reversal stirring hinders the formation of a continuous and stable flow field and prevents momentum from being transmitted to the free surface, so it can effectively reduce the liquid level difference and velocity to within 1 mm and 0.06 m s−1 . Figure 6 shows the influence of intermittent time on free surface velocity and outlet velocity, respectively. Generally speaking, in order to give full play to the role of electromagnetic metallurgy, the shorter the intermittent time of electromagnetic stirring is, the better the efficiency of electromagnetic metallurgy is. However, the results of Fig. 6 indicate that the free surface fluctuates greatly when the resting time is 2 s, while the fluctuation of 5 and 8 s is roughly the same. The reason for the phenomenon is that when the intermittent time is 2 s, the velocity in the mold decreases slightly, and the molten steel still has a certain inertia. The addition of reversed electromagnetic forces will cause certain adverse effects on the free surface. When the intermittent time is greater than 5 s, the influence on the free surface is significantly reduced.

30

X.-L. Yang et al.

(a) Stir without intermittent stirring

(b) End of corotation

(c) Start of inversion

(d) End of inversion

Fig. 5 Fluctuation characteristics of free surface at 400 A: a continuous stirring, b–d intermittent inversion electromagnetic stirring in each period

0.18 0.16 0.06

Maximum velocity of outlet, m⋅s-1

Surface velocity, m⋅s-1

0.20 15s-2s-15s 15s-5s-15s 15s-8s-15s Continuous stirring

0.05 0.04 4

5 cycle, -

(a) Free surface velocity

Fig. 6 Effect of resting time on velocity

6

0.5 0.4 0.3 0.2 0.1 0.0 4

15s-2s-15s 15s-5s-15s 15s-8s-15s Continuous stirring

5 Cycle, -

(b) Outlet velocity of mold

6

Influence of M-EMS Parameters on Flow Characteristics …

31

Figures 7 and 8 show the influence of the electromagnetic force loading time on the maximum liquid level difference (L) and the maximum velocity of the mold outlet, respectively. It can be seen from Fig. 7 that the reduction of the electromagnetic force loading time can reduce the increase in the momentum of the molten steel, the maximum liquid level difference is reduced, and then the fluctuation of meniscus becomes smaller and the level of liquid level activity is reduced, in order to improve the utilization rate of the electromagnetic force, it is necessary to increase the duration of electromagnetic force. From the influence of the loading time on the maximum velocity of the mold outlet in Fig. 8, it can be seen that the velocity of molten steel increases with the increase of the electromagnetic force loading time. However, when the loading time is longer than the 20 s, the electromagnetic force cannot continue to increase the outlet velocity in the later stage of loading but will propagate the momentum to the vicinity. Therefore, under the condition of 400 A, when the current loading time is 15 s, the maximum stirring speed that this current intensity can provide 3

8

7s-5s-7s 10s-5s-10s 15s-5s-15s 20s-5s-20s

2

ΔL, mm

ΔL, mm

6

7s-5s-7s 10s-5s-10s 15s-5s-15s 20s-5s-20s 30s-5s-30s

1 0.0

0.5

4 2

1.0

Cycle, -

1.5

0

2.0

0.0

(a) 400A

0.5

1.0

Cycle, (b) 500A

1.5

2.0

0.7

Maximum velocity of outlet, m⋅s-1

Maximum velocity of outlet, m⋅s-1

Fig. 7 Influence of electromagnetic force loading time on maximum liquid level difference

7s-5s-7s 10s-5s-10s 15s-5s-15s 20s-5s-20s 30s-5s-30s

0.6 0.5 0.4 0.3 0.2 0.1

0.0

0.5

1.0 Cycle, -

(a) 400A

1.5

2.0

0.7

7s-5s-7s 10s-5s-10s 15s-5s-15s 20s-5s-20s

0.6 0.5 0.4 0.3 0.2 0.1

0.0

0.5

1.0 Cycle, -

1.5

2.0

(b) 500A

Fig. 8 Influence of electromagnetic force loading time on the maximum velocity at the outlet

32

X.-L. Yang et al.

can basically be reached, that is, the loading time when the current intensity is 400 A should not be less than 15 s. In contrast, the electromagnetic force at 500 A is greater, and the load time is 10 s to basically reach the peak value, that is, the minimum load time should not be less than 10 s when the current intensity is 400 A. Combined with the free surface fluctuation, it can be seen that the free surface fluctuation is low at 400 A and the loading time should not be less than the 30 s; at 500 A, the free surface fluctuation increases slightly but is still at a low level, and the electromagnetic loading time is not less than 15 s, the peak velocity of the free surface is slightly higher than 0.1 ms−1 . Therefore, when the current intensity is 400 A, the current loading time should not be less than the 30 s; when the current intensity is 500 s, the loading time should not be less than 15 s.

Practical Application The practical application results of the optimized scheme show that when producing GCr15 bearing steel, the A-type coarse inclusions are reduced from grade 1.0 to grade 0.5, the B-type coarse inclusions are reduced from grade 1.5 to grade 0.5, and the C-type inclusions are all reduced from grade 0.5 down to grade 0, which meets the production requirements.

Conclusions Based on numerical simulation, M-EMS pattern on the flow characteristics within the bloom mold sized 250 mm × 300 mm for bearing steel production have been investigated and the new findings are summarized as follows: (1)

(2)

Increasing the current intensity caused the stirring speed in the mold to increase rapidly. The free surface velocity and turbulent kinetic energy also increased from 0.00016 m2 s−2 and 0.11 m s−1 at 300 A and 0.0012 m2 s−2 and 0.27 m s−1 at 400 A to 0.0016 m2 s−2 and 0.32 m s−1 at 500 A and 0.0026 m2 s−2 and 0.40 m s−1 at 600 A. After increasing to 600 A, the velocity in the mold was higher than the empirical value, so 500 A can be regarded as the optimal current intensity. The electromagnetic loading method of intermittent inversion can make the flow in the mold showing periodic changes, and at the same time, the free surface activity and the fluctuation of the meniscus in the mold can be greatly reduced, so as to reduce the liquid level difference and obtain a more stable flow field. The increase of the electromagnetic force loading time can increase the peak value of the stirring speed and the free surface velocity in the mold. In order to ensure the minimum free surface activity, the current loading time

Influence of M-EMS Parameters on Flow Characteristics …

33

should be longer than the 30 s at 400 A; the current loading time should not be less than 15 s at 500 A. Acknowledgements The authors gratefully acknowledge the National Natural Science Foundation of China (No. 51974080, 52174301), which has made this research possible.

References 1. He ML, Wang N, Chen M et al (2017) Physical and numerical simulation of the fluid flow and temperature distribution in bloom continuous casting mold. Steel Res Int. https://doi.org/10. 1002/srin.201600447 2. Yin YB, Zhang JM, Wang B et al (2019) Effect of in-mould electromagnetic stirring on the flow, initial solidification and level fluctuation in a slab mould: a numerical simulation study. Ironmak Steelmak 46(7):682–691. https://doi.org/10.1080/03019233.2018.1454998 3. Geng X, Li X, Liu FB et al (2015) Optimization of electromagnetic field and flow field in round billet continuous casting mould with electromagnetic stirring. Ironmak Steelmak 42(9):675– 682. https://doi.org/10.1179/1743281215Y.0000000015 4. Fang Q, Ni HW, Wang B et al (2017) Effects of EMS induced flow on solidification and solute transport in bloom mold. Metals. https://doi.org/10.3390/met7030072 5. Chen W, Wang BX, Zheng N et al (2013) Coupled numerical simulation study on electromagnetic field, flow field and heat transfer in ∅210mm round billet mold with M-EMS. Adv Mater Res 651:722–727. https://doi.org/10.4028/www.scientific.net/AMR.651.722 6. Li YG, Sun YH, Bai XS (2021) Numerical simulation of the flow, solidification, and solute transport in a billet mold under electromagnetic stirring. Metall Res Technol. https://doi.org/ 10.1051/metal/2021015. 7. Liu HP, Xu MG, Qiu ST et al (2012) Numerical simulation of fluid flow in a round bloom mold with in-mold rotary electromagnetic stirring. Metall Mater Trans B 43:1657–1675. https://doi. org/10.1007/s11663-012-9737-0 8. Asad A, Kratzsch C, Schwarze R (2016) Numerical investigation of the free surface in a model mold. Steel Res Int 87(2):181–190. https://doi.org/10.1002/srin.201400600 9. Li YJ, Li L, Zhang JQ (2017) Study and application of a simplified soft reduction amount model for improved internal quality of continuous casting bloom. Steel Res Int. https://doi.org/ 10.1002/srin.201700176 10. Zhang J, Wang EG, Deng AY et al (2010) Numerical simulation of flow field in bloom continuous casting mold with electromagnetic stirring. Adv Mater Res 146–147:272–276. https:// doi.org/10.4028/www.scientific.net/AMR.146-147+272

Numerical and Physical Simulations of Bottom Blowing Process Optimization of 120t Refining Ladle Shan Wang, Min Chen, Ming-Tao Xuan, and Xiang-Lan Yang

Abstract Numerical and hydraulic models were established for a 120 ton elliptical ladle, and the influence of the arrangement of the purging plugs and the flow rate of argon blowing on the flow field and mixing time in the ladle was studied. The results showed that the mixing time was 66 s and dead volume percentage of 18% for the prototype location of the purging plugs. The arrangement of the double plugs with 120° at 0.6R of the major axis off-centered positions in the bottom of ladle was the considered to be the optimal scheme, with the mixing time decreased from to 46 s and dead volume percentage of 16% under the flow rate of 500 NL/min, and the practical application showed the satisfied performance. Keywords Refining ladle · Bottom argon blowing · Flow field · Mixing time · Numerical and physical simulation

Introduction Argon stirring is one of the most widely used metallurgical methods in the ladle refining process for the production of clean steel [1] due to its performance on uniform composition and temperature of liquid steel, removal of impurities as well as the advantage of facilitating operation and low cost [2]. To improve the performance of the bottom argon blowing ladle refining, many studies have been carried out about the influence of the purging plugs arrangement and gas supplying system on mixing time and temperature distribution of liquid steel. Till now, most of the studies have been conducted concerning the circular (bottom) ladle [3, 4], and there is little research on elliptical ladle. It is easy to understand, the flow field in an elliptical ladle should be different to the circular ones, and thus the refining performance is influenced. Therefore, the present work investigated the influence of the purging plugs arrangement on flow field and temperature distribution of liquid steel by numerical and physical simulation based on a practical elliptical ladle, for the purpose of S. Wang · M. Chen (B) · M.-T. Xuan · X.-L. Yang School of Metallurgy, Northeastern University, Shenyang 110819, People’s Republic of China e-mail: [email protected] © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_4

34

Numerical and Physical Simulations of Bottom Blowing … Table 1 Correspondence of the bottom blowing gas flow rate between the model and prototype

35

Prototype (m3 h−1 )

Model (m3 h−1 )

Prototype (m3 h−1 )

Model (m3 h−1 )

6

0.02

36

0.14

12

0.05

42

0.16

18

0.07

48

0.18

24

0.09

54

0.21

30

0.12

60

0.23

providing some key technological parameters to improve the performance of bottom argon blowing in practical production.

Study Methods Physical Simulation Based on the similarity principle [5], a 1: 4 scale model ladle of transparent plastic was fabricated. In the water modeling process, water was used to simulate liquid steel and air was used to simulate argon. The mixing time was obtained by measuring the electric conductivity of the aqueous solution using stimulus-responsiveness method, and the flow field was displayed using color ink as tracer. According to the similar third law, under the condition that the modified Froude criterion Fr’ is equal [6], the corresponding relationship between the ladle prototype and the bottom blowing flow rate of the model is shown in Table 1.

Numerical Simulation Mathematical Model and Model Hypothesis The Euler-Euler method was used for numerical simulation. During the process of bottom argon blowing, the molten steel was in a complex turbulent flow state. In this paper, 1/2 of the ladle was selected for equal-scale modeling, and the hexahedral structure grid was used for division. The total number of grids was about 270,000. This experiment was based on the following assumptions: (1) (2)

The molten steel in the ladle is an incompressible viscous fluid, and the ladle molten pool is full of the liquid phase in the initial state. Bubble buoyancy is the driving force that drives the circulating flow of molten steel. Ignoring the influence of slag on the flow on the upper surface of the

36

S. Wang et al.

Fig. 1 Ladle meshing

(3)

molten steel, the upper surface of the ladle is regarded as a non-slip horizontal plane. The bubbles are rigid spheres of uniform size with the same diameter. The aggregation and fragmentation of bubbles are ignored (Fig. 1).

Control Equations and Boundary Conditions The continuity equation, the momentum conservation equation, and the standard k-ε model [7] were used in the numerical simulation. The boundary conditions were given according to the experimental conditions. (1)

(2)

(3)

Each argon blowing hole was set as the inlet boundary condition. The volume fraction of molten steel at the entrance was 0, and the volume fraction of argon was 1. The flow of argon was determined according to the flow of working conditions. The vacuum chamber and the free liquid surface of the ladle were set as the exit boundary conditions, and the shear stress on the surface was ignored. The gas reaches the exit boundary and escapes freely at the above floating velocity. It was the standard that the molten steel cannot flow out there. The non-slip boundary condition was adopted for all walls, and the pressure normal gradient was set to zero. The wall function method was used to process the nodes in the near-wall area.

Study Scheme Figure 2 is a schematic diagram of the original arrangement of purging plugs and the arrangement of the ladle bottom venting elements in the design scheme. The position marked in red represents the layout of the prototype, which is a major axis off-centered of 0.67R and an included angle of 100°. The blue position is the

Numerical and Physical Simulations of Bottom Blowing …

37

Fig. 2 Schematic diagram of the dimensions of the model ladle and the arrangement of purging plugs

optimized scheme, located at the major axis off-centered of 0.6R and an included angle of 120°.

Results and Discussion Influence of Bottom Blowing Arrangement on Mixing Time Figure 3 shows the influence of the arrangement of purging plugs on mixing time under various flow rates of the bottom gas blowing. It is observed that the mixing time was changed with the included angels under the various blowing gas flow rate for all the three radial position of the purging plugs, and it is observed that the included angel of 100°–120° showing minimum values of time. In addition, it is observed that the mixing time slightly decreased with increasing the radial position of the purging plugs from 0.5R to 0.67R. As well known, it is necessary to consider the circulation of molten steel in the ladle and the washing of the gas on the ladle wall, and the position of the purging plugs is generally arranged at 0.5R–0.67R. Comprehensively considering the stirring performance and gas washing to the ladle wall, it is considered that the arrangement of the double purging plugs with 0.6R and the included angle of 120° is the optimized scheme. Figure 4 shows the comparison of the mixing time between the original scheme and the optimized ones under various argon blowing flow rates. It can be seen that the mixing time decreased sharply with increasing argon blowing flow rate to 300 NL/min for both schemes, and then decreased gradually. Therefore, it is considered

38

S. Wang et al. 400 200NL/min 600NL/min 1000NL/min

300

Mixing time (sec)

Mixing time (sec)

400

200 100 0 40

60

80

200NL/min 600NL/min 1000NL/min

300 200 100 0 40

100 120 140 160 180 200

60

80

100 120 140 160 180 200

Include angle (°)

Included angle (°)

(a) 0.5R

(b) 0.6R

Mixing time (sec)

400 200NL/min 600NL/min 1000NL/min

300 200 100 0 40

60

80

100

120 140

160 180

200

Included angle (°)

(c) 0.67R Fig. 3 Influence of bottom blowing arrangement on mixing time under various gas flow rates

500

Mixing time sec

Fig. 4 Effect of different arrangement on mixing time under various gas flow rates

0.67R-100°(original scheme) 0.6R-120°(optimal scheme)

400 300 200 100 0

200

400

600

800

1000

1200

Flow rate NL/min

that the proper flow rate of argon blowing should be around 300 NL/min for ladle refining. It is also observed from the figure that the mixing time is obviously shorter for the optimized scheme comparing to the original ones in the low argon blowing flow rate region (less than 300 NL/min), this result indicates that the optimized scheme should have a better refining performance.

Numerical and Physical Simulations of Bottom Blowing …

(a) 0.6R

39

(b) 0.67R

Fig. 5 Influence of the purging plug arrangement on circles of the flow field

Figure 5 shows the streamline diagram of molten steel in the original scheme and the optimal scheme. The molten steel forms a gas–liquid two-phase flow driven by the bottom blowing argon gas, and the surrounding molten steel was continuously drawn in due to the large center velocity of the stream. At the liquid level on the top of the ladle, the gas escapes the molten pool. After the molten steel reaches the liquid level, it returns to the molten pool and flows down the sidewall of the ladle. The molten steel is driven by the gas to form a cycle again and again. The velocity of the molten steel in the gas–liquid two-phase flow area is the highest, but the circulation has a small effect on the center area of the vortex and the angle area between the bottom of the bag and the bag wall. The molten steel has a low flow rate and poor activity, that is, a dead zone is formed. The ladle is elliptical, so the influence of the circulation on the molten steel in the long axis direction is very important, and the corrosion of the ladle wall in the short axis direction should be considered. When the angle between the two purging plugs was 0.6R at 120°, the formed circulation can effectively improve the flow of molten steel in the long axis direction and reduce the erosion of the ladle wall by the gas in the short axis direction. Since the purging plugs of the original scheme were installed on the circumference of 0.67R, the formed eddy current corrodes the ladle wall surface more seriously, and the included angle of the purging plugs of 100° was not conducive to improving the mixing effect. When the optimized scheme was adopted, it was beneficial to reduce the dead volume percentage and improve the mixing effect to reduce the corrosion of the ladle wall surface. When the angle between the two purging plugs was 0.6R at 120° the argon blowing flow rate is 500 NL/min, the dead volume percentage was the smallest, which is 16%.

Practical Application Considering the mixing time in the bottom blowing process, the scouring of the gas– liquid mixed flow on the ladle wall, the fluctuation of the liquid level, and the volume

40

S. Wang et al.

of the dead zone, the bottom blowing hole are selected at a distance of 0.6R from the center of the bottom of the ladle, and the included angle of the purging plugs is 120°. Renovate the original ladle bottom blowing system. Using the optimized scheme increases the inclusion removal rate of Q345B steel by 4% and the inclusion removal rate of 82B steel by 3.5%, which is a significant improvement compared with the original scheme.

Conclusions Through the establishment of the physical model and numerical, the influence of the bottom blowing flow rate on mixing time and dead volume percentage were studied. Through the analysis and calculation of the data, the mixing time and dead volume percentage of the ladle were obtained and the conclusions are as follow: (1)

(2)

(3)

(4)

The mixing time and dead volume percentage in the ladle were affected by the angle of the purging plugs and blowing flow rate. Under the same Circle radius and blowing conditions, the mixing time decreased first and then increased, with the increase of the angle from100° to 180°. The mixing effect was perfect when the double plugs with 120° at 0.67R of the major axis off-centered positions and the flow rate of 500 NL/min. By measuring the mixing time at different angles of 0.5R, 0.6R, and 0.67R, it can be seen when double plugs were arranged at a distance of 0.6R from the major axis off-centered positions of 120°, and the flow rate of 500 NL/min, the mixing time in the ladle Shorten to 46 s. The mixing efficiency of the ladle has been significantly improved, and the erosion of the ladle wall can be effectively reduced. When purging plugs were arranged at 0.6R and the angle of 120°, the proportion of dead zone in the ladle decreases first and then increases with the increase of the argon blowing flow rate. When the argon blowing flow rate was 500 NL/min, the proportion of dead volume percentage in the ladle was 16%. The dead volume percentage of the original scheme was 18%. Compared with the original scheme, this scheme reduces the dead volume percentage. The impurity removal rate of Q345B steel produced by the optimized scheme was increased by 4%, and the impurity removal rate of 82B steel was increased by 3.5%.

Acknowledgements The authors gratefully acknowledge the National Natural Science Foundation of China (No. 51974080, 52174301), which has made this research possible.

Numerical and Physical Simulations of Bottom Blowing …

41

References 1. Yang J, Jin HY, Zhu MY et al (2019) Physical simulation of molten steel homogenization and slag entrapment in argon blown ladle. Processes 7(8):14–15 2. Maldonado-Parra FD, Ramirez-Argaez MA, Conejo AN et al (2011) Effect of both radial position and number of purging plugs on chemical and thermal mixing in an industrial ladle involving two phase flow. ISIJ Int 51(7):1110–1118 3. Wang Y, Ai XG, Liu F et al (2017) Physical simulation of mixing behavior of symmetrical alternate bottom blowing in ladle with double plugs. China Metall 27(7):18–21 (in Chinese) 4. Kaizawa A, Kamano H, Kawai A et al (2010) Thermal and flow behaviors in heat transportation container using phase change material. Energy Convers Manag 49(4):698–706 5. Zhan ZH, Qiu ST, Yin SB et al (2017) Research on bottom argon blowing process water model of 135t LF ladle furnace. Hot Work Technol 46(15):98–101 (in Chinese) 6. Llanos CA, Garcia-hernandez S, Ramos-Banderas JA et al (2010) Multiphase modeling of the fluid dynamics of bottom argon bubbling during ladle operations. ISIJ Int 50(3):396–402 7. Xiao ZQ, Zhu MY (2006) Application of numerical simulation analysis technology in metallurgical process [M]. Metallurgical Industry Press (in Chinese)

Parametric Study of Mold Electromagnetic Stirring: Effects of Load Condition and Copper Resistivity Qilan Li, Lifeng Zhang, and Jing Zhang

Abstract Accuracy and precision of numerical simulation were affected by many factors, such as mesh grid size, material properties, model simplification. Mold electromagnetic stirring was applied widely to steelmaking industry, which had a great influence on phenomena inside mold. To better calculate the magnetic field distribution and the magnetic flux density, a numerical model of 250 mm × 280 mm cast billet was proposed to investigate the influence of load conditions and mold copper resistivity on the magnetic flux density inside the mold under different stirring frequencies. The distribution of the magnetic flux density along the mold axis was studied. There was little influence of load and casting conditions on the magnetic flux density. The magnetic flux density changed more obviously under different copper resistivity. The magnetic flux density increased with the increase of mold copper resistivity due to the reduced loss of the magnetic field from the coils to the mold center. Keywords Mold electromagnetic stirring · Copper resistivity · Materials property · Simulation

Introduction Electromagnetic stirring technology was applied in the continuous casting process to improve the quality of casting billets. According to the installation position of

Q. Li School of Metallurgical and Ecological Engineering, University of Science and Technology Beijing, Beijing 100083, China L. Zhang (B) State Key Laboratory of Metastable Materials Science and Technology, Yanshan University, Qinhuangdao 066004, Hebei, China e-mail: [email protected] J. Zhang School of Vehicle and Energy, Yanshan University, Qinhuangdao 066004, Hebei, China © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_5

42

Parametric Study of Mold Electromagnetic Stirring: Effects of Load …

43

the stirrer, electromagnetic stirring device is divided into three types: Mold Electromagnetic stirring (M-EMS), Secondary Cooling Electromagnetic stirring (S-EMS), and Final Electromagnetic stirring (F-EMS). The M-EMS located around the mold is widely used in the steelmaking industry, which plays a positive role in increasing the casting speed, enhancing the surface quality, reducing bubbles and inclusions inside the solidified shell, improving the cleanliness of the molten steel, controlling the microstructure of billets, and enlarging the equiaxed crystal zone. To determine the best stirring parameters including the current intensity, current frequency, and installing location, numerical simulations were conducted to calculate the magnetic field. A great deal of studies have investigated the influence of stirring parameters, such as the current intensity and frequency, on the magnetic flux density and magnetic force. Trindade [1] built a numerical model showing the relationship between the current and magnetic flux density and observed that the copper mold attenuation was a function of the frequency and torque. Liu [2] described the three-dimensional electromagnetic field distribution and the electromagnetically driven flow characteristics in a round-bloom mold with a low-frequency electromagnetic stirrer. Ren [3] studied the effect of stirring current on induced flow characteristics, heat transfer, and solidification. According to the simulation and experiment results of Wang [4], the superheat dissipation of the molten steel and columnar to equiaxed transition (CET) could be promoted by M-EMS. Ambrish [5] found that frequency had a great influence on the stirring intensity and solidification behavior while the stirrer width and current had little influence by a mathematic model of the in-mold electromagnetic stirrer. The stirring of the molten steel and fluid flow in the mold as well as the magnetic flux density and magnetic force under different current intensity and frequencies were studied by Dang [6] and Li [7], respectively. Wang [8, 9] investigated the electromagnetic field, fluid flow, solidification, inclusions removal, and slag entrainment, then solved the transport of the solute after the fluid flow reached a steady state. Zhang [10] established a bipolar electromagnetic stirrer which enlarged the center equiaxial crystal zone compared with the traditional continuous casting electromagnetic stirrer. Although the application of numerical simulations to study M-EMS was already relatively mature, the main parameters in electromagnetic stirring simulations were different in published studies. The electrical conductivity, which was the reciprocal of resistivity, was always varied. Electrical conductivity of the mold copper mainly used is listed in Table 1. On the basis of the parameters reported, the effect of the mold copper resistivity on the accuracy of the numerical simulation was studied. On the other hand, simulating results under the load condition were usually compared straightly with experimental results under the no-load condition in the industry or laboratory experiments to verify the accuracy of the electromagnetic model.

44

Q. Li et al.

Table 1 Electrical conductivity of the mold copper in published studies Electrical conductivity of the mold copper Resistivity of the mold copper ( m) (S/m)

References

5.814 × 107

1.72 × 10–8

[5]

5.6 ×

107

1.79 × 10–8

[6, 7]

4.7 × 107

2.13 × 10–8

[1, 2, 4]

4.0 ×

2.50 ×

[10]

107

10–8

3.87 × 107

2.58 × 10–8

[6]

3.18 × 107

3.14 × 10–8

[1–4, 7]

1.25 ×

7.81 ×

[8]

107

10–8

Mathematical Formulation The schematic diagram of the mold electromagnetic stirrer installation location and the M-EMS model is shown in Fig. 1. The total height of the mold was 800 mm with an effective height of 700 mm. The section size of the billet was 280 mm × 250 mm, while the inner section size of the mold was 287 mm × 260 mm and the thickness of the copper plate was 14.5 mm. In the current model, the air gap between the mold copper and solidified shell was considered, with the thickness of 3.5 mm in the X direction and 5 mm in the Y direction. The electromagnetic field was calculated by the ANSYS software. The electromagnetic field generated by the alternating current (AC) can be calculated by

Fig. 1 Schematic diagram of a the stirrer and mold, b the M-EMS model

Parametric Study of Mold Electromagnetic Stirring: Effects of Load …

45

Maxwell’s equations as follows: ∇×E =−

∂B ∂t

(1)

∇×H =J

(2)

∇·B=0

(3)

where E is the electric field intensity, V/m; B is the magnetic flux density, T; t is time, s; H is the magnetic field intensity, A/m; J is the current density, A/m2 .

Effect of Load on Calculating Results Parameters selected in the current study are listed in Table 2 to find the difference in the magnetic flux density between the load and no-load conditions. The comparison of the magnetic flux density with and without the molten steel in the mold is shown in Fig. 2. The peak of the magnetic flux density was 21.06 mT, 19.61 mT, and 17.88 mT under the frequency of 2 Hz, 4 Hz, and 6 Hz, respectively, with the molten steel in the mold, and that was 21.21 mT, 20.05 mT, and 18.64 mT under no-load condition. The load condition has little effect on calculation results, which can be ignored in general. However, the effect of frequency on the magnetic flux density without the molten steel was more obvious than that with the molten steel under the same current density of 150 A. The no-load condition, without the molten steel, was recommended to ensuring the accuracy of the numerical model, especially under high-frequency conditions. Table 2 Parameters used in the current study

Parameter

Value

Current intensity of M-EMS (A)

150

Current frequency of M-EMS (Hz)

2, 4, 6

Copper resistivity ( m)

8.0 × 10–8

Molten steel resistivity ( m)

1.4 × 10–6

Relative permeability of copper and steel

1

Relative permeability of iron core

1000

46

Q. Li et al.

Fig. 2 The comparison of the magnetic flux density with and without the molten steel in the mold

Effect of the Copper Resistivity on Calculating Results Mold coppers with different resistivity were selected to study the effect of copper resistivity on the model accuracy. Parameters and material properties are listed in Table 3. Magnetic flux density under the current intensity of 150 A and frequency of 2 Hz, 4 Hz, and 6 Hz with different copper resistivity of 1.7 × 10–8  m, 2.6 × 10–8  m, 4.8 × 10–8  m, and 8.0 × 10–8  m are shown in Fig. 3. At the same frequency, the magnetic flux density increased with the increase of copper resistivity. When the resistivity of the mold copper increased, the current conductivity decreased, leading to the decrease of the eddy current in the mold generated by the magnetic field Table 3 Parameters used in the current study

Parameter

Value

Current intensity of M-EMS (A)

150

Current frequency of M-EMS (Hz)

2, 4, 6

Copper resistivity ( m)

1.7 × 10–8 , 2.6 × 10–8 , 4.8 × 10–8 , 8.0 × 10–8

Molten steel resistivity ( m)

1.4 × 10–6

Relative permeability of copper, steel, coil, air

1

Relative permeability of the 1000 iron core

Parametric Study of Mold Electromagnetic Stirring: Effects of Load …

(a) Copper resistivity of 1.7×10-8 Ω·m

(b) Copper resistivity of 2.6×10-8 Ω·m

(c) Copper resistivity of 4.8×10-8 Ω·m

(d) Copper resistivity of 8.0×10-8 Ω·m

47

Fig. 3 Magnetic flux density under different copper resistivity at 150 A, a 1.7 × 10–8  m, b 2.6 × 10–8  m, c 4.8 × 10–8  m and d 8.0 × 10–8  m

crossing the mold copper. Thus, the magnetic flux density in mold increased due to the reduced loss of the magnetic field from the coils to the mold center. On the other hand, with the increase of the mold copper resistivity, the shielding effect of the copper mold enhanced, resulting in the weakening of the influence of the frequency on the magnetic flux density.

Conclusions (1)

(2)

Load conditions had little effect on calculating results. But the influence would be more obvious with the increase of frequency. It is suggested that build a no-load model to validate the magnetic model. The mold copper resistivity had an obvious effect on calculating results. When the mold copper resistivity increased, the magnetic flux density increased due to the reduced loss of the magnetic field from the coils to the mold center.

48

(3)

Q. Li et al.

Therefore, it is necessary to select an appropriate resistivity value of mold copper plate. The increase of the mold copper resistivity enhanced the shielding effect of the mold copper, leading to a decreased impact of the magnetic flux density by frequency under higher resistivity conditions of the mold copper.

Acknowledgements The authors are grateful for support from the National Nature Science Foundation of China (Grant No. U1860206, No. 51725402), the S&T Program of Hebei (Grant No. 20311006D, 20591001D), the High Steel Center (HSC) at Yanshan University, and Beijing International Center of Advanced and Intelligent Manufacturing of High Quality Steel Materials (ICSM) and the High Quality Steel Consortium (HQSC) at University of Science and Technology Beijing (USTB), China.

References 1. Trindade LB, Vilela ACF, Filho ÁFF, Vilhena MTMB, Soares RB (2002) Numerical model of electromagnetic stirring for continuous casting billets. IEEE Trans Magn 38(6):3658–3660 2. Liu H, Xu M, Qiu S, Zhang H (2012) Numerical simulation of fluid flow in a round bloom mold with in-mold rotary electromagnetic stirring. Metall Mater Trans B 13(6):1657–1675 3. Ren B, Chen D, Wang H, Long M, Han Z (2015) Numerical simulation of fluid flow and solidification in bloom continuous casting mould with electromagnetic stirring. Ironmak Steelmak 42(6):401–408 4. Wang P, Zhang Z, Tie Z, Qi M, Lan P, Li S, Yang Z (2019) Initial transfer behavior and solidification structure evolution in a large continuously cast bloom with a combination of nozzle injection mode and m-ems. Metals 9:1083(15) 5. Ambrish M, Rajneesh K, Kumar JP (2020) Simulation of electromagnetic field and its effect during electromagnetic stirring in continuous casting mold. J Manuf Process 60:596–607 6. Dang A, Zhong B, Tian X, Li Y (2020) Numerical simulation on effect of electromagnetic stirring in 160 mm × 160 mm billet mould on flow field of molten steel. Spec Steel 41(6):6–11 7. Li Y, Wang H, Bai X, Sun Y, Zhang M, Jia J (2020) Numerical simulation of electromagnetic stirring in billet mold. Iron Steel Vanadium Titan 41(04):108–114 8. Wang Y, Chen W, Jiang D, Zhang L (2020) Effect of the gap between copper mold and solidified shell on the fluid flow in the continuous casting strand with mold electromagnetic stirring. Steel Res Int 91(2):1900470(11) 9. Wang Y, Zhang L, Chen W, Ren Y (2021) Three-dimensional macrosegregation model of bloom in curved continuous casting process. Metall Mater Trans 52(4):2796–2805 10. Zhang L, Xu C, Zhang J, Wang T, Li J, Li S (2020) The simulation and optimization of an electromagnetic field in a vertical continuous casting mold for a large bloom. Metals 10(4):516(11)

Phase and Microstructural Analysis of In-Situ Derived Alumina-TiB2 Composites Evangelos Daskalakis, Animesh Jha, Andrew Scott, and Ali Hassanpour

Abstract The paper focusses on the phase and microstructural evolution in ceramic composite based on alumina and titanium diboride, in which the precursor materials are aluminium mixed with TiO2 and B2 O3 . The highly exothermic, self-sustaining, Al + TiO2 + B2 O3 = 53 Al2 O3 + TiB2 , H = and aluminothermic reaction ( 10 3 kJ 2,710 mol ) occurs in the absence of oxygen. Eventually, triggering the self-sustaining reaction on compacts formed from a ball-milled derived mixture, consisted of Al particles coated with TiO2 and B2 O3 nano-particles, resulted in the spontaneous formation of a ceramic phase mixture consisted of alumina particles coated with TiB2 and Ti2 O3 nano-particles. The highly exothermic character of the reaction enabled the sintering of the coated Al2 O3 particles into a dense unique microstructure. Keywords Self-sustaining reaction · Ceramic composites · Al2 O3 –TiB2

Introduction Ceramic matrix composites exhibit advanced mechanical, thermal, and electrical properties compared to the base matrices they derive from. The Al2 O3 /TiB2 ceramic composite is a structurally and thermodynamically compatible structure, which displays high hardness, stiffness, wear resistance, strength, fracture toughness, and sinter-ability [1, 2]. As already known in the literature, alumina is an insulator and has low thermal expansion coefficient, good thermal stability but poor thermal shock resistance, while E. Daskalakis (B) · A. Jha · A. Scott · A. Hassanpour School of Chemical and Process Engineering, University of Leeds, Leeds LS2 9JT, UK A. Jha e-mail: [email protected] A. Scott e-mail: [email protected] A. Hassanpour e-mail: [email protected] © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_6

49

50

E. Daskalakis et al.

Table 1 Thermal, physical and mechanical properties of Al2 O3 and TiB2 [3, 12–14]

Properties Density

Al2 O3

(g/cm3 )

Hardness (GPa)

TiB2

3.95

4.52

5.5–22

25–35

Fracture toughness (MPa/m2 )

3.3–5

6–8

Tensile strength (MPa)

69–665

338–373

Shear strength (GPa)

88–165

182–191

Expansion coefficient

(10–6

7.8

7

Thermal conductivity (W/mK)

/K)

12–38.5

25–90

Electrical conductivity (S/cm)

10–11

105

it is highly resistant to oxidation [3]. By contrast titanium diboride exhibits much higher heat and electrical conductivity [4]. Compared to alumina, TiB2 is a much better conductor of heat and electricity because of the Ti–Ti bonding along the 0001 plane of hexagonal structure. The Ti–Ti layer imparts metallic bonding whereas the B-B layer provides covalent bonding. In the ABAB. stacking of Ti–Ti and B-B layers, the mixed bonding is manifested along the c-axis of the hexagonal structure [5]. Since alumina and TiB2 have closed packed hexagonal structure, their thermal, physical, and mechanical properties are anisotropic [5]. Therefore, the properties compared in Table 1, display minimum and maximum values [3]. Alumina has a higher expansion coefficient value compared to titanium diboride, Table 1, thus alumina is the matrix, while titanium diboride is the dispersion phase of the composite [2, 6, 7]. This composite can find applications in impact-resistance armours, mechanical seals, aerospace, in wear resistance coating in cutting tools, crucibles, and cathode material [8–11]. The composite will be manufactured in situ, from a highly exothermic selfsustaining reaction due to its energy and cost efficiency, high products purity, and advanced dispersion of secondary phase [2, 6–11]. The alumina-thermic reaction of Eq. 1 is thermodynamically favoured at room temperature, since the Gibbs free energy is negative, however reaction kinetics at room temperature are very slow and additional energy is required to trigger the overall reaction [6, 15–26]. Different reactant combinations are present in literature [2, 8, 9, 11, 27–48], however, the oxides of boron and titanium decrease manufacturing cost, as the pure forms of Ti and B are 10 times and 100 times more expensive than TiO2 and B2 O3, respectively. 10 5 kJ Al + TiO2 + B2 O3 = Al2 O3 + TiB2 , H = 2,710 ( ) 3 3 mol

(1)

The adiabatic temperature of the reaction is 2,700 K, calculated by the generalised form of adiabatic temperature, given by Eq. (2) [6, 17, 18, 22–26, 24, 25]. Where a and b are two different solid phases, Tm is the melting point of second phase, Tad is the adiabatic temperature, and Cp is the heat capacity. Thermal analysis data of the reaction is displayed in Table 4 [28].

Phase and Microstructural Analysis of In-Situ Derived …

T t H f,298 =

51

T m Tad C p(a)dT + Ht + C p(b)dT + vHm + C p(liquid)dT

To

Tt

Tm

(2) Milling is a pre-processing technique of ceramic composites, leading to polymorphic transformation of the powder mixture. It aims in the reduction of the particles size, increase of the surface area, and the delivery of a final mixture with specific particle size distribution. Size reduction is a result of fracture and wear, as a function of frequency of stress application and magnitude of stress [51, 52]. Powder pre-processing is feasible through different techniques, which follow mechanical, atomisation, aerosol, physical, chemical, and plasma routes. Mechanical routes of processing metal and ceramic powders include the sole use or combination of impaction, attrition, shear, and compression. Powders subjected to these forces undergo fracture, cold-welding, and polymorphic transformation [52].

Materials and Methods Stoichiometric powder mixture of (0.375) Al–(0.334) TiO2 –(0.291) B2 O3 is ball milled for 1 h, split in two 30 min parts, with a 10 min cooling interval, at 30 Hz. Milling takes place in an iron vessel, with an alumina impact ball, while the powder to ball weight ratio is 0.55. Then compacts with mass of 0.5 g are formed from cold pressing the as-milled mixture at 248 MPa. Thermal analysis of the Al–TiO2 –B2 O3 powder mixture takes place with a Perkins Elmer Simultaneous Thermal Analyser (STA 6000), in argon environment. The maximum temperature is set at 1400 °C, while the heating rate is 20 °C per minute. A tube furnace is used for the compacts’ sintering, in argon atmosphere, with flowrate 2 L/min. Characterisation of sintered samples’ is completed with X-Ray diffraction (XRD), scanning electron microscopy (SEM), and energy dispersive XRay spectroscopy (EDS). For XRD, a Burker D8 Advance with monochromatic CuKα radiation (λ = 0.154 nm) and a 2θ range of 10–90° is used, for phases identification. Using Scherrer’s equation, the average crystallite size of reactants and product phases is feasible, by calculating the area at FWHM, under the 100% peak of each of the phases, utilising equation (nm) = (0.9λ)/(Bcosθ ) where B is the full width half maximum (in radians) of the XRD peak at angle 2θ and λ is the X-ray wavelength. For SEM–EDS, a Hitachi SU8230 scanning electron microscope at 2–20 kV in backscattered and secondary electrons imaging modes was used. Samples were loaded on SEM metal stubs, carbon paint created a conductive path and along with carbon coating prevented sample’s charging.

52

E. Daskalakis et al.

Results and Discussion Milling of the stoichiometric powder mixture resulted in particles’ size reduction. X-Ray diffraction displays reduction in the peaks’ intensity of the milled samples, as a result of particles’ size reduction, Fig. 1. Similar results are reported in literature [27–31, 29]. It is observed, that hard TiO2 and B2 O3 particles are being shattered, eventually coating the ductile Al particles, Fig. 2, as seen in literature [27–31, 29]. Aluminium particles have sizes of approximately 200 μm, while TiO2 and B2 O3 particle sizes range from 10 to 600 nm. From Scherrer’s equation, crystallite sizes of reactants and products are calculated in Table 2. Crystallite sizes of TiO2 and B2 O3 oxides are smaller compared to the ones of ductile aluminium. Eventually, the crystallite sizes of TiB2 and Ti2 O3 , deriving from the smaller sized oxides are smaller than alumina’s. Crystallite sizes of alumina and titanium diboride, deriving from in-situ planetary ball milling, in literature, are shown in Table 3. The time required for the triggering of the reaction ranged between 1.5–60 h, as a result of the difference in energy input during milling, Table 3 [27–30]. In the cases where planetary ball milling continued after the reaction completion, resulted in alumina displaying relatively lower crystallite size, compared to the cases where milling stopped right after reaction completion [28, 29]. Crystallite sizes of alumina and titanium diboride calculated in Table 2, match the ones of literature, for mixtures where milling ceased right after reaction’s completion. Thermal analysis on the milled powder, Fig. 3, displays 4 endotherm peaks which correspond to water removal from the powder mixture at 120 °C, decomposition of (a)

3000

TiO2

Milled (1h-30Hz)

Al

2500 2000 1500

B2O3

Intensity

1000

B2O3

500

Al TiO2

TiO2

TiO2 TiO2

Al TiO2

Al TiO2

TiO2

0

(b)

25000

Un-Milled

Al

20000

TiO2

15000 Al

10000

B2O3

5000

TiO2

TiO2 B2O3

Al TiO2

TiO2

Al TiO2

TiO2

0 10

20

30

40

50

60

70

80

Pos. [°2Th.] Fig. 1 Phases identification of (a) milled and (b) un-milled powder mixtures using X-ray diffraction with monochromatic CuKα radiation (λ = 0.154 nm) and a 2θ range of 10–90°

Phase and Microstructural Analysis of In-Situ Derived …

53

Fig. 2 Al particles coated with shattered TiO2 and B2 O3 particles, observed with SEM and EDS Table 2 Crystal systems, cell volumes, and crystallite sizes of reactants and products Milled 1 h–30 Hz

Sintered 1000 C

Compound

Al

TiO2

B2 O3

Crystal system

Cubic

Tetragonal

Cubic

Cell volume (106 pm3 )

66

136.03

1016

Crystallite size (nm)

55.69

49.19

26.79

Compound

Al2 O3

TiB2

Ti2 O3

Crystal system

Rhombohedral

Hexagonal

Rhombohedral

Cell volume (106 pm3 )

254.7

25.63

312.4

Crystallite size (nm)

53

47

33

Table 3 Crystallite sizes of alumina and titanium diboride Name

Final crystal size (nm)

Milling time (h)

Reaction time (h)

Mohammad Sharifi et al. [27]

50 nm Al2 O3 and TiB2

60

60

Sharifi et al. [28]

30 nm Al2 O3 , 46 nm TiB2 40

31.5

Khaghani-Dehaghani et al. [29]

20 nm Al2 O3 , 32 nm TiB2 20

1.5

Rabiezadeh et al. [30]

Less than 500 nm Al2 O3 and TiB2

30

30

54

E. Daskalakis et al.

Fig. 3 Thermal analysis data of the Al–TiO2 –B2 O3 in argon environment, with heating rate 20 C/min, up to 1400 C

DT

2.0 1.5

DT

1.0 0.5 0.0 -0.5 -1.0 0

200

400

600

800

1000 1200 1400

Temperature (C)

H3 BO3 to B2 O3 at 176 °C, melting of boron at 450 °C, and melting of aluminium at 660 °C, during the heating phase. Melting of aluminium is a result of the breaking of the thin Al2 O3 layer coating the Al particles, due to the difference in their expansion coefficients [32, 54]. Three exotherm peaks are present, at 876 °C, 958 °C, and 1167 °C, which correspond to alumina-thermic reduction of TiO2 and the formation of intermediate product AlTi3 [32], formation of TiB2 [29] and formation of Al18 B4 O33 from AlTi3 and Al2 O3 [32], respectively, Table 4. In literature, the exotherm at 958 °C, for the formation of TiB2 , corresponds to the reaction of (AlTi3 + 6B = 3TiB2 + Al) [32]. During the cooling phase, seems that leftover aluminium solidifies at 660 °C. Table 4 Analysis of the peaks arising from the thermal analysis plots Peak type

Temperature (o C) Reference [29]

Temperature (o C)

Reason [29, 32]

Endotherm

110

120

Water removal

Endotherm

180

176

H3 BO3 to B2 O3

Endotherm

–

460

Reduction of B2 O3

Endotherm

660

660

Al melting

Exotherm

730

876

Aluminothermic reduction of TiO2 (TiAl3 )

958

Ti + 2B = TiB2

Exotherm

960

Exotherm

1134 [32]

Exotherm

–

1167 660

Al18 B4 O33 from AlTi3 and Al2 O3 Al solidification (Cooling cycle)

Phase and Microstructural Analysis of In-Situ Derived …

55 Milled (1h-30Hz)

3000

TiO2

Al

2500 2000 B2O3

1500

Al

Intensity

1000

TiO2

TiO2

B2O3

TiO2 TiO2

500

TiO2

Al

Al TiO2

TiO2

0 4000

Al2O3

Sintered (1000 C)

Al2O3 Al2O3

3000

TiB2

Al2O3

2000

TiB2 Ti2O3

1000

TiB2

Al2O3

Ti2O3 TiB2

Ti2O3

Al2O3

Al2O3 Ti2O3

Al2O3 Al2O3

TiB2

Al2O3 TiB2

Ti2O3

0 10

20

30

40

50

60

70

80

Pos. [°2Th.]

Fig. 4 Phase identification of a un-reacted powder mixture, b reacted sintered material, using X-ray diffraction with monochromatic CuKα radiation (λ = 0.154 nm) and a 2θ range of 10°–90°

In-situ sintering of the powder mixture in a tube furnace, under inert conditions, at 1000 C for 1 h, triggered the alumina-thermic reduction for the formation of the composite. X-Ray diffraction analysis displays 3 phases in the final composite, the ones of Al2 O3 , TiB2, and Ti2 O3 . It is believed that the atmospheric water absorbed by B2 O3 , formed a mixture of H3 BO3 –B2 O3 , which slightly altered the stoichiometric ratio of the milled reactants [9, 35]. Eventually, abundance of TiO2 was present, resulting in the formation of Ti2 O3 phase in the final composite, Fig. 4. Regarding morphological aspects of the microstructure, transfer of oxygen from the oxides on the surface of aluminium particles, to aluminium, during the reaction, resulted in alumina particles formation, while free Ti and B, formed TiB2 nano-particles on the surface of Al2 O3 , Fig. 5, at approximate temperature 958– 1000 °C. Sintering, which was a result of the highly exothermic character of reaction 1, resulted in a microstructure consisted of alumina clusters, surrounded by agglomerated clusters of TiB2 and Ti2 O3 nano-particles.

56

E. Daskalakis et al.

Fig. 5 Al2 O3 particles coated with TiB2

Conclusions In-situ sintering of Al–TiO2 –B2 O3 resulted in a matrix composite displaying Al2 O3 , TiB2 , and Ti2 O3 phases, as the atmospheric water absorption by B2 O3 , altered the stoichiometric composition, allowing excess TiO2 . During sintering, the presence of moisture in B2 O3 also increases boron loss during ignition, enabling larger concentration of Ti2 O3 phase in the final composite. Thermal analysis data show that the reaction completes at around 958–1000 °C. Milled aluminium particles, coated with nano-particles of TiO2 and B2 O3 , when heat treated at 1000 °C, resulted in alumina particles coated with TiB2 and Ti2 O3 , after the completion of exothermic reaction 1. Acknowledgements The financial support for this work from the DSTL office and lab facilities provided by the University of Leeds are gratefully acknowledged.

References 1. Kingery WD (1976) Introduction to ceramics. Wiley, New York 2. Dorri Moghadam A, Omrani E, Lopez H, Zhou L, Sohn Y, Rohatgi PK (2017) Strengthening in hybrid alumina-titanium diboride aluminum matrix composites synthesized by ultrasonic assisted reactive mechanical mixing. Mater Sci Eng A 702:312–321 3. Azom (2001) Alumina–Aluminium Oxide–Al2 O3 —a refractory ceramic oxide. https://www. azom.com/. Accessed on 4 Sept 2021 4. Munro RG (2000) Material properties of titanium diboride. J Res Nat Inst Stand Technol 105(5):709 5. Goldschmid HJ (2013) Interstitial alloys. Springer, Heidelberg

Phase and Microstructural Analysis of In-Situ Derived …

57

6. Munir ZA, Anselmi-Tamburini U, Ohyanagi M (2006) The effect of electric field and pressure on the synthesis and consolidation of materials: a review of the spark plasma sintering method. J Mater Sci 41(3):763–777 7. Palicka R, Rubin J (1991) Development, fabrication and ballistic testing of new, novel low-cost high performance ballistic materials. Cercom Inc., Vista, CA 8. Wang D (2009) Effects of additives on combustion synthesis of Al2 O3 –TiB2 ceramic composite. J Eur Ceram Soc 29(8):1485–1492 9. Rabiezadeh A, Ataie A, Hadian AM (2012) Sintering of Al2 O3 –TiB2 nano-composite derived from milling assisted sol–gel method. Int J Refract Metal Hard Mater 33:58–64 10. Wang W, Fu Z, Wang H, Yuan R (2002) Influence of hot pressing sintering temperature and time on microstructure and mechanical properties of TiB2 ceramics. J Eur Ceram Soc 22(7):1045– 1049 11. Liu G, Yan D, Zhang J (2011) Microstructure and mechanical properties of TiB2 -Al2 O3 composites. J Wuhan Univ Technol-Mater Sci Ed 26(4): 696–699 12. Goodfellow (2001) Titanium diboride (TiB2 )—properties and applications. https://www.azom. com/. Accessed on 6 Sept 2021 13. Kok M (2005) Production and mechanical properties of Al2 O3 particle-reinforced 2024 aluminium alloy composites. J Mater Process Technol 161(3):381–387 14. Königshofer R, Fürnsinn S, Steinkellner P, Lengauer W, Haas R, Rabitsch K, Scheerer M (2005) Solid-state properties of hot-pressed TiB2 ceramics. Int J Refract Metal Hard Mater 23(4):350–357 15. Munir ZA, Holt JB (1990) Combustion and plasma synthesis of high-temperature materials. VCH Publishers Inc., New York 16. Subrahmanyam J, Vijayakumar M (1992) Self-propagating high-temperature synthesis. J Mater Sci 27(23):6249–6273 17. Borovinskaya IP (1992) Chemical classes of the SHS processes and materials. Pure Appl Chem 64:919. https://www.degruyter.com/. Accessed Sept 2021 18. Mossino P (2004) Some aspects in self-propagating high-temperature synthesis. Ceram Int 30(3):311–332 19. Moore JJ, Feng HJ (1995) Combustion synthesis of advanced materials: Part I. Reaction parameters. Prog Mater Sci 39(4):243–273 20. Wang LL, Munir ZA, Maximov YM (1993) Thermite reactions: their utilization in the synthesis and processing of materials. J Mater Sci 28(14):3693–3708 21. Munir Z, Anselmi-Tamburini U, Ohyanagi M (2006) The effect of electric field and pressure on the synthesis and consolidation of materials: a review of the spark plasma sintering method. J Mater Sci 41(3):763–777 22. Feng A, Munir ZJ (1995) The effect of an electric field on self-sustaining combustion synthesis: Part I. Modeling studies. Metall Mater Trans B 26(3):581 23. Yi HC, Moore JJ (1990) Self-propagating high-temperature (combustion) synthesis (SHS) of powder-compacted materials. J Mater Sci 25(2):1159–1168 24. Feng A, Graeve OA, Munir ZA (1998) Modeling solution for electric field-activated combustion synthesis. Comput Mater Sci 12(2):137–155 25. Merzhanov AG (1976) Problems of Combustion in Chemical Technology and in Metallurgy. Russ Chem Rev 45(5):409–420 26. Levashov EA, Mukasyan AS, Rogachev AS, Shtansky DV (2017) Self-propagating hightemperature synthesis of advanced materials and coatings. Int Mater Rev 62(4):203–239 27. Mohammad Sharifi E, Karimzadeh F, Enayati MH (2010) Synthesis of titanium diboride reinforced alumina matrix nanocomposite by mechanochemical reaction of Al–TiO2 –B2 O3 . J Alloy Compd 502(2):508–512 28. Sharifi EM, Karimzadeh F, Enayati MH (2011) Preparation of Al2 O3 –TiB2 nanocomposite powder by mechanochemical reaction between Al, B2 O3 and Ti. Adv Powder Technol 22(4):526–531 29. Khaghani-Dehaghani MA, Ebrahimi-Kahrizsangi R, Setoudeh N, Nasiri-Tabrizi B (2011) Mechanochemical synthesis of Al2 O3 –TiB2 nanocomposite powder from Al–TiO2 –H3 BO3 mixture. Int J Refract Metal Hard Mater 29(2):244–249

58

E. Daskalakis et al.

30. Rabiezadeh A, Hadian A, Ataie A (2012) Preparation of alumina/titanium diboride nanocomposite powder by milling assisted sol–gel method. Int J Refract Metals Hard Mater 31:121– 124 31. Yang W, Dong SJ, Yangli AZ, Xie ZX (2013) Synthesis of Al2 O3 -TiB2 composite powder by planetary milling from Al-TiO2 -B2 O3 -Ni mixture. In: Applied mechanics and materials. Trans Tech Publ, Stafa, pp 734–737 32. Sundaram V, Logan K, Speyer R (1997) Aluminothermic reaction path in the synthesis of a TiB2 –Al2 O3 composite. J Mater Res 12(7):1681–1684 33. Plovnick RH, Richards EA (2001) New combustion synthesis route to TiB2 -Al2 O3 . Mater Res Bull 36(7–8):1487–1493 34. Li J, Cai Z, Guo H, Xu B, Li L (2009) Characteristics of porous Al2O3–TiB2 ceramics fabricated by the combustion synthesis. J Alloy Compd 479(1):803–806 35. Kecskes LJ, Niiler A, Kottke T, Logan KV, Villalobos GR (1996) Dynamic consolidation of combustion-synthesized alumina-titanium diboride composite ceramics. J Am Ceram Soc 79(10):2687–2695 36. Meyers MA, Olevsky EA, Ma J, Jamet M (2001) Combustion synthesis/densification of an Al2 O3 –TiB2 composite. Mater Sci Eng A 311(1):83–99 37. Popov AYu, Sivak AA, Borodianska HYu, Shabalin IL (2015) High toughness TiB2–Al2 O3 composite ceramics produced by reactive hot pressing with fusible components. Adv Appl Ceram 114(3):178–182 38. Kurto˘glu A (2004) Aluminum oxide and titanium diboride reinforced metal matrix composite and its mechanical properties. Thesis, Citeseer 39. Xu J, Zou B, Tao S, Zhang M, Cao X (2016) Fabrication and properties of Al2 O3 –TiB2 –TiC/Al metal matrix composite coatings by atmospheric plasma spraying of SHS powders. J Alloy Compd 672:251–259 40. Chatterjee S, Shariff SM, Datta Majumdar J, Roy Choudhury A (2008) Development of nanostructured Al2 O3 -TiB2 -TiN coatings by combined SHS and laser surface alloying. Int J Adv Manuf Technol 38(9):938–943 41. Kimura I, Hotta N, Hiraoka Y, Saito N, Yokota Y (1989) Sintering and characterization of Al2 O3 -TiB2 composites. J Eur Ceram Soc 5(1):23–27 42. Tang W, Fu Z, Zhang J, Wang W, Wang H, Wang Y, Zhang Q (2006) Fabrication and characteristics of TiB2 /Al2 O3 core/shell particles by hybridization. Powder Technol 167(3):117–123 43. Cheng H, Li Z, Shi YJ (2008) Microstructure and wear resistance of Al2 O3 –TiB2 composite coating deposited by axial plasma spraying. Surf Eng 24(6):452–457 44. Eskandari H, Taheri R (2015) A novel technique for development of aluminum alloy matrix/TiB2 /Al2 O3 hybrid surface nanocomposite by friction stir processing. Procedia Mater Sci 11:503–508 45. Sajedi Alvar F, Heydari M, Kazemzadeh A, Vaezi M, Nikzad L (2018) Al2 O3 -TiB2 nanocomposite coating deposition on titanium by air plasma spraying. Mater Today: Proc 5(7, Part 3):15739–15743 46. Mousavian RT, Sharafi S, Shariat M (2011) Preparation of nano-structural Al2 O3 -TiB2 in-situ composite using mechanically activated combustion synthesis followed by intensive milling. Iranian J Mater Sci Eng 8(2):1–9 47. Reed JS (1995) Principles of ceramics processing. Wiley-Interscience Publication, Hoboken 48. Meyers M, Olevsky E, Ma J, Jamet M (2001) Combustion synthesis/densification of an Al2 O3 – TiB2 composite. Mater Sci Eng A 311(1–2):83–99 49. Danˇek V (2006) Chapter 3—Phase equilibria. In: Danˇek V (ed) Physico-chemical analysis of molten electrolytes. Elsevier Science, Amsterdam, pp 107–219 50. Mills AF (1999) Basic heat and mass transfer. Prentice Hall, Upper Saddler River 51. Chauruka SR, Hassanpour A, Brydson R, Roberts KJ, Ghadiri M, Stitt H (2015) Effect of mill type on the size reduction and phase transformation of gamma alumina. Chem Eng Sci 134:774–783 52. Seetharaman S (2005) Fundamentals of metallurgy. Taylor & Francis, London

Phase and Microstructural Analysis of In-Situ Derived …

59

53. Mohammad Sharifi E, Karimzadeh F, Enayati MH (2010) Mechanochemically synthesized Al2 O3 –TiC nanocomposite. J Alloy Compd 491(1):411–415 54. Logan K, Sparrow J, McLemore W (1990) Experimental modeling of particle-particle interactions during SHS of TiB2 -Al2 O3 . In: Combustion and plasma synthesis of high-temperature materials. VCH Publishers Inc, New York, pp 219–228

Phase Equilibria in the Ag–Ge–Bi–Te System and Thermodynamic Properties of the nGeTe•mBi2 Te3 (n, m = 1–4) Layered Compounds Mykola Moroz, Fiseha Tesfaye, Pavlo Demchenko, Myroslava Prokhorenko, Orest Pereviznyk, Bohdan Rudyk, Lyudmyla Soliak, Daniel Lindberg, Oleksandr Reshetnyak, and Leena Hupa Abstract Phase equilibria of the Ag–Ge–Bi–Te system in the part GeTe– Ag8 GeTe6 –Te–Bi2 Te3 (I) were established by the electromotive force (EMF) method with a solid Ag+ conducting electrolyte. It was shown that the concentration space of (I) in the part 4GeTe·Bi2 Te3 –Ag8 GeTe6 –Te–Bi2 Te consists of 8 four-phase regions, formed of the layered compounds of the nGeTe·mBi2 Te3 (n, m = 1–4) homologous range, as well as Ag8 GeTe6 , Bi2 Te3 , and Te. Equations of overall potentialforming reaction of the decomposition and synthesis of compounds of the homologous range were written for each region. Reactions were performed in the following electrochemical cells (ECCs) (−)IE | Ag | SE | PE | IE(+), where IE is the inert electrode (graphite), Ag is the negative (left) electrode, SE is the solid-state Ag+ ion-conducting electrolyte, PE is the positive (right) electrode. PEs of ECCs were prepared by melting of a mixture of the high-purity elements Ag, Ge, Bi, and Te. The component ratios in samples were determined based on the equations of the potential-forming reactions in respective phase Phase equilibria regions. The finely grounded samples were used as PEs of ECCs. The synthesis of an equilibrium set M. Moroz (B) · B. Rudyk · L. Soliak Department of Chemistry and Physics, National University of Water and Environmental Engineering, Rivne 33028, Ukraine e-mail: [email protected] M. Moroz · O. Pereviznyk · O. Reshetnyak Department of Physical and Colloid Chemistry, Ivan Franko National University of Lviv, Lviv 79005, Ukraine F. Tesfaye · L. Hupa Johan Gadolin Process Chemistry Centre, Åbo Akademi University, 20500 Turku, Finland P. Demchenko Department of Inorganic Chemistry, Ivan Franko National University of Lviv, Lviv 79005, Ukraine M. Prokhorenko Department of Cartography and Geospatial Modeling, Lviv Polytechnic National University, Lviv 79013, Ukraine D. Lindberg Department of Chemical and Metallurgical Engineering, Aalto University, 02150 Espoo, Finland © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_7

60

Phase Equilibria in the Ag–Ge–Bi–Te System and Thermodynamic Properties …

61

of phases was performed in the part of PE that is in contact with SE of ECC at T = 580 K for 50 h. Silver cations that shift from the left to the right electrode acted as the nucleation centers of equilibrium compounds in the corresponding phase regions. Linear dependences E versus T of ECCs in the range of T = (440– 500) K were used to calculate values of the Gibbs energies, enthalpies of formation, and entropies of compounds GeTe·4Bi2 Te3 , GeTe·3Bi2 Te3 , GeTe·2.5Bi2 Te3 , GeTe·2Bi2 Te3 , GeTe·Bi2 Te3 , 2GeTe·Bi2 Te3 , 3GeTe·Bi2 Te3 , and 4GeTe·Bi2 Te3 . The differences in the values of the thermodynamic functions of GeTe-rich and Bi2 Te3 rich compounds with respect to GeTe·Bi2 Te3 correlate well with the literature data on the differences of their crystal structures. Keywords Layered compounds · Phase equilibria · Thermodynamic properties · EMF method · Gibbs energy

Introduction The GeTe–Bi2 Te3 system features the formation of a sequence of the compounds of the formula compositions nGeTe·mBi2 Te3 (n = 1–9, m = 1–4) with layered long-periodic crystal structure [1, 2]. All of these compounds belong to class of thermoelectric materials [3–5]. Moreover, the GeBi2 Te4 is three-dimensional topological insulator [6–8]. According to phase diagram of the GeTe–Bi2 Te3 system [2, 9], the 3GeTe·Bi2 Te3 , GeTe·Bi2 Te3 , and GeTe·2Bi2 Te3 compounds are formed by peritectic reactions at 923 K, 857 K, and 837 K, respectively. The 4GeTe·Bi2 Te3 , 2GeTe·Bi2 Te3 , 2GeTe·5Bi2 Te3 , GeTe·3Bi2 Te3 , and GeTe·4Bi2 Te3 compounds were obtained in sub-solidus temperature range by homogenizing melts of a mixture of elements at 570–620 K for the period of 300–500 h [2]. The crystal structure of compounds was established by the results of electron diffraction on thin films, Xray diffraction on poly- and single-crystal samples. Tellurium atoms are grouped by the principle of dense cubic packing in all compounds. By superimposing layers of tellurium atoms, the packages are formed. In these packages, the octahedral cavities are occupied by germanium and bismuth atoms. Hexagonal elementary cells of compounds are formed from multilayer packages of different types, the planes of which are perpendicular to the c axis. The packets are divided by the Van der Waals slits. The number of packages in the unit cell depends on the relative amount of germanium and bismuth tellurides in the compound. In the GeTe-rich compounds (n/m > 1), the unit cell is formed from packages with the same number of layers. The packages differ in the placement of Ge and Bi in the cationic layers. In the Bi2 Te3 -rich compounds (m/n > 1), the unit cell consists 5- and 7-layered packets. Compounds of the nGeTe·mBi2 Te3 (n, m = 1–4) homologous range are phases of variable composition [10]. Deviation from the stoichiometric composition of one or more elements significantly affects the numerical values of the thermo-EMF coefficients α, the electrical conductivity σ , and the thermal conductivity ktot = kph + kel , where

62

M. Moroz et al.

kph and kel are phonon and electronic components of ktot , respectively, which determine the thermoelectric efficiency of material Z = α2 σ/ktot [11–13]. The thermoelectric efficiency of compounds of the homologous series depends on the peculiarities of filling octahedral cavities. Layers with mixed cationic filling in the GeBi2 Te4 [1] and GeBi4 Te7 [14, 15] compounds were established by XRD measurements. The presence of mixed cationic positions causes fluctuations of masses and stresses, which contributes to the strong scattering of phonons. The consequence of the formation of such layers is a decreasing of kph values. The thermoelectric properties of ternary compounds can be optimized by doping with electroactive impurities. According to Konstantinov et al. [11], a significant increase of the σ and decrease of kph values of the GeBi4 Te7 compound can be explained by the intercalation of copper into the space between multilayer packets. Another way to reduce the lattice thermal conductivity in the compounds is the formation of solid solutions with cationic and anionic substitution [12, 16]. Herein, we present thermodynamic properties of the layered equilibrium compounds nGeTe·mBi2 Te3 (n, m = 1–4) in the 4GeTe·Bi2 Te3 –Ag8 GeTe6 –Te– Bi2 Te3 part of the Ag–Ge–Bi–Te system below T = 500 K, determined by the EMF method described in [17, 18]. Information on the thermodynamic properties of the compounds can be used for modeling by CALPHAD methods [19] of the chemical composition of a multinary solid solution based on compounds of homologous series with optimal values of ZT parameter.

Experimental The starting materials for synthesis were high-purity elements: Ag, 99.99 wt% (AlfaAesar, Germany); Ge, 99.999 wt% (Alfa-Aesar, Germany); Bi, 99.99 wt% (AlfaAesar, Germany); S, 99.999 wt% (Alfa-Aesar, Germany); and Te, 99.999 wt% (AlfaAesar, Germany). For the EMF measurements [18, 20–24], the following electrochemical cells (ECCs) were assembled: (−)IE | Ag | SE | PE | IE(+),

(A)

where IE is the inert electrode (graphite), SE is the solid-state electrolyte, PE is the positive (right) electrode. A pure Ag in powder form was used as a negative (left) electrode. As SE we used Ag2 GeS3 -glass which is the purely Ag+ ionic conductor material [25–27]. The Ag2 GeS3 -glass [25, 28] was obtained by melt quenching of the corresponding elements from T = 1200 K in ice water. The PEs of the EECs (A) were synthesized by melting of a mixture of the high-purity elements Ag, Ge, Bi, and Te in evacuated quartz glass ampoules at T = 1070 K for 5 h. Slowly cooled to room temperature samples were ground into a fine powder with the particle size of ≤5 μm and were used as PEs of ECCs. The composition of the pure elements of PEs

Phase Equilibria in the Ag–Ge–Bi–Te System and Thermodynamic Properties …

63

were calculated based on equations of electrochemical reactions for each of 8 fourphase regions of the 4GeTe·Bi2 Te3 –Ag8 GeTe6 –Te–Bi2 Te3 system. The synthesis of an equilibrium set of phase was performed in the part of the PE that is in contact with SE of ECC at 580 K for 50 h. Ag+ that shift from the left to the right electrode acted as the nucleation centers of equilibrium compounds in the corresponding phase regions [29]. Components of the ECCs in powder form were pressed at 108 Pa through a 2 mm diameter hole arranged in the fluoroplast matrix up to the density ρ = (0.93 ± 0.02)·ρ 0 , where ρ 0 is the experimentally determined density of cast samples. Three-fold thermal cycling of ECCs in the temperature between 400 and 500 K was performed to eliminate possible defects due to plastic deformation during sample pressing [30, 31]. The heating and cooling rates were 2 K min–1 . Experiments were performed in a horizontal resistance furnace, similar to that described in [32, 33]. As protective atmosphere, we used a continuously flowing highly purified (99.99 vol%) Ar(g) at 0.12 MPa, with a flow rate of 2·10–3 m3 h–1 from the left to right electrode of the ECCs. The temperature was maintained with an accuracy of ±0.5 K. The EMF of the cells were measured by the compensation method with high-resistance universal U7-9 digital voltmeter with input impedance of >1012 . The equilibrium in ECCs at each temperature was achieved within ≤3 h. After equilibrium has been attained, the EMF values were constant or their variation did not exceed ±0.2 mV. The criterion for achieving the equilibrium state is the reproducibility of the E versus T dependences in the heating–cooling cycles. In our previous works [34, 35], we have described in details the scheme of ECCs and procedure of the EMF measurements.

Results and Discussion The division of concentration space of the Ag–Ge–Bi–Te system in the part 4GeTe·Bi2 Te3 –Ag8 GeTe6 –Te–Bi2 Te3 (II) into 8 four-phase regions is shown in Fig. 1 and corresponding phase regions is listed in Table 1. The division of (II) is based on our investigations of the phase regions boundaries by the EMF as well as the data published in [36–38]. The correctness of the division of the equilibrium concentration space in Fig. 1 is confirmed by the following: (a) (b)

the ECCs with positive electrodes of three-phase regions are characterized by different EMF values at constant temperature in the range 445–500 K, the three-phase region that is more distant from the point of silver is characterized by higher EMF value [18], see Table 1.

The established division of (II) relative to the position of Ag can be used for the calculation of thermodynamic properties of the ternary compounds, equilibrium in these phase regions. The GeTe·4Bi2 Te3 –Te–Ag8 GeTe6 –Bi2 Te3 phase region (PR(1)) is the nearest to the point of Ag. For this region the electrochemical process of the synthesis of compounds Ag8 GeTe6 and Bi2 Te3 from Ag, Te, and GeTe·4Bi2 Te3 can

64

M. Moroz et al.

Fig. 1 The phase equilibria of the Ag–Ge–Bi–Te system in the 4GeTe·Bi2 Te3 –Ag8 GeTe6 –Te– Bi2 Te3 part, below T = 500 K. 1 is GeTe·4Bi2 Te3 , 2 is GeTe·3Bi2 Te3 , 3 is GeTe·2.5Bi2 Te3 , 4 is GeTe·2Bi2 Te3 , 5 is GeTe·Bi2 Te3 , 6 is 2GeTe·Bi2 Te3 , 7 is 3GeTe·Bi2 Te3 , and 8 is 4GeTe·Bi2 Te3

Table 1 Ternary phase regions of the Ag–Ge–Bi–Te system in the 4GeTe·Bi2 Te3 –Ag8 GeTe6 –Te– Bi2 Te3 part and the EMF values of ECCs in corresponding phase regions at 460 K Number of the phase region

Phase region

E/mV

PR(1)

GeTe·4Bi2 Te3 –Te–Ag8 GeTe6 –Bi2 Te3

237.83

PR(2)

GTe·3Bi2 Te3 –Te–Ag8 GeTe6 –GeTe·4Bi2 Te3

241.40

PR(3)

GeTe·2.5Bi2 Te3 –Te–Ag8 GeTe6 –GeTe·3Bi2 Te3

245.58

PR(4)

GeTe·2Bi2 Te3 –Te–Ag8 GeTe6 –GeTe·2.5Bi2 Te3

250.18

PR(5)

GeTe·Bi2 Te3 –Te–Ag8 GeTe6 –GeTe·2Bi2 Te3

253.81

PR(6)

2GeTe·Bi2 Te3 –Te–Ag8 GeTe6 –GeTe·Bi2 Te3

263.19

PR(7)

3GeTe·Bi2 Te3 –Te–Ag8 GeTe6 –2GeTe·Bi2 Te3

267.64

PR(8)

4GeTe·Bi2 Te3 –Te–Ag8 GeTe6 –3GeTe·Bi2 Te3

280.12

be expressed by: 8Ag = 8Ag+ + 8e− left electrode (reference system), 8Ag+ + 8e− + 5Te + GeTe · 4Bi2 Te3 = Ag8 GeTe6 + 4Bi2 Te3 − right electrode (sample system), 8Ag + 5Te + GeTe · 4Bi2 Te3 = Ag8 GeTe6 + 4Bi2 Te3 (overall cell reaction) (R1) In the positive electrodes, the overall cell reactions of ECCs (A) in the PR(2)–(8) can be expressed as:

Phase Equilibria in the Ag–Ge–Bi–Te System and Thermodynamic Properties …

65

8Ag + 5Te + 4(GeTe · 3Bi2 Te3 ) = Ag8 GeTe6 + 3(GeTe · 4Bi2 Te3 ),

(R2)

8Ag + 5Te + 6(GeTe · 2.5Bi2 Te3 ) = Ag8 GeTe6 + 5(GeTe · 3Bi2 Te3 ),

(R3)

8Ag + 5Te + 5(GeTe · 2Bi2 Te3 ) = Ag8 GeTe6 + 4(GeTe · 2.5Bi2 Te3 ),

(R4)

8Ag + 5Te + 2(GeTe · Bi2 Te3 ) = Ag8 GeTe6 + GeTe · 2Bi2 Te3 ,

(R5)

8Ag + 5Te + 2GeTe · Bi2 Te3 = Ag8 GeTe6 + GeTe · Bi2 Te4 ,

(R6)

8Ag + 5Te + 3GeTe · Bi2 Te3 = Ag8 GeTe6 + 2GeTe · Bi2 Te3 ,

(R7)

8Ag + 5Te + 4GeTe · Bi2 Te3 = Ag8 GeTe6 + 3GeTe · Bi2 Te3 .

(R8)

In accordance with reactions (R1)–(R8), the composition of the PE of the ECCs were determined by the following Ag: Ge: Bi: Te component ratios 4: 1: 8: 18, 4: 4: 24: 45, 2: 3: 15: 28, 4: 5: 20: 40, 4: 2: 4: 13, 2: 1: 1: 5, 4: 3: 2: 11, and 2: 2: 1: 6, respectively. The measured EMF values (E) of the ECCs at different temperatures (T ) are listed in Table 2 and plotted in Fig. 2. The upper and lower limits of the temperature range of the measurements were determined by the linear part of the E versus T dependences that were reproducible in the heating–cooling cycles. Table 2 The measured values of temperature and EMF of the ECCs from different phase regions of the Ag–Ge–Bi–Te system T/ K

E (R1) /mV E (R2) /mV E (R3) /mV E (R4) /mV E (R5) /mV E (R6) /mV E (R7) /mV E (R8) /mV PR(1)

PR(2)

PR(3)

PR(4)

PR(5)

PR(6)

PR(7)

PR(8)

445.2 234.6

237.8

241.4

245.7

248.9

259.6

263.9

276.6

450.1 235.7

239.1

242.7

247.2

250.6

260.8

265.1

277.7

455.1 236.7

240.2

244.1

248.8

252.2

262.0

266.3

278.9

460.0 237.8

241.4

245.6

250.2

253.8

263.2

267.6

280.1

465.0 238.8

242.6

247.2

251.7

255.4

264.5

268.9

281.3

469.9 239.9

243.9

248.6

253.3

257.1

265.8

270.3

282.6

474.8 240.9

245.2

250.0

254.8

259.0

267.1

271.7

283.8

479.8 242.0

246.4

251.3

256.3

260.5

268.3

272.9

284.7

484.7 243.1

247.6

252.4

258.0

262.1

269.6

274.0

285.8

489.6 244.0

248.8

253.6

259.5

263.6

270.8

275.2

287.1

494.5 245.1

249.9

255.3

261.0

265.2

272.1

276.5

288.3

499.4 246.1

251.0

256.7

262.5

266.8

273.3

277.6

289.3

66

M. Moroz et al.

Fig. 2 Temperature dependences of EMF (E) on temperature (T ) of the ECCs with PEs of the phase regions PR(1)–(8). The phase regions are defined in Table 1

The treatment of the E versus T dependencies for reactions (R1)–(R8) in the range 446–499 K was performed by the least-squares method [39] and can be expressed as [40]:  E = a + bT ± k

  2 u 2E , + u 2b T − T n

(1)



where n is number of experimental pairs E i and Ti , T = nTi . The measured E versus T values for the PR(1)–(8) presented in Table 2 were used to calculate the coefficients (a, b, and k) and dispersions (u E and u b ) of Eq. 1. The obtained results are listed in Table 3. Thermodynamic properties of the layered compounds were calculated based on overall potential-forming reactions (R1)–(R8), the equations of the temperature dependences of EMF E (R1) –E (R8) , and the standard thermodynamic properties of the pure elements and Ag8 GeTe6 , Bi2 Te3 compounds [18, 41]. The calculations were performed according to the methodology proposed by Osadchii et al. [42]. The Gibbs energies, entropies, and enthalpies of the reactions (R1)–(R8) can be calculated by combining the measured EMF values of each ECCs and the thermodynamic Eqs. (2)–(4): r G = −n · F · E,

(2)

r H = −n · F · [E − (d E/dT )T ],

(3)

r S = n · F · (d E/dT ),

(4)

Phase Equilibria in the Ag–Ge–Bi–Te System and Thermodynamic Properties …

67

Table 3 EMF (mV) versus T (K) relations of the ECCs of type (A) for PR(1)–(8) of the Ag–Ge– Bi–Te system in the range of 446–499 K   2  u 2E 2 Phase region E = a + bT ± k n + ub T − T PR(1)

E (R1) = 

2.40·10−3 140.63 + 211.21 · 10−3 T ± 2.179 + 6.91 · 10−7 (T − 472.34)2 12

PR(2)

E (R2) = 

7.33·10−3 128.45 + 245.68 · 10−3 T ± 2.179 + 2.11 · 10−6 (T − 472.34)2 12

PR(3)

E (R3) = 

3.11·10−2 116.75 + 280.15 · 10−3 T ± 2.179 + 8.94 · 10−6 (T − 472.34)2 12

PR(4)

E (R4) = 

4.89·10−3 107.30 + 310.75 · 10−3 T ± 2.179 + 1.41 · 10−6 (T − 472.34)2 12

PR(5)

E (R5) = 

1.24·10−2 101.49 + 331.21 · 10−3 T ± 2.179 + 3.56 · 10−6 (T − 472.34)2 12

PR(6)

E (R6) = 

2.68·10−3 146.18 + 254.58 · 10−3 T ± 2.179 + 7.72 · 10−7 (T − 472.34)2 12

PR(7)

E (R7) = 

1.54·10−2 149.97 + 255.88 · 10−3 T ± 2.179 + 4.42 · 10−6 (T − 472.34)2 12

PR(8)

E (R8) = 

1.23·10−2 + 3.53 · 10−6 (T − 472.34)2 171.91 + 235.23 · 10−3 T ± 2.179 12

where n = 8 is the number of electrons involved in the reactions (R1)–(R8), F = 96485.33289 C mol –1 is Faraday constant, and E in V is the EMF of the ECCs. The thermodynamic functions of reactions (R1)–(R8) at 298  K were calculated  rS = 0 [20, 22]. The using Eqs. (2)–(4) in the approximation ∂∂ Tr H p = 0 and ∂ ∂T p results of the calculations are presented in Table 4.

68

M. Moroz et al.

Table 4 Standard thermodynamic values of the reactions (R1)–(R8) in the ECCs at 298 Ka

Reaction

−r G ◦

−r H ◦

kJ mol −1

r S ◦ J (mol K) −1

(R1)

157.1 ± 0.3

108.6 ± 0.7

163.0 ± 1.4

(R2)

155.7 ± 0.5

99.2 ± 1.2

189.6 ± 2.5

(R3)

154.6 ± 0.9

90.1 ± 2.4

216.2 ± 5.1

(R4)

154.3 ± 0.4

82.8 ± 1.0

239.9 ± 2.0

(R5)

154.5 ± 0.6

78.4 ± 1.5

255.7 ± 3.2

(R6)

171.4 ± 0.3

112.8 ± 0.7

196.5 ± 1.5

(R7)

174.6 ± 0.7

115.8 ± 1.7

197.5 ± 3.5

(R8)

186.8 ± 0.6

132.7 ± 1.5

181.6 ± 3.2

Uncertainties for r G ◦ , r H ◦ , and r S ◦ are standard uncertainties

a

Standard Gibbs energy and entropy of reaction (R1) are related to the Gibbs energy of formation and entropy of compounds and pure elements the following equations: r(R1) G ◦ = f G ◦Ag8 GeTe6 + 4f G ◦Bi2 Te3 − f G ◦GeTe·4Bi2 Te3 ,

(5)

◦ ◦ ◦ r(R1) H ◦ = f HAg + 4f HBi − f HGeTe·4Bi , 2 Te3 2 Te3 8 GeTe6

(6)

◦ ◦ ◦ ◦ ◦ r(R1) S ◦ = SAg + 4SBi − 8SAg − SGeTe·4Bi − 5STe . 2 Te3 2 Te3 8 GeTe6

(7)

It follows from Eqs. (5)–(7) that: f G ◦GeTe·4Bi2 Te3 = f G ◦Ag8 GeTe6 + 4f G ◦Bi2 Te3 − r(R1) G ◦ ,

(8)

◦ ◦ ◦ f HGeTe·4Bi = f HAg + 4f HBi − r(R1) H ◦ , 2 Te3 2 Te3 8 GeTe6

(9)

◦ ◦ ◦ ◦ ◦ SGeTe·4Bi = SAg + 4SBi − 8SAg − 5STe − r(R1) S ◦ . 2 Te3 2 Te3 8 GeTe6

(10)

For GeTe·3Bi2 Te3 , GeTe·2.5Bi2 Te3 , GeTe·2Bi2 Te3 , GeTe·Bi2 Te3 , 2GeTe·Bi2 Te3 , 3GeTe·Bi2 Te3 , and 4GeTe·Bi2 Te3 compounds the corresponding reactions to determine f G ◦ , f H ◦ , and S ◦ can be written similar to Eqs. (8)–(10) with their respective moles. By combining Eqs. (8)–(10) using thermodynamic data of the pure elements [41], compounds Ag8 GeTe6 [18], Bi2 Te3 [41], and the thermodynamic data listed in Table 4, the standard Gibbs energies, enthalpies of formations, and entropies of the layered tetradymite-like compounds of the homologous series nGeTe·mBi2 Te3 were calculated. A comparative summary of the calculated values is shown in Fig. 3 and listed in Table 5.

Phase Equilibria in the Ag–Ge–Bi–Te System and Thermodynamic Properties …

69

Fig. 3 Concentration changes of thermodynamic functions of the layered tetradymite-like compounds of the GeTe–Bi2 Te3 system

Table 5 Summary of the standard thermodynamic properties of layered tetradymite-like compounds of the GeTe–Bi2 Te3 system at 298 K determined in this work Phase

−f G ◦ kJ

−f H ◦

−T f S ◦

mol –1

S◦ J mol –1 K –1

GeTe·4Bi2 Te3

419.2 ± 2.8

446.3 ± 3.0

27.1 ± 0.2

GeTe·3Bi2 Te3

342.5 ± 5.6

371.2 ± 5.9

28.7 ± 0.3

1037.6 ± 29.4 770.0 ± 22.1

GeTe·2.5Bi2 Te3

304.3 ± 10.1

335.2 ± 10.5

30.9 ± 0.4

631.8 ± 18.5

GeTe·2Bi2 Te3

266.3 ± 11.4

300.7 ± 12.0

34.4 ± 0.6

488.9 ± 14.6

GeTe·Bi2 Te3

189.9 ± 11.7

233.8 ± 12.2

43.9 ± 0.7

195.1 ± 7.1

2GeTe·Bi2 Te3

286.5 ± 18.2

366.2 ± 21.6

79.7 ± 1.5

155.6 ± 4.8

3GeTe·Bi2 Te3

379.8 ± 32.5

495.6 ± 34.5

115.8 ± 3.1

115.0 ± 6.9

4GeTe·Bi2 Te3

461.0 ± 39.2

608.2 ± 48.4

147.2 ± 5.0

90.5 ± 8.1

The temperature dependences of the Gibbs energies of formation of the layered compounds are described as:   f G ◦GeTe·4Bi2 Te3 / kJ · mol−1 = −(446.3 ± 3.0) + (90.8 ± 0.8) · 10−3 T /K, (11)

70

M. Moroz et al.

  f G ◦GeTe·3Bi2 Te3 / kJ · mol−1 = −(371.2 ± 5.9) + (96.4 ± 0.9) · 10−3 T /K, (12)   f G ◦GeTe·2.5Bi2 Te3 / kJ · mol−1 = −(335.2 ± 10.5) + (103.7 ± 1.1) · 10−3 T /K, (13)   f G ◦GeTe·2Bi2 Te3 / kJ · mol−1 = −(300.7 ± 12.0) + (115.6 ± 1.4) · 10−3 T /K, (14)   f G ◦GeTe·Bi2 Te3 / kJ · mol−1 = −(233.8 ± 12.2) + (147.4 ± 2.3) · 10−3 T /K, (15)   f G ◦2GeTe·Bi2 Te3 / kJ · mol−1 = −(366.2 ± 21.6) + (267.6 ± 6.5) · 10−3 T /K, (16)   f G ◦3GeTe·Bi2 Te3 / kJ · mol−1 = −(495.6 ± 34.5) + (388.7 ± 9.3) · 10−3 T /K, (17)   f G ◦4GeTe·Bi2 Te3 / kJ · mol−1 = −(608.2 ± 48.4) + (493.8 ± 15.6) · 10−3 T /K. (18) As can been seen in Fig. 3, the thermodynamic functions of the layered compounds are connected by smooth lines with an inflection at the point of 50 mol% GeTe, that corresponds to the composition of the GeTe·Bi2 Te3 . The depicted dependences correlate well with the results of studies of the crystal structure of the compounds. Compounds of the homologous series are divided into two groups according to the method of forming from slabs the crystal lattice period along the c axis [2]. For GeTe-rich compounds, the parameter c determines the number of slabs with the same number of layers. The layers differ in the way they fill octahedral cavities with Ge and Bi cations. For Bi2 Te3 -rich compounds, the parameter c is determined by a combination of 5- and 7-layer packets, which also differ in the way cations fill octahedral voids. According to Fig. 3, established in [2] compound of the formulaic composition 2GeTe·5Bi2 Te3 is a high-temperature modification of the GeTe·2.5Bi2 Te3 compound with a double lattice parameter c.

Conclusions The synthesis of the equilibrium set of phases for electrochemical cells of each of the eight-phase regions 4GeTe·Bi2 Te3 –Ag8 GeTe6 –Te–Bi2 Te3 was carried out by melting of pure elements at 1070 K with subsequent annealing of the fine mixture at 550 K for 250 h. The ratios of the components were established according to equations of the overall potential-forming reactions. The equilibrium set of phases determine the value of the EMF of the cell at constant temperature. Based on the EMF versus T

Phase Equilibria in the Ag–Ge–Bi–Te System and Thermodynamic Properties …

71

dependences of ECCs, the standard values of Gibbs energies, enthalpies of formation, and entropies of compounds of homologous series nGeTe·mBi2 Te3 (n, m = 1–4) were calculated for the first time. The polymorphism of the compound GeTe·2.5Bi2 Te3 was established. The 2GeTe·5Bi2 Te3 compound of the homologous series is actually a high-temperature modification of the GeTe·2.5Bi2 Te3 phase with a double lattice parameter c. The experimental thermodynamic data of the layered compounds determined in this work can be used in thermodynamic modeling of the Ag–Ge–Bi–Te system and contribute to the search for new materials with high ZT values in different temperature ranges. Acknowledgements This research was supported by the national projects of the Ministry of Education and Science of Ukraine: “Synthesis, physico-chemical and thermodynamic properties of nano sized and nanostructured materials for electrochemical systems” (No. 0120U102184) and “Scientific and experimental bases of manufacturing composite oxide, chalcogenide materials with prolonged service life”. This work was partly supported by the Academy of Finland project (Decision number 311537), as part of the activities of the Johan Gadolin Process Chemistry Centre at Åbo Akademi University. Conflict of Interest The authors declare that they have no conflict of interest.

References 1. Karpinsky OG, Shelimova LE, Kretova MA, Fleurial J-P (1998) X-ray study of the nGeTe·mBi2 Te3 mixed layered tetradymite-like compounds. J Alloys Compd 265:170–175. https://doi.org/10.1016/S0925-8388(97)00294-6 2. Shelimova LE, Karpinskii OG, Zemskov VS, Konstantinov PP (2000) Structural and electrical properties of layered tetradymite-like compounds in the GeTe–Bi2 Te3 and GeTe–Sb2 Te3 systems. Inorg Mater 36:235–242. https://doi.org/10.1007/BF02757928 3. Perumal S, Roychowdhury S, Biswas K (2016) High performance thermoelectric materials and devices based on GeTe. J Mater Chem C 4:7520–7536. https://doi.org/10.1039/C6TC02501C 4. Bailey TP, Uher C (2017) Potential for superionic conductors in thermoelectric applications. Curr Opin Green Sustain Chem 4:58–63. https://doi.org/10.1016/j.cogsc.2017.02.007 5. Shi Y, Sturm C, Kleinke H (2019) Chalcogenides as thermoelectric materials. J Solid State Chem 270:273–279. https://doi.org/10.1016/j.jssc.2018.10.049 6. Kane C, Moore J (2011) Topological insulators. Phys World 24:32–36. https://doi.org/10.1088/ 2058-7058/24/02/36 7. Okamoto K, Kuroda K, Miyahara H, Miyamoto K, Okuda T, Aliev ZS, Babanly MB, Amiraslanov IR, Shimada K, Namatame H, Taniguchi M, Samorokov DA, Menshchikova TV, Chulkov EV, Kimura A (2012) Observation of a highly spin-polarized topological surface state in GeBi2 Te4 . Phys Rev B 86:195304–195308. https://doi.org/10.1103/PhysRevB.86.195304 8. Babanly MB, Chulkov EV, Aliev ZS, Shevelkov AV, Amiraslanov IR (2017) Phase diagrams in materials science of topological insulators based on metal chalcogenides. Russ J Inorg Chem 62:1703–1729. https://doi.org/10.1134/S0036023617130034 9. Abrikosov NKh, Danilova-Dobryakova GT (1965) Bi2 Te3 -GeTe phase diagram. Izv Akad Nauk SSSR Neorg Mater 1:57–61 [in Russian] 10. Shelimova LE, Karpinsky OG, Kretova MA, Avilov ES (1996) Phase equilibria in the Ge-Bi-Te ternary system at 570–770 K temperature range. J Alloys Compd 243:194–201. https://doi.org/ 10.1016/S0925-8388(96)02394-8

72

M. Moroz et al.

11. Konstantinov PP, Shelimova LE, Avilov ES, Kretova MA, Fleurial J-P (1999) Transport phenomena in mixed layered tetradymite-like compounds in the GeTe–Bi2 Te3 system. J Solid State Chem 146:305–312. https://doi.org/10.1006/jssc.1999.8340 12. Shelimova LE, Konstantinov PP, Kretova MA, Avilov ES, Zemskov VS (2004) Thermoelectric properties of cation-substituted solid solutions based on layered tetradymite-like compounds. Inorg Mater 40:461–467. https://doi.org/10.1023/B:INMA.0000027591.50936.18 13. Tesfaye F, Moroz M (2018) An overview of advanced chalcogenide thermo-electric materials and their applications. J Electron Res Appl 2:28–41. https://doi.org/10.26689/jera.v2i2.337 14. Zhukova TB, Zaslavskii AI (1971) The crystal structures of the PbBi4 Te7 , PbBi2 Te4 , SnBi4 Te7 , SnBi2 Te4 , SnSb2 Te4 , GeBi4 Te7 compounds. Crystallografia 16:918–922 [in Russian] 15. Shelimova LE, Karpinskii OG, Konstantinov PP, Avilov ES, Kretova MA, Zemskov VS (2004) Crystal structures and thermoelectric properties of layered compounds in the ATe–Bi2 Te3 (A = Ge, Sn, Pb) systems. Inorg Mater 40:451–460. https://doi.org/10.1023/B:INMA.0000027590. 43038.a8 16. Shelimova LE, Karpinskii OG, Konstantinov PP, Svechnikova TE, Avilov ES, Kretova MA, Zemskov VS (2006) Layered chalcogenides in the quasi-binary systems AIV BVI AV 2 BVI 3 (AIV -Ge, Sn, Pb; BVI -Te, Se; AV -Bi, Sb)—promising thermoelectric materials for thermogenerators. Perspektivnye materialy 3:5–17 [in Russian] 17. Ipser H, Mikula A, Katayama I (2010) Overview: the CMF method as a source of experimental thermodynamic data. Calphad 34:271–278. https://doi.org/10.1016/j.calphad.2010.05.001 18. Babanly M, Yusibov Y, Babanly N (2011) The EMF method with solid-state electrolyte in the thermodynamic investigation of ternary copper and silver chalcogenides. In: Kara S (ed) InTech, pp 57–78. https://doi.org/10.5772/28934 19. Kroupa A (2013) Modelling of phase diagrams and thermodynamic properties using Calphad method—development of thermodynamic databases. Comput Mater Sci 66:3–13. https://doi. org/10.1016/j.commatsci.2012.02.003 20. Voronin MV, Osadchii EG (2013) Thermodynamic properties of silver and bismuth sulfosalt minerals, pavonite (AgBi3 S5 ) and matildite (AgBiS2 ) and implications for ore deposits. Econ Geol 108:1203–1210. https://doi.org/10.2113/econgeo.108.5.1203 21. Aliev ZS, Musayeva SS, Imamaliyeva SZ, Babanly MB (2017) Thermodynamic study of antimony chalcoiodides by EMF method with an ionic liquid. J Therm Anal Calorim 133:1115–1120. https://doi.org/10.1007/s10973-017-6812-4 22. Voronin MV, Osadchii EG, Brichkina EA (2017) Thermochemical properties of silver tellurides including empressite (AgTe) and phase diagrams for Ag–Te and Ag–Te–O. Phys Chem Miner 44:639–653. https://doi.org/10.1007/s00269-017-0889-y 23. Babanly NB, Orujlu EN, Imamaliyeva SZ, Yusibov YA, Babanly MB (2019) Thermodynamic investigation of silver-thallium tellurides by EMF method with solid electrolyte Ag4 RbI5 . J Chem Thermodyn 128:78–86. https://doi.org/10.1016/j.jct.2018.08.012 24. Imamaliyeva SZ, Musayeva SS, Babanly DM, Jafarov YI, Taghiyev DB, Babanly MB (2019) Determination of the thermodynamic functions of bismuth chalcoiodides by EMF method with morpholinium formate as electrolyte. Thermochim Acta 679:178319(1)–(7). https://doi.org/10. 1016/j.tca.2019.178319 25. Robinel E, Carette B, Ribes M (1983) Silver sulfide based glasses (I): glass forming regions, structure and ionic conduction of glasses in GeS2 –Ag2 S and GeS2 –Ag2 S–AgI systems. J Non-Cryst Solids 57:49–58. https://doi.org/10.1016/0022-3093(83)90407-6 26. Moroz MV, Demchenko PY, Prokhorenko SV, Moroz VM (2013) Physical properties of glasses in the Ag2 GeS3 -AgBr system. Phys Solid State 55:1613–1618. https://doi.org/10.1134/S10637 83413080209 27. Moroz MV, Demchenko PYu, Mykolaychuk OG, Akselrud LG, Gladyshevskii RE (2013) Synthesis and electrical conductivity of crystalline and glassy alloys in the Ag3 GeS3 Br-GeS2 system. Inorg Mater 49:867–871. https://doi.org/10.1134/S0020168513090100 28. Moroz M, Tesfaye F, Demchenko P, Prokhorenko M, Lindberg D, Reshetnyak O, Hupa L (2018) Determination of the thermodynamic properties of the Ag2 CdSn3 S8 and Ag2 CdSnS4 phases in the Ag–Cd–Sn–S system by the solid-state electrochemical cell method. J Chem Thermodyn 118:255–262. https://doi.org/10.1016/j.jct.2017.12.001

Phase Equilibria in the Ag–Ge–Bi–Te System and Thermodynamic Properties …

73

29. Moroz M, Tesfaye F, Demchenko P, Prokhorenko M, Kogut Y, Pereviznyk O, Prokhorenko S, Reshetnyak O (2020) Solid-state electrochemical synthesis and thermodynamic properties of selected compounds in the Ag–Fe–Pb–Se system. Solid State Sci 107:106344(1)–(9). https:// doi.org/10.1016/j.solidstatesciences.2020.106344 30. Moroz MV, Prokhorenko MV, Reshetnyak OV, Demchenko PYu (2017) Electrochemical determination of thermodynamic properties of saturated solid solutions of Hg2 GeSe3 , Hg2 GeSe4 , Ag2 Hg3 GeSe6 , and Ag1.4 Hg1.3 GeSe6 compounds in the Ag–Hg–Ge–Se system. J Solid State Electrochem 21:833–837. https://doi.org/10.1007/s10008-016-3424-z 31. Moroz MV, Demchenko PYu, Prokhorenko MV, Reshetnyak OV (2017) Thermodynamic properties of saturated solid solutions of the phases Ag2 PbGeS4 , Ag0.5 Pb1.75 GeS4 and Ag6.72 Pb0.16 Ge0.84 S5.20 of the Ag-Pb-Ge-S system determined by EMF method. J Phase Equilibria Diffus 38:426–433. https://doi.org/10.1007/s11669-017-0563-6 32. Tesfaye F, Taskinen P (2014) Electrochemical study of the thermodynamic properties of matildite (β-AgBiS2 ) in different temperature and compositional ranges. J Solid State Electrochem 18:1683–1694. https://doi.org/10.1007/s10008-014-2395-1 33. Moroz M, Tesfaye F, Demchenko P, Prokhorenko M, Lindberg D, Reshetnyak O, Hupa L (2018) Phase equilibria and thermodynamics of selected compounds in the Ag–Fe–Sn–S system. J Electron Mater 47:5433–5442. https://doi.org/10.1007/s11664-018-6430-3 34. Moroz MV, Prokhorenko MV, Demchenko PYu (2013) Thermodynamic properties of the intermediate phases of a Ag-Te-AgBr system. Russ J Phys Chem A 87:6–9. https://doi.org/10.1134/ S0036024413010147 35. Prokhorenko MV, Moroz MV, Demchenko PYu (2015) Measuring the thermodynamic properties of saturated solid solutions in the Ag2 Te-Bi-Bi2 Te3 system by the electromotive force method. Russ J Phys Chem A 89:1330–1334. https://doi.org/10.1134/S0036024415080269 36. Shelimova LE, Karpinskii OG, Konstantinov PP, Kretova MA, Avilov ES, Zemskov VS (2001) Composition and properties of layered compounds in the GeTe–Sb2 Te3 system. Inorg Mater 37:342–348. https://doi.org/10.1023/A:1017519625907 37. Blachnik R, Gather B (1978) Mischungen von GeTe, SnTe und PbTe MIT Ag2 Te lin beitrag zur klärung der konstitution der ternären Ag-IVb-Te systeme (IVb = Ge, Sn, Pb). J Common Met 60:25–32. https://doi.org/10.1016/0022-5088(78)90086-3 38. Wu H-J, Chen S-W (2011) Phase equilibria of Ag–Sb–Te thermoelectric materials. Acta Mater 59:6463–6472. https://doi.org/10.1016/j.actamat.2011.07.010 39. Gordon AJ, Ford RA (1972) The chemist’s companion: a handbook of practical data, techniques, and references. Wiley, New York 40. Mashadieva LF, Kevser JO, Aliev II, Yusibov YA, Tagiyev DB, Aliev ZS, Babanly MB (2017) The Ag2 Te-SnTe-Bi2 Te3 system and thermodynamic properties of the (2SnTe)1–x (AgBiTe2 )x solid solutions series. J Alloys Compd 724:641–648. https://doi.org/10.1016/j.jallcom.2017. 06.338 41. Barin I (1995) Thermochemical data of pure substance. VCH, Weinheim 42. Osadchii EG, Rappo OA (2004) Determination of standard thermodynamic properties of sulfides in the Ag-Au-S system by means of a solid-state galvanic cell. Am Mineral 89:1405–1410. https://doi.org/10.2138/am-2004-1007

Potentiostatic Electrodeposition of Ti–Al Alloy with 40% Titanium from the Lewis Acidic 1-Butyl-3-Methylimidazolium Chloride-Aluminum Chloride Ionic Liquid Electrolyte Pravin S. Shinde, Yuxiang Peng, and Ramana G. Reddy Abstract Ti–Al alloys were electrodeposited from the Lewis acidic electrolyte containing 1-butyl-3-methylimidazolium chloride (BMIC) ionic liquid (IL) and aluminum chloride (AlCl3 ). Constant potential electrodeposition was performed in a two-electrode configuration on copper cathode for 4 h at 383 K from BMICAlCl3 electrolyte with a fixed AlCl3 mole fraction and deposition potential. Titanium was served as an anode and also the source of Ti ions. Ti–Al alloys deposited on Cu substrate at different synthesis conditions were analyzed using scanning electron microscopy, energy-dispersive X-ray spectroscopy, and X-ray diffraction techniques. A Ti0.4 Al0.6 phase with 40-atom % Ti was obtained in the final deposit at optimized process parameters and was confirmed by repeating electrodeposition experiments with identical synthesis conditions. After each electrolysis experiment, the Cu cathode weight gain and Ti anode weight loss were measured to determine the Faradaic current efficiency of the Ti–Al electrodeposition process. The current efficiency and energy consumption values were 49.93 ± 0.95 and 23.77 ± 0.89 kWh kg−1 , respectively. Keywords Ti–Al alloy · BMIC-AlCl3 ionic liquid · Potentiostatic Electrolysis

Introduction Titanium (Ti) production by the Kroll process has been the longest known primary method [1], which is an expensive process due to the difficulty of extracting and machining the Ti [2]. Therefore, Ti is usually obtained in the form of alloys. Ti and its alloys (Al, Mo, and Fe) find potential applications in aerospace (aircraft, spacecraft, and missiles) and medical (bone-compatible and surgical tools) industries because of their greater strength-to-weight ratio, excellent corrosion-resistant P. S. Shinde · Y. Peng · R. G. Reddy (B) Department of Metallurgical and Materials Engineering, The University of Alabama, Tuscaloosa, AL 35487, USA e-mail: [email protected] © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_8

74

Potentiostatic Electrodeposition of Ti–Al Alloy with 40% Titanium …

75

properties [3–6]. Electrodeposition has been regarded as one of the inexpensive technologies for the deposition of alloy coatings. Electrochemical synthesis of Ti metal or its alloys from low-temperature ionic liquid (IL) electrolytes is one of the fascinating methods [7]. However, the electrodeposition of pure Ti is challenging compared to other metals like aluminum because of its different oxidation states (II, III, and IV). The synthesis of Ti in the form of Ti–Al alloy from ionic liquids is particularly promising as AlCl3 can easily form eutectic composition at room temperature and increase the electrolyte’s conductivity. The molten chloride salt electrolytes have been employed to electrodeposit of Ti and Ti–Al alloys via an energy-efficient and cost-effective extraction process [8–12]. Several attempts have been made to understand the electrochemistry of Ti ions in the ionic liquid electrolytes to improve the Ti–rich deposition of Ti–Al alloy [13–21]. All the attempts of obtaining pure Ti resulted in the co-deposition of Al and Ti. Nevertheless, the low-temperature electrodeposition of Ti–Al alloys using ionic liquid electrolyte is not only energy-efficient but also eliminates high-temperature melting and consolidation processes. The percentage of Ti in the Ti–Al alloy varies depending on the concentration of Ti species, the Lewis acidity, electrolyte type, and the applied electrochemical potential/current density. About 28 at.% Ti was obtained from NaCl:AlCl3 (1:2 mol ratio) melts at 423 K containing Ti2+ ions, which were electrochemically dissolved by varying the Ti(II) concentration and the applied current. Potentiostatic deposition of Al–Ti on the GC electrode from equimolar AlCl3 -NaCl melt containing anodically dissolved Ti at −0.085 V versus Al/Al(III) for 2 h resulted in different alloys AlTi3 , Al2 Ti, and Al3 Ti alloys, with AlTi3 dominating at a lower temperature (473 K) and Al2 Ti dominating at a higher temperature (573 K). EDS suggested Ti content in the range of 12.6–21.9 at.% [22]. The Al–Ti alloy electrodeposited from AlCl3 NaCl-KCl eutectic molten salt electrolyte in a flowing cell at 473 K improved with the addition of TiCl3, resulting in a smooth deposit of 40.2% Ti content at current density as high as 200 mA cm−2 . Ti co-deposition could only proceed in the presence of Ti2+ species in the electrolyte that were obtained by reducing Ti3+ with the help of Al powder [23]. Al–Ti alloy with 1 wt.% Ti was obtained using TiO2 feeds at high temperatures (1233–1253 K) during aluminum electrolysis for 4 h from fluoride-based melts at a cathodic current density of 900 mA cm−2 and a NaFAlF3 cryolite ratio of 2.2 [24]. About 24.1 at.% Ti was obtained from NaCl:AlCl3 (1:2 mol ratio) melts at 423 K containing Ti2+ ions, which were electrochemically dissolved by varying the Ti(II) concentration and/or the applied current [12, 25]. The constant potential (−1.5 V vs. Pt) and constant current (10 mA cm−2 ) electrolysis from EMIC-AlCl3 (1:2 mol ratio) at 383 K for 1-h duration resulted in Ti deposits of 14 at.% and 16.5 at.%, respectively. Relatively higher Ti content for later was attributed to the relatively higher concentration of Ti species [17]. Stafford et al. reported about 18.4 at.% Ti by electrodeposition from chloroaluminate EMIC-AlCl3 electrolyte containing Ti2+ concentration of 170 mmol L−1 at 353 K [26]. Tsuda et al. studied electrochemistry and the dissolution effect of different Ti ions (Ti2+ , Ti3+ , and Ti4+ ) in EMIC-AlCl3 melts with AlCl3 mole fractions of 0.6 and 0.667 on the electrodeposition of Al–Ti alloys. Ti content in Ti–Al alloy obtained at the AlCl3 mole fraction of 0.667 by dissolving 150 mmol L−1 of Ti2+ species in EMIC-AlCl3

76

P. S. Shinde et al.

at 353 K was 19.1 at.% at a current density of 2.5 mA cm−2 and decreased to 12.4 at.% with further increase of current density to 20 mA cm−2 [20]. Ti–Al deposits obtained from BMIC-AlCl3 (1:2 molar ratio) electrolyte at 373 K appears to be smooth and dendrite-free at relatively low potentials (−1 to −1.3 V vs. Ti) for 1 h, however, results in with lower Ti content (3 at.%), due to lower electrochemically dissolved Ti concentration at lower applied potential [14]. Electrodeposition of dense and adhesive Ti–Al alloy coatings are reported on mild steel from BMIC-AlCl3 ionic liquid electrolyte containing TiCl4 at 343 K. The composition, microstructure of obtained Ti–Al coatings depended on the applied current density and TiCl4 concentration. The Ti concentration in the Ti–Al coatings increased to an optimum level of 11.4 at.% for 0.22 mol L−1 TiCl4 as evidenced from XRD studies [27]. Ti content decreased at higher current densities beyond 5 mA cm−2 , corroborating a similar phenomenon observed by Tsuda et al. [20]. Thus, the primary electrodeposition process proceeds at more negative potentials to deposit microcrystalline or bulk Ti–Al alloy. In our previous study, the chronopotentiometric electrodeposition of Ti–Al alloy was achieved from BMIC-AlCl3 (1:2 molar ratio) electrolyte at 373 K at different current densities (13.5–89.1 mA cm−2 ) for 4 h. The obtained Ti–Al consisted of compact and dendritic deposits with disordered face-centered cubic (FCC) lattice containing 14.56–20.75 at.% Ti [28]. Electrodeposition of Ti–Al alloys was further investigated from the BMIC-AlCl3 (1:2 molar ratio) system by including Ti4+ ions (0.019 molar ratio of TiCl4 ) at different temperatures in the range of 343–498 K and at various potentials in the range of 1.5–3.0 V. Ti–Al alloys with 15–27 at.% Ti contents were produced with current efficiencies in the range of 25–38%. Higher potentials (2.5–3.0 V) and higher temperatures (373–398 K) resulted in non-uniform and coarse-grained Ti–Al deposits. Lower potentials (1.5–2.5 V) produced smooth, bright, and finer particle-sized Ti–Al alloys. The estimated energy consumption of Ti–Al alloys production varied from 16.63 to 31.98 kWh kg−1 of Ti–Al alloy [19, 29]. Motivated from these results, these studies were revisited. Ti content in the Ti–Al alloy was observed to be interdependent regardless of TiCl4 addition because sufficient Ti ions are electrochemically dissolved in the ionic liquid electrolyte at higher electrochemical potential and electrolysis duration. In this work, Ti–rich Ti–Al alloy is deposited on copper substrate by constantpotential electrolysis from the Lewis acidic ionic liquid mixture of 1-butyl-3methylimidazolium chloride and aluminum chloride (BMIC-AlCl3 ) using the electrochemical dissolution of Ti anode. This research work is aimed to obtain the optimum percentage of Ti in the Ti–Al alloy at optimum deposition potential, duration, and the molar ratio of BMIC ionic liquid and AlCl3 . The electrodeposited material is characterized by XRD and SEM-SEM to yield 40 at.% Ti in the final Ti–Al deposit.

Potentiostatic Electrodeposition of Ti–Al Alloy with 40% Titanium …

77

Experimental Preparation of BMIC-AlCl3 Ionic Liquid Electrolyte The chemicals such as anhydrous aluminum chloride (AlCl3 , 95%, Alfa Aesar) and 1butyl-3-methyl imidazolium chloride (BMIC, 98%, Sigma Aldrich) ionic liquid were purchased and used without further treatment. The copper sheet (99%, 0.25 cm thick) was purchased from Sigma Aldrich. A pure titanium sheet (2 mm thick, 99.99%) was obtained from Alfa Aesar® . The ultrahigh pure (UHP) argon gas (99.999%) was obtained from Airgas. All chemical reagents were handled in a dry atmosphere. The eutectic mixture of Lewis acidic BMIC-AlCl3 electrolyte was prepared by mixing a 1:2 molar ratio (AlCl3 mole fraction, XAlCl3 = 0.667) of BMIC IL and AlCl3 in a Pyrex beaker at room temperature under constant stirring for 30 min until a clear homogeneous solution was obtained. XAlCl3 = 0.667 was chosen to maximize the concentration of Al2 Cl7 − anion species crucial for metal electrodeposition from the BMIC-AlCl3 electrolyte. The desired amount of clear electrolyte was then transferred to the 50 mL electrochemical Pyrex cell placed on a hot plate, and the electrolyte was stirred for several minutes using a magnetic stirrer at 120 RPM to achieve a stable temperature of 383 K for electrochemical deposition.

Electrochemical Deposition The electrodeposition experiments were performed at constant potential from BMICAlCl3 electrolyte at 383 K using a KEPCO power supply and a KEITHLEY multimeter controlled by LabVIEW software. The electrochemical cell for the measurements consisted of a 40 mL Pyrex® glass beaker fitted with Teflon/Perspex cover, which has provisions for inserting the electrodes, thermometer, and inert gas inlet/outlets shown schematically in Fig. 1. The Ti ions were incorporated in the BMIC-AlCl3 electrolyte from the Ti anode electrode during electrolysis. The source of Al ions was from AlCl3 . The two-electrode electrolysis setup consisted of a copper sheet (2 × 2 × ~0.25 cm, 99%, Sigma Aldrich) as working electrode (WE) and a Ti plate (2 × 2 × ~2 cm, 99.99%, Alfa Aesar® ) as the counter electrode (CE). The working distance between WE and CE was kept constant at 2 cm. The constant potential electrolysis experiments were performed and repeated at least three times for 4 h duration. The temperature of the hot plate was precisely controlled and monitored by the inserted thermometer. The Ar gas flow was continuously maintained over the electrolyte’s surface throughout the experiment (through the alumina tube). Before the electrolysis experiment, both the Cu and Ti electrodes were polished with 800-grit SiC abrasive paper, rinsed thoroughly with deionized water, cleaned in an ultrasonic bath for 5 min, and dried by air to remove any residual impurities. The height of the electrodes immersed in the BMIC-AlCl3 electrolyte was measured after each experiment for efficiency and energy consumption calculations. Ti–Al

78

P. S. Shinde et al.

Fig. 1 Schematic of a two-electrode experimental setup for electrolysis from BMIC-AlCl3 electrolyte

alloy was deposited on copper plate by constant potential electrolysis from BMICAlCl3 (XAlCl3 = 0.667) electrolyte at 383 K at a fixed applied potential of −3.0 V for 4 h duration. The electrodeposited Ti–Al deposits obtained on Cu electrodes were characterized using structural, morphological, and compositional techniques such as scanning electron microscopy (SEM) on Thermo Scientific™ Apreo scanning electron microscope equipped with energy-dispersive spectroscopy (EDS), and Xray diffraction (XRD) on a Bruker D8 Discover X-ray diffractometer with GADDS by employing monochromatic Co Kα radiation.

Results and Discussion Constant Potential Electrodeposition of Ti–Al from BMIC-AlCl3 Electrolyte The Ti–Al electrodeposits were obtained from BMIC-AlCl3 electrolyte at constant potential conditions for a 4-h duration at electrolyte temperature of 383 K and electrolyte rotation speed of 120 rpm. The constant-potential experiments were performed and repeated at least three times at identical conditions to check the reproducibility of results (morphology, composition, current efficiency, and energy consumption). The preparation of ionic liquid, polishing, and cleaning of electrodes were done identically. The electrode size, electrode separation distance, depth of

Potentiostatic Electrodeposition of Ti–Al Alloy with 40% Titanium …

-3.0V/4h (1st) -3.0V/4h (2nd) -3.0V/4h (3rd)

25

-Current density (mA cm-2)

Fig. 2 The current density-time plots for 2-electrode electrolysis for 4 h using KEPCO power supply at a constant potential of –3.0 V from BMIC-AlCl3 electrolyte at 383 K

79

20

15

10

5

0

0

1

2

3

4

Time (h)

electrode immersion, and electrolyte volume (30 mL) were all the same for all the experiments. To rule out differences in electrolyte composition, the sufficient stock solution of electrolyte with AlCl3 mole fraction of 0.667 was prepared in a bulk quantity and stored correctly to avoid contamination. A 30 mL of electrolyte was taken from this stock solution for all sets of experiments. All the samples were then analyzed by XRD, SEM, and EDS to determine the phase, morphology, uniformity, and elemental composition of the synthesized Ti–Al deposits. The Ti ions were incorporated in-situ into the BMIC-AlCl3 ionic liquid by electrochemical dissolution of the Ti anode during the 4-h electrolysis experiment. The Cu cathode and Ti anode electrodes were weighed before and after each electrolysis experiment to account for Ti ions stripped from the Ti anode and the amount of Ti–Al material deposited onto the Cu cathode electrode. Figure 2 shows current density-time plots recorded from 2-electrode electrolysis from BMIC-AlCl3 (AlCl3 mole fraction of 0.667) electrolyte by applying a constant potential of −3.0 V for 4 h at 383 K and with electrolyte flow of 120 rpm. The electrochemical deposition of Al from such chloroaluminate ionic liquid electrolytes has been reported to be mainly due to contribution from the diffusion of Al2 Cl7 − species [30–32]. The initial current density is driven by the Al2 Cl7 − ions in the ionic liquid electrolyte leading to the reduction of aluminum atoms. Once the Ti ions are anodically dissolved into the electrolyte, they form a complex with Al2 Cl7 − ions to obtain Ti[(Al2 Cl7 )4 ]2− complex. The possible reaction mechanisms of co-deposition of Al and Ti have been reported previously [14]. In short, the co-deposition of 3electron reduction of Al and 2-electron reduction of Ti proceeds according to the reactions (1, 2): 4[Al2 Cl7 ]− + 3e− ↔ Al (At Cathode) + 7[AlCl4 ]−

(1)

 2− Ti(Al2 Cl7 )4 + 2e− ↔ Ti (At Cathode) + 4[Al2 Cl7 ]−

(2)

80

P. S. Shinde et al.

Table 1 The weight gain of Ti–Al deposit on Cu cathode, the weight loss of Ti ions from Ti anode, Faradaic efficiency, and the energy consumption values for 4-h constant potential electrolysis at −3.0V from BMIC-AlCl3 electrolyte at 383 K at 120 RPM Potential

−3 V (1st)

−3 V (2nd)

−3 V (3rd)

Total charge (C)

2348

1915

2003

Total atomic weight (g mol−1 )

35.336

35.336

35.336

Total electron transferred

2.600

2.600

2.600

Theoretical weight gain (g)

0.331

0.270

0.282

Actual weight gain (g)

0.162

0.139

0.139

Initial current density, J t=0 h (mA cm−2 )

−23.04

−21.83

−20.48

Current density after 4 h, J t=4 h (mA cm−2 )

−14.92

−15.06

−13.87

Efficiency (%)

48.98

51.53

49.27

Energy consumption (kWh kg−1 )

24.66

22.28

24.37

Ti anode weight loss

0.306

0.257

0.264

Initially, the current density (in Fig. 2) reaches a maximum (in the range of 22– 24 mA cm−2 ). It then decreases to ~15 mA cm−2 over time for all the repeated experiments at the end of 4 h electrolysis. The average current density for three electrodes is ~18 mA cm−2 . The decrease in current density over time could be due to lower diffusivity and availability of electroactive species (Al2 Cl7 − or Ti-complex) in the vicinity of the electrode surface. Table 1 shows the weight loss and gain measured for Ti anode and Cu cathode for three experiments. The weight gain of Cu cathode increases with electrolysis time due to the deposition of Ti–Al. At the same time, the Ti ions released by Ti anode in BMIC-AlCl3 electrolyte also increases with time, as indicated by weight loss of Ti anode. Interestingly, the amount of Ti ions dissolved from the Ti anode is more significant than the percentage of Ti in the electrodeposited Ti–Al material (coating on Cu+ stripped powder). Although a higher concentration of Ti ions is released in electrolyte upon anodic dissolution of Ti anode, significantly fewer Ti[(Al2 Cl7 )4 ]2− ions are available for electrodeposition as there may be a saturation limit at equilibrium to form the Ti-complex with Al2 Cl7 − ions. As a result, there is a drop in current density over the electrolysis time. The current efficiency and energy consumption values align with our previously published reports [19, 29].

Characterization of Ti–Al Deposits by XRD and SEM-EDS The Ti–Al electrodeposits are characterized by XRD to determine the alloy composition. Figure 3 shows the normalized XRD patterns of Ti–Al electrodeposits on Cu substrate. All the Ti–Al deposits are crystalline and show five distinct crystallographic peaks (dark circles) such as (111), (200), (220), (311), and (222) belonging to the Ti0.40 Al0.60 phase of Ti–Al alloy with the cubic crystal structure (space group:

Potentiostatic Electrodeposition of Ti–Al Alloy with 40% Titanium … 3.0V for 4h (1st) 3.0V for 4h (2nd) 3.0V for 4h (3rd)

Ti0.4Al0.6(PDF#04-020-2451) Cu(PDF#00-04-0836)

Normalized Intensity

Fig. 3 XRD patterns of Ti–Al electrodeposits obtained from BMIC-AlCl3 on copper substrate by 2-electrode-electrolysis at 383 K using KEPCO power supply at −3.0 V for 4 h for three repeated experiments (Electrodes: 1st, 2nd, and 3rd). The vertical lines represent the lines from standard diffraction patterns of Ti–Al (Ti0.40 Al0.60 ) and Cu

81

20

40

60

80

100

2theta (o)

Fm-3 m (225), lattice parameter, a: 4.022 Å) as identified according to PDF4+ card (ICDD#:04-020-2451). The rest of the peaks (open squares) are due to the copper substrate (ICDD#:00-04-0836), and no other impurity or low-titanium Ti–Al alloy phases are seen. This suggests that Ti–Al electrodeposits obtained at − 3.0 V are reproducible, producing 40 at.% Ti. The SEM and EDS analyses were performed to examine the surface morphology and chemical composition of the electrodeposits. Figure 4 (left) shows the surface morphology of Ti–Al deposits on copper substrates obtained by constant potential electrolysis from BMIC-AlCl3 electrolyte at − 3.0 V for 4 h at 383 K. The morphology of Ti–Al deposits is uniform and porous. The Ti–Al growth is dendritic with uniformly covered spherical grains (20–25 μm) enclosing the smaller grains. The spherical grains break open over a prolonged period of electrolysis time. On top of the dendritic growth of spherical grains, few white overgrowths are crystallized. Such crystallization could be due to a slight rise in temperature of the

Fig. 4 Surface morphology and composition of representative Ti–Al alloy coating obtained on Cu substrate by constant-potential electrodeposition from BMIC-AlCl3 (1:2) at −3.0 V for 4 h

82

P. S. Shinde et al.

Al: Ti:

At.% 60.4 39.6

Fig. 5 Surface morphology and composition of representative Ti–Al alloy powder collected by scrapping/stripping the Ti–Al coating from the Cu substrate

BMIC-AlCl3 electrolyte leading to accumulation of diffused species of Al as well as Ti, eventually covering few regions of the Ti–Al deposits. Due to their porous nature and large surface area, these overgrowths appear to trap the anions, especially Ti-complex ions. Figure 4 (right) shows the representative EDS spectrum of Ti–Al deposit, revealing the chemical compositions of Ti and Al in Ti–Al electrodeposit to be 33.2 at.% and 66.8 at.%, respectively. It should be noted that the obtained values are not precisely accurate due to the semi-quantitative nature of the EDS measurement. The SEM–EDS analysis was also performed on the powders stripped/scraped from the Cu substrate, and the Ti content was close to 40% (38.9 ± 0.5 at.%), as shown in Fig. 5. Thus atomic percentage composition observed from SEM–EDS agrees reasonably with the XRD studies.

Calculation of Current Efficiency and Energy Consumption The weight gain of Ti–Al deposit on Cu cathode, the weight loss of Ti ions from Ti anode, Faradaic efficiency, and the energy consumption values shown earlier in Table 1 are calculated according to the procedure discussed below. The total weight of Ti–Al that included the weight of Ti–Al powder stripped and Ti–Al deposit on Cu substrate itself was considered for efficiency and energy consumption calculations. The cathodic current efficiency represents how efficiently the applied electricity is utilized to deposit the metals and is defined as the percentage of the weight gain of the electrodeposit relating to the actual Ti–Al alloy produced to the Ti–Al alloy that would theoretically be obtained based on Faraday’s law. The current efficiency (ï) is calculated using Eq. (3). =

W × 100% Wt

(3)

Potentiostatic Electrodeposition of Ti–Al Alloy with 40% Titanium …

83

The experimental weight gain (W ) was obtained by adding the weight difference of the cathode before and after electrodeposition, and the weight of the precipitate/powder stripped off from the cathode during the electrolysis/cleaning process. The theoretical weight gain (W t ) of the deposit is calculated using Faraday’s law, given by Eq. (4). Wt =

Qa jta A = nF nF

(4)

where j is the current density (A cm−2 ), t is time (s), a is the total weight of all the atoms in the deposit (g mol−1 ), A is the area of the electroactive region on the working electrode (cm2 ), n is the total number of transferred electrons, and F is Faraday constant. Since current density is not a constant during the electrodeposition process, the total charge (Q), a product (j × t) for a given area A, is obtained by integrating the current versus time plot in Fig. 2 using Eq. (5). t

Q = ∫ A × jdt

(5)

0

The total atomic weight (Al and Ti) and the total number of transferred electrons involving the reduction of Al and Ti are calculated based on the final composition of Ti and Al using Eqs. (6, 7). atotal = a Al × X Al + aT i × X T i

(6)

n total = n Al × X Al + n T i × X T i

(7)

where atotal is the total weight of Al and Ti atoms, ntotal is the total number of transferred electrons, aAl is the atomic weight of Al (26.982 g mol−1 ), aTi is the atomic weight of Ti (47.867 g mol−1 ), nAl is electrons transferred for producing 1 mol of Al (3), nTi is electrons transferred for producing 1 mol of Ti (2), X Al and X Ti are the corresponding atomic fractions of Al and Ti in the final Ti–Al deposit obtained from XRD. The energy consumption, E (kWh kg−1 ), of the electrodeposition process, is then determined using Eq. (8). E=

V×Q  × W

(8)

where V is the applied potential, Q is the total charge (Coulomb or Ampere second) calculated from Eq. (5). Thus, through this work, it is demonstrated that it is feasible to synthesize Ti– Al alloy with as high as 40 at.% Ti by low-temperature electrolysis from BMICAlCl3 ionic liquid electrolyte at relatively lower deposition potential by the in-situ electrochemical anodic dissolution of Ti ions from Ti anode during electrolysis.

84

P. S. Shinde et al.

Conclusions Electrodeposition of Ti–Al alloy with 40% Ti was accomplished for the first time from the Lewis acidic electrolyte containing 1-butyl-3-methylimidazolium chloride (BMIC) ionic liquid and aluminum chloride (AlCl3 ) with a 0.667-mol fraction of AlCl3 . The Ti–Al electrodeposits were obtained on Cu substrate by employing a constant potential of −3.0 V for 4 h in a two-electrode configuration at 383 K from BMIC-AlCl3 electrolyte. Both XRD and SEM–EDS confirmed the formation of crystalline Ti0.4 Al0.6 phase with cubic crystal structure on Cu. Three identical 4-electrolysis experiments confirmed the repeatability of producing 40-at.% Ti in the Ti–Al electrodeposit. The morphology of Ti–Al deposits consisted of uniform coverage of spherical grains with dendritic growth with few white overgrowths. The Faradaic current efficiency and energy consumption values of the Ti–Al electrodeposition process were obtained using cathode weight gain, anode weight loss, and the concentration of Ti obtained from XRD. The current efficiency and energy consumption values were 49.93 ± 0.95 and 23.77 ± 0.89 kWh kg−1 , respectively. More work is underway further to improve the percentage of Ti in the Ti–Al deposits. Acknowledgements The authors acknowledge the financial support from the National Science Foundation (NSF) award number 1762522 and ACIPCO for this research project. The authors also thank the Department of Metallurgical and Materials Engineering, The University of Alabama, for providing the experimental and analytical facilities.

References 1. Kroll W (1940) The production of ductile titanium. Trans Electrochem Soc 78:35 2. Crowley G (2003) How to extract low-cost titanium. Adv Mater Processes 161:25–27 3. Inagaki I, Takechi T, Shirai Y, Ariyasu N (2014) Application and features of titanium for the aerospace industry. Nippon Steel Sumitomo Metal Tech Rep 106:22–27 4. Peters M, Kumpfert J, Ward CH, Leyens C (2003) Titanium alloys for aerospace applications. Adv Eng Mater 5:419–427 5. Elias C, Lima J, Valiev R, Meyers M (2008) Biomedical applications of titanium and its alloys. Jom 60:46–49 6. Niinomi M, Nakai M, Hieda J, Cho K, Akahori T, Hattori T et al (2013) Research and development of low-cost titanium alloys for biomedical applications. Key Engineering Materials: Trans Tech Publ, pp 133–139 7. Zhang M, Kamavaram V, Reddy RG (2006) Ionic liquid metallurgy: novel electrolytes for metals extraction and refining technology. Mining Metall Explor 23:177–186 8. Fung KW, Mamantov G (1972) Electrochemistry of titanium (II) in AlCl3 -NaCl melts. J Electroanal Chem 35:27–34 9. Girginov A, Tzvetkoff TZ, Bojinov M (1995) Electrodeposition of refractory-metals (Ti, Zr, Nb, Ta) from molten-salt electrolytes. J Appl Electrochem 25:993–1003 10. Rolland W, Sterten A, Thonstad J (1987) Electrodeposition of titanium from chloride melts. Proc—Electrochem Soc 1987–7:775–785

Potentiostatic Electrodeposition of Ti–Al Alloy with 40% Titanium …

85

11. Head RB (1961) Electrolytic production of sintered titanium from titanium tetrachloride at a contact cathode. J Electrochem Soc 108:806–809 12. Stafford GR (1994) The electrodeposition of Al3 Ti from chloroaluminate electrolytes. J Electrochem Soc 141:945–953 13. Shinde P, Peng Y, Reddy RG (2020) Electroanalytical study of active species to deposit ti alloy from 1-butyl-3-methylimidazolium chloride-aluminum chloride ionic liquid. ECS Trans 98:231–243 14. Shinde PS, Peng Y, Reddy RG (2020) Electrodeposition of titanium aluminide (TiAl) alloy from AlCl3 -BMIC ionic liquid at low temperature. Springer International Publishing, Cham, pp 1659–1667 15. Bogala MR (2015) Electrodeposition of titanium aluminides from aluminum chloride: 1-butyl3-methyl imidazolium chloride ionic liquid. University of Alabama Libraries 16. Reddy RG, Shinde PS, Liu A (2021) The emerging technologies for producing low-cost titanium. J Electrochem Soc 17. Shinde PS, Reddy RG (2021) Effect of dissolution of titanium ions on Ti alloys electrodeposition from EMIC-AlCl3 ionic liquid at low temperature. Springer International Publishing, Cham, pp 141–153 18. Stafford GR, Moffat TP (1995) Electrochemistry of titanium in molten 2AlCl3 -NaCl. J Electrochem Soc 142:3288–3296 19. Pradhan D, Reddy R, Lahiri A (2009) Low-temperature production of Ti-Al alloys using ionic liquid electrolytes: effect of process variables on current density, current efficiency, and deposit morphology. Metall Mater Trans B 40:114–122 20. Tsuda T, Hussey CL, Stafford GR, Bonevich JE (2003) Electrochemistry of titanium and the electrodeposition of Al-Ti alloys in the lewis acidic aluminum chloride-1-Ethyl-3methylimidazolium chloride melt. J Electrochem Soc 150:C234–C243 21. Endres F, Zein El Abedin S, Saad AY, Moustafa EM, Borissenko N, Price WE et al (2008) On the electrodeposition of titanium in ionic liquids. Phys Chem Chem Phys 10:2189–2199 22. Cvetkovi´c VS, Vuki´cevi´c NM, Mili´cevi´c-Neumann K, Stopi´c S, Friedrich B, Jovi´cevi´c JN (2020) Electrochemical deposition of Al-Ti alloys from equimolar AlCl3 + NaCl containing electrochemically dissolved titanium. Metals 10:88 23. Uchida J-I, Seto H, Shibuya A (1995) Electrodeposition of Al-Ti alloy from chloroaluminate bath. J Surface Finish Soc Jpn 46:1167–1172 24. Awayssa O, Saevarsdottir G, Meirbekova R, Haarberg GM (2021) Electrodeposition of aluminium-titanium alloys from molten fluoride-oxide electrolytes. Electrochem Commun 123:106919 25. Janowski G, Stafford GR (1992) The microstructure of electrodeposited titanium-aluminum alloys. Metall Trans A 23:2715–2723 26. Stafford GR, Tsuda T, Hussey C (2003) Order/disorder in electrodeposited aluminum-titanium alloys. J Min MetallSect B 39:23–42 27. Xu C, Hua Y, Zhang Q, Li J, Lei Z, Lu D (2017) Electrodeposition of Al-Ti alloy on mild steel from AlCl3 -BMIC ionic liquid. J Solid State Electrochem 21:1349–1356 28. Pradhan D, Reddy RG, Electrodeposition of titanium using BmimCl ionic liquid at higher cathode current densities. Unpublished work: Unpublished work, p 1 29. Pradhan D, Reddy RG (2009) Electrochemical production of Ti-Al alloys using TiCl4 -AlCl3 -1butyl-3-methyl imidazolium chloride (BmimCl) electrolytes. Electrochim Acta 54:1874–1880 30. Pradhan D, Reddy RG (2014) Mechanistic study of Al electrodeposition from EMIC-AlCl3 and BMIC-AlCl3 electrolytes at low temperature. Mater Chem Phys 143:564–569

86

P. S. Shinde et al.

31. Jiang T, Chollier Brym MJ, Dubé G, Lasia A, Brisard GM (2006) Electrodeposition of aluminium from ionic liquids: Part I—electrodeposition and surface morphology of aluminium from aluminium chloride (AlCl3 )–1-ethyl-3-methylimidazolium chloride ([EMIm]Cl) ionic liquids. Surf Coat Technol 201:1–9 32. Tang J, Azumi K (2011) Optimization of pulsed electrodeposition of aluminum from AlCl31-ethyl-3-methylimidazolium chloride ionic liquid. Electrochim Acta 56:1130–1137

Prediction of Distribution of Composition of Inclusion in Continuous Casting Bloom of the Heavy Rail Steel Coupling Element Segregation, Heat Transfer, and Kinetics Yuexin Zhang, Wei Chen, Jujin Wang, Yadong Wang, Wen Yang, and Ying Ren Abstract A kinetic model had been used to predict the distribution of composition of inclusions in continuous casting (CC) bloom of the heavy rail steel. On this basis, the elemental segregation model was introduced to predict the distribution of inclusions in CC bloom. Samples from the inner arc to the outer arc of the bloom were selected for the investigation. Calculation of mold transient flow field, heat transfer, solidification, thermodynamic analysis, and element diffusion were combined to predict the composition distribution of inclusions in the whole section of the bloom. The experimental results showed that CaO, Al2 O3 , SiO2 , MgO, and CaS of inclusions fluctuated in the range of 15%–30%, 22%–30%, 27%–30%, 8%–12%, and 5%–20%, respectively. The model predicted that CaO, Al2 O3 , SiO2 , MgO, and CaS of inclusions varied in the range of 18.9%–28.6%, 23.1%–30.7%, 25.9%–31.5%, 5.9%–10.9%, and 7.2%–16.3%, respectively, which indicated that the predicted results were in good agreement with the measured ones. Keywords Heavy rail steels · Kinetics · Segregation · Inclusions

Introduction As the main component of railway rails, the quality of the heavy rail steel is closely related to the safety and efficiency of railway transportation. Non-metallic inclusions have a non-negligible influence of the fatigue, machinability, tensile, welding, or corrosion properties of the final product of steels [1]. For this reason, the nonmetallic inclusions in the heavy rail steel at all stages must be strictly controlled. The

Y. Zhang · J. Wang · Y. Wang · W. Yang · Y. Ren School of Metallurgical and Ecological Engineering, University of Science and Technology Beijing, Beijing 100083, China W. Chen (B) School of Mechanical Engineering, Yanshan University, Qinhuangdao 066004, China e-mail: [email protected] © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_9

87

88

Y. Zhang et al.

heavy rail steel is deoxidized by silicon and manganese, resulting in complex SiO2 Al2 O3 -CaO deoxidized products. In order to clarify and reduce the content of nonmetallic inclusions in steel, it is very important to control inclusions in the process of steelmaking, refining, and continuous casting [2–4]. The removal and modification of inclusions in the molten steel by deoxidation, slag refining, secondary oxidation, calcium treatment, and rare earth treatment have been extensively studied [5–8]. The prediction of metal, slag, composition of inclusions in the molten steel using thermodynamic [9, 10] and kinetic models [11–15] is a current research topic. In addition, the evolution of inclusion in the solid steel during solidification, cooling, and heat treatment has also been investigated [16–18]. Wintz et al. [19] reported that heat treatment of type 304 stainless steel in the range of 1273–1473 K can transform manganese silicate inclusions to manganese chromite (spinel) inclusions. Up to now, some researchers have developed coupling models combining heat transfer and solidification of steel, thermodynamic simulations, and kinetic simulations to predict the distribution of the quantity, size, and composition of inclusions in continuous casting (CC) billets. However, these models did not take into account the influence of element segregation in the steel [20–23]. In this work, a model was established to predict the distribution of inclusion composition within a heavy rail steel CC bloom by coupling thermodynamic, element segregation, heat transfer, and kinetics of inclusions.

Evolution of Inclusions Considering the U75V high-speed heavy rail steel produced by a domestic steel plant, the production route was as follows: KR hot metal pretreatment desulfurization → BOF(150 t) → LF → VD → CC. BaCaSi alloys, MnSi alloys, and FeSi alloys were added for deoxidation during converter tapping. The CC bloom of the heavy rail steel was scanned by automatic scanning SEM–EDS system, and the quantity, size, composition, and morphology of non-metallic inclusions were analyzed. Figure 1 shows the distribution of composition of inclusions at the surface and one quarter of the outer arc of CC bloom, respectively. In Fig. 1a, the scanning area, average size, number density, and area fraction of the sample were 63.40 mm2 , 1.69 μm, 26.65 #/mm2 , and 59.67 ppm, respectively. The main composition of inclusions was 23.46wt%CaO29.22%wtSiO2 -27.25wt%Al2 O3 -9.76wt%MgO-10.31wt%CaS. In Fig. 1b, the scanning area, average size, number density, and area fraction of the sample were 73.54 mm2 , 3.7 μm, 5.44 #/mm2 , and 58.62 ppm, respectively. The main composition of inclusions was 18.17wt%CaO-29.49wt%SiO2 -26.74wt%Al2 O3 -9.90wt%MgO15.70wt%CaS. Compared with inclusions at 1/4 of the outer arc of the bloom, there were many inclusions on the surface of the bloom, and the size was small. The scanning results of the element distribution of typical inclusions at the surface and the 1/4 of the heavy rail steel CC bloom are shown in Fig. 2. The inclusion was

Prediction of Distribution of Composition of Inclusion … Average composition 1673K Isoline 1773K Isoline 1873K Isoline

Dmax= 11.06 µm Dmin = 1.01 µm Number : 1860

Average composition 1673K Isoline 1773K Isoline 1873K Isoline

SiO2

0

100

25

Dmax= 15.13 µm Dmin = 1.06 µm Number : 400

75

50

0

CaO

50

(wt%)

75

(a) at the surface

100

25

75

100

50

75

25

25

SiO2

0

50

50

75

100

89

0

Al2O3

100

0

CaO

25

25

50

(wt%)

75

100

0

Al2O3

(b) at 1/4 of the outer arc

Fig. 1 Distribution of composition of inclusions at the surface (a) and one quarter of the outer arc (b) of CC bloom

unevenly distributed, mainly CaO-MgO-SiO2 -Al2 O3 and CaS phases. The CaOSiO2 -Al2 O3 -MgO inclusion was located inside the phase, and the outermost layer was wrapped by CaS.

Model Introduction A comprehensive model had been established to predict the spatial distribution of composition of inclusions on the entire cross section of a heavy rail steel CC bloom. The detailed description of the model had been described in [24]. In the study, influence of element segregation was considered. A model coupling the heat transfer and element segregation of CC bloom, thermodynamics and kinetics of inclusion were used to predict the distribution of composition of inclusion in CC bloom, as shown in Fig. 3. Element segregation in the bloom is related to the mass, momentum, and solute transfer. Firstly, the element content in the steel was calculated by using segregation model, and then composition of inclusions was calculated by thermodynamic and kinetic simulation. Thermodynamic equilibrium between steel matrix and inclusions was simulated theoretically using FactSage7.1 [25] databases with FactPS, FToxide, FSstel. Considering the spatial differences of elements in steel, the kinetic model was established based on Fick’s first law to imitate the evolution of composition of inclusion caused by the reaction between elements and inclusions. Diffusion coefficients of the dissolved S in liquid γ steel were 0.1–4.2 × exp (−223,426/RT)/10000 m2 /s. Mathematical description of the segregation model had been described elsewhere in detail [26].

90

Y. Zhang et al.

(a) at the surface

(b) at 1/4 of the outer arc Fig. 2 Element distribution of typical inclusions at the surface (a) and the 1/4 of the outer arc (b) of the heavy rail steel CC bloom

Model Validation The distribution of C, Si, Mn, and other elements in the steel at different distances from the meniscus was calculated using the segregation model. Then, the composition distribution of inclusions along the inner arc to outer arc at the center of bloom width was calculated by using the comprehensive model, as shown in Fig. 4. With the increase of the distance from the center of continuous casting bloom to the surface layer, the content of CaO of inclusions first decreased from 28.6 to 18.9% and then increased to 26.3%. On the contrary, the content of CaS increased from 7.2 to 16% and then decreased to 8.2%. The content of SiO2 , Al2 O3 , and MgO varied little,

Prediction of Distribution of Composition of Inclusion …

91

Fig. 3 Schematic of the model to predict the composition of inclusions in CC bloom

Distance from center of thickness (mm)

140

Meassured CaO Al2O3 SiO2 MgO CaS Predicted CaO Al2O3 SiO2 MgO CaS

120 100 80 60 40 20 0 -20 -40 -60 -80 -100 -120 -140 0

10

20

30

40

50

60

Composition of inclusions (wt%) Fig. 4 Comparison between measured and predicted composition of inclusions from the loose side to the fixed side of heavy rail steel CC bloom

92

Y. Zhang et al.

fluctuating in the range of 28%–34%, 23%–26%, and 8%–12%, respectively. The predicted results of the model were basically consistent with the experimental ones. The result not only verified the accuracy of the model, but also can be used to provide guidance to the actual steelmaking process. The combination of element segregation, heat transfer, mass transfer, and the transformation of inclusions to predict the distribution of inclusion composition on the whole section of CC bloom of different steel grades needs to be further studied.

Conclusions (1)

(2)

(3)

Inclusions in heavy rail steel CC bloom were CaO-SiO2 -Al2 O3 -MgO-CaS, of which CaO-SiO2 -Al2 O3 -MgO was located inside the phase, and the outermost layer was wrapped with CaS. A comprehensive model was developed by coupling the heat transfer, element segregation of the CC bloom, and the thermodynamics and kinetics of inclusions. Along the cross section of the bloom, as the distance from the center of the bloom to the surface increased, the content of CaO of inclusions first decreased from 28.6 to 18.9% and then increased to 26.3%. On the contrary, the content of CaS increased from 7.2 to 16% and then decreased to 8.2%. The content of SiO2 , Al2 O3 , and MgO varied little. The model coupling thermodynamics, kinetics, and segregation calculated the distribution of inclusions along the loose side to the fixed side at the center of the width direction of heavy rail steel, and the predicted result was basically consistent with the experimental one. The combination of element segregation, heat transfer, mass transfer, and the transformation of inclusions to predict the distribution of inclusion composition on the whole section of CC bloom of different steel grades needs to be further studied.

Acknowledgements The authors are grateful for support from the National Natural Science Foundation of China (Grant No. U1860206, No. 51725402, 51874031, 51874032), the S&T Program of Hebei (Grant No. 20311004D), the High Steel Center (HSC) at Yanshan University, Hebei Innovation Center of the Development and Application of High Quality Steel Materials, Hebei International Research Center of Advanced and Intelligent Manufacturing of High Quality Steel Materials, Beijing International Center of Advanced and Intelligent Manufacturing of High Quality Steel Materials (ICSM) and the High Quality Steel Consortium (HQSC) at University of Science and Technology Beijing, China.

References 1. Simachev A, Oskolkova T, Temlyantsev M (2016) Influence of nonmetallic inclusions in rail steel on the high-temperature plasticity. Steel Trans 46(2):112–114

Prediction of Distribution of Composition of Inclusion …

93

2. Zhang L, Guo C, Yang W, Ren Y, Ling H (2018) Deformability of oxide inclusions in tire cord steels. Metall Mater Trans B 49(2):803–811 3. Zhang L, Thomas BG (2003) State of the art in evaluation and control of steel cleanliness. ISIJ Int 43(3):271–291 4. Chu Y, Chen Z, Liu N, Yang W, Wang J, Zhang L (2018) Behavior evolution of nonmetallic inclusions during production of U75V heavy rail steel (in Chinese). China Metall 28(Supplement1):83–89 5. Zhang L (2019) Non-metallic inclusions in steels: fundamentals (in Chinese). Metallurgical Industry Press, Beijing 6. Godik L, Kozyrev N, Korneva L (2009) Optimizing the oxygen content in rail steel. Steel Trans 39(3):240–242 7. Holappa L, Hamalainen M, Liukkonen M, Lind M (2003) Thermodynamic examination of inclusion modification and precipitation from calcium treatment to solidified steel. Ironmak Steelmak 30(2):111–115 8. Han ZJ, Liu L, Lind M, Holappa L (2006) Mechanism and kinetics of transformation of alumina inclusions by calcium treatment. Acta Metall Sin (English Letters) 19(1):1–8 9. Ren Y, Zhang L, Fang W, Shao S, Yang J, Mao W (2016) Effect of slag composition on inclusions in Si-deoxidized 18Cr-8Ni stainless steels. Metall Mater Trans B 47(2):1024–1034 10. Ha H, Park C, Kwon H (2006) Effects of misch metal on the formation of non-metallic inclusions and the associated resistance to pitting corrosion in 25% Cr duplex stainless steels. Scripta Mater 55(11):991–994 11. Suito H, Inoue R (1996) Thermodynamics on control of inclusions composition in ultraclean steels. ISIJ Int 36(5):528–536 12. Tang H, Li J (2010) Thermodynamic analysis on the formation mechanism of MgO·Al2 O3 spinel type inclusions in casing steel. Int J Miner Metall Mater 17(1):32–38 13. Harada A, Maruoka N, Shibata H, Kitamura SY (2013) A kinetic model to predict the compositions of metal, slag and inclusions during ladle refining: part 1. Basic concept and application. ISIJ Int 53(12):2110–2117 14. Harada A, Maruoka N, Shibata H, Kitamura SY (2013) A kinetic model to predict the compositions of metal, slag and inclusions during ladle refining: part 2. Condition to control the inclusion composition. ISIJ Int 53(12):2118–2125 15. Jamieson BJ, Tabatabaei Y, Barati M, Coley KS (2019) Kinetic modeling of the silicothermic reduction of manganese oxide from slag. Metall Mater Trans B 50(1):192–203 16. Ende M-AV, Kim Y-M, Cho M-K, Choi J, Jung I-H (2011) A kinetic model for the ruhrstahl heraeus (RH) degassing process. Metall Mater Trans B 42(3):477–489 17. Zhang Y, Ren Y, Zhang L (2018) Kinetic study on compositional variations of inclusions, steel and slag during refining process. Metall Res Technol 115(4):1–7 18. Cheng G, Li W, Zhang X, Zhang L (2019) Transformation of inclusions in solid GCr15 bearing steels during heat treatment. Metals 9(6):642 19. Wintz M, Bobadilla M, Lehmann J, Gaye H (1995) Experimental study and modeling of the precipitation of non-metallic inclusions during solidification of steel. ISIJ Int 35(6):715–722 20. Zhang Y, Zhang L, Chu Y, Ren Q, Wang J, Liu N, Chen Z, Zhi J (2020) Transformation of inclusions in a complicated-deoxidized heavy rail steels during heating. Steel Res Int 91(9):2000120 21. Ren Y, Zhang L, Pistorius PC (2017) Transformation of oxide inclusions in type 304 stainless steels during heat treatment. Metall Mater Trans B 48(5):2281–2292 22. Y Zhang, L Zhang, J Wang, K Niu, Y Wang (2021) Prediction of composition distribution of non-metallic inclusions in a billet (in Chinese). Iron Steel 56(10):74–82 23. Wang J, Zhang L, Zhang Y, Cheng G, Wang Y, Ren Q, Yang W (2021) Prediction of spatial composition distribution of inclusions in the continuous casting bloom of a bearing steel under unsteady casting. ISIJ Int 61(3):824–833 24. Ren Q, Zhang Y, Zhang L, Wang J, Chu Y, Wang Y, Ren Y (2020) Prediction on the spatial distribution of the composition of inclusions in a heavy rail steel continuous casting bloom. J Market Res 9(3):5648–5665

94

Y. Zhang et al.

25. Ren Y, Zhang L, Li S (2014) Transient evolution of inclusions during calcium modification in linepipe steels. ISIJ Int 54(12):2772–2779 26. Y Wang, L Zhang, W Yang, Y Ren (2021) Effect of nozzle type on fluid flow, solidification, and solute transport in mold with mold electromagnetic stirring. J Iron Steel Res Int (Online)

Strategies for the Upgrade of a TBZC Product (Tetra Basic Zinc Chloride) by Selective Removal of the Impurity Chlorine L. Höber, R. Ahmed, T. Hofbauer, and S. Steinlechner

Abstract When recycling materials containing zinc in hydrochloric acid processes, the precipitation of zinc can lead to the formation of various compounds which can be assigned to the group of zinc hydroxide chlorides. This paper compares different approaches for the selective removal of chlorine from tetra basic zinc chloride to obtain a cleaned zinc product. The removal of chlorine via soda leaching at atmospheric conditions as well as under overpressure in an autoclave was investigated. Furthermore, concepts were considered in which the chlorine can be evaporated as a compound and thus separated via selective temperature and atmospheric control. Thereby, a focus is on the simulation of pyrohydrolysis and clinkering for the separation of chlorine via gaseous compounds whereby zinc remains and can be brought to further processing. The simulations with multivariant parameters are carried out using the thermochemical calculation software package FactSage. Keywords Chlorine removal · Tetra basic zinc chloride · Hydrometallurgical chlorine extraction · Pyrometallurgical chlorine extraction

Introduction When zinc is precipitated from hydrochloric acid solutions, depending on the conditions such as the pH, zinc hydroxide chlorides can be formed. In processes that serve to obtain high-purity zinc oxides, this circumstance can lead to problems, since significant amounts of zinc evaporate as chlorides during the necessary dehydration/clinker L. Höber (B) · S. Steinlechner Christian Doppler Laboratory for Selective Recovery of Minor Metals Using Innovative Process Concepts, Montanuniversität Leoben, Franz Josef-St. 18, 8700 Leoben, Austria e-mail: [email protected] R. Ahmed Nonferrous Metallurgy, Montanuniversität Leoben, Franz Josef-St. 18, 8700 Leoben, Austria T. Hofbauer Andritz AG, Eibesbrunnergasse 20, 1120 Vienna, Austria © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_10

95

96

L. Höber et al.

step and are thus lost. In order to counteract this problem, various possibilities are available for separating the contained chlorine. A number of approaches were evaluated in the course of this research in thermodynamic simulations as well as in experiments on a laboratory scale. In general, they can be divided into hydrometallurgical and thermal treatments. In the field of hydrometallurgical processes, soda-leaching treatments were carried out under atmospheric pressure as well as overpressure. In the pyrometallurgical field, the chlorine removal was investigated in clinker tests and experiments using pyrohydrolysis. Pyrometallurgical treatment steps of selected compounds were simulated using the thermodynamic software FactSage. Figure 1 gives an overview of possible approaches for the removal of chlorine from zinc precipitates. The purpose of the chlorine removal is to keep zinc losses due to ZnCl2 evaporation in the clinker step low and generate a material which is suitable for the production of high-purity zinc oxide. Zinc Oxide (ZnO) is a widely studied material which is of high importance in the chemical and electrical industry [1]. It has a wide range of uses and applications and it can be prepared by different pyro- as well as hydrometallurgical methods [1, 2]. ZnO thin films preparation and synthesis have been studied since the 1960s due to their important applications in sensors, transducers [3], optoelectronics applications [4, 5], space applications [4], thin films transistors [5], facial powders, varistors, transparent conducting films [6], solar cells [7], piezoelectric devices, and photocatalysis [8]. Considering its unique physical and chemical properties and wide range of use, ZnO can be called a multifunction material [9]. ZnO has distinctive characteristics which can be used in many fields of industry which lead to rising interest in many research fields. It has a high thermal, mechanical, and chemical stability and it is nontoxic with environmental compatibility [2, 10]. In addition, it has good physico-chemical characteristics, good

Fig. 1 Schematic overview of different treatment possibilities for the removal of chlorine from zinc precipitates with significant amounts of zinc hydroxide chlorides (TBZC)

Strategies for the Upgrade of a TBZC Product (Tetra Basic Zinc Chloride) …

97

electrical, ferromagnetic, piezoelectric, and optoelectric properties [2]. The properties of ZnO materials can be affected by the shape, defect structure, size, and crystallinity of its particles [2, 7]. Some of the developed morphologies of ZnO are nanorods, ultralong nanobelts with various facets, nanoneedles, hollow structures, nanocombs, and nanohelixes [3, 8]. The preparation method is important to create well-oriented ZnO nanoparticles [8], with controlling the preparation method and with the development of many methods for the synthesis of ZnO powders and films it is possible to produce its particles in different sizes and shapes making it easier to create different products [9]. In the course of specific treatment methods to win ZnO from residual materials using hydrochloric solutions, Zn can precipitate, depending on the pH, in the form of zinc hydroxide chlorides in significant extents. Zinc hydroxide chloride monohydrate (ZHC) or simonkolleite is a layered hydroxide salt [11]. It has the chemical formula Zn5 (OH)8 Cl2 ·H2 O and can be used in many commercial applications [12]. While the usage of zinc nitrate, and zinc acetate were the most reported precursor materials, the usage of ZHC in preparing ZnO is currently rarely reported [8]. ZHC can be used in many applications such as flame retardants, antacids, drug fillers, anticorrosion agents, clay modified electrodes, and wastewater cleaning [12] and as layered hydroxide salt as a nanostructured oxide precursor [13]. ZHC can be synthesized or can be found in nature, for example, in zinc mines or as deposits of zinc and galvanized steel corrosion products in marine environments [14]. The thermal decomposition mechanism of ZHC has been studied since 1967, it is reported to be complex with overlapping steps, this decomposition was reported to lead to the formation of ZnO [15, 16]. However, there is still no entirely accepted mechanism for this process and it is reported that the process is highly dependent on the experimental conditions [1, 12, 17]. In 1967, two of the first studies about the decomposition of ZHC were reported by a team of Spanish researchers which suggested a decomposition mechanism which is represented in Formula (1) and claimed that the characteristic lines of ZnO were visible in XRD at 140 °C which means that the ZHC decomposed at a relatively low temperature. Another research proposed a different mechanism which consisted of 2 steps and claimed that the mass loss was due to the vaporization of zinc chloride (ZnCl2 ) and that its melting caused an endothermic peak at 262 °C [12, 18–20]. A different research was dedicated to the same thermal decomposition which confirmed the formation of ZnCl2 but did not detect the two steps of dehydration [21]. Zn5 (OH)8 Cl2 · H2 O → 5ZnO + 2HCl(g) + 4H2 (g)

(1)

In 1980, an intermediate compound was reported to be formed. This compound is zinc hydroxide chloride (Zn(OH)Cl) which was found to decompose at ~270 °C with endothermic behavior, which was assigned to the melting of ZnCl2 [12, 22]. The formation of the same intermediate compound was confirmed later in 1994, and it was proposed that it decomposes according to Formula 2 [17]. An important note on Formula 2 is that ZnCl2 is present as a hydrated salt and its transformation involves vaporization and thermal hydrolysis, which was reported to be highly affected by the

98

L. Höber et al.

Fig. 2 General schematic of the thermal decomposition process of TBZC [12]

experimental conditions [12, 17]. Providing a low heating rate, the decomposition is associated with a thermal hydrolysis of ZnCl2 as represented by Formula (3). This hydrolysis was reported to be intensive leading to the full conversion of ZnCl2 to ZnO [17]. 2Zn(OH)Cl → ZnCl2 · xH2 O + ZnO + (1 − x)H2 O(g), x < 1

(2)

ZnCl2 · H2 O → ZnO + 2HCl(g)

(3)

In 2010 it was confirmed that the formation of the intermediate compound Zn(OH)Cl and its decomposition is taking place between 200 and 225 °C as represented by Formula (4) [16]. Zn(OH)Cl → ZnO + HCl(g)

(4)

Figure 2 shows a schematic of a general thermal decomposition mechanism according to the mentioned studies [12]. Based on the information in literature a number of approaches were investigated on a laboratory scale to remove the chlorine from materials containing TBZC. The studied materials were synthesized in the laboratory via precipitation from a synthetic iron-containing hydrochloric zinc solution after purification.

Materials and Methods Production of the Zinc Precipitate The investigated zinc hydroxide chloride material was produced in a reproducible experiment on a laboratory scale. The target was a quantity of 600 g of material, for

Strategies for the Upgrade of a TBZC Product (Tetra Basic Zinc Chloride) … Table 1 Initial amount of chemicals utilized for the preparation of the zinc chloride solution

Substance

99

Quantity [g]

MnO

11.72

PbO

1.6

Fe2 O3

3.3

CaO

20.72

ZnO

622.08

HCl (37%)

1536.41

H2 O

963.59

which 2.5 l of zinc hydroxide chloride solution were prepared. Laboratory chemicals were used to prepare the synthetic solution, including typical impurities, in the amounts shown in Table 1. The initial solution was prepared to be close to one that would realistically be present in a hydrometallurgical recycling process for steel mill dusts. When dissolving the oxides in the hydrochloric acid solution, it is important to add the poorly soluble oxides such as PbO and MnO at the beginning to ensure that sufficient free acid is available. Subsequently, the other metal compounds are added. In a first treatment step, the added iron is precipitated from the solution. The addition of hydrogen peroxide converts any Fe2+ ions into the trivalent form. Subsequently, a Mg(OH)2 suspension is added to precipitate the zinc present as zinc hydroxide, whereby significant amounts of zinc hydroxide chlorides are formed. The obtained precipitated zinc products used in the further tests to investigate the zinc removal capability have a chemical analysis as shown in Table 2. Table 2 Chemical analysis of the zinc precipitates

Element

Batch 1

Batch 2

[wt-%]

[wt-%]

Ca

60 MPa, with a cohesive failure. • X-ray diffraction in the as sprayed condition showed peak broadening indicative of micro-strains, which go relieved in the heat-treated condition. • The residual stress of the heat-treated coating is compressive, ~100 MPa uniform near the interface, as measured using the Sin2 ψ technique. • As sprayed coating shows nearly a 1 ppm/o C lower CTE than the substrate up to 300 °C, above which temperature it matches with the substrate. • Coating microstructure showed severe deformation, in the as sprayed condition. Coating was seen to recover post-heat treatment, having a grain size of ~2 μm along with niobium carbide precipitates. • The coating exhibits lower yield strength and little ductility in comparison with the substrate (base material), after heat treatment, measuring 0.2% YS of 452 MPa and UTS of 900 MPa. Based on these characterization and tests, it is observed that the cold spray coating process (using Nitrogen gas) may be a good candidate for repair restoration of component (for those less critical on ductility) to achieve a near net shape formation. Overall, based on the analysis, it appears feasible to repair the IN718 stationary parts using cold spray additive technique to repair worn surfaces or to modify dimensions, after adjusting the heat treatment, without having to scrap the component. Acknowledgements This analysis and research work were performed with the support from GE Aviation, GE Gas Power, Repair Development Center (RDC), and Material Processes Engineering (MPE), USA. We acknowledge Impact Innovations GmbH for the cold spray coating coupons. Repair Engineering Team Members Joseph Bamonte, Warren Grossklaus, Timothy Rasch, Michael Schulte, and Chris Lambert are gratefully acknowledged for supporting and funding this work. Dr. Ramar Amuthan, Shivanandappa Meti, Lakshmikanth S., and Praveen R., are thanked for their help with the analysis work, laboratory support, and mechanical tests. The authors would like to thank GE India Industrial Private Limited, GEKTC Kuwait, and GE Gas Power Engineering for the support in presenting this work.

Residual Stress, Microstructure, and Characterization of Self-Mated Additive Repair …

237

References 1. Wong W, Irissou E, Legoux J-G, Bernier F (2012) Cold Spray Forming Inconel 718. In: Thermal spray 2012: proceedings from the international thermal spray conference and exposition, Houston, Texas, USA 2. Stephen JB (2006) Buckets and nozzles. Gas Turbine Handbook, Department of Energy, USA, pp 411–419. https://netl.doe.gov/coal/turbines/handbook 3. Yin S et al (2018) Cold spray additive manufacturing and repair: fundamentals and applications. Addit Manufact 21:628–650. https://dx.doi.org/10.1016/j.addma.2018.04.017 4. Stoltenhoff T et al. (2002) An analysis of the cold spray process and its coatings. H. J. J Thermal Spray Tech 11:542 5. Widener CA et al (2008) Structural repair using cold spray technology for enhanced sustainability of high value steels. Procedia Manuf 21:361–368 6. Gartner F, Schmidt T, Kreye H (2007) Present status and future prospects of cold spraying. Mater Sci Forum 534–536, 433–436. https://doi.org/10.4028/www.scientific.net/MSF.534-536.433 7. Intrater J (2002) Cold spray technology—prospects and applications. Surf Eng 18(5):321–323 8. Gartner F, Stoltenhoff T, Schmidt T, Kreye H (2006) The cold spray process and its potential for industrial applications. J Therm Spray Technol 210:223–232

Reverse Engineering of Aerospace Components Utilizing Additive Manufacturing Technology Balakrishnan Subeshan, Abdulaziz Abdulaziz, Zeeshan Khan, Md. Nizam Uddin, Muhammad Mustafizur Rahman, and Eylem Asmatulu

Abstract Using conventional manufacturing methods for product development typically involves a relatively long lead time and cost, especially for obsolete, wornout, or broken parts. Reverse engineering is a preferred solution for reproducing obsolete parts and has been increasingly utilized to advance additive manufacturing technology. It is a combined process of laser scanning the obsolete parts where engineering design has become unavailable. These designs are then converted into patterns for sand casting to manufacture three-dimensional (3D) prototypes for further product development. The combination of reverse engineering and additive manufacturing is being utilized to manufacture the pattern for sand casting to produce the final product faster and distribute that to the industry more rapidly. Additive manufacturing technology has had a significant impact on the manufacturing industry throughout the world. There are various application fields where additive manufacturing has had a significant impact, and one of these fields is the aerospace industry. This study presents technologies and methodologies for reverse engineering, illustrated by a stainless-steel lever part from an aircraft control assembly. It involves the reconstruction of part geometry using laser scanning, fabrication of the pattern for sand casting using material extrusion additive manufacturing technology, and reversed part fabrication using sand casting. It was found that the fabrication of patterns directly from reverse-engineered computer-aided design (CAD) data using a suitable additive manufacturing technique provides a reliable and economic path B. Subeshan · Z. Khan · M. M. Rahman · E. Asmatulu (B) Department of Mechanical Engineering, Wichita State University, 1845 Fairmount St., Wichita, KS 67270, USA e-mail: [email protected] M. M. Rahman e-mail: [email protected] A. Abdulaziz Department of Industrial, Systems, and Manufacturing Engineering, Wichita State University, 1845 Fairmount St., Wichita, KS 67270, USA Md. N. Uddin College of Business, Engineering, and Technology, Texas A & M University, Texarkana, TX 75503, USA © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_21

238

Reverse Engineering of Aerospace Components Utilizing Additive …

239

for rapid product development of complicated parts for replacement purposes. The process of reverse engineering comprises of laser scanning, CAD data, and all regular manufacturing steps to make that part. Keywords Reverse engineering · Additive manufacturing · 3D printing · Material science · Data capturing · Material extrusion · CAD reconstruction

Introduction Reverse engineering refers to the process of generating engineering design information from existing parts. It refabricates an existing part by obtaining its surface information using a laser scanning or measurement device. It is valuable to reconstruct the three-dimensional (3D) prototype of an existing part when the engineering design data is missing or when the model has disappeared from numerous design variations, especially when the part becomes obsolete, worn-out, or broken. Furthermore, reverse engineering consists of breaking down an innovation explicitly to determine how it works and how it was designed. This breaking down draws a useful learning process [1, 2]. As a technique, reverse engineering is not set for a specific reason. Nevertheless, reverse engineering is frequently a significant part of logical strategy for innovative product development. The way of dismantling a part and finding out how it works is frequently a fascinating method to determine how to manufacture the part or upgrade it [3]. Reverse engineering is a way of producing a prototype using a computer-aided design (CAD) model from an existing part that has already been manufactured. It allows for the reproduction of a part by understanding its physical and geometrical measurements [4]. This reverse configuration approach begins with a physical part and works backward through the procedure to plot the part’s measurements and structure, which allow the part to rationally recreate the structural views that were originally produced [5]. Reverse engineering has unique characteristics in the typical design process. It takes a physical part and makes it into a CAD model to change or adjust the structure. It can similarly characterize the procedure or copy an existing part by gathering segments of the physical measurements [6]. This type of engineering is typically incorporated to upgrade the design of a product for better viability or to deliver a duplicate of a design without having access to the design plan from which the part was initially created. On many occasions, it is useful for improved maintenance, or when the technical data is misplaced, imprecise, or obsolete. Examples include handmade prototypes and the reproduction of obsolete engineering objects used in aerospace, automotive, medical, and dental applications. There are many different processes used to obtain data, but most of them are time-consuming. Technical information is essential for the smooth operation and continuous work of any generation of additive manufacturing, helping to manage the model of the part’s computer-aided manufacturing (CAM) [7, 8]. A slight insufficiency or inaccuracy in data has repressed reverse engineering. By simplifying CAM operations of these physical models, it is essential to create their CAD models

240

B. Subeshan et al.

[9]. Reverse engineering is the fastest method to get data into any system. Technical data is essential to any industry’s smooth and continuous relationship with a manufacturing facility. Presently, there has been a demand for reverse engineering within many industries, such as aerospace, automotive, medical, robotics, military, and various other research environments. This study aims to minimize the knowledge gap regarding reverse engineering and achieve a complete understanding of how to deal with reverse engineering for future studies [10–13]. This study aims to cover the details of reverse engineering, showing a step-by-step process of the fabrication of patterns for sand casting a lever part from a control assembly of an aircraft using the additive manufacturing technique with polymeric materials. The study describes the various methods of creating CAD data and scanning technologies used to capture data for part geometry, a key component in the reverse engineering process. The digitized component can be manipulated in CAD modeling software to generate the files needed for additive manufacturing. This study also demonstrates the different additive manufacturing technology methods applied to create a casting pattern. Replacing older components that are no longer in production can result in substantial manufacturing costs involving complexity when using a variety of molds, machines, and tools for sand casting. Employing additive manufacturing for creating a pattern for the casting of a part has been shown to reduce these costs significantly and minimize material waste.

Experimental Procedure Equipment and Materials The experimental setup consisted of various pieces of equipment: HandySCAN 700 scanner, Stratasys Dimension BST 1200 printer, and Fusion 360 and VX Element 7.0 software. The HandySCAN 700 scanner was attached to a USB 3.1 connector, which facilitated getting all the cloud points faster. VX Elements 7.0 software, which is used to capture the geometry utilizing the HandySCAN 700, is fully integrated with other 3D software platforms, such as Fusion 360, and allows for the export of files. Moreover, VX Elements 7.0 is a post-processing software that integrates VX elements and enables processing of the 3D point cloud to transfer in CAD modeling software. Fusion 360 is the CAD modeling software used to edit the mesh file and finalize the 3D scanned data that is employed directly in most rapid prototyping techniques, preferably for metal 3D printing. This software allows objects to be exported in stereolithography (STL) format, which is required for 3D printing the casting pattern. HandySCAN 700 is a portable laser 3D scanner that considers multiple measurements with different speed and accuracy. It uses seven lasers, which create a grid and an additional line for increased accuracy. The scanning area is approximately 275 mm × 250 mm, the volumetric accuracy is 0.02 mm ± 0.06 mm/m, 56~60 images are scanned per second, each image collected is about 600 points, and the measurement

Reverse Engineering of Aerospace Components Utilizing Additive …

241

Fig. 1 Equipment used to perform reverse engineering of an obsolete part: a HandySCAN 700 laser scanner and b Dimension 1200es BST 3D printer

rate can meet 480,000 measures. HandySCAN 700 consists of a laser projector, and a CCD sensor triangulation system is used for self-positioning. The illuminated light emitted by the laser projector is spatially modulated by the surface of the object to be measured. Changing the angle of the imaging beam, the position of the structured light on the CCD sensor results in a change. Figure 1a displays a commercial HandySCAN 700. The Dimension 1200es BST 3D printer is based on material extrusion technology. This printer has space to build 3D designs up to 254 × 254 × 305 mm, which allows more than one model to be created at a time. Figure 1b shows a commercial Dimension 1200es BST 3D printer.

Experimental Methods The HandySCAN 700 is an entirely handheld laser scanning device, which applies targets from its reference system to the part. These targets achieve up to 20-micron accuracy specifications. It is feasible to collect data on both sides of the part by simply moving the device around the part. A computer system connected to the device helps to virtually spin the part itself to capture data from all sides. HandySCAN 700 laser scanning is unique to previous 3D laser scanning technologies. This device uses seven laser crosses, which give a precise scanning speed of 480,000 data points per second, gathering a considerable amount of data extremely quickly and consequently minimizing scan time. Not only a seven-laser cross-system but the device can also be optionally turned into a single-line mode by simply double-clicking a button on the back of the scanner. In a single-line system, only one laser line that does not require both lenses on the scanner is used, so it only requires whichever lenses are looking at the reflecting end. These single lines quickly reflect data from the tight areas of the part, such as holes. This study used the general area coverage, which is a quick way of gathering data, and the most accurate way to map out targets and tell

242

B. Subeshan et al.

the scanner where it is moving around in space, thereby also assisting with a depthof-field indicator. Holding the scanner too far away from the part indicates blue and holding the scanner too close to the part indicates red. Holding the scanner directly in the middle indicates green, which is the scanner’s optical scanning distance from the part. After scanning the entire part, VX Elements 7.0 software processes the part’s point cloud. The transformation of a 3D point cloud file to a CAD file is performed using Fusion 360 software, which analyzes the part geometrically and breaks it into meshes, thereby recognizing all the part’s critical geometrical features. Mesh is used for extracting the surfaces to generate the CAD model. This software supports critical features to modify the CAD model to look the same as the obsolete part. In this study, additive manufacturing was employed to manufacture the sandcasting pattern for the obsolete part. Due to accessibility, 3D printing was performed with polymeric materials, using the actual 3D part in hand to measure the geometric features and dimensions of the sand-casting pattern for the obsolete part. This technique is viable for complex parts with many geometric features. The casting pattern was fabricated using ABS plastic and PLA plastic on a Dimension 1200es BST 3D printer by incorporating the material extrusion additive manufacturing technique. This fabrication took 2 h and was followed by clearing the supports manually, using xylene for a better surface finish, and drying in an airstream for 30 min to improve the strength. The ABS plastic was closer to the more desired material used in the sandcasting process with its mechanical properties compared to the PLA plastic. The ABS P400 plastic used is a durable ABS-based material appropriate for concept models and testing of form, fit, and some function. It is impact-resistant, has a relatively high tensile strength, and is heat, scratch, and chemical resistant. The material has a relatively high thermal expansion for plastic and is lower in cost than most engineering thermoplastics. However, ABS plastic has limited weather resistance and is not very resistant to solvents. During 3D printing, the Dimension 1200es BST 3D printer operates with an extrusion system—an extrusion head with a nozzle to deposit build material onto a build platform. During material extrusion, a filament (ABS plastic) is fed through the extrusion head by a drive roll and past a heating element. The material liquefies and is then deposited by the nozzle onto a build platform. The build platform, which usually moves vertically, lowers itself automatically so the nozzle and can dispense a new material layer on top of the previous layer. To build each layer, the extrusion head deposits the layer’s outline first and then fills in each layer according to master dimensions set by the operator. As each layer is deposited, it fuses to the previous one, creating a solid model. This process is repeated until the final product is completed. While adding layers, material extrusion systems sometimes dispense support material along with build material. Support material occupies negative spaces and provides stability to disconnected part features, such as overhangs or holes, during the building process. Once the part is completed and used, the support material and break-away support systems are manually removed. To remove soluble support materials, the parts are submerged in a chemical bath or a solvent, such as a xylene, which quickly dissolves the support material. 3D printing machines exclusively use thermoplastics, a group of polymers, as build materials. Thermoplastics melt or soften quickly when heated, and they retain their desired shape upon cooling.

Reverse Engineering of Aerospace Components Utilizing Additive …

243

A large variety of thermoplastics are available for use in the 3D printing additive manufacturing industry. Many modern 3D printing machines still rely on thermoplastics, usually in pellets or filaments, to build parts today. The sand-casting process is relatively simple. A sand casting of the stainless-steel lever part from an aircraft control assembly was prepared using a 3D printed ABS pattern, applying material extrusion additive manufacturing technology. The pattern-development time was reduced by speeding up the sand-casting application. The reverse-engineered part should be resistant to chemicals in the sand, abrasion-resistant, and able to withstand compaction forces applied to pack the sand. Sand casting offers an economical process for the mass production of complex metal part structures. Stainless steel was the metal used for sand casting in this study. The progression starts with the design of the cast part and the metal delivery paths in the mold. The 3D printed ABS pattern is placed in a square box called a flask. Sand is then poured into the flask and placed firmly against the 3D printed ABS pattern. Clay, which is green sand and dry sand, holds the compacted sand together. Molten metal is poured into the mold over the sprue, flowing through the path onto the part cavity. The metal correspondingly fills the riser, a reservoir that continues to supply the part cavity as the metal cools and shrinks. The metal is then allowed to cool and solidify, and the sand is separated from the part. Subsequently, the manufactured part is removed and subjected to surface treatment processes such as sanding, sealing, and painting (optional) to improve its appearance and durability. By using additive manufacturing technology for pattern creation, sand casters use an effective prototyping approach, resulting directly in production.

Results and Discussion The CAD model obtained from laser scanning consists of many points on the surface, stitched together on the cloud data. It is very tedious to make a complete whole volume model due to unavoidable discontinuities among the stitched surfaces. However, some reverse engineering software facilitates surface evaluation and model inspection by measuring the difference between the surface model and cloud data, automatically filling and correcting gaps. Overall, the results of models created using 3D CAD modeling software are better than those obtained through reverse engineering. Reverse engineering is suitable when drawings are not available, and the part has a complex geometry. The lever part, when reversed, is a medium-complexity part due to the holes and fillets on the top portion and its curved geometry. CAD modeling with traditional CAD software typically takes at least a week. In contrast, it took only 3 h for laser scanning and 2 h for 3D printing this pattern for casting the part. Figure 2a–c show the results of the study. Figure 2a shows the obsolete physical part, and Fig. 2b shows the CAD model of the obsolete part, which was obtained using the HandySCAN 700 laser scanner. The pattern for sand casting of the reversed part was created using the material extrusion additive manufacturing technique. Figure 2c shows the 3D printed part, which is the pattern for sand casting of the reversed part

244

B. Subeshan et al.

Fig. 2 Reverse engineering process: a obsolete physical part, b CAD model of obsolete part, and c pattern for sand casting of the reversed part using material extrusion additive manufacturing

manufactured with the Dimension 1200es BST 3D printer, an affordable process with low energy usage. Figure 3 shows a visual comparison of the obsolete part and the 3D printed part, a fabricated pattern for sand casting. The 3D printed fabricated pattern for sand casting was made precisely like the actual obsolete part, which incorporated the dimensional analysis. All figures display a complete understanding of the reverse engineering progression. When the feature extraction of the five major characteristic designators on the obsolete part was tested, the design value of the hole distance varied, depending on the location. A comparison of the measured value and design value of the hole distance is shown in Table 1, which also includes the deviation of the pattern manufactured using the material extrusion system from the original obsolete part. Dimensional accuracy measures the fabricated pattern’s closeness for sand casting to the corresponding obsolete part. For dimensional comparison, Fig. 3 shows the three locations measured on both the obsolete and 3D printed parts fabricated for sand casting. Each dimension was measured twice. The absolute deviation was calculated relative to the obsolete part’s dimensions since the same model was used for all reverse

Fig. 3 Dimensional comparison of three locations measured on both obsolete part (left), and 3D printed part (right), a fabricated pattern for sand casting

Reverse Engineering of Aerospace Components Utilizing Additive …

245

Table 1 Comparison of measured values and characteristic designator values Reference location D1

Original obsolete part value (mm)

Fabricated part value (mm)

Deviation value (mm)

Error (%)

5

4.9

0.1

5

4.4

0.6

0.02 0.12

D2

60

62.4

2.4

4.00

60

54.4

5.6

9.30

D3

15

11.9

3.1

20.60

15

12.3

2.7

18.00

Fig. 4 Final reverse-engineered part using sand casting from 3D printed pattern

engineering processes. The percentage deviation for measuring the dimensional accuracy of the fabricated pattern for sand casting is defined as the percentage ratio of deviation from the obsolete part’s corresponding dimension. Dimensional errors may be introduced during laser scanning, point-data processing, solid modeling, and manufacturing pattern casting. Figure 4 shows the final sand-casted part, which is similar in geometry to the obsolete lever from the aircraft’s control assembly.

Conclusion There has been an increasing interest in reverse engineering of obsolete parts to replace parts that are worn out or whose original drawings (geometric, material, and manufacturing details) are not available. With reverse engineering, the obsolete part is measured to obtain new data and permit remanufacture of the part, which is possible in several ways with today’s modern techniques of using point cloud data from the obsolete part in digital CAD data files. This study showed how to overcome product development issues for traditional metal sand casting by combining reverse engineering and additive manufacturing technologies. This approach has been successfully demonstrated by conducting an industrial case study on an obsolete stainless-steel lever from an aircraft control assembly. Every step in the reverse

246

B. Subeshan et al.

engineering process, such as data capturing, refining the CAD model, and manufacturing the casting pattern, was achieved correctly. The comparative analysis of several routes for pattern fabrication yielded valuable data regarding dimensional accuracy and surface quality. Material extrusion additive manufacturing was used to make the part with sufficient quality to be used as a pattern for sand casting. By using additive manufacturing instead of traditional manufacturing methods, time and money can be saved. The proposed approach is particularly suited to the urgent replication of worn-out parts with complex shapes and those needed only once or in small quantities. Additive manufacturing is pertinent to the aerospace industry when the manufacturing of engineering products and tooling is involved. Innovation in additive manufacturing technology is expected to bring benefits in terms of accuracy and cost soon, especially for small and complex components. Acknowledgements The authors gratefully acknowledge Wichita State University for the technical and financial support for this research study.

References 1. Bhatti A, Syed NA, John P (2018) Reverse engineering and Its applications. In: Omics technologies and bio-engineering. Academic Press, pp 95–110 2. Buonamici F, Carfagni M, Furferi R, Governi L, Lapini A, Volpe Y (2018) Reverse engineering of mechanical parts: a template-based approach. J Comput Des Eng 5:145–159 3. Samuelson P, Scotchmer S (2002) The law and economics of reverse engineering. Yale Law J 111:1575–1663 4. Payal H, Tomer R (2020) Review on reverse engineering. J Crit Rev 7:1408–1412 5. Chilton J, Chuang CC (2017) Rooted in nature: aesthetics, geometry and structure in the shells of Heinz Isler. Nexus Netw J 19:763–785 6. Potabatti NS (2019) Photogrammetry for 3D Reconstruction in solidworks and its applications in industry (Doctoral dissertation) 7. Jiménez M, Romero L, Domínguez IA, Espinosa MDM, Domínguez M (2019) Additive manufacturing technologies: an overview about 3D printing methods and future prospects. Complexity. https://doi.org/10.1155/2019/9656938 8. Tong W, Chen M (2021) A sufficient condition for 3D typical curves. Comput Aided Geom Des 87:101991 9. Wang X, Bi Z (2019) New CAD/CAM course framework in digital manufacturing. Comput Appl Eng Educ 27:128–144 10. Subeshan B, Baddam Y, Asmatulu E (2021) Current progress of 4D-printing technology. Prog Addit Manuf 1–22 11. Asmatulu E, Alonayni A, Subeshan B, Rahman MM (2018) Investigating compression strengths of 3D printed polymeric infill specimens of various geometries. In: Nano-, Bio-, Info-Tech Sensors, and 3D Systems II, p 21 12. Baddam Y, Uddin MN, Don TN, Asmatulu E (2019) Integrating 4D printing processes into STEM education. ASEE Midwest Section Conference, Wichita, KS 13. Chen Z, Chen W, Guo J, Cao J, Zhang YJ (2018) Orientation field guided line abstraction for 3D printing. Comput Aided Geom Des 62:253–262

Part VI

Additive Manufacturing of Refractory Metallic Materials

Laser Metal Deposition of Nickel Silicide on S355 Structural Steel Mohammad Ibrahim, Tor Oskar Sætre, and Ragnhild E. Aune

Abstract Components with excellent oxidation, corrosion, and wear resistance are constantly employed in turbines, offshore systems, and power generation setups. The parts do often have to withstand extreme conditions during operation (e.g., high temperatures) making the choice of material a crucial step. Nickel silicides have proven to be possible candidate materials, but their inherent brittleness at room temperature prevents them from being produced by conventional processes such as casting and machining. However, with today’s advances in Additive Manufacturing (AM), novel materials can be produced by exposing metal powders to high-power lasers and electron beams. The goal of the present study is to explore the feasibility of using laser metal deposition (LMD) technique to produce nickel silicide beads on S355 structural steel. The microstructure and properties of the deposited beads are characterized using various analytical techniques (i.e., SEM, XRD, EDS, and hardness measurements), and changes in microstructure and composition are also discussed. Keywords Additive manufacturing · Laser metal deposition · Nickel silicide · Silicon · Parameters

Introduction Silicides of transition metals are the largest group of intermetallic compounds. They have an abundance of properties which make them ideal candidates for the mechanical components that require high abrasive and temperature resistance. However, their lack of ductility at room temperature combined with poor workability does not allow M. Ibrahim (B) · T. O. Sætre · R. E. Aune Department of Engineering and Science, Faculty of Engineering Sciences, University of Agder, Grimstad, Norway e-mail: [email protected] R. E. Aune Department of Materials Science and Engineering, Norwegian University of Science and Technology (NTNU), Trondheim, Norway © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_22

249

250

M. Ibrahim et al.

their production in bulk and consequently shuts down any chance of their use as structural or tribological components in industrial applications [1, 2]. Ordinary shapes can be created using techniques like spark plasma sintering (SPS) and hot pressing. However, these processes cannot be used to produce complex geometries. Additive manufacturing (AM) presents a solution to this problem it is known to fabricate complex shapes which cannot be accomplished using conventional techniques like casting or forming. There are already studies showing successful manufacture of silicides of Molybdenum [3] and Manganese [4] using AM techniques. There are many types or classes of additive manufacturing technologies. One of them is laser metal deposition (LMD). It was invented due to the research activities carried out at laboratories in New Mexico, along with the directed light fabrication and the famous laser engineered net shaping (LENS) method. These techniques are capable of building components from metals, alloys, or composites into the final desired shape [5]. The process takes place by constantly supplying raw materials directly to the area where the laser is focused; this causes these raw materials to melt and forms a melt pool at that spot. This melt pool solidifies when the laser beam moves away from that spot. The constant repetition of this process leaves a track of deposited material wherever the beam goes. It possesses the ability to use both wire fed and powdered raw materials [6]. LMD is unlike other classes of AM as it not only is able to manufacture of new 3D parts, but it also allows the repair of high valued parts. In conventional techniques, addition of parts or components onto an existing part takes place by joining them together using bolts, nuts, rivets, or welding. This gives rise to areas of high stress concentration where the joints exist due to conflict in properties. These areas are prone to phenomenon like cracking or other material failures [7]. LMD eliminates these issues as it builds upon the existing feature with metallurgical bonding with the added advantage of no extra weight and less material usage. Hence making this technique ideal for industries like aerospace and automobiles. When compared with other classes of AM, laser metal deposition is also known to have a higher build rate [8]. Another important requirement in these industries are functionally graded materials, which vary in composition and structure gradually over volume giving rise to corresponding changes in properties of the materials. If LMD is used to manufacture such parts, the energy-intensive and time-consuming conventional processes are no longer required because LMD has the flexibility to use more than one material at a time. Which makes it ideal for production of functionally graded parts [9, 10]. Despite having these advantages, there are challenges associated with AM in general and with LMD in particular which have limited their use in afore-mentioned industries to produce critically important parts. Some of the underlying physics of the LMD process are still to be understood completely as it is important that properties of the produced parts must be predictable and controllable. Parts produced from LMD are known to require post-processing steps like machining and polishing because of poor surface finish and heat treatment for stress relief [11, 12]. As the nozzle for feeding the raw material and laser source is mounted on a 4- or 5-axis robotic arm that deposits the material on the substrate, there is no longer a powder bed which keeps

Laser Metal Deposition of Nickel Silicide …

251

the part suspended during printing. Therefore, support structures are also required to provide stability [12]. There are quite a few processing parameters in LMD process which have varying effects on the properties that develop the printed parts. Some of them have more significant impact than the rest. Laser power is one such parameter; it has great influence on surface finish and microstructural properties. Its selected magnitude is dependent on the material being used since it is responsible for the melting of feedstock material [13]. Another important parameter is scanning velocity which determines the amount of time a laser interacts with the substrate and deposited material. If laser power and scanning velocity are not adjusted properly, it could lead to issues like incomplete melting, porosity, and vaporization of the deposited material [14]. The rest of the essential process parameters include powder and gas flow rates, laser beam diameter, and overlap percentage. Powder flow rate is measure of the amount of powder leaving the delivery nozzle in grams per unit time. It determines the density of the printed part and influences the way material distribution takes place [15]. Laser beam diameter corresponds to the spot of the laser at a given focal length. Laser spot size dictates the laser energy density as the spot size increases; there is decrease in the laser energy density if laser power remains constant [16, 17]. Hatch distance or overlap percentage is the percentage of the last printed track which will be covered by the next track that is to be printed. It is therefore necessary to select a proper overlap percentage so that inter layer and inter track porosity can be avoided [18, 19]. In this study, 4 single bead samples of nickel silicide are 3D printed on S355 construction steel using LMD process. The process parameters are varied in each of these samples to help determine the most optimum set of parameters for additive manufacturing of nickel silicide powders. These printed samples were analyzed using scanning electron microscope, electron dispersive spectroscopy, and X-ray diffraction. Their hardness was measured using Vickers hardness testing for this purpose.

Materials and Characterization The additively manufactured samples investigated in this study were printed from NiSi16 powder via laser metal deposition technique using DMG Mori Lasertec 65 3D system at Mechatronics Innovation Lab (MiL) in University of Agder on top of S355 structural steel substrate. The NiSi16 powder was produced by Phoenix Scientific Industries Ltd (PSI/MPP) using gas atomization technique. Following sets of parameters shown in Table 1 were used while printing these beads and then smaller pieces were sectioned for analysis. X-ray diffraction (XRD) was performed using the D8 Focus X-ray diffractometer located in the X-ray Diffraction Lab at Norwegian University of Science and Technology (NTNU), Trondheim. The diffractometer operates using a copper X-ray cathode tube with a wavelength of 1.54 Å operated at 40 kV and 44 mA. The beam

252

M. Ibrahim et al.

Table 1 Parameters for printed beads using LMD technique No

Printed beads

Laser power (W)

Deposition speed (m/s)

Feeding rate (g/min)

1

Sample 1P

1390

0.9

14

2

Sample 2P

1707

1

20

3

Sample 3P

2500

2.5

13

was aligned following placement of the samples on the diffractometer stage. A scan range of 10–90º was selected with a step size of 0.1. Spectral analysis and phase identification were performed using the software package Bruker AXS DIFFRAC SUITE EVA. Scanning electron microscopy (SEM) images were taken using a Low Voltage Field Emission Scanning Electron Microscope (LVFESEM), Zeiss Supra 55VP belonging to Electron Microscopy Laboratory of the Materials Department at Norwegian University of Science and Technology (NTNU), Trondheim. Secondary electron detector was used to take images and provide information on surface topography. Elemental analysis was performed via energy dispersive X-ray spectroscopy (EDS) using an EDAX detector. Hardness measurements were performed on a Innovatest–Vickers hardness tester Nova 240 under a load of 300 g for a dwell time of 10 s. Hardness measurements were done along the entire length of the printed bead except for the cracks to ensure uniformity.

Results and Discussion From the macroscopic overview of the samples in the Fig. 1, it is visible that the sample 1P possesses the most rough surface and the surface finish gets better as we

Fig. 1 Surface finish of printed beads of nickel silicide

Laser Metal Deposition of Nickel Silicide …

253

move to sample 2P and finally the best out of the three in sample 3P. This can have a direct relation to the laser power used while printing these samples. Higher laser power allows for better Marangoni flow, which is where the melt flows from regions of low surface tension to higher surface tension within the melt pool [20]. Generally blown powder laser deposition techniques have lower powder efficiency, but we can still observe the number of un-melted particles decreasing from sample 1P to sample 3P as we observe scanning electron images in Figs. 2, 3, and 4; this is again attributed to the increased laser power in sample 3P [21]. These partially melted particles can be a source of roughness increase and can also affect the wetting of the subsequent layer [22]. Another reason for bad surface finish can be the spatter observed on the walls of these printed beads. Most of the times this spatter is formed by the localized boiling of the metal melt droplets, and it’s known to form on the edges of the melt pool [23]. Sample 2P also shows some pores which are thought to be because of the trapped gas used during deposition [21] or the gas trapped in the particles during their manufacture via gas atomization. The earlier mentioned Marangoni effect is also responsible for the dilution of the substrate into the printed bead. In these samples, considerable amount of Iron (Fe)

(a)

(b)

Fig. 2 Scanning electron microscope image of printed sample 1P: a 15 × magnification b 35 × magnification

(a)

(b)

Fig. 3 Scanning electron microscope image of printed sample 2P: a 15 × magnification b 35 × magnification

254

M. Ibrahim et al.

(a)

(b)

Fig. 4 Scanning electron microscope image of printed sample 3P: a 15 × magnification b 35 × magnification

Table 2 Quantitative electron dispersive spectroscopy results of printed beads No

Printed beads

Iron content (weight%)

Silicon content (weight%)

Nickel content (weight%)

1

Sample 1P

20.28

22.89

56.82

2

Sample 2P

6.01

19.43

74.56

3

Sample 3P

4.75

19.62

75.64

made it into the printed bead. This is shown in Table 2 in the EDS results of printed beads. The higher amount of Iron in sample 1P can be because of the low deposition speed; this causes the laser to stay at a point for longer time period, which allows numerous melting cycles and hence higher dilution. Relatively lower Fe content is observed in sample 2P, which can be related to its high feeding rate due to which a cloud of NiSi16 powder is formed in front of the laser; this reduces the intensity with which the laser interacts with the steel substrate. This ultimately led to a smaller substrate melt pool with a higher powder content. If we observe the Fe content value in sample 3P, we can speculate that the low dilution was caused by the high deposition speed, which was responsible for the shallow penetration of the laser in the substrate [24]. Table 3 shows the results from EDS analysis of the cross-sections of printed beads from the top and bottom of the samples. These results support the earlier EDS analysis (shown in Table 2) from above printed samples. An obvious trend is seen where the amount of dilution decreases as the distance of the point of acquisition Table 3 Quantitative electron dispersive spectroscopy results from cross section of printed beads No Printed beads Iron content at peak point (atomic %) Iron content at base point (atomic %) 1

Sample 1P

28.63

34.12

2

Sample 2P

11.34

20.01

3

Sample 3P

2.15

8.12

Laser Metal Deposition of Nickel Silicide … Table 4 Width and height of single NiSi16 printed bead on steel substrate with different parameter sets

255

No

Printed beads

Width of printed bead (mm)

Height of printed bead (mm)

1

Sample 1P

2.67

0.56

2

Sample 2P

2.86

1.07

3

Sample 3P

2.61

0.76

increases from the substrate. Table 4 shows the width and height measurements of these printed samples. Sample 2P has the largest width and height due to a moderate deposition speed with a comparatively high feeding rate. This allowed the bead to expand and grow beyond the other two during solidification. X-ray diffraction results given in Fig. 5 show multiple phases identified in the process. The parent phase of the NiSi16 powder, Ni2Si, was detected in sample 1P and sample 3P where as sample 2P showed some presence of Manganese as well along with an Iron Nickel Silicon phase with varying quantities of elements. This Mn signal could have originated from close to the substrate as making sure that the signal comes from only the printed bead and not the substrate is quite difficult (Fig. 6). The SEM images in Figs. 2, 3, and 4 show cracks in all three samples. Although cracks can be attributed to the inherent brittle nature of NiSi, yet a closer observation reveals that the crack in sample 3P is larger than sample 2P which is larger than the crack in sample 1P. This trend can be associated with the increasing scan speeds in these samples [22]. Hardness value shown in Table 5 did not follow a trend in these samples. Sample 1P had the lowest hardness whereas sample 2P had the highest closely followed by sample 3P. Generally, hardness values are known to increase with increase in deposition speed [25].

Fig. 5 Cross-sectional schematic displaying points of acquisition for quantitative EDS data of iron content in the bead

256

M. Ibrahim et al.

Fig. 6 X-ray diffraction scan of printed samples on S355 construction steel, Black-Sample 1P, Red-Sample 2P, Sample 3P (Color figure online)

Table 5 Hardness values of LMD printed beads

No

Printed bead

Hardness value

1

Sample 1P

574.5 HV

2

Sample 2P

1130 HV

3

Sample 3P

1068 HV

Conclusion Further characterization and analysis are required to confirm the preliminary findings of this study. It is difficult to attribute a certain behavior to a change in one process parameter if the rest of the parameters are also changing so it will be beneficial to observe the change of one parameter at a time. Possibility of an inclusion element should be explored to reduce the inherent brittleness of the NiSi system. Dilution by Iron from the substrate seems like the most obvious matter of concern in this preliminary examination. So multilayer deposits in case of cladding operations can provide results with lower dilution. However, there is ample evidence available that suggests even diluted claddings perform better in corrosive environments than HVOF sprayed coatings [26]. The most certain conclusion that can be drawn from this study is that laser power is the most dominant parameter among the ones discussed here. Increase in laser power can result in reduction of conventional surface defects but the possibility of unconventional defects like spatter formation cannot be ignored. Acknowledgements This study was possible due to the funding from the Norwegian Research Council (NFR). The parts were printed in Mechatronics Innovation Lab (MiL) at University of Agder (UiA) by Morten Kollerup. Characterization took place at the Materials Department in Norwegian

Laser Metal Deposition of Nickel Silicide …

257

University of Science and Technology (NTNU). Constant technical support and expertise were provided by Professor Geir Grasmo and Dr. Naureen Akhtar at University of Agder (UiA).

References 1. Vasudevan AK, Petrovic JJ (1992) A comparative overview of molybdenum disilicide composites. Mater Sci Eng 155(1–2):1–17 2. Liu X, Yu L, Wang HM (2001) Synthesis of a nickel silicide-base composite coating on austenitic steel by laser cladding. J Mater Sci Lett 20(16):1489–1492 3. Hagihara K et al. (2017) Successful additive manufacturing of MoSi2 including crystallographic texture and shape control. J Alloys Compounds 696:67–72 4. Thimont Y et al. (2018) Thermoelectric higher manganese silicide: synthetized, sintered and shaped simultaneously by selective laser sintering/melting additive manufacturing technique. Mater Lett 214:236–239 5. Mueller B (2012) Additive manufacturing technologies–rapid prototyping to direct digital manufacturing. Assembly Autom 6. Li F et al. (2016) Microstructural study of MMC layers produced by combining wire and coaxial WC powder feeding in laser direct metal deposition. Opt Laser Technol 77:134–143 7. Mahamood RM (2018) Laser metal deposition process of metals, alloys, and composite materials. Springer 8. Marchese G et al. (2017) Characterization and comparison of Inconel 625 processed by selective laser melting and laser metal deposition. Adv Eng Mater 19(3):1600635 9. Mahamood RM, Akinlabi ET (2015) Laser metal deposition of functionally graded Ti6Al4V/TiC. Mater Design 84:402–410 10. Naebe M, Shirvanimoghaddam K (2016) Functionally graded materials: a review of fabrication and properties. Appl Mater Today 5:223–245 11. Rombouts M et al. (2013) Surface finish after laser metal deposition. Phys Procedia 41:810–814 12. Yadollahi A, Shamsaei N (2017) Additive manufacturing of fatigue resistant materials: challenges and opportunities. Int J Fatigue 98:14–31 13. Mahamood RM, Akinlabi ET (2014) Effect of laser power on surface finish during laser metal deposition process. In: Proceedings of the world congress on engineering and computer science 14. Senthilkumaran K et al. (2009) Influence of building strategies on the accuracy of parts in selective laser sintering. Mater Design 30(8):2946–2954 15. Brandl E et al. (2011) Deposition of Ti–6Al–4V using laser and wire, part I: Microstructural properties of single beads. Surface Coatings 206(6):1120–1129 16. Wolf M (2016) Improving the efficiency of the DMLD process: how particle size and laser spot size influence process quality and efficiency. Laser Technik J 13(4):32–34 17. Francis Z, Beuth J (2016) The effect of beam spot size on melt pool geometry in direct metal additive manufacturing processes. Mater Sci Technol 18. Francis L (2016) Materials processing: a unified approach to processing of metals, ceramics and polymers. Academic Press, San Diego 19. Schneider MF, Schneider MF (1998) Laser cladding with powder. Ph.D. thesis, University of Twente 20. Naesstroem H (2021) Phenomena in laser based material deposition. Ph.D. thesis, Luleå University of Technology 21. Selcuk C (2011) Laser metal deposition for powder metallurgy parts. Powder Metall 54(2):94– 99 22. Badiru AB, Valencia VV, Liu D (2017) Additive manufacturing handbook: product development for the defense industry. CRC Press, Boca Raton 23. Prasad HS, Brueckner F, Kaplan AF (2020) Powder incorporation and spatter formation in high deposition rate blown powder directed energy deposition. Additive Manuf 35:101413

258

M. Ibrahim et al.

24. Segerstark A, Andersson J, Svensson L-E (2017) Investigation of laser metal deposited Alloy 718 onto an EN 1.4401 stainless steel substrate. Optics Laser Technol 97:144–153 25. Zhang K et al (2014) Characterization of stainless steel parts by laser metal deposition shaping. Mater Des 55:104–119 26. Tuominen J et al. (2003) Microstructure and corrosion behavior of high power diode laser deposited Inconel 625 coatings. J Laser Appl 15(1):55–61

Refractory Metals—Some Historical Observations Jeffrey Wadsworth

Abstract In 1980, the Metallurgy Department of the Lockheed Palo Alto Research Laboratory, which was part of the Lockheed Missiles and Space Company, had tremendously exciting programs underway that required metallurgical insights. The Trident Missile program was an exemplar; the so-called Post Boost Control System (PBCS), operated for a few minutes (technically in space, to position nuclear warhead payloads) at a temperature of 1650 °C; the system used Mo, Ta, Nb, and W alloys, as well as high temperature coatings to execute its mission. Some of the issues that arose in that mission will be described, including impacts on part of the USA Nuclear Submarine Fleet due to the room temperature fracture of Mo fasteners upon refurbishment of the PBCS. This led to studies on the role of oxygen in embrittlement of Mo alloys. The advent of techniques such as in situ Auger Spectroscopy fracture was key in understanding the true origins of failure, but so were fundamental thermodynamics. Other curious decisions had been made regarding Nb alloys. In addition, the political role of key decision makers and how they influenced technical options will be described. Keywords Molybdenum · Embrittlement · Auger spectroscopy

Introduction In the early 1980s, research was underway into refractory metals and alloys at Lockheed Missiles and Space Company (LMSC) at their Palo Alto Research Laboratory (LPARL) which subsequently became the Research and Development Division (RD&D) of LMSC. The problems facing Lockheed centered on the mechanical behavior of refractory metal alloys at both room and high temperatures and in some cases on their environmental resistance to complex atmospheres. These alloys included Molybdenum, Columbium (Niobium in Europe), Tantalum, and Tungsten. One of the principal J. Wadsworth (B) Montecito, CA, USA © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_23

259

260

J. Wadsworth

products at LMSC at that time was the Trident Missile. This was a very complex enterprise with many challenges, but metallurgically the areas of concern were the mechanical properties of the alloys at both their operating temperatures, often at 1650 °C which is above the melting point of steel, and just as important, at room temperature where manufacturing issues could arise. In this paper, some of the research findings will be summarized and it is proposed that they will be of importance in attempts to manufacture these materials by 3D printing.

Background Entering the 1980s, the bulk of data regarding the properties of refractory metals and alloys had been funded by the Air Force. This was understandable as the USA recognized the need to win a space and missile race against known potential enemies. This work was largely empirical and consisted of massive amounts of raw data, often exploring the range of compositions and creep behavior of refractory metal alloys at various strain rates and temperatures. The results were to be found in very thick books, literally held together with copper rivets. This work was admirable and used hundreds of creep machines at various companies, including my former company Battelle (today there is a single creep machine there, maybe). The data generated by these studies is incredibly important although I suspect it is hard to find copies of the books. By today’s standards, the data are very empirical and relatively speaking not fully investigated or understood against contemporary models of deformation. An early modern study was the creep of Hot Isostatically Pressed (HIPPED) C103 (Nb-10Hf-1Ti) alloy [1, 2]. This was an example of the attempts to understand the behavior of wrought versus power metallurgy processing routes. Subsequently, it was possible to identify the creep behavior of C103 as that of a Class 1 solid solution, i.e., a glide-controlled alloy [3]. Another example was the reevaluation of the role of carbide formers, such as Hf in strengthening Mo and W alloys [4, 5]. In all these examples, contemporary studies of high temperature behavior allowed a reevaluation of properties.

Background on Molybdenum and Its Alloys Molybdenum has a very high melting point of 2623 °C. As a result, Mo was the subject of intense work in the 1950s and 1960s for Space and Nuclear applications. Deleterious effects of oxygen on the properties of Mo had long been recognized at a macro-level. In fact, to overcome this embrittlement problem, a program called INFAB (Inert Fabrication) had been created to process Mo alloys in an inert Argon atmosphere. Workers wore “Space Suits” inside a facility with Mo processing equipment. But a tragic death due to asphyxiation occurred and the program was terminated

Refractory Metals—Some Historical Observations

261

after a year. It should be noted that to avoid oxygen embrittlement of Mo an environment of 10–9 of oxygen is needed, which is several orders of magnitude higher than is available in commercial Argon. Although processing in Argon eliminated the “smoking” effects of processing Mo because of MoO3 formation, it did not eliminate embrittlement. Processing of Mo evolved to accept surface oxidation as inevitable and to recognize the need to perform surface removal. Nonetheless, this enormous body of work led to the establishment of several major alloys such as Unalloyed Mo; TZM (Ti–Zr–Mo); TZC (Ti–Zr–C–Mo); and Mo–Re (45–55%Re).

The Role of Oxygen at the Atomic Level As a result of the attempts to control oxygen in Mo, an assumption was developed that any brittle behavior in Mo was the result of oxygen segregation to grain boundaries. But this is far from the case as will be shown. It turns out that the C:O ratio is critical in predicting Mo embrittlement by oxygen, and in commercial practice, the amounts of C and oxygen are part of the composition requirements. The most elegant scientific work in this area is that of Kumar and Eyre and their findings on the necessary C:O ratios to avoid embrittlement are enlightening [6]. It also matches the empirically developed, commercial practice regarding C and oxygen levels mentioned above. Kumar and Eyre developed bamboo structures of Mo alloys and varied the C:O ratios dissolved in the Mo at high temperatures. They then fractured the samples at room temperature using in situ Auger Spectroscopy and determined that an atomic ratio of C:O of 2:1 was necessary to prevent grain boundary segregation of oxygen. Examples of the role of oxygen in Mo Alloys include Welding of Molybdenum, the Transverse Properties of Molybdenum Bar stock, and Dilute Molybdenum-Rhenium Alloys.

Welding of Molybdenum Mo alloys were welded using Electron Beam, Lasers, and TIG [7]. In all cases, this resulted in very large grain size differences in the Heat Affected Zone (HAZ). Tests in the transverse (or isostress) direction exhibited zero ductility. Traditionally, oxygen embrittlement was assumed to be the cause. But this study showed that oxygen could be eliminated as the reason using in situ Auger Electron Spectroscopy. The origin of the brittle behavior was that deformation was confined to the HAZ, and therefore, the real strain rates were much higher than nominal (and the strong grain size dependency in Group VI amplified this result). Thus, strain rates of 10–5 /sec were effectively significantly increased. By reducing the strain rate so that the real strain rate in the HAZ was nominal, local ductile behavior was observed. (Note: Even in the ductile necked region, the fracture surface is a mix of grain boundaries and cleavage, which is helpful in Auger studies since the cleavage areas are an oxygen

262

J. Wadsworth

free surface for comparison with the grain boundaries.). This observation suggests possible solutions: remove the grain size dependency by recrystallization (but this reduces strength) or add strengthening elements to the welding alloys.

Transverse Properties of Molybdenum Bar Stock A major study area involving Mo bar stock was in the Post Boost Control System (PBCS) of the Trident Missile System. The PBCS operates when solid Propellants in Ti cases ignite and gas at 1650 °C flows through a Ta-10 W alloy plumbing system to Cb (C103) valves that provide positioning and direction of the payload. The plumbing system is connected by unalloyed Mo fittings. These fittings consist of Mo rings cut from cylindrical bar stock that has been hollowed out. The Mo rings seal the fittings by deformation at room temperature during which the flat rings deform to a conical section. This causes stresses and deformation in the transverse direction of the bar. The bars are the product of a process involving initial arc melting of Mo during which C is added to tie up Oxygen. Although this is effective, any excess C causes the formation of Molybdenum carbides. In commercial practice, the subsequent processing of Mo bars involves high temperature extrusion, rotary swaging, and round rolling. Because the molybdenum carbides do not deform as easily as the Mo matrix, stringers of carbides and associate cracks are present [8–10]. An associated strong BCC fiber texture develops which leads to unusual transverse microstructures; instead of the expected small uniform grains, wide ribbons appear and are conspicuous in transverse sections but are not apparent in longitudinal microstructures. The texture does not have an impact on flow properties [11]. The stringers of carbides and cracks, however, cause brittle failure of the rings in the transverse direction. In practice, the Post Boost Control Systems in Trident Missiles are refurbished periodically. New Mo rings are needed to reconstitute the PBCS and are supplied in bags of 20 or so. Brittle fractures are audible during installation, and any rings in the bag containing a failed ring are discarded. Consequently, rejection rates were very high. At one point, the level of failure rates became unacceptable. Our investigation concluded that the acceptance tests were inappropriate because tensile tests from bars examined in the longitudinal direction were irrelevant relative to the stress state undergone in the sealing step. Tensile tests in the transverse direction were relevant but had never been done. The ring failure rates were inevitably high, and minor statistical variations in composition or processing could take the failures to very high levels. (It should be noted that the major applications of Mo bars were for electrodes in glass making so the application of their use described here was unique.) A simple solution was developed: an approach of cutting bar sections, and heating, and upset forging effectively both reduced the carbide/crack stringers, but more importantly turned the microstructure through 90° so that any remaining cracks were no longer normal to the stress state during sealing. Rings were then machined from these forged pieces. At the time of the C4 missiles, in a major meeting, this new process was rejected. The feedback was: “This worked before, make it work

Refractory Metals—Some Historical Observations

263

again”. Well, it didn’t work before—clearly 40–80% failure rates are not acceptable. BUT: It is dangerous to make changes in complex systems that are built on multiple subsystem tests. The transition from the Trident C4 to D5 starting in 1990 allowed new processes to be included because the subsystem tests were now performed and this new process for manufacturing Mo rings was introduced. Not a single failure of a ring has occurred since.

Dilute Molybdenum-Rhenium Alloys A paper was published, along with press coverage, that new dilute Mo-Re alloys showed enhanced ductility. The data suggest maximum ductility at several percent additions of Re. However, at zero percent Re, Mo alloys showed very high ductility. Reevaluation of the data revealed that all low-ductility, dilute Mo-Re alloys, had poor C:O ratios [12]. So, the apparent maximum in ductility in the dilute alloys was a specious result.

Implications for 3D Printing of Mo Alloys In 3D printing machines, the oxygen partial pressure will not be low enough to avoid oxygen absorption. Therefore, C will need to be added so a 2:1 atomic ratio with oxygen exists in the printed materials. But then beware the consequential effects of Molybdenum carbides on fracture resistance. Also, care must be taken to avoid stress localization due to grain size differences. Alloying roles of Ti, Zr, (Re) should be considered.

The Need for Columbium Alloy Coatings As described, the Post Boost Control System (PBCS) of the Trident Missile System consists of gas flow at 1650 °C through Ta-10 W alloy plumbing to Columbium (C103) Valves that provide positioning and direction. The C103 Valves had historically been coated with a complex silicide, processing that required two high temperature treatments. Examination of a retrieved C103 valve showed no internal oxidation which raised the question of why was a coating required? The testing protocol at Ford Aerospace was reviewed and the Oxygen/Carbon environment in the testing protocol was oxidizing and quite unlike that produced in the operation of the PBCS. It was determined that coatings were not needed for the real environment. Going forward these complex silicide coatings were cancelled at significant cost savings. Basic thermodynamics evaluations and an understanding of the different testing protocols would have been sufficient to avoid this coating requirement.

264

J. Wadsworth

Other Refractory Metal Topics Various reviews were published after the above studies [13–16], including topics such as “Weld embrittlement of Silicide Coatings on Ta-10 W alloys” [17], “Interfacial reactions between refractory metals and other materials” [18, 19], “A review of the properties of Tungsten alloys” [20], “A study of the properties of Mo at elevated temperatures” [21], and “A history of the evolution of structural materials” [22]. Acknowledgements Much of the work summarized in this paper was possible because of the leadership and contributions of the (Late) Charles M. Packer. Roger Perkins was also a key contributor as were the co-authors of the cited papers.

References 1. Wadsworth J, Roberts CA, Rennhack EH (1982) Creep behavior of hot isostatically pressed niobium alloy powder compacts. J Mater Sci 17: 2539–2546 2. Himmelblau CD, Kibrick M, Runkle J, Joshi A, Wadsworth J, Moncur J (1984) Mechanical properties of HIP columbium C-103 Alloy. In: Nayar HS et al. (eds) Progress in powder metallurgy (1983 Ann. Powder Metallurgy Conf. Proc.), vol 39. Metal Powder Industries Federation. Princeton, New Jersey, pp 525–542 3. Wadsworth J, Kramer PA, Dougherty SE, Nieh TG (1992) Evidence for dislocation glide mechanism in niobium base alloys. Scripta Metall Mater 27:71–76 4. Wadsworth J (1983) A reevaluation of the mechanical properties of molybdenum- and tungstenbased alloys containing hafnium and carbon. Metall Trans 14A:285–294 5. Wadsworth J, Klopp WD (1985) The influence of the atomic ratios of hafnium to carbon on high temperature strength in molybdenum and tungsten alloys. In: Miska KH et al (eds) Physical metallurgy and technology of molybdenum and its alloys. AMAX Metals. Ann Arbor, Michigan, pp 127–133 6. Kumar A, Eyre BL (eds) Grain boundary segregation and intergranular fracture in molybdenum. Proc R Soc Lond A370:431–458 7. Wadsworth J, Morse GR, Chewey PM (1983) The microstructure and mechanical properties of a welded molybdenum alloy. Mater Sci Eng 59:257–273 8. Wadsworth J, Packer CM, Chewey PM, Coons WC (1984) A microstructural examination of the origin of brittle behavior in the transverse direction in mo-based alloy bars. Metall Trans 15A:1741–1752 9. Wadsworth J, Packer CM, Coons WC (1985) The mechanical properties of molybdenum and TZM bar stock. In: Miska KH et al (eds) Physical metallurgy and technology of molybdenum and its alloys. AMAX Metals. Ann Arbor, Michigan, pp 13–19 10. Wadsworth J, Nieh TG, Henshall CA, Coons WC, Stephens JJ (1988) The role of oxygen on the mechanical behavior of molybdenum alloys. In: El-Genk MS, Hoover MD (eds) Space nuclear power systems 1987. Orbit, Malabar, Florida, pp 313–318 11. Oyama T, Wadsworth J (1987) Anisotropy of microstructure and strength in fiber textured molybdenum alloys. Textures Microstruct 7:1–10 12. Wadsworth J, Nieh TG, Stephens JJ (1986) Dilute Mo-Re alloys—a critical evaluation of their comparative mechanical properties. Scripta Metall 20:637–642 13. Wadsworth J, Nieh TG, Stephens JJ (1988) Recent advances in aerospace refractory metal alloys. Int Mater Rev 33:131–150

Refractory Metals—Some Historical Observations

265

14. Nieh TG, Wadsworth J (1994) Recent advances and developments in refractory alloys. In: Briant CL, Petrovic JJ, Bewlay BP, Vesudevan AK, Lipsitt HA (eds) High temperature materials, materials research society symposium, vol 332. Materials Research Society. Pittsburgh, Pennsylvania, pp 315–327 15. Wittenauer J, Nieh TG (1993) Strengthening and toughening in refractory metal alloys. In: Stoloff NS, Duquette DJ, Giamei AF (eds) Critical issues in the development of high temperature materials. The Minerals, Metals & Materials Society, Warrendale, Pennsylvania, pp 189–202 16. Wadsworth J, Wittenauer JP (1994) The history of the development of molybdenum alloys for structural applications. In: Dalder ENC, Grobstein T, Olsen CS (eds) Evolution of refractory metals and alloys. The Minerals, Metals & Materials Society, Warrendale, Pennsylvania, pp 85–108 17. Wadsworth J, Packer CM (1983) Weld embrittlement in a silicide-coated tantalum alloy. Int J Refractory Hard Metals 2:164–169 18. Joshi A, Hu HS, Wadsworth J (1990) Interfacial reactions of refractory metals niobium and tantalum with ceramics silicon carbide and alumina. In: Pantano CG, Chen EJH (eds) Tailored interfaces in composite materials, MRS Symp. Proc., vol 170. Materials Research Society, Pittsburgh, Pennsylvania, pp 149–154 19. Joshi A, Hu HS, Jesion L, Stephens JJ, Wadsworth J (1990) High temperature interactions of refractory metal matrices with selected ceramic reinforcements. Metall Trans 21A:2829–2837 20. Wittenauer JP, Nieh TG, Wadsworth J (1992) Tungsten and its alloys. Adv Mater Process 142(3):28–37 21. Nieh TG, Wadsworth J (1998) Improved understanding of the mechanical behavior of molybdenum alloys at elevated temperatures. In: Crowson A, Chen ES, Shields JA, Subramanian PR (eds) Molybdenum and molybdenum alloys. The Minerals, Metals Materials Society, Warrendale, Pennsylvania, pp 89–98 22. Wadsworth J (2007) The evolution of technology for structural materials over the last 50 years. JOM 59(2):41–47

Part VII

Additive Manufacturing: Beyond the Beam III

Microstructure Evolution and Mechanical Properties of Friction Stir Metal Deposited SS304 Nikhil Gotawala, Neeraj Kumar Mishra, and Amber Shrivastava

Abstract The objective of this work is to analyse the microstructure and its effect on mechanical properties of friction stir deposited SS304. Friction stir deposition is a solid-state additive manufacturing technique, where the material does not melt during the process. This process has shown potential for applications like large scale repairing of steel infrastructure. In this work, friction stir deposition of SS304 is performed at 1000 rpm rotation speed, 0.75 mm/s plunge feed rate, and 4 mm/s forward feed rate. The results suggest that a layer of about 0.5 mm thickness is deposited per pass. The continuous dynamic recrystallization and discontinuous dynamic recrystallization occurred during friction stir metal deposition. The twin boundaries observed in the as-received material are also reflected in the deposited material. The average tensile strength and ductility along the longitudinal direction of the deposited region are 647 ± 63.5 MPa and 70.7 ± 27.2%, respectively. The interface between the successively deposited regions is the weakest region of the deposition. This work shows the feasibility of the direct deposition of steel by friction stir metal deposition. Further, the challenge posed by the interface of the regions deposited by successive rods is highlighted. Keywords Friction stir metal deposition · Dynamic recrystallization · Microstructure evolution

Introduction Stainless steel has good corrosion resistance properties and high tensile strength. Due to these properties, steel finds applications in the petrochemical, automobile, aerospace, and many other industries. However, the severe environment of usage continuously degrades stainless steel over time. Therefore, stainless steel components may need to be repaired once in a while. Friction stir metal deposition (FSMD) N. Gotawala · N. K. Mishra · A. Shrivastava (B) Department of Mechanical Engineering, Indian Institute of Technology Bombay, Powai, Mumbai, MH 400076, India e-mail: [email protected] © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_24

269

270

N. Gotawala et al.

is a solid-state process which has shown potential to repair such components [1]. There are some similarities FSMD and friction stir welding (FSW) [2]. In both the processes, severe plastic deformation takes place at high temperatures. However, during FSMD, a consumable rod is used which is deposited on the substrate material, whereas during FSW, a non-consumable tool is used to stir the regions of workpieces and create the joint. Previously, dynamic recrystallization is noticed in the deposited material, due to severe plastic deformation at high temperatures [3, 4]. The dynamic recrystallization leads to a refined grain structure in the deposited region [5, 6]. The mechanical properties of the deposited material depend on this refined grain structure. This makes the analysis of microstructure evolution in friction stir metal deposition very interesting. Wedge et al. deposited Al2219 using additive friction stir metal deposition (additive-FSMD) and analysed the microstructure of the deposited material. The deposited material consisted of grains 5.5 times smaller the grains of the wrought material. The precipitate dissolution was also noticed in the deposited material, which reduced the fatigue life [7]. Similarly, Avery et al. investigated the microstructure and mechanical properties for the deposited region after additive-FSMD of the Al7075. Similar to the previous study, grain refinement with precipitate dissolution was observed in the deposited region [8]. Similar observations have been reported for other alloys of aluminium upon additive-FSMD [3, 5–8]. Very limited work is available on the additive-FSMD of Inconel [4] and copper [9]. However, to the author’s knowledge, there is no work available for the additive-FSMD of steels. The friction surfacing is attempted for the deposition of stainless steel [10–12], which involves the deposition of a single layer of material on the substrate. Presently, the multi-layered deposition of steels is primarily explored with wire arc additive manufacturing and laser-based additive manufacturing. Wang et al. deposited SS304 using direct energy deposition additive manufacturing and analysed the resulting microstructure and mechanical properties. The deposited region consisted of a heterogeneous microstructure with columnar grain growth towards the build direction [13]. Lima and Sankare also used laser-based additive manufacturing for the deposition of SS316 and reported the formation of δ-phase in the deposited region [14]. Chen et al. fabricated SS316 deposition by gas metal arc additive manufacturing and observed dendritic grain structure of austenite and columnar grain structure with σ and δ phases. The presence of σ phase in the deposited region deteriorated the ductility [15]. Wang et al. also found the presence of σ and δ phases upon wire arc additive manufacturing of SS316 [16]. This suggests that the liquification and solidification during wire arc additive manufacturing and laser-based additive manufacturing lead to the formation of σ and δ phases. The present work focuses on the analysis of microstructure and mechanical properties upon the friction stir metal deposition of SS304.

Microstructure Evolution and Mechanical Properties …

271

Experimental Work Figure 1a, b show the schematic diagram of FSMD of SS304 and an image of the deposited SS304, respectively. BD, LD, and TD are build direction, longitudinal direction, and transverse direction, respectively. The friction stir metal deposition was performed on a 3-axis CNC milling machine (LMW LV55). As Fig. 1 shows SS304 rod was used for deposition. The SS304 rods of 16 mm diameter and 150 mm length were used for this purpose. About 100 mm of each rod was deposited, before replacing the same with a new rod. The mild steel block was used as substrate. Initially to encourage the heat generation, the SS304 rod was plunged into the substrate up to 0.2 mm at 2000 rpm rotational speed and 5 mm/min feed rate. After initial plunge, a dwell time of 5 s was provided, allowing the temperature to build up. After dwell, the rotational speed of SS304 rod was lowered to 1000 rpm and friction stir deposition was performed with 45 mm/min plunge feed rate (towards substrate along BD) and 360 mm/min forward feed rate (along LD), over a total travel length of 40 mm (Fig. 1a). Accounting for the radius of the rods, the final build length and height were 50 mm (Fig. 1b) and 30 mm, respectively. The average height of deposition achieved per pass is 0.55 mm. To analyse the microstructure in the deposited region, the cross-section of the deposited region was machined using wire EDM. Further, to analyse the mechanical properties of the deposited region, the tensile specimens were machined along longitudinal and build directions. The gauge length and width of the tensile specimens were 7.5 mm and 1.5 mm, respectively. MTS landmark servohydraulic test system was used for tensile testing. The cross-section of the deposited material was mechanical polished with grit papers up to 2500 grit size. Then, the specimen was electropolished with Buehler Electromet 4 at 20 V voltage, 15 s time, and −15 °C temperature. The concentration of methanol, perchloric acid, and butanol in the

Fig. 1 a Schematic diagram of friction stir metal deposition of SS304 and b image of deposited SS304

272

N. Gotawala et al.

electrolyte was 70%, 20%, and 10%, respectively. Electron backscattered diffraction (EBSD) scan was carried out on Zeiss Gemini SEM 300 with Oxford EBSD detector. The scan size and step size for EBSD scan were 400 μm × 400 μm and 0.5 μm, respectively. Aztec hkl software was used to post-process the data. XRD 2 theta scan (between 30° and 100°) was also performed for the deposited region.

Results and Discussion Microstructure of Deposited SS304 Figure 2 shows the microstructure of the as-received SS304 rod. The as-received SS304 consists of equiaxed grains with twin boundaries. In SS304, the twin boundaries are present with crystallographic definition of (60°, ). It can be noticed that the twin boundaries are homogeneously distributed in the microstructure of the as-received SS304. The misorientation angle profile of the as-received SS304 also shows a higher fraction of twin boundaries, with the maximum fraction near 60° misorientation angle (Fig. 5). The average grain size in the as-received SS304 is 3.63 μm. As discussed previously, FSMD of SS304 was performed by successive deposition of layers over the previously deposited layers. In order to analyse the effect of this deposition pattern, the cross-section of the deposited region was studied with an optical microscope. The micrograph in Fig. 3 shows the optical image of the crosssection of deposited SS304. Figure 3 shows two successive regions with multiple Fig. 2 Micrograph of the as-received SS304

Microstructure Evolution and Mechanical Properties …

273

Fig. 3 Optical image of a cross-section of deposited SS304

layers of deposited material. Each region corresponds to the material deposited by a single rod over multiple (8) passes/layers. The thickness of the layers in the deposited region varies from bottom to top. Within each region, thicknesses of the bottom and top layers are the maximum and minimum, respectively. The width of the layers in a region also varies from top to bottom, such that minimum width at the bottom layer and maximum width at the top layer. The change in width from across the layers is attributed to the flash formation during FSMD. The flash formation during FSMD would depend on the viscoplastic boundary layer between the consumable rod and previously deposited region [17]. As the FSMD process progresses, the viscoplastic boundary layer temperature increases due to heat accumulation over previous passes/layers. This softens the material and increases the flash formation. Therefore, the deposition thickness of material near the centre decreases and flash formation at the periphery increases. This led to minimum flash formation for the bottom passes and increases for the later passes, for each successive region. Figure 4 shows the microstructure of the deposited region at location A and location B in Fig. 3. Figure 5 shows the XRD 2 theta scan for the SS304 deposited region, which indicates the presence of only austenite phase. Therefore, the σ and δ phases are not found in the deposited region, which are observed previously in deposited regions from laser-based additive manufacturing [14] and wire arc additive manufacturing [15, 16]. To capture the subgrain formation, the boundaries with misorientation angle between 2° and 10° are shown with red line. As Fig. 4 shows, the grain size varies, such that the finer grains are surrounding the coarser grains. The subgrain boundary formation suggests that the fine grains formed because of continuous dynamic recrystallization (CDRX) in the deposited region. The CDRX in the deposited region expected to be driven by the shearing of material in the viscoplastic boundary layer due to rotation of the consumable rod. The high-temperature plastic deformation in the viscoplastic boundary layer would lead to subgrain boundaries, which upon further deformation evolve into the grain boundaries. Apart from that, there are coarse grains without any subgrain boundaries. The bulging of grain boundaries for the coarse grains is observed. The bulging of grain boundaries indicates discontinuous dynamic recrystallization (DDRX) [18]. This suggests that DDRX

274

N. Gotawala et al.

Fig. 4 Microstructure of SS304 deposited region at a Location A and b Location B in Fig. 3

Fig. 5 Misorientation angle distribution of a as-received material and b deposited region at Location A

also occurred in the deposited region. Therefore, the fine and coarse grains formed in the deposited region due to CDRX and DDRX in the viscoplastic boundary layer, respectively. From Fig. 4a, the presence of interface suggests that all the layers are diffusion bonded with each other, due to high normal pressure during friction stir metal deposition (Fig. 6). Figure 5 shows the misorientation angle distribution of the as-received material and deposited region at Location A. As previously mentioned, twin boundaries were uniformly present in the as-received material. A significant fraction of grain boundaries has 60° misorientation angle (Fig. 5a). However, the distribution of misorientation angle for deposited region is quite different from the as-received material (Fig. 5b). This is due to CDRX and DDRX during FSMD. As Fig. 5b shows, the

Microstructure Evolution and Mechanical Properties …

275

Fig. 6 XRD 2 theta scan for SS304 deposited region

grain boundary misorientation angle distribution is relatively uniform between 10° and 59°. However, there is a sharp rise near 60° misorientation angle. This suggests that some of twin boundaries remain stable after the friction stir metal deposition. The average misorientation angles for as-received material and deposited region at Location A are 47.7° and 35.4°. This reduction in average misorientation angle between grain boundaries is attributed to the DDRX and CDRX during FSMD.

Mechanical Properties of Deposited SS304 Figure 7 shows the stress–strain curves from tensile testing of three specimens from deposited regions. Specimens 1 and 2 were prepared from the region near the top Fig. 7 Stress–strain curves of tensile specimens along longitudinal direction

276 Table 1 Comparison of tensile strength deposited region with previous work

N. Gotawala et al. Material

Tensile strength

Ductility

Present work

SS304

647 ± 63.5 MPa

70.7 ± 27.3%

Wang et al. [13]

SS304

609 ± 18 MPa

48.2 ± 2.5%

Wang et al. [16]

SS316

553 ± 2 MPa

-

surface, and specimen 3 was prepared to include the interface between the two topmost successive regions. All three specimens in Fig. 7 were along the longitudinal direction. The tensile strength and ductility of specimens 1 and 2 are considerably higher compared to specimen 3. This suggests that the interface between the successively deposited regions leads to the weakness and sufficient diffusion is not achieved at the initial layer (of successive region). This can be due to lower temperature in the viscoplastic boundary layer upon introducing a new rod to the previously deposited region. As a result, any considerable tensile strength and ductility is not achieved along the build direction, at present. The tensile strength of specimen 1, specimen 2, and specimen 3 is 689 MPa, 678 MPa, and 574 MPa, respectively. Similarly, the ductility of specimen 1, specimen 2, and specimen 3 is 92%, 80%, and 40%, respectively. Table 1 compares the average tensile strength achieved in deposited region against the tensile strengths along longitudinal direction reported previously for wire arc additive manufacturing. It can be noticed that the tensile strength and ductility along longitudinal direction achieved compare well against the same from a different additive manufacturing technique (Table 1). This can be attribute to the absence of σ and δ phases in the deposited region, which was observed in the previous works [15, 16].

Conclusions The present work analyses the microstructure and mechanical properties of friction stir metal deposited SS304. The friction stir metal deposition of SS304 was carried out at 1000 rpm rotational speed, 45 mm/min plunge feed rate, and 360 mm/min forward feed rate. The deposition consists of successively deposited regions along the build direction, which correspond to each SS304 rod. These regions consist of multiple layers, which correspond to the number of passes and the layer thickness reduces from bottom to top within each successively deposited region. The continuous dynamic recrystallization and discontinuous dynamic recrystallization occurred during friction stir metal deposition, due to high-temperature plastic deformation in the viscoplastic boundary layer. The twin boundaries observed in the as-received material are also reflected in the deposited material. The average tensile strength and ductility along the longitudinal direction of the deposited region are 647 ± 63.5 MPa and 70.7 ± 27.2%, respectively. The interface between the successively deposited regions is the weakest region of the deposition. This work shows the feasibility of

Microstructure Evolution and Mechanical Properties …

277

the direct deposition of steel by friction stir metal deposition. Further, the challenge posed by the interface of the regions deposited by successive rods is highlighted. Acknowledgements The authors gratefully acknowledge the partial support of this work by the Science & Engineering Research Board, Department of Science & Technology, Government of India (File no. ECR/2017/000727/ES), Department of Mechanical Engineering, Microstructural Mechanics and Microforming Lab, and Machine Tools Lab at Indian Institute of Technology Bombay, Mumbai.

References 1. Heidarzadeh A, Mironov S, Kaibyshev R, Çam G, Simar A, Gerlich, Withers PJ (2020) Friction stir welding/processing of metals and alloys: a comprehensive review on microstructural evolution. Progress Mater Sci 100752 2. Khodabakhshi F, Gerlich AP (2018) Potentials and strategies of solid-state additive friction-stir manufacturing technology: a critical review. J Manuf Processes 36:77–92 3. Phillips BJ, Avery DZ, Liu T, Rodriguez OL, Mason CJT, Jordon JB, Allison PG (2019) Microstructure-deformation relationship of additive friction stir-deposition Al–Mg–Si. Materialia 7:100387 4. Avery DZ, Rivera OG, Mason CJT, Phillips BJ, Jordon JB, Su J, Allison PG (2018) Fatigue behavior of solid-state additive manufactured inconel 625. JOM 70(11):2475–2484 5. Rivera OG, Allison PG, Brewer LN, Rodriguez OL, Jordon JB, Liu T, Hardwick N (2018) Influence of texture and grain refinement on the mechanical behavior of AA2219 fabricated by high shear solid state material deposition. Mater Sci Eng, A 724:547–558 6. McClelland Z, Avery DZ, Williams MB, Mason CJT, Rivera OG, Leah C, Hardwick N (2019) Microstructure and mechanical properties of high shear material deposition of rare earth magnesium alloys WE43. In: Magnesium technology 2019. Springer, Cham, pp 277–282 7. Anderson-Wedge K, Avery DZ, Daniewicz SR, Sowards JW, Allison PG, Jordon JB, Amaro RL (2021) Characterization of the fatigue behavior of additive friction stir-deposition AA2219. Int J Fatigue 142:105951 8. Avery DZ, Phillips BJ, Mason CJT, Palermo M, Williams MB, Cleek C, Jordon JB (2020) Influence of grain refinement and microstructure on fatigue behavior for solid-state additively manufactured Al-Zn-Mg-Cu Alloy. Metall Mater Trans A 51(6):2778–2795 9. Griffiths RJ, Garcia D, Song J, Vasudevan VK, Steiner MA, Cai W, Hang ZY (2021) Solidstate additive manufacturing of aluminum and copper using additive friction stir deposition: process-microstructure linkages. Materialia 15:100967 10. Yamashita Y, Fujita K (2001) Newly developed repairs on welded area of LWR stainless steel by friction surfacing. J Nucl Sci Technol 38(10):896–900 11. Guo D, Kwok CT, Chan SLI (2019) Spindle speed in friction surfacing of 316L stainless steel– how it affects the microstructure, hardness and pitting corrosion resistance. Surf Coat Technol 361:324–341 12. Rafi HK, Babu NK, Phanikumar G, Rao KP (2013) Microstructural evolution during friction surfacing of austenitic stainless steel AISI 304 on low carbon steel. Metall Mater Trans A 44(1):345–350 13. Wang Z, Palmer TA, Beese AM (2016) Effect of processing parameters on microstructure and tensile properties of austenitic stainless steel 304L made by directed energy deposition additive manufacturing. Acta Mater 110:226–235 14. de Lima MSF, Sankaré S (2014) Microstructure and mechanical behavior of laser additive manufactured AISI 316 stainless steel stringers. Mater Des 55:526–532

278

N. Gotawala et al.

15. Chen X, Li J, Cheng X, He B, Wang H, Huang Z (2017) Microstructure and mechanical properties of the austenitic stainless steel 316L fabricated by gas metal arc additive manufacturing. Mater Sci Eng, A 703:567–577 16. Wang L, Xue J, Wang Q (2019) Correlation between arc mode, microstructure, and mechanical properties during wire arc additive manufacturing of 316L stainless steel. Mater Sci Eng, A 751:183–190 17. Gandra J, Krohn H, Miranda RM, Vilaça P, Quintino L, Dos Santos JF (2014) Friction surfacing—a review. J Mater Process Technol 214(5):1062–1093 18. Huang K, Logé RE (2016) A review of dynamic recrystallization phenomena in metallic materials. Mater Des 111:548–574

Part VIII

Additive Manufacturing: Materials Design and Alloy Development IV: Rapid Development

A 3D Multiple-Slip Crystal-Plasticity Model for Precipitate Hardening in Additively Manufactured High Strength Steels Moustafa M. AbdelHamid and Tarek M. Hatem

Abstract Additive Manufacturing (AM) revolutionized the manufacturing of complex geometry products, especially in medical and aerospace fields. Highstrength precipitate hardened (PH) stainless steels provide unique properties in term of strength and corrosion resistance for critical applications in both fields. In the current study, a 3D multiple-slip crystal-plasticity dislocation densities-based model is used to study the effect of copper precipitate hardening in high-strength stainless steels. The proposed approach accurately predicts the complex structure of martensite and properly represents the precipitates, based on its characteristics, such as texture, morphology, secondary phases, and initial dislocation densities. The results show the effect of materials’ characteristics on mechanical properties and failure of AM-PH high-strength steels. The current work lays the groundwork for more extensive work of AM modeling. Keywords Crystal plasticity · Metal additive manufacturing · Multiscale material modeling

Introduction Additive Manufacturing (AM) is a remarkable process to build up heterogeneous multi-material microstructure by the control of material distribution at different scale. In particular, Metal Additive Manufacturing has numerous contributions to produce low-volume and cost-effective parts with high hardness and strength in medical and aerospace fields, like stator parts of engines, fitting gears, compressor impeller, and M. M. AbdelHamid Mechanical Engineering Department, Nile University, Juhayna Square, 26th of July Corridor, Sheikh Zayed, Giza 12588, Egypt e-mail: [email protected] T. M. Hatem (B) Faculty of Energy and Environmental Engineering, The British University in Egypt, El-Sherouk City 11837, Cairo, Egypt e-mail: [email protected] © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_25

281

282

M. M. AbdelHamid and T. M. Hatem

fasteners [1, 10]. By different AM processing techniques, the physical and mechanical properties of parts have affected by metallurgical differences such as mechanical anisotropy, residual stress, and defects. The microstructural defects in result of the AM processing techniques are loss of alloying elements, porosity and lack of fusion (LOF) defects, surface roughness, and cracking and delamination [2]. To suppress these defects, precipitation hardening is an effective strengthening mechanism that resists the plastic deformation in metals, at crystal level. High-strength precipitate hardening (PH) stainless steel is an outstanding ferrous alloy whose mechanical properties are the combination of excellence in high-strength and corrosion resistance. 17–4 PH is a martensitic PH stainless steel that can be strengthened by precipitation of fine copper (Cu) particles. A fully martensitic matrix is a body-centered cubic (BCC) that has poor ductility and high susceptibility for stress corrosion cracking, as a result it has strengthened by a fine Cu-rich particle, face-centered cubic (FCC) [3]. By [12], the aged-specimens at 620 °C for 4 h are investigated, and transmission electron microscopy (TEM) images have revealed large quantities of precipitates in the martensite matrix, whose average particle size was about 30 nm. These precipitates had a Kurdjumov–Sachs (K–S) orientation relations (OR) with the body-centered cube (BCC) matrix. Therefore, the current study aims to present the computational framework for AM material modeling at multi-scale. A 3D multiple-slip crystal-plasticity dislocation densities-based model and specialized finite-element formulations are used to study the effect of Cu-rich precipitates hardening in 17–4 PH stainless steel. To investigate the evolution of shear-strain localization, the model of 17–4 PH stainless steel matrix crystal is subjected to high impact loading condition. This paper is organized as follows: the methodology of the framework of AM modeling, which consists of the crystal-plasticity dislocation-based model and finite-element model, is given in section “Methodology”, the results and discussion are given in section “Results and Discussion”, and the conclusion is given in section “Conclusion”.

Methodology AM has potential to produce high-performance parts which are high-demand in industry market. To achieve optimal process-structure–property relationships for AM, it should be hold material modeling at different length scale. At microstructure scale, the model of 17–4 PH presents as proposed by [4–9] in which microstructural model presents martensitic microstructure accurately in terms of orientation, morphology, and transformation dislocation density. Microstructural models are studied for four cases as follows: (1) microstructure without voids, (2) microstructure with voids by 4% of the area fraction, (3) microstructure with copper inclusion by 4%, and (4) microstructure with voids by 2% and copper inclusions by 2%. At precipitate-matrix crystal scale, the proposed model is studied as follows:

A 3D Multiple-Slip Crystal-Plasticity Model …

283

Crystal-Plasticity Dislocation-Based Model A 3D multiple-slip crystal-plasticity dislocation densities-based constitutive laws, which are coupled with the evolutionary equations for the mobile and immobile dislocation densities, are used. A detailed formulation is given by [11]. First, multiple-slip crystal-plasticity kinematics present as follows: the velocity gradient is related to the deformation gradient and is decomposed into a symmetric deformation rate tensor, Dij, and a skew-symmetric deformation rate tensor, Wij. Then, Dij and Wij are additively decomposed into elastic and plastic parts as Vi j = Di j + Wi j

(1)

Di j = DiPj + Di∗j

(2a)

Wi j = WiPj + Wi∗j

(2b)

The inelastic components are formulated in terms of the crystallographic slip rates, summed over all slip-systems, α as (α) DiPj = Pi(α) j γ˙

(3a)

(α) WiPj = ωi(α) j γ˙

(3b)

(α) where the tensors Pi,(α) j and ωi, j are the symmetric and skew-symmetric components of the Schmid tensor, defined in terms of the unit normal and the unit slip vectors. For a rate-dependent inelastic formulation, the constitutive description on each slip system can be characterized by power law formulations where the slip rates are the functions of the resolved shear and the reference stresses.

 γ˙

(α)

=

(α) γ˙ref

τ (α) (α) τref

   1 τ (α)  m −1 (α) τref

(4)

(α) no sum on α is used. The reference shear-strain rate, γ˙ref , corresponds to a reference (α) shear stress, τref and m describe the material rate sensitivity. Second, the evolutionary equations for the mobile and immobile dislocation densities are utilized to depict the dislocation structure, motion, interaction, transmission, and its effect on material deformation modes. The total dislocation density, ρ (α) , is assumed that is decomposed into a mobile dislocation density, ρm(α) ,and an immobile (α) dislocation density, ρim , as

284

M. M. AbdelHamid and T. M. Hatem (α) ρ (α) = ρm(α) + ρim

(5)

Then, a set of nonlinear of evolutionary equation is as follows:       (α) gimmob H dρm(α) ρim gminter (α) (α) gsource = γ˙ − exp − ρim − (6) dt b2 b2 κT b ρm(α)  

   (α) dρim gimmob H H (α) (α) (α) gminter + ρ exp − ρim − grecov exp − = γ˙ dt b2 κT b κ T im (7) where gsource is a coefficient pertaining to an increase in the mobile dislocation density due to dislocation sources, gminter is a coefficient related to the trapping of mobile dislocations due to forest intersections, cross-slip around obstacles, or dislocation interactions, grecov is a coefficient related to the rearrangement and annihilation of immobile dislocations, gimmob is a coefficient related to the immobilization of mobile dislocations, H is the activation enthalpy, and κ is Boltzmann’s constant.

Finite-Element Model The plastic response due to the presence of a fine Cu-rich particle in 17–4 PH stainless steel matrix under large compressive dynamic load was investigated. The material properties used in this model are summarized in Table 1. The precipitate-matrix model presents for three cases as follows: (1) BCC matrix without FCC particle, (2) BCC matrix embedded with a FCC particle, and (3) BCC matrix with four FCC particles. The area fraction between the matrix size and the particle size was calculated by the weight percentage for copper in 17–4 PH used by [13]. BCC matrix is a matrix crystal that be assumed to have dimension of 0.1 μm × 0.1 μm as shown in Fig. 1. FCC particle was placed at the center of BCC matrix with K-S orientation relation to the matrix. For the three cases, the boundary condition is periodic for a two-dimensional (2D) plane-strain deformation mode as shown in Fig. 1. The loading pressure value had been investigated at 1.5 GPa. A convergent mesh of four-node bilinear plane-strain quadrilateral, one-point integration, and hourglass control elements is used.

A 3D Multiple-Slip Crystal-Plasticity Model … Table 1 Material properties of 17–4 PH stainless steel and copper particles used in finite element model

285

Property

Matrix

Precipitates

E(GPa)

196

120

ν

0.3

0.34

τ y (MPa)

587.5

80

ρ(g/cm3 )

8.03

8.93

C p (J/Kg · K)

550

390

H/k(K )

3290

3700

γ˙ref (s −1 )

0.001

0.001

0 (m−2 ) ρim

1010

1010

0 (m−2 ) ρmo

107

107

T0 (K )

753

753

m

0.01

0.01

gsource

2.76e − 5

2.76e − 5

gimmob

0.0127

0.0127

gminter

5.53

5.53

grecov

6.63

6.63

Fig. 1 Model of 17–4 PH martensitic matrix crystal embedded with a fine Cu-rich precipitate. The left and right boundaries are periodic, the upper edge is subjected to a compressive dynamic loading of 1.5 GPa, and the lower edge is fixed in [010] direction. To eliminate the rigid body motion, the left-lower corner node is fixed in [010] and [100] directions

Results and Discussion Precipitate-Matrix Scale The nominal strain and the strain rate to the applied load are calculated from the average nodal displacement and velocity on the top boundary. The results are as

286

M. M. AbdelHamid and T. M. Hatem

follows: the nominal strain at 12.5 μs is 1.21% for case (1), 1.24% for case (2), and 1.22% for case (3). The maximum nominal strain rate is 2.8 × 10∧ 5/sec for case (1), 2.0 × 10∧ 5/sec for case (2), and 2.54 × 10∧ 5/sec for case (3). Results for shear slip, the accumulated maximum plastic shear slip is 0.035 at 12.5 μs for case (1), 0.31 for case (2), and 0.48 for case (3) as shown in Fig. 2. The normalized normal stresses for three cases are given in Figure 3. For case 1, the maximum normalized value is −1.76, and the minimum normalized value is −3.2. For case 2, the maximum normalized value is −0.51, and the minimum normalized value is −3.36 attached to the precipitate in [100] direction. For case 3, the maximum normalized value is −1.49 at the particles closed to the loading boundary, and the minimum normalized value is −3.2.

Fig. 2 Shear slip contours at dynamic loading (1.5 GPa), at 12.5 μs a case 1, BCC matrix without particle; b case 2, BCC matrix with a FCC particle; and c case 3, BCC matrix with four FCC particles

Fig. 3 Normalized normal stresses for a dynamic pressure loading of 1.5 GPa, a case 1, BCC matrix without particle; b case 2, BCC matrix with a FCC particle; and c case 3, BCC matrix with four FCC particles

A 3D Multiple-Slip Crystal-Plasticity Model …

287

Fig. 4 Shear slip contours at dynamic loading (1.5 GPa), at 12.5 μs a case 1, microstructure without voids; b case 2, microstructure with voids by 4% of the area fraction; c case 3, microstructure with copper inclusion by 4%; and d case 4, microstructure with voids by 2% and copper inclusions by 2%

Microstructure Scale The results are as follows: the nominal strain at 12.5 μs is 1.78% for case (1), 2.45% for case (2), 1.93% for case (3), and 2.21% for case (4). The maximum nominal strain rate is 6.33 × 10∧ 4/sec for case (1), 8.06 × 10∧ 4/sec for case (2), 6.38 × 10∧ 4/sec for case (3), and 7.63 × 10∧ 4/sec for case (4). Results for shear slip, the accumulated maximum plastic shear slip is 0.11 at 12.5 μs for case (1), 0.56 for case (2), 0.34 for case (3), and 0.34 for case (4) as shown in Fig. 4.

Conclusion AM material modeling, at microstructure scale and at crystal scale, has prospective to improve the mechanical properties of 17–4 PH stainless steel. At crystal scale, the distribution of copper precipitates has affected on the plastic response of material as well as the precipitate size and the number of precipitates per matrix. At microstructure scale, the percentage of porosity, the size of copper inclusion, and the volume fraction of them are significant modeling parameters that influence on the shear-localization due to the existence of voids and inclusions.

References 1. Rashid R, Masood SH, Ruan D, Palanisamy S, Rashid RR, Brandt M (2017) Effect of scan strategy on density and metallurgical properties of 17–4PH parts printed by Selective Laser Melting (SLM). J Mater Process Technol 249:502–511

288

M. M. AbdelHamid and T. M. Hatem

2. DebRoy T, Wei HL, Zuback JS, Mukherjee T, Elmer JW, Milewski JO, Zhang W (2018) Additive manufacturing of metallic components–process, structure and properties. Prog Mater Sci 92:112–224 3. Cheruvathur S, Lass EA, Campbell CE (2016) Additive manufacturing of 17–4 PH stainless steel: post-processing heat treatment to achieve uniform reproducible microstructure. JOM 68(3):930–942 4. Hatem TM, Zikry MA (2009) Dislocation density crystalline plasticity modeling of lath martensitic microstructures in steel alloys. Phil Mag 89(33):3087–3109 5. Hatem TM, Zikry MA (2009) Modeling of lath martensitic microstructures and failure evolution in steel alloys. J Eng Mater Technol 131(4):041207 6. Hatem TM (2009) Microstructural modeling of heterogeneous failure modes in martensitic steels. North Carolina State University. 7. Hatem TM, Zikry MA (2010) Deformation and failure of single-packets in martensitic steels. Comput Mater Continua 17(2):127–147 8. Hatem TM, Zikry MA (2010) Dynamic shear–strain localization and inclusion effects in lath martensitic steels subjected to high pressure loads. J Mech Phys Solids 58(8):1057–1072 9. Hatem TM, Zikry MA (2009) Shear pipe effects and dynamic shear-strain localization in martensitic steels. Acta Mater 57(15):4558–4567 10. Shehata M, Hatem TM, Samad WA (2019) Experimental study of build orientation in direct metal laser sintering of 17–4 pH stainless steel. 3D Print Additive Manuf 6(4):227–233 11. Zikry MA, Kao M (1996) Inelastic microstructural failure mechanisms in crystalline materials with high angle grain boundaries. J Mech Phys Solids 44(11):1765–1798 12. Hsiao CN, Chiou CS, Yang JR (2002) Aging reactions in a 17–4 PH stainless steel. Mater Chem Phys 74(2):134–142 13. Lebrun T, Tanigaki K, Horikawa K, Kobayashi H (2014) Strain rate sensitivity and mechanical anisotropy of selective laser melted 17–4 PH stainless steel. Mech Eng J 1(5):SMM0049– SMM0049

Development of Al-Cu-Mg and Al-Mg-Si-Zr Alloys with Improved L-PBF Processability F. Belelli, R. Casati, and M. Vedani

Abstract Many Al alloys are susceptible to hot cracking when manufactured by Laser Powder Bed Fusion (L-PBF). In this study, small batches of Al powders were processed using a Reduced Build Volume device to target the optimal chemical composition of the alloy able to suppress hot cracks during solidification. Specifically, batches with increasing content of Cu and Zr were obtained through mechanical mixing of Al-4wt.%Cu-Mg and pure Cu and Al-Mg-Si and Al-Mg-Si-2wt.%Zr powders, respectively. The design strategy based on Cu relies on the segregation of an abundant Al-Al2 Cu eutectic phase mixture during final stages of solidification, whereas the Zr addition promotes a fine equiaxed microstructure induced by heterogeneous nucleation of grains triggered by the precipitation of L12 -Al3 Zr crystal nuclei in the liquid phase. The design of the new alloys was supported by thermodynamic simulations. The microstructures and phase transformations of the alloys were investigated through electron microscopy, X-ray diffraction, and differential scanning calorimetry. Keywords Additive manufacturing · Alloy design · Simulation · Material characterization

Introduction Additive Manufacturing (AM) technologies are widely used in several industrial sectors for the production of light-weight components. Laser Powder Bed Fusion (LPBF) is the most widespread technique among AM technologies [1]. L-PBF relies on the production of metallic parts by means of a layer-by-layer strategy. Specifically, a powder bed is selectively melted by a high-energy laser beam involving the formation and rapid solidification of tiny melt pools. The deposition of several powder layers leads to the production of a three-dimensional component [2]. Even though F. Belelli · R. Casati (B) · M. Vedani Department of Mechanical Engineering, Politecnico Di Milano, Via G. La Masa 1, 20156 Milano, Italy e-mail: [email protected] © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_26

289

290

F. Belelli et al.

L-PBF systems are commonly used in the manufacturing industry, few materials are available for this technique [3]. High-strength Al alloys belonging to 2xxx, 6xxx, and 7xxx series are characterized by a scarce L-PBF processability [4, 5]. Indeed, when these alloys are melted and rapidly solidified suffer the phenomenon of solidification cracking, commonly found during casting and welding of metallic materials. Hot cracks arise along grain boundaries during the final stages of solidification due to tensile stresses induced by thermal shrinkages [6]. Factors affecting the solidification cracking susceptibility of Al alloys are mainly the chemical composition, solidification temperature range, and microstructure [7]. In this work, we review two different strategies that can be adopted to increase the solidification cracking resistance of high-strength Al alloys. Specifically, the effect of Cu and Zr on L-PBF processability of Al-Cu-Mg and Al-Mg-Si alloys, respectively, was investigated in this study. The design strategy based on Cu relies on the segregation of an abundant Al-Al2 Cu eutectic phase mixture during the final stages of solidification [8, 9], whereas the Zr addition promotes a fine equiaxed microstructure induced by heterogeneous nucleation of grains triggered by the precipitation of L12 -Al3 Zr crystal nuclei in the liquid phase [5, 6, 7, 10]. To target the optimal content of Cu and Zr, small batches of Al powders were processed using a Renishaw AM250 machine equipped with Reduced Build Volume device. Specifically, batches with increasing content of Cu and Zr were obtained through mechanical mixing of Al-4wt.%Cu-Mg and pure Cu and Al-Mg-Si and Al-Mg-Si-2wt.%Zr powders, respectively. The effect of Cu and Zr on the L-PBF processability of the new alloys was supported by thermodynamic simulations. Microstructures and phase transformations of the alloys were investigated through electron microscopy, X-ray diffraction, and differential scanning calorimetry.

Material and Methods A Renishaw AM 250 system equipped with Reduced Build Volume device was used to produce cubic samples (10 mm × 10 mm × 10 mm). Nine specimens with different set of process parameters were manufactured for each powder. The nominal chemical composition of the investigated alloys is reported in Table 1. Optimization of L-PBF parameters was carried out based on the effect of hatch distance and point distance on relative density results. Table 2 resumes the investigated set of process parameters. Layer thickness, laser power, and exposure time were set at 25 μm, 200 W, and 140 μs, respectively. The relative density of samples was computed by image analysis through ImageJ software. Three images of the section parallel to the building direction were taken for each sample. ThermoCalc software (using TCAL 5: Al-Alloys v5.1 database) was used to perform simulations of the solidification curves of all alloys, according to Scheil solidification hypotheses [7]. Nikon Eclipse LV150NL light optical microscope and Zeiss Sigma 500 field emission scanning electron microscope (FE-SEM) were used for microstructure investigations. Chemical etching was performed using Keller’s

Development of Al-Cu-Mg and Al-Mg-Si-Zr Alloys with Improved …

291

Table 1 Nominal chemical composition of the investigated alloys Cu

Mg

Si

Zr

Mn

Fe

Al

Al-Mg-Si

–

1.2

1.1

0.2

0.6

0.1

Bal

Al-Mg-Si-1wt.%Zr

–

1.1

1.2

1.1

0.7

0.3

Bal

Al-Mg-Si-2wt.%Zr

–

1.0

1.2

2.0

0.8

0.5

Bal

Al-4wt.%Cu-Mg

4.0

1.4

0.1

–

0.4

0.1

Bal

Al-6wt.%Cu-Mg

6.0

1.4

0.1

–

0.4

0.1

Bal

Al-8wt.%Cu-Mg

8.0

1.3

0.1

–

0.4

0.1

Bal

Al-10wt.%Cu-Mg

10.0

1.3

0.1

–

0.4

0.1

Bal

Al-12wt.%Cu-Mg

12.0

1.3

0.1

–

0.4

0.1

Bal

Table 2 Set of process parameters used to manufacture specimens for density and microstructural investigations Hatch distance (μm)

Point distance (μm)

Volumetric energy density (J/mm3 )

A

80

40

350

B

100

40

280

C

120

40

233

D

80

60

233

E

100

60

187

F

120

60

156

G

80

80

175

H

100

80

140

I

120

80

167

reagent after polishing and grinding of metallographic samples. A Rigaku SmartLab SE multipurpose X-ray diffractometer operating at 40 kV and 40 mA and equipped with copper radiation source (Cuα1 of 1.54060 Å and Cuα2 of 1.54439 Å) and D/teX Ultra 250 detector was used for XRD analysis. All diffraction patterns were collected in Bragg–Brentano geometry with a scan rate of 1°/min and a step size of 0.02°. Differential Scanning Calorimetry (DSC) analysis was performed in Ar atmosphere with a cooling rate of 30 °C/min using a Setaram Labsys TG-DSC equipment.

Results Al-Cu-Mg Thermodynamic simulations of the solidification curve of the investigated alloys are reported in Fig. 1. The formation of the primary α-Al phase starts at 642 °C

292

F. Belelli et al.

and 620 °C in the Al-4Cu-Mg and Al-12Cu-Mg alloy, respectively, whereas the solidification ends at 508 °C in all Al-Cu-Mg alloys. The addition of Cu is responsible for an abundant precipitation of α-Al-Al2 Cu and α-Al-Al2 Cu-Al2 CuMg eutectic mixtures during the final stages of solidification. Specifically, the mass fraction of solid corresponding to α-Al-Al2 Cu precipitation decreases from 89.0% to 66.7% increasing the Cu content in the alloy (Fig. 1b). Relative density results of Al-Cu-Mg alloys are depicted in Fig. 2a as a function of volumetric energy density (VED). Both Al-10Cu-Mg and Al-12Cu-Mg alloys showed the best results for all set of process parameters, with the highest relative density of 99.71% and 99.52%, respectively, for samples processed with VED = 280 J/mm3 . Representative LOM images are reported in Fig. 2b and show the beneficial effect of Cu on the solidification cracking resistance of Al-Cu-Mg alloys. Micrographs collected on the vertical section of Al-10Cu-Mg and Al-12Cu-Mg specimens depict no hot cracks within the microstructure, whereas long and interconnected flaws are visible in samples with lower content of Cu. FE-SEM investigations were performed on Al-4Cu-Mg and Al-12Cu-Mg samples and results are shown in Fig. 3. Discrete second phases homogeneously dispersed in the Al matrix are visible in the microstructure of the Al-4Cu-Mg alloy (Fig. 3a),

Fig. 1 a Simulations of solidification curves of Al-Cu-Mg alloys. b A magnification of solidification curves in the range of high mass fraction of solid (final stages of solidification)

Fig. 2 a Relative density results of Al-Cu-Mg alloys as a function of VED. b Representative LOM images collected on the vertical section of Al-Cu-Mg samples

Development of Al-Cu-Mg and Al-Mg-Si-Zr Alloys with Improved …

293

Fig. 3 FE-SEM images collected on the vertical section of a Al-4Cu-Mg and b Al-12Cu-Mg samples

Fig. 4 a XRD and b DSC analyses performed on Al-Cu-Mg alloys

whereas the addition of Cu led to abundant segregations of second phases at cell boundaries. XRD analyses were performed on the investigated alloys and diffractograms are plotted in Fig. 4a. Characteristic peaks of Al2 Cu phase are noticeable (Card number: 9012196, database COD), with higher intensity with the increase of Cu content in the alloy. DSC investigations were carried out with a cooling rate of 30 °C/min and curves are shown in Fig. 4b. Exothermic peaks A and B indicate the precipitation of α-Al-Al2 Cu and α-Al-Al2 Cu-Al2 CuMg eutectic phase mixture, respectively, in accordance with literature results [11, 12]. Furthermore, low intensity peaks called C and D can be ascribed to Fe-rich phases and Mn/Si-based precipitates, respectively.

Al-Mg-Si Simulations of the solidification curve of Al-Mg-Si and Al-Mg-Si-Zr alloys were performed and shown in Fig. 5. The formation of L12 -Al3 Zr phase in the liquid starts at 720 °C, 900 °C, and 990 °C in the Al-Mg-Si, Al-Mg-Si-1Zr, and Al-Mg-Si-2Zr

294

F. Belelli et al.

Fig. 5 a Simulations of solidification curves of Al-Mg-Si alloys. b A magnification of solidification curves in the range of low mass fraction of solid (beginning of solidification)

alloy, respectively, with a mass fraction that increases with the Zr content (Fig. 5b). The onset of primary α-Al is detected at 647 °C for all investigated alloys, followed by precipitation of Si- and Mg-based phases during the final stages of solidification. Figure 5a and 6b depict relative density results as a function of volumetric energy density and representative LOM images of the investigated alloys, respectively. The highest relative density corresponding to 99.84% and 99.61% was detected in AlMg-Si-1Zr and Al-Mg-Si-2Zr samples processed with the set of process parameters B (VED = 280 J/mm3 ), respectively, whereas the alloy without Zr shows relative densities systematically lower than 99.0%. LOM images confirm the evidence of hot cracks in the vertical section of Al-Mg-Si samples, whereas no large flaws were detected in Al-Mg-Si-1Zr and Al-Mg-Si-2Zr alloys. FE-SEM images collected on the vertical section of specimens are reported in Fig. 7. The solidification microstructure of the Al-Mg-Si-1Zr and Al-Mg-Si-2Zr alloys show fine cells with size in the order of 500 nm and nano-sized precipitates

Fig. 6 a Relative density results of Al-Mg-Si alloys as a function of VED. b Representative LOM images collected on the vertical section of Al-Mg-Si samples

Development of Al-Cu-Mg and Al-Mg-Si-Zr Alloys with Improved …

295

at cell boundaries, whereas slightly coarser and discrete second phases are visible in the Al-Mg-Si samples.

Discussion High-strength Al alloys belonging to 2xxx, 6xxx, and 7xxx series show a scarce L-PBF processability and high tendency to hot cracking [4, 7]. During the final stages of solidification, tensile stresses build-up in the melt pool and hot cracks can nucleate and propagate along grain boundaries [6, 7]. In this study, two strategies to suppress the formation of hot cracks in high-strength Al alloys were reviewed. The first route relies on the addition of Cu in the chemical composition of a Al-Cu-Mg alloy. Simulations of the solidification curves of Al-Cu-Mg alloys (Fig. 1) showed that the addition of Cu is able to decrease the solidification temperature range and stimulate the precipitation of an abundant Al-Al2 Cu eutectic phase mixture during the final stages of solidification. Both phenomena are beneficial in terms of solidification cracking resistance [6, 7, 8, 9]. Indeed, abundance of liquid during the final stages of solidification is capable to “back-fill” and heal hot cracks, as shown by LOM images and relative density results depicted in Fig. 2. FE-SEM investigations reveal that the solidification microstructure changes with the Cu content. Indeed, the Al-12Cu-Mg alloy features a cellular-like microstructure rich of precipitates at cell boundaries (Fig. 3). It is reasonable to state that those precipitates are rich of Cu, in accordance with diffractograms and DSC curves confirming the precipitation of Al2 Cu phase with higher intensity with the increase of Cu content in the alloy (Fig. 4). Another strategy to increase the hot cracking resistance of high-strength Al alloys is based on the formation of L12 -Al3 Zr crystal nuclei in the liquid phase at the beginning of solidification, able to promote heterogeneous nucleation of fine equiaxed grains [5, 10]. Indeed, an equiaxed microstructure promotes a delay of the dendrite coherency point, a uniform distribution of low melting point segregates and lower tensile stresses during the solidification of the melt pool [13, 14]. Figure 5 depicts the simulations of the solidification curve of Al-Mg-Si alloys with and without Zr. The addition of Zr promotes the formation of Al3 Zr phase in the molten liquid, with higher mass fraction with the increase of Zr in the alloy (Fig. 5b). LOM images reported in Fig. 6b show its beneficial effect, with no flaws in Al-Mg-Si-Zr samples and long cracks noticeable in the Al-Mg-Si alloy. Relative density results (Fig. 6a) are in accordance with LOM investigations, and they reveal a higher processability of the Zr-rich alloys in mostly all processing conditions. The solidification microstructure is also modified with the addition of Zr, with fine cells and nano-sized precipitates at cell boundaries in the Al-Mg-Si-Zr alloys (Fig. 7).

296

F. Belelli et al.

Fig. 7 FE-SEM images collected on the vertical section of a Al-Mg-Si, b Al-Mg-Si-1Zr, and c Al-Mg-Si-2Zr alloy

Conclusion In this study, a method to develop Al alloys processable by L-PBF using small batches of powders was investigated. Al-Cu-Mg and Al-Mg-Si-Zr alloys showing no hot cracks and high relative density results were developed and their microstructures and phase transformations were investigated through electron microscopy, Xray diffraction, and differential scanning calorimetry. The formation of abundant α-Al-Al2 Cu and α-Al-Al2 Cu-Al2 CuMg eutectic mixtures during the final stages of solidification revealed to be beneficial in terms of solidification cracking resistance of Al-Cu-Mg alloys with a content of Cu higher than 10wt.%. Simultaneously, 1wt.%Zr promoted the suppression of hot cracks in Al-Mg-Si systems with the precipitation of Al3 Zr crystal seeds at the beginning of solidification. The as-built microstructure of both Al-Cu-Mg and Al-Mg-Si alloys was modified by the addition of Cu and Zr, respectively, with the formation of fine cells and copious segregation at cell boundaries.

References 1. Conner BP, Manogharan GP, Martof AN, Rodomsky LM, Rodomsky CM, Jordan DC, Limperos JW (2014) Making sense of 3-D printing: creating a map of additive manufacturing products and services. Addit Manuf 1:64–76. https://doi.org/10.1016/j.addma.2014.08.005 2. Yap CY, Chua CK, Dong ZL, Liu ZH, Zhang DQ, Loh LE, Sing SL (2015) Review of selective laser melting: materials and applications. Appl Phys Rev 2:041101. https://doi.org/10.1063/1. 4935926 3. Gu D (2015) Laser additive manufacturing of high-performance materials. Springer. https:// doi.org/10.1007/978-3-662-46089-4 4. DebRoy T, Wei HL, Zuback JS, Mukherjee T, Elmer JW, Milewski JO, Beese AM, WilsonHeid A, De A, Zhang W (2018) Additive manufacturing of metallic components—process, structure and properties. Prog Mater Sci 92:112–224. https://doi.org/10.1016/j.pmatsci.2017. 10.001 5. Martin JH, Yahata BD, Hundley JM, Mayer JA, Schaedler TA, Pollock TM (2017) 3D printing of high-strength aluminium alloys. Nature 549:365–369. https://doi.org/10.1038/nature23894 6. Kou S (2003) Solidification and liquation cracking issues in welding. Jom 55:37–42. https:// doi.org/10.1007/s11837-003-0137-4 7. Kou S (2003) Welding metallurgy. New Jersey, USA, 431(446):223-225. https://doi.org/10. 1533/9781855737631.10

Development of Al-Cu-Mg and Al-Mg-Si-Zr Alloys with Improved …

297

8. Kou S (2015) A criterion for cracking during solidification. Acta Mater 88:366–374. https:// doi.org/10.1016/j.actamat.2015.01.034 9. Casati R, Coduri M, Riccio M, Rizzi A, Vedani M. (2019) Development of a high strength Al-Zn-Si-Mg-Cu alloy for selective laser melting, J Alloys Compd 801:243–253. https://doi. org/10.1016/j.jallcom.2019.06.123. 10. Zhang H, Zhu H, Nie X, Yin J, Hu Z, Zeng X (2017) Effect of Zirconium addition on crack, microstructure and mechanical behavior of selective laser melted Al-Cu-Mg alloy. Scr Mater 134:6–10. https://doi.org/10.1016/j.scriptamat.2017.02.036 11. Qiang Han J, Sheng Wang J, Shan Zhang M, Min Niu K (2020) Relationship between amounts of low-melting-point eutectics and hot tearing susceptibility of ternary Al−Cu−Mg alloys during solidification. Trans Nonferrous Met Soc China (English Ed.) 30:2311–2325. https:// doi.org/10.1016/S1003-6326(20)65381-X 12. Zamani M, Toschi S, Morri A, Ceschini L, Seifeddine S (2020) Optimisation of heat treatment of Al–Cu–(Mg–Ag) cast alloys. J Therm Anal Calorim 139:3427–3440. https://doi.org/10. 1007/s10973-019-08702-x 13. Belelli F, Casati R, Riccio M, Rizzi A, Kayacan MY, Vedani M (2021) Development of a novel high-temperature Al alloy for laser powder bed fusion, metals (Basel). 11:35. https://doi.org/ 10.3390/met11010035 14. Belelli F, Casati R, Larini F, Riccio M, Vedani M (2021) Investigation on two Ti – B-reinforced Al alloys for laser powder bed fusion. Mater Sci Eng A 808:140944. https://doi.org/10.1016/ j.msea.2021.140944

Effect of Hot Isostatic Pressing on the Microstructure of Directionally Solidified Nickel Alloy After SLM Evgenii Borisov, Anna Gracheva, Vera Popovich, and Anatoly Popovich

Abstract The paper investigates the effect of hot isostatic pressing of single-crystal nickel-based alloy manufactured by selective laser melting (SLM) with a hightemperature substrate preheating. A study of the structure and phase composition of the material before and after treatment has been carried out. It was found that as a result of such treatment, the ratio and proportion of the strengthening phases change; however, due to slow cooling after treatment, the optimal ratio and shape of the inclusions are not fixed. In addition, the hardening particles are precipitated. Keywords Selective laser melting · Single-crystal alloys · Powder metallurgy · Additive manufacturing

Introduction The current level of additive technologies development allows them to be seen as an alternative to existing subtractive manufacturing methods and to be integrated more and more deeply into various applications [1–11]. However, selective laser melting (SLM) technology has not yet become fully implemented in industry. The study of the possibilities of this technology will provide a fuller understanding of the prospects and feasibility of its implementation in various technological processes and expand the scope of its possible application. Most of the scientific papers are mainly aimed at expanding the capabilities of additive technologies and the limits of their application [12–20]. Post-treatment methods also have a significant impact on the capabilities of additive technologies [21–23]. This article includes the results of an experiment aimed at studying the effect of hot isostatic pressing (HIP) on the structure and phase composition of a heat-resistant nickel-based superalloy produced E. Borisov (B) · A. Gracheva · V. Popovich · A. Popovich Peter the Great St. Petersburg Polytechnic University, St. Petersburg, Russia e-mail: [email protected] V. Popovich Faculty of Mechanical, Maritime, and Materials Engineering, Delft University of Technology (TU Delft), Mekelweg 2, 2628 CD Delft, The Netherlands © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_27

298

Effect of Hot Isostatic Pressing on the Microstructure of Directionally …

299

by SLM. The SLM process produces a large number of micro-defects, which reduce the strength and load bearing capabilities. One possible solution to this problem is hot isostatic pressing, but this method of post-processing has a negative effect on the structure of the resulting products. Various HIP effects on single-crystal nickel-based alloy are reflected in this paper.

Experimental Methods For the manufacture of test samples, a powder of a heat-resistant nickel alloy obtained by plasma atomization of a rotating electrode was used. Alloy features a three-phase system consisting of a nickel-based γ-solid solution with an FCC crystal lattice, dispersion precipitates of the reinforcing γ’-phase based on the Ni3 Al intermetallic and MC type carbides [24]. The chemical composition of the alloy is shown in Table 1. Samples were manufactured by Aconity3D MIDI selective laser melting machine (Aconity3D GmbH, Germany). SLM parameters resulting in the absence of cracks were selected based on our previous study [24]. The machine is equipped with laser source with variable focal spot diameter with Gauss power distribution and a maximum power of 1000 watts. Moreover, machine is equipped with a module to enable operation with platform preheating up to 1200 °C. Rectangular specimens manufactured from the nickel-based superalloy (the chemical composition is given in Table 1) were exposed to heat treatment. The hot isostatic pressing of the specimens was carried out according to the conditions: heating up to 1200 °C, pressure 160 MPa, holding for 3 h, followed by furnace cooling (Fig. 1). After hot isostatic pressing, specimens were cut and polished along the building direction (BD). To highlight the microstructure, specimens were etched with CuSO4 , Table 1 Chemical composition of nickel alloy powder (wt.%) Ni

Cr

Al

Mo

W

Co

Re

Ta

Nb

C

B

Balance

4.9

5.9

1.1

8.4

9.0

1.93

4.1

1.6

0.12

0.01

Fig. 1 Heat treatment parameters

300

E. Borisov et al.

H2 SO4 , and HCl. Carl Zeiss Supra 55VP scanning electron microscope was used for microstructure analysis. X-ray diffraction analysis was carried out using a diffractometer Bruker D8 Advance (CuKα = 0.15406 nm) in the 2θ-range of 30–100° with a scanning step of 0.020 and exposition of 1.5 s at every step. Structural parameters were refined by the Rietveld method; crystal density for equiatomic alloys was calculated from the mass and lattice parameter using the TOPAS5 program.

Results and Discussion The microstructure of the obtained samples consists of elongated cells located mainly along the growing direction of the samples γ—solid solution with scattered particles γ’—phase formed on the basis of the intermetallic compound Ni3 Al (Fig. 2), which in turn consists of cuboid microparticles with an average size of ~200 nm (Fig. 2). After selective laser melting, the structure of the samples is represented by grains elongated mainly in the direction of heat removal. The main phase is γ, with a distribution of inclusions of γ’—phase (Fig. 2). These inclusions are cuboids with an average size of about 0.2 μm. Inclusions in the form of carbides of alloying elements (Nb and Mo) can be seen along the boundaries of the γ-phase cells. Such precipitates impede the movement of the grain boundaries of the base alloy matrix at high temperatures and prevent the appearance of plastic deformation (Fig. 3). The presence of such carbides, however, can lead to the formation of defects in the form of hot cracks and pores [24, 25]. The microstructure of the sample after hot isostatic pressing is shown in Fig. 4. It was found that the γ’—phase inclusions increased in size and changed their shape. At the same time, the total γ’—phase content in the alloy structure decreased. Phase composition of the sample before (Figs. 4, 1) and after (Figs. 4, 2) hot isostatic pressing shows that all peaks belong to the HCC phase. The structure of the samples is characterized by a strong texture. The strong (110) peak of sample Fig. 2 SEM image of the γ/γ’ microstructure after SLM

Effect of Hot Isostatic Pressing on the Microstructure of Directionally …

301

Fig. 3 SEM images of the γ/γ’ microstructure: a after SLM; b after hot isostatic pressing

Fig. 4 XRD pattern of the nickel-based superalloy samples. 1-before HIP, 2-after HIP

1 indicates that the crystallites are preferably (110)-oriented along the surface. In sample 2, the orientation of crystallites along the direction perpendicular to the (110) plane is observed, and the March-Dallas coefficient was 0.21. The detailed diffractogram of the sample examined before heat treatment shows that the separation of the γ and γ’ phases is observed in the region of 75°. The crystal lattice parameter of the γ-phase is a = 0.3589 nm, and the γ’-phase is a = 0.3579 nm. The misfit value a = (ay – ay’)/ay’ for the sample before heat treatment was 0.003. In the second sample, reflexes from all planes are visible after the hot isostatic pressing, which may be due to recrystallization of grains in the places of defects (cracks) in the process of hot isostatic pressing in the sample. In the samples after hot isostatic pressing, in contrast to the initial samples, there is no peak separation. This effect can be caused due to a decrease in the lattice parameter of the gamma of the solid solution, which is caused by the release of carbides of elements Ta, W, etc. Thus, a decrease in the volume fraction of γ’— phase cells can be associated with a change in the crystal lattice parameter and a change in the chemical composition of the phases.

302

E. Borisov et al.

Conclusions The effect of hot isostatic pressing of a heat-resistant nickel alloy samples fabricated by selective laser melting on the structure and phase composition was investigated. It was found that during hot isostatic pressing, the proportion of the gamma prime phase in the alloy structure decreases. In addition, inclusions change their shape and increase in size. This may be caused by a change in the period of the crystal lattice of the matrix γ—phase, which occurs due to the precipitation of carbides of alloying elements. This alignment can occur due to the release of carbides TaC and WC and, as a consequence, depletion of the γ—solid solution with alloying elements Ta and W, which to the greatest extent increases the lattice period of the γ—solid solution. Local recrystallization is also observed after hot isostatic pressing. Such recrystallization can occur in the areas of micropores and cracks in the process of their closure. Acknowledgements This research was supported by Russian Science Foundation grant (project No. 19-79-30002).

References 1. Kozlova EV, Yakinchuk VV, Starikov KA et al (2021) Life cycle management of the 3D-printer technology to design an underwater drone hull to study the arctic. IOP Conf Ser Earth Environ Sci 625:012016. https://doi.org/10.1088/1755-1315/625/1/012016 2. Kozlova EV, Starikov KA, Konakhina NA, Aladyshkin IV (2020) Usage of additive technologies in the Arctic region. IOP Conf Ser Earth Environ Sci 539:012140. https://doi.org/10.1088/ 1755-1315/539/1/012140 3. Sotov A, Kantyukov A, Popovich A, Sufiiarov V (2021) LCD-SLA 3D printing of BaTiO3 piezoelectric ceramics. Ceramics Int 47(21):30358–30366 4. Ekaterina Kozlova ND (2021) The impact of technological development factors on the quality of life: a comparative analysis OF E7 and G7. Int J Qual Res 16:03. https://doi.org/10.24874/ IJQR16.02-18 5. Javaid M, Haleem A (2018) Additive manufacturing applications in orthopaedics: a review. J Clin Orthop Trauma 9:202–206. https://doi.org/10.1016/j.jcot.2018.04.008 6. Salmi M (2021) Additive manufacturing processes in medical applications. Materials (Basel) 14:191. https://doi.org/10.3390/ma14010191 7. Javaid M, Haleem A (2018) Additive manufacturing applications in medical cases: a literature based review. Alexandria J Med 54:411–422. https://doi.org/10.1016/j.ajme.2017.09.003 8. Patel P, Gohil P (2021) Role of additive manufacturing in medical application COVID-19 scenario: India case study. J Manuf Syst 60:811–822. https://doi.org/10.1016/j.jmsy.2020. 11.006 9. Otero JJ, Vijverman A, Mommaerts MY (2017) Use of fused deposit modeling for additive manufacturing in hospital facilities: European certification directives. J Cranio-Maxillofacial Surg 45:1542–1546. https://doi.org/10.1016/j.jcms.2017.06.018 10. Sacco E, Moon SK (2019) Additive manufacturing for space: status and promises. Int J Adv Manuf Technol 105:4123–4146. https://doi.org/10.1007/s00170-019-03786-z 11. Farré-Guasch E, Wolff J, Helder MN et al (2015) Application of additive manufacturing in oral and maxillofacial surgery. J Oral Maxillofac Surg 73:2408–2418. https://doi.org/10.1016/ j.joms.2015.04.019

Effect of Hot Isostatic Pressing on the Microstructure of Directionally …

303

12. Sufiiarov V, Kantyukov A, Polozov I (2020) Reaction sintering of metal-ceramic AlSi-Al2O3 composites manufactured by binder jetting additive manufacturing process, pp 1148–1155 13. Polozov I, Sufiiarov V, Kantyukov A et al (2020) Microstructure, densification, and mechanical properties of titanium intermetallic alloy manufactured by laser powder bed fusion additive manufacturing with high-temperature preheating using gas atomized and mechanically alloyed plasma spheroidized powder. Addit Manuf 34:101374. https://doi.org/10.1016/j.addma.2020. 101374 14. Maximov M, Nazarov D, Rumyantsev A et al (2020) Atomic layer deposition of lithium– nickel–silicon oxide cathode material for thin-film lithium-ion batteries. Energies 13:2345. https://doi.org/10.3390/en13092345 15. Popovich AA, Masaylo DV, Sufiiarov VS et al (2016) A laser ultrasonic technique for studying the properties of products manufactured by additive technologies. Russ J Nondestruct Test 52:303–309. https://doi.org/10.1134/S1061830916060097 16. Novikov PA, Kim AE, Ozerskoi NE et al (2019) Plasma chemical synthesis of aluminum nitride nanopowder. Key Eng Mater 822:628–633. https://doi.org/10.4028/www.scientific.net/KEM. 822.628 17. Polozov I, Sufiiarov V, Starikov K, Popovich A (2021) In situ synthesized Ti2AlNb-based composites produced by selective laser melting by addition of SiC-whiskers. Mater Lett 297:129956. https://doi.org/10.1016/j.matlet.2021.129956 18. Sufiiarov V, Kantyukov A, Popovich A, Sotov A (2021) Structure and properties of barium titanate lead-free piezoceramic manufactured by binder jetting process. Materials (Basel) 14:4419. https://doi.org/10.3390/ma14164419 19. Borisov EV, Popovich VA, Popovich AA et al (2020) Selective laser melting of Inconel 718 under high laser power. Mater Today Proc 30:784–788. https://doi.org/10.1016/j.matpr.2020. 01.571 20. Polozov I, Starikov K, Popovich A, Sufiiarov V (2021) Mitigating inhomogeneity and tailoring the microstructure of selective laser melted titanium orthorhombic alloy by heat treatment, hot isostatic pressing, and multiple laser exposures. Materials (Basel) 14:4946. https://doi.org/10. 3390/ma14174946 21. Polozov I, Sufiiarov V, Kantyukov A, Popovich A (2019) Selective Laser Melting of Ti2AlNbbased intermetallic alloy using elemental powders: effect of process parameters and posttreatment on microstructure, composition, and properties. Intermetallics 112:106554. https:// doi.org/10.1016/j.intermet.2019.106554 22. Sufiiarov VS, Popovich AA, Borisov EV, Polozov IA (2017) Evolution of structure and properties of heat-resistant nickel alloy after selective laser melting, hot isostatic pressing and heat treatment. Tsvetnye Met 77:77–82 23. Popovich VA, Borisov EV, Sufiyarov VS et al (2019) Tailoring the properties in functionally graded alloy inconel 718 using additive technologies. Met Sci Heat Treat 60:701–709. https:// doi.org/10.1007/s11041-019-00343-z 24. Borisov E, Starikov K, Popovich A, Popovich V (2020) CRACKs formation in nickel-based single crystal alloy manufactured by selective laser melting, pp 875–879 25. Evgenii B, Kirill S, Anatoly P, Vera P (2021) Melt pool evolution in high-power selective laser melting of nickel-based alloy, pp 142–148

Functionally Graded Alloys from 316 Stainless Steel to Inconel 718 by Powder-Based Laser Direct Energy Deposition Kun Li, Jianbin Zhan, Peng Jin, Qian Tang, David Z. Zhang, Wei Xiong, and Huajun Cao Abstract Extreme serving conditions are demanding on materials with functional microstructure and properties. Additive manufacturing (AM) is an efficient method to fabricate complex geometry functionally graded materials (FGMs) with gradually variable composition and structures as a function of position. In this work, a laser-based directed energy deposition (DED) process was carried out to develop a series of compositionally graded joints from 316 stainless steel to Inconel 718 alloy through computational analysis and experimental characterization. The microstructure, composition, and phases were investigated as a function of position in FGMs. Compared to the traditionally fabricated joint, AM graded materials had more gradient composition and microstructure. The computational-experimental approach is a promising method to design good properties of dissimilar metal joints. The gradient zone that can be flexibly tuned by AM process provides a high throughput design through local tailoring of properties to develop new functional materials. K. Li (B) · Q. Tang · D. Z. Zhang · H. Cao (B) State Key Laboratory of Mechanical Transmission, Chongqing University, Chongqing 400044, China e-mail: [email protected] H. Cao e-mail: [email protected] K. Li · Q. Tang · D. Z. Zhang Chongqing Key Laboratory of Metal Additive Manufacturing (3D Printing), Chongqing University, Chongqing 400044, China K. Li · J. Zhan · P. Jin · Q. Tang College of Mechanical and Vehicle Engineering, Chongqing University, Chongqing 400044, China D. Z. Zhang College of Engineering, Mathematics and Physical Sciences, University of Exeter, North Park Road, Exeter EX4 4QF, UK W. Xiong Department of Mechanical Engineering and Materials Science, University of Pittsburgh, Pittsburgh, PA 15261, USA © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_28

304

Functionally Graded Alloys from 316 Stainless Steel to Inconel 718 …

305

Keywords Functionally graded materials · Inconel 718 · Laser-based directed energy deposition · High throughput design

Introduction Functionally graded materials (FGMs) possess spatially varying properties with gradual transitions in chemical compositions or structures, which can overcome the defects resulting from the sharp dissimilar joints [1, 2]. The laser-based direct energy deposition (LDED) with the powder feedstock draws the attention on the manufacturing of FGMs because of the flexibility in geometrical structure design, control space, and so on [3–5]. In the present work, we developed successful builds of FGM bulks from 316 stainless steel (SS316) to Inconel 718 superalloy (IN718) by LDED. The element composition, microstructure, and mechanical properties in asdeposited FGMs were evaluated. The aging behaviors on the FGMs were for the first time studied to elucidate the phase transformation and precipitation. The CALPHADbased (CALculation of PHAse Diagrams) high throughput modeling was performed to predict and compare the phases observed in the experiments, which provides a fundamental guidance for further development on FGMs.

Materials and Experimental Procedure The IN718 and SS316 powders with a particle size range of 80–125 μm were used for the gradient component building. They were deposited on a substrate plate of 316L stainless steel (SS316L) through two feeders in a directed energy system embedded with a Nd:YAG laser. The building parameters were optimized as a laser power of 300 W, a scan speed of 5 mm/s, a layer thickness of 0.25 mm, and a hatch spacing of 0.50 mm with a scan pattern of 90° between each layer. The printing system is flexible to deposit powders with variable fraction of powders by controlling the feed rates. The FGMs were built from pure SS316 to pure IN718 with an increased weight percent of 25% (25 wt.%). Each composition of the material was applied to deposit four layers. The deposited blocks were cut into several samples along cross sections parallel to building direction by wire electric discharge machine for metallurgical and mechanical characterization. Some pieces of samples were encapsulated into vacuumed quartz tubes filled with pure argon for post-aging treatments at 718 °C for 15 h, followed by water quenching. All the samples were mounted and polished with standard metallographic methods. The macro-morphology and microstructure as a function of location were observed using OM, SEM, and EBSD. The phases in this FGM were performed using CALPHAD-based modeling to predict the phases and help microstructural analyses. The TCNI8 database from Thermo-Calc Software [6] was used to compare the calculation of the phase formation

306

K. Li et al.

in the FGM. P and S were excluded in the calculation as a simplification. The δ phase was suspended in the calculation of TCNI8 to make sure the presence of γ and γ phases.

Results and Discussion FGM Forming Morphologies as a Function of Composition The overall forming morphologies are revealed in Fig. 1. There were no cracks and big forming distortions in these FGM blocks by our in-house processing control. The representative block was cut along the A-A section, as shown in Fig. 1b. The corresponding OM morphology is presented in Fig. 1c. The melt pools and fusion lines between each laser pass are clear to see after the chemical etching. 20 layers were built up with five compositional gradients, i.e., pure SS316, 25 wt.% IN718,

Fig. 1 Overall forming morphologies of as-deposited FGMs a final FGM blocks, b the investigated FGM block showing cross section A-A of sample slicing, c OM morphologies of the FGM along cross section A-A, and d layer height distribution

Functionally Graded Alloys from 316 Stainless Steel to Inconel 718 …

307

50 wt.% IN718, 75 wt.% IN718, and 100 wt.% IN718. Each compositional gradient has four layers. However, each layer has different actual layer height (H L ), which is increasing gradually as the FGM builds up (Fig. 1d). The accumulative height of layers (H A ) is non-linear and deviated from the theoretical line of the ideal case, because of this mismatch between the actual layer height and the set layer thickness. However, the total built height (H T ) was well controlled at around 5 mm. The increase of H L can be explained by the changes in the absorption of laser energy and heat accumulation of each layer.

Microstructural Evolution in As-Deposited FGM The backscatter electron (BSE) morphology and EDS profile of the whole asdeposited FGM along building direction by the multi-step seaming observation are shown in Fig. 2a, b. The EDS analysis was conducted on the finely polished flat surface (Fig. 2a) of the as-deposited FGM without etching. It is obvious that the major elements of Ni and Cr have equal proportional increases, and Fe has an equal proportional decrease from pure SS316 to pure IN718. The alloying element of Nb has a corresponding increase gradually (Fig. 2b). Other alloying elements cannot be quantified accurately due to their low values of contents. However, because the gradient compositions were mixed by two pure powders of SS316 and IN718 with certain weight ratios, all of the elements have the same ratio change in each mixed composition. The major elements of Fe and Ni can represent the real controlling accuracy for the whole gradient compositions. It can be observed that there is a

Fig. 2 Electron observation of the as-deposited FGM along building direction a BSE morphology, b EDS profile, and c EBSD profile

308

K. Li et al.

flat composition profile in each component that changes from pure SS316 to pure IN718, except few small deviations between each layer like in the compositions of 25 and 5 wt.%. Further comparing the measured composition to the nominal (designed) composition by two representative elements of Fe and Ni, it reveals a well agreement with each other, which demonstrates the successful build of this FGM with a decent control on composition. The stitched overall EBSD mapping exhibits the grain evolution as the composition changes (Fig. 2c). Only one face center cubic (FCC) γ phase of the matrix exists in the whole FGM, because both SS316 and IN718 have FCC matrix. There are no sharp interfaces between each component with different composition. Each interface has a small gradient zone, especially for those near the higher content of IN718, as demonstrated in EDS profiles of Fe and Ni in Fig. 2b. The gradient zone along each interface lowers the mismatch of printability for two individual compositions, which reduces the possibility of cracks. Another phenomenon worth to notice is that the grains grow from columnar shape to equiaxial form as the weight percent of IN718 increases (Fig. 2b), which is due to the effect of heat capacity on the solidification and grain growth [7–9]. In the as-deposited FGM, it is obvious that the grain orientation shows a strong texture and anisotropy along the building direction. The grain size decreases first and then increases as the weight percent of IN718 increases. The component with 75 wt.% IN718 has the finest grain size (Fig. 2c). The found of this critical composition on the grain size might be explained by the entropy of this component [10, 11], which will be further studied in the next work. This results in the finest grain in the as-deposited component with the composition of 75 wt.% IN718. Moreover, it can be seen that the low angle grain boundary (LAGB) has the highest value at 75 wt.% IN718, which reflects the residual stress and distortion in the grains [12–14]. This 75 wt.% IN718 component with fine and equiaxial grains directly obtained from the laser deposition has a great potential for use in AM industry with the advantage of avoiding post-heat treatments.

Precipitation Behaviors in FGM During Aging Because of the synergistic effect of gradient compositions and non-equilibrium status from laser deposition, the precipitation behaviors presented in FGMs during aging would have a significant difference from those in the alloys fabricated by the traditional methods. After aged at 718 °C for 15 h, i.e., a peak aging status for IN718 alloy [15, 16], the SEM morphologies of the microstructure in the components with different compositions were observed, respectively, as shown in Fig. 3. It shows that the Laves phase remains in the matrix, even occurring in the matrix of pure SS316 after the long aging time (Fig. 3a). The segregation along columnar dendrites and cellular dendrites becomes serious because of the increased precipitation. Nevertheless, there are new precipitates occurring around the Laves phase, as seen at the high magnification (Fig. 3a3–e3).

Functionally Graded Alloys from 316 Stainless Steel to Inconel 718 …

309

Fig. 3 SEM morphologies of the microstructure in different compositional components of the aged FGM at 718 °C for 15 h a Pure SS316, b 25 wt.% IN718, c 50 wt.% IN718, d 75 wt.% IN718, e 100 wt.% IN718, and f the interface zone between (b) and (c). (“1–4” present the images at different magnifications, the observed location is in the center zone of each component)

To clarify the precipitates in the aged FGM, the thermodynamic simulation via the CALPHAD database was performed to predict the phase fraction as a function of temperature for different compositions. Because aging for 15 h is a relative equilibrium status which is different from the as-deposited condition, the equilibrium database was used for this simulation. Figure 4 shows the simulation result from TCNI8 from pure SS316 to pure IN718 components. In the pure SS316 component, there are plenty of irregular phases along the grain boundaries and many circular Fig. 4 Phase fraction versus composition in the FGM from high throughput thermodynamic modeling by TCNI8

310

K. Li et al.

particles in the matrix (Fig. 3a2). They are only supposed to be the secondary phases of Laves and M23 C6 [17, 18]. In the 25 wt.% IN718 component, Laves phase still forms along the grain boundaries, with an increase volume (Fig. 3b). Meanwhile, there are numerous needle-like secondary phase precipitated near the Laves phase, occupying the grain boundaries (Fig. 3b3). They are supposed to be the NbNi3 . The NbNi3 phase is defined as δ phase in the TCNI8 database. Because δ phase is the stable phase and γ is the metastable phase in Ni-based superalloy, the δ phase is usually suspended in TCNI simulations to guarantee the formation of γ phase. That is why δ phase does not show in Fig. 4, replaced by γ and γ phase. When the weight percent of IN718 increases to 50 wt.%, the NbNi3 phase is not found in the microstructure. Tremendous ultra-fine nanoprecipitates occur in the matrix, which is supposed to be γ and γ phases (Fig. 3c). According to the simulation result, γ and γ phases are formed replacing δ phase, with chemical formulas of Ni3 (Ti, Al) and Ni3 Nb, respectively. It implies that the finer nanoprecipitates surrounding the Lave phase are γ phase and the relatively bigger ones away from the Laves phase are γ phase, which is also demonstrated in the following analyses for 75 wt.% and 100wt.% IN718 components (Fig. 3d4, e4). The TEM analysis of γ and γ phases in IN718 was studied in our previous work [19, 20]. As the increased content of IN718, γ and γ phases become separated from each other. The fraction of γ phase increases, which is well predicted in the simulations using TCNI8 database, as shown in Fig. 4. However, since Laves phase is a metastable phase, it is difficult to get a good prediction from thermodynamic simulation using the AM components with directly aging. Besides the analyzed phases above, MX secondary phase is always formed in the matrix as long as the matrix has the alloying elements, which has been confirmed in the SEM observations (Fig. 3a2–e2) and simulation result (Fig. 4). As elucidated above, the gradient components with different compositions have remarkably different precipitation behaviors and transformations by the impact of the thermal aging treatment. This kind of precipitation transformations can be found in the gradient interface between 25 and 50 wt.% IN718, as presented in Fig. 3f. NbNi3 or δ phase is reduced and partially transformed to γ /γ phase. Another intriguing precipitation behavior is that the formation of γ is always near the Laves phase, and the formation of γ is far away from the Laves phase. Nonetheless, they are all nucleated and precipitated at grain boundaries. This is because the alloying elements like Nb, Ti, and Al at grain boundaries are high due to the segregation, which is suitable to form alloying element-rich γ and γ phases. Furthermore, during the isothermal aging heat treatment, the Laves phase gets dissolved gradually and releases the alloying elements, especially the element of Nb. Therefore, the Nb-rich γ phase is easy to form in the Nb-enriched area near the Laves phase. The Ti-rich γ phase is formed far away from the Laves phase because of the poverty of Nb. These nucleation and transformation between different phases are important and worth to exploit for the development of new FGMs using AM in the following work.

Functionally Graded Alloys from 316 Stainless Steel to Inconel 718 …

311

Conclusions In the present work, the functionally graded material (FGM) from 316 stainless steel to Inconel 718 is fabricated using the powder-based LDED. The gradient microstructure and aging precipitation behaviors have been studied via experimental characterization and thermodynamic modeling. The FGM without cracks and distortions is successfully built using the LDED fabrication method. The composition of this FGM from 316 stainless steel to Inconel 718 is well controlled using hybrid powder feed system. The 75 wt.% IN718 component with fine and equiaxial grains is directly obtained from the laser deposition, which has a great potential for use in AM industry with the advantage of avoiding post-heat treatments. The aging precipitation behaviors in the FGM from Fe-based alloy to Ni-based alloy are for the first time studied. The phase transformations in these gradient components with different compositions are elucidated in depth. The diffusion and segregation of Ni, Nb, and Ti elements underly the transformation mechanism between Laves, δ, γ , and γ phases for the new development of FGMs using AM. Acknowledgements The authors acknowledge all the researchers and labs to provide the experimental facilities. K.L. gratefully acknowledges the support from the Fundamental Research Funds for the Central Universities, under the award number 2021CDJQY-024.

References 1. Mortensen A, Suresh S (1995) Functionally graded metals and metal-ceramic composites: Part 1 processing. Int Mater Rev 40:239–265 2. Liu Z, Meyers MA, Zhang Z, Ritchie RO (2017) Functional gradients and heterogeneities in biological materials: design principles, functions, and bioinspired applications. Prog Mater Sci 88:467–498 3. Hinojos A, Mireles J, Reichardt A, Frigola P, Hosemann P, Murr LE, Wicker RB (2016) Joining of Inconel 718 and 316 stainless steel using electron beam melting additive manufacturing technology. Mater Des 94:17–27 4. Yusuf SM, Gao N (2017) Influence of energy density on metallurgy and properties in metal additive manufacturing. Mater Sci Technol 33:1269–1289 5. Reichardt A, Dillon RP, Borgonia JP, Shapiro AA, McEnerney BW, Momose T, Hosemann P (2016) Development and characterization of Ti-6Al-4V to 304L stainless steel gradient components fabricated with laser deposition additive manufacturing. Mater Des 104:404–413 6. Sundman B, Jansson B, Andersson J-O (1985) The thermo-calc databank system. Calphad 9:153–190 7. Albrecht GF, Sutton SB, George EV, Sooy WR, Krupke WF (1998) Solid state heat capacity disk laser. Laser Part Beams 16:605–625 8. Künzel HM, Kiessl K (1996) Calculation of heat and moisture transfer in exposed building components. Int J Heat Mass Transf 40:159–167 9. Fann WS, Storz R, Tom HWK, Bokor J (1992) Direct measurement of nonequilibrium electronenergy distributions in subpicosecond laser-heated gold films. Phys Rev Lett 68:2834 10. Miracle DB, Senkov ON (2017) A critical review of high entropy alloys and related concepts. Acta Mater 122:448–511

312

K. Li et al.

11. Chen YY, Duval T, Hung UD, Yeh JW, Shih HC (2005) Microstructure and electrochemical properties of high entropy alloys-a comparison with type-304 stainless steel. Corros Sci 47:2257–2279 12. Sun R, Li L, Zhu Y, Guo W, Peng P, Cong B, Sun J, Che Z, Li B, Guo C (2018) Microstructure, residual stress and tensile properties control of wire-arc additive manufactured 2319 aluminum alloy with laser shock peening. J Alloys Compd 747:255–265 13. Winning M, Gottstein G, Shvindlerman LS (2001) Stress induced grain boundary motion. Acta Mater 49:211–219 14. Powell-Do˘gan CA, Heuer AH, Ready MJ, Merriam K (1991) Residual-stress-induced grain pullout in a 96% alumina. J Am Ceram Soc 74:646–649 15. Chaturvedi MC, Han Y (1983) Strengthening mechanisms in Inconel 718 superalloy. Met Sci 17:145–149 16. Slama C, Abdellaoui M (2000) Structural characterization of the aged Inconel 718. J Alloys Compd 306:277–284 17. Boes J, Röttger A, Becker L, Theisen W (2019) Processing of gas-nitrided AISI 316L steel powder by laser powder bed fusion—microstructure and properties. Addit Manuf 30:100836 18. Saboori A, Aversa A, Marchese G, Biamino S, Lombardi M, Fino P (2020) Microstructure and mechanical properties of AISI 316L produced by directed energy deposition-based additive manufacturing: a review. Appl Sci 10:3310 19. Zhao Y, Li K, Gargani M, Xiong W (2020) A comparative analysis of Inconel 718 made by additive manufacturing and suction casting: microstructure evolution in homogenization. Addit Manuf 36:101404 20. Zhao Y, Sargent N, Li K, Xiong W (2020) A new high-throughput method using additive manufacturing for materials design and processing optimization. Materialia 13:100835

In-Situ LENS Fabricated Ti–Al–Si Alloy Phase Transformation and Microstructural Evolution After Isothermal Annealing Heat Treatments Sadiq Abiola Raji, Abimbola Patricia Idowu Popoola, Sisa Leslie Pityana, Olawale Muhammed Popoola, Nasirudeen Kolawole Raji, and Monnamme Tlotleng Abstract Gamma titanium aluminide (γ-TiAl) alloys are lightweight materials with potential application for high-temperature components. But their ductility at room temperature impedes widespread production of parts via traditional processing routes. In this work, intermetallic Ti–Al–Si alloy was produced via laser in situ alloying from elemental powders by applying the laser engineered net shaping (LENS) technique. Isothermal annealing heat treatment was carried out at 1200, 1300, and 1400 °C for 1 h, followed by furnace cooling (FC). A second homogenization heat treatment was done at 850 °C for 6 h followed by FC. The microstructure was characterized by optical microscopy, (OM), scanning electron microscopy (SEM) S. A. Raji (B) · A. P. I. Popoola Department of Chemical, Metallurgical and Materials Engineering, Tshwane University of Technology, Staatsartillerie Road, Pretoria West, Pretoria, South Africa e-mail: [email protected]; [email protected] A. P. I. Popoola e-mail: [email protected] S. A. Raji · N. K. Raji Department of Metallurgical Engineering, Yaba College of Technology, P.M.B. 2011 Yaba, Lagos, Nigeria S. L. Pityana · M. Tlotleng National Laser Centre, Council for Scientific and Industrial Research (NLC-CSIR), Meiring Naude Road, Pretoria, South Africa e-mail: [email protected] M. Tlotleng e-mail: [email protected] O. M. Popoola Department of Electrical Engineering, Centre for Energy and Electric Power (CEEP), Tshwane University of Technology, Staatsartillerie Road, Pretoria West, Pretoria, South Africa e-mail: [email protected] M. Tlotleng Department of Mechanical Engineering Science, University of Johannesburg, Auckland Park Campus, Johannesburg, South Africa © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_29

313

314

S. A. Raji et al.

equipped with an electron dispersion spectroscopy (EDS), and electron backscattered diffraction (EBSD) technique. The result shows precipitates of silicide (ζ-Ti5 Si3 ) grains with lamellae microstructure in the as-built Ti–Al–Si samples, while dense columnar grains of fully lamellar (FL) microstructure comprising of α2 -Ti3 Al and γ-TiAl were observed for the 1300 °C/1 h/FC/850 °C/6 h/FC heat-treated sample with ζ-Ti5 Si3 -phase at the grain boundaries. The high microhardness values of the samples were ascribed to the presence of ζ-Ti5 Si3 -phase being formed. This study established that laser in-situ alloying with standard heat treatment is feasible for the development of TiAl-based alloys. Keywords Additive manufacturing (AM) · Phase transformation · Materials science and engineering · Gamma-titanium aluminides (γ-TiAl) · Composites · Laser engineered net shaping (LENS) · Silicides (ζ-Ti5 Si3 )

Introduction Titanium aluminide-based (TiAl-based) alloys have become materials of key interest, particularly in the field of automobile and aerospace. This is primarily owing to their lightweight, high strength, good corrosion resistance, high stiffness, and strength retention at elevated temperatures [1–5]. TiAl with Al contents between 45 and 48 at.% is generally referred to as gamma-titanium aluminide (γ-TiAl). The alloy has a dual-phase containing a major phase of γ-TiAl and a minor phase of α2 -Ti3 Al of about 90 vol% and 5 vol%, respectively [3]. Depending on the processing route, thermal history, and post-treatment of wrought alloy material, four different microstructures could be possibly achieved [6, 7]. These are fully lamellar (FL), near lamellar (NL), duplex phase (DP), and near gamma (NG) phases. The different microstructures have varied strengths and these structures are basically determined by the brittle-ductile transition temperature (BDTT) [8]. The large grains and anisotropic lath observed by Kim and Kim [3] was said to influence the creep resistance of lamellar microstructure and duplex phase TiAl-based alloys. In Si-containing TiAl-based alloys, it has been reported that fine silicide (ζ-Ti5 Si3 ) particles precipitated within the diminishing α2 -phases which mitigate dislocation motions leading to efficient property improvements in FL materials [9]. According to Karthikeyan and Mils [10], improved creep resistance is suggested to induce strength reduction at temperatures below the BDTT for FL materials by slowing the highcycle fatigue. TiAl-based alloys containing Si have gain attention due to enhanced mechanical properties [9–12] as a result of the strengthening ζ-Ti5 Si3 phase formed. The precipitation of ζ-Ti5 Si3 -phase formation is due to the minimal solubility of Si in the γ-phase [13]. Several authors [4, 9, 11, 12, 14–16] have investigated the effects of Si in TiAl-based intermetallic alloys. Si addition as an alloying element was reported to improve oxidation resistance [4, 17–19], creep resistance [11, 20], and stabilization of fine lamellar microstructure [20]. It is understood that Si addition

In-Situ LENS Fabricated Ti–Al–Si Alloy Phase Transformation …

315

of about 0.5 at.% in TiAl-based alloys precipitates particles of fine Ti5 Si3 while 2–6 at.% Si produces whiskers or large Ti5 Si3 particles [7, 9]. Traditional processing of TiAl-based alloys through forging and casting are very problematic most especially when producing complex-shaped parts [21, 22]. This arises from high energy consumption and machining costs [23, 24], thereby impacting heavily on the production costs. Likewise, TiAl-based alloys are inherently brittle, have low fracture toughness and poor resistance to oxidation above 800 °C. Recently, additive manufacturing (AM) technologies have been developed to overcome the shortcomings associated with conventional manufacturing routes [22, 25]. Laser engineered net shaping (LENS) is a type of laser AM technique that is very attractive and has been adopted in the fabrication of metals, ceramics, and composites [26]. This technology is reported to be suitable for producing complex-shaped components with thin walls [5]. This is achieved by melting and directly depositing metallic powders on a metallic substrate forming a melt pool on a layer-by-layer basis according to predesigned computer-aided design (CAD) model. Thus, in this work, LENS was adopted to build sample coupons from elemental metal powders. However, the aim of this present study is to examine the effects of heat treatment on in-situ LENS fabricated Ti-45Al-5Si alloy. Emphasis will be on the microstructure, phase formation, and microhardness of the alloy after heat treatment.

Experimental Procedure The Ti–Al–Si alloy was fabricated with a 1 kW laser power Optomec 850R LENS system. The processed powders were commercially pure Ti and Al powders supplied from the hoppers of the Optomec LENS system. The pure Si powder was externally supplied from a GTV powder hopper feeder. All powders (Ti, Al, and Si) used were 45–90 μm particle sizes with spherical shape. The powders were all deposited in-situ from all three feeding hoppers to produce alloy samples of 15 mm by 15 mm cubes on Ti6Al4V substrate. Argon gas was used as a carrier gas to purge off oxygen during processing from the chamber. A laser power of 450 W, scan speed of 10.58 mm/s, centre purge of 25 l/min while the carrier gas and powder feed rate for Ti is 4.2 l/min and 2.21 g/min, Al is 2.4 l/min and 0.48 g/min while Si is 2.0 l/min and 0.043 g/min, respectively. The sample fabrication and experimental setup are the same as our previous work, see Ref. [12]. After the fabrication, a two-stage heat treatment was carried out in a muffle furnace. The isothermal annealing heat treatment was performed at 1200, 1300, and 1400 °C for 1 h, followed by furnace cooling (FC). The second step was an ageing homogenization heat treatment done at 850 °C for 6 h followed by FC. The as-built and heat-treated LENS fabricated Ti–Al–Si alloy samples were cut along the build direction for microstructural examinations and phase analysis. The samples were mounted with the aid of an automatic mounting press, ground with emery papers of grits P80 to P4000 and subsequently polished with OP-S suspension using a Struers TegrsForce-5 auto/manual polisher. Kroll’s reagent was used

316

S. A. Raji et al.

as etchant after polishing to reveal the microstructure of the alloy for investigations. The microstructure was analysed by Olympus optical microscope and Joel JSM-6010PLUS/LA SEM equipped with EDS. The composition was investigated with SEM–EDS while the phase present in the alloy for the sample heat treated at 1300 °C/1 h/FC/850 °C/6 h/FC was studied via EBSD analysis. The alloy microhardness values were measured by Zwick/Roell Indentec (ZHVμ) for the as-built and heat-treated LENS fabricated Ti–Al–Si alloy with a 500 gf load.

Results and Discussion Microstructure and Phase Analysis The optical micrograph of the LENS fabricated as-built and heat-treated Ti–Al– Si alloy is shown in Fig. 1. The image of the as-produced alloy by the optical microscope shows bright particles suspected to be mostly unmelted Al as identified in ref [12]. The 1200 °C/1 h/FC/850 °C/6 h/FC alloy sample displays bright precipitates believed to be silicides (ζ-Ti5 Si3 ) which is heterogeneous throughout the microstructure. However, as observed in Fig. 1c, d, the ζ-Ti5 Si3 particles of 1300 °C/1 h/FC/850 °C/6 h/FC sample seems fine and evenly dispersed while in the

Fig. 1 Optical micrograph (×50) of a As-Built, b 1200 °C/1 h/FC/850 °C/6 h/FC, c 1300 °C/1 h/FC/850 °C/6 h/FC and d 1400 °C/1 h/FC/850 °C/6 h/FC LENS fabricated Ti–Al–Si Alloy

In-Situ LENS Fabricated Ti–Al–Si Alloy Phase Transformation …

317

1400 °C/1 h/FC/850 °C/6 h/FC sample, the ζ-Ti5 Si3 particles appeared to have been dissolved in the matrix throughout the microstructure. The heat treatment effects on the microstructural evolution of the in-situ LENS fabricated Ti–Al–Si alloy were studied. The images in Figs. 2, 3, 4, and 5 shows the SEM and EDS mapping of as-built, 1200 °C/1 h/FC/850 °C/6 h/FC, 1300 °C/1 h/FC/850 °C/6 h/FC, and 1400 °C/1 h/FC/850 °C/6 h/FC Ti–Al–Si alloy samples, respectively, that further emphasizes the heterogeneous distribution of the

Fig. 2 Showing a SEM image, b Al, c Si, and d Ti EDS mapping, and e EDS analysis of as-built LENS fabricated Ti–Al–Si alloy

Fig. 3 Showing a SEM image, b Al, c Si, and d Ti EDS mapping, and e EDS analysis of LENS fabricated Ti–Al–Si alloy heat treated at 1200 °C/1 h/FC/ 850 °C/6 h/FC

318

S. A. Raji et al.

Fig. 4 Showing a SEM image, b Al, c Si, and d Ti EDS mapping, and e EDS analysis of LENS fabricated Ti–Al–Si alloy heat treated at 1300 °C/1 h/FC/ 850 °C/6 h/FC

Fig. 5 Showing a SEM image, b Al, c Si, and d Ti EDS Mapping, and e EDS analysis of LENSV fabricated Ti–Al–Si Alloy heat treated at 1400 °C/1 h/FC/ 850 °C/6 h/FC

bright ζ-Ti5 Si3 precipitates mostly around the grain boundaries of the microstructure of supersaturated α2 grains. From the backscattered electron (BSE) SEM images of the Ti–Al–Si alloy microstructure γ(dark), α2 (grey), and ζ-Ti5 Si3 (bright) phases were suspected. The mapped areas (in Figs. 2, 3 and 4) showing high concentrations of Si could easily be matched with the bright portion suspected to be ζ-Ti5 Si3 particles because Si promotes the development of the ζ-Ti5 Si3 -phase in TiAl-based alloys. While the Al and Ti concentrated areas show the formation of γ and α2 , respectively,

In-Situ LENS Fabricated Ti–Al–Si Alloy Phase Transformation …

319

since Al promotes the formation of γ-phase and Ti aids α2 -phase both producing alternating phases of α2 /γ colonies of lamellae. According to the Ti–Al phase diagram, the as-built Ti–Al-Si alloy can be classified as dual (α2 + γ) phase microstructure due to the Al content of 39.67 at.%. The heat treatments were carried out for homogenization and complete transformation of the phases present. It has been reported that γ-TiAl-based alloys produce ζ-Ti5 Si3 either through eutectoid of α2 /γ + ζ or eutectic reactions of L/β + ζ depending on the amount and distribution of Si within the alloy microstructure [15]. The lamellar structure of γ and α2 alternating laths with precipitates of ζ-Ti5 Si3 phase at the grain boundaries was noticed in the heat-treated samples. The amount of ζ-Ti5 Si3 phase tends to increase with temperature from 1200, 1300 to 1400 °C as higher temperatures allows for precipitation of more ζ-Ti5 Si3 phase due to α2 -phase acting as nucleation sites for the formation of ζ-Ti5 Si3 phase. This occurs as the Al content is being depleted favouring more α2 -phase formation within the microstructure. The α2 -phase has high solubility ζ-Ti5 Si3 while the γ-phase has limited or no solubility for ζ-Ti5 Si3 -phase [7, 13, 15]. The SEM images of 1200 °C/1 h/FC/ 850 °C/6 h/FC and 1400 °C/1 h/FC/ 850 °C/6 h/FC sample in Fig. 3a and Fig. 5a, respectively, depict lamellar microstructure of γ and α2 alternate laths with precipitates of ζ-Ti5 Si3 phases at the grain boundaries. However, it was noticed in 1300 °C/1 h/FC/ 850 °C/6 h/FC sample (Fig. 4a) that the lamellar was more apparent and precipitates of ζ-Ti5 Si3 formed. This can be ascribed to the Al content of 44.49 a.% for this sample promoting the formation of more γ-phase needed to produce more γ/α2 lamellae colonies. The ζ-Ti5 Si3V phases were dispersed at the boundaries of the lamellae colonies and within the matrix, causing refinements of considerable colonies to fine grained [27]. Thus, the basis for the microstructural grain refinement observed for the heat-treated samples with silicon demonstrating both positive and negative effects of α2 -phase stability [16]. The amount of the ζ-Ti5 Si3 phase in the heat-treated samples tend to reduce from with increase in the isothermal annealing heat treatment despite the 6 h hold time at 850 °C, due to the stabilising effects of Si on the lamellar microstructure nevertheless more ζ-Ti5 Si3 particles can be seen for the 1300 °C/1 h/FC/ 850 °C/6 h/FC sample. The EDS results of the LENS fabricated Ti–Al–Si alloys samples (Figs. 2, 3, 4 and 5) clearly show the affinity of Si in replacing Al sites in the presence of α2 -phases due to its high solubility for Si. Therefore, it may be inferred that the Si addition causes a reduction of the overall Al content of the alloy. Also, the two-step heat treatment at 1300 °C/1 h/FC/850 °C/6 h/FC showed a more refined lamellar microstructure with approximately 5 at.%. The slight increase in the ζ-Ti5 Si3 precipitates causes both the γ-TiAl and lamellae γ/α2 grains to become finer in size, though the amount of lamellar slightly increases [28]. The one-hour holding time of the isothermal annealing process results in a microstructure consisting of supersaturated α2 -grains and ζ-Ti5 Si3 and γ-phases. The ageing heat treatment allows precipitation of lamellar grains from the supersaturated α2 -phase to guarantee the thermal stability of the final microstructures [12, 24]. The suspected phases of α2 , γ, and ζ-Ti5 Si3 were confirmed via the electron backscatter diffraction (EBSD) analysis presented in Fig. 6. The EDS layered image

320

S. A. Raji et al.

Fig. 6 Showing a SEM Image in Backscattered Electron (BSE) mode, EBSD analysis showing b EDS layered image, c Phase statistics table, d Phase mapping, e Pole Figures (PF), and f PF orientation of phases for the LENS fabricated Ti–Al–Si alloy heat treated at 1300 °C/1 h/FC/ 850 °C/6 h/FC

(Fig. 6b) shows the grain arrangements while the phase orientation mapping (Fig. 6d) displays the ζ-Ti5 Si3 precipitates evenly dispersed in the matrix and along the microstructural grain boundaries with small quantities of γ-phase largely in regions parallel to the build direction. The EBSD analysis reveals that the microstructure is dominated α2 -phase. This was ascertained with phase composition statistics table (Fig. 6c) indicating that the α2 -Ti3 Al is 71.55%, γ-TiAl is 9.10% and ζ-Ti5 Si3 is 16.42% of the phase fraction. Moreover, the EBSD analysis reveals phases in the 1300 °C/1 h/FC/ 850 °C/6 h/FC sample contains mainly dense columnar grains of α2 laths. This dominates the matrix of the microstructure with the γ and ζ-Ti5 Si3 phases in smaller quantities. Consequently, attaining a more balanced phase ratio between α2 and γ via improving the amount of γ-phase while decreasing ζ-Ti5 Si3 would be an objective for future work. The pole figures (PF) in Fig. 6e and PF orientations show the phase texture components to the build direction which follows the morphological orientation observed in the SEM microstructure (Fig. 6a). However, it was noticed that the γ-phase is positioned in clusters between ζ-Ti5 Si3 particles and grains boundaries. It was also observed from the PF that γ-phase displays strong texture followed by ζ-Ti5 Si3 . This is perpendicular to the build direction with the weak texture of α2 -phase. This is attributed to the change during LENS processing and heat transfer. In Fig. 6e, f, the misorientation distribution angle between the phases (γ, α2, and ζ-Ti5 Si3 ) shows no preferential misorientations observed for 1300 °C/1 h/FC/ 850 °C/6 h/FC alloy sample. Thus, it can be inferred that nucleation of lamellar appeared through gradual cooling, i.e. FC.

In-Situ LENS Fabricated Ti–Al–Si Alloy Phase Transformation …

321

Microhardness Properties Figure 7 shows the microhardness of the as-built and heat-treated LENS fabricated Ti–Al–Si alloy. The microhardness properties tend to vary subject on the heat treatment temperature. Also, the microhardness values vary significantly throughout the alloy samples. The as-built Ti–Al–Si alloy has high microhardness values with an averages value of 712 Hv0.5 . However, samples heat-treated at 1200 °C/1 h/FC/850 °C/6 h/FC and 1400 °C/1 h/FC/850 °C/6 h/FC have average microhardness values between 714 Hv0.5 and 752 Hv0.5 greater than the as-built alloy. But the 1300 °C/1 h/FC/850 °C/6 h/FC sample had the least value of microhardness, 661 Hv0.5 . The high microhardness values for all the samples in this study were suggested to be due to the presence of ζ-Ti5 S3 formed with the prolonged holding time allowing for the silicide growth at such high temperatures. Klein et al. [15] and Klein, Clemens and Mayer [13], reported that α2 -Ti3 Al greatly increases TiAl-based alloys mechanical properties and creep resistance. This can easily be observed as the hardness values of all the samples were higher than the 1300 °C/1 h/FC/ 850 °C/6 h/FC sample. As the isothermal heat treatment at 1200 °C/1 h/FC, 1300 °C/1 h/FC and 1400 °C/1 h/FC might have lowered the microhardness values but the ageing heat treatment at 850 °C/6 h/FC allowed for the growth of the ζ-Ti5 Si3 phase favoured by the presence of a large amount of α2 Ti3 Al phase within the matrix of the Ti–Al-Si alloy. Hence, an increase in ζ-Ti5 Si3

Fig. 7 Microhardness values of as-built and heat-treated LENS fabricated Ti–Al–Si alloy

322

S. A. Raji et al.

prevails owing to its dissolution and coarsening phenomena in the α2 -Ti3 Al phase, after annealing at 850 °C [15]. These ζ-Ti5 Si3 grains contributes tremendously to the increased microhardness value of all the samples compared to an average microhardness value of 322.9 Hv for Ti-48Al-2Nb-0.7Cr-0.3Si by Mathabathe et al. [6]. The 1300 °C/1 h/FC/ 850 °C/6 h/FC sample had higher lamellae colonies which can be attributed to the heat treatment temperature of 1300 °C for isothermal transformation to occur, leading to the reduced value in microhardness. The results clearly show that the two-step heat treatments were more effective at temperature of 1300 °C.

Conclusion In this work, heat treatment effects on microstructure and microhardness of Ti–Al–Si alloy produced by LENS were investigated. Based on this study it was concluded that. 1.

2.

3.

The volume fraction of the ζ-Ti5 Si3 phases tends to increase with isothermal heat treatment temperature from 1200 °C, 1300 °C to 1400 °C as the higher temperature allows for precipitation of more ζ-Ti5 Si3 phase. The two-step heat treatment at 1300 °C/1 h/FC/850 °C/6 h/FC showed a more refined lamellar microstructure with a moderate volume fraction of ζ-Ti5 Si3 phases. The increased microhardness values were attributed to the presence of silicide formed.

Acknowledgements The authors acknowledge Mr Nana Arthur, Mr Samuel, Mr Paul, and Dr Iphi all of NLC-CSIR for their assistance while doing the experiments. Also, the authors appreciate financial support of African Laser Centre-National Laser Centre; Council of Scientific and Industrial Research (ALC-NLC;CSIR); Project Number LHIP500 Task ALC S100.

References 1. Pond RC, Shang P, Cheng TT, Aindow M (2000) Interfacial dislocation mechanism for diffusional phase transformations exhibiting martensitic crystallography: formation of TiAl+Ti3 Al lamellae. Acta Mater 48(5):1047–1053 2. Qu SJ, Tang SQ, Feng AH, Feng C, Shen J, Chen DL (2018) Microstructural evolution and hightemperature oxidation mechanisms of a titanium aluminide-based alloy. Acta Mater 148:300– 310 3. Kim YW, Kim SL (2014) Effects of microstructure and C and Si additions on elevated temperature creep and fatigue of gamma TiAl alloys. Intermetallics 53:92–101 4. Jiang HR, Wang ZL, Feng XR, Dong ZQ, Zhang L, Yong LIU (2008) Effects of Nb and Si on high temperature oxidation of TiAl. Trans Nonferrous Metals Soc China 18(3):512–517 5. Seidel A, Saha S, Maiwald T, Moritz J, Polenz S, Marquardt A, Kaspar J, Finaske T, Lopez E, Brueckner F, Leyens C (2019) Intrinsic heat treatment within additive manufacturing of gamma titanium aluminide space hardware. JOM 71(4):1513–1519

In-Situ LENS Fabricated Ti–Al–Si Alloy Phase Transformation …

323

6. Mathabathe MN, Govender S, Bolokang AS, Mostert RJ, Siyasiya CW (2018) Phase transformation and microstructural control of the α-solidifying γ-Ti-45Al-2Nb-0.7Cr-0.3Si intermetallic alloy. J Alloy Compd 757:8–15 7. Raji SA, Popoola API, Pityana SL, Popoola OM (2020) Characteristic effects of alloying elements on β solidifying titanium aluminides: a review. Heliyon 6(7):e04463 8. Wang Q, Chen R, Yang Y, Guo J, Su Y, Ding H, Fu H (2018) Improvement of the creep lifetimes and microstructural stability of β-solidifying γ-TiAl by cold crucible directional solidification. Intermetallics 100:104–111 9. Sun FS, Froes FS (2002) Precipitation of Ti5 Si3 phase in TiAl alloys. Mater Sci Eng, A 328(1–2):113–121 10. Karthikeyan S, Mills MJ (2005) The role of microstructural stability on compression creep of fully lamellar γ-TiAl alloys. Intermetallics 13(9):985–992 11. Du XW, Wang JN, Zhu J (2001) The influence of Si alloying on the crept microstructure and property of a TiAl alloy prepared by powder metallurgy. Intermetallics 9(9):745–753 12. Raji SA, Popoola API, Pityana SL, Tlotleng M (2021) Microstructure and mechanical properties of heat-treated Ti-Al-Si alloy produced via laser in situ alloying. J Mater Eng Perform 30(5):3321–3332 13. Klein T, Clemens H, Mayer S (2016) Advancement of compositional and microstructural design of intermetallic γ-TiAl based alloys determined by atom probe tomography. Materials 9(9):755 14. Karthikeyan S, Viswanathan GB, Gouma PI, Vasudevan VK, Kim YW, Mills MJ (2002) Mechanisms and effect of microstructure on creep of TiAl-based alloys. Mater Sci Eng, A 329:621–630 15. Klein T, Rashkova B, Holec D, Clemens H, Mayer S (2016) Silicon distribution and silicide precipitation during annealing in an advanced multi-phase γ-TiAl based alloy. Acta Mater 110:236–245 16. Huang ZW (2013) Thermal stability of Ti-44Al-4Nb-4Hf-0.2 Si-1B alloy. Intermetallics 37:11– 21 17. Kastenhuber M, Klein T, Clemens H, Mayer S (2018) Tailoring microstructure and chemical composition of advanced γ-TiAl based alloys for improved creep resistance. Intermetallics 97:27–33 18. Ostrovskaya O, Badini C, Baudana G, Padovano E, Biamino S (2018) Thermogravimetric investigation on oxidation kinetics of complex Ti-Al alloys. Intermetallics 93:244–250 19. Mathabathe MN, Bolokang AS, Govender G, Mostert RJ, Siyasiya C (2018) Structure-property orientation relationship of a γ/α2 /Ti5 Si3 in as-cast Ti-45Al-2Nb-0.7Cr-0.3Si intermetallic alloy. J Alloy Compd 765:690–699 20. Kastenhuber M, Rashkova B, Clemens H, Mayer S (2015) Enhancement of creep properties and microstructural stability of intermetallic β-solidifying γ-TiAl based alloys. Intermetallics 63:19–26 21. Baudana G, Biamino S, Klöden B, Kirchner A, Weißgärber T, Kieback B, Pavese M, Ugues D, Fino P, Badini C (2016) Electron beam melting of Ti-48Al-2Nb-0.7Cr-0.3Si: feasibility investigation. Intermetallics 73:43–49 22. Raji SA, Popoola API, Pityana SL, Popoola OM, Aramide FO, Tlotleng M, Arthur NKK (2019) Laser based additive manufacturing technology for fabrication of titanium aluminidebased composites in aerospace component applications. In: Mofid Gorji-Bandpy M, Aly A (ed) Aerodynamics. IntechOpen, London, pp 193–218 23. Wu X (2006) Review of alloy and process development of TiAl alloys. Intermetallics 14(10– 11):1114–1122 24. San Juan J, Simas P, Schmoelzer T, Clemens H, Mayer S, Nó ML (2014) Atomic relaxation processes in an intermetallic Ti–43Al–4Nb–1Mo–0.1B alloy studied by mechanical spectroscopy. Acta Mater 65:338–350 25. Kenel C, Dasargyri G, Bauer T, Colella A, Spierings AB, Leinenbach C, Wegener K (2017) Selective laser melting of an oxide dispersion strengthened (ODS) γ-TiAl alloy towards production of complex structures. Mater Des 134:81–90

324

S. A. Raji et al.

26. Hong C, Gu D, Dai D, Alkhayat M, Urban W, Yuan P, Cao S, Gasser A, Weisheit A, Kelbassa I (2015) Laser additive manufacturing of ultrafine TiC particle reinforced Inconel 625 based composite parts: tailored microstructures and enhanced performance. Mat Sci Eng A 635:118– 128 27. Neelam NS, Banumathy S, Bhattacharjee A, NageswaraRao GVS (2019) The effect of Cr and Mo addition on the oxidation behaviour of Ti-46.5Al-3.5Nb-2Cr-0.3B. Mater Today Proceed 15:30–35 28. Hsu FY, Klaar HJ, Wang GX, Dahms M (1996) Influence of Si content on microstructure of TiAl alloys. Mater Charact 36(4–5):371–378

Part IX

Advanced Functional and Structural Thin Films and Coatings

In-Air Polymerization and Crosslinking of Monomers During Electrospray Deposition Catherine J. Nachtigal, Michael J. Grzenda, and Jonathan P. Singer

Abstract Electrospray deposition (ESD) is a coating technique in which a solution is passed through a charged capillary, causing the solution to disperse into child droplets through a series of Coulomb fissions until the droplets reach a grounded target. This process is advantageous due to its ability to create nanostructured selflimiting electrospray deposition (SLED) coatings with certain solutions. A large disadvantage is its use of a large amount of solvent. This is wasteful and causes sprays to take a long time to deposit a film. To alleviate this, this study focuses on spraying monomers, which can be sprayed at a much higher weight percentage than their corresponding polymers, blended with a photoactivated polymerizing agent and crosslinker under ultraviolet light to create polymers mid-spray. It was found that the spray, using methyl methacrylate as the monomer, deposited a film consisting of oligomeric polymers that could be optionally crosslinked. Keywords Electrospray deposition · Self-limiting electrospray deposition · Thin-film · Monomer · Polymer · Crosslink · Polymerization · Efficiency

Introduction Electrospray deposition is a promising coating technique that has several advantages in its method of deposition, including its ability to create complex nanotextured coatings, ability to adjust solutions and their corresponding spray morphologies, ability to create SLED coatings due to the build-up of charge in the sprayed film, and its ability to be completed in any environment as well as an ambient environment [1–3]. C. J. Nachtigal (B) · M. J. Grzenda · J. P. Singer Rutgers University, 98 Brett Rd, Piscataway, NJ 08854, USA e-mail: [email protected] M. J. Grzenda e-mail: [email protected] J. P. Singer e-mail: [email protected] © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_30

327

328

C. J. Nachtigal et al.

This versatile process has several applications, ranging from use in biomedical engineering and pharmacological settings, to sensors and other complex nanotechnology [4–6]. This process involves spraying a solution by passing it through a conductive capillary and applying a voltage to it, resulting in the formation of a Taylor cone as the solution is drawn out through a build-up of charge. At a critical voltage, the solution then begins to disperse into child droplets repeatedly through a series of Coulomb fissions until there is an equilibrium between the surface tension and charge of the droplet [7–9]. The droplets produced during ESD have been correlated with several properties and is found as: 1

Q 3 ε0 ρ 6 ) + d0 d = α( 4 π σγ where α is a constant related to the fluid’s dielectric permittivity, ρ is the density, γ is the surface tension, σ is the electrical conductivity, Q is the flow rate, ε0 is the permittivity of free space, and d o is a small droplet diameter only significant at low flow rates [10]. Efficiency is an important element of making sure that any method of manufacturing is feasible on a large scale. Several studies have been completed in order to increase the efficiency of ESD and its deposition onto given targets. This includes but is not limited to attempts to increase the spray efficiency of solutions on to nonconductive surfaces, as well as controlling the areas in which the sprays are being deposited [11, 12]. Though these studies have taken a large step in making the process scalable, no studies have focused on a key aspect of efficiency—the weight percentage (wt%) of solute that can be loaded into and sprayed using ESD. It has been found that the stability of these sprays is greatly affected by the viscosity of the solution, and an increase in the wt% of polymers in particular has a large effect on the viscosity of the solution [13]. Usually, sprays previously completed using polymers could only handle at most 8.2 wt% polymers in the solution [14]. This necessity to use a large amount of solvent to spray a small amount of solute forces sprays to not only be wasteful and not environmentally friendly, but it also causes them to take a long time in order to deposit a given film thickness. The process of ESD requires the use of a high voltage power supply in order to cause a build-up of charge in the solution to create the spray. This means that the longer that a spray needs to run, the more power consumed solely by the usage of the high voltage power supply, as well as power used by the syringe pump or other devices used to feed the solution into the capillary, any ventilation equipment used such as ventilation, and other devices required for the operation of an ESD system. This study focused on determining a method of ESD that would allow polymers to be sprayed at a higher wt% in order to avoid these drawbacks of the process. To combat this, monomers were sprayed in place of their polymer counterparts. In order to transition these monomers into polymers, first a photoinitiated polymerizing agent was loaded into the solution, which was a mix of the monomer at a

In-Air Polymerization and Crosslinking of Monomers During Electrospray Deposition

329

high wt% in a sprayable solvent. Then, a crosslinking agent was loaded in as well in order to create a complex connected nanostructure. Following the spray of this solution under ultraviolet light, the photoinitiator (PI) would cause the monomers to polymerize into polymer chains mid-spray, and these chains would then (optionally) crosslink, depositing a polymer formation onto the desired spray substrate. In order to determine if the deposited film was truly polymers, gel permeation chromatography and differential scanning calorimetry were used to determine the molecular weight of the film deposited. A scanning electron microscope (SEM) was also used to determine the morphology of the films, and microscopic reflectometry was used to measure the thickness of the film over time in order to determine if the spray was SLED.

Experimental Solutions Acrylic acid, vinyl pyrrolidone, polyethylene glycol diacrylate, bis-acrylamide, and phenylbis phosphine oxide were obtained from Sigma Aldrich and used as received. Methyl methacrylate (min 99.5%) was obtained from Polysciences, Inc. and used as received. Ethanol (Koptec, 200 proof pure ethanol) and methyl ether ketone (Sigma Aldrich, ACS Reagent, ≥9.0%) were used as the carrier solvents for the ESD solutions. Acrylic acid (AA) and vinyl pyrrolidone (VP) were blended with ethanol individually in order to test their maximum weight loading when electrosprayed. Methyl methacrylate (MMA) was blended with methyl ether ketone (MEK) in order to determine the maximum weight loading when electrosprayed. Phenylbis phosphine oxide (PI) was loaded into each monomer or monomer and solvent blend in order to determine polymerization rate and minimum PI loading amount in each blend. Polyethylene glycol diacrylate (PEGDA) and bis-acrylamide (BIS) were individually blended with each monomer, solvent, and PI blend in order to determine the minimum crosslinker loading amount in each blend.

Solution Formulation To determine the minimum PI needed for each solution in order to successfully polymerize the monomer, first MMA was blended with PI in 100:1, 100:5, 100:10, 100:15, and 100:20 ratios and placed in droplets under a direct UV lamp for a period of time of 30 s, 1, 3, 5, 10, 20, or 30 min. Polymerization was determined when a thinfilm formed over the droplet during the given period of time. For all other monomer solutions, the minimum PI needed was determined by first determining the maximum monomer loading in solvent that was capable of ESD, then determining the minimum

330

C. J. Nachtigal et al.

PI needed to polymerize the spray. Polymerization was determined by either surface appearance or by heating the sprayed substrate and observing the melting point. The maximum monomer loading in solvent was determined by loading either the monomer or the monomer and PI blend into its soluble solvent and using ESD and determining the maximum monomer amount that could successfully undergo ESD and produce a stable spray. The minimum crosslinker loading in each blend was determined by loading either BIS or PEGDA into the blend and then using the blend in ESD. The sprayed wafer was then treated with a droplet of the solution’s solvent, and dissolution would indicate that the solution had not crosslinked and more crosslinker was necessary.

Electrospray Set-Up The set-up of this ESD process consisted of the use of a syringe pump, syringes, silicon wafers, a high voltage power supply, a wafer holder, a UV light, and a ring stand to hold the UV light in place, as displayed in Fig. 1. For sprayed on 3D objects, the silicon wafer and wafer holder would be replaced with a 3D object, such as a brass hedgehog or silver-coated plastic anchor charm, and a tin/lead soldering wire to hold the object in place and serve as a conductive piece to apply the ground charge to. The syringe (5 mL NORM-JECT® ) was loaded with the given solution and placed in the syringe pump (KD Scientific Syringe Pump 780–100) at a desired flow rate. The wafer or 3D object was placed directly in front of the syringe needle, and in the case of the wafer it would be secured in the wafer holder. A high voltage power supply (Gamma High Voltage Research, HV Power Supply ES30P–5 W/DAM) was connected to the system with the positive voltage wire attached to the needle on the syringe and the grounded wire attached to the wafer or 3D object’s securing soldering wire. Silicon wafers and 3D objects were cleaned and degreased with acetone and ethanol before each spray. The UV light (Dymax Model PC-3) was secured using

Fig. 1 Schematic of experimental ESD set-up

In-Air Polymerization and Crosslinking of Monomers During Electrospray Deposition

331

a ring stand and a clamp to point directly at the spray path as well as the substrate during the ESD process at a distance of 8 cm.

Spray Conditions All sprays were sprayed in the Taylor-cone jet regime in this study. Each spray was completed in ambient 23–54% relative humidity. The positive voltage for each was set between 4.5 and 8.0 kV in order to produce a stable spray and a Taylor cone. The spray distance for each spray was constant at 8 cm between the tip of the syringe needle and the silicon wafer or 3D object. Each spray was conducted at room temperature, ranging from 18 to 23 °C under a fume hood. All silicon wafer samples were sprayed at 0.25 mL/hr or 0.5 mL/hr. For each time series, each spray was completed for either 10, 50, or 90 min. For each spray solution test, each spray was completed for 5 min.

Sample Analysis The thickness of each sample was measured using microscopic reflectometry using a Filmetrics microscope model F40-EXR with custom translation stage. Prior to thickness measurement, each sample was smoothed at 125 °C on a hot plate for 5–20 s. The sample central thickness was determined by measuring the thickness of the sample across the diameter of the spray spot on the substrate at 100 points equally spaced across and taking the mean and standard deviation of the center centimeter. The data was processed using a MATLAB program to determine the central thickness from these points. The morphologies of these spray films were characterized using a Zeiss Sigma Field Emission Scanning Electron Microscope (SEM) and a normal microscope. The glass transition temperature of the film samples was determined using a TA Instruments Q1000 differential scanning calorimeter (DSC). The molecular weight of the film samples was determined using gel permeation chromatography using an Agilent 1260 Infinity II.

Results and Discussion Spray Conditions and Formulation Figure 2a–c shows each MMA:PI blend droplet after being exposed to direct UV light. The sample shown in Fig. 2b was able to form a solid film over the surface of the droplet in the 1 m light exposure, while the sample shown in Fig. 2a was unable

332

C. J. Nachtigal et al.

a

b

c

Fig. 2 MMA and PI blends droplet post UV light exposure in a 100:1 for 20 min, b 100:10 for 1 min, and c 100:20 for 1 min MMA:PI ratios

to form a film even with 20 min of UV light exposure. Similar results were found for MMA:PI ratios under 100:10 as in Fig. 2a, making the 100:10 ratio the minimum amount of PI needed in order to polymerize the MMA in a reasonable amount of time, making it likely that the much smaller (~3 orders of magnitude) spray droplets would be able to sufficiently polymerize mid-spray. Figure 3a–d demonstrates the necessity of deliberate selection of the UV light path when polymerizing the MMA MEK PI blended solution. As shown in Fig. 3a, when the UV light does not point through the spray path the solution is unable to properly

Fig. 3 1:1 (10% PI in MMA):(MEK) ESD samples sprayed at 0.5 mL/hr for 5 min at 5.5 kV a Silicon wafer spray spot with UV light pointed at silicon wafer. b Sample in a examined under 5× microscope objective. c Silicon wafer spray spot with UV light pointed with spray path. d Sample in c examined under 5 × microscope objective

In-Air Polymerization and Crosslinking of Monomers During Electrospray Deposition

333

Fig. 4 a 50% MEK, 41% MMA, 5% PI, and 4% PEGDA and b 50% MEK, 43% MMA, 5% PI, 2% PEGDA ESD films with MEK droplet applied to film post-spray

polymerize and create a powdery film as seen in Fig. 3c that is more consistent with a polymer. This blend was determined and used to find the best UV light set-up as it was the maximum amount of monomer in solvent that was able to spray in a stable manner and form a stable Taylor cone. This solution, containing 50% MEK, 45% MMA, and 5% PI, was then crosslinked using BIS and PEGDA. For each of these blends, the crosslinker replaced a portion of the MMA in the blend to determine how much crosslinker was needed to sufficiently crosslink the polymerized monomer. When sprayed, the minimum amount of PEGDA needed in order to crosslink the solution was 4% PEGDA with 50% MEK, 41% MMA, and 5% PI, while the minimum amount of BIS needed in order to crosslink was more than what could be dissolved into the solution. Crosslinking in the case of PEGDA was determined by placing a droplet of MEK on the sprayed film. Upon dissolution, it was determined that the film had not been able to crosslink, while a lack of dissolution indicated the film had crosslinked, as displayed in Fig. 4.

Electrospray Thickness and Morphology Figure 5 shows the evolution of the thickness of the film over time in order to determine the SLED properties of the film. As shown, the thickness increases nearly linearly over time and does not asymptote as the time increases. This data indicates that the spray is not strongly SLED. This data, however, is inconsistent with the observation that the spray spots were increasing in size significantly over time, indicating that the spray was spreading out over time and that the film’s charge was deflecting further spray. As this film was determined to be an oligomer, it makes sense that it had difficulty displaying SLED characteristics, though it may be possible to enhance its SLED properties further through loading in additional materials or through better polymerization of the particle surfaces. Figure 6 shows a 3D anchor nearly-completely coated in a spray film. Though the back of the anchor does not face the spray needle at all and the electric field would

334

MMA+MEK+PI Spray Film Thickness 18 16

Thickness (um)

Fig. 5 50% MEK, 45% MMA, and 5% PI solution ESD film thickness over time Each spray was completed at 0.25 mL/hr with a positive voltage of 6.5–8.0 kV and sprayed on to a silicon wafer. Each individual series was completed at the same humidity, with the humidity for the series ranging from 26 to 46%

C. J. Nachtigal et al.

14

Series 1 Series 2 Series 3 Series 4

12 10 8 6 4 2 0 20

40

60

80

Time (min)

Fig. 6 Sprayed 3D anchor charm. a Front side facing spray needle. b back side facing away from spray needle. Coated with 50% MEK, 45% MMA, 5% PI for 30 min at 0.25 mL/hr and 7.0 kV

not favor the spray to go to the back of the anchor, it does so likely because the charge formation in the film on the front of the anchor begins to deflect further spray towards the back of the anchor, making the spray, at least to some extent, SLED. Figure 7 shows the spray morphology of the standard spray sample, showing a density gradient in the droplets formed and deposited on the silicon wafer.

Polymeric Analysis Figure 8 shows the DSC results for the standard sample in this study. This dip in the graph shows a low glass transition temperature (Tg) of approximately 90 °C. Alongside this study, a gel permeation chromatography (GPC) study was completed in order to determine the molecular weight (MW) of the film. This data found that the polymer was below the 2 kDa detection threshold. Taken together, the results of

In-Air Polymerization and Crosslinking of Monomers During Electrospray Deposition

335

Fig. 7 SEM image of 50% MEK, 45% MMA, and 5% PI spray completed at 0.25 mL/hr for 10 min

MMA+MEK+PI Spray DSC Data

Fig. 8 DSC results from 50% MEK, 45% MMA, and 5% PI sample sprayed for 50 min

Heat Flow (mW)

0.0

-0.2

-0.4

-0.6

80

100

Temperature (°C)

the GPC and DSC studies show that the film contained an oligomeric polymer, or a polymer with a low MW. As the GPC and DSC studies confirm, the studied process of using UV light to polymerize and crosslink a monomer mid-spray was able to successfully convert the given monomer into a polymer. The spray behavior of varying SLED abilities, though troublesome to need to control humidity in order to maintain a SLED spray, is consistent with the spray properties of low MW polymers, as these have been shown to vary significantly with varying humidity as well [12]. Future work will look to increase the MW of the sprayed polymer for more reliable results through use of more-optimized initiators. Despite this, these results indicate that this method can be useful in circumventing the issue of requiring a low wt% of solids to produce a stable electrospray sample consisting of polymers. This allows the sprays to be completed in a much shorter period of time, as they can be sprayed at moderate flow rates with a large wt%, and also waste less monomers. Because of the necessity

336

C. J. Nachtigal et al.

for smaller droplets though and a high surface area to completely polymerize the droplets under UV light, the flow rate cannot be set as high as other polymer sprays. Higher flow rates in this study were also found to make the Taylor cone unstable, making a more moderate flow rate necessary for proper spray deposition. Even with the sprays being completed at about half of the flow rate as their polymer counterpart sprays, the deposition rate is still up to an order of magnitude higher because of the high wt% loading, making this process still a significant optimization of the standard polymer ESD process.

Conclusion This new method of utilizing in-air polymerization and crosslinking during ESD to deposit polymers using an exceptionally low wt% of solvents can be extremely promising in the industrial use and scale-up capabilities of ESD. Studies have found ways to scale-up the ESD process to make it more feasible in a manufacturing environment, but few have done so in a way to improve efficiency to make the process affordable enough to make it a candidate in thin-film deposition in an industrial setting [15]. Because of the SLED capabilities that these films could display with future development, this method is even more promising in industrial use as the spray target and spray capillary do not have to be moved to coat the entire surface evenly, making high-quality automated ESD coatings possible [16]. Moreover, this method has the potential to be applied to several different polymers beyond MMA, making it possible to become a standard in efficiently coating with polymers using ESD. It was found to work with vinyl pyrrolidone and acrylic acid, though not as efficiently as MMA, further justifying the need to study this method more to determine its possible uses and applications. These polymers can likewise be optionally crosslinked through loading in crosslinkers, adding in an additional capability of this spray method. Most importantly, beyond the scope of an industrial setting, this method greatly reduces the environmental impact of spray coatings by using fewer toxic solvents, which have been shown to have significant negative health effects in people [17]. Overall, this process provides a less wasteful, more efficient, and more environmentally friendly way to utilize ESD in polymer thin-film deposition that has the capability to be scaled-up to industrial settings.

References 1. Matsumoto H, Mizukoshi T, Nitta K, Minagawa M, Tanioka A, Yamagata Y (2005) Organic/inorganic hybrid nano-microstructured coatings on insulated substrates by electrospray deposition. J Colloid Interface Sci 286(1):414–416 2. Morota K, Matsumoto H, Mizukoshi T, Konosu Y, Minagawa M, Tanioka A, Yamagata Y, Inoue K (2004) Poly(ethylene oxide) thin films produced by electrospray deposition: morphology control and additive effects of alcohols on nanostructure. J Colloid Interface Sci 279(2):484–492

In-Air Polymerization and Crosslinking of Monomers During Electrospray Deposition

337

3. Novak S, Lin P-T, Li C, Borodinov N, Han Z, Monmeyran C, Patel N, Du Q, Malinowski M, Fathpour S, Lumdee C, Xu C, Kik P, Deng W, Hu J, Agarwal A, Luzinov I, Richardson K (2016) Electrospray deposition of uniform thickness Ge23Sb7S70 and As40S60 chalcogenide glass films. J Vis Exp 114:54379 4. Kavadiya S, Biswas P (2018) Electrospray deposition of biomolecules: applications, challenges, and recommendations. J Aerosol Sci 125:182–207 5. Li W, Lin J, Wang X, Jiang J, Guo S, Zheng G (2018) Electrospray deposition of ZnO thin films and its application to gas sensors. Micromachines 9(2):66 6. Jaworek A, Sobczyk AT (2008) Electrospraying route to nanotechnology: an overview. J Electrostat 66(3–4):197–219 7. Gomez A, Tang K (1994) Charge and fission of droplets in electrostatic sprays. Phys Fluids 6(1):404–414 8. Collins R, Sambath K, Harris M, Basaran O (2013) Universal scaling laws for the disintegration of electrified drops. Proc Natl Acad Sci 110(3):4905–4910 9. Duft D, Achtzehn T, Huber MR, B, Leisner T, (2003) Coulomb fission: Rayleigh jets from levitated microdroplets. Nature 421(6919):128–128 10. Gañán-Calvo AM, Dávila J, Barrero A (1997) Current and droplet size in the electrospraying of liquids scaling law. J Aerosol Sci 28(2):249–275 11. Rahman MK, Phung TH, Oh S, Kim SH, Ng TN, Kwon K-S (2021) High-efficiency electrospray deposition method for nonconductive substrates: applications of superhydrophobic coatings. ACS Appl Mater Interfaces 13(15):18227–18236 12. Lei L, Gamboa AR, Kuznetsova C, Littlecreek S, Wang J, Zou Q, Zahn Z, Singer JP (2020) Self-limiting electrospray deposition on polymer templates. Sci Rep 10:17290 13. Kuwahata Y, Takehara H, Ichiki T (2020) Comprehensive study on electrospray deposition in the single Taylor cone-jet mode by changing the spatial electric potential using a ring-shaped ternary electrode. AIP Adv 10:045107 14. Almería B, Gomez A (2014) Electrospray synthesis of monodisperse polymer particles in a broad (60 nm–2 μm) diameter range: guiding principles and formulation recipes. J Colloid Interface Sci 417:121–130 15. Martinez-Vazquez B, Sanchez DG, Castillo JL, Friedrich KA, Garcia-Ybarra PL (2015) Scaling-up and characterization of ultralow-loading MEAs made-up by electrospray. Int J Hydrogen Energy 40(15):5384–5389 16. Lei L, Kovacevich D, Nitzsche MP, Ryu J, Al-Marzoki K, Rodriguez G, Klein LC, Jitianu A, Singer JP (2018) Obtaining thickness-limited electrospray deposition for 3D coating. ACS Appl Mater Interfaces 10(13):11175–11188 17. Ojo TO, Onayade AA, Afolabi OT, Ijadunola MY, Esan OT, Akinyemi PA, Awe OO (2020) Work practices and health problems of spray painters exposed to organic solvents in Ile-Ife, Nigeria. J Health Pollut 10(28):201208

Incorporation of Metallic Nanoparticles Into Alkyd Resin: A Review of Their Coating Performance I. H. Ifijen, M. Maliki, S. O. Omorogbe, and S. D. Ibrahim

Abstract There is a dramatic increase in alkyd nanocomposites research over the years with the emergence of diverse experimental techniques. This has led to the redefining synthetic processes, analysis, and cost control techniques of nanocomposites. The incorporation of nanomaterials into polymer composites can improve properties such as tensile strength, impact and scratch resistance, mechanical properties, drying properties, chemical resistance, thermal stability, electrical conductivity, and fire resistance. This improvement is expected due to the collaborative properties of the metallic nanomaterials (high surface area to volume ratio, small size, extremely small sizes with high density, and great functionality per unit space) and alkyd resins (biodegradability, great gloss retention, adaptability, flexibility, durability, good drying properties, and weathering resistance). This review examined alkyd resin nanocomposites, possible utilization, and the performance of varying types of nano-metallic materials in modifying alkyd resin polymer. The use of metallic nanomaterials has generated alkyd nanocomposites with better coating features which will find increasing applications in anticorrosion, antifouling/antibacterial application, superhydrophobic application, self-cleaning, antiwear, and electronics. Keywords Alkyd resin · Nanoparticles · Nanocomposites · Coating performance · Metallic nanomaterials

I. H. Ifijen (B) · S. O. Omorogbe Department of Research Operations, Rubber Research Institute of Nigeria, Iyanomo, Benin City, Nigeria M. Maliki Department of Chemistry, Edo University Iyamho, Iyamho, Edo State, Nigeria S. D. Ibrahim Plant Protection Division, Department of Agronomy, Rubber Research Institute of Nigeria, Benin City, Nigeria © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_31

338

Incorporation of Metallic Nanoparticles Into Alkyd Resin …

339

Introduction The versatility and good coating properties of alkyd resins have made them the most well-known synthetics resins employed in the paint industry. They are generally known for their remarkable gloss and corrosion protection and also their ease of application [1]. Alkyds are polyesters which are obtained from the reaction between polybasic acids, polyhydric alcohols, and fatty acids [1, 2]. The use of various types of seed oils like sesamum indicum L. seed oil, linseed oil, jatropha curcas oil, castor oil, and soybean oil have been used for the preparation of vegetable-based coatings like alkyds [3–7]. In recent times, the use of nanomaterials in improving the physical and chemical properties of alkyd resins for better overall coating and more durable automotive clear coats has attracted so much attention. As a result, the demands of alkyd resin nanocomposites have increased in the market [8]. Due to the small particle diameter, nanoparticles an improved specific surface area and their presence in alkyd resins can enhance the optical, mechanical, barrier, and thermal characteristics of their coatings [9]. Such improvement in characteristics can be actualized by uniformly dispersing the nanoparticles through the polymer matrix, thereby preventing nanoparticles aggregation which can occur due to their high surface tension energy and interfacial reactivity. The aggregation of nanoprticles can be modifying the surface of nanoparticles thereby enhancing the compatibility between nanoparticles and polymer. The most conventional way to stabilize the colloidal nanoparticles and control the aggregation of nanoparticles via surface modification is to attach suitable organic groups to the surface atoms of the nanoparticles [10, 11]. There is decrease in the surface tension and an improved compatibility of the nanoparticles with the polymer matrix due to the conferment of hydrophobic property on the nanoparticles after necessary surface modification [12]. This review focused on the current progress in eco-friendly alkyd resins modified with metallic nanoparticles and their coating performance.

Factors Affecting Nanocomposite Coatings The coatings properties nanocomposite is largely influenced by the morphologies, sizes, and types of nanoparticles. The effect of some inorganic oxide nanoparticles (i.e. titania, silica, and alumina) on the performance of acrylic/CNT coating was investigated by Kugler et al. [13]. Findings from their investigation reveals that the silica-based coating containing has a greater transparency and an extraordinarily lower electrical surface resistivity than the blank coating (with CNT only). Besides, CNT and silica or titania nanoparticles-based coatings exhibited a lower gloss and greater hardness than the blank coating. It was established by Zamfirova et al. [14] that the incorporation of CNTs into poly (methyl methacrylate) (PMMA)

340

I. H. Ifijen et al.

and poly (hexyl methacrylate) composites can influence their mechanical characteristics. The addition of 1% of CNTs into polymers matrix greatly improves all the micro-hardness properties. The incorporation level of the dispersed particles is majorly influenced by their concentration. Several researchers have studied the incorporation of metallic oxide nanoparticles such as Al2 O3 , TiO2 , etc., into coating matrix [15]. They were able to establish that an increase in incorporated particles leads to a corresponding increase in their concentration to critical concentration. At the critical concentration, any further addition leads to a decrease in the incorporated particles. Dang et al. [16] examined the impacts of Ag concentration on the coating performance of TiSiN-Ag nanocomposite. The addition of 1.4% of Ag showed a weak wear resistance and high-level of hardness. Elevation in the wear resistance in artificial seawater and ambient air was observed to increase as the Ag content increase to 5.3% and 7.9%, respectively. Mallikarjuna et al. [17] investigated the impacts of silicon carbide (SiC) and CNTs contents on the coating capability of hybrid nanocomposites. They observed an 80% improvement in the coating hardness by incorporating 4.0% of CNTs into the hybrid nanocomposites. Mixing and agitation during the synthesis of nanocomposites is one major factor to be considered in the establishment of particle incorporation. Generally, the dispersion of nanoparticles into polymer matrix is negatively affected by poor mixing. On the contrary, a deformation in the nanoparticles and a decrease of contact between the polymers and the nanoparticles can be facilitated by a very high mixing condition. Kucharska et al. [18] investigated the impact of stirring conditions on the coating capabilities of Ni/Al2 O3 nanocomposite. They examined several stirring techniques such as ultrasonic and mechanical techniques. They stated that the stirring conditions can greatly influence the properties of the synthesized Ni/Al2 O3 nanocomposite coatings. Lecina et al. [19] obtained irregular surface of Ni/IF-WS2 composite coatings by agitating mechanically. Interestingly, a more uniform and compact surface of composite coatings was obtained by the use of ultrasound agitation. Another factor that play a vital role on the modality of the incorporation of nanoparticles into the polymer resin is the addition of surfactants. The surfactants prevent the agglomeration of the particles inside the polymer matrix by generating positive charges over the surface of the nanoparticles, thereby yielding smooth composite coatings. Shirani et al. [20]) examine the function of adding surfactants to Co-TiO2 nanocomposite coatings on its performance. They obtained a better quality of Co-TiO2 nanocomposite coatings by the addition of cetyltrimethyl ammonium bromide and sodium dodecyl sulfate during the nanocomposite coatings fabrication.

Alkyd Resin-Based Metallic Nanocomposite Coatings The use of polymeric matrix in nanocomposite coatings based on alkyd resin alkyd has attracted so much attention in the coating industry. The modification of alkyd resins by incorporating several types of metallic nanomaterials into their structure has been embarked on by several researchers because of the short comings observed

Incorporation of Metallic Nanoparticles Into Alkyd Resin …

341

during the coating applications. Some of these short comings are low drying properties, low resistance to chemical attack, poor thermal and mechanical properties, and inadequate anticorrosive properties. Bhanvase et al. [21] described that the blending of alkyd resin with nano CaCO3 help to improve the strength of alkyd resin. As a consequence, nano CaCO3 alkyd resin composites are able to protect the coated iron surface from corrosion attack from hydrochloric acid (HCl), sodium hydroxide (NaOH), and sodium chloride (NaCl) solutions. They speculated that the nano calcium carbonate (CaCO3 ) particles in the alkyd resin matrix led to the generation of a well-adhered, dense, and a continuous coating layer. This layer hinders the penetration of the corrosive ions through the resin matrix. Also, the roughness of the surface and the surface area is increased by the presences of the nano calcium carbonate (CaCO3 ) particles resulting to an improved adhesion. Permeability in their texture is also among the major issues faced by these polymers when applied in anticorrosive coatings applications. The corrosion processes can be initiated by the permeability which causes the corrosion ions transit to get to the surface of metal. Deyab [22] attempted to solve this problem by the incorporating carbon nanotubes (CNTs) into alkyd resins. His study came to a conclusion that the addition of CNTs to alkyd resin can reduce its permeability and also enhance the anticorrosion effect of the alkyd resin films. Besides, the adhesion and cohesion of the alkyd have been improved by the addition of CNTs. Indeed, the fundamental step for the decrease in adhesion strength is the shear stress at the alkyd resin/metal surface interface. The incorporation of the CNTs into the alkyd resin matrix can relieve the residual stress inside the alkyd resin by shearing the weakly bound carbon collection against the alkyd resin. This leads to the secure adhesion of the CNTs/alkyd resin composite coating on the metal surface. Ong et al. [23] developed palm oil-based alkyd/epoxy resin incorporated with copper oxide nanoparticles and a neat palm oil-based alkyd resin. They also investigated the curing of the prepared alkyd resin with varying ratios of epoxy resin in the presence of poly (amidoamine) as a hardener (Scheme 1). The alkyd incorporated with copper oxide nanoparticles showed about 18 and 41% improvement in tensile and flexural strength respectively, compared to the neat blend. FTIR confirmed the generation of a network between oxirane and amine groups and interaction between polymer matrix and copper oxide nanoparticles. The modification of the palm oilbased alkyd/epoxy resin with copper oxide nanoparticles drastically enhanced its thermal stability and hydrophobicity compared to the neat composite. The study showed that copper oxide nanoparticles fabricated in the glycerol phase promoted their homogeneous distribution in the resin phase and significantly modified the thermal and mechanical characteristics of the nanocomposite. A novel anticorrosion nanocomposite coating was developed by Deyab et al. [24] by incorporating MPorphyrins into alkyd resin. The result revealed that the blending of M-Porphyrins with alkyd resin drastically improve its mechanical and anticorrosive properties. Water uptake by the alkyd resin was reduced by the M-Porphyrins. They attributed it to the difference in the anticorrosive and mechanical efficiency caused by the central metals in M-Porphyrins particles. A significant improvement in the corrosion

342

I. H. Ifijen et al.

Scheme 1 Schematic diagram of the phenomenon occurring during the cross linking of the blend [23]

impedance of steel was observed by Deyab et al. [25] in acidic solution by the incorporation of lanthanide bis-phthalocyanine into alkyd resin. A substantial decrease in the permeability of lanthanide bis-phthalocyanine/alkyd composites was seen when compared to the neat alkyd resin. Besides, the adhesion strength of the alkyd resin lanthanide bis-phthalocyanine was also significantly increased. The magnetic properties of Ln (3+ ) ions were ascribed to be the main reason for the reasonable adhesion of alkyd@LnPc2 nanocomposite coatings. Rahman et al. [26] reported the synthesis (Schemes 2 and 3) and corrosion inhibition efficiency of hyperbranched soya alkyd-based nanocomposite coatings. The synthesis was carried out using phthalic anhydride, soya oil, and pentaerythritol. They dispersed the magnetite (Fe3 O4 ) nanoparticles in butylated melamine–formaldehyde (BMF) modified Hyperbranched alkyd (HBA) (HBA−BMF) using a sonication

Incorporation of Metallic Nanoparticles Into Alkyd Resin …

343

Scheme 2 Synthesis of a Monoglyceride, b Hyperbranched Alkyd (HBA), c HBA−BMF and in Inset 3D View of Hyperbranched HBA [26]

process to fabricate the anticorrosive nanocomposite coatings (HBA−BMF−Fe3 O4 ). The morphological, physico-mechanical, structural, anticorrosive, thermal, and electrochemical properties of these coatings were evaluated using ASTM methods. A globular structure with a considerable degree of branching (DOB = 0.69) was observed for the HBA. The nanocomposites of HBA−BMF and HBA−BMF− Fe3 O4 showed more improved properties such as toughness, mechanical robustness, and flexibility and produced exceptional outcomes as corrosion protective coatings. The incorporation of Fe3 O4 nanoparticles improved the load-bearing potential of nanocomposite coatings by disseminating the instantaneous energy in impact tests and scratches. The corrosion resistance performance exhibited by the HBA−BMF−Fe3 O4 nanocomposite coatings (impedance = 10  and corrosion rate 1.0 × 10−4 mils per year) is better than that of HBA−BMF and other similar reported

344

I. H. Ifijen et al.

Scheme 3 Synthesis of HBA−BMF−Fe3 O4 nanocomposite [26]

coating systems according to the electrochemical corrosion investigations carried out. They compared their study with earlier studies and concluded that the generated HBA−BMF−Fe3 O4 nanocomposite has potential applications as an environmentally friendly, low cost, and VOC-free coatings system. Lotfizadeh et al. [27] investigated the influence of coating aluminium plates with Alkyd Melamine (ES-665) car paint which comprises silver nanoparticles (10 nm) in a wind tunnel utilizing two separate air velocities parallel to the surface of the aluminium plate. Employing two dissimilar air convection velocities of 1.7 and 2.6 m/s, the impact on the temperatures of the surfaces of the coated samples and the solvent vaporization velocities in the drying process is inversely linked to the Reynolds number confirming that a 52.94% increment in the Reynolds number led to a reduction in the samples maximum temperatures by an average of 21.6%. They also investigated the effect of the amounts of silver nanoparticles on the drying process of the samples by utilizing paint with five varying amounts of nanoparticles (5, 10, 15, 20, and 25 ppm) and paint without nanoparticles. The paint with 10 ppm silver nanoparticles exhibited the most reliable outcomes for the drying process, recording a maximum temperature of 80.5 and 69 °C at an air velocity of 1.7 and 2.6 m/s, respectively. This nanoparticle amount also exhibited the most suitable solvent vaporization, probably because the silver had a higher thermal conductivity than the base paint giving rise to enhanced paint’s total thermal conductivity. A better drying rate of the paint was observed by a higher convection coefficient and lower diameter of

Incorporation of Metallic Nanoparticles Into Alkyd Resin …

345

nanoparticles. The research successfully established that the incorporation of metallic nanoparticles into an alkyd resin can improve the drying process and other properties of automotive-based paint and coated aluminium plates. Radoman et al. [9] fabricated alkyd resin/TiO2 nanocomposites (NC) from short oil alkyd resin and TiO2 nanoparticles (NPs) surface modified with in situ developed imine ligand, based on oleylamine, or with three gallic acid esters and 3,4-dihydroxybenzaldehyde (Scheme 4). Sunflower oil, trimethylolpropane, and phthalic anhydride were employed in the synthesis of the short oil alkyd resin. Rheological examination showed that the generated alkyd resin/TiO2 nanocomposites (NCs) have a more elevated dynamic viscosity than pure alkyd resin. Applying several characterization techniques it was realized that the NCs has lower glass transition temperature, better barrier properties, enhanced hardness, comparable thermal properties, chemical resistance, oxidative stability as alkyd resin, and lower adhesion to the metal, except in the case of NC developed utilizing TiO2 NPs surface altered with imine, which presented more reliable adhesion to the metal than pure alkyd resin. With the better filler dispersibility, better chemical resistivity, and adhesion to the metal, as well as good barrier properties recorded from the results of this study, it can be inferred that NCs developed from TiO2 NPs surface modified with imine and short oil alkyd resin is an excellent candidate for protective coatings of metal substrates. A novel composite coating composed up of alkyd@lanthanide bis-phthalocyanine was examined by Deyab et al. [24] for its corrosion protection characteristics on carbon steel pipelines. Meaningful corrosion resistance was observed in steel in 0.5 M HCl solution. The performance of the alkyd-based resin was considerably improved by the incorporated lanthanide compound by decreasing water permeability and improving physico-mechanical properties/adhesion strength. Grozdanov et al. [28] generated two types of alkyd resin-based nanocomposite by incorporating different content of silicon oxide (SiO2 ) and titanium oxide (TiO2 ) nanoparticles (2 and 4wt/wt %) into alkyd resin separately using the solvent casting method. The SiO2 and TiO2 alkyd resin samples both exhibited improved flexibility, elasticity, and wear resistance of the nanocomposite coatings as the nanoparticle content increased to a certain point. However, the nanocomposite coatings with SiO2 nanoparticles exhibited a better abrasion resistance when compared to nanocomposites with the same content of TiO2 . The resistance to chemical attacks for both alkyd resin-based nanocomposite coatings prepared by Grozdanov et al. [28] in organic solvents, acids, and alkali media were observed to poor. The drying times of the synthesized alkyd samples were not investigated, even though the drying time is a very vital parameter to be considered when recommending resins in the coating industries. Sarmin et al. [29] investigated the catalytic effect of using strontium oxide/hydroxide [SrO/Sr(OH)2 ] nanoparticles on palm oil (PO)-based alkyd resin synthesized using alcoholysis-polyesterification technique. Palm oil (PO)-based alkyd resin was also synthesized using sodium hydroxide (NaOH) catalyst for comparative purpose. The degree of polymerization and the extent of polyesterification reaction were examined using a time-course acid value data. The dispersion of as-synthesized SrO/Sr(OH)2 nanoparticles in the reaction mixture developed a long-lasting suspension which in turn efficiently improved the rate of reaction for the alcoholysis and polyesterification

346

I. H. Ifijen et al.

process when compared to NaOH. The use of SrO/Sr(OH)2 as a catalyst for the alcoholysis reaction increased the rate of the reaction by decreasing the reaction time to 40 min compared to the use of NaOH-catalyst in the polyesterification process which usually gets to completion at about 120 min. Sumi et al. [30] incorporated PANIFe2 O3 nanoparticles into alkyd resin. They generated PANI-Fe2 O3 /alkyd composite showed good corrosion resistance that can be attributed to the formation of a dense passive layer and high coverage.

Conclusion This article reported the up-to-date notable attempts made by scientists towards the synthesis of non-agglomerated alkyd resins-based metallic nanocomposites for utilization in coating industries. We presented an extensive evaluation of the consequences of incorporating several types of metallic nanomaterials into an alkyd resin polymeric matrix. In addition, factors that can impact the performance of nanocomposite coatings were also discussed. To avoid the generation of agglomerated alkydbased metallic nanocomposites, appropriate mixing techniques and synthetic procedures should be employed. The prospect of generating coatings composites with new features using metallic nanomaterials to modify alkyd resin has immensely changed the world of coating. The data and the comprehensive report associated with this topic establish the potential of nanotechnology in improving the coating performance of alkyd-based metallic nanocomposites.

Incorporation of Metallic Nanoparticles Into Alkyd Resin …

347

Scheme 4 Schematic representation of surface modification of TiO2 NPs with DHBA and OA, and in situ formation of imine ligand [9]

Conflict of Interest On behalf of all authors, the corresponding author states that there is no conflict of interest.

References 1. Ifijen IH, Odi HD, Maliki M, Omorogbe SO, Aigbodion AI, Ikhuoria EU (2020) Correlative studies on the properties of rubber seed and soybean oil based-alkyd resins and their blends. J Coat Techn Res 18(5):459–467 2. Otabor GO, Ifijen IH, Mohammed FU, Aigbodion AI, Ikhuoria EU (2019) Alkyd resin from rubber seed oil/linseed oil blend: a comparative study of the blend properties. Heliyon 5(5)15:e01621 3. Malik M, Ikhuoria EU, Ifijen IH (2020) Extraction and physiochemical characterization of oils obtained from selected under-utilized oil bearing seeds in Nigeria. Chem Search J 11(1):110– 117 4. Maliki M, Ifijen IH (2020) Extraction and characterization of rubber seed oil. Int J Sci Eng Sci 4(6):24–27 5. Nkwor AN, Ukoha PO, Ifijen HI (2021) Synthesis of sulfonated sesamum indicum L. seed oil and its application as a fatliquor in leather processing. J Leather Sci Eng 3(16):1–13 6. Nkwor AN, Ukoha PO, Ifijen HI, Ikhuoria EU (2020) The use of sulfonated jatropha curcas oil for the processing of mechanically improved leather. Chem Afri 3:911–925 7. Ifijen HI, Nkwor AN (2020) Selected under-exploited plant oils in Nigeria: a correlative study of their physiochemical properties. Tanz J Sci 46(3):817–827 8. Jalili MM, Moradian S (2009) Deterministic performance parameters for an automotive polyurethane clear coat loaded with hydrophilic or hydrophobic nano-silica. Prog Org Coat 66:359–366

348

I. H. Ifijen et al.

9. Radoman TS, Džunuzovi´c JV, Spasojevi´c PM, Marinovi´c- MT, Jeremi´c KB, Popovi´c IG, Džunuzovi´c ES (2018) Preparation and Properties of short oil alkyd resin/Tio2 nanocomposites based on surface modified Tio2 nanoparticles. Polym Compos 39:1488–1499 10. Neouze MA, Schubert U (2008) Surface modification and functionalization of metal and metal oxide nanoparticles by organic ligands. Monatshefte fur Chemie 139(3):183–195 11. Elbasuney S (2014) Dispersion characteristics of dry and colloidal nano-titania into epoxy resin. Powder Tech 268:158–164 12. Kango S, Kalia S, Celli A, Njuguna J, Habibi Y, Kumar R (2013) Surface modification of inorganic nanoparticles for development of organic-inorganic nanocomposites—a review. Prog Polym Sci 38(8):1232–1261 13. Kugler S, Kowalczyk K, Spychaj T (2016) Influence of dielectric nanoparticles addition on electroconductivity and other properties of carbon nanotubes-based acrylic coatings. Prog Org Coat 92:66–72 14. Zamfirova G, Gaydarov V, Faraguna F, Vidovi´c E, Juki´c A (2016) Influence of carbon nanotubes functionalization on the mechanical properties of polymethacrylate nanocomposites. Colloids Surf A Physicochem Eng Asp 510:169–175 15. Susumu A, Akihiro F, Masami M, Morinobu E (2008) Excellent solid lubrication of electrodeposited nickel-multiwalled carbon nanotube composite films. Mater Lett 62:3545–3548 16. Dang C, Li J, Wang Y, Yang Y, Wang Y (2017) Influence of Ag contents on structure and tribological properties of TiSiN-Ag nanocomposite coatings on Ti–6Al–4V. J Chen Appl Surf Sci 394:613–624 17. Mallikarjuna HM, Ramesh CS, Koppad PG, Keshavamurthy R, Kashyap KT (2016) Effect of carbon nanotube and silicon carbide on microstructure and dry sliding wear behavior of copper hybrid nanocomposites. T Nonferr Metal Soc 26:3170–3182 18. Kucharska B, Popławski K, Jezierska E, Oleszak D, Sobiecki JR (2016) Influence of stirring conditions on Ni/Al2 O3 nanocomposite coatings. Surface Eng 32:457–463 19. García E, García-Urrutia I, Díez JA, Fornell J, Pellicer E (2013) Codeposition of inorganic fullerene-like WS2 nanoparticles in an electrodeposited nickel matrix under the influence of ultrasonic agitation. J Sort Electrochim Acta 114:859–867 20. Shirani A, Momenzadeh M, Sanjabi S (2012) Surfactant effect on electrochemical behavior of Co–Tio2 nanocomposite coatings. Surf Coat Tech 206:2870–2876 21. Bhanvase BA, Sonawane SH (2010) New approach for simultaneous enhancement of anticorrosive and mechanical properties of coatings: application of water repellent nano CaCO3 –PANI emulsion nanocomposite in alkyd resin. Chem Eng J 156:177–183 22. Deyab MA (2015) Effect of carbon nano-tubes on the corrosion resistance of alkydcoating immersed in sodium chloride solution. Prog in Org Coat 85:146–150 23. Ong HR, Khan MMR, Ramli R, RM, (2015) Effect of CuO nanoparticle on mechanical and thermal properties of palm oil based alkyd/epoxy resin blend. Procedia Chem 16:623–631 24. Deyab MA, Mele G, Al-Sabagh AM, Bloise E, Lomonaco D, Mazzetto SE, Clemente CDS (2017) Synthesis and characteristics of alkyd resin/m-porphyrins nanocomposite for corrosion protection application. Prog in Org Coat 105:286–290 25. Deyab MA, Słota R, Bloise E, Mele G (2018) Exploring corrosion protection properties of alkyd@lanthanide bis-phthalocyanine nanocomposite coatings. RSC Adv 8:1909–1916 26. Obaid (2017) Hyperbranched soya alkyd nanocomposite: a sustainable feedstock-based anticorrosive nanocomposite coatings. ACS Sustainable Chem Eng 5:9725–9734 27. Lotfizadeh H, Rezazadeh S, Fathollahi MR, Jokar J, Mehrizi AA and Soltannia B (2018) the effect of silver nanoparticles on the automotive-based paint drying process: an experimental study. Int J Adv Multi Eng Sci 2(1):7–14 28. Grozdanov A, Gentile G, Avella M, Dobreva T, Kotsilkova R (2019) Nanocomposite coatings based on alkyd resin with TiO2 and SiO2 nanoparticles. Mater Sci Eng 3(6):210–215 29. Sarmin S, Ethiraj B, Tarek M, Rengaraju B, Karim KMR, Huei Ruey Ong HR, Abdullah H, Khan MMR (2020) Palm oil derived alkyd resin synthesis catalyzed by Sro/Sr(OH)2 nanoparticles. JCR 7(6):2131–2138

Incorporation of Metallic Nanoparticles Into Alkyd Resin …

349

30. Sumi VS, Arunima SR, Deepa MJ, Ameen Sha M, Riyas, Meera MS, Saji VS and Shibli SMA (2020) PANI-Fe2 O3 composite for enhancement of active life of alkyd resin coating for corrosion protection of steel. Mater Chem Phys 2471:122881

Materials for Antireflection Coatings in Photovoltaics—An Overview Vishal Mehta, Cory Conkel, Andrew Cochran, and N. M. Ravindra

Abstract The ability to maximize the reflectance losses due to silicon is of paramount importance in the design, fabrication, and operation of silicon solar cells. Optimally designed antireflection coatings are required to improve photon collection in solar cells. For efficient performance, solar cells need to have low reflectance and high absorptance in the visible to near-infrared region. In this study, reflectance due to varying thicknesses of various dielectrics such as aluminum oxide (Al2 O3 ), silicon dioxide (SiO2 ), titanium dioxide (TiO2 ), magnesium fluoride (MgF2 ), and silicon nitride (Si3 N4 ) has been simulated in the range of visible to near infrared by mathematical modelling using MATLAB simulations. The results of the evolution of spectral properties, as a function of dielectric material thickness, on silicon substrates are presented. Keywords Silicon · Solar cell · Antireflection coating · Optical properties

Introduction Antireflection coatings (ARC) have been used in solar cells to improve the light collection efficiency, short circuit current density (Jsc ) and in some cases, for passivating the front surface of silicon [1]. Various ARC materials such as aluminum oxide (Al2 O3 ), silicon dioxide (SiO2 ), titanium dioxide (TiO2 ), magnesium fluoride (MgF2 ), and silicon nitride (Si3 N4 ) have been used as ARCs [2–5]. The ARC coatings V. Mehta (B) · C. Conkel · A. Cochran Ohio Northern University, Ada, OH, USA e-mail: [email protected] C. Conkel e-mail: [email protected] A. Cochran e-mail: [email protected] N. M. Ravindra New Jersey Institute of Technology, Newark, NJ, USA © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_32

350

Materials for Antireflection Coatings …

351

can be applied as single layer, double layer (DARC), or triple/multilayer antireflection coatings (MARC). These coatings can be manufactured using different fabrication techniques such as sol–gel/spin-coating, atmospheric pressure chemical vapor deposition, thermally grown [6], thermal evaporation, plasma-enhanced chemical vapor deposition [7], reactive evaporation, e-beam evaporation, liquid phase deposition [8], and magnetron sputtering. Each layer of ARC is tailored for a specific wavelength of the solar spectrum. Multiple coatings are used to increase the amount of incoming light by increasing the spectral absorptance. In conventional solar cells, further increase in absorptance is achieved by using textured surface. Light reflection from an antireflection layer coated solar cell is a function of refractive index, layer thickness, light incident angle, incident light wavelength, and light polarization [9]. Optical properties of antireflection coatings on silicon can be modelled by either using Transfer matrix theory modelling [10], Airy’s expression, Differential Evolution (DE) algorithm [11], Multi-rad [12], Dupoisot & Morizet, method [13], etc. In the present study, optical properties of single layer and double layer coating stacks of various candidate materials on silicon have been simulated. Plots of reflectance, transmittance, and absorptance in the wavelength range from 300 to 900 nm have been presented and analyzed.

Optical Properties The fundamental understanding of the optical properties of semiconductors have been increasing over the years [14]. However, there is lack of sufficient available data for different stacks of coatings and underlying semiconductors. ARC on solar cells are comparable to those used in other applications such as filters, lenses, photonics. They consist of a thin layer of dielectric material, with a specially chosen thickness so that interference effects in the coating cause the wave reflected from the antireflection coating top surface to be out of phase with the wave reflected from the semiconductor surface. These out-of-phase reflected waves destructively interfere with one another, resulting in zero net reflected energy. The interaction of light in a material depends on the incident photon energy (E) or wavelength of light (λ), as well as on the material properties and device fabrication conditions [15]. The relationship between E (eV) and λ is given by: E (eV) =

1.24 (micron) λ

Optical constants (refractive index n and extinction coefficient κ) describe the optical behavior of the material. The refractive index (n) is the ratio of the speed of light in a vacuum to the speed of light in material. The extinction coefficient (κ) is a measure of how much light is absorbed in the material [16]. Together, they can be stated as the real and imaginary parts of the complex index of refraction (N) as:

352

V. Mehta et al.

N = n−iκ The thickness of the antireflection coating is chosen so that the wavelength in the dielectric material is one quarter the wavelength of the incoming wave. For a quarter wavelength antireflection coating of a transparent material, with a refractive index n1 and light incident on the coating with a free-space wavelength λ0 , the thickness t 1 which causes minimum reflection is calculated by: t1 =

λ0 4n 1

Reflection is further minimized if the refractive index of the antireflection coating is the geometric mean of that of the materials on either side, that is, glass or air and the semiconductor. This is stated by: n1 =

√ n0n2

Calculations of the optical properties were performed for normal incidence and room temperature. The reflectivity, R, between two materials of different refractive indices, is determined by:  R=

n 0 − n Si n 0 + n Si

2

where n0 is the refractive index of the surroundings and nSi is the complex refractive index of silicon. In order to calculate the optical properties of different types of stacks, two types of stacks were designed: one layer of antireflection coating and double layer antireflection coating on silicon. The interface at the double layer was assumed to be abrupt. The refractive index within a given layer was considered to be constant. Figure 1 is the set up used for equations for DARC. For single layer, similar but simpler equations were used. Reflectance, Transmittance, and absorptance curves were calculated with coatings and silicon together as a unit. A series of parameters were used: r1 , r2 , r3 , θ1, and θ2 [17, 18]. The refractive index of air is represented as n0 , the first antireflection layer has a refractive index of n1 and a thickness of t1 , the second antireflection layer below the first has a refractive Fig. 1 Schematic of the layers for antireflective coating

Air with refractive index of n0 First layer with refractive index n1 Second layer with refractive index n2 Silicon with refractive index n3

Materials for Antireflection Coatings …

353

index of n2 and a thickness of t2 , and the silicon has a refractive index of n3 . The equation was modified to accommodate single antireflection layer. r1 =

n0 − n1 n0 + n1

r2 =

n1 − n2 n1 + n2

r3 =

n2 − n3 n2 + n3

θ1 =

2π n 1 t1 λ

θ2 =

2π n 2 t2 λ

The reflectivity is then calculated from the above parameters using the following formula:   R = r 2  =

r12 + r22 + r32 + r12 r22 r32 + 2r1 r2 (1 + r32 ) cos 2θ1 + 2r2 r3 (1 + r12 ) cos 2θ2 + 2r1 r3 cos 2(θ1 + θ2 ) + 2r1 r22 r3 cos 2(θ1 + θ2 ) 1 + r12 r22 + r12 r32 + r22 r32 + 2r1 r2 (1 + r32 ) cos 2θ1 + 2r2 r3 (1 + r12 ) cos 2θ + 2r1 r3 cos 2(θ1 + θ2 ) + 2r1 r22 r3 cos 2(θ1 − θ2 )

The absorption coefficient measures how much light of a given color (wavelength) is absorbed by a material of given thickness and is calculated by: α=

4π κ λ

Here, κ is the extinction coefficient corresponding to the respective wavelength. Transmittance at each interface was then calculated by:   T = (1 − R) e−αt The total Transmittance was calculated using: T = T1*T2*T3. Absorptance was then calculated by: A = (1− R − T)

354

V. Mehta et al.

Fig. 2 Reflectance-Transmittance-Absorptance (R-T-A) graphs for bare (i.e. uncoated silicon) untextured silicon

Results and Discussion Bare Silicon As a first step, Reflectance-Transmittance-Absorptance (R-T-A) graphs for bare (i.e. uncoated silicon) untextured silicon were plotted. In Fig. 2, one can see that there is more than 30 percent reflectance in the visible wavelength range.

Single Layer Antireflection Coatings In order to find the optimal thickness values for SiO2 , Si3 N4 , and Al2 O3 , ReflectanceTransmittance-Absorptance (R-T-A) plots for various thickness were simulated for visible wavelength range. Figure 3 shows the R-T-A plot of the single layer SiO2 ARC on silicon. Here, it is seen that, for various thickness of SiO2 (40-70 nm), the reflectance values have been reduced compared to bare silicon. The lowest reflectance

Materials for Antireflection Coatings …

355

Fig. 3 R-T-A plot of SiO2 ARC on silicon

can be found at 480 nm. SiO2 , as a native oxide, will also act as a passivating layer for surface defects on silicon. Figure 3 indicates that SiO2 /Si does not have zero reflectance at any wavelength in the selected range. Figure 4 shows R-T-A values for Si3 N4 on silicon. It can be seen that zero reflectance is found between 70–80 nm. This is also the wavelength range where peak power of the solar spectrum resides. Indeed, most silicon solar cells have Si3 N4 ARC thickness of ~75 nm. Low frequency plasma-enhanced chemical vapor deposition (L-PECVD) of Si3 N4 also results in similar properties. Figure 5 is the plot of Al2 O3 on silicon. Here, it can be seen that, for 110– 120 nm thick ARC layer of Al2 O3 , the reflectance is lowest. Also, it can be seen that absorptance values for the same wavelength range are near 1. Optical property graph indicates that Al2 O3 can also be used as band pass filter for the same thickness range. It can be noted that the zero reflectance occurs at more than 700 nm. This indicates its less-than-optimal ARC compared to Si3 N4 . Table 1 shows the reflectance peaks and reflectance valleys of various stacks of single layer ARCs on silicon, studied in the spectral range of 300–1000 nm. Table 1 indicates that, in order to obtain minimum reflectance, around 70 nm of TiO2 needs to be deposited. The reflectance minima are close to 600 nm (i.e. peak solar

356

V. Mehta et al.

Fig. 4 R-T-A values for Si3 N4 on Silicon

power). In the case of SiO2 , a coating thickness of 70 nm yields reflectance minima of 0.14. In the case of Si3 N4 , between 70–80 nm of coating thickness is required to achieve reflectance minima closer to near 600 nm. 80 to 90 nm thick Al2 O3 -based coating thickness is necessary to attain reflectance minima. These results suggest Si3 N4 , Al2 O3 and TiO2 are good candidates for single layer antireflection coatings on Silicon.

Dual Layer Antireflection Coatings R-T-A simulation plots of two layer of ARCs on silicon were also determined. It is important to note that the two stacks and silicon are being considered as a unit for calculating the optical properties. Hence among R-T-A plots, absorptance is mostly zero between 300–900 nm. Figure 6 shows the R-T-A plots for Si3 N4 /TiO2 /Silicon. In this simulation, the TiO2 thickness was kept constant (70 nm) and Si3 N4 thickness was varied between 60–90 nm in 10 nm increments. It can be seen that at about 90 nm thick Si3 N4 , the dip in reflectance curve can be seen at around 502 nm. This is closer to the dip seen when only Si3 N4 is used as ARC. The simulation data indicates that,

Materials for Antireflection Coatings …

357

Fig. 5 R-T-P plot of single layer of Al2 O3 on top of silicon

compared to Si3 N4 /Silicon stack (Fig. 4), the Si3 N4 /TiO2 /Si stack is not giving any real advantage in reducing the reflectance values. Figure 7 shows the R-T-A plots for SiO2 /Si3 N4 /silicon. In this experiment, the Si3 N4 thickness was kept constant (80 nm) and SiO2 thickness was varied between 10–40 nm in 10 nm increments. The silicon thickness was kept constant at 125 microns. It can be seen that R values were almost zero between 680–820 wavelength range. This is a promising result for PV cell manufacturers due to the improvements in the spectral performance. Figure 8 shows the R-T-A plots for Al2 O3 /TiO2 /Si structure. In this simulation, the TiO2 thickness was kept constant (89 nm) and SiO2 thickness was varied between 70– 130 nm in 15 nm increments. The silicon thickness was kept constant at 125 microns. It can be seen that reflectance is zero at ~600 nm when the Al2 O3 thickness is 115 nm. The R values increase with wavelength. When compared to Fig. 3, Al2 O3 /TiO2 layer does not exhibit any benefit for solar cell applications. Figure 9 shows the R-T-A plots for MgF2 /Si3 N4 /Si structure. In this simulation, the Si3 N4 thickness was kept constant (70 nm) and MgF2 thickness was varied between 100–125 nm in 12 nm increments. The silicon thickness was kept constant at 125 microns. From a materials perspective, MgF2 is abrasion resistant. As seen from the graph, the reflectance values are extremely low in a broad wavelength range for

358

V. Mehta et al.

Table 1 Single layer ARC on Silicon Coating layer type Max wavelength (nm) Reflectance Min wavelength (nm) Reflectance 70 nm TiO2

378

0.54

671

0.05

80 nm TiO2

403

0.46

766

0.05

90 nm TiO2

445

0.41

359

0.04

100 nm TiO2

330

0.44

379

0.01

40 nm SiO2

370

0.38

300

0.19

50 nm SiO2

370

0.31

300

0.15

60 nm SiO2

370

0.27

330

0.17

70 nm SiO2

370

0.28

480

0.14

60 nm Si3 N4

310

0.38

492

0.00

70 nm Si3 N4

310

0.45

568

0.00

80 nm Si3 N4

370

0.53

645

0.00

90 nm Si3 N4

370

0.55

724

0.00

100 nm Si3 N4

380

0.52

803

0.00

80 nm Al2 O3

300

0.44

555

0.03

90 nm Al2 O3

370

0.49

615

0.03

100 nm Al2 O3

370

0.54

679

0.02

110 nm Al2 O3

370

0.55

741

0.02

120 nm Al2 O3

376

0.53

808

0.02

stack containing 112 nm thick MgF2 layer. Specifically, zero reflectance is observed at 466 nm and in the wavelength range of 900-1000 nm and 11% reflectance at 633 nm. The MgF2 /Si3 N4 dual layer coating on silicon is a promising candidate as antireflection coating. Table 2 presents the reflectance peaks and reflectance valleys of various stacks of double layer ARCs on silicon in the spectral range of 300–1000 nm. The Si3 N4 /TiO2 stack on silicon does not exhibit any advantages compared to Si3 N4 on silicon. The 10 nm SiO2 /80 nm Si3 N4 stack on silicon shows zero reflectance which is a promising result. 115 nm Al2 O3 /89 nm TiO2 stack on silicon can be an alternative to single layer Si3 N4 /silicon stack as reflectance is zero at 600 nm. Based on the simulation results, the 30% reflection loss from the bare/polished silicon (Fig. 2) can be effectively reduced by selecting optimal thickness of the antireflection coatings tailored for broad solar spectrum. Al2 O3 , TiO2 , MgF2, and Si3 N4 are good candidates for effective suppression of reflectance of silicon. The results obtained can also be useful in other applications that utilize optical coatings such as optical detectors, filters, imagers, lenses, optical coatings, photonic crystals, sensors, and waveguides.

Materials for Antireflection Coatings …

Fig. 6 The R-T-A plots for Si3 N4 /TiO2 /Silicon

359

360

Fig. 7 R-T-A plots for SiO2 /Si3 N4 /Silicon

V. Mehta et al.

Materials for Antireflection Coatings …

Fig. 8 R-T-A plots for Al2 O3 /TiO2 /Si structure

361

362

Fig. 9 R-T-A plots for MgF2 /Si3 N4 /Si structure

V. Mehta et al.

Materials for Antireflection Coatings …

363

Table 2 Dual layer coatings on silicon Coating layer type

Wavelength (nm) Reflectance Wavelength (nm) Reflectance max min

60 nm Si3 N4 /70 nm TiO2

365

0.37

427

576

0.22

0

0

370

0.49

453

0

619

0.21

0

0

370

0.54

477

0

650

0.2

660

0.2

379

0.54

502

0.01

660

0.2

10 nm SiO2 /80 nm Si3 N4

370

20 nm SiO2 /80 nm Si3 N4 30 nm SiO2 /80 nm Si3 N4

70 nm Si3 N4 /70 nm TiO2 80 nm Si3 N4 /70 nm TiO2 90 nm Si3 N4 /70 nm TiO2

0

0

0

0.54

675

0

370

0.52

704

0

370

0.49

734

0

40 nm SiO2 /80 nm Si3 N4

376

0.43

765

0

70 nm Al2 O3 /89 nm TiO2

380

0.28

341

0.01

633

0.15

474

0.05

393

0.35

520

0.03

694

0.11

348

0.04

408

0.4

562

0.01

752

0.09

355

0.05

425

0.42

600

0

802

0.08

362

0.03

444

0.42

368

0

332

0.19

636

0

310

0.43

888

0.01

550

0.11

432

0.06

310

0.45

950

0.02

588

0.1

449

0.03

328

0.44

466

0.02

633

0.11

0

85 nm Al2 O3 /89 nm TiO2 100 nm Al2 O3 /89 nm TiO2 115 nm Al2 O3 /89 nm TiO2 130 nm Al2 O3 /89 nm TiO2 100 nm MgF2 /70 nm Si3 N4 112 nm MgF2 /70 nm Si3 N4 125 nm MgF2 /70 nm Si3 N4

0

364

V. Mehta et al.

Conclusion Optical properties relating to varying thicknesses of different dielectrics such as aluminum oxide (Al2 O3 ), silicon dioxide (SiO2 ), titanium dioxide (TiO2 ), magnesium fluoride (MgF2 ), and silicon nitride (Si3 N4 ) have been simulated in the wavelength range of visible to near infrared by mathematical modelling using MATLAB simulations. The results for the single layer and dual layer coatings show that SiO2 , Si3 N4 MgF2 , and Al2 O3 are excellent candidates for reducing reflectance of silicon either as single layer or as dual layer. Acknowledgements Authors thank the Department of Mechanical Engineering, Ohio Northern University, Ada, Ohio for its support of this study. This study has been carried out as part of an independent study course (ME 2951) entitled “Simulation of optical properties of semiconductor multilayers from extreme ultraviolet to far infrared”. Senior Undergraduate Student, Cory Conkel, is the beneficiary of this course.

References 1. Zhao J, Wang A, Green MA (1994) Double layer antireflection coating for high-efficiency passivated emitter silicon solar cells. IEEE Trans Electron Devices 41(9):1592–1594. https:// doi.org/10.1109/16.310110 2. Raut HK, Ganesh VA, Nair AS, Ramakrishna S (2011) Anti-reflective coatings: A critical, in-depth review. Energy Environ Sci 4(10):3779–3804. https://doi.org/10.1039/C1EE01297E 3. Eshavarz Hedayati M, Elbahri M (2016) Antireflective coatings: conventional stacking layers and ultrathin plasmonic metasurfaces a mini-review. Materials 9(6):497. https://doi.org/10. 3390/ma9060497 4. Rajan G, Karki S, Collins RW, Podraza NJ, Marsillac S (2020) Real-time optimization of antireflective coatings for CIGS solar cells. Materials (Basel, Switzerland) 13(19):4259. https:// doi.org/10.3390/ma13194259 5. Kaminski PM, Lisco F, Walls JM (2014) Multilayer broadband antireflective coatings for more efficient thin film CdTe solar cells. IEEE J Photovolt 4(1):452–456. https://doi.org/10.1109/ JPHOTOV.2013.2284064 6. Richards BS (2004) Comparison of TiO2 and other dielectric coatings for buried-contact solar cells: a review. Prog Photovolt: Res Appl 12:253–281. https://doi.org/10.1002/pip.529 7. Chen Z, Sana P, Salami J, Rohatgi A (1993) A novel and effective PECVD SiO/sub 2//SiN antireflection coating for Si solar cells. IEEE Trans Electron Devices 40(6):1161–1165. https:// doi.org/10.1109/16.214744 8. Wuu D-S, Lin C-C, Chen C-N, Lee H-H, Huang J-J (2015) Properties of double-layer Al2 O3 /TiO2 antireflection coatings by liquid phase deposition. Thin Solid Films 584:248–252. https://doi.org/10.1016/j.tsf.2015.01.042 9. Sikder U, Zaman MA (2016) Optimization of multilayer antireflection coating for photovoltaic applications. Opt Laser Technol 79:88–94. https://doi.org/10.1016/j.optlastec.2015.11.011 10. Abeles F (1948) Transmission of light by a system of alternate layers C. R Acad Sci 226:1808 11. Hebb JP, Jensen KF (1996) The effect of multilayer patterns on thermal stress during rapid thermal processing. MRS Online Proc Libr 429(1):43–49. https://doi.org/10.1557/PROC429-43 12. Storn R, Price K (1995) Differential evolution—a simple and efficient adaptive scheme for global optimization over continuous spaces

Materials for Antireflection Coatings …

365

13. Dupoisot H, Morizet J (1979) Thin film coatings: algorithms for the determination of reflectance and transmittance, and their derivatives. Appl Opt 18(15):2701–2704. https://doi.org/10.1364/ AO.18.002701 14. Shanmugam N, Pugazhendhi R, Madurai Elavarasan R, Kasiviswanathan P, Das N (2020) Anti-reflective coating materials: a holistic review from PV perspective. Energies 13(10):2631. MDPI AG. Retrieved from http://dx.doi.org/10.3390/en13102631 15. Leqi L, Ravindra NM (2020) Simulation of optical properties of semiconductor multilayers from extreme ultraviolet to far infrared. Mater Sci Eng 4(5):131–137. https://doi.org/10.15406/ mseij.2020.04.00139 16. Bhandari KP, Lamichhane A, Maenle T, Bastola E, Ellingson RJ (2019) Optical properties of organic inorganic metal halide perovskite for photovoltaics. In: 2019 IEEE 46th Photovoltaic Specialists Conference (PVSC), Chicago, IL, USA, pp 0359–0362. https://doi.org/10.1109/ PVSC40753.2019.8981333 17. Honsberg CB, Bowden SG (2019) Photovoltaics education website. https://www.pveducati on.org. Accessed 6 July 2020 18. Wang EY, Yu FTS, Sims VL, Brandhorst EW, Broder JD (1973) Optimum design of antireflection coating for silicon solar cells. 10th IEEE Photovoltaic Specialists Conference, pp 168–171

Nanostructured Graphene Thin Films: A Brief Review of Their Fabrication Techniques and Corrosion Protective Performance Ikhazuagbe H. Ifijen, Oscar N. Aghedo, Ifeanyi J. Odiachi, Stanley O. Omorogbe, Ekebafe L. Olu, and Innocent C. Onuguh Abstract Graphene oxide has attracted so much attention over the last few years owing to its astonishing features and has proven to have a major contribution to the anticorrosive coating industry. A great deal of this attention is motivated by the necessity to realize additional functionalities, to boost the anti-corrosion performance of the graphene oxides, and to eventually lengthen the service life of metallic structures. This review covers the properties, fabrication techniques, corrosion protective performance, realistic problems, and modification of graphene corrosion protective films. Keywords Anticorrosive · Coating · Nanostructures · Graphene oxide · Films

Introduction Material efficiency can be enhanced by depositing several nanostructured coatings on the surface area of a matrix for numerous utilizations [1]. As such, remarkable growth has been displayed by these nanostructured materials in coatings applications due to two major factors: (1) availability of numerous types of nanostructured and I. H. Ifijen (B) · S. O. Omorogbe Department of Research Operations, Rubber Research Institute of Nigeria, Iyanomo, Benin City, Nigeria O. N. Aghedo Department of Laboratory Technology, Faculty of Life Sciences, University of Benin, P.M.B. 1154, Benin City, Nigeria I. J. Odiachi Department of Science Laboratory Technology, Delta State Polytechnic Ogwashi-Uku, Ogwashi-Uku, Nigeria E. L. Olu Department of Chemistry, University of Lagos, Lagos State, Nigeria I. C. Onuguh Department of Chemistry, Igbinedion University, Okada, Edo State, Nigeria © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_33

366

Nanostructured Graphene Thin Films: A Brief Review …

367

(2) drastic improvements in techniques employed in controlling the structures of coating at the nano-scale. In addition, the possibility of nanotechnology in solving several challenges resulting from the poor performance of coatings applied to a vast range of products is another vital reason for this growth [2]. Due to the small particle sizes of 100 nm or less, nanostructured coating materials can be utilized for the achievement of interaction between coating and surface, better and higher durability of the coating and higher opacity [3]. These coatings proffer much more improved material and processing characteristics than the conventional coatings (e.g. high elasticity, increased indentation resistance, no expansion after contact with water, high water vapour permeability, and fast-drying) [4–10]. Coatings can be utilized in interior furnishings, interior and exterior house paints, all kinds of transportation vehicles and structures (aeroplanes, bridges, automobiles, marine vessels, spacecraft, road markings, etc.), a wide variety of industrial and non-industrial maintenance coatings and glass and facade coatings for high-rise buildings, ceramics, and other proprietary applications specific to industry [11]. The efficacy of nanostructured coatings in materials provides specialized functionality to the product (e.g., insulation, heat reflection and water repellency, improve product’s esthetic appeal, and protect the substrate from a wide range of abuses (e.g., damage due to impact or scratches, long-term weathering, bio-fouling, and corrosion). Previous studies have developed anticorrosive coating materials such as plant seed oils based materials such as alkyd resin [12, 13], plant seed oil-based fatliquor [14, 15], and nanostructure materials like polymeric nanostructures [4–8], metallic oxides nanostructures [1], grapheme nanostructures [2], carbon nanotubes, silica-titania nanostructures [11], etc. Graphene is a newly found allotrope form of carbon that inspired the scientific world and expanded the scope of applicability for composite materials due to its distinct high mechanical, electrical, thermal properties, and specific surface area [16]. The possible application of graphene is limited because of its tendency to agglomerate when used in composite formulations, its poor solubility and the preparation techniques are very expensive [16]. For graphene oxide nanostructures to be a very effective anticorrosive coating material, they must be modified by incorporating other compounds into their matrix. Therefore, graphene oxide was widely adopted in the last few years due to its exceptional assets and demonstrated to have a meaningful contribution to composite materials. This study presents a brief review of graphene oxide nanostructured thin films, focusing in particular on their coating features; the objective of this review is to give a concise view of the functional characteristics and the associated applicability of the aforementioned nanostructured coatings.

368

I. H. Ifijen et al.

Properties of Graphene Graphene is among the several allotropic carbon varieties, viewed as single-layer carbon atoms, arrayed in a honeycomb lattice. It is a material that exhibits exceptional characteristics, i.e., high thermal conductivity, the high mechanical strength of 1100 GPa, high surface area, very dense network, high carrier mobility, and others. As a result of the aforementioned properties, graphene can be utilized in various branches of industry, including de-oiling high salinity emulsions, electronics, catalysis, gas permeability, biomimetics, electrodialysis, energy, hydrogen, filtration, as anti-corrosion layer and storage, and biologic sensors [18–23]. Graphene possessed so many properties that make it an extraordinary protective layer. For instance, the optical appearance of surfaces coated with graphene can be retained as a result of the high light transmission of graphene (97%) in a wide range of the electromagnetic spectrum. Furthermore, the flexible nature of graphene allows graphene coatings to conform to the curvature and roughness of the surface of the substrate. The aromatic C=C bond network extends across the entire basal plane due to the delocalization of the electron cloud in graphene, making graphene thermodynamically stable. This chemical inertness is a vital property that favours its usefulness as a protective coating. When exposed to superheated water and high pressure, graphene has been established to be more stable than diamond [24]. Graphene is impermeable, apart from its chemical inertness, the dense graphene lattice function as a barrier, restricting even helium atoms, the smallest of atoms [25].

Fabrication Techniques of Graphene and Graphene-Based Materials The preparation of the graphene films was usually carried out on metal surfaces, such as steel, aluminium alloy, copper alloy, magnesium alloy and nickel, titanium. It was first isolated successfully by successive peelings using scotch tape via mechanical exfoliation of graphite [26]. This technique generated one atom thick, high-quality, single-crystal graphene flakes. Notwithstanding, its throughput is unreasonably low, and its upscaling is irrational. One of the first synthetic options that came into being was that of epitaxial graphene fabricated by silicon carbide (SiC) graphitization [27]. The sublimation of Si atoms occurs, with the remaining C atoms undergoing graphitization by the heating up of single-crystal silicon carbide at atmospheric pressure or in ultrahigh vacuum conditions [28]. This process can take place on both the Cterminated face and the Si-terminated one. In the earlier case, the interaction with the substrate is to a great extent weak, while in the second case, a C-rich buffer layer covalently bonded to the substrate is present beneath the fabricated graphene [29]. As a result of this synthetic method, it was swiftly concluded that the generated graphene is of extraordinary feature, with charge carrier mobility of about 27,000 cm−2 V−1 s−1

Nanostructured Graphene Thin Films: A Brief Review …

369

for graphene generated on the C-terminated face [30]. Nevertheless, the major shortcoming of graphene synthesized from SiC is the expensive nature of the substrates [31] and limitations in terms of sizes and difficulties in micromachining the fabricated material [32]. In recent times the generation of graphene films on metal surfaces have been achieved using mechanical transfer technology, chemical vapour deposition (CVD), rapid thermal annealing, pyrolysis, electrodeposition, etc. however, the simplicity of CVD make it the most universally utilized technique [33]. Graphene films synthesized using the CVD technique is broad in area, extraordinary in quality, and simple to transfer. More notably, the CVD-generated graphene films could be directly applied to prevent metals corrosion. This technique is one of the best methods for the surface segregation of carbon atoms into graphene domains and continuous layers from transition metals [34]. For instance, Yu et al. [33] generated super-high graphene layers by exposing Ni foil to a mixture of H2 , Ar, and CH4 at 1000 °C (atmospheric pressure). The process can be described based on the decomposition of hydrocarbon gas, accompanied by the diffusion of carbon atoms into the metal foil. Carbon segregation to form graphene layers at the surface is achieved by controlled cooling of the substrate. Likewise, it was also revealed that a support layer (silicone rubber) applied to the grown film can be used to transfer the graphene onto insulating substrates, accompanied by the etching of the metal and the utilization of the graphene/support stack onto the preferred substrate. The features of the substrate material have a way of influencing the characteristics and structure of the fabricated graphene films during the implementation of the CVD process. There is a strong relationship between the bonding strength of metal crystal faces and graphene to the distance of the carbon atoms from the metal. As a consequence, the stability of the structures of graphene films fabricated on several metal surfaces varies from one another. In comparison with gold, palladium, or silver, the force existing between nickel, copper, carbon, and cobalt atoms is weaker, so the latter was the major substrate for the fabrication of graphene films by CVD [35, 36]. Though to fabricate high-quality graphene films, the CVD technique could only be applied on metal substrates such as nickel, copper, cobalt, etc., the mechanical transfer technique favours the covering of the surface of metallic materials by the graphene film. For instance, the copper substrate is generally known to have low carbon solubility and catalytic effect on the hydrocarbon precursor breakdown and is also capable of facilitating the growth of continuous single-layer graphene (SLG) films (Fig. 1 [36]). In a typical synthesis, the generation of graphene on a copper substrate using the CVD technique is achieved at temperatures close to the latter’s melting point. Conversely, the use of low-temperature assisted microwave (300– 400 °C) plasma growth has been demonstrated to assist the decomposition of the hydrocarbon precursors [37]. Presently, the CVD technique permits the generation of graphene domains with arbitrarily huge single-crystal films and above millimetric diameters [38, 39] as well as graphene with high charge carrier mobilities (350,000 cm2 V−1 s−1 ) [3] with growth at both low and possible high atmospheric pressures [40]. At this point, it must be noted that, in several cases, graphene developed using a catalyst or epitaxially have to be transferred onto a dissimilar substrate where it can be processed or directly

370

I. H. Ifijen et al.

Fig. 1 a Schematic representation of the CVD process of graphene growth on copper. b Schematic representation of the electrochemical bubbling transfer process

incorporated for the preferred use. A good illustration of this process is the deposition of graphene on top of insulating substrates for lithographic production of electronic devices, or onto substrates that are transparent for optical utilizations. For CVDgenerated graphene on a copper substrate, the graphene can be separated from the copper or the copper can either be etched away [41]. The former can be achieved electrochemically, by producing bubbles of hydrogen gas, via hydrolysis, between copper and graphene, exfoliating the latter from the former [42]. On the other hand, the intercalation of water molecules between the substrate and graphene can be carried out to weaken their interaction in order to collect the sample by means of a stamp [43]. Whichever way, a supporting material must be employed in most cases to prevent damage to the graphene film following its removal from the copper. Poly (methyl methacrylate) (PMMA), which is regularly spin-coated on the as-grown graphene at the commencement of the transfer process, is the most generally applied supporting material. Nevertheless, the PMMA residues which are usually hard to completely eliminate are left behind by the taking away of the PMMA support layer after the completion of the transfer process, typically completed by inserting the transferred sample in acetone [44]. The characteristics of graphene are influenced by these residues, principally by forming p-type doping. To avoid this problem several studies have taken advantage of plasma treatments and high-temperature annealing [45]. Others have avoided the use of PMMA in general by using other supporting materials that can be detached without leaving any traces of residues [46].

Corrosion Protective Performance of Graphene Corrosion Films One of the major roles of an anti-corrosion coating agent is to act as a hindrance between the metal surface and the corrosive electrolyte by the inhibition of the formation of chemical compounds that promote corrosion. Pu et al. fabricated graphene

Nanostructured Graphene Thin Films: A Brief Review …

371

films on the surface of nickel-plated stainless steel and stainless steel via the CVD technique [47]. Microscopic analysis of the nickel-plated stainless steel covered with graphene films was observed to 100% coverage, but only a minute quantity of graphene was seen on the stainless steel surface. The difference in the graphene films coverage was attributed to the reduced formation of metal carbides facilitated by the nickel plating layer. This observation, therefore, implies that the problem of poor graphitization of stainless steel surfaces has been solved. By investigating polarization curves, it was established that the nickel-plated stainless steel coated with graphene films has excellent corrosion resistance, and its corrosion currents were observed to be merely one-fifth of the nickel-plated stainless steel that was not coated with graphene. The electrochemical corrosion of copper was reduced by 1.5 orders of magnitude when Raman et al. generated graphene films on a copper surface using the CVD method [48]. After comprehensive investigations via polarization curves and electrochemical impedance spectroscopy, graphene film was observed to considerably amplify the impedance of copper in chloride ion solutions, reducing the cathode and anode currents by a magnitude of 2. An interesting thing about this is that the graphene films on the surface of the copper can also be transferred mechanically to other metal surfaces such as silver (Fig. 2) and nickel [49, 50]. Prasai establishes that the corrosion rate of nickel covered with four layers of graphene films in the sodium sulfate solution was four times lesser than that of uncovered nickel and the corrosion rate of copper covered with graphene film in sodium sulfate solution was seven times lower than that of uncovered copper [49]. The increased number of layers of graphene films drastically enhanced the corrosion resistance of nickel. Electrochemical impedance spectroscopic analysis and other analyses revealed that the graphene films principally acted as physical barriers between the corrosive medium and metal, thereby protecting the metal from corrosion. Although the unstable nature of silver promotes its easy oxidization and corrosion, silver films have been extensively employed in the optical field. The regular protective coatings can be used to

Fig. 2 Preparation and corrosion inhibition effect of graphene film on metal nickel

372

I. H. Ifijen et al.

inhibit the corrosion of silver, but can also affect the optical properties in the process. Zhang generated transparent, ultra-thin protective monolithic graphene films on the copper surface using by CVD technique and then used a mechanical transfer method to graft them onto the surface of silver [50]. The corroding rate of silver covered with graphene films was observed to be more than sixty times lighter than that of uncovered silver. Apart from showing a strong barrier to water oxygen and other corrosive media, it also shows tremendous transparency. As a consequence, the intrinsic optical features of silver covered with graphene films are retained. At the same time, the oxidation resistance and chemical stability in harsh environments are improved. The inhibition of corrosion under biological action can also be achieved using graphene films. The corrosion of copper in different biological environments was suppressed by a bio-friendly graphene film generated on a copper surface via the CVD method [51]. By analyzing the biological survival rate, it was established that the survival rate of cells on the uncovered copper surface was roughly zero after one day, while a 100% survival rate of cells was observed for the surface covered with graphene. This is an indication that the graphene films hindered the corrosion of copper and drastically lowered the production of copper ions which could exterminate the microbes. Furthermore, it was observed that the assembly of mercapto derivatives on graphene films successfully hindered corrosive media diffusion at defects in graphene films. This to a great extent enhanced the shielding effect of graphene film on copper.

Realistic Problems of Graphene Films The corrosion rate of metals is considerably decreased by coating with a graphene film, because of its barrier influence on corrosive mediums (Fig. 3). Although graphene films are proficient in blocking corrosive particles and not easily damaged

Fig. 3 Barrier effect on corrosive media of graphene film

Nanostructured Graphene Thin Films: A Brief Review …

373

by corrosive media, the presence of defects and cracks in the films can cause corrosion of the metal surfaces after some time [49, 52]. The existence of these defects reduces the timeline of the protective performance of the graphene films on metals. To overcome this issue, researchers have improved on the long-term protective capability of the graphene films by using atomic layer precipitation technology to repair the defects [52]. A study concluded that since it is almost unattainable to prevent defects in the fabrication of thin-film, applying graphene films as a long-term anti-corrosion technology is not practicable [53]. This shows that there is a huge gap between the performance of graphene in metal protection and the actual demand, particularly in long-term protection.

Modification of Graphene Corrosion Protective Film Despite the practical problems associated with graphene films, several researchers attempted to improve the corrosion protective performance of graphene films. The obtained results were very hopeful. Current investigations in the fabrication of coatings centre on the approach of combining dissimilar corrosion inhibition mechanisms in [54] a synergetic approach through a single coating system aiming to boost the life span of the structural components and to proffer shielding against corrosive environments. Incorporating graphene with appropriate nanostructured materials can be a useful means to regulate and reinforce the anti-corrosion assets and adsorption features of graphene to function as an exceptional corrosion inhibitor. The resistance to external factors and barrier effect can be improved by inorganic compounds incorporated onto the surface of graphene oxide [55]. Ye et al. [55] generated a covalently improved polyhedral oligomeric silsesquioxane–graphene oxide (POSS-GO) for application in epoxy composite coatings in a marine environment for the inhibition of corrosion. Comparison of the neat epoxy system with the GO-POSS/epoxy nanocomposite reveals that the barrier effect and long-term anti-corrosion potential was considerably improved by the addition of only 0.5% of nanomaterial. The distinctiveness of this system resides in the potential of filling the defects usually found in the coating film and in addition to hinder the diffusion pathway for the corrosive species. [56] et al. observed an improvement of 12.39% protection efficiency when graphene oxide that was chemically modified with dodecylamine (GO-DDA) in a composite formulation was applied in aircraft to prevent corrosion. Still, 1.2% of GO-DDA has the potential of reducing the path of corrosion media to the metal surface and impeding the development of cracks. A vital factor that needs to be considered when. formulating an anticorrosive coating is the material’s hydrophilicity. The diffusion and absorption of water which is not favourable to the barrier property can be facilitated by the high content of polar groups (epoxy, hydroxyl, and carboxyl) on the surface of the graphene oxide. The effectiveness of this approach was examined by analyzing with electrochemical impedance spectroscopy (EIS) and potentiodynamic

374

I. H. Ifijen et al.

polarization after immersion in 3.5 wt. % NaCl. The outcome revealed that both nanomaterials act as an obstacle to hamper the permeation of corrosive electrolytes.

Conclusion Graphene coatings without modifications demonstrate protective features; nevertheless, these coatings suffer from the presence of pores, defects, and cracks, resulting in reduced protective performance. The modification of graphene oxide with diverse chemical compounds (organic or inorganic) drastically enhanced the anti-corrosion properties. The available investigations in literature reported the exceptional characteristics possessed by graphene and its derivatives which, jointly with the performance of industrial processes to fabricate satisfactory amounts of these nano-coating structured materials, will facilitate the creation of novel solutions for corrosion protection in the future.

References 1. He J, Schoenung JM (2002) Nanostructured coatings. Mater Sci Eng A 336:274–319 2. Santos J, Moschetta M, Rodrigues J, Alpuim P, Capasso A (2021) Interactions between 2D materials and living matter: a review on graphene and hexagonal boron nitride coatings. Front Bioeng Biotechnol 9:612669 3. Ferhan AR, Ma GR, Jackman JA, Sut TN, Park JH, Cho NJ (2017) Probing the interaction of dielectric nanoparticles with supported lipid membrane coatings on nanoplasmonic arrays. Sensors (Basel) 17(7):1484 4. Ifijen HI, Ikhuoria EU (2020) A simple technique for the fabrication of colour tunable P(StBAAA) colloidal crystal microdots on ink-jet paper. Heliyon 6:e04196 5. Ifijen IH, Odi HD, Maliki M, Omorogbe SO, Aigbodion AI, Ikhuoria EU (2020) Correlative studies on the properties of rubber seed and soybean oil based-alkyd resins and their blends. J Coating Techn Res 18:459–467 6. Ifijen IH, Ikhuoria EU, Omorogbe SO, Aigbodion AI (2019) Ordered colloidal crystals fabrication and studies on the properties of poly (Styrene–Butyl Acrylate–Acrylic Acid) and polystyrene latexes. In: Srivatsan T, Gupta M (eds) Nanocomposites VI: nanoscience and nanotechnology in advanced composites. The minerals, metals & materials series. Springer, Cham, pp 155–164 7. Ifijen IH, Maliki M, Ovonramwen OB, Aigbodion AI, Ikhuoria EU (2019) Brilliant coloured monochromatic photonic crystals films generation from poly(Styrene-Butyl Acrylate-Acrylic Acid) latex. J Appl Sci Environ Mgt 23(9):1661–1664 8. Ifijen IH, Ikhuoria EU (2019) Generation of highly ordered 3D vivid monochromatic coloured photonic crystal films using evaporative induced technique. Tanz J Sci 45(3):439–449 9. Ifijen IH, Omorogbe SO, Maliki M, Odiachi IJ, Aigbodion AI, Ikhuoria EU (2020) Stabilizing capability of gum Arabic on the synthesis of poly (Styrene-Methylmethacrylate-Acrylic Acid) latex for the generation of colloidal crystal films. Tanz J Sci 46(2):345–435 10. Ifijen IH, Ikhuoria EU (2020) Monodisperse polystyrene microspheres: studies on the effects of reaction parameters on particle diameter. Tanz J Sci 46(1):19–30 11. Jaya Verma AS, Sahney KR, Bhattacharya A (2020) Super protective anti-bacterial coating development with silica–titania nano core–shells. Nanoscale Adv 2:4093–4105

Nanostructured Graphene Thin Films: A Brief Review …

375

12. Otabor GO, Ifijen IH, Mohammed FU, Aigbodion AI, Ikhuoria EU (2019) Alkyd resin from rubber seed oil/linseed oil blend: a comparative study of the blend properties. Heliyon 5(5)15:e01621 13. Ifijen HI, Nkwor AN (2020) Selected under-exploited plant oils in Nigeria: a correlative study of their physiochemical properties. Tanz J Sci 46(3):817–827 14. Nkwor AN, Ukoha PO, Ifijen HI (2021) Synthesis of sulfonated Sesamum indicum L. seed oil and its application as a fatliquor in leather processing. J Leather Sci Eng Springer Nature 3(16):1–13 15. Nkwor AN, Ukoha PO, Ifijen HI, Ikhuoria EU (2020) The use of sulfonated Jatropha Curcas oil for the processing of mechanically improved leather. Chem Afri. 3:911–925 16. De Lima LRM, Martins FP, Lagarinhos JN, Santos L, Lima P, Torcato R, Marques PAAP, Rodriguez DL, Melo S, Grilo J et al (2020) Characterization of commercial graphene-based materials for application in thermoplastic nanocomposites. Mater Today Proc 20:383–390 17. Calovi M, Rossi S, Deflorian F, Dirè S, Ceccato R (2020) Effect of functionalized graphene oxide concentration on the corrosion resistance properties provided by cataphoretic acrylic coatings. Mater Chem Phys 239:121984 18. Alammar A, Park SH, Williams CJ, Derby B, Szekely G (2020) Oil-in-water separation with graphene-based nanocomposite membranes for produced water treatment. J Memb Sci 603:118007 19. Cataldi P, Steiner P, Raine T, Lin K, Kocabas C, Young RJ, Bissett M, Kinloch IA, Papageorgiou DG (2020) Multifunctional biocomposites based on polyhydroxyalkanoate and graphene/carbon nanofiber hybrids for electrical and thermal applications. ACS Appl Polym Mater 2:3525–3534 20. Khazaee A, Jahanshahi R, Sobhani S, Skibsted J, Sansano JM (2020) Immobilized piperazine on the surface of graphene oxide as a heterogeneous bifunctional acid-base catalyst for the multicomponent synthesis of 2-amino-3-cyano-4: H-chromenes. Green Chem 22:4604–4616 21. Kweon H, Lin CW, Faruque Hasan MM, Kaner R, Sant GN (2019) Highly permeable polyaniline–grapheme oxide nanocomposite membranes for CO2 separations. ACS Appl Polym Mater 1:3233–3241 22. Ma L, Zhou M, He C, Li S, Fan X, Nie C, Luo H, Qiu L, Cheng C (2019) Graphene-based advanced nanoplatforms and biocomposites from environmentally friendly and biomimetic approaches. Green Chem 21:4887–4918 23. Cseri L, Baugh J, Alabi A, AlHajaj A, Zou L, Dryfe RAW, Budd PM, Szekely G (2018) Graphene oxide-polybenzimidazolium nanocomposite anion exchange membranes for electrodialysis. J Mater Chem 6:24728–24739 24. Xuan Lim CHY, Sorkin A, Bao Q, Li A, Zhang K, Nesladek M, Loh KP (2013) A hydrothermal anvil made of graphene nanobubbles on diamond. Nat Commun 4:1556–1558 25. Kyhl L, Nielsen SF, Cabo AG, Cassidy A, Miwa JA, Hornekær L (2015) Graphene as an anti-corrosion coating layer. Faraday Discuss 180:495–509 26. Novoselov KS, Geim AK, Morozov SV, Jiang D, Zhang Y, Dubonos SV, Grigorieva IV, Firsov AA (2004) Electric field effect in atomically thin carbon films. Science 306:666–669 27. Berger C, Song Z, Li X, Wu X, Brown N, Naud C, Mayou D, Li T, Hass J, Marchenkov AN, Conrad EH, First PN, de Heer WA (2006) Electronic confinement and coherence in patterned epitaxial graphene. Science 312:1191–1196 28. Balandin AA (2020) Phononics of graphene and related materials. ACS Nano 14:5170–5178 29. Emtsev KV, Bostwick A, Horn K, Jobst J, Kellogg GL, Ley L, McChesney JL, Ohta T, Reshanov SA, Reohrl J et al (2009) Towards wafer-size graphene layers by atmospheric pressure graphitization of silicon carbide. Nat Mater 8:203–207 30. Hass J, Feng R, Li T, Li X, Zong Z, de Heer WA, First PN, Conrad EH, Jeffrey CA, Berger C (2006) Highly ordered graphene for two dimensional electronics. Appl Phys Lett 89:143106 31. Ferrari AC, Bonaccorso F, Falko V, Novoselov KS, Roche S, Bøggild P, Borini S, Koppens FHL, Palermo V, Pugno N et al (2015) Science and technology roadmap for graphene, related two-dimensional crystals and hybrid systems. Nanoscale 7:4598–4810

376

I. H. Ifijen et al.

32. Mishra N, Boeckl J, Motta N, Iacopi F (2016) Graphene growth on silicon carbide: a review. Phys Status Solidi A 213:2277–2289 33. Yu Q, Lian J, Siriponglert S, Li H, Chen YP, Pei SS (2008) Graphene segregated on Ni surfaces and transferred to insulators. Appl Phys Lett 93:113103 34. Giovanneteti G, Khomyakov PA, Brocks G, Karpan VM, Vam DBJ, Kelly PJ (2008) Doping grapheme with metal contacts. Phys Rev Lett 101:026803 35. Ramezanzadeh M, Ramezanzadeh B, Mahdavian M, Bahlakeh G (2020) Development of metalorganic framework (MOF) decorated graphene oxide nanoplatforms for anti-corrosion epoxy coatings. Carbon 161:231–251 36. Li X, Zhu Y, Cai W, Borysiak M, Han B, Chen D, Piner RD, Colombo L, Ruoff RS (2009) Transfer of large-area graphene films for high-performance transparent conductive electrodes. Nano Lett 9:4359–4363 37. Kim Y, Song W, Lee SY, Jeon C, Jung W, Kim M, Park CY (2011) Low-temperature synthesis of large-area graphene-based transparent conductive films using surface wave plasma chemical vapor deposition. Appl Phys Lett 98:263106 38. Kulyk B, Carvalho AF, Fernandes AJS, Costa FM (2020) Millimeter sized graphene domains through in-situ oxidation/reduction treatment of the copper substrate. Carbon 169:403–415 39. Xu X, Zhang Z, Dong J, Yi D, Niu J, Wu M, Lin L, Yin R, Li M, Zhou J et al (2017) Ultrafast epitaxial growth of metre-sized single-crystal graphene on industrial Cu foil. Sci Bull 62:1074–1080 40. Munoz R, Gomez-Aleixandre C (2013) Review of CVD synthesis of graphene. Chem Vap Deposition 19:297–322 41. Suk JW, Kitt A, Magnuson CW, Hao Y, Ahmed S, An J, Swan AK, Goldberg BB, Ruoff RS (2011) Transfer of CVD-grown monolayer graphene onto arbitrary substrates. ACS Nano 5:6916–6924 42. Wang Y, Zheng Y, Xu X, Dubuisson E, Bao Q, Lu J, Loh KP (2011) Electrochemical delamination of CVD grown graphene film: toward the recyclable use of copper catalyst SI. ACS Nano 5:9927–9933 43. Banszerus L, Schmitz M, Engels S, Goldsche M, Watanabe K, Taniguchi T, Beschoten B, Stampfer C (2011) Ballistic transport exceeding 28 lm in CVD grown graphene. Nano Lett 16:1387–1391 44. Pirkle A, Chan J, Venugopal A, Hinojos D, Magnuson CW, McDonnell S, Colombo L, Vogel EM, Ruoff RS, Wallace RM (2011) The effect of chemical residues on the physical and electrical properties of chemical vapor deposited graphene transferred to SiO2 . Appl Phys Lett 99:122108 45. Zhang Z et al (2018) Rosin-enabled ultraclean and damage-free transfer of graphene for largearea flexible organic light-emitting diodes. Nat Commun 8:1–9 46. Luo D, You X, Li BW, Chen X, Park HJ, Jung M, Ko TY, Wong K, Yousaf M, Chen X et al (2017) Role of graphene in water-assisted oxidation of copper in relation to dry transfer of graphene. Chem Mater 29:4546–4556 47. Pu NW, Shi GN, Liu YM, Sun X, Chang JK, Sun CL, Ger MD, Chen CY, Wang PC, Peng YY (2015) Graphene grown on stainless steel as a high-performance and ecofriendly anticorrosion coating for polymer electrolyte membrane fuel cell bipolar plates. J Power Sources 282:248–256 48. Singh Raman RK, Chakraborty BP, Lobo DE, Gullapalli H, Sumandasa M, Kumar A, Choudhary L, Tkacz R, Ajayan PM, Majumder M (2012) Protecting copper from electrochemical degradation by graphene coating. Carbon 50:4040–4045 49. Prasai D, Tuberquia JC, Harl RR, Jennings GK, Rogers BR, Bolotin KI (2012) Graphene: corrosion-inhibiting coating. ACS Nano 6:1102–1108 50. Zhao Y, Xie Y, Hui YY, Tang L, Jie W, Jiang Y, Xu L, Shu PL, Chai Y (2013) Highly impermeable and transparent graphene as an ultra-thin protection barrier for Ag thin films. J Mater Chem C 1:4956–4961 51. Zhang W, Lee S, Mcnear KL, Chung TF, Lee S, Lee K, Crist SA, Ratliff TL, Zhong Z, Chen YP (2014) Use of graphene as protection film in biological environments. Sci Rep 4:4097

Nanostructured Graphene Thin Films: A Brief Review …

377

52. Hsieh YP, Hofmann M, Chang KW, Jian GJ, Li YY, Chen KY, Yang CC, Chang WS, Chen LC (2014) Complete corrosion inhibition through graphene defect passivation. ACS Nano 8(1):443–448 53. Schriver M, Regan W, Gannett WJ, Zaniewski AM, Crommie MF, Zettl A (2013) Graphene as a long-term metal oxidation barrier: worse than nothing. ACS Nano 7:5763–5768 54. Mirmohseni A, Azizi M, Dorraji MSS (2020) Cationic graphene oxide nanosheets intercalated with polyaniline nanofibers: a promising candidate for simultaneous anticorrosion, antistatic, and antibacterial applications. Prog Org Coat 139:105419 55. Ye Y, Chen H, Zou Y, Ye Y, Zhao H (2020) Corrosion protective mechanism of smart graphenebased self-healing coating on carbon steel. Corros Sci 174:108825 56. Zhang H, Zhou H, Yang J, Song Y, Zhao L (2019) Synergism of modified graphene oxide to aircraft structural corrosion inhibiting compounds coatings. Prog Org Coat 132:490–498

Numerical Study of Intrinsic Stresses in Perovskite-on-Si Solar Cells with Intermetallic Bonding Seif Tarek and Tarek M. Hatem

Abstract Organic-based perovskite semiconductors provide a suitable material candidate to fabricate optoelectronic devices such as solar cells due to its low deposition cost and high efficiency. On the other hand, silicon wafers provide a stable and cost-effective technology for electronics and opto-electronics industries. Combining both materials in photovoltaic solar cells, in what is called Multijunction (MJ) solar cells, allows the absorbance of a broader range of wavelengths, improving the cell’s photo to electrical energy conversion efficiency. Nevertheless, thermal stresses are generated during processing and operation of the perovskite/silicon layer due to the thermal mismatch between both materials. In the current study, different interfaces have been explored to reduce intrinsic stresses and therefore decrease defects and therefore enhance stability and increase efficiency of perovskite thin films. In particular, intermetallic bonding interfaces will be explored which give different results and different stress distribution. The effect of intermetallic bonding will be investigated in this paper. Keywords Solar cells · Multijunction · Perovskite · Silicon · Efficiency

S. Tarek · T. M. Hatem (B) Centre for Simulation Innovation and Advanced Manufacturing, The British University in Egypt, El-Sherouk City, Cairo 11837, Egypt e-mail: [email protected] S. Tarek e-mail: [email protected] T. M. Hatem Faculty of Energy and Environmental Engineering, Renewable Energy Department, The British University in Egypt, El Sherouk City, Cairo 11837, Egypt © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_34

378

Numerical Study of Intrinsic Stresses in Perovskite-on-Si Solar Cells …

379

Introduction Perovskite solar cells efficiency is growing rapidly and progressively, which makes it a highly demanded topic for researchers. During the ten-year gap between 2009 and 2019, the efficiency of organo-metallic halide perovskite jumped from 3.5 to 23.3% [1]. There are multiple ways to fabricate Perovskites such as such as, spin-coating, inverse temperature crystallization, the ligand-assisted reprecipitation technique, hydrothermal synthesis, and self-template-directed synthesis. More than 90% of commercial solar panels, use silicon as its base material. During the last 15 years, the silicon solar cells efficiency improved to 25.6% which is close to the theoretical efficiency of 29.4% [2]. An alternative approach can be taken to improve the efficiency of solar panel, by introducing another cheap material cell with different band gap width [3]. Those characteristics are offered by organic–inorganic perovskite which offers an efficiency as high as 20.1%. The organic component within the perovskite ease the precipitation on the silicon substrate due to its solubility, furthermore, it enhances the mechanical flexibility of the silicon solar cells. The choice of which perovskite to use to provide optimum qualities and properties is still under investigation [4]. In this work, the effect of processing and operating silicon/perovskite layer is examined on thin film using finite element analysis to model the effect of thermal mismatch on the structure and to investigate the effect of the thermal loading on the solar cell made of perovskite and silicon, and study which interface has a better effect in reducing the thermal loading on the solar cells.

Theoretical Work and Results Methodology To investigate the effect of thermomechanical loading on the silicon/perovskite solar cells, thin-film layer representing a multijunction is used for this study. The perovskite used is organic–inorganic methylammonium lead halogen (CH_3 NH_3 PbX_3). CH_3 NH_3 PbI_3 is a methylammonium lead which was chosen and used due to its efficiency and mechanical properties [5]. The thin films have two different interfaces separating the two materials, straight and irregular lines as shown in Fig. 1. Also, the effect of different interfaces is being studied. Model A has a straight interface as shown in Fig. 1a while model B has an irregular interface as shown in Fig. 1b. Section “Introduction” in the upper part in each thin film represents the perovskite region, while section “Theoretical Work and Results” in the lower part represents the silicon region. The substrate thin film has a geometry of 1 μm × 2 μmsilicon layer and 3 μm × 2 μmperovskite layer. Periodic boundary conditions are applied at the nodes located

380

S. Tarek and T. M. Hatem

Fig. 1 Shows the suggested models a shows model A with straight interface, b shows model B with irregular interface

at the outer edges of the thin film. Thermal mismatch of both materials is tested using fundamental linear hyperelastic constitutive model with thermal effects (DuhamelNeuman thermoelastic constitutive law). The law follows the following equations [6]. σi j = Ci jkl kl − βi j (T − T0 )

(1)

where σ is the second order Cauchy stress tensor, ε is the second order linearized strain tensor, β is the second order thermoelastic moduli tensor, C is the fourth order elasticity tensor, T is the current temperature, and T0 is the reference temperature. Thermal effects on both the silicon and perovskite layers are approximated to be isotropic. Therefore, βi j is reduced to the second order isotropic tensor (αi j ) and becomes [7]: αi j = α(T − T0 )δi j

(2)

And the second order thermal strain tensor (εiTj ) becomes: εiTj = αi j (T − T0 )

(3)

After conducting the thermoelastic model, a homogeneous temperature similar to the thermal stresses generated during processing and operating of perovskite/silicon layer is applied. The initial temperature of the model is chosen to be at 423 K [8–11], then the model is cooled to room temperature at 300 K and then reheated again to the operating temperature at 358 K [8, 12].

Results The results of each phase in the process are studied individually and respectively for both types of multijunction. The cooling effect on the multijunction from 423 K to room temperature is shown in Fig. 2. Figure 2a shows that the stress distribution for model A in the perovskite

Numerical Study of Intrinsic Stresses in Perovskite-on-Si Solar Cells …

381

Fig. 2 Shows stress distribution on both models after the cooling phase a shows model A while b shows model B

section with an average of 50 MPa, while in the silicon section to be averaging a little more than 50 MPa. For model B, Fig. 2b shows that the perovskite section stress values range from 38 to 100 MPa, while the silicon section ranges from 78 to 100 MPa with the exception of 38 MPa at the bottom right corner, with the stress is highly concentrated on the interface between both sections. Figure 3 shows the heating effect after the temperature was raised to 358 K. Figure 3a shows the perovskite section in model A with an average stress of 58 MPa. For the silicon section, it has an average of 28 MPa. Figure 3b shows the perovskite section in model B to range between 49 and 86 MPa with the stress concentrated at the border between the perovskite and the silicon. For the silicon section, the bottom right corner had the least thermal effect with a stress that ranges between 22 and 44 MPa, while the left border had the highest with stress ranging between 60 and 76 MPa.

Conclusion Model A after the cooling phase showed higher stress at the silicon section; however, after the heating phase, the perovskite section had higher stresses. For both phases, the stress distribution was the same at both sections. The results from the TFA method used, based on a linear constitutive model, show good accordance with the manufacturer tabulated data on low-medium voltage, suitable to the applications of sensing and wave generation commonly used in structural health monitoring.

382

S. Tarek and T. M. Hatem

Fig. 3 Shows the stress distribution for both models after the reheating phase a shows model A while b shows model B

For model B, after the cooling phase the perovskite section had a relatively low stress values except for areas surrounding the interface, while the silicon section had moderate stress values with higher stress values at the interface. However, after the heating phase the stress values was slightly higher at the perovskite section than the silicon but with higher stress values at the perovskite side of the interface. For both phases, model A which is the model with the straight-line interface had better results. Acknowledgements The authors would like to thank H. Pelletier, G. Bertrand, for the technical help, the Young Investigators Research Grant (No. YIRG05) at the British University in Egypt, JESOR Research Grant from the Academy of Science and Technology (ASRT), the Natural Sciences and Engineering Research Council of Canada (NSERC), and the Fonds de Recherche du QuebecNature et Technologies (FRQNT) for financial support

References 1. Jung EJ, Jeon NJ, Park EY, Moon CS et al (2019) Efficient, stable and scalable perovskite solar cells using poly(3-hexylthiophene. Nature 567:511–515 2. Swanson RM (2005) Approaching the 29% limit efficiency of silicon solar cells. Conference record of the thirty-first IEEE photovoltaic specialists conference 2005, pp 889−894 3. Asadpour R, Chavali RVK, Khan MR, Alam MA (2014) Appl Phys Lett 106:24 4. Brenner P, Bar-On O, Jakoby M, Allegro I, Richards BS, Paetzold UW, Howard IA, Scheuer J, Lemmer U (2019) Nat Commun 10:988

Numerical Study of Intrinsic Stresses in Perovskite-on-Si Solar Cells …

383

5. Hörantner MT, Leijtens T, Ziffer ME, Eperon GE, Greyson Christoforo M, McGehee MD, Snaith HJ (2017) The potential of multijunction perovskite solar cells. ACS Energy Lett 2(10):2506–2513 6. Simo JC, Marsden JE (1986) On the rotated stress tensor and the material version of the Doyle-Ericksen formula. In: The breadth and depth of continuum mechanics. Springer, Berlin, Heidelberg, pp 453–471 7. Sadd MH (2009) Elasticity: theory, applications, and numerics. Academic Press, Oxford 8. Jacobsson TJ, Josef Schwan L, Ottosson M, Hagfeldt A, Edvinsson T (2015) Determination of thermal expansion coefficients and locating the temperature-induced phase transition in methylammonium lead perovskites using x-ray diffraction. Inorg Chem 54(22):10678–10685 9. Bett AJ, Schulze PSC, Winkler K, Gasparetto J, Ndione PF, Bivour M, Hinsch A et al (2017) Low temperature perovskite solar cells with an evaporated TiO2 compact layer for perovskite silicon tandem solar cells. Energy Procedia 124:567–576 10. Sum T-C, Mathews N (eds) Halide perovskites: photovoltaics, light emitting devices, and beyond. Wiley, Weinheim 11. Mesquita I, Andrade L, Mendes A (2019) Temperature impact on perovskite solar cells under operation. Chemsuschem 12(10):2186–2194 12. Holzapfel AG (2000) Nonlinear solid mechanics II. Wiley

Part X

Advanced Magnetic Materials for Sensors, Power, and Multifunctional Applications

Custom-Designed Miniature-Coil Winding/Wrapping Machine Balraj S. Mani, Bilal Adra, and Nuggehalli M. Ravindra

Abstract Magnetic field assisted assembly techniques are gaining acceptance as a novel concept in the intricate placement of devices on substrates at the wafer-level. The research into such innovative techniques requires creation of a uniform magnetic field to begin with. This will require a large array of milli-scale electromagnets needing a large number of miniature electromagnetic coils of uniform size. Such coils are required in varying lengths from 5.0 to 25.0 mm, core diameters ranging from 0.5 to 2.0 mm and wire sizes ranging from AWG 24 to 32. Dual-layer coils will more than double the magnetic field strength compared to monolayer coils. In addition, dual-layer coils will result in a coil structure in which the leads will terminate at the same end of the coil. Keywords Coil winding · Coil wrapping · Milli robots

Background Liu et al. demonstrated a novel milli-scale robotic assembly machine with parallel processing capabilities, assisted by programmable magnetic field [1]. Liu’s prototype consisted of a 16 × 16 array of electromagnets using coils of the type produced by the Coil Winding Machine discussed in this paper. In every such experiment in this field of study, one will require electromagnetic coils in different lengths, wire diameter and other characteristics and in large quantities. The set of coils used in such experiments must all be of uniform characteristics. One of the primary requirements that is important in the design of such Coil Winding Machines is tension control to keep the coil wire taut while it is wound on the core without causing any damage or breaks to the coil wire. Voss and Heft [2], likewise, point out the need to produce B. S. Mani (B) · B. Adra Department of Mechanical and Industrial Engineering, New Jersey Institute of Technology, Newark, NJ 07102, USA e-mail: [email protected] N. M. Ravindra Department of Physics, New Jersey Institute of Technology, Newark, NJ 07102, USA © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_35

387

388

B. S. Mani et al.

“small or large lots, very much varying in dimensions and shape” but of uniform dimensions. While proposing statistical process control, the authors reflect on the need to maintain uniform tension and the need to monitor “wire break.” Voss and Heft discuss the need for sophisticated feedback control systems to electronically monitor and control wire tension. Florian Sell-Le Blanc et al. [3] have proposed controls for Coil Winding Machines for electric motors. The authors propose the use of piezo-electric actuators and electromechanical tension control mechanisms. Amit Jaywantrao Somwanshi et al. [4] provide a tutorial for the design of the various components of a Coil Winding Machine.

Introduction Most applications in the industry refer to Coil Winding Machines that are used in the construction of electric motors and transformers. The Coil Winding Machine discussed in this paper refers to the machines producing free miniature coils. Comprehensive set of components that constitute the unique Coil Winding Machine and simple tension and pitch control implemented in this laboratory application are discussed. Adaptation of coil wrapping technique derived from “coil wrapping” tools used in digital electronics area is proposed as a solution to produce miniature coils of core diameters less than 1 mm.

Design Intent Plan A custom-built Coil Winding Machine was pursued to establish a long-term solution that is capable of producing custom designed electromagnetic coils. Coils produced by this machine will form the backbone of the ongoing research in magnetic field assisted assembly needed in varying lengths, core diameters, wire gauges, and in large quantities, several hundred in any single experiment. Robots assembled using such coils are capable of intricate placement of miniature components. First, we synthesized the Design Intent Plan for the prototype machine. Figure 1 represents the initial Design Intent Plan. The prototype was designed to perform six required and distinct operations namely: a. releasing the wire, b. maintaining the wire tension when a coil is wound, c. winding the coils over a core, d. supporting the rotating core, e. automating the pitch control and forming the dual-layer, and f. guiding the wire as it is fed over the rotating core. Units designed to deliver the above functions, respectively, are the Bobbin Support, the Dancer, the Motor Drive, the Idler, the Coil Steering Unit (CSU), and the Riding Wire Guide. The Coil Winding Machine has been an ongoing development. Figure 2 shows the CAD image of the machine designed. Creo Parametric 6.0™ [5] was used for the CAD design.

Custom-Designed Miniature-Coil Winding/Wrapping Machine

389

Fig. 1 Design intent plan

Fig. 2 Dual-layer coil winding machine, CAD image

Bobbin Support A Coil Winding Machine used in a laboratory should be capable of accommodating different reel sizes with minimal change to the design or the parts. The Bobbin Support has been designed to accommodate such varying reel sizes. This is accomplished by adopting a pair of adjustable Teflon end supports. In some cases, this may require the use of an additional Teflon or Nylon spacer. A second important requirement for the Bobbin Support is the ability to release the wire in a controlled manner without allowing the reel to retract and rotate in the reverse direction causing

390

B. S. Mani et al.

problems in feeding the wire properly. This is accomplished by a simple ratchet design. Figure 2 shows the special Bobbin Support designed to ensure continuous, coordinated release, and dispensing of the coil wire from the reel.

Spring Loaded Dancer Subassembly Winding is realized by feeding coil wire onto the rotating mandrel/core. As the wire is being fed, it should be kept taut in order to ensure a tightly wound coil conforming to the final size desired. This ensures quality and productivity, minimizing rejection rates of the final coils produced. Expensive Dancers have been traditionally used in web control for the same purpose. Sophisticated tension control systems have been used in similar Coil Winding Machines that are used for motors and transformers. Such systems use dedicated feedback control system to control the tension provided by the dancer rolls. A simple adaptation is commonly seen in a manual sewing machine which uses a take-up lever and thread tensioner unit. The prototype designed incorporates two Dancers with opposing forces. One of the Dancers is located near the Bobbin Support at the unwind end and the other near the rotating mandrel/core at the wind end. Torsion springs suitably wound with left-hand and right-hand windings are incorporated in the design to deliver the opposing forces and provide efficient handling of the coil wire. Figure 2 shows the two Dancers.

Motor Drive Subassembly Winding is realized by feeding the coil wire onto the rotating mandrel/core with its end secured. The mandrel/core is driven by a stepper motor. Integrating an Arduino and motor driver controls the speed of the motor and the number of turns of the core. The eventual prototype will use Lin Engineering Stepper motor. However, initial evaluations are accomplished using an inexpensive stepper motor without any sophisticated speed and step control.

Idler Subassembly Support at either ends of the mandrel/core is required during the winding process to prevent unwanted deflection of the core due to the tension arising from the coil wire being wound. An Idler subassembly was developed to secure and support the free end of the mandrel/core allowing it to rotate freely with the Motor Drive. Miniature Dremel 4486 keyless chuck is used to support the ends of the rotating core. Figure 2

Custom-Designed Miniature-Coil Winding/Wrapping Machine

391

shows the Motor Drive subassembly and the Idler subassembly in relation to the rest of the machine.

The Coil Steering Unit The Coil Steering Unit is the heart of the Coil Winding Machine. Figure 3 shows the Coil Steering Unit in its current design. The Coil Steering Unit automates the pitch control and produces dual-layers. As the coil wire is wound onto the mandrel/core, the wire translates laterally to form a tight-coil structure. The Coil Steering Unit is also able to achieve the dual-layer or double layer coil structure. A dual-layer coil structure is one in which the wire is first wound over the mandrel/core for a desired length of coil. The winding will follow a right-hand winding configuration based on the configuration of the design. In continuation, a coaxial, second layer of coil is wound over the primary winding, with the coil progressing in the reverse direction. However, the winding will be maintained in the same configuration (right hand) as the primary layer. A dual-layer coil constitutes more than double the length of the coil wire or conductor and therefore allows for more than double the electromagnetic energy generated compared to a monolayer coil of the same length. In addition, a duallayer coil will result in a coil structure in which the leads will physically terminate at the same end of the coil. This will enhance manufacturability by reducing clutter in the future integration of hundreds of such coils in a single application.

Fig. 3 Coil steering unit—CAD image

392

B. S. Mani et al.

In the current design, the Coil Steering Unit integrates a pulley drive, a gear train, and leadscrew, all in one assembly. The translation motion is derived from the primary motor driving the mandrel/core. In the first stage, the rotation of the primary drive shaft is transmitted through a pulley drive. The driving pulley diameter is 0.672 and the driven pulley diameter is 4 . The second stage comprises of a gear train using three spur gears of 48 Diametral Pitch (DP), and pressure angle 14.5°. An idler gear in the gear train maintains the direction of rotation identical to the direction of the rotation of the primary motor driving the mandrel/core. The number of teeth in the gear train is 24-54-21. This will concurrently drive a single start, double-helix lead screw with left hand and right hand threads of pitch 0.125 . The left-hand and right-hand threads are machined in continuation in order to provide seamless reversal of the direction of winding to form the second and continuous layer of the coil. Such double-helix threaded lead-screws are currently in commercial use in intricate mechanisms such as a fishing reel [6]. Following calculation shows the translation speed per one revolution of the core: = 0.168. Ratio of speed at the pulley drive = 0.672 4 54 Ratio of speed at the Gear train = 24 × = 1.14286 54 21 Lead screw Pitch = 0.125 . Lateral translation of guide per one revolution of the primary motor driving the mandrel/core = 0.168 × 1.14286 × 0.125 = 0.024 This will be suitable for AWG 23 wire of diameter 0.0226 . The Coil Steering Unit is a classic example of diverse drive techniques integrated in one subassembly to provide translation of the coil wire coordinated with the rotation of the mandrel/core. For each 360° rotation of the mandrel/core, the coil wire should advance progressively and continuously and translate over a pitch equal to the diameter of the coil wire. A translation, more than the diameter of the coil wire, will result in a coil with dispersed windings. On the other hand, translation less than the diameter of the wire like in this design is likely to result in a coil with somewhat crowded layer of windings depending on the length of the coil produced. Ideally, the speed should be controlled electronically to accommodate multiple wire diameters with flexibility. The current design of the prototype depends on mechanical motion transfer. Eventually, the pulley drive would be replaced with a second stepper motor with integrated electrical speed control.

The Riding Wire Guide The Riding Wire Guide has a guide with an eye (orifice) that rides with the guide of the Coil Steering Unit riding on the groove of the lead screw. The riding wire guide incorporates a chute made of Teflon that feeds the wire delivered by the Coil Steering Unit onto the mandrel/core. The Teflon chute guarantees the prompt and immediate delivery of the wire onto the mandrel/core with minimum slack. Special bearing designed aides in friction free riding (rolling) of the riding wire guide in conjunction with the guide of the Coil Steering Unit.

Custom-Designed Miniature-Coil Winding/Wrapping Machine

393

Fig. 4 Prototype in progress. Arduino and motor controller shown on the left

Status The Coil Winding Machine is currently being manufactured and has been partially assembled. Preliminary trials of the partially assembled machine has shown promising results. Figure 4 shows portion of the assembly completed along with the prototype Arduino and Motor controller. The rest of the machine will be ready for evaluation in the next few weeks. However, the machine is able to produce dual-layer coils on mandrels of diameter 1.5 and 3 mm using automatic reversal of winding to form the dual or second layer. Figure 5 shows the samples of coils produced using the prototype in progress. AWG 24 and AWG 28 wires were used in the trials. Mandrels of diameter 1.5 and 3.0 mm were used. The primary contributor to the quality of outcome, the coils produced, is the drive transmission to the Coil Steering Unit.

Future Work Trials using mandrel/core of diameter less than 1 mm open up a significant challenge. The challenge associated with this reduction in diameter of the mandrel/core comes from the deflection of the mandrel/core due to the tension of the coil wire applied on such a thin mandrel/core. In order to solve this problem, a different approach will be considered. In this new configuration, the machine will be referred to as “Coil Wrapping Machine,” deriving its name from the innovative technique proposed to be used in creating the coil. The 0.5–2 mm diameter mandrels/cores will still be used, but rather than winding on it, the wire will be wrapped around it with minimal induced stress. This method involves a specially designed “bit” and “sleeve”.

394

B. S. Mani et al.

Fig. 5 Dual-layer coils produced. Wire diameter AWG 24 and 28; mandrel diameter 1.5 mm and 3 mm

A conceptual design is shown in Fig. 6. The bit, a thin shaft, will include an axial hole for the mandrel/core to be placed, serving as a support for the free end of the mandrel/core. The bit itself will be encased in a sleeve. The bit will revolve around the mandrel/core feeding the coil wire to be wrapped, eccentrically, as it revolves. In order to facilitate wire feed, the bit will consist of a semi-circular groove along its outside surface machined eccentric to the hole. This semi-circular groove will seat, hold, and guide the coil wire to be wrapped as it is dispensed to wrap the coil. The sleeve will enclose the semi-circular groove on the bit to form a conduit. The wire will be dispensed from a special spool (not shown in Fig. 6) which will also revolve along with the spinning bit and sleeve. The dispensed coil wire will enter a small hole on the sleeve and into the semi-circular groove machined on the surface of the bit encased by the sleeve. As the bit holding the wire in its semi-circular groove revolves around the mandrel/core, the wire will wrap onto the mandrel/core producing a coil. Two dancer rolls will be used to control the path of the wire through maintaining proper tension. One dancer roll is to control the path of the dispensed wire from the spool while the second will be positioned near the hole in the sleeve where the coil wire will enter the sleeve. Arduino Uno and Motor control will again be used to control a stepper motor, which will drive the rotation of the bit, sleeve, and the coil wire-dispensing spool simultaneously. Such concepts have been adopted in

Custom-Designed Miniature-Coil Winding/Wrapping Machine

395

Fig. 6 Wire wrap head—a conceptual design

special wire wrap tools used to secure and wrap wires onto terminals [7, 8] in digital electronics wiring. The innovative design of automating the wire-wrapping concept will continue to develop through further analysis of the ideas implemented though the wire wrapping tools. Few challenges are anticipated. One challenge is to control the pitch of the coil while wrapping. The pitch can be controlled by gradually withdrawing the mandrel/core in order for the wire to be wrapped evenly and uniformly. However, the rate at which the mandrel/core is withdrawn needs to accommodate the rate at which the bit rotates and wraps the wire in order to achieve the desired pitch. The current design using the lead screw with the double-helix threads will be adopted to accomplish the translation.

Conclusion The Coil Winding Machine is a relatively inexpensive and flexible means to produce miniature coils in large quantities, few hundred, as and when needed for research application in a laboratory.

History of the Project and Acknowledgements The design and construction of the Coil Winding Machine is an ongoing project to support our research on Magnetic field assisted assembly. The need for a dedicated Coil Winding Machine was originally envisioned by Dr. Nuggehalli Ravindra when Ravindra et al. filed and secured their original patents of the concept of Magnetic

396

B. S. Mani et al.

field assisted assembly [9, 10]. A simple and preliminary version of the prototype was built by Mr. Peter Kaufman but it was not commissioned into use. Later, the current version was designed by B. S. Mani a few years ago. Bilal Adra, one of the coauthors prepared almost all production drawings from the original design during the course of his individual research during summer 2020 and summer 2021, which facilitated the fabrication. Bilal Adra also followed up the machining, assembly, and troubleshooting along with B. S. Mani. The authors acknowledge their sincere gratitude to the following individuals for their willing and generous support, which made the realization of the project possible. First and foremost, Dr. Moshe Kam, Dean, Newark college of Engineering generously provided major funding for the project at the recommendation and request of Dr. Joga Rao, Chairman, Mechanical and Industrial Engineering Department. Dr. Pushpendra Singh, Professor of Mechanical Engineering, provided funds to procure the stepper motor from Lin Engineering. Dr. Atam Dhawan, Senior Vice Provost for Research and Distinguished Professor of Electrical and Computer Engineering, supported the project by sponsoring Bilal Adra under the Provost’s Undergraduate Research Innovation (URI) Fellowship during summer 2021. (Late) Mr. Michael Insabella past President, State Tool Gear, and his team of able technicians including Mr. Evan Hoff provided dedicated support in the actual machining of the intricate parts. Mr. Doug Meyer, Application Engineer, Boston Gears provided the set of 48 DP gears needed for the Coil Steering Unit. To all those who helped us in the realization of this valuable tool, “we thank you!”.

References 1. Yan L, Ravindra NM (2018) A magnetic-field-assisted milli-scale robotic assembly machine: an approach to parallel robotic automation systems. Micromachines 9(4):144. https://doi.org/ 10.3390/mi9040144 2. Voss G, Hefti K (1997) Importance and methods of tension control, pp 517–523. In: Proceedings: electrical insulation conference and electrical manufacturing and coil winding conference. IEEE, 25 Sept 1997 3. Sell-Le Blanc F et al (2015) Analysis of wire tension control principles for highly dynamic applications in coil winding. In: 5th international electric drives production conference (EDPC). IEEE, 15–16 Sept 2015 4. Jaywantrao Somwanshi A et al (2018) Design and development of coil winding machine. Int J Adv Res Sci Eng 07(03):50–57 5. Creo Parametric 6.0TM is the Trade Mark of the commercial solid modelling CAD software of Parametric Technology Corporation, USA 6. Chang L-J (2008) Fishing reel with improved transmission efficiency. US 7,429,011 B1, 30 Sept 2008 7. Rivera T (2005) Wire wrapping hand tool. US 7,261,128 B1, 5 July 2005 8. Kober M (1977) Electric wire wrapping tool. US 194,700, 24 June 1977 9. Ravindra NM et al (2007) Method of magnetic field assisted self-assembly. US Patent, US 7217592B2 10. Ravindra NM et al (2010) Assembly using programmable magnets. US Patent, US 7737515B2

Effect of Hot Band Annealing and Final Annealing Temperatures on the Texture, Grain Size, and Magnetic Properties of 1.2 wt% Si Non-oriented Electrical Steel Youliang He, Mehdi Mehdi, Tihe Zhou, Chad Cathcart, Peter Badgley, and Afsaneh Edrisy Abstract A 1.2 wt% Si non-oriented electrical steel (NOES) was processed using conventional rolling-annealing routes. The hot-rolled steel was annealed at various temperatures from 850 to 1000 °C for 4 h and cold rolled to a thickness of 0.5 mm. The steel was final annealed at temperatures varying from 700 to 850 °C for 24 h. The textures and magnetic properties of the steel sheets were characterized by electron backscatter diffraction (EBSD) and Epstein frame techniques, respectively. It was found that hot band annealing at 850 °C for 4 h followed by final annealing at 800 °C for 24 h resulted in the lowest core loss (60 Hz, 1.5 T) of 3.59 W/kg, which is 26% lower than the core loss obtained without hot band annealing (final annealing at 700 °C). Correlations among the magnetic properties, grain size, and texture factor were evaluated for all the annealing conditions to discuss the effect of the annealing conditions on the magnetic properties. Keywords Non-oriented electrical steel · Texture · Core loss · Recrystallization · Rolling · EBSD · Magnetic properties

Introduction Due to their high magnetic permeability, high magnetization saturation, and relatively low core loss, non-oriented electrical steels are the most commonly used soft magnetic materials for electric motors, generators, and alternators [1–3]. The final magnetic properties of the NOES sheets are closely related to the silicon content, Y. He (B) · M. Mehdi CanmetMATERIALS, Natural Resources Canada, Hamilton, ON, Canada e-mail: [email protected] M. Mehdi · A. Edrisy Department of Mechanical, Automotive, and Materials Engineering, University of Windsor, Windsor, ON, Canada T. Zhou · C. Cathcart · P. Badgley Stelco Inc., Hamilton, ON, Canada © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_36

397

398

Y. He et al.

the cleanliness, the sheet thickness, and the microstructure and texture of the steel, the latter being highly dependent on the thermomechanical processing parameters employed during the production of the steel sheets [1–5]. Texture control is an effective method of improving the magnetic properties of non-oriented electrical steels, but it has proven to be difficult in industrial practices because all the thermomechanical processing procedures involved in the steel production will affect the final texture [5, 6]. The initial solidification texture, which is formed due to the directional cooling of the melt, will be changed in the subsequent hot rolling, cold rolling, and annealing, which are associated with metallurgical processes such as plastic deformation, recrystallization, and phase transformation that significantly alter the crystallographic texture [5–7]. In addition, in most cases, the texture of a previous step will affect the textures of the subsequent steps, which makes the control of the final texture very challenging. Hot band annealing, i.e., annealing after hot rolling, is normally not a required step for the production of electrical steel sheets, since the hot-rolled plates can usually be directly cold rolled to achieve the final thickness. However, several studies [8– 10] have shown that annealing after hot rolling can considerably change the grain size and crystallographic texture of the final sheets, due to the formation of large recrystallized grains in the microstructure before cold rolling. The change of the orientations of these large grains and the formation of substructures in these grains during cold rolling will alter the final recrystallization texture. On the other hand, the final annealing temperature also affects the final texture, grain size, and magnetic properties of the electrical steel. Thus, in this study, a 1.2 wt% Si NOES was hot band annealed at various temperatures from 850 to 1000 °C after hot rolling, which was then cold rolled and final annealed at temperatures varying from 700 to 850 °C. The final textures of the steel sheets processed under different conditions were characterized by EBSD, and the magnetic properties were measured by standard Epstein frame method. The magnetic quality of the texture was evaluated using a texture factor calculated from the orientation distribution functions (ODFs), which was directly compared to the core loss. The optimal processing conditions, i.e. the hot band annealing and final annealing temperatures, were determined based on the minimum core loss obtained. The correlations between the magnetic properties and the grain size/texture factor are discussed.

Material and Experimental Procedure The chemical composition of the non-oriented electrical steel investigated in this paper is given in Table 1. The steel was melted in a vacuum furnace and cast into ingots of 200 mm × 100 mm × 265 mm (width × thickness × length). The ingots were then heated to 1050 °C and hot rolled to a thickness of 25 mm (~75% reduction) in a reversing rolling mill. The hot rolling entry temperature was ~950 °C and the finishing temperature was ~875 °C. A second hot rolling step (also heated to 1050 °C) was applied to reduce the thickness to ~2.5 mm (~90% reduction). In this case the entry

Effect of Hot Band Annealing and Final Annealing Temperatures …

399

Table 1 Chemical composition of the investigated electrical steel (wt%) C

Mn

P

Sn

Si

Al

Ni

Mo

Fe

0.0015

0.17

0.0017

0.048

1.2

0.45

0.0023

0.0022

Balance

and finishing temperatures were ~935 and ~720 °C, respectively. After hot rolling, the steel plates were pickled in a hydrochloric acid solution to remove surface oxides and defects. Four hot-rolled plates were then annealed at various temperatures, i.e., 850, 900, 950, and 1000 °C, for 4 h in argon-protected atmosphere. The 4 annealed and 1 unannealed plates were cold rolled to a final thickness of 0.5 mm (~80% reduction). The cold-rolled sheets were finally annealed at different temperatures, i.e., 700, 750, 800, and 850 °C, for 24 h, also in argon-protected atmosphere. The textures of the steel sheets after final annealing were characterized by EBSD using an EDAX OIM 8.1 system. The samples were prepared using conventional metallographic procedures plus a final polishing step using a 0.05 μm colloidal silica solution followed by ultrasonic cleaning. The EBSD scans were performed on the middle-thickness planes of the sheets, which enabled large scans of 3–4 mm × 4 mm to ensure a good statistical representation of the texture. ODFs were calculated using a harmonic series expansion method with a series rank of 22 and a Gaussian half-width of 5°. The ϕ2 = 45° sections of the Euler space (Bunge notation) were used to display the typical body-centered cubic (bcc) texture components and fibers [6, 7, 11, 12]. The grain sizes were evaluated from the grain orientation data using the OIM software. The texture factor is computed from the ODFs using a MATLAB script [13]. The magnetic properties of the final annealed steel sheets (in the rolling direction, RD) were measured by standard Epstein frame method [14]. A total of 16 strips (280 mm × 30 mm × 0.5 mm) were cut from the steel sheets after final annealing, with the longitudinal axis parallel to RD. The core loss at 60 Hz and 1.5 T was evaluated and compared to the texture factor in RD, as well as to the average grain size after final annealing.

Results Variations of the Final Texture Due to Hot Band Annealing Steel Without Hot Band Annealing The textures of the final annealed steel without hot band annealing are shown in Fig. 1. After final annealing at 700 °C, the texture is dominated by three major components, i.e. a {001} (~15° from the cube) on the θ-fiber (//ND), a {331} , and a {112} . There is also a minor Goss ({110} ) component. The γfiber (//ND) is fairly weak (only the {111} component appears). The α-fiber (//RD) commonly observed in annealed NOES is not seen, while a strong α*-fiber ({11h} ) is observed. When the final annealing temperature is increased to 750 °C, the major component, {001} , is considerably weakened,

400

Y. He et al.

Fig. 1 The textures after final annealing at different temperatures: a 700 °C, b 750 °C, c 800 °C, d 850 °C, for the steel without hot band annealing. ϕ2 = 45° sections of the ODF

while the other two components, {331} and {112} , are moved to {552} and {335} , respectively. The α*-fiber is also weakened, and a minor Goss is noticed. Further increasing the annealing temperature to 800 °C, the major component is changed to {110} , a component that is not normally seen in bcc metals. Final annealing at 850 °C produces a strong {113} component and a Goss, while the γ-fiber is essentially eliminated. It is noted that, without hot band annealing, the final annealed samples all show a θ-fiber (the desired texture for nonoriented electrical steels), but the intensity of this fiber reduces with the increase of the final annealing temperature, so does the α*-fiber. The α-fiber and γ-fiber, which are common texture fibers in non-oriented electrical steels after annealing, are very weak in the samples, no matter what is the final annealing temperature. Steels After Hot Band Annealing At Various Temperatures The final annealing textures for the steel after hot band annealing at 850 °C for 4 h are shown in Fig. 2a. A major difference from those without hot band annealing is that the θ-fiber becomes individual components instead of a continuous fiber at most of the final annealing temperatures (except at 850 °C). Final annealing at 700 °C produces {110} and {001} components, while the α-fiber and γ-fiber are eliminated. The α*-fiber is quite weak. Final annealing at 750 °C creates strong {335} and {112} components as well as γ- and α*-fibers. The {001} component is moved to {001} . The {110} texture is significantly weakened. Increasing the final annealing temperature to 800 °C brings the {001} back to {001} , while the {110} component is strengthened. The strongest component is now close to {332} . The γ- and α*- fibers are both weakened. At the above final annealing temperatures, there is no α-fiber or cube component. When the final annealing temperature is 850 °C, the θ-fiber is continuous, i.e. spreading from {001} to cube. The α*- and γ-fibers are strengthened, and there is also a weak {335} component on the α-fiber. The strongest component is at {331} . If the steel was hot band annealed at 900 °C for 4 h, the final annealing textures vary considerably (Fig. 2b) at different final annealing temperatures. Final annealing at 700 and 800 °C produces a relatively strong θ-fiber, a {221} /{332} texture, and an α*-fiber, while final annealing at 750 and 850 °C essentially eliminates

Effect of Hot Band Annealing and Final Annealing Temperatures …

401

Fig. 2 The final annealing textures for steels after hot band annealing at different temperatures: a 850 °C, b 900 °C, c 950 °C, d 1000 °C. ϕ2 = 45° sections of the ODF

the θ-fiber, but creates a strong {110} texture with only a minor α*-fiber. It is noticed that final annealing at 850 °C produces a Goss texture, while at all the other temperatures, there is essentially no Goss. In all the cases, the γ- and α- fibers are very weak or essentially eliminated. Hot band annealing at 950 °C before cold rolling (Fig. 2c) produces a very strong {001} component (intensity 6.7) on the θ-fiber when the final annealing temperature is 850 °C. At all the other final annealing temperatures, the θ-fiber components are relatively weak and the textures are dominated by uncommon components, e.g., {110} , {118} , {441} , etc. After hot band annealing at 1000 °C for 4 h (Fig. 2d), a common feature of the final annealing textures is that there is a strong α*-fiber component, i.e. {114} , {113} or {112} . The sample after final annealing at 700 °C

402

Y. He et al.

shows a strong {111} component on the γ-fiber, while in all the other samples, the γ-fiber is essentially eliminated. The θ-fiber components are relatively weak in all the cases. A texture component close to Goss develops after final annealing at 800 °C.

Texture Factor and Core Loss It is seen that the textures after final annealing at different temperatures show considerable differences among steels without hot band annealing and after hot band annealing at different temperatures. To investigate the correlation between the magnetic properties (e.g., core loss) of the steel and the crystallographic texture, a texture factor is calculated from the obtained ODFs using the following Equation [13]:  TF = f (g)h(g)dg (1) where h(g) is the minimum angle between the magnetization direction and the three easy axes of a given orientation, g, and f(g) is the texture intensity of g. Apparently, the texture factor evaluates the “average” minimum angle between the magnetization direction and the easy axes of all the crystals in the sample (weighted by the intensity), i.e. it indicates how closely the crystals’ “overall” easy axis is aligned to the applied magnetic field. Thus, it is expected that the smaller the texture factor, the better the magnetic properties (e.g., the lower the core loss). Since in this study the magnetic properties were only measured in RD, only the texture factors along RD are calculated. Without hot band annealing (Fig. 3a), the texture factor and core loss show a fairly good correlation, i.e. generally the smaller the texture factor, the lower the core loss. However, the lowest core loss (3.82 W/kg, at 850 °C) is not found at the smallest texture factor (24.5°, at 800 °C). Hot band annealing at 850 °C for 4 h reduces the core losses of all the samples (Fig. 3b), with a minimum core loss (3.59 W/kg) observed at a final annealing temperature of 800 °C. However, in this case, the texture factors do not match the core losses, i.e. at lower final annealing temperatures of 700 and 750 °C, the texture factor and the core loss show opposite trends, while at 800 and 850 °C, they show the same trend. Hot band annealing at 900 °C for 4 h (Fig. 3c) also reduces the core losses of all the samples as compared to those without hot band annealing. In this case, the texture factor and core loss show a fairly good match. It is noted that, the variation of the core losses among different final annealing temperatures is quite small, i.e. from 3.72 W/kg (750 °C) to 3.92 W/kg (700 °C). Increasing the hot band annealing temperature to 950 °C (Fig. 3d) again reduces the core losses of all the samples. In this case, the texture factor and the core loss essentially display opposite trends, i.e. the smallest texture factor corresponds to the highest core loss (at 700 °C), and the largest texture factor corresponds to the lowest core loss (at 850 °C). The core

Effect of Hot Band Annealing and Final Annealing Temperatures …

403

Fig. 3 The relationships between the core loss (60 Hz, 1.5 T) and the texture factor for steels: a without hot band annealing, and hot band annealed at b 850 °C, c 900 °C, d 950 °C, e at 1000 °C. f Shows the comparison of the core losses for all the hot band annealing and final annealing temperatures

losses also show a very small spread from 3.80 to 3.93 W/kg when the final annealing temperatures vary from 700 to 850 °C. When the hot band annealing temperature is increased to 1000 °C (Fig. 3e), the core losses are only reduced (compared to the steel without hot band annealing) when the final annealing temperature is low, i.e. at 700 and 750 °C. At high final annealing temperatures (800 and 850 °C), the core losses are higher than those without hot band annealing. The texture factor and the core loss again essentially show opposite trends, i.e. the smallest texture factor corresponds to the highest core loss (800 °C). The core losses of all the steels processed using different annealing strategies are compared in Fig. 3f. It is seen that hot band annealing at 850 °C for 4 h can generally lead to the lowest core losses if the final annealing temperature is higher than 700 °C. The lowest core loss, 3.59 W/kg, is observed in the steel hot band annealed at 850 °C and final annealed at 800 °C, while the highest core loss, 4.87 W/kg, is found in the steel without hot band annealing and final annealed at 700 °C. The core loss may be reduced by up to 26% if the steel is hot band annealed and final annealed at appropriate temperatures. It is also noted that the texture factor (or crystallographic texture) alone cannot determine the core loss of the steel, as a clear correlation is not observed in all the samples.

Average Grain Size and Core Loss The relationship between the average grain size and the annealing conditions is shown in Fig. 4a. For the same final annealing temperature, hot band annealing

404

Y. He et al.

Fig. 4 The relationships between core loss and average grain size: a variation of the average grain size with respect to the hot band annealing temperature, b–f the correlations between core loss and average grain size for steels: b without hot band annealing, and c, d, e, and f hot band annealed at 850, 900, 950, and 1000 °C, respectively

at all temperatures can increase the final average grain size of the steel. This is because, after hot rolling, ~1/3 of the microstructure (not shown here) near the middle-thickness plane is composed of elongated (deformed) grains that contain substructures such as deformation bands, microbands, and grain fragmentation [13, 15]. Cold rolling of these grains develops high stored energy and generates a large density of dislocations, which, together with the substructures and grain boundaries, lead to a large number of recrystallization nuclei [9, 15, 16]. Thus, the steel without hot band annealing produces smaller recrystallized grains after final annealing. Hot band annealing destroyed the hot rolling microstructure and led to large recrystallized grains before cold rolling; thus the grain boundaries are fewer and the dislocation density is lower after cold deformation, which leads to fewer nuclei and larger grain sizes after recrystallization. It is also seen from Fig. 4a that, the higher the final annealing temperature, the larger the average grain size. This is due to the fact that the higher the annealing temperature, the larger the driving force for grain growth. However, the final grain size is also dependent on the stored energy, dislocation density, microstructure/substructure, and texture of the cold-rolled material before final annealing, which are dependent on the hot band annealing conditions. It is noted that although in most cases, the final grain size also increases with the hot band annealing temperature, the steels after hot band annealing at higher temperatures (950 and 1000 °C) show smaller grain sizes than that hot band annealed at 900 °C if they are final annealed at relatively low temperatures (700 and 750 °C). One possible reason is that highertemperature hot band annealing produces larger recrystallized grains before cold rolling, which leads to higher tendency of shear banding [15, 17]. As shear bands are preferred nucleation sites during recrystallization, the number of nuclei will be

Effect of Hot Band Annealing and Final Annealing Temperatures …

405

larger. If the final annealing temperature is low, there is not enough driving force to cause significant grain growth, which will result in smaller grain sizes. The relations between the core loss and the average grain size of the electrical steels with or without hot band annealing are shown in Fig. 4b–f. It is seen that, in some cases, e.g., without hot band annealing or hot band annealed at 950 °C, there is a clear correlation between the core loss and the average grain size, i.e. the larger the grain size, the lower the core loss, which agrees with well-known magnetization theory, i.e. the larger the grain size, the fewer the grain boundaries (pinning sites), thus the lower the resistance to magnetization. However, in all the other cases, such a correlation is not obvious or does not exist. Nevertheless, it does illustrate that in most cases, the smallest average grain size corresponds to the largest core loss (except hot banding annealing at 1000 °C), while the largest average grain size usually leads to the lowest core loss (except hot band annealing at 850 and 900 °C). Apparently, grain size is also not the only factor that determines the core loss of the material.

Discussion From the above analysis, it is seen that hot band annealing has considerable effects on the texture, grain size, and core loss of the steel after final annealing at different temperatures. Theoretically, hot band annealing is beneficial to the core loss as the average grain size after final annealing is usually increased, which can normally reduce the core loss by reducing the pinning sites (grain boundaries). However, it should be noted that the relationship between the core loss and grain size is not a simple linear relation, since the three core loss components, i.e., hysteresis loss (Wh ), eddy current loss (Wc ), and excess loss (We ), show √ different correlations √ with respect to the grain size (D), i.e. Wh ∝ 1/D, Wc ∝ D [18], and We ∝ D [19]. Nevertheless, Lee et al. [18] have shown that if the grain size is smaller than ~100 μm, the hysteresis loss (Wh ) will dominate the core loss, and the core loss decreases with the increase of the grain size. Figure 5a plots the core losses from this study with

Fig. 5 Relationship between core loss and grain size: a average grain size smaller than ~120 μm, b average grain size greater than ~120 μm

406

Y. He et al.

average grain sizes smaller than ~120 μm. It is seen that the core loss approximately decreases linearly with the increase of grain size. If the grain size is greater than ~120 μm (Fig. 5b), both the hysteresis loss (Wh ) and eddy current loss (Wc ) will contribute to the total core loss. Since they display opposite trends with respect to the grain size, there is no clear dependence of the core loss on the grain size (Fig. 5b). It is also noticed that, in some cases, hot band annealing (e.g., hot band annealing at 950 °C and final annealing at 700 °C) can considerably decrease the texture factor (Fig. 3d) and benefit the core loss, but the grain size produced under these conditions is small (Fig. 4e); thus the core loss is still relatively high. Apparently, the effects of grain size and texture on core loss are convoluted. A practical way of elucidating the individual effects of these two parameters is to produce microstructures having the same grain size but with different textures or having the same texture but with different grain sizes. However, the results from this study do not provide such microstructures for statistical analysis. It should also be noted that, the core loss is closely related to the number, size, and distribution of particles and defects within the microstructure, which are not reflected in the grain size or texture, but are affected by the hot band annealing and final annealing parameters employed. This should also be considered in the evaluation of the core loss with respect to the annealing conditions. Further investigation on the effects of all these parameters is needed.

Summary and Conclusions In this study, the effect of hot band annealing and final annealing temperatures on the texture, grain size, and core loss of a 1.2 wt% Si non-oriented electrical steel was investigated. The findings can be summarized as follows: • At the same final annealing temperature, hot band annealing at different temperatures can all increase the final average grain size of the electrical steel; generally, the higher the hot band annealing temperature, the larger the final grain size. • At the same hot band annealing temperature, the higher the final annealing temperature, the larger the final average grain size of the steel; the final average grain sizes vary from ~30 to ~220 μm. • If the average grain size is smaller than about ~120 μm, there exists an approximately linear relation between the core loss and the grain size: the larger the grain size, the smaller the core loss; when the average grain size is greater than ~120 μm, such a relation does not exist anymore. • Hot band annealing at 850 °C for 4 h followed by final annealing at 800 °C for 24 h resulted in the lowest core loss, which is 26% lower than that of the steel without hot band annealing and final annealed at 700 °C. • Although the variations of hot band annealing and final annealing temperatures cause considerable variations of the crystallographic texture, the core loss and the texture factor do not show a clear correlation, due to the convoluted effect of the grain size.

Effect of Hot Band Annealing and Final Annealing Temperatures …

407

Acknowledgements Funding for this research was provided by Stelco Inc., Mitacs, and the Office of Energy Research and Development, Natural Resources Canada. The authors are grateful to Peter Newcombe, Doug McFarlan, Howard Webster, and David Saleh for casting the steel, and to Michael Attard for rolling and annealing the steel. Jian Li and Renata Zavadil are gratefully acknowledged for their assistance in EBSD measurements.

References 1. Fiorillo F, Bertotti G, Appino C, Pasquale M (2016) Soft magnetic materials. In: Peterca M (ed) Wiley encyclopedia of electrical and electronics engineering. Wiley, Hoboken, New Jersey 2. Moses AJ (1990) Electrical steels: past, present and future developments. IEEE Proc A (Phys Sci Meas Instrum Manag Educ) 137(5):233–245 3. Lyudkovsky G, Rastogi PK, Bala M (1986) Nonoriented electrical steels. JOM 38(1):18–26 4. Matsumura K, Fukuda B (1984) Recent developments of non-oriented electrical steel sheets. IEEE Trans Magn 20(5):1533–1538 5. Mehdi M, He Y, Hilinski EJ, Edrisy A (2017) Effect of skin pass rolling reduction rate on the texture evolution of a non-oriented electrical steel after inclined cold rolling. J Magn Magn Mater 429:148–160 6. Kestens L, Jacobs S (2008) Texture control during the manufacturing of non-oriented electrical steels. Text Stress Microstruct. 1–9. Article ID 173083 7. Humphreys FJ, Hatherly M (2012) Recrystallization and related annealing phenomena. Elsevier, New York 8. Yasiki H, Kaneko T (1992) Effect of hot-band annealing on anisotropy of magnetic properties in low-Si semi-processed electrical steels. J Magn Magn Mater 112:200–202 9. de Campos MF, Landgraf FJG, Takanohashi R, Chagas FC, Falleiros IGS, Fronzaglia GC, Kahn H (2004) Effect of the hot band grain size and intermediate annealing on the deformation and recrystallization textures in low silicon electrical steels. ISIJ Int 44(3):591–597 10. Takanohashi R, Landgraf FJG (2006) Effect of hot-band grain size and intermediate annealing on magnetic properties and texture of non-oriented silicon steels. J Magn Magn Mater 304:e608–e610 11. Hölscher M, Raabe D, Lücke K (1991) Rolling and recrystallization textures of bcc steels. Steel Res Int 62(12):567–575 12. Hu H (1974) Texture of metals. Text Stress Microstruct 1(4):233–258 13. Mehdi M, He Y, Hilinski EJ, Kar NC, Edrisy A (2019) Non-oriented electrical steel with core losses comparable to grain-oriented electrical steel. J Magn Magn Mater 491:165597 14. ASTM A343/A343M-14 (2014) Standard test method for alternating-current magnetic properties of materials at power frequencies using wattmeter-ammeter-voltmeter method and 25-cm Epstein test frame. ASTM International, West Conshohocken, PA 15. Mehdi M, He Y, Hilinski EJ, Kestens LAI, Edrisy A (2020) The evolution of cube ({001}) texture in non-oriented electrical steel. Acta Mater 185:540–554 16. Mehdi M, He Y, Hilinski EJ, Kestens LA, Edrisy A (2019) The origins of the Goss orientation in non-oriented electrical steel and the evolution of the Goss texture during thermomechanical processing. Steel Res Int 90(7):1800582 17. Ridha AA, Hutchinson WB (1982) Recrystalllization mechanisms and the origin of cube texture in copper. Acta Metall 30:1929–1939 18. Lee KM, Park SY, Huh MY, Kim JS, Engler O (2014) Effect of texture and grain size on magnetic flux density and core loss in non-oriented electrical steel containing 3.15% Si. J Magn Magn Mater 354:324–332 19. Bertotti G, Di Schino G, Ferro Milone A, Fiorillo F (1985) On the effect of grain size on magnetic losses of 3% non-oriented Si-Fe. J Phys Colloq 6 46:C6–385:1–4

Magneto-Mechanical Properties and Magnetocaloric Behaviour of Rapidly Solidified Melt-Spun Ni50 Mn28 Ga22 Heusler Alloy D. K. Satapathy, P. D. Babu, I. A. Al-Omari, and S. Aich

Abstract Rapidly solidified melt-spun ribbons of Ni50 Mn28 Ga22 Heusler alloy were prepared at two different wheel speeds of 1300 and 1600 RPM and were annealed at different temperatures over different time periods to study the magneto-mechanical properties and magnetocaloric behaviour of the alloy. Annealed ribbons showed higher MFIS (magnetic field induced strain) than as-spun ribbons; ribbon prepared at 1300 RPM and annealed at 900 °C for 5 h (1300 NMG5 900 ) showed the highest MFIS value ~1547 με. From the texture analysis, a higher magnetization value at 90° angle of alignment of the sample surface with respect to the field direction indicates that a easy axis might be a possibility. Magnetocaloric behaviour of the ribbons were investigated with the help of SQUID magnetometer. Sm values were calculated from the SQUID data which were further used to calculate the RC (refrigeration capacity) values. The highest RC value was obtained for 1300 NMG5 800 ; 273 J/kg. Keywords Heusler alloy · Magnetic field induced strain · Magneto-mechanical properties

Introduction NiMnGa-based alloys are well known for their magnetic field induced strain (MFIS) as well as for their magnetocaloric properties. However, most of these reports focussed on single crystals while very little is known about the behaviour of the D. K. Satapathy · S. Aich (B) Department of Metallurgical and Materials Engineering, Indian Institute of Technology Kharagpur, Kharagpur 721302, India e-mail: [email protected] P. D. Babu UGC-DAE Consortium for Scientific Research, BARC, Mumbai centre, Mumbai 400085, India I. A. Al-Omari Department of Physics, Sultan Qaboos University, P.O. Box 36, PC 123 Muscat, Oman © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_37

408

Magneto-Mechanical Properties and Magnetocaloric Behaviour …

409

polycrystalline materials. Magnetic field strain values as high as 12% have been reported on single crystals of NiMnGa-based alloys [1]. However, in polycrystalline materials the maximum strain achieved was 1% with an external stress aiding the magnetic field [2]. Similar studies done on NiMnGa-based melt-spun ribbons doped with other elements such as Al have generated strain values up to 550 ppm [3]. To achieve better magnetic field induced strain, the fraction of martensitic twins (especially modulated twins) should be high. Modulated martensitic twins are reported to possess low twinning stress and high mobility which enhances the magnetic shape memory effect. Similarly, Ni-based Heusler alloys were found to have superior magnetocaloric properties. Refrigeration capacity value of 341 J/kg was obtained at a field of 90 kOe [4]. Magnetocaloric effect can be conventional (negative value of Sm ) or inverse (positive value of Sm ). Sm is the change in the magnetic entropy value when an external field is applied. A change in the magnetic entropy value results in an increase or decrease in the temperature, depending on whether, the magnetocaloric effect is inverse or conventional. Inverse magnetocaloric effect has been observed in NiMnSn-based alloys [5]. In this report, magneto-mechanical as well as magnetocaloric properties of melt-spun ribbons were discussed.

Experimental Methods Ingots of Ni50 Mn28 Ga22 were prepared by TIG arc melting technique. These ingots were then used to make ribbons via melt spinning at 1300 and 1600 RPM followed by annealing at different temperatures over different time periods (mentioned in Table 1) followed by quenching in ice cooled water. Bulk specimens were separately annealed at 900 °C for 5 h followed by quenching in ice cooled water and also furnace cooled to study the magnetocaloric properties and compare them with those of the ribbons. The melt-spun samples are nomenclated as RPM NMGT θ , where RPM is the wheel rotation speed, T is the time of annealing and θ is the annealing temperature. The annealing parameters are mentioned in Table 1. The bulk specimens are simply referred as Q-NMG5 900 and FC-NMG5 900 for quenched and furnace-cooled samples, respectively. The MFIS values of the ribbons were measured using strain gauges and strain indicator at a field of 9000 Oe. To investigate the magnetocaloric properties, field of 50 kOe was used. The temperature range was chosen to be 355 to 385 K with the temperature step of 3 K. The magnetic measurements were done in a Physical Property Measurement System (PPMS) from Quantum Design.

Results and Discussion Phase analysis and microstructural studies of the melt-spun ribbons confirmed the presence of modulated martensitic phases in the annealed ribbons [6]. Both 5 M and 7 M modulated structures were observed in the electron micrographs of the annealed ribbons.

410

D. K. Satapathy et al.

The twinned martensitic features can be seen in Fig. 1. These twins have a two variant arrangement with different orientations. Hence, when a magnetic field is applied, the favourable variant grows at the expense of the unfavourably oriented variant resulting in a net strain. Figure 1d shows the presence of secondary twinning which is a desirable feature of shape memory alloys. Multiple twinning happens to accommodate the strain which originates during martensitic transformation. Multiple twinning results in reduction of the twin sizes which reduces the elastic energy. These twins have higher mobility and lower twinning stress than the primary twins. The martensitic twins are responsive to magnetic field as well as external stress. To measure the MFIS, strain gauges were attached to the ribbons using adhesive followed by measurement in a strain indicator. The field direction was normal to the surface of the ribbons. Figure 2 contains the magnetic field induced strain plots of ribbons melt spun at 1300 and 1600 RPM. The absolute values of the MFIS are given in Table 2. MFIS values as high as 1547 μE have been observed in the ribbons. The annealed ribbons have performed better than the as-spun ribbons due to the higher fraction of martensitic phases. The field induced strain is directly dependent on the lattice parameters of the selected composition since the difference in the lattice parameters manifests itself as the strain. This field induced strain is reversible and emanates from the reorientation of the twin variants. Since, the lattice parameters [6] in the concerned specimens are almost similar in magnitude, the maximum strain that can be expected is around 4.3% (the maximum strain is achievable in a single crystal). The maximum strain that has been achieved in the current work is 0.15% for 1300 NMG5 900 . Magnetocrystalline anisotropy can also play an important role in influencing the recoverable strain since Table 1 Annealing parameters of the melt-spun ribbons with temperatures and times of annealing. The ribbons with blank entries are the as-spun ribbons

Sample ID

Temperature (°C)

Time (h)

1300 NMG

–

–

1300 NMG 400 5 1300 NMG 800 5 1300 NMG 900 1 1300 NMG 900 5 1300 NMG 900 10 1300 NMG 1000 5 1600 NMG

400

5

800

5

900

1

900

5

900

10

1600 NMG 400 5 1600 NMG 800 1 1600 NMG 800 5 1600 NMG 900 1 FC-NMG5 900 Q-NMG5 900

1000

5

–

–

400

5

800

1

800

5

900

1

900

5

900

5

Magneto-Mechanical Properties and Magnetocaloric Behaviour …

411

Fig. 1 SEM micrographs of a Q-NMG5 900 , b FC-NMG5 900 showing extensively twinned microstructural features. c Twinned microstructure in 1300 NMG5 400 d primary and secondary twins observed in 1300 NMG1 900 : the red line represents the primary twin while the lamellar region between the red arrows is the secondary twin. (Color figure online)

Fig. 2 Semi log plots of MFIS vs H of some of the selected ribbons. MFIS values are in logarithmic scale

412 Table 2 Absolute values of magnetic field induced strain (MFIS) in the melt-spun ribbons

D. K. Satapathy et al. Sample id

MFIS (μE)

1300 NMG

167

1300 NMG 400 5 1300 NMG 800 5 1300 NMG 900 1 1300 NMG 900 5 1300 NMG 900 10 1300 NMG 1000 5 1600 NMG

526

1600 NMG 400 5 1600 NMG 800 1 1600 NMG 800 5 1600 NMG 900 1

166 337 1547 1201 730 122 86 286 844 68

without it there will not be any magnetically induced stress. Ideally, the stress generated should be small enough to mobilize the twin boundaries to facilitate twin variant selection. Generally, the twinning stress or the critical stress required to mobilize the twin boundaries should be low for a good shape memory alloy. The MFIS has not achieved saturation as the applied field is small. Ribbons melt-spun at 1300 RPM have better MFIS values than those melt-spun at 1600 RPM. As mentioned earlier, the fraction of martensitic twins (modulated) contribute to high MFIS values. Correlating the MFIS values with the grain size distribution in the ribbons can also give an insight into high values for 1300 RPM [6]. Annealing had resulted in higher average grain size in 1300 RPM melt-spun ribbons. However, in 1600 RPM melt-spun ribbons, the average grain size was somewhat similar [6]. Besides grain size variations, extensive precipitation was also observed in the ribbons annealed at higher temperatures. The precipitations were rich in manganese and concentrated along the grain boundaries [6]. Manganese segregation was common in most of the annealed ribbons. To study the magnetocaloric properties, Sm values were calculated. Sm values are dependent on the dM/dT which is obtained from the magnetic isotherms [7]. Table 3 contains the Sm as well as the RC values of all the specimens. Since the dM/dT values are almost similar for all the specimens, not much difference was observed in the Sm values. The only variation is in the RC values. The highest RC was 273 J/kg which is at par with the results reported in literature. The difference in the RC values can be attributed to TFWHM which is calculated from the Sm vs T plots [7]. High TFWHM is a feature of second order phase transformation (ferromagnetic to paramagnetic) which was confirmed from the absence of any negative slopes in the Arrott plots [7]. A negative slope would point towards inverse magnetocaloric effect which is mostly associated with first order phase transformation. First order phase transformation can result in very high Sm values but smaller TFWHM . Second

Magneto-Mechanical Properties and Magnetocaloric Behaviour …

413

Fig. 3 Refrigeration capacities (RC) of ribbons melt-spun at a 1300 RPM and b 1600 RPM. c Refrigeration capacity of bulk annealed specimens Table 3 RC and − ΔS m values of as-spun and annealed ribbons at a field of 50 kOe

Sample id

−Sm (J kg−1 K−1 )

RC (J kg−1 )

1300 NMG

7.4

119

1300 NMG 400 5 1300 NMG 800 5 1300 NMG 900 1 1300 NMG 900 5 1300 NMG 900 10 1300 NMG 1000 5 1600 NMG

8.2

154

7.8

273

7.5

113

8.3

142

8.4

160

8.4

144

8.2

162

8.1

166

8.6

141

8.4

142

7.7

132

46

184

4.8

56

1600 NMG 400 5 1600 NMG 800 1 1600 NMG 800 5 1600 NMG 900 1 FC-NMG5 900 Q-NMG5 900

414

D. K. Satapathy et al.

order phase transformation is mostly associated with small Sm but high TFWHM . However, high Sm was observed in FC-NMG5 900 : 46 J/kg K. The contribution of annealing is manifold. Effect of annealing on the magnetic isotherms cannot be ruled out since the annealed ribbons showed the best RC. But somewhat similar values obtained for ribbons melt spun at 1600 RPM brings in the influence of grain size on the magnetic properties. The grain sizes for ribbons melt spun at 1300 RPM varied greatly after annealing while for 1600 RPM the grain sizes were almost similar [6]. Similarly, the bulk specimens have completely different phase and microstructural features [6]. The effect of microstructure on the behaviour of the magnetic isotherms needs to be studied more in order to achieve a better understanding of the phenomenon. Changing the composition can influence the feature of the magnetic isotherms [8] as well as the microstructural features such as grain size distribution [6]. Hence, it will be quite interesting to see the effect of microstructural variations on the magnetocaloric effect since magnetic features such as TC and saturation magnetization are greatly dependent on composition [9]. Also, composition can decide if the material will be ferromagnetic or anti-ferromagnetic in nature. For example, increasing the amount of manganese will reduce the distance between the Mn–Mn bonds thus imparting anti-ferromagnetic behaviour. An antiferromagnetic to ferromagnetic transition is a first order transition and will result in an inverse magnetocaloric effect. A second order transition is better since there is literally negligible thermomagnetic hysteresis and the TFWHM is wide. Also, the Tpk (peak temperature used to obtain universal plots [7, 10]) is independent of the field in second order transition.

Conclusions The magneto-mechanical properties and magnetocaloric behaviour of melt-spun and annealed bulk specimens were studied. The magnetic field induced strain (MFIS) was measured using strain gauges and the MFIS value as high as 1547 μE was obtained. The microstructural features revealed a highly twinned structure especially in the bulk annealed specimens. Martensitic twins were also observed in the melt-spun ribbons. Magnetic entropy as well as refrigeration capacities was also calculated. High Sm value was observed for annealed bulk specimen. Refrigeration capacity of 273 J/kg was observed.

References 1. Sozinov A, Lanska N, Soroka A, Zou W (2013) 12% magnetic field-induced strain in Ni-MnGa-based non-modulated martensite. Appl Phys Lett 102: 021902. 2. Gaitzsch U, Potschke M, Roth S, Rellinghaus B, Schultz L (2009) A 1% magnetostrain in polycrystalline 5M Ni–Mn–Ga. Acta Mater 57:365–370

Magneto-Mechanical Properties and Magnetocaloric Behaviour …

415

3. Singh S, Roy RK, Ghosh M, Mitra A, Panda AK (2013) Martensitic transformation and magneto-strain in melt spun NiMnGaAl Ferromagnetic Shape Memory Alloys. Intermetallics 43:147–151 4. Zhang X, Zhang B, Yu S, Liu Z, Xu W, Liu G, Chen J et al (2007) Combined giant inverse and normal magnetocaloric effect for room-temperature magnetic cooling. Phys Rev B 76:132403 5. Krenke T, Duman E, Acet M, Wassermann EF, Moya X, Mañosa L, Planes A (2005) Inverse magnetocaloric effect in ferromagnetic Ni–Mn–Sn alloys. Nat Mater 4:450–454 6. Satapathy DK, Biswas S, Aich S (2021) Microstructure and micro-texture evolution in rapidly solidified melt-spun Ni50 Mn28 Ga22 ribbons. J Mag Mag Mater 527:167784 7. Satapathy DK, Al-Omari IA, Aich S (2021) Magnetocaloric properties of N150 Mn28 Ga22 meltspun ribbons. Phil Mag Lett. https://doi.org/10.1080/09500839.2021.1962015 8. Pavan KN, Sagar E, Babu PD, Srinivas A, Manivel Raja M (2019) Investigation of low temperature magnetization, specific heat and magnetocaloric effect in Ho doped TbMnO3 multiferroic system. Solid State Sci 94:54–63 9. Sarkar SK, Sarita BPD, Biswas A, Siruguri V, Krishnan M (2016) Giant magnetocaloric effect from reverse martensitic transformation in Ni-Mn-Ga-Cu ferromagnetic shape memory alloys. J Alloys Comp 670:281–288 10. Franco V, Conde A (2010) Scaling laws for the magnetocaloric effect in second order phase transitions: from physics to applications for the characterization of materials. Int J Refrig 33:465–473

Part XI

Advanced Materials for Energy Conversion and Storage 2022

Characterization of AlCl3 -Urea Electrolyte for Speciation, Conductivity, and Electrochemical Stability and Its Application in Al-Ion Batteries Monu Malik, Kok Long Ng, and Gisele Azimi

Abstract In the present study, the physicochemical properties of AlCl3 and urea mixtures, a potential electrolyte for Al-ion batteries, are investigated by changing the molar ratio of AlCl3 /urea in the range of 1.0–1.6. In recent years, Al-ion batteries are receiving growing attention due to the high abundance and low cost of Al, ease of handling in an ambient environment, and high theoretical capacities. A urea-based electrolyte is a cost-effective and environmentally friendly alternative to expensive 1-Ethyl-3-methylimidazolium chloride ([EMIM]Cl)-based electrolyte for Al-ion batteries. Several characterization techniques such as nuclear magnetic resonance spectroscopy, electrochemical impedance spectroscopy, and linear sweep voltammetry are used to determine the speciation of ionic moieties, ionic conductivity, and electrochemical stability of this complex system. Based on the obtained results, the best composition is used as an electrolyte in an Al-ion battery, which delivered a specific capacity of 73 mAh g–1 at 100 mA g–1 . Keywords Aluminum-ion battery · Ionic liquid analogues · Ionic conductivity · Electrochemical stability · Graphene nanoplatelets

Introduction With growing attention to climate change, the focus around the world is moving towards sustainable development, where electrification of the transportation sector and renewable energy storage will play an important role. This move has led to a research wave on the development of low-cost and high-performing batteries using earth-abundant resources. Among the emerging candidate for post-lithium-ion M. Malik · G. Azimi (B) Department of Chemical Engineering and Applied Chemistry, University of Toronto, 200 College Street, Toronto, ON M5S35, Canada e-mail: [email protected] K. L. Ng · G. Azimi Department of Materials Science and Engineering, University of Toronto, 184 College Street, Toronto, ON M5S34, Canada © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_38

419

420

M. Malik et al.

batteries, aluminum (Al)-based batteries are of great interest due to high abundance, low cost, and high volumetric and gravimetric capacity. The electrodeposition of aluminum significantly affects the battery performance but is only viable in organic solvent and ionic liquids because of its high reactivity and high negative reduction potential [1]. Therefore, the investigation of the electrodeposition of aluminum at room temperature is highly important for the development of high-performing aluminum ion batteries. Numerous electrolytes have been proposed by researchers for electrodeposition of aluminum including molten salts eutectics [2] and aprotic polar organic solvents that were previously used in Li and Na batteries [3, 4]. However, the application of proposed organic solvents remains limited due to their low thermal stability and high volatility, the very low solubility of aluminum salts, and narrow electrochemical windows [5]. In recent years, ionic liquids ionic liquid analogues (ILs) have been proposed and utilized for Al electrodeposition to overcome the shortcomings of the organic solvent. Several ionic liquids ionic liquid analogues were suggested using an aluminum salt and an organic solvent that have low volatility and flammability, high thermal stability, and high solvability and remain liquid at room temperature. For aluminum salt, AlCl3 is mainly used as a source by researchers due to its better stability compared with Br− and I− sources [5–7]. For the organic component, a list of molecules were investigated such as 1-butyl-1-methylpyrrolidinium bis (trifluoromethylsulfonyl) amide, 1-ethyl-3-methylimidazolium bis (trifluoromethylsulfonyl) amide, and dicyanamide [8–11]. Among these compounds, 1-ethyl-3-methylimidazolium chloride ([EMIm]Cl) is the most widely utilized candidate due to its high electrical conductivity, low vapor pressure, liquid state over a large composition range, and wide stable electrochemical window [12, 13]. These chloroaluminate electrolytes made of AlCl3 and [EMIm]Cl can be classified as acidic, basic, and neutral depending on the AlCl3 content. However, the only acidic composition can participate in the aluminum plating and stripping through the reversible reaction 4Al2 Cl7 − + 3e−  Al + 7AlCl4 − [12, 13]. Despite the high performance of AlCl3 /[EMIm]Cl ILs as the electrolyte in aluminumion batteries, the high cost of [EMIm]Cl and its hazardous nature impelled the search for alternative organic solvents. This leads to the development of a new class of ionic liquids obtained from mixing of an oxygen donor amide ligand and AlCl3 and commonly known as ionic liquids analogues (ILAs) or deep eutectic solvents (DESs) [14]. After Aboot et al. [15] showed that the electrodeposition of aluminum can take place in the mixtures of AlCl3 and urea-based ILAs, a few groups have investigated the application of AlCl3 /urea ILAs as the electrolyte in aluminum battery systems and reported excellent performance [16–18]. A urea-based electrolyte for aluminum-ion batteries offers both economic and environmental benefits over ionic liquids because of the large production of urea and its environmental friendliness as a commercial fertilizer [19]. Although AlCl3 /urea ILAs have been studied by few researchers, only a few physicochemical properties of this system such as electrical conductivity, viscosity, and density have been reported [1] and other properties are missing. Therefore, the objective of the present study is to investigate the physicochemical properties of AlCl3 -urea ILAs both in neural

Characterization of AlCl3 -Urea Electrolyte for Speciation, Conductivity …

421

and acidic regions. Several characterization techniques including nuclear magnetic resonance (NMR), linear sweep voltammetry (LSV), electrochemical impedance spectroscopy (EIS) were employed to find speciation, ionic conductivity, and electrochemical stability of the electrolyte. The physical state of the electrolyte at different temperature was also investigated. Based on the obtained results, best composition of AlCl3 /urea ILAs was identified and utilized as electrolyte. The performance of the selected composition was tested in an electrochemical cell with aluminum as the anode and graphene nanoplatelets as the cathode.

Experimental Procedures Chemicals and Materials Anhydrous aluminum chloride (99.985%) and aluminum shots (99.999%) were obtained from Alfa Aesar (USA). Urea and sodium alginate (≥99.5%) were purchased from Bioshop Canada Inc. (Canada) and Landor Trading Co. Ltd. (Canada), respectively. Molybdenum sheet (130 µm thick, 99.95%) and aluminum foil (50 µm thick, 99.999%) were obtained from Beijing Loyaltarget Tech. Co., Ltd. (China). Whatman Glass microfiber separators (GF/A) were purchased from SigmaAldrich Co. (USA). Graphene nanoplatelets (GNP H15) were purchased from XG Science (USA).

AlCl3 -Urea Electrolyte The AlCl3 -urea ILAs were prepared inside an argon-filled glovebox by slowly mixing urea powder with anhydrous AlCl3 (AlCl3 /urea = 1.0–1.6) in a glass beaker under constant magnetic stirring. Since the mixing process was exothermic, the temperature of the mixture was regulated by wrapping the beaker with an ice gel pad to prevent electrolyte decomposition. The mixture was stirred overnight inside the glovebox and the obtained product was a transparent, yellowish, and viscous liquid. Then the liquid was heated at 60 °C for 30 min and subjected to vacuum for 10 min to remove gaseous impurities. The steps of heating and vacuuming were repeated before storing the liquid in a glass vial at room temperature.

422

M. Malik et al.

Linear Sweep Voltammetry and Electrochemical Impedance Spectroscopy A Linear Sweep Voltammetry (LSV) was performed in a three-electrode configuration with a potentiostat using a Teflon cell to investigate the stability of the electrolyte. An aluminum wire was used as the reference electrode, and molybdenum (Mo) wire and a Mo rod were used as working and counter electrodes, respectively. The Teflon cell was assembled inside the glovebox with appropriate ILA compositions and the edges were further sealed with a Teflon tap to avoid contact with moisture or air. The LSV curve was collected in a voltage range of –0.20 to 3.00 V (vs. Al) to investigate the oxidation/reduction potential of aluminum and the electrochemical stability window of the ILAs. The same Teflon cell with a two-electrode arrangement was used to measure the ionic conductivity of the considered ILA compositions at various temperatures (25–85 °C) through electrochemical impedance spectroscopy (EIS) technique. For the EIS measurements, two Mo rods of the same diameter were used as both working and counter electrodes and the measurements were compiled in a frequency range of 0.1 Hz–100 kHz versus open-circuit potential.

Nuclear Magnetic Resonance Analyses The samples to collect 27 Al NMR spectra were prepared inside the glovebox using a coaxial insert and 5 mm NMR tubes, where chloroform with 0.05% Tetramethylsilane (TMS) was used as deuterium. The 1D 27 Al NMR spectra were acquired for different compositions at 182.345 MHz using an Agilent DD2 spectrometer with 5.0 s recycle delay and 512 transients. The obtained spectra were further processed in MestreNova software for baseline and phase correction, where 27 Al chemical shifts were indirectly referenced to the spectrometer 2H lock.

Battery Preparation and Testing The battery cathode was prepared using graphene nanoplatelets, carbon black, and sodium alginate slurry coated on Mo current collector using a doctor blade following the procedure described elsewhere [16]. The prepared cathode was dried at 80 °C under vacuum for 24 h and weighed immediately to find the amount of active mass coated on the current collector. The coated surface area was determined using Image J software to find the loading of the active material. The prepared cathode was partially heat-sealed in an aluminum-laminated film pouch with a piece of L-shaped pure aluminum sheet as anode and glass fiber membranes in between the two. Around 1.5 mL of selected ILA as electrolyte was injected into the assembled cell inside the glovebox using a pipette and the remaining open end was quickly heat-sealed outside

Characterization of AlCl3 -Urea Electrolyte for Speciation, Conductivity …

423

the glovebox. The cells assembled with different compositions of ILA were independently used for electrochemical and cyclic voltammetry (CV) tests. The CV test was performed in a voltage range of –0.50 to 2.40 V (vs. Al) using a potentiostat and considering cathode as the working electrode and aluminum anode as both auxiliary and reference electrodes. For electrochemical performance testing, cells were charged/discharged to 2.20 V/1.00 V, respectively, using a battery tester, and the current density during charging/discharging was varied from 100 to 1000 mA g–1 .

Results and Discussion Physical State The obtained 1.0–1.5 ILA compositions were clear liquids at room temperature and some white solids were observed above 1.5 molar composition due to supersaturation of AlCl3 [20], as shown in Fig. 1a. However, the 1.0 and 1.1 molar compositions solidified after a week of storage at room temperature, which leaves only 1.2–1.5 molar compositions as a clear liquid. To evaluate the operating temperature range of the prepared ILAs compositions, they were subjected to a controlled temperature environment in the range of –25 to 50 °C with an interval of 25 °C. Figure 1b represents the physical state of investigated ILAs indicating that only 1.2–1.5 molar compositions remained clear liquid at –25 to 25 °C and all compositions including 1.6 molar becomes clear liquid at 50 °C.

Fig. 1 a Prepared AlCl3 -urea ILAs compositions at room temperature, b physical state of 1.0–1.6 molar composition in a temperature range of –25 to 50 °C

424

M. Malik et al.

Speciation The type of species and their concentration define the properties of the ILA, which were investigated through NMR analysis of 1.0, 1.1, 1.2, and 1.4 molar compositions of AlCl3 -urea ILAs, and results are presented in Fig. 2. According to the literature, the type of charged species possible in AlCl3 -urea ILAs can include AlCl2 ·(urea)n + , AlCl4 – , Al2 Cl7 – , and Al3 Cl10 – [18, 20–22]. As shown in Fig. 2a, four clear peaks were observed from the 27 Al NMR spectrum of 1.0 composition (neutral system) that belongs to neutral species (AlCl3 ·(urea)2 (chemical shift: 54.54 ppm) and AlCl3 ·(urea) (89.64 ppm)) and charged species (AlCl2 ·(urea)2 + (73.75 ppm) and AlCl4 – (103.00 ppm), while a missing peak at ~97 ppm indicates the absence of Al2 Cl7 – in this composition. However, peaks that were observed in the 1.0 composition start to broaden with an increase in the molar compositions and some of them even merge as shown in Fig. 2b, c. There are several reasons for this peak broadening

Fig. 2 27 Al NMR spectra for AlCl3 -urea ILA of a 1.0 molar at 25 °C, b 1.1 molar, and c 1.4 molar at –10 °C, with deconvoluted curves. c Comparison of the relative amount of species obtained from deconvoluted NMR peaks

Characterization of AlCl3 -Urea Electrolyte for Speciation, Conductivity …

425

such as the presence of Al2 Cl7 – at higher molar composition and its dynamic equilibrium with other species, increased sample viscosity, and changes in the electronic environment of the aluminum nucleus [23]. To evaluate the relative concentration of the available species in the ILAs, the obtained NMR spectra were deconvoluted in the Origin software by considering their chemical shift in the 1.0 molar system for the selected compositions as shown in Fig. 2b, c. The chemical shift for Al2 Cl7 – was referenced from the 1.1 molar composition of AlCl3 -[EMIm]Cl IL as data for this species in the present system is not available in the literature. Based on the deconvoluted results presented in Fig. 2d, it was observed that the relative amount of AlCl4 – species remains almost stable with an increase in molar ratio, but a significant increase in Al2 Cl7 – from 0.0 to 11.3% was noticed especially from 1.0 to 1.1 molar composition. At the same time, the relative amount of neutral species (AlCl3 ·(urea)) also reduced from 55.1 to 25.0%, while cation species (AlCl2 ·(urea)2 + ) increased from 22.1 to 33.0%. This is because of the conversion of neutral species AlCl3 ·(urea) to AlCl2 ·(urea)2 + and AlCl4 – and then further conversion of AlCl4 – to Al2 Cl7 – with additional AlCl3 [20]. The relative amount of the Al2 Cl7 – keeps on increasing with further addition of the AlCl3 , suggesting that Al2 Cl7 – become more dominating species contributing to the electrodeposition of aluminum at higher molar composition. Overall, the relative amount of electroactive species significantly increased with an increase in molar composition, while the neural species are substantially decreased.

Ionic Conductivity and Electrochemical Stability The ionic conductivity of the electrolyte affects the charging and discharging rate of a battery. Therefore, the ionic conductivity of prepared ILAs was measured using the EIS technique. The results show that ionic conductivity first increases with increasing composition from 0.91 mS cm–1 at 1.1 molar to 1.45 mS cm–1 at 1.3 molar due to the increase in Al2 Cl7 – concentration, the major ionic transport carrier [24] and then decreases due to the increase in the electrolyte viscosity at higher compositions [1] (see Fig. 3a). The measurements at different temperatures (25–85 °C) show that similar to other conventional electrolytes, the ionic conductivity of the considered AlCl3 -urea ILA compositions (1.0–1.5) increased with the increase in temperature as shown in Fig. 3b. The linear scan voltammetry (LSV) analysis of the considered compositions was performed in the –0.20 to 3.00 V (vs. Al/Al3+ ) potential range. As shown in Fig. 4a, an increasing cathodic current (corresponding to Al plating) with onset potential at –0.20 V (vs. Al/Al3+ ) and an anodic wave (Al stripping) starting at ~0.00 V (vs. Al/Al3+ ) with the highest current at ~0.20 V (vs. Al/Al3+ ) was observed for all compositions except 1.0 (neutral). No appreciable electrochemical activity in the neutral composition (AlCl3 /urea = 1.0) is due to the absence of Al2 Cl7 − in this composition and the phenomenon is well explained in our previous study [17].

426

M. Malik et al.

Fig. 3 Ionic conductivity of 1.1–1.5 molar AlCl3 -urea ILAs at a 25 °C, b 25–85 °C

Fig. 4 Electrochemical stability of the AlCl3 -urea ILAs: a obtained LSV curve showing the electrodeposition/stripping of aluminum, b differential capacity as a function of voltage on a logarithmic scale showing the electrochemical stability of the AlCl3 -urea ILAs

The studied ILAs were electrochemically stable between 0.25 and ~2.40 V as no apparent oxidation or reduction peaks were observed in this voltage range as shown in Fig. 4a. The ILAs starts to oxidize above 2.40 V due to the oxidation of AlCl4 − (4AlCl4 −  2 Al2 Cl7 − + Cl2 + 2e− ) [25]. To find the initial oxidation point for these ILAs, differential capacity (dQ/dV) data were plotted against potential on a logarithmic scale which shows a steep increase in the differential capacity curve at ~2.40 V indicating that the studied ILAs oxidize beyond this potential (see Fig. 4b). The obtained results in Fig. 4b also suggest that the oxidation potential of studied ILAs decreases with increasing molar composition. This can be linked with the increased concentration of Al2 Cl7 − and increased activity of electrochemically active species at higher AlCl3 content, confirmed by NMR analysis, and can be explained through the Nernst equation for oxidation [16].

Characterization of AlCl3 -Urea Electrolyte for Speciation, Conductivity …

427

Electrochemical Cell Testing To verify the feasibility of the studied AlCl3 -urea ILAs as potential electrolytes for aluminum-ion batteries, cells were assembled with different compositions of ILAs (1.1–1.5) using aluminum foil as anode and GNP as the cathode. The cyclic voltammetry confirms the reversibility of the aluminum insertion/desertion into the cathode where multiple oxidation and reduction peaks were observed as shown in Fig. 5a. The cells were charged and discharged to 2.20 V/1.00 V, respectively, at different current densities to investigate the electrochemical performance of the studied ILAs. Figure 5b shows the voltage profile during charging and discharging of the cell at different current densities with 1.3 molar composition electrolyte where the obtained voltage plateaus are consistent with CV results. From Fig. 5c, d, it can be observed that the specific discharge capacity of the cell is increased with an increase in the concentration of AlCl3 in the electrolyte up to 1.4 compositions and then decrease. This happens is due to the mixed effect of increased concentration of electroactive, particularly Al2 Cl7 – , and increased viscosity of the electrolyte at higher AlCl3 content. To achieve the optimum performance of the battery, a balance is required between the increased viscosity (negative effect)

Fig. 5 Electrochemical performance analysis of AlCl3 -urea electrolyte GNP cathode; a cyclic voltammograms, and b voltage profile during charge/discharge at different current densities in 1.3 molar composition. The specific discharge capacity and Coulombic efficiency with different molar compositions at c 100 mA g–1 , d 400 mA g–1

428

M. Malik et al.

and the increased concentration of ionic species (positive effect) [16]. On the other hand, the Coulombic efficiency slightly decreases with an increase in the molar composition because of a decrease in electrolyte oxidation potential at higher AlCl3 content leading to loss of electrons with electrolyte oxidation and other side reactions. As shown in Fig. 5c, the highest specific discharge capacity of ~73 mAh g–1 was delivered by 1.4 molar composition ILA followed by 1.3 composition (~60 mA g–1 ) at 100 mA g–1 , while the Coulombic efficiency with 1.3 compositions was higher (84%) than 1.4 (78%). Similar trends were observed at higher current density (400 mA g–1 ) as shown in Fig. 5d, where Coulombic efficiency improved due to reduced side reactions and discharge capacity reduces due to a decrease in the fraction of the accessible cathode material available for intercalation/de-intercalation.

Conclusions In the present study, the physicochemical properties of AlCl3 -urea ILA were investigated in both neural and the acidic range (AlCl3 /urea = 1.0–1.6). The type of neutral species (AlCl3 ·urea, AlCl3 ·(urea)2 ) and electrochemical active species (AlCl2 ·(urea)2 + , AlCl4 – , Al2 Cl7 – ) and their relative amount was determined through NMR analysis. The results suggest that the relative amount of AlCl4 – remains almost the same across the studied compositions of AlCl3 -urea ILA, while the relative amount of Al2 Cl7 – significantly increased with an increase in molar composition. This indicates that Al2 Cl7 – is the dominating species contributing to the electrodeposition of aluminum at higher molar compositions. The EIS analysis shows that 1.3 molar composition has the highest ionic conductivity of 1.45 mS cm–1 among all compositions. Lastly, the electrochemical performance of the prepared ILAs was investigated in a full cell using aluminum as anode and GNP graphene nanoplatelets cathode. Based on the obtained results, 1.4 and 1.3 molar compositions were selected as the best compositions delivering a specific capacity of 73 and 60 mAh g–1 , respectively, at 100 mA g–1 . Overall, AlCl3 -urea ILAs can be used as a cost-effective electrolyte in Aluminum-ion batteries. Acknowledgements The authors acknowledge the financial support provided by Potent Group (No. 503355) and Ontario Centres of Excellence (No. 503760) for this project.

References 1. Liu C, Chen W, Wu Z, Gao B, Hu X, Shi Z, Wang Z (2017) Density, viscosity and electrical conductivity of AlCl3 -amide ionic liquid analogues. J Mol Liq 247:57–63 2. Zhao Y, VanderNoot T (1997) Electrodeposition of aluminium from nonaqueous organic electrolytic systems and room temperature molten salts. Electrochim Acta 42(1):3–13 3. Bai L, Conway B (1990) Complex behavior of Al dissolution in non-aqueous medium as revealed by impedance spectroscopy. J Electrochem Soc 137(12):3737–3747

Characterization of AlCl3 -Urea Electrolyte for Speciation, Conductivity …

429

4. Licht S, Levitin G, Yarnitzky C, Tel-Vered R (1999) The organic phase for aluminum batteries. Electrochem Solid-State Lett 2(6):262–264 5. Gali´nski M, Lewandowski A, St˛epniak I (2006) Ionic liquids as electrolytes. Electrochim Acta 51(26):5567–5580 6. Wang H, Gu S, Bai Y, Chen S, Zhu N, Wu C, Wu F (2015) Anion-effects on electrochemical properties of ionic liquid electrolytes for rechargeable aluminum batteries. J Mater Chem A 3(45):22677–22686 7. Giffin GA (2016) Ionic liquid-based electrolytes for “beyond lithium” battery technologies. J Mater Chem A 4(35):13378–13389 8. Rocher NM, Izgorodina EI, Rüther T, Forsyth M, MacFarlane DR, Rodopoulos T, Horne MD, Bond AM (2009) Aluminium speciation in 1-butyl-1-methylpyrrolidinium bis (trifluoromethylsulfonyl) amide/AlCl3 mixtures. Chem A Eur J 15(14):3435–3447 9. El Abedin SZ, Moustafa E, Hempelmann R, Natter H, Endres F (2005) Additive free electrodeposition of nanocrystalline aluminium in a water and air stable ionic liquid. Electrochem Commun 7(11):1111–1116 10. Fang Y, Yoshii K, Jiang X, Sun XG, Tsuda T, Mehio N, Dai S (2015) An AlCl3 based ionic liquid with a neutral substituted pyridine ligand for electrochemical deposition of aluminum. Electrochim Acta 160:82–88 11. Wang H, Gu S, Bai Y, Chen S, Wu F, Wu C (2016) High-voltage and noncorrosive ionic liquid electrolyte used in rechargeable aluminum battery. ACS Appl Mater Interfaces 8(41):27444– 27448 12. Lin MC, Gong M, Lu B, Wu Y, Wang DY, Guan M, Angell M, Chen C, Yang J, Hwang BJ (2015) An ultrafast rechargeable aluminium-ion battery. Nature 520(7547):324 13. Jiang T, Brym MC, Dubé G, Lasia A, Brisard G (2006) Electrodeposition of aluminium from ionic liquids: Part I—electrodeposition and surface morphology of aluminium from aluminium chloride (AlCl3 )–1-ethyl-3-methylimidazolium chloride ([EMIm] Cl) ionic liquids. Surf Coat Technol 201(1–2):1–9 14. Smith EL, Abbott AP, Ryder KS (2014) Deep eutectic solvents (DESs) and their applications. Chem Rev 114(21):11060–11082 15. Abood HM, Abbott AP, Ballantyne AD, Ryder KS (2011) Do all ionic liquids need organic cations? Characterisation of [AlCl2 ·nAmide]+ AlCl4 − and comparison with imidazolium based systems. Chem Commun 47(12):3523–3525 16. Malik M, Ng KL, Azimi G (2020) Physicochemical characterization of AlCl3 -urea ionic liquid analogs: speciation, conductivity, and electrochemical stability. Electrochim Acta, 136708 17. Ng KL, Malik M, Buch E, Glossmann T, Hintennach A, Azimi G (2019) A low-cost rechargeable aluminum/natural graphite battery utilizing urea-based ionic liquid analog. Electrochim Acta 327:135031 18. Jiao H, Wang C, Tu J, Tian D, Jiao S (2017) A rechargeable Al-ion battery: Al/molten AlCl3 – urea/graphite. Chem Commun 53(15):2331–2334 19. Antonetti E, Iaquaniello G, Salladini A, Spadaccini L, Perathoner S, Centi G (2017) Waste-tochemicals for a circular economy: the case of urea production (waste-to-urea). Chemsuschem 10(5):912–920 20. Coleman F, Srinivasan G, Kwasny MS (2013) Liquid coordination complexes formed by the heterolytic cleavage of metal halides. Angew Chem Int Ed 52(48):12582–12586 21. Wang C, Li J, Jiao H, Tu J, Jiao S (2017) The electrochemical behavior of an aluminum alloy anode for rechargeable Al-ion batteries using an AlCl3 –urea liquid electrolyte. RSC Adv 7(51):32288–32293 22. Angell M, Zhu G, Lin MC, Rong Y, Dai H (2019) Ionic liquid analogs of AlCl3 with urea derivatives as electrolytes for aluminum batteries. Adv Funct Mater, 1901928 23. Estager J, Holbrey J, Swad´zba-Kwa´sny M (2014) Halometallate ionic liquids–revisited. Chem Soc Rev 43(3):847–886

430

M. Malik et al.

24. Ferrara C, Dall’Asta V, Berbenni V, Quartarone E, Mustarelli P (2017) Physicochemical characterization of AlCl3 –1-Ethyl-3-methylimidazolium chloride ionic liquid electrolytes for aluminum rechargeable batteries. J Phys Chem C 121(48):26607–26614 25. Wang S, Kravchyk KV, Filippin AN, Müller U, Tiwari AN, Buecheler S, Bodnarchuk MI, Kovalenko MV (2018) Aluminum chloride-graphite batteries with flexible current collectors prepared from earth-abundant elements. Adv Sci 5(4):1700712

Multi-layered Thin-Film Metal Contacts for New Generation Solar Cells I. Kruhlov, A. Orlov, V. Zakiev, I. Zakiev, S. Prikhodko, and S. Voloshko

Abstract The physical and mechanical properties of Cr(30 nm)/Cu(30 nm) /Ni(30 nm) thin films magnetron sputtered on Si (001) were studied. A novel microtribological tests have been proposed to evaluate the wear resistance and adhesion of the thin films. Microtribological characteristics such as coefficient of friction, wear resistance, and adhesion were quantified for samples in the as-deposited state and also after their low-energy (1000 eV) Ar+ ion irradiation, after annealing at 450 °C in vacuum for 15 min, and finally after their combined treatment when the ion irradiation followed by the annealing. The best microtribological properties among all the samples tested were demonstrated by thin-films after their combined treatment. The results of the SIMS depth profile show that the diffusion redistribution of the major components over the entire depth of the film occurs after annealing. After ion bombardment, the redistribution of the main components was not observed, but the chemistry of the close surface layer was modified by the introduced Ar. The result of the combined treatment (ion bombardment + annealing) showed that chemical modification occurs in the surface layer and mainly in Ar, while the total distribution of the main components over the depth of the film is similar to the case of annealed sample. The improvement in thin-film mechanical properties was explained by their surface hardening associated with dislocation dynamics modified by the implanted argon ions. Therefore, pre-irradiation with ions can be recommended for long-term stability of Cu-based thin-film metal contacts for new generation solar cells that are exposed to elevated temperatures. Keywords Microtribology · Thin films · Ion irradiation · Adhesion · Long-term stability

I. Kruhlov · A. Orlov · V. Zakiev · S. Voloshko National Technical University of Ukraine “Igor Sikorsky Kyiv Polytechnic Institute”, 37 Peremogy ave., Kyiv 03056, Ukraine S. Prikhodko (B) University of California Los Angeles, 2121K-Engineering 5, 420 Westwood Plaza, Los Angeles, CA 90095-1595, USA e-mail: [email protected] V. Zakiev · I. Zakiev National Aviation University, 1 Liubomyra Huzara Ave., Kyiv 03058, Ukraine © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_39

431

432

I. Kruhlov et al.

Introduction Thin metal films based on Cu are widely used in modern technologies of micro- and nanoelectronics, which is foremost attributed to the Cu high electrical and thermal conductivity, as well as resistance to electromigration [1]. Recently, an increased interest in Cu-based materials has been also associated with photovoltaics, where Cu is one of the most promising candidates for replacing the high-cost Ag paste contacts, which currently is used in about 90% of the solar cells’ devices [2]. However, the current trend towards smaller contact sizes (line width and thickness) for electronic devices is accompanied with the long-term reliability issues associated with adhesion, diffusion, and oxidation of such materials. One of the conventional approaches to stabilize the adhesion, diffusion, and corrosion properties of Cu is the use of additional functional layers, which leads to the formation of multilayer thin-film systems. In addition, various energetic influences, such as heat treatment, are usually applied for governing the thin films’ structure, composition, or functional properties [3]. Additional processing, such as ion irradiation, could significantly change the physicomechanical characteristics of the thin-film materials, affecting the performance and long-term reliability of the final product. Wear resistance and friction coefficient, along with hardness, elasticity, and adhesion are the key parameters that determine the performance characteristics of final products with deposited thin films or coatings. Advanced nanotechnologies require the functional element’s thicknesses decreasing to tens or even several nanometers, which significantly complicates the correct examination of the physicomechanical characteristics of such systems. When studying the mechanical characteristics of thin films and coatings by the nanoindentation and scratching techniques, numerous issues arise due to the fact that calculated values do not always correspond to the actual characteristics of the material. Despite the fact that modern nanoindentation testers are equipped with sensors that measure the penetration of an indenter into the sample surface with resolution of less than a nanometer, there are objectively a number of factors that have a huge impact on the extracted quantitative information on mechanical characteristics. For example, the measurement of hardness and elastic modulus of thin films and coatings by nanoindentation is rather challenging due to the imperfection of the indenter and influence of the substrate on the test results. To exclude this influence, it is needed to restrict the depth of the indenter penetration, whereas the actual radius of the indenter tip, even a new one, is approximately 100 nm. Numerous different techniques have been proposed for correction of the indentation data [4] and adequate hardness and elasticity modulus measurements, especially for thin films. On the other hand, the performance characteristics of thin films are governed by microtribological characteristics, namely adhesion between the layers and the substrate, wear resistance of the film-substrate system, fracture resistance, and friction coefficient. Microtribological properties of interacting bodies substantially depend on their structure, surface topography, chemical and phase composition,

Multi-layered Thin-Film Metal Contacts for New Generation …

433

elastoplastic, and other characteristics of both the coating and the substrat [5]. Traditional macroscopic principles cannot be fully applied directly to microtribological tests of thin films, mainly due to the influence of various factors on the processes of friction and wear in macro- and microscales and different specific loads. A large number of studies have been carried out in the field of nanotribology using atomic force microscopes and their modifications [6]. However, the results obtained with such equipment cannot be directly applied for real engineering systems. Such studies are fundamental, the main task of which is to explain the processes of friction and wear at the nanoscale, the mechanisms of atomic interaction of contacting bodies, etc., which may lead to discoveries and the introduction of new technologies. At present, tribometers operating according to the pin-on-disk scheme are widely utilized for testing friction and wear of new materials [7] and coatings [8]. At the same time, ball counterbodies of a sufficiently large diameter and relatively large loads are used, which is not entirely applicable to thin films. In addition, the use of thin films with translational or contact interaction with other surfaces puts forward high requirements for their surface properties, such as fracture resistance, low friction and high wear resistance, adhesive strength, etc. The objectives of this study were (a) the approbation of the original technique of pin-on-disk microtribological test using indentation tester for investigation of Cr/Cu/Ni thin films after various types of their energetic treatment and (b) the estimation of the expediency of samples ion irradiation prior to thermal treatment in terms of their wear resistance and adhesion strength characteristics.

Materials and Methods Cr(30 nm)/Ni(30 nm)/Cu (30 nm) thin films were deposited on Si(001) single crystal substrates by magnetron sputtering technique at room temperature using high-purity targets of Ni (99.995 at.%), Cu (99.99 at.%), and Cr (99.95 at.%). The substrates were ultrasonically cleaned before deposition without removing of already formed native oxide layer. Low-energy Ar+ ion bombardment of films surfaces was performed using OMI0010 accelerator with energy of 1000 eV and fluence of 5 × 1016 ion/cm2 , which was pre-calibrated using standard Faraday cup. Since low beam energies were applied, no noticeable heating of the samples’ during irradiation was observed, which was controlled by a K-type thermocouple mounted at the film surface. SRIM2013 software [9] was used to evaluate the theoretical projectile range and stopping in solid after ion irradiation at applied energy. Heat treatment of the samples was carried out in a vacuum atmosphere of 10–3 Pa pressure at a temperature of 450 °C for 15 min using a heating rate of 2 °C/s. Microtribological characteristics, such as the coefficient of friction and wear resistance, were examined by the pin-on-disk method (Fig. 1a) with microcircular friction of the Rockwell indenter with simultaneous recording of the friction force on a multifunctional indentation tester Micron-gamma [10]. For this purpose, a special

434

I. Kruhlov et al.

Fig. 1 The pin-on-disk friction model (a) and the mechanism of its realization using the multifunctional indentation tester Micron-gamma (b). Key parameters of the pin-on-disk test are shown, where D is the wear scar diameter, V is the volume loss, and h is the track depth

precision rotating stage was designed allowing sample (marked in Fig. 1b as 3) to rotate relative to indenter (4) at a speed of 150–900 rpm. The rotating surface of the stage was fixed to a massive plane-parallel steel base (2) by means of precision bearings and is driven by a microelectric motor (6) through a rubber belt with a square section of 1.2 × 1.2 mm2 . Microfriction tests could be performed with friction tracks of relatively small diameter ranging from 50 to 3000 µm. Such small track diameters significantly reduce the moment of inertia and allow reliably fix studied sample to the rotating stage by pressing it with a hand press using plasticine (5) or wax. Such sample mounting provides strict parallelism and significantly reduces vertical beats during sample rotation. In addition, small friction paths the increases the counterbody resource, which is of high importance for the results repeatability, since the contact area is almost not worn during the test. The revolutions of the electric motor are regulated and stabilized by the control unit. Microtribological tests of all studied samples were performed at 300 rpm rotation speed of the precision stage and the diameter of the track was 150 µm. At the same time, the load on the indenter was gradually increased from zero to 500 mN and then decreased. This loading mode allows to estimate not only the wear resistance, but also the critical failure load at which the thin-film coating is destroyed (peeled off) from the substrate as well as its adhesion strength. The moment of film destruction during the test is accompanied by a sharp leap in the recorded friction force. 3D topography of the formed friction tracks was investigated using a non-contact optical profilometer “Mikron-alpha” with nanometric accuracy along the vertical direction [11] from the scanning area of 200 × 200 µm2 . The average depth of friction track was evaluated at few different mostly destroyed regions of the track. Chemical composition of all samples has been examined by secondary ion mass spectrometry technique using PHI ADEPT-1010 device with Cs+ 2 keV primary ion beam. Chemical analysis has been taken from a local detection area of 75 × 75 µm2 to minimize the negative effect of crater walls. Surface morphology of thin films was studied by atomic force microscopy (AFM) technique, using the Bruker Dimension Icon device.

Multi-layered Thin-Film Metal Contacts for New Generation …

435

Experimental Results 3D images of friction tracks and corresponding friction force plots of Cr(30 nm)/Ni(30 nm)/Cu(30 nm) thin films in the as-deposited state and after various treatments are shown in Fig. 2. For comparative analysis, the calculated volume losses and the depth of the friction tracks were summarized in Table 1. For all samples the track depth does not exceed the thickness of the original film, which indicates that the applied mode of microtribological pin-on-disk test directly characterizes a film, not a substrate. It follows from the analysis of the friction track of the as-deposited sample, that the film failure turned out at a load of 360 mN. 3D topography data of the wear track revealed that the destruction and flaking of the film did not occur completely, but only partially. The depth of the track in the fracture region was about 72 nm, while the volume loss was 0.5808 × 106 µm3 . The film after ion irradiation was destroyed at a lower load of 260 mN. In this case, the edge of the track shows numerous chips and delamination of the film from the substrate, which indicates the worst adhesive strength among all studied samples. The formed track had the largest average depth among the studied samples (76 nm)

Fig. 2 3D topography images of the wear tracks and friction force as a function of load of the as-deposited (a), ion irradiated (b), annealed (c), and annealed with ion pre-irradiation (d) samples

436

I. Kruhlov et al.

Table 1 Microtribological characteristics of the studied samples Sample

Critical load, mN

Volume loss, µm3 (× 106 )

Average track depth, nm

As-deposited

360

0.5808

72

Ion irradiated

260

2.1689

76

Annealed

250

1.0176

72

Annealed with ion pre-irradiation

No destruction

0.1863

27

with the volume loss of 2.1689 × 106 µm3 , which confirms the degradation of the tribological characteristics of irradiated sample in comparison with the initial one. Data of pin-on-disk microtribological test of the sample after heat treatment indicated that the critical load of its destruction was 250 mN, with a track depth of 72 nm and a volume loss of 1.0176 × 106 µm3 . Figure 3 shows the results of scanning electron microscopy, which demonstrate in more detail how the destruction of the nanoscale coating down to the substrate occurs under the pin-on-disk test. However, the application of ion irradiation followed by annealing resulted in substantially improved tribological characteristics of the film. The evidence of the high both wear resistance and adhesion after complex treatment is the fact that the film was not destroyed during the test, its friction track depth was only about 27 nm, as well as the wear volume was the smallest (0.1863 × 106 µm3 ) among studied samples. Therefore, the most encouraging tribological properties were found for the sample after the complex treatment, and the penetration depth of the indenter in this case was comparable to the thickness of upper layer (Cr) of the thin-film system. A substantial difference in the results of tribological tests of samples in the as-received state and Fig. 3 SEM images of the annealed sample after pin-on-disk microtribological test. Different areas of the sample are marked in yellow (Color figure online)

Multi-layered Thin-Film Metal Contacts for New Generation …

437

after ion irradiation, as well as samples after annealing and complex treatment, is apparently associated with ion-stimulated modification of the films surface. Figure 4 shows the AFM images of the surface morphology of the as-received (a), ion irradiated (b), annealed (c), and annealed with ion pre-irradiation (d) samples. It can be seen that irradiation led to a smoothing (polishing) of the surface resulted in a halving of the root mean square roughness (0.39 nm) as compared to the initial state (0.87 nm). Heat treatment caused the growth of grains and surface irregularities, which was accompanied by an increase of the roughness value to 2.64 nm. At the same time, the measured surface roughness of sample after complex treatment was slightly lower (2.09 nm). However, considering that the depth penetration of the indenter during microtribological tests significantly exceeded the surface roughness, the difference in its values for irradiated and non-irradiated samples is negligible [12] and cannot be a key factor causing improved tribological characteristics of the thin film after complex processing. To explore the layer-by-layer distribution of chemical elements through the depth of the films in the as-received state and after various treatments, the SIMS technique was employed (Fig. 5). For the initial film (Fig. 5a), clear interfaces between layers can be seen indicating the absence of diffusion mixing of the components. The depth profile of the irradiated sample (Fig. 5b) is similar; however, it should be emphasized that in this case the signal from Ar in the near-surface area of the film is detected. This is the evidence that ion irradiation of the thin film by the kinetic projectiles beam produced Ar ions implantation into the surface layers. According to the theoretical calculations performed using the SRIM2010 software, the depth of ions penetration (ion range) into the upper Cr layer does not exceed 1.5 nm at used energy of the beam. It follows from the depth profiling data, that in our case, the depth of ion implantation after irradiation exceeds the theoretical

Fig. 4 Surface morphology 2D and 3D scans of the as-deposited (a), ion irradiated (b), annealed (c), and annealed with ion pre-irradiation (d) samples

438

I. Kruhlov et al.

Fig. 5 SIMS depth profiles of the as-deposited (a), ion irradiated (b), annealed (c), and annealed with ion pre-irradiation (d) samples

estimations of the ion range by an order of magnitude. However, it should be noted that theoretical calculations do not take into account the influence of specific peculiarities of the crystal structure, number of defects and crystallinity, surface roughness, inhomogeneity of the chemical composition distribution, as well as the effects of ion channeling or recombination. In addition, it is specifically characteristic to thin films which have an imperfect crystal structure due to the enlarged contribution of interfaces. Therefore, the difference of the theoretical and experimental estimations of ion penetration depth into the film should be associated to the combination of these factors. The analysis of the depth profile of the annealed sample (Fig. 5c) showed that the heat treatment at a temperature of 450 °C resulted in the activation of diffusion intermixing. Diffusion of Cr atoms towards the substrate is observed, as well as mutual diffusion of Cu and Ni, which is commonly accompanied by the formation of substitutional solid solutions due to their unlimited solubility [13]. It should be also noted that after heat treatment a significant oxidation of the Cr surface with the formation of a Crx Oy oxide layer is observed. The application of ion irradiation followed by annealing (Fig. 5d) did not lead to noticeable changes of the main elements depth distribution compared to the annealed sample (Fig. 5c). However, the Cr layer got thinner due to the surface sputtering, and a moderate signal of Ar was detected, which apparently did not completely desorb during annealing due to the growth of surface oxide.

Multi-layered Thin-Film Metal Contacts for New Generation …

439

As has been reported by Onorati et al. [14], in most cases, a monotonic decrease in the sputtering rate is observed with an increase in the hardness of the coating. Moreover, the authors believe that the sputtering rate of SIMS can be applied to evaluate the hardness of thin coatings, together with nanoindentation. We have received the confirmation of this hypothesis. The comparison of Fig. 5c, d shows that a real decrease in the thickness of the Cr layer upon the low-energy ion irradiation before heat treatment practically does not change its thickness according to SIMS data, i.e., the ion sputtering rate of the Cr layer decreases. This is consistent with the microtribology measurements data.

Discussion It is known that thermal annealing of thin metal films is accompanied by a decrease in their strength characteristics, which is usually associated with decreased residual stresses and annealing of dislocations [15]. However, the combination of ion and thermal treatments led to an enlargement of the tribological characteristics of Cr(30 nm)/Cu(30 nm)/Ni(30 nm) thin films. An improvement in physicomechanical properties is detected not only compared to the annealed sample, but also compared to the initial one. It is acknowledged, that nanoscale thin films are characterized by a less perfect crystal structure than the corresponding bulk materials. It can be assumed, that in the case of a consecutive ion and thermal treatments, the implanted upon irradiation Ar atoms are strongly bonded to the atoms of the metal lattice, and the relatively low annealing temperature (450 °C) used in this study is insufficient for their complete desorption from the film. In this instance, the implanted atoms could serve as a barrier for the dislocations movement and annealing upon heat treatment and form the so-called Cottrell atmospheres [16], contributing to the hardening of the material and determining the enlarged microtribological characteristics of the films after the combined action. In turn, ion irradiation and heat treatment separately do not give a positive effect. For instance, mictotribological test data of the annealed film showed a decreased critical load value compared to the initial sample. Such tribological characteristics are typical for thin films subjected to annealing in vacuum or other atmospheres, which is accompanied by the processes of recrystallization, crystallites growth, decrease of strength, change of Young’s modulus, etc. [17]. Chemical analysis data of the irradiated only sample indicated an increased Ar content as a result of ion implantation during ion irradiation, which is a wellacknowledged effect [18]. However, ion irradiation also leads to a significant increase in number of point structural defects in the near-surface area, which is most likely responsible for the experimentally observed degradation of the adhesion and wearresistant characteristics of the irradiated sample. It should be emphasized, that in the literature there is rather contradictory information regarding the effect of ion irradiation of metal surfaces on their strength properties. It has been reported either about the positive and negative effect of irradiation on the strength characteristics

440

I. Kruhlov et al.

associated with ion-induced modifications of residual stresses, packing factor of the surface atoms, and density of the metal surface lattice governed by the concentration of implanted ions [19, 20].

Conclusions An original method of microtribology test is proposed, which allowed to estimate the wear resistance and adhesive strength of the nanosized Cr(30 nm)/Cu(30 nm)/Ni(30 nm) thin films. To interpret the observed phenomenon of enhanced tribological characteristics for the case when ion and thermal treatments were subsequently combined, the mass spectrometry of secondary ions, atomic force, and scanning microscopy techniques were additionally applied. The ion-induced modification of the near-surface area with the incorporation of Ar atoms most likely underlies the observed effect. Acknowledgements This publication is based on work supported by a grant (#G-202108-68019) from the U.S. Civilian Research & Development Foundation (CRDF Global). Any opinions, findings and conclusions, or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of CRDF Global. This study has been also partially supported by the Ministry of Education and Science of Ukraine grant (project #0121U110283).

References 1. Bahramian A, Eyraud M, Vacandio F, Knauth P (2019) Cu/Ni/Au multilayers by electrochemistry: a crucial system in electronics-a critical review. Microelectron Eng 206:25–44. https:// doi.org/10.1016/j.mee.2018.12.008 2. Lennon A, Colwell J, Rodbell KP (2019) Challenges facing copper-plated metallisation for silicon photovoltaics: insights from integrated circuit technology development. Prog Photovoltaics Res Appl 27(1):67–97. https://doi.org/10.1002/pip.3062 3. Ma H, Zou Y, Sologubenko AS, Spolenak R (2015) Copper thin films by ion beam assisted deposition: strong texture, superior thermal stability and enhanced hardness. Acta Mater 98:17– 28. https://doi.org/10.1016/j.actamat.2015.07.013 4. Firstov SA, Ignatovich SR, Zakiev IM (2009) Strength Mater 41(2):155. https://doi.org/10. 1007/s11223-009-9116-5 5. Bhushan B (ed) (2017) Nanotribology and nanomechanics: an introduction. Springer. https:// doi.org/10.1007/978-3-319-51433-8 6. Stoyanov P, Chromik RR (2017) Scaling effects on materials tribology: from macro to micro scale. Materials 10(5):550. https://doi.org/10.3390/ma10050550 7. Mechnik VA, Bondarenko NA, Kolodnitskyi VM, Zakiev VI, Zakiev IM, Ignatovich SR, Kuzin NO (2020) Effect of vacuum hot pressing temperature on the mechanical and tribological properties of the Fe–Cu–Ni–Sn–VN composites. Powder Metall Met Ceram 58(11):679–691. https://doi.org/10.1007/s11106-020-00125-w 8. Efremenko VG, Chabak YG, Fedun VI, Shimizu K, Pastukhova TV, Petryshynets I, Efremenko BV (2021) Formation mechanism, microstructural features and dry-sliding behaviour

Multi-layered Thin-Film Metal Contacts for New Generation …

9. 10. 11.

12. 13.

14.

15.

16.

17. 18. 19.

20.

441

of “Bronze/WC carbide” composite synthesised by atmospheric pulsed-plasma deposition. Vacuum 185:110031. https://doi.org/10.1016/j.vacuum.2020.110031 Ziegler JF, Ziegler MD, Biersack JP (2010) emph SRIM—the stopping and range of ions in matter Nucl. Instrum. Meth. B 268:1818. https://doi.org/10.1016/j.nimb.2010.02.091 Zakiev I, Storchak M, Gogotsi GA, Zakiev V, Kokoieva Y (2021) Instrumented indentation study of materials edge chipping. Ceram Int. https://doi.org/10.1016/j.ceramint.2021.07.133 Zakiev V, Markovsky A, Aznakayev E, Zakiev I, Gursky E (23 September 2005) Micromechanical properties of bio-materials. In: Medical imaging. International society for optics and photonics, VOL 5959, P 595916. https://doi.org/10.1117/12.628396 Jiang WG, Su JJ, Feng XQ (2008) Effect of surface roughness on nanoindentation test of thin films. Eng Fract Mech 75(17):4965–4972. https://doi.org/10.1016/j.engfracmech.2008.06.016 Kruhlov IO et al (2020) Effect of barrier underlayer on diffusion and phase composition of Ni/Cu thin films under annealing. Paper presented at the IEEE 40th international conference on electronics and nanotechnology (ELNANO), April 2020. https://doi.org/10.1109/ELNANO 50318.2020.9088920 Onorati E, Iacob E, Bartali R, Barozzi M, Gennaro S, Bersani M (2017) Experimental study by Secondary Ion Mass Spectrometry focused on the relationship between hardness and sputtering rate in hard coatings. Thin Solid Films 625:35–41. https://doi.org/10.1016/j.tsf.2017.01.038 Khojier K, Mehr MRK, Savaloni H (2013) Annealing temperature effect on the mechanical and tribological properties of molybdenum nitride thin films. Nanostruct Chem 3(1):1–7. https:// doi.org/10.1186/2193-8865-3-5 Waseda O, Veiga RG, Morthomas J, Chantrenne P, Becquart CS, Ribeiro F, Perez M (2017) Formation of carbon Cottrell atmospheres and their effect on the stress field around an edge dislocation. Scripta Mater 129:16–19. https://doi.org/10.1016/j.scriptamat.2016.09.032 Du S, Li Y (2015) Effect of annealing on microstructure and mechanical properties of magnetron sputtered Cu thin films. Adv Mater Sci Eng. https://doi.org/10.1155/2015/969580 Ryssel H, Ruge I (1986) Ion implantation Kobata J, Miura KI (2016) Effects of Ar ion bombardment by unbalanced magnetron sputtering on mechanical and thermal properties of Ti-Cu-Zr-Ni-Hf-Si thin film metallic glass. Mater Des 111:271–278. https://doi.org/10.1016/j.matdes.2016.09.005 Lin YW, Huang JH, Yu GP, Hsiao CN, Chen FZ (2015) Influence of ion bombardment on structure and properties of TiZrN thin film. Appl Surf Sci 354:155–160. https://doi.org/10. 1016/j.apsusc.2015.02.190

On Recent Development in Two-Dimensional Transition Metal Dichalcolgenides for Applications in Hydrogen Evolution Reaction Chukwudike Ukeje

Abstract The evolution of hydrogen from water through Hydrogen Evolution Reaction (HER) is considered a stimulating strategy for the production and storage of clean energy. However, a militating factor against the full implementation of HER is the shortage of electrocatalysts required to trigger the reaction. While noble metal such as Pt has been found to provide significant catalytic activity for HER implementation, nonetheless, the scarcity and cost of this rare metal pose a challenge for its utilization in large scale HER processes. Two-Dimensional Transitional Metal Dichalcolgenides (2D TMDs) have been found to possess significant catalytic properties capable of driving HER. This paper highlights the recent progress in the development of 2D TMDs for HER applications. Major highlights include recent advancements in the synthesis of 2D TMDs and strategies employed in improving their performance in the catalysis of HER. Challenges in their utilization for HER application and possible future developments are also discussed. Keywords Sustainable enenrgy · Hydrogen evolution reaction · Transition metal dichalcolgenides · Two-dimensional materials · Electrocatalysts

Introduction Sustainable energy infrastructure development is a vital topic that is of great importance in modern times. Considering the diverse range of sustainable issues facing the human race with specific regard to population growth and increase in global energy demand, our current energy infrastructure needs to be strengthened to meet the energy target of the next decade [1]. To this effect, the development of functional materials that are capable of driving energy-related technologies is an important discus in the rapidly growing field of material science and engineering. Following the success of graphene, a lot of interest has been geared towards the development of other 2D materials with exceptional properties. These include C. Ukeje (B) Department of Metallurgical and Materials Engineering, Federal University of Technology, Akure, Nigeria © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_40

442

On Recent Development in Two-Dimensional Transition Metal …

443

2D TMDs (e.g. molybdenum disulfide (MoS2 ), molybdenum diselenide (MoSe2 ), tungsten disulfide (WS2 ), and tungsten diselenide (WSe2 )), hexagonal boron nitride (h-BN), borophene (2D boron), silicene (2D silicon), germanene (2D germanium), and MXenes (2D carbides/nitrides) [2]. Beyond their intrinsically semiconducting properties, this range of materials has been studied and observed to find potential application in a wide range of areas including the fabrication of solar cells, transistors, sensors, and in the catalysis of Hydrogen Evolution Reaction (HER). With particular regard to HER, 2D TMDs have been predicted to find useful applications as catalysts for HER. Ultimately, the successful deployment of these materials in HER development will solve the obstacles facing the full implementation of HER as a result of the scarcity and cost of the rare metal Pt, which has been found to provide significant catalytic activity for HER. In line with this, it becomes important to discuss the recent advancements in the processing-structure–property relationship of 2D TMDs and how their electrocatalytic property is affected, with specific regard to HER applications.

Advancement in the Synthesis of 2D TMDs for HER Applications Enhancing charge transport properties in 2D TMDs is one of the strategies employed by researchers to fully take advantage of the exceptional properties of 2D TMDs when synthesized for energy applications. The differential in electrical conductivity or resistivity of a material as a function of charge carrier density can provide useful insights into the electronic properties of materials [3]. Carrier mobility μ is often used as a yardstick for determining the electronic quality or on/off speed of semiconductor devices and can be mathematically defined as μ = σ/en, where σ is conductivity, n is density of charge carrier, and e is the elementary charge [3]. It has been observed that carrier density is not often precisely known, thus, in literatures, many researchers often quote field effect mobility, an effective approximation for mobility derived by a transfer curve plot of σ against gate voltage (Vg) [3]. The field effect mobility provides a medium for estimating the actual carrier mobility in 2D TMDs, making it possible to describe the charge transport characteristics of these materials. Aside sharing similarities of a bandgap in the visible-near IR range with traditional silicon, 2D TMDs have been shown to exhibit higher carrier mobility [4]. The distinct change in quantum confinement and surface chemistry that accompanies materials that are scaled from their bulk form to monolayers give rise to the superb transport properties exhibited by 2D TMDs. While 2D materials such as graphene have shown extensive electron mobility at room temperature (15,000 cm2 V−1 s−1 ), however, the lack of a bandgap in this material limits its use in FET for applications in electronics and energy storage devices. Nonetheless, 2D TMDs makes it possible to surmount this challenge with their considerable band gap and electron mobility. For example, MoS2 apart from having a direct band gap of ~1.8 eV also show good mobility at ~700 cm2 V−1 s−1 [5].

444

C. Ukeje

Carrier conduction mechanism in 2D TMD has been strongly linked with carrier density or evenly, the Fermi energy EF . The changes in conduction mechanism of 2D TMDs have been found to be dependent on device resistivity or conductivity temperature [6]. Accordingly, studies have shown that effective conduction occurs in these materials as a result of a positive gate bias been able to move EF to the conduction band edge where electrons starts to gain mobility through the help of thermal excitation [6]. It has been observed that mobility values in 2D TMDs devices vary across a wide range, depending strongly on extrinsic effects such as carrier type, flake thickness, measurements temperature, and measurement conditions. 2D TMDs are extremely sensitive to the influence of minute differences in material quality, measurements conditions, device configuration, fabrication procedure, choice of substrate, mobility suppression through grain boundaries, as well as metal contacts [7, 8]. Wang et al. in their study of the electronics and optoelectronics properties of 2D TMDs suggest the absence of distinction between the mobility values of mechanically exfoliated MoS2 and CVD-grown samples. An observation which suggests that grain boundaries in CVD-TMDs have less influence on charge transport compared to the point defects that exist within each grain in CVD-TMDs [9]. As stated earlier, layer type and choice of substrates increasing affect transport properties in 2D TMDs. In this regard, room temperature mobility has been discovered to increase with increasing number of layers in 2D TMDs, an observation that was attributed to the extrinsic effect of charged impurities on substrate, which influenced thinner samples more strongly [10]. Similarly, Najmaei et al. synthesized MoS2 monolayer triangular flakes on Si/SiO2 substrate using vapor–phase reaction and observed an average mobility value of 4.3cm2 V−1 S−1 [11]. MoS2 has also been shown to exhibit higher mobility values on SiO2/Si substrate with Scandium contact (700 cm2 V−1 s−1 ) than in BN/Si substrate (33–151 cm2 V−1 S−1 ) encapsulated at room temperature [12], while MoSe2 has been found to show enhanced mobility when deposited on perylene C-coated SiO2 as opposed to bare SiO2 [13]. More so, hBN substrate has been observed to enhance mobility in WS2 [14]. In recent studies, researchers have tried investigating the charge transport properties of TMDs heterostructures. Xu et al., in their study of the role of Anderson’s rule in determining electronic, optical, and transport properties of transition metal dichalcogenide heterostructures, found that carrier mobilities in van der Waals stacked MX2 heterostructures formed by two individual monolayers depended on factors such as elastic modulus, effective mass, and deformation potential constant. It was observed that carrier mobilities of hetero-bilayers MX2 were higher than those of their monolayers [15], a result which was attributed to the higher elastic modulus associated with hetero-bilayers MX2 . In similar research, An et al. studied the electronic transport properties of transition metal dichalcogenides lateral heterojunctions. Relying on first-principle techniques, the researchers investigated the electronic properties of several types of zigzag MoS2 -WS2 lateral heterojunctions. Results from their study suggest that MoS2 -WS2 lateral heterojunctions show similar current–voltage characteristics, as well as an interesting negative differential resistive effect, owing to their identical band structure near the Fermi level [16]. It was

On Recent Development in Two-Dimensional Transition Metal …

445

also observed that electron current always propagates through the heterojunctions along the metal-termination and never along the S-termination [16]. Thermodynamically, the development of high efficient HER electrocatalysts is usually associated with reduction of overpotential (μc ) in electrocatalytic reactions [17]. Many 2D TMDs materials have been observed to posses’ active sites capable of enhancing higher HER activities. More so, electrocatalytic properties in 2D TMDs have been discovered to be largely tunable by chemical modification such as doping. Kim et al., in their study of enhanced electrocatalytic properties of MoS2 and WS2 by spontaneous gold particle decoration, found that as a result of selective decoration of gold nanoparticles at the edges and line defects in the basal planes of bulk single crystals, the MoS2 /Au and WS2 /Au hybrids showed significantly improved electrocatalytic performance for HER activities [18]. Similarly, the HER performance of MoS2 doped with varying additions of Fe, Co, Ni, Mn has been studied. Results from this study, through the electrochemical impedance spectroscopic measurements, showed that Fe doped MoS2 had the best conductivity and charge transferability [19], while Co-doped MoS2 showed better catalytic properties when compared to Mn-doped MoS2 , a result attributed to the higher number of catalytically active sites observed in Co-doped MoS2 when compared to Mn-doped MoS2 nanocrystals at room temperature.

Challenges Facing the Application of 2D TMDs in HER Despite the observed advancement in the research and exploitation of 2D TMDs for HER applications, there are, however, certain challenges encountered in the deployment of these functional materials at scale for the HER processes. One of the major concerns is the electrochemical stability of 2D TMDs when used as electrocatalysts for HER. Generally, group-VI TMD has been found to exhibit good stability under HER conditions over 10,000 cycles with minimal increase in the Tafel slope [20], while for group-V TMDs, electrochemical stability has been observed for over 12,000 cycles with a slight improvement in catalytic activity [21]. Also, it is imperative to note that atomic arrangement and stacking affects the electrochemical stability of these materials under HER conditions. For instance, the 2H phase of group-VI TMDs is known to be more stable than their IT phase [22], while the IT phase of MoS2 and IT phase of WS2 were discovered to be stable for over 100 h under HER [23]. Moreover, studies have shown that there is an activation barrier of ca. 1 eV that prevents the rapid restoration of the 2H phase under HER conditions [24].

446

C. Ukeje

Conclusion and Future Outlook The search for renewable energy sources to help solve human fundamental energy challenges has compelled researchers across the world to seek and develop functional materials capable of driving renewable energy devices. 2D TMDs is one such material which has shown promising properties as electrocatalyst for driving HER—a major source of renewable energy. In particular, recent advancement in the research and understanding of 2D TMDs towards HER has been discussed. Extrinsic parameters such as layer type, choice of subtract, temperature, doping, and how they affect the electrocatalytic and charge transport properties of these materials have been highlighted. Similarly, the development in the manipulation of 2D TMDs heterostructures to enhance charge transport properties towards HER has been highlighted. However, electrochemical stability of these materials in HER conditions remains a major challenge. Hopefully, continued progress in research will eventually lead to the synthesis of efficient 2D TMDs for improved HER applications.

References 1. Pitchaimuthu S, Marappan S, Kharton V (2019) Materials for energy technologies: recent developments and trends. Mater Lett 253:195 2. Mahatha S, Patel K, Menon K (2012) Electronic structure investigation of MoS2 and MoSe2 using angle-resolved photoemission spectroscopy and ab initio band structure studies. J Phys Conden Matter Instit Phys J 24:475504. https://doi.org/10.1088/0953-8984/24/47/475504 3. Schmidt H, Giustiniano F, Eda G (2015) Electronic transport properties of transition metal dichalcogenide field-effect devices: surface and interface effects. Chem Soc Rev 44. https:// doi.org/10.1039/c5cs00275c 4. Cheng R, Jiang S, Chen Y et al (2014) Few-layer molybdenum disulfide transistors and circuits for high-speed flexible electronics. Nat Commun 5:5143. https://doi.org/10.1038/ncomms6143 5. Fuhrer M, Hone J (2013) Measurement of mobility in dual-gated MoS2 transistors. Nat Nanotechnol 8:146–147. https://doi.org/10.1038/nnano.2013.30 6. Yin Z, Li H, Li H, Jiang L, Shi Y, Sun Y, Lu G, Zhang Q, Chen X, Zhang H (2012) Single-layer MoS2 phototransistors. ACS Nano 6:74 7. Choi W, Choudhary N, Han G, Park J, Akinwande D, Lee YH (2017) Recent development of two-dimensional transition metal dichalcogenides and their applications. Mater Today 20. https://doi.org/10.1016/j.mattod.2016.10.002 8. McDonnell S, Addou R, Buie C, Wallace RM, Hinkle CL (2014) Defect-dominated doping and contact resistance in MoS2 . ACS Nano 8:2880 9. Wang QH, Kalantar-Zadeh K, Kis A, Coleman JN, Strano MS (2012) Electronics and optoelectronics of two-dimensional transition metal dichalcogenides. Nat Nanotechnol 7:699 10. Li S-L, Wakabayashi K, Xu Y, Nakaharai S, Komatsu K, Li W-W, Lin Y-F, Aparecido-Ferreira A, Tsukagoshi K (2013) Thickness-dependent interfacial coulomb scattering in atomically thin field-effect transistors. Nano Lett 13:3546 11. Najmaei S, Liu Z, Zhou W, Zou X, Shi G, Lei S, Yakobson B, Idrobo J-C, Ajayan P, Lou J (2013) Vapour phase growth and grain boundary structure of molybdenum disulphide atomic layers. Nat Mater 12. https://doi.org/10.1038/nmat3673 12. Das S, Chen HY, Penumatcha AV, Appenzeller J (2013) High performance multilayer MoS2 transistors with scandium contacts. Nano Lett 13(1):100–105. https://doi.org/10.1021/nl3 03583v

On Recent Development in Two-Dimensional Transition Metal …

447

13. Bao W, Cai X, Kim D, Sridhara K, Fuhrer MS (2013) High mobility ambipolar MoS2 field-effect transistors: substrate and dielectric effects. Appl Phys Lett 102:042104 14. Withers F, Bointon TH, Hudson DC, Craciun MF, Russo S (2014) Electron transport of WS2 transistors in a hexagonal boron nitride dielectric environment. Sci Rep 4:4967 15. Xu K, Xu Y, Zhang H, Peng B, Shao H, Ni G, Li J, Yao M, Lu H-L, Zhu H, Soukoulis C (2018) Role of anderson rule in determining electronic, optical and transport properties of transition-metal dichalcogenides heterostructure. Phys Chem Chem Phys 20. https://doi.org/ 10.1039/C8CP05522J 16. An Y, Zhang M, Wu D, Fu Z, Wang K (2016) The electronic transport properties of transitionmetal dichalcogenide lateral heterojunctions. J Mater Chem C 4. https://doi.org/10.1039/C6T C04327E 17. Devadoss A, Nagarajan S, Devarajan VP, Grace A, Sudhagar P (2020) Electrocatalytic properties of two-dimensional transition metal dichalcogenides and their hetrostructures in energy applications. https://doi.org/10.1016/B978-0-12-817678-8.00008-7 18. Kim J, Byun S, Smith A, Yu J, Huang J (2013) Enhanced electrocatalytic properties of transitionmetal dichalcogenides sheets by spontaneous gold nanoparticle decoration. J Phys Chem Lett 4:1227–1232. https://doi.org/10.1021/jz400507t 19. Martinez L, Delgado J, Saiz C, Cosio A, Singamaneni S (2018) Magnetic and electrocatalytic properties of transition metal doped MoS2 nanocrystals. J Appl Phys 124(15):153903 20. Chen Z, Cummins D, Reinecke BN, Clark E, Sunkara MK, Jaramillo TF (2011) Coreshell MoO3 -MoS2 nanowires for hydrogen evolution: a functional design for electrocatalytic materials. Nano Lett 11(10):4168–4175. https://doi.org/10.1021/nl2020476 21. Yuan J, Wu J, Hardy WJ, Loya P, Lou M, Yang Y, Najmaei S, Jiang M, Qin F, Keyshar K, Ji H, Gao W, Bao J, Kono J, Natelson D, Ajayan PM, Lou J (2015) Adv Mater 27:5605 22. Qian X, Liu J, Fu L, Li J (2014) Solid state theory. Quantum spin Hall effect in two-dimensional transition metal dichalcogenides. Science 346(6215):1344–1347. https://doi.org/10.1126/sci ence.1256815 23. Guo Y, Sun D, Ouyang B, Raja A, Song J, Heinz TF, Brus LE (2015) Probing the dynamics of the metallic-to-semiconducting structural phase transformation in MoS2 crystals. Nano Lett 15(8):5081–5088. https://doi.org/10.1021/acs.nanolett.5b01196 24. Voiry D, Yamaguchi H, Li J, Silva R, Alves DC, Fujita T, Chen M, Asefa T, Shenoy VB, Eda G, Chhowalla M (2013) Enhanced catalytic activity in strained chemically exfoliated WS2 nanosheets for hydrogen evolution. Nat Mater 12(9):850–855. https://doi.org/10.1038/ nmat3700

Photoabsorbers with Hybrid Organic–Inorganic Structures for Optoelectronics and Solar Cells Mohin Sharma, Mritunjaya Parashar, and Anupama B. Kaul

Abstract Hybrid organic–inorganic structures are explored here for their use in optoelectronics and toward solar cell platforms. In the first study, graphene quantum dots (GQDs) are studied once they are integrated with inorganic twodimensional (2D) MoS2 . This hybrid structure displays exciting optoelectronics properties, exceeding the limitations of the bare MoS2 photodetectors, where the GQDs enhance optical absorption properties into the near-UV regime. Raman and photoluminescence (PL) spectroscopy analysis, along with atomic force microscopy, were used to characterize the GQDs and MoS2 films. In the second study, hybrid structures comprising of organo-halide perovskites are analysed, where similar such optoelectronic properties were examined. Here, solution-processed 2D (CH3 (CH2 )3 NH3 )2 (CH3 NH3 )n−1 Pbn I3n+1 (n = 4) organo-halide layered perovskites were inspected in terms of their photoluminescence behaviour and optical absorption spectra, and the photo response of spin-coated films was studied as a function of incoming laser wavelength of the optical excitation toward future solar cell platforms. Keywords GQDs · Graphene · MoS2 · Photo response · 2D perovskite · Photodetectors

Introduction In recent years, perovskite solar cells (PSCs) based on organic–inorganic halide perovskites have received a lot of attention [1]. According to the National Renewable Energy Laboratory’s (NREL) efficiency chart, since the first report of a perovskite material used in solar cells in 2009, the power conversion efficiencies (PCEs) of M. Sharma · M. Parashar · A. B. Kaul Department of Electrical Engineering, PACCAR Technology Institute, University of North Texas, Denton, TX 76203, USA A. B. Kaul (B) Department of Materials Science and Engineering, University of North Texas, Denton, TX 76203, USA e-mail: [email protected] © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_41

448

Photoabsorbers with Hybrid Organic–inorganic Structures …

449

PSCs have now reached a certified value higher than those of solar cells based on multi-crystalline Si, cadmium telluride, or copper indium gallium diselenide [2]. Organic–inorganic halide perovskites have numerous benefits over traditional semiconductors for photovoltaics, such as long carrier lifetimes, high light absorption, ease of processing, and low fabrication cost [3]. To synthesize these materials, heavy organic cations are inserted into the three-dimensional (3D) arrangement. Because of its hydrophobic nature and van der Waals binding, the so-called two-dimensional (2D) perovskites are more stable than their 3D counterparts. A 2D perovskite has the following configuration as a rule: An-1 Bn X3n+1 , where n is the thickness of the inorganic layers (n = 1, 2, 3, 4, −∞) [4]. A solution method selectively modifies many 2D materials, including monoelemental black phosphorus (BP), binary MoS2 , and the multi-component 2D perovskites. Many of the 2D materials exhibit high carrier mobilities, ultrathin thicknesses, and smooth surfaces free of dangling bonds [5–9]. Physisorption and chemisorption are used in two methodologies for forming the hybrid architectures, which are enabled by synthesis processes such as thermal evaporation and chemical vapor routes, spin casting, as well as immersion operations [10, 11]. In this work, we focus on two hybrid structures, the first being graphene quantum dots (GQDs) integrated with inorganic MoS2, while the second structure is based on 2D organohalide perovskites to explore their light-matter interactions towards optoelectronics and solar cell platforms. Hybrid devices constructed from MoS2 when combined with other lowdimensionality materials, in this case zero-dimensional (0D) graphene quantum dots (GQDs) are expected to have high quantum efficiency because the built-in field working in ultrathin heterostructures across vast contact surfaces enables photogenerated electron–hole pairs to be separated more quickly and efficiently [12]. In addition to their unique optoelectronic features, GQDs have substantial transition energies and weak coupling to excitonic states below the fundamental band gap, which allows them to have extended carrier lifetimes and effective carrier separation and collection. The GQDs demonstrate semiconducting characteristics with donor-like or acceptor-like behaviour depending on the synthesis method used. This confinement may lead to quantum emission with Coulomb-correlated electron–hole pairs in GQDs where the excitons are trapped in quantum dots. In addition, GQDs appear to interact directly with 2D materials to facilitate charge transfer, carrier separation, and collection, where quantum effects are critical for modulating charge carrier dynamics [13]. In the first study reported here, we describe our structures of hybrid GQD/MoS2 photodetectors made with drop-cast GQDs on MoS2 membranes, with Au/Ti as the top contact metal. With broadband white light illumination and a tuneable laser source ranging from ultraviolet (UV) to infrared (IR), the photo response of the hybrid GQD/MoS2 photodetectors was studied at room temperature. In the second study, we have explored the optical properties of solution-processed RuddlesdenPopper 2D perovskites in the context of optoelectronics and solar cells as effective photoabsorbers.

450

M. Sharma et al.

Experimental Synthesis and Fabrication of GQD/MoS2 Photodetector Devices In the first study, bulk MoS2 flakes were transferred to p-doped Si substrates with 300 nm thick thermally grown SiO2 . Conventional optical photolithography (Karl Suss MJB-3 mask aligner) was used to pattern the electrodes on the MoS2 membranes. After e-beam evaporating the electrodes (∼100 nm Au/10 nm Ti) using lift-off in acetone, the surface was further cleaned using a stripper (AZ Kwik strip remover). To fabricate the GQD/MoS2 hybrid structure, the GQD solution was drop-cast on the MoS2 surface and then annealed at 150 °C for 2 h. The hybrid GQD/MoS2 photodetectors were fabricated by drop-casting the GQD solution onto bulk mechanically exfoliated MoS2 membranes on SiO2 /Si substrates, as shown in Fig. 1. The GQDs were synthesized by sonicating coal in concentrated sulfuric and nitric acid, followed by heat treatment at 120 °C for 24 h using an approach that has been reported previously [14, 15]. The solution was cooled to room temperature and poured into a beaker containing an ice—water bath, followed by the addition of NaOH until the pH reaches 7. The neutral mixture was then filtered through a 0.45 μm polytetrafluoroethylene membrane, and the filtrate was dialyzed in dialysis bags against deionized water for 5 days. After purification, the solution was concentrated using a rotary evaporator. Characterization was conducted using Raman and photoluminescence (PL) spectroscopy. Specifically, the Horiba LabRAM HR Evolution microscope was used for the Raman/PL with a ∼532 nm laser for incoming optical excitation. The atomic force microscopy (AFM) measurements were performed using the HORIBA Scientific System (AIST-NT). For the fabrication process, after mechanically exfoliating Fig. 1 Schematic diagram of the hybrid GQD/MoS2 device. The crystal structure representation of the GQDs (top view) and MoS2 (side view) i shown in the bottom left and right insets, respectively

Photoabsorbers with Hybrid Organic–inorganic Structures …

451

MoS2, Au (∼100 nm) and Ti (∼10 nm) electrodes were deposited using e-beam evaporation to define the source (S) and drain (D) electrodes by lift-off.

Synthesis of 2D Organo-Halide Perovskites In the second study, we synthesized the (CH3 (CH2 )3 NH3 )2 (CH3 NH3 )n-1 Pbn I3n+1 (n = 4) perovskite photoabsorbers, which are a family of layered compounds with tunable semiconducting properties. The PbO (part #: 402982), HI (part #: 210021), H3 PO2 (part #: 214906), CH3 NH3 Cl (part #: 8060200250), and n-CH3 (CH2 )3 NH2 (part #: 471305) were purchased from Sigma-Aldrich. The PbO powder was dissolved in a mixture of 57% w/w aqueous HI solution and 50% aqueous H3 PO2 by heating and subsequent boiling under constant magnetic stirring for about 5 min, that led to the formation of a bright yellow solution. Subsequent addition of solid CH3 NH3 Cl to the hot yellow solution initially caused the precipitation of a black powder that rapidly re-dissolved under stirring to afford a clear bright-yellow solution. Additionally, the n-CH3 (CH2 )3 NH3 I was then added dropwise under rigorous stirring over a period of 5 min, which did not result in any visible changes to the solution. Stirring was then stopped, and the solution was left to cool to room temperature during which time, deep red, rectangular-shaped platelets started to crystallize. The precipitation was deemed to be complete after ∼2 h. The crystals were isolated by suction filtration and thoroughly dried under reduced pressure [16, 17]. Afterwards, perovskite inks were formed using a magnetically stirred solvent engineering process. The 0.118 M solution of (CH3 (CH2 )3 NH3 )2 (CH3 NH3 )n-1 Pbn I3n+1 (n = 4) was made in dimethylformamide (DMF). The obtained solution was then spin coated by using the hot-casting method. The SiO2 /Si was preheated to 100 °C before coating and then was instantly transferred to a spin coater and then spin coated at 5000 rpm for 20 s, where the overlying perovskite film turned brown during spinning [18].

Results and Discussion GQD/MoS2 Ensembles for Photodetectors The AFM image in Fig. 2a depicts bulk MoS2 flakes (thickness ∼53.8 nm) mechanically exfoliated on the SiO2 /Si substrate. After fabricating the bare MoS2 devices, the GQD solution (1 mg/ mL in DMF) was drop-cast on the bare MoS2 surface and annealed at 150 °C for 2 h at ambient temperature. The thickness of the hybrid structure of GQDs and MoS2 was determined to be ∼78 nm, as shown by the AFM image in Fig. 2b. The GQDs decorated the MoS2 surface in the form of islands ∼24.2 nm in height.

452

M. Sharma et al.

Fig. 2 AFM image of a fabricated MoS2 device and b GQD/MoS2 on SiO2 /Si substrates

To probe the optical properties of the hybrid GQDs/MoS2 , Raman spectroscopy was used, where the spectra are shown in Fig. 3a, and the results were compared to the bare MoS2 . In the hybrid GQD/MoS2 sample, a blue shift in both the E2g 1 and A1g modes was evident. The Raman shift ω(+) was ∼1.7 cm−1 for the in-plane E2g 1 mode while for the out-of-plane A1g mode, this blue shift ω(+) was ∼0.4 cm−1 , as shown in Fig. 3b, c, respectively. The blue shift for the A1g and E2g1 mode is an artifact of the strong electron−phonon coupling, which provides direct evidence for the n-type doping of MoS2 by the GQDs. The PL spectra of the GQDs, the bare MoS2 film, and the hybrid GQDs/MoS2 after annealing are shown in Fig. 4a–c. Interestingly, the PL emission peak of the GQDs in Fig. 4a occurs over a broad range, centred around ∼433 nm in the near-UV region. The emission peak for the bare MoS2 occurs at ∼690.7 nm, as shown in Fig. 4b, and this peak red shifts to ∼693.1 nm when the GQDs adsorb on the MoS2 surface, depicted by the spectrum in Fig. 4c, with slight peak broadening and a reduced PL signal intensity. A red shift of ∼2.4 nm in the PL peak for the GQDs/MoS2 arises from the Stokes shift, and its magnitude is also a signature of n-type doping in MoS2 . The photo response of the GQD/MoS2 hybrids was then studied with optical illumination. The electronic transport behavior of the hybrid GQD/MoS2 devices was measured and compared to the bare MoS2 -contacted devices. In Fig. 5a, the Ids − Vds characteristic is exhibited for the same MoS2 device before (black circles) and after (red circles) GQDs were deposited on the MoS2 membranes. The measurements were conducted in both the dark (open circles) and under illumination (filled circles) with a

Fig. 3 a Raman spectra of the hybrid GQDs/MoS2 (red) and the bare MoS2 (black) on the SiO2 /Si substrate. Expanded Raman spectra for b E2g 1 and c A1g modes (Color figure online)

Photoabsorbers with Hybrid Organic–inorganic Structures …

453

Fig. 4 a The PL data of GQDs showing a peak occurring at ∼433 nm, b the bare MoS2 with a peak occurring at ∼690.7 nm, and c hybrid GQDs/MoS2

Fig. 5 a Ids–Vds plots for the hybrid GQD/MoS2 (red) and bare MoS2 (black) devices in the dark (open circles) and under broadband white light illumination (solid circles). The inset shows the Ids–Vds curves on an expanded scale for both devices for V ranging from − 4.2 to 2.5 V. b The Iph of the hybrid GQD/MoS2 and the bare MoS2 devices for Vds = 5 V as a function of Po (Color figure online)

broadband white light source at an optical power density (or light intensity) Po of ∼3.2 mW/cm2 . The Ids for the hybrid GQD/MoS2 device was ∼20.7 μA at 10 V, which is higher than ∼11.9 μA for the bare MoS2 device. The Ids of the hybrid GQD/MoS2 device exhibits rectifying behaviour under illumination, and no rectification is seen for the bare MoS2 . To evaluate the photo response, the photocurrent Iph was measured, where Iph = Ilight − Idark ; here, Ilight is the current under illumination and Idark is the dark current. The Iph of the hybrid GQD/MoS2 and the bare MoS2 devices are shown in Fig. 5b for Vds = 5 V at increasing P0 from 0.64 to 3.2 mW/cm2 . The Iph of the hybrid GQDs/MoS2 was larger than the bare MoS2 devices as Po increased; for example, when the device is irradiated at ∼3.2 mW/cm2 . The Iph of the hybrid GQD/MoS2 device was ∼11.3 μA, which is 1.6 times higher than the Iph of the bare MoS2 device (∼7.16 μA) at room temperature.

454

M. Sharma et al.

2D Perovskites and Their Optoelectronic Properties Toward Solar Cell Platforms The 2D (BA)2 (MA)3 Pb4 I13 perovskite films can be utilized for optoelectronic devices due to their strong light-matter interactions. Spin coating of the 2D (BA)2 (MA)3 Pb4 I13 perovskite was done to yield a continuous film, in order to then study its morphological and optical properties. Figure 6a, b show the scanning electron microscopy (SEM) images of the synthesized perovskite thin film at low magnification (scale bar = 10 μm) and at higher magnification (scale bar = 1 μm), respectively. The higher magnification image in Fig. 6b reveals the presence of voids and cracks in the as synthesized films which can be optimized further to form denser films. The optical absorbance and PL spectra of the 2D perovskite thin films are illustrated in Fig. 7, where excitonic features are evident in our absorber at ~600 nm, as

Fig. 6 a and b SEM images of the spin coated 2DP films

Fig. 7 a The UV–Vis spectra of the 2D perovskite thin film. b PL spectra of the 2D perovskite thin film having peak at 750.12 nm

Photoabsorbers with Hybrid Organic–inorganic Structures …

455

Fig. 8 The Iph of the spin coated 2D perovskite fim

shown in Fig. 7a. The understanding of excitonic properties is important since the nature of carriers (excitons versus free carriers) may affect the charge transport. For (BA)2 (MA)3 Pb4 I13 (n = 4) compounds, the exciton binding energy is closer to that of MAPbI3 (n → ∞), for which the excitons are expected to be almost ionized at room temperature and charge-carrier transport is expected to be dominated by free carriers. The photoluminescence spectra of our spin coated 2D perovskite thin film is shown in Fig. 7b which reveals an absorption peak centred at ~750.26 nm, consistent with prior reports [17]. In order to conduct the photo response measurements, the device layout was similar to our previous reports [19]. We spin coated the 2DP perovskite on top of SiO2 /Si substrates. Afterwards, silver electrodes were deposited on top of the perovskite film using physical vapor deposition with e-beam evaporation. The photo response measurements were conducted using a tuneable laser source from 400 to 1100 nm. To evaluate the optoelectronic response, the photocurrent (Iph ) was determined using Iph = Ilight − Idark, where Ilight represents measurements under illumination and Idark denotes the currents measured without illumination, in the dark. The Iph data of the spin coated 2D perovskite is shown in Fig. 8, where the peak amplitude occurs at close to 700 nm, in the vicinity of the bandgap for this 2D perovskite composition, consistent with our expectations.

Conclusions The results obtained in this work for the use of 0D-2D hybrid structures and the 2D perovskites indicate that both of these can be combined as effective absorber layers and charge transport layers for better charge extraction towards optoelectronics and solar cell platforms.

456

M. Sharma et al.

Acknowledgements We thank the Office of Naval Research (grant number ONR N00014-20-12597) that enabled us to pursue this work. A.B.K. is also grateful to the support from the PACCAR Technology Institute at UNT and the Endowed Professorship support.

References 1. Kojima A, Teshima K, Shirai Y, Miyasaka T (2009) Organometal halide perovskites as visiblelight sensitizers for photovoltaic cells. J Am Chem Soc 131(17):6050–6051 2. NREL (n.d.) Best-Research-Cell-Efficiencies-Rev210726.Pdf 3. You P, Tang G, Cao J, Shen D, Ng TW, Hawash Z et al (2021) 2D materials for conducting holes from grain boundaries in perovskite solar cells. Light Sci Appl 10(1):23–26 4. Ortiz-Cervantes C, Carmona-Monroy P, Solis-Ibarra D (2019) Two-dimensional halide perovskites in solar cells: 2D or not 2D? Chemsuschem 12(8):1560–1575 5. Liu Z, Lau SP, Yan F (2015) Functionalized graphene and other two-dimensional materials for photovoltaic devices: device design and processing. Chem Soc Rev 44(15):5638–5679 6. Tang G, You P, Tai Q, Yang A, Cao J, Zheng F et al (2019) Solution-phase epitaxial growth of perovskite films on 2D material flakes for high-performance solar cells. Adv Mater 31(24):1–9 7. You P, Tang G, Yan F (2019) Two-dimensional materials in perovskite solar cells. Mater Today Energy 11:128–158 8. Wang B, Iocozzia J, Zhang M, Ye M, Yan S, Jin H et al (2019) The charge carrier dynamics, efficiency and stability of two-dimensional material-based perovskite solar cells. Chem Soc Rev 48(18):4854–4891 9. Agresti A, Pazniak A, Pescetelli S, Di Vito A, Rossi D, Pecchia A et al (2019) Titaniumcarbide MXenes for work function and interface engineering in perovskite solar cells. Nat Mater 18(11):1228–1234 10. Gough JJ, McEvoy N, O’Brien M, Bell AP, McCloskey D, Boland JB et al (2018) Dependence of photocurrent enhancements in quantum dot (QD)-sensitized MoS2 devices on MoS2 film properties. Adv Funct Mater 28(13):1706149 11. Chen C, Qiao H, Lin S, Man Luk C, Liu Y, Xu Z et al (2015) Highly responsive MoS2 photodetectors enhanced by graphene quantum dots. Sci Rep 5:1–9 12. Wang J, Fang H, Wang X, Chen X, Lu W, Hu W (2017) Recent progress on localized field enhanced two-dimensional material photodetectors from ultraviolet—visible to infrared. Small 13(35):1–17 13. Min M, Sakri S, Saenz GA, Kaul AB (2021) Photophysical dynamics in semiconducting graphene quantum dots integrated with 2D MoS2 for optical enhancement in the near UV. ACS Appl Mater Interfaces 13(4):5379–5389 14. Ye R, Xiang C, Lin J, Peng Z, Huang K, Yan Z et al (2013) Coal as an abundant source of graphene quantum dots. Nat Commun 4:1–7 15. Ye R, Peng Z, Metzger A, Lin J, Mann JA, Huang K et al (2015) Bandgap engineering of coal-derived graphene quantum dots. ACS Appl Mater Interfaces 7(12):7041–7048 16. Stoumpos CC, Cao DH, Clark DJ, Young J, Rondinelli JM, Jang JI et al (2016) Ruddlesden-popper hybrid lead iodide perovskite 2D homologous semiconductors. Chem Mater 28(8):2852–2867 17. Cao DH, Stoumpos CC, Farha OK, Hupp JT, Kanatzidis MG (2015) 2D homologous perovskites as light-absorbing materials for solar cell applications. J Am Chem Soc 137(24):7843–7850 18. Tsai H, Nie W, Blancon J-C, Stoumpos CC, Asadpour R, Harutyunyan B et al (2016) Highefficiency two-dimensional Ruddlesden-Popper perovskite solar cells. Nature 536(7616):312– 316 19. Min M, Hossain RF, Adhikari N, Kaul AB (2020) Inkjet-printed organohalide 2D layered perovskites for high-speed photodetectors on flexible polyimide substrates. ACS Appl Mater Interfaces 12(9):10809–10819

Simulating Microstructure Evolution in Ni-YSZ Electrodes of Solid Oxide Cells Under Operating Conditions Yinkai Lei, William Epting, Jerry Mason, Tian-Le Cheng, Harry Abernathy, Gregory Hackett, and Youhai Wen

Abstract A model for simulating microstructure evolution in Ni-YSZ electrodes of solid oxide cells under fuel cell, electrolysis, and reversible mode has been developed by coupling our recently developed phase field model, multi-physics model, and microstructure analyzing tools. The mechanisms of Ni(OH)2 diffusion and Ni-YSZ wettability change have been considered. The model has been used to investigate the effect of temperature, current density, and gas compositions on the degradation of NiYSZ electrode. Both Ni coarsening and redistribution are found to be affected by gas composition and temperature, while current density only affects the Ni redistribution. The results are compared to available experiments. It shows both mechanisms cannot fully explain the Ni redistribution observed experimentally. Keywords Solid oxide cell · Microstructure evolution · Ni-YSZ electrode · Phase field simulation

Introduction The solid oxide cell (SOC) is a promising electrochemical device that converts between the electricity and chemical energy with high efficiency [1–5] and low pollution emissions [2, 5, 6]. However, due to high operating temperatures, the porous electrodes of SOCs are prone to microstructure evolution which in turn leads to performance degradation over time [7, 8]. Ni coarsening and redistribution have Y. Lei (B) · W. Epting · T.-L. Cheng · Y. Wen U.S. DOE National Energy Technology Laboratory, Albany, OR 97321, USA e-mail: [email protected] Y. Lei · W. Epting · T.-L. Cheng NETL Support Contractor, Albany, OR 97321, USA J. Mason · H. Abernathy · G. Hackett U.S. DOE National Energy Technology Laboratory, Morgantown, WV 26507, USA J. Mason · H. Abernathy NETL Support Contractor, Morgantown, WV 26507, USA © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_42

457

458

Y. Lei et al.

been reported in operated Ni-yttria stabilized zirconia (YSZ) electrodes which are commonly used in SOCs [9–15]. The former refers to the growth of average Ni particle size at the expense of small Ni particles. The latter refers to the migration of Ni phases from one region to others. Both types of microstructure evolutions cause the loss of triple phase boundaries (TPBs) and Ni surface area [9, 16], which reduces the active sites for the electrochemical reactions in the electrode. They also lead to increased Ohmic resistance [17, 18] by altering the conducting path in the electrode. Therefore, it is imperative to understand the microstructure evolution in Ni-YSZ electrodes for the purpose of mitigating the performance degradation in SOCs. Both Ni coarsening and redistribution are known to be affected by the operating conditions. For example, the rate of Ni coarsening has been found to be orders of magnitude higher in humid atmosphere than that in dry atmosphere [9]. Ni redistribution has been found in both fuel cell and electrolysis modes, despite reports in the literature being inconsistent about the direction of the redistribution. In fuel cell mode, Ni has been found to redistribute toward the electrolyte layer near open-circuit voltage (OCV) [13] but toward the support layer otherwise [12, 19]. In electrolysis mode, Ni has been found to redistribute toward the electrolyte layer at 950 °C [11] but toward the support layer below 850 °C [14, 15, 20]. These works show that both Ni coarsening and redistribution are affected by the steam level, temperature, and overpotential in the electrode. Despite the existing works on Ni coarsening and redistribution, the mechanism that drives the microstructure evolution in Ni-YSZ electrodes is still under debate. The formation and diffusion of volatile species of Ni, e.g. Ni(OH)2 , has been proposed to explain the fast coarsening under humid atmosphere [9]. However, our recent work shows that it is not able to explain the redistribution observed in experiments due to the low equilibrium concentration of Ni(OH)2 in the pore phase [21]. In addition, the microstructure evolution due to Ni(OH)2 diffusion would lead to Ni being redistributed from the region with high steam partial pressure to that with low steam partial pressure. This is opposite to the Ni redistribution observed in electrolysis mode below 850 °C [14, 15, 20]. Trini et al. [14] proposed that the wettability change between Ni and YSZ [22, 23] might explain the Ni redistribution found in electrolysis mode. However, it is still unclear how the Ni-YSZ wettability change affects the microstructure evolution in the Ni-YSZ electrode. Nevertheless, both mechanisms may contribute to the microstructure evolution in the Ni-YSZ electrode with each contributing differently under different operating conditions. We have previously developed a phase field model for simulating microstructure evolution in the Ni-YSZ electrodes through Ni(OH)2 diffusion [21]. However, the Ni-YSZ wettability change was not considered nor was the steam partial pressure used in our previous work coupled to actual operating conditions. In this work, we extend our model to incorporate both the mechanisms of Ni(OH)2 diffusion and Ni-YSZ wettability change. This model is coupled to our previously developed microstructure analysis tools [24] and multi-physics model for cell performance [25, 26] which allows the simulation of microstructure evolution under different operating conditions. We investigated the effect of operating mode, temperature,

Simulating Microstructure Evolution in Ni-YSZ Electrodes of Solid Oxide Cells …

459

current density, and gas composition on the microstructure evolution. The comparison between the simulation results and existing experiments is discussed.

Methodology The phase field model for Ni(OH)2 diffusion is discussed in detail in Ref. [21]. We extend it to model the Ni-YSZ wettability change by linking the contact angle to local steam partial pressure. The Ni-YSZ contact angle in our phase field model is determined by the boundary condition of φ on the Ni-YSZ interface [21, 27]:   → −κ ∇φ · − n = −6γ φ(1 − φ)cosθ

(1)

→ where − n is the normal vector of the interface, θ is the contact angle between Ni and YSZ, and γ is the surface energy in the model. The contact angle θ can be determined by the work of adhesion Wad and Ni surface energy γ N i by [22] Wad = γ N i (1 + cosθ )

(2)

where the Ni surface energy can be linked to the local oxygen partial pressure, which is in turn determined by the steam partial pressure. The model proposed by Gheribi et al. [28, 29] is used in this work to link Ni surface energy to the oxygen partial pressure. This model takes the saturated oxygen partial pressure on the Ni surface into consideration, which is lacking in the Belton model used in Ref. [22]. Figure 1 shows the Ni-YSZ contact angle as a function of the steam partial pressure assuming a binary gas of steam and hydrogen. Wad is assumed to be 1.4 J/m2 [22, 30]. It shows that the model used in this work predicts a greater contact angle change with increased steam partial pressure. Fig. 1 Ni-YSZ contact angle as a function of steam partial pressure at 1 atm and 800 °C predicted by the model in this work and Ref. [22]. The dashed line shows the condition when Ni and NiO are in equilibrium

460

Y. Lei et al.

The multi-physics model in Refs. [25, 26] is used to obtain the steam partial pressure that is needed by the phase field model. To avoid the time-consuming solution of a full three-dimensional steam partial pressure in the electrode, we make the assumption that the steam partial pressure is uniform in the plane parallel to the electrolyte–electrode interface and solve the steam partial pressure distribution in 1D using the average microstructure properties along the axial direction. The required microstructure properties, namely the TPB density, pore size, and volume fraction of each phase, are evaluated by the microstructure analysis tools in Ref. [24] as a function of the distance from the electrolyte layer. The tortuosity factor of each phase, however, is only well defined for the whole simulation cell. Therefore, we use the tortuosity factor of the whole simulation cell as the uniform tortuosity factor along the axial direction. To couple the phase field model to the multi-physics model, we first solve the steam partial pressure on a given microstructure at the specific operating condition, then evolve the microstructure until the distribution of certain microstructure properties along the axial direction is changed, then the steam partial pressure is solved again on the updated microstructure. To avoid evaluating all microstructure properties at every time step, the volume fractions and TPB density are chosen to determine when to pause the phase field model to obtain an updated steam distribution in this work. We append a support layer with fixed microstructure properties to the active layer in the axial direction. This introduces a concentration change between the gas composition at the gas inlet and active layer, which reflects the concentration overpotential due to the gas diffusion in the support layer. A subvolume of the anode active layer from a button cell originally obtained from a commercial supplier (Materials and System Research Inc., Salt Lake City, UT) [31] is used in this work as the initial microstructure. The method combining Xe plasma focused-ion-beam and scanning electron microscopy is used to reconstruct the microstructure. The spatial resolution is 65 nm and the volume of the subvolume is 8.3 × 8.3 × 8.3 µm3 . The overall composition of the subvolume is 42% YSZ, 39% Ni, and 19% pore. Periodic boundary conditions are used in the directions in the plane of electrolyte–electrode interface. Zero-flux boundary condition is used at the interface between the active layer and the support layer, assuming that the Ni(OH)2 flux in the support layer is negligible compared to that in the active layer. The support layer is 780 µm in thickness with a uniform composition of 50% Ni and 50% Pore. Binary gas of hydrogen and steam with a total pressure of 1 atm is applied in all simulations. Finite difference method with an iterative fully-implicit backward Euler solver is used to solve the phase field model while the multi-physics model is solved by the finite-volume method.

Simulating Microstructure Evolution in Ni-YSZ Electrodes of Solid Oxide Cells …

461

Results Microstructure Evolution in Different Operating Modes We first compare the microstructure evolution under fuel cell, electrolysis, and reversible mode. Figure 2a shows the evolution of overpotential in the simulations in these three operating modes under the conditions of 800 °C, 0.5 A/cm2 , and 50% steam by mole fraction. The current is reversed every 100 h in reversible mode. The overpotential evolution up to 2 kh is plotted for clarity while all simulations run up to 10 kh. It shows that the performance degradation in all three modes is similar: a fast degradation stage in the first 100 h followed by a slower but steady degradation stage. The later degradation stage continues for the duration of the simulation. Figure 2b shows the initial microstructure and the evolved microstructures after 10 kh in these simulations. The coarsening of Ni particles can be seen clearly. However, no visible difference can be seen between the evolved microstructures at 10 kh in all three operating modes. This indicates that the microstructure evolution is not very sensitive to the operating mode, which is consistent with the fact that the overpotential in reversible mode is identical to that in either fuel cell or electrolysis mode in the corresponding time period as shown in Fig. 2a. This is confirmed by the evolutions of Ni particle size and TPB density shown in Fig. 3a, b. The Ni particle size increased from ~0.54 µm to ~0.63 µm, while the TPB density decreased from ~8 µm−2 to ~4 µm−2 in all three operating modes. Further analysis shows that the average steam molar content in the active layer is ~51% in

Fig. 2 a The evolution of the overpotential in fuel cell, electrolysis, and reversible mode under the conditions of 800 °C, 0.5 A/cm2 , and 50%H2 O-50%H2 , and b the initial microstructure and the evolved microstructures after 10 kh in the fuel cell, electrolysis, and reversible mode. The red and green phases are YSZ and Ni, respectively (Color figure online)

462

Y. Lei et al.

Fig. 3 The evolution of a Ni particles size and b TPB density, c the Ni volume fraction distribution along the axial direction at 0 h and 10 kh in fuel cell, electrolysis, and reversible modes, and d the evolution of the linear-fitting slope of the Ni volume fraction distribution in three different operating modes under the condition of 800 °C, 0.5 A/cm2 , and 50% H2 O-50% H2 molar fractions

fuel cell mode and ~49% in electrolysis mode. Our previous work [21] shows that Ni bulk diffusion dominants the Ni coarsening in this steam content range due to the low equilibrium concentration of Ni(OH)2 . In addition, the Ni-YSZ contact angles differed by less than 1° under these two steam partial pressures. Thus it is no surprise that the operating mode does not have a visible effect on the Ni coarsening and TPB degradation. We further analyze the Ni redistribution along the axial direction by plotting the distribution of Ni volume fraction as a function of the distance to the electrolyte layer as shown in Fig. 3c. The change of the Ni volume fraction distribution shows that some big peaks grow at the expense of smaller peaks, which is a typical behavior of coarsening driving Ni redistribution. The Ni volume fraction distributions at 10 kh are very similar in all three operating modes with the only difference occurring near the support layer. We fit the Ni volume fraction to a linear function and use the slope to quantify the Ni redistribution. The evolution of the slope is plotted in Fig. 3d. It shows that Ni migrates toward the support layer (slope increases) in all three modes in the early stage, which is likely due to the local Ni coarsening. Then Ni keeps migrating toward the support layer in fuel cell mode while it changes course and migrates toward the electrolyte layer (slope decreases) in electrolysis mode. This is

Simulating Microstructure Evolution in Ni-YSZ Electrodes of Solid Oxide Cells …

463

consistent with our previous work [21] which shows that Ni(OH)2 diffusion drives Ni away from the electrolyte layer in fuel cell mode and toward the electrolyte layer in electrolysis mode despite not having much effect on Ni coarsening overall. The Ni redistribution in reversible mode lies between that in fuel cell and electrolysis modes. The results in Fig. 3 show that the microstructure evolution in reversible mode can be considered as a superposition of the microstructure evolution in fuel cell and electrolysis modes. Therefore, only fuel cell and electrolysis mode are discussed in the later sections.

Temperature Effect We investigate the temperature effect by running the simulations at 800 °C, 850 °C, and 900 °C while keeping the gas composition of 50% steam and 50% hydrogen by mole fraction, and the current density of 0.5 A/cm2 . The evolution of the Ni particle size and TPB density is given in Fig. 4a, b, respectively. Only the evolutions in fuel cell mode are plotted as the evolutions in electrolysis mode are not discernible from that in fuel cell mode. It shows that both the Ni coarsening and the degradation of TPB density are faster at higher temperature, which is expected as both Ni and Ni(OH)2 diffusions are faster at higher temperature. The evolution of the Ni redistribution is evaluated by plotting the slope of the linear fitting of the Ni volume fraction distribution as a function of time as shown in Fig. 4c. It clearly shows that Ni migrates away from the electrolyte layer in fuel cell mode and toward electrolyte layer in electrolysis mode at all temperatures. The Ni redistribution is more significant at higher temperature, which is due to the faster Ni(OH)2 diffusion under a steady steam gradient at higher temperature.

Current Density Effect The current density effect on Ni coarsening and TPB degradation is plotted in Fig. 5a, b, respectively. The temperature and gas composition are kept at 800 °C and 50% steam-50% hydrogen by molar fraction, respectively, while the current densities of 0.1 A/cm2 , 0.5 A/cm2 , and 1 A/cm2 are applied. Again, only the evolutions in fuel cell mode are plotted as the evolutions in electrolysis mode are nearly equivalent. It shows that current density has no significant effect on Ni coarsening and TPB degradation. The average steam content in the active layer is ~52%, ~51%, and ~50% by molar fraction in fuel cell mode under the current densities of 0.1 A/cm2 , 0.5 A/cm2 , and 1 A/cm2 . Such a small change in the steam partial pressure does not affect the Ni coarsening and TPB degradation that are mostly driven by the Ni bulk diffusion. However, the evolutions of the linear-fitting slope of the Ni volume fraction distribution in Fig. 5c show that the current density does affect the Ni redistribution in the active layer. Ni redistribution is more significant with higher current density. A

464

Y. Lei et al.

Fig. 4 The evolutions of a Ni particle size and b TPB density at 800 °C, 850 °C, and 900 °C in fuel cell mode, and c the evolution of the Ni redistribution measured by the slope of the linear fitting of the Ni volume fraction distribution at different temperatures

detailed check shows that the gradient of the steam partial pressure increased from 4.5 × 10−5 atm/µm to 4.5 × 10−4 atm/µm when the current density changes from 0.1 A/cm2 to 1 A/cm2 . This increased gradient at higher current density enlarges the driving force for Ni redistribution which leads to the faster Ni redistribution.

Gas Composition Effect The effect of gas composition is explored by setting the gas composition at the gas inlet to be 25% steam-75% hydrogen, 50% steam-50% hydrogen, and 75% steam-25% hydrogen, while keep the temperature and current density at 800 °C and 0.5 A/cm2 , respectively. The evolutions of the Ni particle size and TPB density under different gas compositions are shown in Fig. 6a, b, respectively. Similar to Figs. 4 and 5, only the evolutions in fuel cell mode are plotted as the evolutions in electrolysis mode are nearly equivalent. The gas composition shows no significant effect on Ni coarsening, but the degradation of TPB is slightly faster with higher

Simulating Microstructure Evolution in Ni-YSZ Electrodes of Solid Oxide Cells …

465

Fig. 5 The evolution of a Ni particle size and b TPB density at 0.1 A/cm2 , 0.5 A/cm2 , and 1 A/cm2 in fuel cell mode, and c the evolution of the Ni redistribution measured by the slope of the linear fitting of the Ni volume fraction distribution at different current densities

steam partial pressure. This is consistent with our previous work showing that the effect of Ni(OH)2 diffusion on coarsening can only be observed at a relatively high steam partial pressure due to its low equilibrium concentration. The evolutions of the linear-fitting slope of the Ni volume fraction distribution at different gas compositions are given in Fig. 6c. It shows that the Ni redistribution is more significant at higher steam partial pressure. On the one hand this is because the equilibrium Ni(OH)2 concentration is higher at higher steam partial pressure. On the other hand, this is also because the gradient in Ni(OH)2 concentration is greater at higher steam partial pressure even when the gradient of the steam partial pressure is the same. The non-linear relation between the equilibrium Ni(OH)2 concentration and the steam partial pressure means that there is both greater loading and a larger driving force of Ni redistribution through Ni(OH)2 diffusion.

466

Y. Lei et al.

Fig. 6 The evolution of a Ni particle size and b TPB density at 25% steam-75% hydrogen, 50% steam-50% hydrogen, and 75% steam-25% hydrogen in fuel cell mode, and c the evolution of the Ni redistribution measured by the slope of the linear fitting of the Ni volume fraction distribution at different current densities

Discussions Both Ni(OH)2 diffusion and Ni-YSZ wettability change have been proposed to explain the Ni redistribution observed in experiments. It is hypothesized that Ni(OH)2 diffusion drives Ni downward along the steam partial pressure gradient [15], while Ni-YSZ wettability change drives Ni upward along the steam partial pressure gradient as higher steam partial pressure favors the wetting between Ni and YSZ [14]. In our simulations, only the effect of Ni(OH)2 diffusion has been found as Ni redistribution always occurs downward along the steam partial pressure gradient. This suggests that the mechanism of Ni-YSZ wettability change is not comparable to the Ni(OH)2 diffusion under the assumptions of current model. In fact, the contact angle difference is less than 5° across the active layer in all simulations performed, which is not expected to cause a significant change in microstructure evolution. Note that we only considered the change in Ni surface energy under humid atmosphere in the current model for the Ni-YSZ wettability change. Recent works show that Ni-YSZ interface energy may also change under polarization [23]. Once a more sophisticated

Simulating Microstructure Evolution in Ni-YSZ Electrodes of Solid Oxide Cells …

467

Fig. 7 The comparison between the amount of Ni redistribution predicted by this work and available experiments in fuel cell (left) and electrolysis (right) mode

model between the Ni-YSZ contact angle and the operating condition is established, our model will be used to re-evaluate the effect of Ni(OH)2 diffusion and Ni-YSZ wettability change on the microstructure evolution in Ni-YSZ electrode. The comparison between this work and available experiments [13–15, 32] on Ni redistribution has been summarized in Fig. 7. The amount of Ni redistribution is measured by the change in the linear-fitting slope of the Ni volume fraction distribution. Only the simulations at 800 °C are used here as the temperatures in the experiment range from 710 to 800 °C. It shows that the amount of Ni redistribution predicted by the current model with Ni(OH)2 diffusion and Ni-YSZ wettability change are less than most experiments except for the fuel cell experiment in Ref. [15]. In addition, it shows that the available experiments on Ni redistribution are not consistent, especially in fuel cell mode. Further investigation combining the phase field simulations and well-controlled experiments are needed to understand the Ni redistribution in the Ni-YSZ electrodes.

Conclusions We developed a model that is able to simulate the microstructure evolution in Ni-YSZ electrodes under different operating conditions by coupling our recently developed phase field model, multi-physics model, and microstructure analyzing tools. Both Ni(OH)2 diffusion and Ni-YSZ wettability change are considered in the model. The model has been used to investigate the effect of operating mode, temperature, current density, and gas composition on Ni coarsening and Ni redistribution. It is found that only temperature and gas composition affects the Ni coarsening while all four factors affect the Ni redistribution. Our model predicts that Ni redistribution always occurs downward along the steam partial pressure gradient, which indicates that Ni-YSZ wettability change is not able to compete with Ni(OH)2 diffusion under the assumptions of current model. The amount of Ni redistribution predicted by the current model is less than which

468

Y. Lei et al.

has been observed experimentally, which indicates that currently proposed mechanisms are not adequate for explaining these observations. Further studies combining simulations and well-controlled experiments are needed to provide insights into the microstructure evolution in Ni-YSZ electrodes under various operating conditions. Disclaimer This work was funded by the Department of Energy, National Energy Technology Laboratory, an agency of the United States Government, through a support contract with Leidos Research Support Team (LRST). Neither the United States Government nor any agency thereof, nor any of their employees, nor LRST, nor any of their employees, makes any warranty, expressed or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise, does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof. Acknowledgements This work was performed in support of the U.S. Department of Energy’s Fossil Energy Crosscutting Technology Research Program. The Research was executed through the NETL Research and Innovation Center’s Solid Oxide Fuel Cell Research.

References 1. Jensen SH, Graves C, Morgensen M, Wendel C, Braun R, Hughes G, Gao Z, Barnett SA (2015) Energy Environ Sci 8:2471–2479 2. Jiang SP (2016) Asia-Pac J Chem Eng 11:386–391 3. Ding D, Li X, Lai SY, Gerdes K, Liu M (2014) Energy Environ Sci 7:552–575 4. Lee TS, Chung J, Chen Y-C (2011) Energy Convers Manag 52(10):3214–3226 5. Singhal SC (2014) Solid oxide fuel cells for power generation, Wiley Inter-disciplinary Reviews. Energy Environ 3(2):179–194 6. Choudhury A, Chandra H, Arora A (2013) Renew Sustain Energy Rev 20:430–442 7. Knibbe R, Hauch A, Hjelm J, Ebbesen SD, Morgensen M (2011) Green 1:141–169 8. Mocoteguy P, Brisse A (2013) Intl J Hyd. Energ 38:15887–15902 9. Holzer L, Iwanschitz B, Hocker T, Mnch B, Prestat M, Wiedenmann D, Vogt U, Holtappels P, Sfeir J, Mai A, Graule T (2011) J Power Sour 196(3):1279–1294 10. De Angelis S, Jorgensen PS, Tsai EHR, Holler M, Kreka K, Bowen JR (2018) J Power Sour 383:72–79 11. Hauch A, Ebbesen SD, Jensen SH, Morgensen M (2008) J Electrochem Soc 155:B1184–B1193 12. Zekri A, Herbrig K, Knipper M, Parisi J, Plaggenborg T (2017) Fuel Cells 17:359–366 13. Parikh H (2015) PhD Dissertation Case Western University 14. Trini M, Hauch A, De Angelis S, Tong X, Vang Hendriksen P, Chen M (2020) J Power Sour 450:227599 15. Monaco F, Hubert M, Vulliet J, Ouweltjes JP, Montinaro D, Cloetens P, Piccardo P, LefebvreJoud F, Laurencin J (2019) J Electrochem Soc 166:F1229–F1242

Simulating Microstructure Evolution in Ni-YSZ Electrodes of Solid Oxide Cells … 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32.

469

Yokokawa H, Tu H, Iwanschitz B, Mai A (2008) J Power Sour 182:400–412 Simwonis D, Tietz F, Stover D (2000) Solid State Ion 132:241–251 Pihlatie M, Kaiser A, Mogensen M (2012) Solid State Ion 222–223:38–46 Sciazko A, Miyahara K, Komatsu Y (2019) Takaaki Shimura, Zhenjun Jiao and Naoki Shikazono. J Electrochem Soc 166:F44–F52 Trini M, Jorgensen PS, Hauch A, Bentzen JJ, Hendriksen PV, Chen M (2019) J Electrochem Soc 166:F158–F167 Y. Lei, T.-L. Cheng, H. Abernathy, T. Kalapos, G. Hackett, Y. Wen, Journal of Power Sources, 482, 2021, 228971. Jiao Z, Shikazono N (2017) Acta Mater 135:124–131 Jiao Z, Busso EP, Shikazono N (2020) J Electrochem Soc 167:024516 Epting WK, Mansley Z, Menasche DB, Kenesei P, Suter RM, Gerdes K, Litster S, Salvador PA (2017) 100:2232–2242 Mason J, Celik I, Lee S, Abernathy H, Hackett G (2019) J Electrochem Soc 165:F65–F74 Yang T, Sezer H, Celik I, Finklea H, Gerdes K (2017) Int J Electrochem Sci 12:6801–6828 J. A. Warren, T. Pusztai, L. Kornyei, and L. Granasy, Physical Review B, 79, 2009, 014204 Gheribi AE, des Roches MV, Chartrand P (2019) J Non-Cryst Solids 505:154–161 Gheribi AE, Chartrand P (2019) Sci Rep 9:7113 Tsoga A, Naoumidis A, Nikolopoulos P (1996) Acta Mater 44:3679–3692 Mahbub R, Epting WK, Hsu T, Mason JH, Feng M, Nuhfer MT, Abernathy HW, Hackett GA, Litster S, Rollet AD, Solvardor PA (2020) J Electrochem Soc 167:054506 Geng J, Jiao Z, Yan D, Jia L, Pu J, Li J (2021) J Power Sour 495:229792

Study on a Short Process Method for Preparation of 3.5 Valence Vanadium Electrolyte Zhengtuan Li, Chunjing Wu, Heli Wan, and Lanjie Li

Abstract This study reports on the preparation of high-purity electrolyte containing vanadium of 3.5+ . It can be realized in a short process to prepare the electrolyte of the all-vanadium redox flow battery through the optimization of different parameters. The results show that a sample with stable reduction valence (V: 3.5+ ) can be obtained when vanadium pentoxide is used as the raw material, the reducing gas volume ratio is N2 : CO: H2 = 80: 17: 3, and the reduction temperature and time are controlled at 700 °C and 2 h, respectively. In addition, the concentration of sulfuric acid 4 M is selected as the reaction solution, and the vanadium ion concentration of the electrolyte solution is between 1.5 and 1.6. The single cell made of vanadium electrolyte can run stably for 500 h, charge and discharge 100 times, and basically maintain energy efficiency at around 82%. It has very good industrial application prospects. Keywords Vanadium pentoxide · Gas–solid reduction · 3.5-valent vanadium · Vanadium electrolyte

Introduction Vanadium battery is a redox battery energy storage system based on metallic vanadium. It has a series of unique advantages such as high power, large capacity, high efficiency, low cost, long life, and environmental protection. It is very suitable for largescale static energy storage [1–3]. Vanadium battery (VRB) is the world’s largest, most technologically advanced, and highly efficient reversible fuel cell closest to industrialization [4–6]. It has a good application prospect in photovoltaic power generation, wind power generation, power grid peak shaving, communication base stations, electric buses, military power storage, and other broad fields [7–10]. Z. Li · C. Wu School of Materials Science and Engineering, University of Science and Technology Beijing, Beijing 100083, China Z. Li · H. Wan · L. Li (B) Chengde Iron and Steel Group Co., Ltd., HBIS Group Co., Ltd., Chengde 067102, Hebei, China e-mail: [email protected] © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_43

470

Study on a Short Process Method for Preparation of 3.5 Valence … Table 1 Raw materials and composition of the experiment

471

Raw material

Purity

V2 O5

≥99.9%

H2 SO4

≥98%

H2 C2 O4

≥98%  >15 M .cm

Deionized water

The electrolyte is an important part of the vanadium battery, which directly affects the cost and performance of the vanadium battery. This article reports that high-purity ammonium polyvanadate is used as the starting material to prepare high-purity 3.5+ vanadium electrolyte through gas–solid phase controlled reduction. This method has a low-cost, low-carbon, and high-efficiency preparation method.

Experimental The all-vanadium redox flow battery (VRFB, hereinafter referred to as “vanadium battery”) has not been used in large-scale industrial applications. This is mainly due to the long preparation process of the energy storage medium electrolyte and the difficulty of completely removing the reductant residue, resulting in production high cost. In this study, the intermediate product (high-purity ammonium polyvanadate) prepared by “targeted impurity removal-structural control” is used as the initial raw material, and the low-cost, low-carbon, and efficient preparation. The raw materials and ingredients used in the experiment are shown in Table 1. The comparison between this method and the traditional vanadium electrolyte preparation method is shown in Fig. 1, and its advantages such as the carbon emissions are decreased by 30%; the preparation process is reduced; in the assisted reduction process of ammonium, the valence state of tin is easily controlled at 4.

Results and Discussion Research on the Preparation Mechanism The high-efficiency preparation of high-purity 3.5+ vanadium oxide compounds through gas–solid controllable reduction is the key to the preparation of all-vanadium redox flow battery electrolytes with low cost and short process. In the reduction process of ammonium polyvanadate, different reducing agents are selected for the reaction, and the reaction formula is shown in formulas (1–3). (NH4 )2 V6 O16 = 6VO2 + N2 + 4H2 O

(1)

472

Z. Li et al.

Fig. 1 Comparison diagram between this method and the traditional vanadium electrolyte preparation method

(NH4 )2 V6 O16 + 3CO = 3V2 O3 + 3CO2 + N2 + 4H2 O

(2)

(NH4 )2 V6 O16 + 3H2 = 3V2 O3 + N2 + 7H2 O

(3)

The equilibrium partial pressure of the gas–solid phase thermodynamic calculation is shown in Fig. 2. This shows that the ammonium vanadate deamination reaction to decompose ammonia can undergo a self-reduction reaction to generate VO2 , the equilibrium partial pressure of the reaction is extremely low, and the reaction can be complete. However, the conversion of ammonium polyvanadate into V2 O3 equilibrium partial pressure CO presents an exponential upward trend, and H2 tends to be flat. By adjusting the partial pressure of each atmosphere of the ternary reduction system, the precise preparation of 3.5+ vanadium oxide can be achieved.

Research on Gas–Solid Phase Reduction Process By adjusting the partial pressure of each atmosphere of the N2 , CO, and H2 ternary reduction system, and comparing the valence state of the final reduction of vanadium oxide, the law is shown in Fig. 3. The results show that the final valence state of

Study on a Short Process Method for Preparation of 3.5 Valence …

473

Fig. 2 Comparison of the equilibrium partial pressure of vanadium oxide reduction at different temperatures Fig. 3 The influence of different reducing agent partial pressures on the valence state of vanadium oxide compounds

474

Z. Li et al.

vanadium oxide compound fails to reach 3.5+ under the condition of self-reduction and the ratio of reducing gas is N2 : CO: H2 = 80:20:0, which does not meet the technical requirements. When the reducing gas ratio is N2 : CO: H2 = 80:15:5 and N2 : CO: H2 = 80:10:10, the vanadium in the product at 800–850 °C is 3.5+ vanadium, but it is difficult to accurately control. When the reducing gas ratio is N2 : CO: H2 = 80:17:3, and the reaction temperature is 850–1000 °C, the valence state of vanadium oxide is between 3.55 and 3.4. In the later stage, by adding a small amount of pentavalent vanadium oxide compound, the precise preparation of 3.5+ vanadium oxide can be realized. When the fixed reducing gas ratio is determined to be N2 : CO: H2 = 80:17:3, the effects of reduction temperature and reduction time on vanadium oxide compounds are studied, and the results are shown in Fig. 4. This shows that as the reduction temperature and time increase, the reduction valence gradually decreases. When the temperature is lower than 900 °C, the expected reduced valence state cannot be reached. Therefore, a reduction temperature of 900 °C and reduction of 2 h were selected as the optimal reaction conditions.

Fig. 4 The effect of reduction temperature and reduction time on vanadium oxide compounds

Fig. 5 The stability of V4+ -sulfuric acid solution

Study on a Short Process Method for Preparation of 3.5 Valence …

475

Fig. 6 The stability of V5+ -sulfuric acid solution

Fig. 7 The stability of V3+ -sulfuric acid solution

Design of Valence Regulation and Concentration Matching It is reported that the concentration of sulfuric acid and the temperature of the solution are very important for electrochemical performance[11–15]. In this study, in order to obtain the expected concentration of electrolyte, the addition of different solutes is further optimized. First, prepare V(IV) solutions with different vanadium concentrations and H2 SO4 concentrations, and then prepare corresponding V(III) and V(IV) electrolytes through equal volume electrolysis. The samples were placed in different temperature environments (20, 40, 0 and −10 °C), and the stability of different vanadium electrolytes was investigated, the results are shown in Figs. 5, 6 and 7. It was found that when the vanadium concentration was 1.55 M and the H2 SO4 concentration was below 4.5 M, precipitation of pentavalent vanadium occurred when placed at 40 °C. This shows that the lower the concentration of H2 SO4 , the easier it is for V(V) to be precipitated. When the concentration of H2 SO4 increased to 6 M, trivalent vanadium V(III) began to precipitate after being placed at 40 °C for 21 days. When the H2 SO4 concentration is controlled above 5 M, the tetravalent vanadium V(IV) begins to precipitate after being placed at −10 °C for 28 days. It shows that

476

Z. Li et al.

at this concentration of vanadium, when the concentration of H2 SO4 is controlled at 4.5–6.0, the temperature must be controlled above −10 °C in the electrolyte to maintain stability. When the vanadium concentration is 1.6 M, when the sulfuric acid concentration is 6 M, pentavalent vanadium begins to precipitate after being placed for 25 days. When the vanadium concentration is 1.7 M and the sulfuric acid concentration is 6 M, tetravalent vanadium is easy to be precipitated. Therefore, when the vanadium solution concentration is as high as 1.5 M, the sulfuric acid concentration is about 5 M, and the vanadium solution is more stable when the temperature is above −10 °C. In summary, to ensure the stability of the vanadium electrolyte, both positive and negative electrolytes must be considered. In order to suppress the formation of V2 O5 precipitation from pentavalent vanadium in the positive electrode, the concentration of H2 SO4 needs to be increased. However, in order to reduce the precipitation of hydrated sulfate produced by the trivalent vanadium in the negative electrode, the concentration of H2 SO4 should not be too high. This also further confirms that the matching relationship between the vanadium concentration and the medium H2 SO4 concentration is very important for the stability of the vanadium electrolyte. In order to avoid the sensitivity of vanadium electrolyte to impurities, reduce the cost of vanadium batteries, and improve their practicability, it is necessary to appropriately increase the concentration of H2 SO4 to ensure the stability of the vanadium battery cathode solution. At the same time, the vanadium concentration should not be too high to ensure the stability of the vanadium battery anode solution. The above results indicate that in order to ensure the stability of the vanadium electrolyte, while pursuing high purity of the vanadium electrolyte, the concentration of vanadium ions and sulfuric acid should not be too high. Therefore, when the vanadium ion concentration is selected between 1.5 and 1.6, the H2 SO4 concentration is about 5 M, and the temperature cannot be lower than −10 °C, and a more ideal vanadium electrolyte can be obtained. Vanadium electrolytes of different concentrations were used to equip vanadium single cells to investigate the stability of their charge and discharge cycles. After charging and discharging the battery, perform powder X-ray diffraction (XRD) test on the precipitates appearing on the positive/negative electrodes, and the results are shown in Fig. 8. It can be seen that the XRD spectrum of the negative electrode precipitate is basically consistent with the standard spectrum of V2 (SO4 )3 •10H2 O. The XRD spectrum of the positive electrode precipitate is basically consistent with the standard spectrum of V2 O5 •1.6H2 O. The precipitation of pentavalent vanadium in the positive electrode electrolyte is suppressed by increasing the concentration of H+ , however, appropriately reducing the sulfuric acid concentration can stabilize the effect of the negative electrode trivalent vanadium electrolyte. When the concentration of H2 SO4 is increased to 4 M, the vanadium cell has been running stably for nearly 500 h, charging and discharging cycles 100 times, and the energy efficiency is basically maintained at around 82%, as shown in Fig. 9.

Study on a Short Process Method for Preparation of 3.5 Valence …

477

Fig. 8 XRD spectra of positive/negative electrode deposits during charging and discharging. a The deposit of the positive electrode, b The deposit of the negative electrode

Fig. 9 The charge–discharge cycleability of the prepared vanadium electrolyte (1.55 M V(IV)+ , 4 M H2 SO4 )

Conclusions The conclusions of this article are as follows: (1)

(2)

This research has developed a new low-carbon preparation technology of vanadium redox flow battery electrolyte with gas–solid controllable reduction as the core. The partial pressure ratio of each atmosphere of the reduction system is finally determined to be N2 : CO: H2 = 80:17:3. At the same time, the preferred temperature is 900–1000 °C; when the reaction time is 2 h, the precise preparation of 3.5-valent vanadium oxide is achieved. Increasing the concentration of H2 SO4 can inhibit the formation of V2 O5 from the precipitation of vanadium in the positive electrode. However, the excessive concentration of H2 SO4 leads to the precipitation of hydrated sulfate in the negative electrode of trivalent vanadium. Therefore, the best electrolyte composition is obtained when the vanadium ion concentration is controlled

478

(3)

Z. Li et al.

between 1.5 and 1.6, the H2 SO4 concentration is controlled at about 5 M, and the temperature is controlled at −10 °C to 40 °C. In addition, the results also show that the precipitation of pentavalent vanadium in the cathode electrolyte is suppressed by increasing the H+ concentration, however, the trivalent vanadium of the negative electrode is stabilized by appropriately reducing the sulfuric acid concentration. This phenomenon shows that a reasonable increase in the concentration of sulfuric acid and a proper decrease in the concentration of vanadium can effectively improve the stability of the vanadium electrolyte.

Acknowledgements The work was carried out under financial support from the National key Research and development program (No. 2016YFC0400403), Sustainable Development Demonstration Zone special project of Chengde (No. 202008F003), and Construction of Vanadium and titanium Industry Technology Innovation Research Institute (No. 202008F027). Conflict of Interest The authors declare no conflicts of interest.

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15.

Nikiforidis G,van de Sanden MCM, Tsampas MN (2019) RSC Adv 9:5649–5673 Doetsch C, Pohlig A (2020) Futur Energy, 263–277 Minke C, Turek T (2018) J Power Sour 376:66–81 Rahman F, Skyllas-Kazacos M (2009) J Power Sour 189:1212–1219 Vijayakumar M, Li L, Graff G, Liu J, Zhang H, Yang Z, Hu JZ (2011) J Power Sour 196:3669– 3672 Park SK, Shim SKJ, Yang JH, Jin C-S, Lee BS, Lee YS, Shin KH, Jeon JD (2014) Electrochim Acta 121:321–327 Balducci A (2018) In: Springer International Publishing, Cham, pp 1–27 Mallakpour S, Dinari M (2012) In: Mohammad A, Inamuddin D (eds) Springer Netherlands, Dordrecht, pp 1–32 Yang Q, Zhang Z, Sun X-G, Hu Y-S, Xing H, Dai S (2018) Chem Soc Rev 47:2020–2064 Anouti M, Couadou E, Timperman L, Galiano H (2012) Electrochim Acta 64:110–117 Cao L, Skyllas-Kazacos M, Menictas C, Noack J (2018) J EnergyChem 27:1269–1291 Wu X, Liu S, Wang N, Peng S, He Z (2012) Electrochim Acta 78:475–482 Drillkens J, Schulte D, Sauer DU (2010) ECS Trans 28:167 Balducci A (2018) In: Springer International Publishing, Cham, p 127 Mallakpour S, Dinari M (2012) In: Mohammad A, Inamuddin D (eds) Springer Netherlands, Dordrecht, pp 1–32

Thermoelectric Generators System Made with Low-Cost Thermoelectric Modules for Low Temperature Waste Heat Recovery Manuela Castañeda, Andrés A. Amell, and Henry A. Colorado

Abstract One of the most common problems in industrial processes is the loss of energy in the form of heat. In the search to recover this type of thermal energy, thermoelectric modules represent a promising alternative because they allow converting this heat into electrical energy, making it more efficient industrial processes. This article presents the laboratory and industrial tests characterization of low-cost thermoelectric modules for the manufacture of a thermoelectric generator for the recovery of low-temperature residual heat. The modules were evaluated at the laboratory level in different configurations, distances from the source and dissipation systems to find the best recovery conditions. The manufactured thermoelectric generator was tested for low temperature heat recovery in an industrial drying oven. The results obtained both at the laboratory and industrial level were compared showing which modules and configurations had the best power generation capacity for this type of industrial process. Keywords Thermoelectric generator · Thermoelectric modules · Cost-efficiency ratio · Sustainability · Life cycle · Circular economy

Introduction Energy consumption is a problem of global interest due to an increasing demand, mainly due to the increase in population and to the economic models. The 84.3% of this energy comes from non-renewable sources, contributing to the emission of

M. Castañeda · H. A. Colorado (B) CC Composites Laboratory, Universidad de Antioquia UdeA, Calle 70 No. 52-21, Medellin, Colombia e-mail: [email protected] A. A. Amell Grupo de Ciencia Y Techología del Gas y Uso Racional de La Energía, Facultad de Ingeniería, Universidad de Antioquia, Calle 67 N° 53–108, Bloque 19–000, Medellín, Colombia © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_44

479

480

M. Castañeda et al.

CO2 into the atmosphere, therefore producing a detrimental effect in the environment. These factors have led to the search for renewable energy sources to mitigate environmental impacts and to increase the energy supply [1]. In Colombia, the final energy consumption for 2018 was 1308 PJ. The transport, industrial, and residential sectors consume more than 80% of total energy. In addition, all the energy used in the different sectors is only 34% of the final energy, the rest is lost due to factors including technological inefficiency, heat, steam, and diverse waste products [2]. Many innovative strategies are required in order to give solutions to the derived issues, which include alternative energy sources [3], always aware of environmental issues taken care of circular economy strategies [4, 5], and energy harvesting [6, 7], among others. This waste heat generated in different processes represents a possible source of heat for the use with thermoelectric materials, which can be found commercially in solid-state devices known as thermoelectric modules [8]. These modules can generate an electrical voltage by applying a temperature gradient between their faces (Seebeck effect). By applying a voltage step to the module this can cause the faces of the module to increase or decrease the temperature depending on the applied voltage (Peltier effect) [9, 10]. The current thermoelectric modules have several limitations such as low efficiency and high cost, but these disadvantages in many cases can be compensated by their easy maintenance as they do not require moving parts, and that are friendly with the environment as well [11, 12]. The implementation of thermoelectric generators is increasingly well received in the aerospace, automotive, and in many industrial processes mainly due to the development and improvement of the thermoelectric materials that make up the modules. In addition, other studies, other components such as the heat source and the dissipation system, can help make this type of power generation more efficient [13–17]. This paper presents the characterization of a thermoelectric generator built with thermoelectric modules reference TEC1-12,706 (TEC) and with a dissipation system made for the recovery of heat from a furnace, particularly tested in the ceramic industry. The initial laboratory-level characterization of the thermoelectric generator and the in-situ characterization in the industrial furnace are shown.

Methodology Laboratory-Level Characterization of the Thermoelectric Generator To characterize the voltage behavior, current, and power generation presented by the commercial cells TEC1-12,706 (TEC), several tests were carried out at the laboratory level, in which 2 cells were used per test. These cells were connected either in series or in parallel, and were located at two distances from the heat source as well (5 cm

Thermoelectric Generators System Made with Low-Cost Thermoelectric …

481

Table 1 Experiment design for laboratory-level characterization of the thermoelectric generator Source temperature

Average source temperature: 530 ± 3 °C

Parameters

Refrigeration cell ref TEC1-12,706 (TEC-REF) Lab tests

Distance from source

10 cm

Serial connection

TEC-REF-S-10 cm

TEC-REF-S-5 cm

Parallel connection

TEC-REF-P-10 cm

TEC-REF-P-5 cm

5 cm

Table 2 Experiment design for the tests of the thermoelectric generator in furnace in the ceramic industry Source temperature

Average source temperature: 163 ± 54 °C

Parameters

Refrigeration cell ref TEC1-12,706 Tests in the industrial furnace

Distance from source

2 cm

Series connection

6TEC-REF-S

2 groups of 3 cells in series and then in parallel

6TEC-REF-M

or 10 cm) (see Table 1). For the assembly of the thermoelectric generator the cells were located between a metal plate and a fin-type heat sink.

Tests of the Thermoelectric Generator in Furnace in the Ceramic Industry An initial characterization of the temperatures presented by the surface of the industrial furnace was carried out using a Testo 885 thermal chamber. The thermoelectric generator was composed of 6 CELLS TEC1-12,706 (TEC) with a fin-type dissipation system. Two types of connections were used for the cells: serial and mixed (M = combination between serial and parallel connection) and the distance between the surface of the furnace and the thermoelectric generator was 2 cm (see Table 2). For all tests a data collection system was used, which allowed to measure the temperatures of the two sides of the cells and the generating voltage for the thermoelectric generator, every 2.4 s.

Visual Inspection of Mechanical Damage to Cells A visual inspection of the mechanical damage caused to the cells during the tests was carried out. For this, some images will be taken with the help of a microscope.

482

M. Castañeda et al.

Results and Discussion In Fig. 1, you can see how the change in voltage generation (mV) vs the temperature delta (°C) of two commercial cells is joined in parallel (P) or in series (S), and at different distances from the source. The fountain had an average temperature of 530 ± 3 °C. It can be observed how increasing the temperature delta favors the generation of electrical voltage of the generator and the conditions in which a better generation of voltage was obtained in the cells in the parallel connection. Table 3 shows the maximum voltage, current, and electrical power values in each of the laboratory tests, including the respective temperatures of the faces of the cells and their temperature delta. The best observed performance was for the cells connected in parallel and a distance to the source of 5 cm, in which, a maximum voltage value of 506.4 mV, a current of 49.8 mA, and a power of 301.99 mV were reached (see Table 3). The thermal characterization of the surface of the industrial furnace is shown in Fig. 2, where the surface of the furnace and its respective thermography are observed. The average temperature of the surface of the furnace was 163 ± 54 °C. Being the temperature of the surface of the furnace is not homogeneous, it generates that the thermoelectric cells will not have the same temperature on the contact surface, which can generate a decrease in the final power generation of the thermoelectric generator. In the tests at the industrial level, it was observed that the temperature delta reached between the faces of the cells varied between 0 and 8 °C as shown in Fig. 3a. 600

Electrical voltage (mV)

Fig. 1 Results of laboratory-level characterization tests of the thermoelectric generator

2TEC REF P 5cm 2TEC REF S 5cm 2TEC REF P 10cm 2TEC REF S 10cm

500 400 300 200 100 0 0

5

10

15

20

25

30

∆T(°C)

Table 3 Maximum values of voltage, current, and power generation in the laboratory-level characterization tests of the thermoelectric generator Test (cm)

Temp cold (Tc)

2TEC REF P 5

78.75

Temp hot (Th) 96.25

Delta Temp

mV

mA

mW

17.5

506.4

49.8

301.99 245.82

2TEC REF P 10

68.25

90.5

22.25

452.4

44.5

2TEC REF S 5

80.5

105.75

25.25

256

25.5

90.78

2TEC REF S 10

90

27.1

102.71

14.25

279

Thermoelectric Generators System Made with Low-Cost Thermoelectric …

483

Fig. 2 a Hot surface and b thermography of the furnace of the ceramic industry

Fig. 3 Whiskers graphs for a temperature delta (°C), b electrical voltage (mV) of the test results of the thermoelectric generator in the ceramic industry furnace

484

M. Castañeda et al.

Fig. 4 Mechanical failures of thermoelectric cells

Therefore, some atypical data can contribute to the low temperature homogeneity that the hot surface of the furnace presented and a heat dissipation by the type of dissipation system. Figure 3b shows that the mixed connection presented a better voltage generation generates approximately twice as much as the serial connection. Also, it can be observed that despite having three times as many cells with respect to laboratory tests, obtaining such a low temperature delta causes the cells not to have their highest performance in the tests in the industrial furnace. During laboratory tests and in the industrial level, some of the thermoelectric cells presented fractures as can be seen in Fig. 4a and b. In most cases they occurred at the ends of the cells, causing direct damage to their power generation capacity. This fragility of the commercial cells represents a great problem when it comes to the assembly of the thermoelectric generator and its performance, because in many cases the fractures are of a very small size which means that it may not be perceived, producing failures throughout the generation system. Although the effective volume of these cells is relatively low in terms of solid waste, thermoelectrics recycling strategies [18] must be considered if a circular process is desired. The ceramics part can be easily separated via simple mechanical processing using for instance in some cases most common facilities, such as those for the processing of hazardous waste, such as those from battery waste management [19], which enable their use in multifunctional materials applications [20, 21].

Conclusions • The heat dissipation system must be more effective in maintaining the temperature gradient between the cell faces for a longer period, and on the contrary, the voltage value obtained will be reduced due to the decrease in the temperature gradient as could be observed in the tests in the industrial furnace. • The low homogeneity of the temperature of the heat source means that the thermoelectric modules do not present a good generation of energy. Therefore, it is

Thermoelectric Generators System Made with Low-Cost Thermoelectric …

485

of great importance the mitigation of this inconvenience with a good dissipation system that allows to maintain the temperature delta between the faces of the cells. • It was observed that the combination between the serial and parallel connections of the cells (mixed connection) presents a better power generation performance. • It has been noted that the good coupling between the different parts of the system generates a better performance of the cells. • Finally, an effective dissipation system allows better performance of the cells and prevents deterioration of these. Acknowledgements The authors gratefully acknowledge the financial support provided by the Colombia Scientific Program within the framework of the call Ecosistema Científico (Contract No. FP44842- 218-2018).

References 1. Bartholdy K (2019) BP statistical review of world energy. (Jan. 2019). https://doi.org/10.1111/ j.1468-0351.1993.tb00076.x 2. UPME (2019) Primer balance de Energía Útil para Colombia y Cuantificación de las Perdidas energéticas relacionadas y la brecha de eficiencia energética Resumen Ejecutivo BEU Sector Residencial y Terciario, p 20 3. Michaelides EES (2012) Alternative energy sources. Springer Sci. Bus. Media 4. Lopera HC, Lopera DC, Lopera GIE (2019) Logistics as an essential area for the development of the solid waste management in Colombia. Informador Técnico, pp 131–154 5. Colorado GI, Echeverri-Lopera HA (2020) The solid waste in Colombia analyzed via gross domestic product: towards a sustainable economy. Rev Fac Ing Univ Antioquia, 96: 51–63 6. Colorado HA, Colorado SA (2017) Manufacturing of zinc oxide structures by thermal oxidation processes as scalable methods towards inexpensive electric generators. Ceram Int 43(17) 7. Quan JM, Colorado J, Yeh HA, Yang PC (2016) Hybridized ZnO nanostructures on carbon-fiber through combustion synthesis induced by joule heating. Ceram Int 8. Jouhara H, Olabi AG (2018) Editorial: industrial waste heat recovery. Energy 160:1–2. https:// doi.org/10.1016/j.energy.2018.07.013 9. Fernández-Yáñez P, Romero V, Armas O, Cerretti G (2021) Thermal management of thermoelectric generators for waste energy recovery. Appl Therm Eng 196. https://doi.org/10.1016/j. applthermaleng.2021.117291 10. Jouhara H et al (2021) Thermoelectric generator (TEG) technologies and applications. Int J Thermofluids 9. https://doi.org/10.1016/j.ijft.2021.100063 11. Araiz M, Casi Á, Catalán L, Martínez Á, Astrain D (2020) Prospects of waste-heat recovery from a real industry using thermoelectric generators: economic and power output analysis. Energy Convers Manag 205(December 2019): 112376. https://doi.org/10.1016/j.enconman. 2019.112376 12. Patowary R, Baruah DC (2018) Thermoelectric conversion of waste heat from IC engine-driven vehicles: a review of its application, issues, and solutions. Int J Energy Res 42(8):2595–2614. https://doi.org/10.1002/er.4021 13. Kaibe H, Makino K, Kajihara T, Fujimoto S, Hachiuma H (2012) Thermoelectric generating system attached to a carburizing furnace at Komatsu Ltd., Awazu Plant. In: AIP conference proceeding, vol 1449, no 2012, pp 524–527. https://doi.org/10.1063/1.4731609

486

M. Castañeda et al.

14. Kuroki T et al (2015) Research and development for thermoelectric generation technology using waste heat from steelmaking process. J Electron Mater 44(6):2151–2156. https://doi. org/10.1007/s11664-015-3722-8 15. Sztekler K, Wojciechowski K, Komorowski M (2017) The thermoelectric generators use for waste heat utilization from conventional power plant. In: E3S web conference, vol 14, pp 1–10. https://doi.org/10.1051/e3sconf/20171401032 16. Hsu CT, Won CC, Chu HS, Hwang JD (2013) A case study of thermoelectric generator application on rotary cement furnace. In: Proceedings of technical papers-international microsystems, packaging, assembly, circuits technology conference IMPACT, no 195, pp 78–81. https://doi. org/10.1109/IMPACT.2013.6706644 17. Charilaou K, Kyratsi T, Louca LS (2019) Design of an air-cooled thermoelectric generator system through modelling and simulations, for use in cement industries. Mater Today Proc 44:3516–3524. https://doi.org/10.1016/j.matpr.2020.11.392 18. Bahrami K, Schierning A, Nielsch G (2020) Waste recycling in thermoelectric materials. Adv Energy Mater 10(19) 19. Kuchhal UC, Sharma PIYUSH (2019) Battery waste management. In: Environmental science and engineering, vol 5, pp 141–155 20. Cardona-Vivas HA, Correa N, Colorado MA (2021) Composite materials fabricated from a conductive polymer with additions of battery waste powders and recycled copper wires. J Compos Mater 21. Cardona-Vivas HA, Correa N, Colorado MA (2021) Multifunctional composites obtained from the combination of a conductive polymer with different contents of primary battery waste powders. Sustain Mater Technol 28

Part XII

Advanced Real Time Imaging

In Situ Observation and Investigation of the Wetting Behaviors of Mold Flux on Steel Substrate Lejun Zhou, Yang Yang, Wanlin Wang, Hao Luo, and Houfa Hu

Abstract There are multi-interphase phenomena occurring in the continuous casting mold, such as shell versus molten flux, inclusions versus molten steel, and molten steel versus gas bubbles, which makes the interfacial phenomena very complex. The interfacial property between liquid mold flux and steel has significant impact on the quality of casting slab. The slag entrapment in mold tends to cause severe defects on slab surface. Therefore, the wetting behaviors of mold flux on steel substrate were in situ observed and investigated using the sessile drop method. The results obtained in this study can provide a new sight to design mold flux and improve slab quality. Keywords Wetting · Contact angle · Interfacial tension · Mold flux · Steel

Introduction The interfacial properties between mold flux and steel have been investigated by many researchers. Lee et al. [1] focused their research on the evaluation of surface tension for liquid Fe-S alloys, the result showed that the surface tension also decreased greatly with the increase of sulfur contents. So, these non-metallic elements, such as O, S, are of surface actives for molten Fe. As for metallic elements, Nakashima et al. [2] summarized the variations of interfacial tension between the liquid iron alloys and liquid slags based on a lot of previous works, they believed that the addition of most of alloy elements, including Ni, Ti, Mn, Mo, V, etc., could reduce the interfacial tension except tungsten (W). Jung et al. [3] investigated the interfacial tension between solid iron and CaO-SiO2 -MO system, results indicated a decrease in the interfacial tension with increased amphoteric oxide additions. Although lots of works related to the interfacial properties have been conducted, most of them concerned more on iron or steel. The research on the effect of composition of slag, especially the composition of the CaO-Al2 O3 -based mold flux, on interfacial properties between molten flux and steel was very few. Therefore, in this L. Zhou (B) · Y. Yang · W. Wang · H. Luo · H. Hu School of Metallurgy and Environment, Central South University, Changsha 410083, P. R. China © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_45

489

490

L. Zhou et al.

Table 1 Major chemical composition of IF steel (mass %) Subject

C

Si

Mn

Al

Ti

Content

0.0007–0.0020

0.0011–0.0065

0.0950–0.1350

0.0150–0.0685

0.0250–0.0750

Table 2 The pre-and post-melted compositions of the designed mold fluxes Sample

C/A

Melted compositions/(wt.%) (±0.05) CaO

SiO2

Al2 O3

Na2 O

Li2 O

F

BaO

MgO

1

1.50

39.00

7.00

28.00

10.00

3.00

7.00

3.00

3.00

2

1.50

37.00

7.00

28.00

10.00

3.00

7.00

5.00

3.00

3

1.50

35.00

7.00

28.00

10.00

3.00

7.00

7.00

3.00

4

1.50

37.80

7.00

27.20

10.00

3.00

7.00

3.00

5.00

5

1.50

36.60

7.00

26.40

10.00

3.00

7.00

3.00

7.00

study, the wetting behavior of CaO-Al2 O3 -based mold flux with different BaO and MgO contents was carried out using sessile drop method.

Experimental Method Sample Preparation The substrates used in this experiment were made of IF steel (Interstitial Free Steel). Table 1 shows major chemical composition of the IF steel. These substrates were made by cutting the steel into small sheets with a size of 30 mm × 30 mm × 5 mm. The sheets also were ground and polished by SiC sandpapers with a grit size down to 2000 to control their surface roughness. The compositions of the mold fluxes are listed in Table 2.

Sessile Drop Test This wetting behavior of CaO-Al2 O3 -based mold flux on the IF steel substrate was tested by sessile drop method. The schematic figure of the apparatus is shown in Fig. 1 [4]. It is mainly composed of a horizontal furnace with MoSi2 heating elements, an image acquisition system, an atmosphere control system, and a temperature control system.

In Situ Observation and Investigation of the Wetting Behaviors …

491

Fig. 1 The schematic figure of sessile drop method

Results and Discussion Effect of BaO and MgO on Contact Angle Figure 2 shows the change of the contact angle between the IF steel substrate and the CaO-Al2 O3 -based mold fluxes with various BaO and MgO contents. The contact angle increased from 62.4° to 74.5° with the increase of BaO content, which suggests that the addition of BaO could weak the wettability of mold flux on the IF steel. However, the contact angle decreased from 62.4° to 51.3° with the increase of MgO content at the same experimental conditions, so MgO showed the opposite impact and it enhanced the wettability and made the mold flux spreading on the surface of IF steel more easily.

Effect of BaO and MgO on Interfacial Tension The surface tension of IF steel (γ s ) was calculated from Eq. (1), which is the formula of surface tension of Fe as suggested by Brooks [5]. The influence of alloy components on the surface tension of the substrate was ignored since the substrates were made of IF steel, and they contained very little alloys as listed in Table 1. γ s = 1870 − (T − 1811)

(1)

where T is the temperature (K), which should be in the range of 1740 K-1920 K.

492

L. Zhou et al.

Fig. 2 Contact angle at slag/steel interface with various BaO and MgO content. a The images of wetting behavior of the mold fluxes on steel substrate, b the variation of contact angle

The surface tension of the molten mold flux (γ l ) was obtained based on a partial molar approach, called as Boni’s empirical Eq. (2) [3]. γl =



γi · Ni

(2)

where γi is the surface tension factor of a pure substance i,Ni is the molar fraction of the pure substance i as a component of the mold flux. The surface tension factors of these substances are listed in Table 3 [6–8]. Figure 3 shows the calculated results of the interfacial tensions between the CaOAl2 O3 -based mold fluxes and the IF steel substrates. The interfacial tension increased from 1630.27 to 1740.81 mN/m when the BaO content increased from 3 to 7 mass%, as shown in Fig. 3a. The increase of interfacial tension also indicates that the wettability of molten flux on the IF steel substrate got weaker when BaO was added into the mold flux. The larger interfacial tension and weaker wettability due to the addition of BaO are beneficial for avoiding the capture of mold flux by the hooks on the solidified shell in continuous casting mold. In addition, the interfacial tension reduced from 1630.27 to1539.66 mN/m with the increase of MgO content also from

In Situ Observation and Investigation of the Wetting Behaviors … Table 3 Surface tension of the pure substances

Oxide

Surface tension factors (mN/m·mol)

CaO

791–0.0935 T

SiO2

243.2 + 0.031 T

Al2 O3

1024–0.177 T

Na2 O

438–0.116 T

Li2 O

300–0.11 T

CaF2

1604.6–0.72 T

BaO

560(1773 K)

MgO

1770–0.636 T

493

Fig. 3 Interfacial tension between IF steel substrate and mold fluxeswith different contents of a BaO and b MgO

3 to 7 mass%, as shown in Fig. 3b. So, the mold flux with the addition of MgO could make it easier to wet the solidified shell and be captured by the hooks, which may cause more slag inclusion defects in the casting product. The variation trends obtained here are consistent with the results from other researchers which were also shown in Fig. 3 a and b [9–12].

Conclusions The effects of BaO and MgO content on wetting behavior of the CaO-Al2 O3 -based mold flux were investigated using sessile drop method. Some important conclusions were summarized as follows: (1)

The contact angle increased from 62.4° to 74.5° with the increase of BaO content, while it decreased from 62.4° to 51.3° with the increase of MgO content. These trends suggest that BaO could weak the wettability, but MgO

494

(2)

L. Zhou et al.

showed the opposite impact and enhanced the wettability of mold flux on the IF steel. The variation trends of interfacial tension with the BaO and MgO contents were consistent with that of contact angle. It increased from 1630.27 to 1740.81 mN/m when the BaO content increased from 3 to 7 mass% and reduced from 1630.27 to 1539.66 mN/m with the addition of MgO content also from 3 to 7 mass%.

References 1. Lee J, Morita K (2002) Evaluation of surface tension and adsorption for liquid Fe-S alloys. ISIJ Int 42(6):588–594 2. Nakashima K, Mori K (1992) Interfacial properties of liquid iron alloys and liquid slags relating to iron-and steel-making processes. ISIJ Int 32(1):11–18 3. Jung EJ, Kim W, Sohn I, Min DJ (2010) A study on the interfacial tension between solid iron and CaO–SiO 2–MO system. J Mater Sci 45(8): 2023–2029 4. Wang W, Li J, Zhou L, Yang J (2016) Effect of MnO content on the interfacial property of mold flux and steel. Metals Mater Int 22 (4): 700–706. (2016.07.09) 5. Brooks RF, Egry I, Seetharaman S, Grant D (2001) Reliable data for high-temperature viscosity and surface tension: results from a European project. High Temp High Press (UK) 33(6): 631–637 6. Hanao M, Tanaka T, Kawamoto M (2007) Evaluation of surface tension of molten slag in multi-component systems. ISIJ Int 47(7):935–939 7. Mills KC (2011) The estimation of slag properties. In: South African Pyrometallurgy 2011 international conference, pp 1–52 8. Mills KC, Karagadde S, Lee PD, Yuan L, Shahbazian F (2016) Calculation of physical properties for use in models of continuous casting process-part 1: mould slags. ISIJ Int 56:264–273 9. Sun H, Nakashima K, Mori K (2006) Influence of slag composition on slag–iron interfacial tension. ISIJ Int 46(3):407–412 10. Ogino K (1975) Interfacial tension between molten iron alloys and molten slags. Tetsu-toHagane 61(8):2118–2132 11. Park SC, Gaye H, Lee HG (2009) Interfacial tension between molten iron and CaO–SiO2– MgO–Al2O3–FeO slag system. Ironmak Steelmak 36(1):3–11 12. Hagemann R, Heller HP, Lachmann S, Seetharaman S, Scheller PR (2012) Slag entrainment in continuous casting and effect of interfacial tension. Ironmak Steelmak 39(7):508–513

Investigation of Echo Source and Signal Deterioration in Ultrasound Measurement of Metal Melt Bitong Wang, Andrew Caldwell, Antoine Allanore, and Douglas H. Kelley

Abstract Ultrasound is a powerful tool for measuring flow and detecting impurities in opaque liquids such as metal melts. Previously, we successfully demonstrated realtime imaging and flow measurement in gallium melt with this technique. However, ultrasound measurement in metal melts is not completely reliable because its operation depends on phenomena that are poorly understood. In this study, we focus on investigating the source of bulk-echoes in gallium melt and the corresponding mechanism of ultrasound signal deterioration. Through electron microscopy and ultrasound measurements, we determine that oxide inclusions are the main source of bulk-echoes in gallium. By conducting a series of ultrasound measurements under different conditions, we demonstrated that the ultrasound signal deterioration is caused by two distinct factors: the loss of echoing objects and the degradation of wetting at the transducer surface. One possible mechanism of wetting degradation—ultrasound-induced cavitation—is further investigated through simulation and experiments. Keywords Ultrasound · Liquid metal · Wetting · Cavitation

B. Wang · D. H. Kelley (B) University of Rochester, 500 Joseph C. Wilson Blvd, Rochester, NY, USA e-mail: [email protected] B. Wang e-mail: [email protected] D. H. Kelley 218 Hopeman Engineering Building, P.O. Box 270132, Rochester, NY 14627-0132, USA A. Caldwell · A. Allanore Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA, USA e-mail: [email protected] © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_46

495

496

B. Wang et al.

Introduction For a solid metal material, there are various advanced techniques and tools available to characterize its material properties. However, only a few methods allow us to ‘look’ inside the metal when it is in liquid state. In many engineering fields, the realtime imaging and measurement of the internal flow of metal melts are important. For example, in metal casting process, the flow structure of the melt inside the casting mold can affect the quality of the cast product [1]. In liquid metal batteries, the flow occurring in the liquid metal electrodes can affect the performance of the battery [2]. However, the opacity of metal melts makes it impossible to use conventional optical techniques, such as Particle Tracking Velocimetry (PTV), to measure the flow inside metal melts directly. In the past thirty years, the Ultrasound Doppler Velocimetry (UDV) has been developed and became a promising technique that could enable real-time flow measurements in metal melts, and it has been validated in many different kinds of metal melts or liquid metals [3–7]. The working principle of UDV has been introduced in previous work [8]. However, for practical applications, problems and challenges still exist for UDV in metal melts. To get sufficient bulk-echoes for velocity measurement, scattering particles or small bubbles are required [3]. Interestingly, strong bulk-echoes can be measured throughout the bulk of high-purity metal melts without adding any artificial scatter particles. People usually assume that those echoes are from natural metal oxides or impurities in the metal melt [7, 9–11]. However, the type and properties of scattering particles would affect the performance of UDV measurement. Therefore, it is important to determine the source of bulk-echoes in metal melt. Ultrasound signal deterioration in metal melt is another undesired phenomenon. Many studies have observed that ultrasound signal quality is unstable, and the signals deteriorate after a long time of UDV measurement in metal melts [4, 7, 9, 11]. This phenomenon seriously limits the measuring time and degrades the signal quality. Although some efforts have been applied to explain or mitigate the deterioration [4, 5, 8–11], its underlying mechanisms remain unclear. Solving the above problems is of great significance for further development and application of the ultrasound technique in liquid metals. In this work, we select liquid gallium as the material because it has been most commonly used in ultrasound studies of metal melts. In Sect. 2, we focus on investigating the source of bulkechoes in gallium. We combined SEM examinations, UDV measurements, and PTV measurements to determine the type of scattering particles. In Sect. 3, we study ultrasound signal deterioration in gallium. We designed a series of experiments to explore the reasons for ultrasound signal deterioration. We also built a simulation model to study the underlying mechanisms behind this phenomenon. In the last section, we summarize our current work and provide suggestions for future work.

Investigation of Echo Source and Signal Deterioration …

497

The Source of Echoes in Gallium Echoes are formed when there is an acoustic impedance mismatch. Various impurities in liquid gallium could reflect or scatter the ultrasound waves and form echoes. The scanning electron microscope (SEM) combined with energy-dispersive X-ray spectroscopy (EDS) has been used to investigate the impurities in gallium. Under SEM, we found many gallium oxide inclusions at the top surface and cross section of a cast gallium sample. Besides gallium oxide, we also found other metal impurities, such as lead, indium, and gold, from the bottom surface and cross section of the gallium sample. Both gallium oxide inclusions and other heavy metal impurities could cause echoes in gallium.

Experimental Methods To investigate which type of impurities is the main source of bulk-echoes, we designed a special experimental apparatus, as shown in Fig. 1. An 8 MHz UDV transducer (Signal Processing, Switzerland) was placed at the top (or bottom) of a vertical acrylic vessel filled with liquid gallium. The liquid gallium was melted from solid gallium of 99.99% purity. The transducer was connected to a DOP3010 Velocimeter (Signal Processing, Switzerland) and operated in energy-profile mode for data acquisition. Compared with the traditional ultrasound echo measurement, the UDV energy-profile only detects and records echo signals caused by moving particles inside the ultrasound beam path. Since there is no extra force applied to the melt, gravity was the dominant force causing particles to move. Therefore, the Fig. 1 Experiment apparatus for investigating echo source. A vertical vessel made of acrylic is filled with liquid gallium or tracer particle water. A UDV transducer was placed at the top (or bottom) for ultrasound measurement. A laser sheet and a camera were also used for PTV measurement in tracer particle water

498

B. Wang et al.

particle trajectories measured by UDV energy-profile at different times and positions would allow us to observe and track the free movement of particles under gravity (i.e., the relative density between particles and liquid gallium). As comparison, we also conducted a similar experiment in water. Tracer particles with known size and density were seeded into water as the echoing objects. To validate the UDV measurement, Particle Tracking Velocimetry (PTV) was applied in water experiments.

The Role of Gallium Oxide Inclusions Figure 2 shows the results of UDV and PTV measurements in tracer particle water. Figure 2 a shows the measured UDV energy-profile map, in which the high-intensity color dots represent the positions of tracer particles detected at different times. Multiple consecutive dots generated by the same particle form a trajectory of that particle. Therefore, the moving direction of a particle can be determined from the slope of its trajectory in the UDV energy-profile map. The trajectories clearly show that tracer particles in water move downward. Figure 2 b shows a snapshot of PTV tracking at t = 33 s, where the red arrows indicate the moving direction of tracer particles. The results of UDV and PTV are consistent with each other, and also consistent with the fact that the density of the tracer particles we selected is higher than that of water. The PTV measurements in water validate the use of UDV energy-profiles to track particles.

Fig. 2 UDV energy-profile and PTV measurement in water seeded with tracer particles. The PTV measurement was made in a region nearly matching the region where UDV measurement were made. Both the PTV tracking arrows and UDV particle trajectories indicate the particles are moving downwards

Investigation of Echo Source and Signal Deterioration …

499

Fig. 3 UDV energy profile maps measured in liquid gallium. High-intensity color dots represent the position of echoing objects at different times. The slope of trajectories indicates that most of the echoing objects in gallium are moving upwards. (Color figure online)

Figure 3 shows UDV energy-profile maps measured in liquid gallium. The UDV transducer has been placed either at the top or bottom of the vessel. In both cases, the slope of trajectories indicates that most of echoing objects inside gallium are moving upward. Among all the impurities we found under SEM, only gallium oxide (β-Ga2 O3 ) has a density (5.88 g/cm3 ) smaller than liquid gallium (6.095 g/cm3 ) so that it may rise towards top under gravity. Therefore, gallium oxide, not other heavy metal impurities, is likely the main source of bulk-echoes in liquid gallium. The SEM images also indicate that gallium oxide inclusions are likely to form agglomerates.

Ultrasound Signals Deterioration in Gallium From the working principle of UDV, the bulk-echo and flow velocity measurements rely on the presence of echoing objects, either scattering particles or bubbles, in the ultrasound beam path. For gallium, the main source of scattering particles is gallium oxide inclusions. Since gallium oxide has a lower density than gallium, it will move toward the top gradually under the effect of gravity. Therefore, as the amount of gallium oxide inclusions within the ultrasound beam path continuously decreases, the ultrasound signals would also deteriorate over time. Theoretically, if there is a vigorous flow to overcome the influence of gravity, a long-time, high-quality UDV measurement could be achieved. However, we found that the loss of echoing objects is not the only reason causing the deterioration of ultrasound signals in gallium. Changes happened at the transducer-gallium interface also play an important role.

500

B. Wang et al.

Fig. 4 a Experimental apparatus for investigating the ultrasound signal deterioration in gallium. b Experimental apparatus for investigating the ultrasound induced cavitation in gallium. A Passive Cavitation Detector (PCD) was used to detect the characteristic acoustic frequencies

Experimental Methods Figure 4a shows the experiment apparatus for investigating the mechanisms of ultrasound signal deterioration over time in gallium. A container filled with liquid gallium is placed on a hotplate, and the temperature is maintained at 60 °C. To enable a longterm UDV measurement, a rotating flow is induced in the gallium by a rotating magnetic field generated by the stirring magnet of the hotplate. A 4 MHz UDV transducer (Signal Processing, Switzerland) was used in this experiment. The transducer was connected to a DOP3010 Velocimeter (Signal Processing, Switzerland) and operated in echo-profile mode for data acquisition. A series of ultrasound measurements were conducted under different conditions, and the measured bulk-echo intensities and back-wall echo intensities were compared among those tests.

The Mechanisms of Ultrasound Signals Deterioration Figure 5 shows the results of a series of consecutive UDV tests in gallium. The temporal evolution of mean bulk-echo intensity from each test is plotted in Fig. 5a. In test-1, we filled the container with liquid gallium and performed UDV measurement. The measured bulk-echo intensity decayed with time. As analyzed above, if the decay is simply caused by the loss of scattering particles in ultrasound beam path, stirring the particles back should restore the bulk-echo intensity. In test-2, we manually stirred the gallium for a while, in order to mix the oxide inclusions back to the bulk. Although the bulk-echo intensity was recovered in some extend by stirring, it is far weaker than test-1 and its decay is much faster. In test-3, we stirred the gallium again,

Investigation of Echo Source and Signal Deterioration …

501

Fig. 5 UDV measurements from a series of consecutive tests in gallium. a Mean bulk-echo intensity vs. time. The mean bulk-echo intensities are spatially averaged over the region 20–100 mm from the transducer. b Back-wall echo intensity versus time. The bulk-echo intensity and back-wall echo intensity always decay with time. Wiping or cleaning the transducer surface improves or restores the signal

but the signal deterioration became even worse. These results suggest that the loss of scattering particles in the ultrasound beam path is not the only reason for UDV signals to deteriorate. We hypothesize the deterioration of UDV signals also results from some changes at the probe surface (the interface between ultrasound transducer and gallium). To test our hypothesis, we wiped the probe surface with a cotton swab and manually stirred the gallium in test-4. This time, the bulk-echo intensity was improved substantially. In test-5, we poured the gallium out of the container, cleaned the ultrasound probe surface, then refilled the container with gallium and restarted the UDV measurement. The echo signal was restored, so that the measured bulk-echo intensity is almost the same as test-1. The effect of changes at the probe surface can be observed even more clearly from back-wall echo measurements. Figure 5b shows the temporal evolution of back-wall echo intensities from a series of tests in gallium. Since the ultrasound reflection coefficient at the container’s back wall is fixed (given by the acoustic impedance mismatch between gallium and the wall material), the back-wall echo intensity should be affected only by the intensity of the emitting beam. Therefore, as the amount of oxide inclusions in the beam path decreases, more acoustic energy should be transmitted to the back wall so that the back-wall echo intensity becomes stronger over time. However, we observed the opposite: back-wall echo intensities always decrease with time. From test-1 to test-3, the back-wall echo intensities decrease monotonically. In test-4 and test-5, wiping or cleaning the probe surface caused the back-wall echo intensity recover. The back-wall echo measurement results are consistent with the bulk-echo measurements, again suggesting that negative changes at the probe surface degraded the ultrasound signals. We hypothesize that those negative changes are related to the acoustic coupling and wetting between the ultrasound transducer and gallium. The wetting between the transducer surface and the gallium is poor due to the high surface tension of

502

B. Wang et al. test-1

500

500

test-2

500

test-3 ultrasound on

Mean bulk echo intensity (arbitrary)

stirring

400

400

400

300

300

300

200

200

200

100

100

100

0

50

100

Time (min)

150

0

50

100

150

Time (min)

0

after 3 hours

50

100

150

Time (min)

Fig. 6 Temporal evolution of mean bulk-echo intensity from three consecutive UDV tests in gallium. The mean bulk-echo intensities are spatially averaged over the region 20–100 mm from the transducer. The degradation of wetting at the probe surface is correlated to the ultrasound emission

gallium. When pouring gallium out, we observed a layer of gallium oxide covering the transducer probe surface, which could degrade the wetting and acoustic coupling. If the layer is thick enough, it could also damp the ultrasound directly. It is possible that the gallium oxides adhere and accumulate at the probe surface as they circulate inside gallium. However, in our control experiment, we found that the ultrasound signal deterioration caused by surface processes (not particles) only happened when there was ultrasound emission. Figure 6 shows the temporal evolution of mean bulk-echo intensity from three consecutive UDV tests in gallium. In test-1, we filled the container with liquid gallium and performed UDV measurement. Continuing to test-2, we only stirred the gallium without cleaning the transducer surface, and the resulting echo intensity was much worse than test-1. In test-3, we poured the gallium out of the container, cleaned the transducer surface, then refilled the container but did not start the UDV measurement immediately (no ultrasound emission). After three hours, we manually stirred the gallium and restarted the UDV measurement. Interestingly, although the transducer had been placed inside the gallium for three hours, the measured echo signal was still similar to test-1. This control experiment suggests that the degradation of wetting at the probe surface is correlated to the ultrasound emission.

Investigation of Ultrasound-Induced Cavitation Next, we asked what mechanisms would cause ultrasound emission to degrade wetting. One possible mechanism is ultrasound-induced cavitation, which could happen if ultrasound waves reduce the local pressure below the vapor pressure of liquid gallium. If ultrasound-induced cavitation happens near the probe surface, it is possible that some micro-gaps or micro-bubbles are generated during this process. Those micro-gaps or micro-bubbles at the probe surface would strongly block ultrasound waves.

Investigation of Echo Source and Signal Deterioration …

503

Cavitation is the basic working principle of ultrasonic treatment, in which high power ultrasound is used to treat and degas metal melts [12]. One concern is that the acoustic power used in UDV is lower than that in ultrasonic treatment, and the ultrasound frequency used in UDV is higher than that in ultrasonic treatment [13]. However, oxide inclusions, dissolved gas, and other impurities existed in metal melts could lower the cavitation threshold by serving as cavitation nucleation sites. From SEM examinations, we learned that there are ample oxide inclusions and metal impurities in gallium. We also studied the solubility of different gases in different metal melts [14], as shown in Fig. 7, and found that hydrogen is the main dissolved gas in liquid gallium. Ultrasound-induced hydrogen bubble growth dynamics in gallium were modeled using the Nolting-Neppiras approximation to the Rayleigh-Plesset equation, following the treatment by Eskin [15]. The bubble radius (normalized by the initial cavitation nucleus radius, here 1 μm) is plotted as a function of the number of cycles in the ultrasound wave in Fig. 8 for a range of ultrasound pressure amplitudes. The model indicates that the cavitation bubbles could be generated in gallium under a relatively low acoustic pressure. From our estimation, the transient acoustic pressure generated by the UDV transducer we used in gallium could reach up to 0.3 MPa [16]. Note that this model does not take into account heterogeneous nucleation on oxide inclusions and other impurities, which would further reduce the cavitation threshold.

Fig. 7 The positive Gibbs energy shown in (a) indicates that liquid gallium will not spontaneously react with the resulting hydrogen in our experiment conditions. Instead, the hydrogen will dissolve into gallium. The green curve in (b) shows that the solubility of oxygen in gallium is extremely low. At gallium’s melting temperature, the solubility of hydrogen in gallium (~10–7 ) is far greater than the solubility of oxygen (~10–23 ). Thus, the main dissolved gas in gallium melt is almost certainly hydrogen. (Color figure online)

504

B. Wang et al.

Fig. 8 Model of bubble growth dynamics. Cavitation is marked by the sudden decrease in bubble size (bubble collapse). The bubble size varies with ultrasound cycles under different acoustic pressures (PA ) at the hydrogen concentration xH = 1.6*10–7 (close to our lab environment)

We experimentally investigated whether cavitation occurs in gallium during UDV measurement. We applied a Passive Cavitation Detector (PCD) (Precision Acoustics, UK) to detect acoustic signals in gallium. The PCD was installed in a con-focal arrangement and connected to an oscilloscope, as shown in Fig. 4b. The acoustic emissions from cavitation are mainly characterized by its ultra-harmonic and subharmonic signals [13]. As the ultrasound emitting frequency used in this experiment was 4 MHz, if cavitation happens, the PCD should detect its ultra-harmonic (6 MHz) frequency and sub-harmonic (2 MHz) frequency signals, as well as some broadband noises. When we placed the PCD in the bulk part of gallium (far away from the UDV transducer), only the emitting frequency (4 MHz) was detected, which implies that cavitation may not have occurred in those areas. However, when we placed the PCD close to (or right above) the UDV transducer surface, it detected weak cavitation signals, as shown in Fig. 9. Note that the ultrasound beam near the transducer surface has a higher intensity. Therefore, it is likely that under the interaction among ultrasound, oxide inclusions, and dissolved gas, very weak cavitation activities occurred

Fig. 9 Acoustic frequency spectrum. The PCD was placed right above the transducer surface. Weak sub-harmonic signals (2 MHz) and ultra-harmonic signals (6 MHz), as well as harmonic signals (8, 12 MHz, etc.) were detected

Investigation of Echo Source and Signal Deterioration …

505

near the UDV transducer surface, in which some micro-gaps or micro-bubbles were generated that deteriorated the wetting and blocked transmission of the ultrasound beam to the liquid gallium. Some studies have reported that maintaining an inert atmosphere (e.g., argon) or acid treatment could prolong the UDV measuring time in metal melts [5]. The reason behind this may not be simply slowing the oxidation; it is also possible that those treatments prevent cavitation by reducing the concentration of oxide inclusion and dissolved gas in metal melts. However, the cavitation signals detected near the transducer surface were very weak, and there are two possible reasons. One reason is that the cavitation itself is very weak and the PCD’s sensitivity is not high enough. The other reason is that because the PCD is designed for water-based liquids, the wetting between liquid gallium and PCD probe surface is poor, which reduces sensitivity. No matter what, further work is needed to fully investigate the role of cavitation in this process. In-situ imaging techniques, such as synchrotron radiation X-ray imaging [17], to directly observe changes at the transducer surface while running UDV measurement, is an option to solve this problem.

Conclusions In this work, we focused on determining the source of bulk-echoes and the mechanisms of ultrasound signal deterioration in gallium. During our study, we gradually realized that those two problems are related to each other. We combined SEM examination and UDV measurement to determine that the gallium oxide inclusions are the main source of bulk-echoes in gallium. As the density of gallium oxide is different from liquid gallium, when there is no vigorous flow, the loss of gallium oxide in the bulk part of gallium would result in the ultrasound signal deterioration. From this point, the presence of oxide inclusions is important for obtaining ultrasound signals. However, through our control experiments, the loss of scattering particles is not the only reason for ultrasound signal deterioration. The degradation of wetting at the UDV transducer surface plays a more important role. From this aspect, the presence of oxide inclusions is undesired since it is related to the degradation of wetting. Oxide inclusions could affect wetting directly and/or by participating in the ultrasound-induced cavitation process. For example, the bubbles generated from cavitation could create micro-gaps if there is already a rigid oxide layer at the transducer surface. More careful research is needed to study whether ultrasound-induced cavitation actually occurs near the UDV transducer surface. In-situ imaging techniques, such as synchrotron radiation X-ray imaging, could be a potential solution. Acknowledgements We thank URNano of the University of Rochester for the use of SEM and EDS facilities, R. Ibanez of the University of Rochester for help setting the PTV measurements. This work was supported by the National Science Foundation [award number CMMI-1562545].

506

B. Wang et al.

References 1. Timmel K, Eckert S, Gerbeth G (2011) Experimental investigation of the flow in a continuouscasting mold under the influence of a transverse, direct current magnetic field. Metall Mater Trans B 42(1):68–80 2. Kelley DH, Weier T (2017) Fluid mechanics of liquid metal batteries 3. Takeda Y (1991) Development of an ultrasound velocity profile monitor. Nucl Eng Des 126(2):277–284 4. Brito D et al (2001) Ultrasonic Doppler velocimetry in liquid gallium. Exp Fluids 31(6):653– 663 5. Tasaka Y, Takeda Y, Yanagisawa T (2008) Ultrasonic visualization of thermal convective motion in a liquid gallium layer. Flow Meas Instrum 19(3):131–137 6. Eckert S, Gerbeth G (2002) Velocity measurements in liquid sodium by means of ultrasound Doppler velocimetry. Exp Fluids 32(5):542–546 7. Cramer A, Zhang C, Eckert S (2004) Local flow structures in liquid metals measured by ultrasonic Doppler velocimetry. Flow Meas Instrum 15(3):145–153 8. Wang B, Kelley DH (2021) Microscale mechanisms of ultrasound velocity measurement in metal melts. Flow Meas Instrum 81: 102010 9. Losev G, Khalilov R, Kolesnichenko I (2017) UDV study of a liquid metal vortex flow. IOP conference series: materials science and engineering, vol 208, p 12022 10. Dadzis K et al (2016) Directional melting and solidification of gallium in a traveling magnetic field as a model experiment for silicon processes. J Cryst Growth 445:90–100 11. Perez A, Kelley DH (2015) Ultrasound velocity measurement in a liquid metal electrode. J Vis Exp 2015(102):1–12 12. Eskin DG (2017) Ultrasonic processing of molten and solidifying aluminium alloys: overview and outlook. Mater Sci Technol 33(6):636–645 13. Tzanakis I et al (2017) Characterizing the cavitation development and acoustic spectrum in various liquids. Ultrason Sonochem 34:651–662 14. Caldwell AH, Allanore A (2019) Analysis of the partial molar excess entropy of dilute hydrogen in liquid metals and its change at the solid-liquid transition. Acta Mater 173:1–8 15. Eskin GI, Eskin DG (2015) Fundamentals of ultrasonic melt processing. In: Ultrasonic treatment of light alloy melts. CRC Press, pp 17–23 16. Ultrasonic Doppler Velocimeter DOP3010. https://www.signal-processing.com/download/dop 3010-brochure.pdf 17. Xu WW et al (2016) Synchrotron quantification of ultrasound cavitation and bubble dynamics in Al–10Cu melts. Ultrason Sonochem 31:355–361

Real-Time Quantification of Nickel, Cobalt, and Manganese Concentration Using Ultraviolet–Visible Spectroscopy—A Feasibility Study Monu Malik, Ka Ho Chan, and Gisele Azimi

Abstract In the present study, the feasibility of the real-time quantification of nickel, cobalt, manganese, and lithium concentration is investigated using Ultraviolet–Visible spectroscopy as a replacement for the conventional method to measure the concentrations of these elements in battery and other applicable industries. Ultraviolet–Visible spectroscopy is one of the most effective, flexible, inexpensive, and simplest analytical techniques to measure species concentration. This technique has a wide range of applications such as wastewater treatment to colloidal nanoparticle characterization. To carry out this study, samples with different concentrations of selected elements are prepared and analyzed using an Ultraviolet–Visible spectrometer. Mathematical relationships are defined between concentration and absorbance and calculated concentrations are compared with ICP-OES results. The effect of elements concentration and path length on absorbance is analyzed to verify the feasibility of the method in the industry. Keywords Absorbance · Battery material · Cuvette cell · Lithium-ion battery · UV–vis spectroscopy

Introduction With the growing concerns of global warming, our future depends on sustainable developments in which electrification of the transportation sector plays a major role. This will require manufacturing of large number of high-performing lithium-ion batteries, especially Li[Nix Mny Coz ]O2 or NMC batteries, which are dominating the M. Malik · K. H. Chan · G. Azimi (B) Department of Chemical Engineering and Applied Chemistry, University of Toronto, 200 College Street, Toronto, ON M5S35, Canada e-mail: [email protected] G. Azimi Department of Materials Science and Engineering, University of Toronto, 184 College Street, Toronto, ON M5S34, Canada © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_47

507

508

M. Malik et al.

existing rechargeable battery market [1]. The production and development of lithiumion batteries are significantly affected by the purity and supply of raw materials such as lithium, nickel, cobalt, and manganese. At the same time, the composition of these elements significantly affects the performance of the lithium-ion batteries, therefore, they need to be carefully measured and controlled. In the existing battery production or recycling industries, the concentration of lithium, nickel, cobalt, and manganese is usually measured by various laboratorybased analytical techniques such as inductively coupled plasma optical emission spectroscopy (ICP-OES) for process control [2]. This involves the movement of the samples between the processing plant and the laboratory where they are analyzed by the technical staff and the results are reported back to the plant for corrective action if needed. Despite being a well-recognized method for process control in the industry, it is a less effective procedure due to the significant time lag between the sampling, analysis, and corrective feedback [3]. In addition, the process is timeconsuming, expensive, and inefficient for modern plants. These plants preferably require an online analytical technique that is non-destructive and provides onsite real-time data to save time, improve the quality of the material, and reduce operating costs. Ultraviolet–Visible (UV–Vis) spectroscopy is a non-destructive analytical technique that is flexible, fast, and inexpensive and could be an alternative to the existing bulky and inefficient methodology for measuring the concentration of various elements such as nickel, cobalt, and manganese in battery and other applicable industries [4]. In UV–Vis spectroscopy, the absorbance or transmittance of incident light is measured as a function of the wavelength and is appropriate for a wide range of organic and some inorganic compounds. In this technique, a UV– Vis spectrometer is used to direct the light through a sample where the energy of the light is absorbed by the molecule which excites electrons from a lower energy orbital to a higher energy unoccupied orbital [5]. A light detector on the opposite end of the spectrometer records the transmitted light and the difference between the two is calculated by the system in form of absorbance or transmittance at the corresponding wavelengths. The ion concentration in the solution can be calculated from the absorbance value at λmax (highest intensity peak) using the Beer-Lambert law, which states that absorbance is directly proportional to concentration and path length [6]. Although UV–Vis spectroscopy has a wide range of applications from wastewater treatment, characterizing colloidal nanoparticles to polymer impregnation, its application in battery production and recycling, and other related industries, to measure the concentration of various elements such as lithium, nickel, cobalt, and manganese has not been investigated in the literature [7–9]. Therefore, the main objective of the present study is to investigate the feasibility of UV–Vis spectroscopy for the onsite measurement of lithium, nickel, cobalt, and manganese. Samples of lithium, nickel, cobalt, and manganese were prepared by changing their concentrations and absorbance measurement was performed in a quartz cuvette cell at room temperature. The effect of path length and concentration on the absorbance of different elements

Real-Time Quantification of Nickel, Cobalt, and Manganese Concentration …

509

was investigated and mathematical relations were developed to calculate the element concentration in a given solution.

Experimental Procedures Chemicals and Materials Lithium sulfate monohydrate (LiSO4 ·H2 O, ≥ 99% pure), nickel sulfate hexahydrate (NiSO4 ·6H2 O, ≥ 98% pure), cobalt sulfate heptahydrate (CoSO4 ·7H2 O, ≥ 99% pure), and manganese sulfate monohydrate (MnSO4 ·H2 O, ≥ 99% pure) were purchased from Sigma Aldrich Canada (Oakville, Canada). Quartz cuvette cells of 2 and 10 mm path lengths with a volume of 0.7 and 1.2 mL (with slits), respectively, were purchased from Lianyungang Highborn Technology Co. Ltd., China.

Instrumentation and Calibration The samples for the concentration measurements were prepared by using Hamilton Microlab 600 auto diluter system where samples were diluted with 5 wt% HNO3 before elemental analysis with ICP-OES. A Lambda 365 UV/Vis spectrometer with a spectral range of 190–1100 nm was used to measure the absorbance of the samples. The spectrometer was calibrated before the absorbance measurement using two standard solutions of 0.001 M K2 Cr2 O7 and 0.0005 M KMnO4 following the procedure described elsewhere [10]. The cuvette cells used in the study were also calibrated. The cuvette cells were triple washed with deionized water (DI) water followed by washing with the pure nickel or cobalt sample before injecting a fixed amount of sample in each cell using a pipette for UV–Vis measurement.

Sample Preparation and Measurements Samples with different concentrations of selected elements were prepared using individual or a combination of elements, where LiSO4 ·H2 O, NiSO4 ·6H2 O, CoSO4 ·7H2 O, and MnSO4 ·H2 O were used as a source of Li, Ni, Co, and Mn, respectively. The sample concentrations were selected based on the solubility of the selected salts in the DI water and their typical concentration in the battery industry. A total of 32 samples were prepared using the selected salts as shown in Table 1. A 10 mL volumetric flask was used to prepare the sample solution, in which a respective amount of source salt/salts was added as per desired metal concentration and the volume of the flask was adjusted to 10 mL using DI water. A clear liquid was obtained after

510

M. Malik et al.

Table 1 List of samples prepared for the present study and their respective concentration Metal

Li

Ni

Co

Mn

Concentration (g L–1 )

0

1

1

1

1

5

5

5

5

10

10

10

10

30

20

20

15

50

30

30

20

70

40

40

25

90

50

50

110

60

60

70

70

sonicating the salt and DI water solution inside the volumetric flask. For the UV–Vis measurements, the respective sample solution was injected into the quartz cuvette cells using a pipette after triple washing the cells with DI water and then with the actual solution. Before the absorbance measurements using the prepared cuvette cells with the sample, a baseline correction was performed using a cuvette cell filled with DI water as a reference sample and keeping the spectrometer sample holder empty. The absorbance spectrum was collected for each sample using the UV–Vis express software in a scan range of 900–200 nm with a scan rate of 240 nm/min. All cuvette cells used are emptied and washed with DI water multiple times before using them to measure the absorbance of the next sample, while the same cuvette filled with DI water was used as a reference sample for all the measurements.

Effect of Path Length According to Beer-Lambert law, the absorbance (A) of an incident ultraviolet and visible light by the molecules is directly proportional to the molecular concentration (C), molar attenuation coefficient or absorptivity (ε), and the optical path length of the incident light traveled through the sample (). A = log

I0 =C ×ε× I

(1)

where I0 and I are the intensity of the incident light before and after passing through the sample, respectively. Literature suggests that Beer-Lambert law follows well for absorbance up to 2 AU (absorbance units) mainly because more than 99% of the incident light is absorbed by the molecules at absorbance 2 AU and less than 1% is transmitted to the receiver. This makes it very difficult for the instrument to correctly measure the absorbance above 2 AU. As the molar attenuation coefficient remains

Real-Time Quantification of Nickel, Cobalt, and Manganese Concentration …

511

constant for a given system, only path length can be varied to maintain the absorbance below 2 AU for a high concentration sample. Therefore, the two cuvette cells with path length of 10 mm and 2 mm were used to perform the absorbance measurement of prepared samples.

Results and Discussion Absorbance Measurement Using 10 mm Path Length Cuvette Cell The calibration of the Lambda 365 UV/Vis spectrometer was performed using 0.001 M K2 Cr2 O7 and 0.0005 M KMnO4 and the comparison with literature results confirms the credibility of the instrument and ensures that the data obtained from various samples of lithium, nickel, cobalt, and manganese is reliable. The absorbance of the samples of the individual elements of different concentrations was first measured using the 10 mm path length cuvette samples. Nickel: As presented in Table 1, samples with different concentrations of nickel were prepared and the absorbance spectrum was collected between 850 and 250 nm, where three peaks at 721 nm, 657 nm, and 394 nm wavelengths were observed as shown in Fig. 1a. Considering the intensity of the peaks, the peak at 394 nm wavelength (highest intensity) is selected as the λmax for nickel, and the absorbance corresponding to the λmax, Ni was used for the analysis. The obtained results show that a clear peak at λmax, Ni was obtained for nickel concentration up to 10 g L–1 while a distorted peak was observed for samples with a higher concentration of nickel (50 g L–1 ). This is because the absorbance of these high concentration samples (≥30 g L–1 ) reaches beyond absorbance 2 AU where more than 99% of the light is absorbed by the metal ions and the detector could not measure the transmitted light that is required for precise calculation of the absorbance. This makes the absorbance data unreliable at a higher concentration of nickel (≥30 g L–1 ). As the first three data points with nickel concentration below 30 g L–1 have absorbance lower than 2 AU, they were used to define the mathematical relation (Eq. 2) between the concentration of the sample measured using the ICP-OES and the obtained absorbance as shown in Fig. 1b. Y = 12.022x − 0.1572

(2)

where Y is nickel concentration (g L–1 ) and x is absorbance obtained from the measurement. Similar mathematical relations were developed for all the investigated systems and stated in subsequent figures. As R2 value for these there data point was very high (0.9996), it confirms a very linear behavior between the concentration of nickel and absorbance below 2 AU. The above equation (Eq. 2) was used to calculate the concentration of all nickel samples using the obtained absorbance values, and

512

M. Malik et al.

Fig. 1 UV–Vis results of the samples with individual elements analyzed using a 10 mm path length cuvette cell: spectrum of a Ni, c Co, e Mn, and g Li. Comparison of absorbance versus measured and calculated concentration of b Ni, d Co, and f Mn, along with error bar

Real-Time Quantification of Nickel, Cobalt, and Manganese Concentration …

513

compared with obtained results from ICP-OES measurements as shown in Fig. 1b. As shown in Fig. 1b, the difference between the measured and calculated concentration of nickel is trivial below 30 g L–1 but increases significantly with a further increase in the concentration. This shows that a 10 mm path length cuvette cannot be used to measure nickel concentration ≥ 30 g L–1 with the UV–Vis technique. Cobalt: Similar to nickel, samples with different concentrations of cobalt were prepared and the absorbance spectrum was collected between 850 and 250 nm, where only one peak with λmax at 512 nm wavelength was observed as shown in Fig. 1c. Similar to nickel, the obtained results show that a clear peak at λmax, Co for cobalt concentration up to 20 g L–1 , while a distorted peak was observed for samples with a higher concentration of cobalt (50 g L–1 ) due to the same reason, explained earlier for the higher concentration of nickel samples. As the first three data points with cobalt concentration below 30 g L–1 have absorbance lower than 2 AU, they were used to define the mathematical relation (Eq. 3) between the concentration of the sample measured using the ICP-OES and the obtained absorbance as shown in Fig. 1d. Y = 12.182x − 0.1688

(3)

Similar to nickel, the R2 value for these there data point was very high (0.9999), which confirms a linear behavior between the concentration of cobalt and absorbance below absorbance 2 AU. Equation 3 was used to calculate the concentration of all cobalt samples using the obtained absorbance values, and compared with the obtained results from ICP-OES measurements as shown in Fig. 1d. The difference between the measured and calculated concentration of cobalt is insignificant below 30 g L–1 but increases significantly with a further increase in the concentration, as shown in Fig. 1d. This shows that a 10 mm path length cuvette cannot be used to measure cobalt concentration ≥30 g L–1 . Manganese: In the case of manganese, the prepared samples were almost colorless at the low concentration of manganese and become light-yellow at a higher concentration. Although several peaks mainly at 530 nm, 436 nm, 401 nm, 358 nm, and 336 nm wavelengths were observed between 850 and 250 nm as shown in Fig. 1e, the absorbance even at λmax, Mn (401 nm) was much lower compared with nickel and cobalt samples. Therefore, the absorbance of all the manganese samples remains lower than 2 AU and no peak distortion was observed at any concentration. However, as the absorbance for even the highest concentration of manganese (70 g L–1 ) remain lower than 0.1, thus, the application of the UV–Vis spectroscopy for the measurement of manganese concentration is limited. Moreover, as the λmax, Mn (401 nm) is very close to λmax, Ni (394 nm), the measurement of manganese concentration in presence of nickel may not be possible due to a large difference in their absorbance value even for the same concentrations. Figure 1f shows a linear relationship between the absorbance and concentration of manganese with a reasonable R2 value of 0.9921. Equation (Eq. 4) was used to calculate the concentration of all manganese samples

514

M. Malik et al.

using the obtained absorbance values, and compared with obtained results from ICPOES measurements as shown in Fig. 1f. The difference between the measured and calculated concentration of manganese is low for low concentration (except 1 g L–1 ) and insignificant for the higher concentration of manganese, as shown in Fig. 1f. This shows that a 10 mm path length cuvette can be used to measure the manganese concentration with reasonable accuracy. Y = 1426.4x − 0.5363

(4)

Lithium: In the case of lithium, the obtain solution remained colorless at every concentration of lithium. Therefore, no absorbance peak was observed between 850 to 250 nm wavelength as shown in Fig. 1g. Hence, the concentration of pure lithium cannot be measured using UV–Vis spectroscopy.

Absorbance Measurement Using 2 mm Path Length Cuvette Cell The path length was reduced five times using a 2 mm path length cuvette cell to measure samples with a high concentration of selected elements especially nickel and cobalt. The absorbance measurements were carried in the same scan range (850–250 nm) for nickel, cobalt, and manganese. The obtained results show that the absorbance of the samples also reduces five times and remains lower than the 2 AU even for the highest concentration of nickel and cobalt selected for the study, as shown in Fig. 2a–d. Clear peaks were obtained for both nickel and cobalt without any distortion at higher concentrations. Similar to the 10 mm path length case, the obtained results were used to define the mathematical relation (Eqs. 5 and 6) between the concentration of the sample measured using the ICP-OES and the absorbance for both nickel and cobalt as shown in Fig. 2b and d, respectively. Y = 67.716x − 1.7867

(5)

Y = 58.516x + 0.2101

(6)

The high R2 values of 0.9989 and 0.9997 for both nickel and cobalt, respectively, confirm linear behavior between the concentration of both nickel and cobalt with their absorbance. Equations 5 and 6 were used to calculate the concentration of both pure nickel and cobalt samples, respectively, using the obtained absorbance values, and compared with the obtained results from ICP-OES measurements as shown in Fig. 1b, d. For both nickel and cobalt systems, the difference between the measured and calculated concentration remains insignificant regardless of the solution concentration. This shows that UV–Vis spectroscopy can be used to accurately measure the

Real-Time Quantification of Nickel, Cobalt, and Manganese Concentration …

515

Fig. 2 UV–Vis results of the samples with individual elements analyzed using a 2 mm path length cuvette cell: spectrum of a Ni, c Co, and e Mn. Comparison of absorbance versus measured and calculated concentrations of b Ni, d Co, and f Mn, along with error bar

concentration of nickel and cobalt in an aqueous solution using a 2 mm path length cuvette cell. In the case of pure manganese, it is suggested to use a 10 mm or higher path length cuvette as the absorbance of the manganese ions is low. Similar to the nickel and cobalt sample, the absorbance of the manganese sample was also reduced by five times with a 2 mm path length cuvette cell as shown in Fig. 2e, f. Except for the 1 g L–1 sample where the absorbance was below the detection limit of the instrument and recorded

516

M. Malik et al.

as 0 AU, the obtained results for higher concentration were reasonably accurate as shown in Fig. 2f. Using the obtained results, the mathematical relation (Eq. 7) was defined between concentration and absorbance of manganese. The measured and calculated concentrations were compared as shown in Fig. 2f, where a relatively small error (≤ ± 10%) suggests that a 2 mm path length cuvette cell can be still used for the measurement of manganese concentration with reasonable accuracy. However, a 10 mm or higher path length cuvette is suggested for higher accuracy. Y = 8911.7x + 5.7333

(7)

Conclusions The present study investigates the feasibility of the real-time quantification of lithium, nickel, cobalt, and manganese for the battery industry using simple and cost-effective UV–Vis spectroscopy as a replacement for bulky, expensive, and time-consuming laboratory-based analytical techniques such as ICP-OES. Samples with different concentrations of lithium, nickel, cobalt, and manganese were prepared and analyzed with Lambda 365 UV/Vis spectrometer. The effect of path length was studied using two different cuvette cells of 10 mm and 2 mm path lengths and mathematical relations were developed between absorbance and concentration. The obtained results show that the low concentration of nickel and cobalt (up to ~ 30 g L–1 ) can be accurately measured using a 10 mm length cuvette cell. However, for the higher concentration of nickel and cobalt, the absorbance values were beyond 2 AU, which makes the data unreliable and leads to significant error in the calculated concentration. Therefore, it is suggested to use of a 2 mm path length cuvette cell to accurately measure both low and high concentrations of nickel and cobalt. To measure the concentration of manganese, a 10 mm path length cuvette is more suitable to obtain accurate results, especially at low concentrations because the absorbance of manganese samples is much lower than that of nickel and cobalt samples. In the case of lithium, no peaks were observed in the selected scan range regardless of the lithium concentration as the obtained solution was completely colorless. Therefore, pure lithium concentration cannot be measured using UV–Vis spectroscopy. Overall, UV–Vis spectroscopy can be used for real-time quantification of nickel, cobalt, and manganese with high accuracy for batteries and other applicable industries. Further studies are underway to investigate the effect of an element’s presence on the absorbance of the other element. Acknowledgements The authors acknowledge the financial support provided by Hatch and Natural Sciences and Engineering Research Council of Canada.

Real-Time Quantification of Nickel, Cobalt, and Manganese Concentration …

517

References 1. Zubi G, Dufo-López R, Carvalho M, Pasaoglu G (2018) The lithium-ion battery: state of the art and future perspectives. Renew Sustain Energy Rev 89:292–308 2. Corazza M, Baldo F, Pagnoni A, Miscioscia R, Virgili A (2009) Measurement of nickel, cobalt and chromium in toy make-up by atomic absorption spectroscopy. Acta Derm Venereol 89(2):130–133 3. Archery E (2005) Simultaneous absorptiometric determination of copper, nickel, iron and cobalt in refinery process streams: potential on-line application. University of Stellenbosch, Stellenbosch 4. Barbosa-García O, Ramos-Ortiz G, Maldonado J, Pichardo-Molina J, Meneses-Nava M, Landgrave J, Cervantes-Martínez J (2007) UV–vis absorption spectroscopy and multivariate analysis as a method to discriminate tequila. Spectrochim Acta Part A Mol Biomol Spectrosc 66(1):129–134 5. Rocha FS, Gomes AJ, Lunardi CN, Kaliaguine S, Patience GS (2018) Experimental methods in chemical engineering: ultraviolet visible spectroscopy—UV-Vis. Can J Chem Eng 96(12):2512–2517 6. Picollo M, Aceto M, Vitorino T (2019) UV-Vis spectroscopy. Phys Sci Rev 4(4) 7. Quinlan PJ, Grishkewich N, Tam KC (2017) Removal of 2-naphthoxyacetic acid from aqueous solution using quaternized chitosan beads. Can J Chem Eng 95(1):21–32 8. Xiao X, Sun Y, Sun W, Shen H, Zheng H, Xu Y, Zhao J, Wu H, Liu C (2017) Advanced treatment of actual textile dye wastewater by Fenton-flocculation process. Can J Chem Eng 95(7):1245–1252 9. Haiss W, Thanh NT, Aveyard J, Fernig DG (2007) Determination of size and concentration of gold nanoparticles from UV−Vis spectra. Anal Chem 79(11):4215–4221 10. Malik M, Chan KH, Azimi G (2021) Quantification of nickel, cobalt, and manganese concentration using ultraviolet-visible spectroscopy. RSC Adv 11(45):28014–28028

Part XIII

Advances and Discoveries in Non-equilibrium Driven Nanomaterials and Thin Films

Salt-Assisted Chemical Vapor Deposition Synthesis of 2D WSe2 and Its Integration in High Performance Field-Effect Transistors Anupama B. Kaul and Avra S. Bandyopadhyay

Abstract The synthesis of two-dimensional (2D) transitional metal dichalcogenides (TMDCs), including in the monolayer limit with control on crystallinity, is an important factor for their integration into a number of device platforms. Monolayer tungsten diselenide (WSe2 ) has recently attracted a great deal of interest because of its tunable charge transport behavior, making it attractive for a variety of electronic and optoelectronic devices. However, the controlled and efficient synthesis of WSe2 using chemical vapor deposition (CVD) is often challenging because of the high temperatures required to generate a steady flux of tungsten atoms in the vapor phase from the oxide precursors. Here, we use a salt (NaCl)-assisted process within the CVD furnace to reduce the growth temperature to ~750 °C, which is lower than the typical temperatures needed with conventional CVD for realizing monolayer WSe2 . The role of substrates also play an important role in the CVD growth process and we found that sapphire improves the optical and crystalline quality of both CVD-grown and mechanically exfoliated WSe2 when compared with SiO2 /Si substrates. Finally, we fabricated WSe2 -based field-effect transistors using metal contacts of varying work functions and analyzed the interface properties in metal-2D WSe2 junctions by extracting the interface state trap density, showing their promise for state-of-theart electronic, optoelectronic, and quantum-optoelectronic devices using scalable synthesis routes. Keywords Chemical vapor deposition · Tungsten diselenide · Raman and PL spectroscopy · Interface state density

A. B. Kaul (B) · A. S. Bandyopadhyay Department of Electrical Engineering, PACCAR Technology Institute, Denton, TX, USA e-mail: [email protected] A. B. Kaul Department of Materials Science and Engineering, University of North Texas, Denton, TX 76203, USA © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_48

521

522

A. B. Kaul and A. S. Bandyopadhyay

Introduction Two-dimensional (2D) materials, such as graphene and transition metal dichalcogenides (TMDCs), recently have attracted significant attention due to their unique properties and are widely used for fabricating electronics, optoelectronics, flexible and sensing devices [1–13]. Amongst the TMDCs, MoS2 [7, 10] has been widely explored, but studies on WSe2 are still in the relatively early stages. Moreover, WSe2 offers unique attributes such as ultra-broadband detection spectral range, a high photoluminescence (PL) quantum yield (nearly unity), a strong spin–orbit coupling, all of which make it intriguing for high performance optoelectronic device applications such as photodetectors, light emitting diodes, and quantum-optoelectronics devices [4, 8, 14]. At the same time, practical device possibilities for WSe2 rely on breakthroughs in the controlled and efficient growth of large-area films. Controlled CVD growth of WSe2 is often challenging in comparison with MoS2 since the metal oxide precursors employed for nucleating WSe2 , such as WO3 is significantly more difficult to vaporize, that consequently yields a lower vapor pressure P for WO3 , compared to the MoO3 precursor used for MoS2 synthesis [8]. In this work, we have conducted experiments to optimize the synthesis of 1L and bi-layer (2L) WSe2 by using a halide-assisted low-pressure CVD process, where NaCl was the halide of our choice which helps activate the tungsten from the oxide precursor to a lower growth temperature. Growth parameters of interest in our study included the NaCl concentration and growth T, where the role of these parameters on the properties of the as-grown WSe2 crystals was examined. The choice of substrate is also an important parameter in CVD growth, as it directly impacts the crystalline quality of the material synthesized. We used two different substrates, SiO2 /Si and sapphire (Al2 O3 ), for this study and the crystalline quality of WSe2 was found to improve on sapphire. We have also analyzed the phonon lifetime τ in CVD grown and mechanically exfoliated WSe2 and found that τ increases on sapphire substrates, suggestive of its higher crystalline quality due to a more pristine interface on sapphire when compared to SiO2 . At the same time, interfaces play an important role to determine device performance figures-of-merit. Compared to traditional semiconducting materials such as silicon, Ge, or III−V materials, the 2D materials, including TMDCs, exhibit pristine surfaces with minimal dangling bonds which should facilitate the realization of interface states with low-interface trap density. This is particularly pertinent at metal– semiconductor hetero-junctions where a high interface trap density can often degrade device performance. In the second part of this work, we discuss the fabrication of WSe2 field-effect-transistors (FETs) using metals with different work functions, where we have analyzed the metal-2D WSe2 interface properties. The conductivity of WSe2 was found to be p-type, ambipolar, and n-type with Au (with Ti as adhesion layer), Mo, and Al metal contacts, respectively. The interface state properties in the metal-2D WSe2 junctions were also investigated using capacitance-frequency measurements.

Salt-Assisted Chemical Vapor Deposition Synthesis of 2D WSe2 …

523

Experimental In this study, WSe2 nanosheets were grown using a halide-assisted CVD method and the details of the synthesis process is described in our previous work [8]. In short, a CVD furnace with a three-foot quartz reaction tube was used for the WSe2 synthesis on either SiO2 /Si (270 nm) or C-plane (0001) sapphire substrates. The optical image of the setup is shown in Fig. 1a. Selenium powder was placed upstream at a position that yielded a T ~ 270 °C during the growth, while a mixture of WO2.9 and NaCl was placed at the center of the furnace. The substrates were placed downstream and facedown, while a mixture of Ar and H2 was introduced as the carrier gas. The center of the furnace was ramped to 750 °C at a ramp rate of 10 °C/min for the growth phase. When the center heating zone reached 750 °C, this translated to a T ∼ 700 °C at the substrate location. During the growth process, the flow rate of Ar/H2 was kept at 120/30 sccm, and the chamber P was ~ 6 Torr throughout the growth run. The WSe2 nanosheets were also mechanically exfoliated on top of SiO2 /Si (t ox ~ 270 nm) and sapphire substrates using the scotch tape method [1]. The samples were characterized using Raman and PL spectroscopy using a LabRAM HR Evolution NIR microscope equipped with a 532 nm laser for excitation. The WSe2 -based FETs were fabricated using a standard e-beam lithography process using the JEOL JSM-7001F SEM and XENOS XPG 2 EBL pattern writer. The electrical measurements were conducted using a state-of-the-art Lakeshore probe stage (CRX-4 K) interfaced to an ultra-low noise semiconductor parameter analyzer (Keysight B1500A).

Fig. 1 a The optical image of the CVD furnace for the growth of WSe2 . b–d The optical images of the WSe2 nanocrystallites grown due to incomplete nucleation. The e Raman and f PL spectra of monolayer WSe2 . Inset of (e) shows the optical image of monolayer WSe2

524

A. B. Kaul and A. S. Bandyopadhyay

Results and Discussion CVD Growth Analysis of WSe2 In this study we implemented a salt-assisted CVD method to synthesize WSe2 nanosheets. In particular, NaCl was used to reduce the growth T of WSe2 with the formation of volatile tungsten oxyhalides [15]. Several important CVD growth parameters were systematically varied in order to optimize crystalline quality. Among these parameters, the amount of precursors used, such as Se, WO2.9 , and NaCl, were found to be the two variables that appeared to have a significant influence on the growth of our WSe2 crystallites. Figure 1b–d shows the optical images of the WSe2 nanosheets due to incomplete nucleation of the WSe2 seeds, attributed to the low amount of NaCl used in the mixture. For example, Fig. 1b shows the WSe2 nanosheets grown using CVD where the amount of Se, WO2.9 , and NaCl used were 15.9 mg, 22 mg, and 4 mg, respectively, while Fig. 1d shows the WSe2 nanosheets grown using CVD where the amount of Se, WO2.9 , and NaCl used were 15 mg, 22.3 mg, and 5 mg, respectively. However, with careful optimization of the growth parameters, we were successfully able to synthesize WSe2 nanosheets at a growth T of ~750 °C where the ratio of WO2.9 and NaCl was tuned to be in the 7:3 ratio. Thereafter, Raman and PL spectroscopy were used to evaluate the quality of our synthesized WSe2 crystals. The Raman spectra of monolayer WSe2 is shown in Fig. 1e at room T, where the two characteristic peaks for WSe2 at 248 cm−1 , assigned 1 vibrational mode, and at 258 cm−1 , assigned to the A1g out-plane to the in-plane E 2g vibrational mode, are observed [16, 17]. The inset of Fig. 1e shows the optical image of the as-grown monolayer WSe2 nanosheets. Figure 1f depicts a typical PL spectra of monolayer WSe2 nanosheets which exhibits a strong emission at ~1.61 eV and the single, symmetric PL A-peak suggests the direct band gap nature of monolayer WSe2 , which is in excellent agreement with other recent PL reports for monolayer WSe2 [8, 16, 18]. To study the impact of the halide content, the ratio of WO2.9 and NaCl was varied from 7:1 to 7:4, while the growth T was fixed at 750 °C, and the full-width-halfmaxima (FWHM) of the PL A-peak for 1L WSe2 is shown in Fig. 2a. The lowest value of the FWHM in the PL A-peak was also found to be ~0.13 eV when the mixing ratio was 7:3, as shown in Fig. 2a. Additionally, the ratio of the PL and Raman intensity (I Lum /I Raman ) has been utilized as a metric to gauge optical quality of TMDCs to determine the intrinsic luminescence quantum efficiency [8, 19]. In our study, I Lum and I Raman represent the intensities of the PL A-peak and the Raman 1 peak, respectively, and the secondary y-axis of Fig. 2a shows I Lum /I Raman as the E 2g oxide-to-halide ratio is varied. It was found that the I Lum /I Raman was maximized at ~ 0.30 when the mixing ratio was 7:3. The PL spectra for 1L WSe2 at various oxide-tohalide ratios are also shown in Fig. 2b, from which it can be inferred that the lowest full-width-half-maximum (FWHM) was ~0.13 eV obtained when the ratio is 7:3.

Salt-Assisted Chemical Vapor Deposition Synthesis of 2D WSe2 …

525

Fig. 2 a The dependence of the FWHM of the PL A-peak and I Lum /I Raman ratio as a function of the halide ratio for 1L WSe2 at T = 750 °C. b PL spectroscopy of monolayer WSe2 grown using varying oxide-to-halide ratios. c Raman and d PL spectra of monolayer WSe2 grown on SiO2 /Si and sapphire substrates. Comparative analysis of T-dependency of phonon lifetime in e CVD grown and f mechanically exfoliated WSe2 on SiO2 /Si and sapphire substrates

Role of Substrate in WSe2 Nanocrystallites Next, we move to analyze the role of substrates in influencing the crystalline quality of CVD grown WSe2 . Figure 2c and d shows a comparative analysis of the Raman and PL spectra, respectively, of monolayer WSe2 grown on sapphire and SiO2 /Si substrates under the optimized metal oxide-to-halide ratio of 7:3 and a growth T of ~750 °C, as discussed earlier. The intensity of the A-peak PL emission on sapphire was nearly ~2 × higher when compared to SiO2 /Si substrates, as is evident in Fig. 2d. Additionally, I Lum /I Raman was found to be 0.76 for sapphire while it was calculated to be ~0.30 at the most optimal conditions for SiO2 /Si substrates; this data is clear evidence for the improved optical quality of 1L WSe2 grown on sapphire. A further gauge of crystalline quality is the PL FWHM, where the FWHM of the A-peak seen Fig. 2d was narrower for WSe2 on sapphire (~0.08 eV), compared to ~0.13 eV for SiO2 /Si. Phonon lifetime τ is an important parameter to analyze the optical quality in the crystal and it is calculated using the energy uncertainty relationship, with the phonon linewidth  as given by Eq. (1) below [20, 21], τ=

 

(1)

526

A. B. Kaul and A. S. Bandyopadhyay

where τ is in picoseconds, and  is Planck’s constant (~5.3 ps-cm−1 ). The value of 1 mode at τ for CVD grown monolayer WSe2 on SiO2 /Si and sapphire for the E 2g T = 298 K was found to be ~ 0.76 ps and 0.98 ps, respectively. Figure 2e and f show τ for CVD grown and mechanically exfoliated monolayer WSe2 , respectively, on SiO2 /Si and sapphire substrates, where the mechanical exfoliation is conducted using the scotch tape approach, details of which are provided in the Experimental Section. The lifetime τ was found to decrease at higher T due to the higher probability for phonon–phonon and electron–phonon interactions leading to dissipation, as was reported in our earlier works [8, 21]. As seen from Fig. 2e and f, τ is higher for WSe2 on sapphire which is attributed to the superior quality of monolayer WSe2 on sapphire for both CVD-grown and mechanically exfoliated samples.

Role of Metal Contacts in Electrical and Interface State Properties in WSe2 -based FETs In the second part of our work, we report here, WSe2 FETs fabricated using metals with different work functions φ m , such as Au (φ m ~ 5.47 eV), Mo (φ m ~ 4.53 eV), and Al (φ m ~ 4.08 eV). The conductivity of WSe2 was found to be ptype, ambipolar, and n-type with Au, Mo, and Al metal, respectively, which is in accordance with our previous work [10] and other reports [21]. The gate leakage currents for the Au/Ti/WSe2 , Mo/WSe2 , and Al/WSe2 FETs are shown in Fig. 3a– c. The Au/Ti/WSe2 , Mo/WSe2 , and Al/WSe2 FETs were also annealed in vacuum conditions and the mobilities were found to increase post-annealing treatment. Frequency-dependent conductance Gp measurements have been carried out to investigate the interface properties between the metal and 2D WSe2 as it is very important to determine the effects of the interface trap states in the metal–semiconductor (MS) junction. It is possible to calculate the interface trap density, i.e., Dit by investigating the MS Schottky junction at different frequencies f as the filling and refilling of the trap states cause a measurable change in the capacitance of the junction. The values of Dit can be obtained from the Gp /ω- ω plot which is shown in Fig. 4a–c for Au/Ti/WSe2 , Mo/WSe2 , and Al/WSe2 junctions, respectively.

Fig. 3 The gate leakage currents in a Au/Ti/WSe2 , b Mo/WSe2 , and c Al/WSe2 FETs. The leakage currents were found to stay mostly at the noise-floor (~10-10 A-10-14 A)

Salt-Assisted Chemical Vapor Deposition Synthesis of 2D WSe2 …

527

Fig. 4 The normalized conductance vs angular frequency Gp /ω- ω plots for a Au/WSe2 , b Mo/WSe2 , and c Al/WSe2 MS junctions before and after annealing

The normalized conductance is expressed as [22, 23],  GP q Dit  ln 1 + (ωτit )2 = ω 2ωτit

(2)

where τit and ω are the interface trap time constant and angular frequency (ω = 2πf ), respectively. For the maximum value of the peak, the first derivative of Eq. (2) was taken, and the density of interface states and the interface trap time constant were expressed as,   2.5 G P Dit = q ω max

(3)

The values of Dit were calculated to be ~4.51 × 1013 cm−2 eV−1 , 1.11 × 1014 cm−2 eV−1 , and 2.66 × 1013 cm−2 eV−1 for Au/Ti/WSe2 , Mo/WSe2 , and Al/WSe2 junctions, respectively, at room temperature which shows that the Dit is the lowest for Au/Ti/WSe2 FET. It was also found that the value of Dit decreases by an order of magnitude with post-annealing treatment for Au/Ti/WSe2 , Mo/WSe2 , and Al/WSe2 devices, which are lower than the previously reported values for Dit in WSe2 -based devices [24]. These results confirm the role of annealing to improve interface properties in metal-2D WSe2 junctions.

Conclusions In conclusion, we have synthesized WSe2 using a salt-assisted CVD method, where NaCl was added to WO2.9 to lower the growth T to ~750 °C at an optimized 7:3 metal oxide-to-halide ratio. The role of substrates on WSe2 crystallites was also studied and it was found that the optical and crystalline quality of WSe2 improved when sapphire was used as the substrate. Moreover, the value of τ was found to be ~0.78 ps and 0.98 psfor our CVD grown monolayer WSe2 on SiO2 /Si and sapphire substrates,

528

A. B. Kaul and A. S. Bandyopadhyay

respectively, indicating the superior quality of WSe2 on sapphire. Finally, we fabricated WSe2 -based FETs using Au/Ti, Mo, and Al as the metal contacts and the conductivities of WSe2 were found to be p-type, ambipolar, and n-type, respectively. The interface state density Dit was also extracted to be ~4.51 × 1013 cm−2 eV−1 , 1.11 × 1014 cm−2 eV−1 , and 2.66 × 1013 cm−2 eV−1 for Au/Ti/WSe2 , Mo/WSe2 , and Al/WSe2 junctions, respectively. Our results validate the potential of 2D WSe2 as a promising 2D semiconductor for electronics, optoelectronics, and quantum-scale systems. Acknowledgements We are extremely grateful to the Air Force Office of Scientific Research (grant number FA9550-15-1-0200) and the National Science Foundation (grant number NSF ECCS 1753933) who provided funding support that enabled us to pursue this work. A.B.K. also acknowledges support from the PACCAR Technology Institute and the Endowed Professorship support at the University of North Texas.

References 1. Novoselov KS, Geim AK, Morozov SV, Jiang D, Zhang Y, Dubonos SV, Grigorieva IV, Firsov AA (2004) Electric field effect in atomically thin carbon films. Science 306(5696):666–669 2. Novoselov KS, Jiang D, Schedin F, Booth TJ, Khotkevich VV, Morozov SV, Geim AK (2005) Two-dimensional atomic crystals. Proc Natl Acad Sci 102(30):10451–10453 3. Wang QH, Kalantar-Zadeh K, Kis A, Coleman JN, Strano MS (2012) Electronics and optoelectronics of two-dimensional transition metal dichalcogenides. Nat Nanotechnol 7(11):699–712 4. Kaul AB (2014) Two-dimensional layered materials: structure, properties, and prospects for device applications. J Mater Res 29(3):348–361 5. Min M, Hossain RF, Adhikari N, Kaul AB (2020) Inkjet printed organo-halide 2D layered perovskites for high-speed photodetectors on flexible polyimide substrates. ACS Appl Mater Interfaces 12:10809 6. Min M, Sakri S, Saenz GA, Kaul AB (2021) Photophysical dynamics in semiconducting graphene quantum dots integrated with 2D MoS2 for optical enhancement in the near UV. ACS Appl Mater Interfaces 13(4):5379–5389 7. Bandyopadhyay AS, Adhikari N, Kaul AB (2019) Quantum multibody interactions in halide-assisted vapor-synthesized monolayer WSe2 and its integration in a high responsivity photodetector with low-interface trap density. Chem Mater 31(23):9861–9874 8. Jayanand K, Chugh S, Adhikari N, Min M, Echegoyen L, Kaul AB (2020) Sc 3 N@ C 80 and La@ C 82 doped graphene for a new class of optoelectronic devices. J Mater Chem C 8(12):3970–3981 9. Mehta RK, Kaul AB (2021) Black phosphorus-molybdenum disulfide heterojunctions formed with ink-jet printing for potential solar cell applications with indium tin oxide. Curr ComputAided Drug Des 11(5):560 10. Bandyopadhyay AS, Saenz GA, Kaul AB (2020) Role of metal contacts and effect of annealing in high performance 2D WSe2 field-effect transistors. Surf Coat Technol 381:125084 11. Desai JA, Bandyopadhyay A, Min M, Saenz G, Kaul AB (2020) A photo-capacitive sensor operational from 6 K to 350 K with a solution printable, thermally-robust hexagonal boron nitride (h-BN) dielectric and conductive graphene electrodes. Appl Mater Today 20:100660 12. Bandyopadhyay AS, Jayanand K, Kaul AB (2020) Electrical and optoelectronic properties analysis in two-dimensional multilayer WSe2 phototransistor for high speed device applications. In: 2020 IEEE 15th international conference on nano/micro engineered and molecular system (NEMS), pp 18–21

Salt-Assisted Chemical Vapor Deposition Synthesis of 2D WSe2 …

529

13. Bandyopadhyay AS, Jayanand K, Kaul AB (2020) Many-body interactions in halide-assisted CVD grown WSe2 for high performance photodetectors. In: 2020 IEEE 15th international conference on nano/micro engineered and molecular system (NEMS), pp 22–25 14. Chakraborty C, Kinnischtzke L, Goodfellow KM, Beams R, Vamivakas AN (2015) Voltagecontrolled quantum light from an atomically thin semiconductor. Nat Nanotechnol 10:507–511 15. Li S, Wang S, Tang D-M, Zhao W, Xu H, Chu L, Bando Y, Goldberg D, Eda G (2015) Halideassisted atmospheric pressure growth of large WSe2 and WS2 monolayer crystals. Appl Mater Today 1:60 16. Li H, Lu G, Wang Y, Yin Z, Cong C, He Q, Wang L, Ding F, Yu T, Zhang H (2013) Mechanical exfoliation and characterization of single- and few-layer nanosheets of WSe2 , TaS2 , and TaSe2 . Small 9:1974–1981 17. Liu B, Fathi M, Chen L, Abbas A, Ma Y, Zhou CW (2015) Chemical vapor deposition growth of monolayer WSe2 with tunable device characteristics and growth mechanism study. ACS Nano 9:6119–6127 18. Ross JS, Klement P, Jones AM, Ghimire NJ, Yan JQ, Mandrus DG, Taniguchi T, Watanabe K, Kitamura K, Yao W, Cobden DH, Xu X (2014) Electrically tunable excitonic light-emitting diodes based on monolayer WSe2 P-N junctions. Nat Nanotechnol 9:268–272 19. Splendiani A, Sun L, Zhang Y, Li T, Kim J, Chim CY, Galli G, Wang F (2010) Emerging photoluminescence in monolayer MoS2 . Nano Lett 10:1271–1275 20. Beechem T; Graham S (2008) Temperature and doping dependence of phonon lifetimes and decay pathways in GaN. J Appl Phys 103:093507 21. Bandyopadhyay AS, Biswas C, Kaul AB (2020) Light-matter interactions in two dimensional layered WSe2 for gauging evolution of phonon dynamics. Belstein J Nanotechnol 11:782–797 22. Schroder D (1998) Semiconductor material and device characterization, 2nd edn. Wiley, Toronto 23. Hussain I, Soomro MY, Bano N, Nur O, Willander M (2012) Interface trap characterization and electrical properties of Au-ZnO nanorod Schottky diodes by conductance and capacitance methods. J Appl Phys 112:064506 24. Kim Y, Kim AR, Yang JH, Chang KE, Kwon JD, Choi SY, Park J, Lee KE, Kim DH, Choi SM, Lee KH, Lee BH, Hahm MG, Cho B (2016) Alloyed 2D metal-semiconductor heterojunctions: origin of interface states reduction and Schottky barrier lowering. Nano Lett 16:5928–5933

Part XIV

Advances in Biomaterials for 3D Printing of Scaffolds and Tissues

Additive Manufacturing of Natural Materials as a Multidisciplinary Approach in Engineering Education Henry A. Colorado, Elkin I. Gutierrez, and Mery Gomez-Marroquin

Abstract This research shows results from additive manufacturing as an important strategy to produce multidisciplinary skills in engineering students. Case studies are presented with materials and mechanical engineering, and arts as well. Two animals and a natural fiber from Colombia were manufactured with the fused deposition modeling (FDM) technique present at the University of Antioquia Museum and in classroom as well, aiming involve the students in a new learning and technological experience. Results reveal the potential of this technology in education and particularly in the motivation for learning and deep comprehension of details of nature only visible at the micro-scale. Moreover, the virtual models for the printing process also open new possibilities after the Covid new challenges for virtual education. Keywords Additive manufacturing. 3D printing · Education · Engineering

Introduction Additive manufacturing (AM) [1], or 3D printing (3DP), is a group of disruptive technologies that has overpassed all the areas of fabrication and impacted all the materials industry [2]. Some of the AM advantages are costs of equipment has been decreasing significantly to be implemented in many processes; since add materials instead of removing them, the waste is too low and therefore is considered green technology; design freedom is in the core of the technology, reducing constrains and allowing rethinking the design to new concepts such as circular economy and H. A. Colorado (B) CCComposites Lab, Universidad de Antioquia UdeA, Calle 70 No. 52-21, Medellín, Colombia e-mail: [email protected] E. I. Gutierrez Mechanical Engineering, Universidad Antonio Nariño, Medellín, Colombia M. Gomez-Marroquin Facultad de Ingeniería Geológica, Minera Y Metalúrgica, Universidad Nacional de Ingeniería, Lima, Perú © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_49

533

534

H. A. Colorado et al.

sustainability; and it is a rapid prototyping technique. AM still has limitations, such as high production costs; low raw materials availability; surface finishing in some technologies; and dimension control. Among the multiple and novel applications of AM are included many products in areas such as energy [3], aerospace [4], construction industry [5], arts [6], medicine [7], and sustainability [8], among others. Materials science have also suffered a revolution, and materials solutions for AM are also quite diverse, which include traditional [9] and engineering ceramics [10], clays [11], polymer [12] and other composites [13], cements [14], concrete [15], plastics [16], and metals [17]. Education has been also greatly benefitted from AM, not only in formal education, but also in museums and other places. The active methodologies such as project and problem based-learning, case studies and others, are now incorporating AM as a tool to produce animal models, machines, or any class of artefacts that aid learning [18], or simply to teach students about 3D printing or to support outreach activities [19]. Some museums have also used AM for connecting people with the collections and artwork, for a more intensive experience that visitors value more, such as manipulating pieces instead of just watching them, which also enables these spaces as new classrooms [20] using active learning [6], and even as support of experts in areas such as restauration [21]. It is clear then that AM presents a huge opportunity not only for the development of materials and products in many areas, but also for teaching and training in important issues for society such as sustainability and circular economy [22–24], areas where many important products have been developed but with lack of education for people implementation. This research therefore shows some efforts carry out at the Universidad de Antioquia, not only involving AM, teaching, and learning in the museum, but also, using the technology for study bioinspired materials such as natural fibers and some insects.

Materials and Experiments The AM technology used in this research is fused deposition modeling (FDM), a technology where a melt extrusion method is used to deposit filaments, typically of thermoplastics materials, following a specific pattern as other 3DP technologies. The material used is polylactic acid (PLA), also known as poly or polylactide, a thermoplastic polyester with formula (C3 H4 O2 )n. Two (2) insects have been manufactured with AM which were selected based on the importance these have for the sustainability of the planet. An ant and a honeybee. The ant was inspired in an art piece made of balsa wood by biologist and artist A. Berrio, 2003, from the Museum of the University of Antioquia MUUA. This ant was scanned in an EinScan-Pro + 3D scanner, with 0.05 mm single shot accuracy. From this scan, a CAD model was obtained and further prepared using Ultimaker CURA slicer to create a g-code file and later printed via FDM. The honeybee was designed completely at the computer, starting from a scratch as a base model using Autodesk Maya and Inventor, but

Additive Manufacturing of Natural Materials as a Multidisciplinary Approach …

535

inspired in images obtained from internet as presented below. Results were also presented with students to know the perception of this method. On the other hand, a natural fiber typical from the Amazons regions of Colombia, Perú, Brazil, Venezuela, and Ecuador were studied in this work: moriche fiber (Mauritia flexuosa), extracted from the palm, and typically used for ancient communities in handicrafts, bags, hammocks, and roofs. The fibers used in this research has been obtained from Puerto Careño, in the Colombian state of Vichada.

Results Figure 1 shows some of the strategies followed by the Museum MUUA for improving teaching and learning of natural and artwork collections. Figure 1a shows local birds made of wood and polymeric materials made by hand by students and artists. Figure 1b shows a skeleton of a humanoid with some graphs, also made by hand. Figure 1c shows some pieces made in balsa wood by a local artist. The first two didactic strategies in general show a lack in quality from the manufacturing point of view, and the last one, as an artwork, it is delicate and not feasible to manipulate by the public, it is fragile, and material has poor durability. In addition, as materials made by hand, the repeatability is not guaranteed, expensive, and difficult. Therefore, although the strategies are innovative and very useful for teaching and learning, they can be further improved with AM and materials science, which allows to use flexible materials, the model can be printed at any time with high repeatability and low costs, and thus, people can manipulate the objects with any risks for the collections. Figure 2 shows the 3D printing model of an ant developed from the scanning of an artwork from Museum MUUA, inspired by the artists in a common ant from Fig. 2a (from pnglux), leading into an artwork in balsa wood in Fig. 2b, later 3D scanned in Fig. 2c, and finally 3D printed using FDM technique in PLA material in Fig. 2d. All the images are important in education, of course the real object is of most significance in terms of information, but clearly many aspects are difficult to observe and understand because of the low scale of the object. The artwork is valuable as it

Fig. 1 Diverse strategies developed at the Museum of University of Antioquia MUUA for education

536

H. A. Colorado et al.

Fig. 2 3D printing model of an ant developed from the scanning of an artwork from Museum MUUA

represents in detail aspects of the ant, but it lacks potential in education as cannot be touched by people, more importantly by children. The 3D virtual file also opens new areas for education, as it allows integrations in new spaces, including virtual education in museums and schools. And the printed object is a versatile instrument that can be used to show people a microscopic animal to play and learn. Figure 3 shows images of the honeybee production, inspired from images taken from internet, and showing parts of the design and final product. The model was printed in parts, main body was divided in two, and the bee wings separated. The finishing is quite acceptable considering the used technique was FDM. Figure 4 shows the additional supporting material necessary to fabricate parts like the ant legs and antennas, which is very useful to teach students about fabrication processes not only in AM but also in general as most of them require some type of postprocessing for the surface finishing. Figure 5 shows the moriche fiber (Mauritia flexuosa) observed at high resolution under the microscope, with the corresponding model after the scanning process, and further 3D printed using FDM. This is another example where AM is very powerful not only from the technical point of view in copy structures to be used in diverse applications, but also in showing and teaching people about the microworld, showing new tools for learning and motivating people for specific subjects such as conservation, culture, and sustainability. Students’ perception of these experiments was registered as very positive, and they were motivated not only to learn about the printed object itself, but also about the materials and manufacturing technology, which can be used to attract people for STEM majors [25], which has been presenting difficulties with the constant decrease

Additive Manufacturing of Natural Materials as a Multidisciplinary Approach …

Fig. 3 3D printing model of a honeybee developed completely using design software

Fig. 4 A detail of the 3D fabrication process showing part of the support material

537

538

H. A. Colorado et al.

Fig. 5 Natural fiber selected under the SEM microscope, its model after a scanning process, and the 3D printing object via FDM

in the admission applications at Colombia and in many countries. In general, students from university value the experience as positive, where is clear many of them can see an interdisciplinarity valuable, regarding the questions and interests in the new technology of fabrication; although the most interesting aspect is for children visiting the museum, where they really had a different approach in try to play with the objects and thus learning became a funny experience, quite important, as has been reported in other areas before for children education [26].

Conclusion This project shows the versatility of the AM technology in education, not only at schools, but also in other environments such as museums, particularly in cultural aspects and also in showing details and characteristics of animals and plants at low scales, where human eye is not enough and also where it is impossible to have a hands on experience, something that is not only winning popularity among students and community but also is becoming in some areas mandatory for people, as it is a more complete form of learning. Unfortunately, still this must be better known for educators, as most of them do not know about the advantage of using AM in their subject of expertise.

Additive Manufacturing of Natural Materials as a Multidisciplinary Approach …

539

References 1. Gebhardt A, Hötter JS (2016) Additive manufacturing: 3D printing for prototyping and manufacturing. Carl Hanser Verlag GmbH Co KG 2. Kumar LJ, Pandey PM, Wimpenny DI (eds) (2019) 3D printing and additive manufacturing technologies, vol 311. Springer, Berlin, Germany 3. Post BK, Richardson B, Lind R, Love LJ, Lloyd P, Kunc V, Jenne D (2017) Big area additive manufacturing application in wind turbine molds. Solid Free Fabr 4. Lyons B (2014) Additive manufacturing in aerospace: examples and research outlook. The Bridge 44(3) 5. Bhardwaj A, Jones SZ, Kalantar N, Pei Z, Vickers J, Wangler T, Zou N (2019) Additive manufacturing processes for infrastructure construction: a review. J Manuf Sci Eng 141(9):091010 6. Colorado HA, Mendoza DE, Valencia FL (2021) A combined strategy of additive manufacturing to support multidisciplinary education in arts, biology, and engineering. J Sci Educ Technol 30(1):58–73 7. Javaid M, Haleem A (2018) Additive manufacturing applications in medical cases: a literature based review. Alex J Med 54(4):411–422 8. Colorado HA, Velásquez EIG, Monteiro SN (2020) Sustainability of additive manufacturing: the circular economy of materials and environmental perspectives. J Market Res 9(4):8221– 8234 9. Ordoñez E, Gallego JM, Colorado HA (2019) 3D printing via the direct ink writing technique of ceramic pastes from typical formulations used in traditional ceramics industry. Appl Clay Sci 182:105285 10. Zocca A, Colombo P, Gomes CM, Günster J (2015) Additive manufacturing of ceramics: issues, potentialities, and opportunities. J Am Ceram Soc 98(7):1983–2001 11. Revelo CF, Colorado HA (2018) 3D printing of kaolinite clay ceramics using the direct ink writing (DIW) technique. Ceram Int 44(5):5673–5682 12. Restrepo JJ, Colorado HA (2020) Additive manufacturing of composites made of epoxy resin with magnetite particles fabricated with the direct ink writing technique. J Compos Mater 54(5):647–657 13. El Moumen A, Tarfaoui M, Lafdi K (2019) Additive manufacturing of polymer composites: processing and modeling approaches. Compos B Eng 171:166–182 14. Vergara LA, Colorado HA (2020) Additive manufacturing of Portland cement pastes with additions of kaolin, superplastificant and calcium carbonate. Constr Build Mater 248:118669 15. Bos F, Wolfs R, Ahmed Z, Salet T (2016) Additive manufacturing of concrete in construction: potentials and challenges of 3D concrete printing. Virtual Phys Prototyp 11(3):209–225 16. Cicala G, Latteri A, Del Curto B, Lo Russo A, Recca G, Farè S (2017) Engineering thermoplastics for additive manufacturing: a critical perspective with experimental evidence to support functional applications. J Appl Biomater Funct Mater 15(1):10–18 17. Yang L, Hsu K, Baughman B, Godfrey D, Medina F, Menon M, Wiener S (2017) Additive manufacturing of metals: the technology, materials, design and production. Springer, Cham, pp 45–61 18. McMenamin PG, Quayle MR, McHenry CR, Adams JW (2014) The production of anatomical teaching resources using three-dimensional (3D) printing technology. Anat Sci Educ 7(6):479– 486 19. Ford S, Minshall T (2019) Where and how 3D printing is used in teaching and education 20. Hancock M (2015) Museums and 3D printing: more than a workshop novelty, connecting to collections and the classroom. Bull Assoc Inf Sci Technol 42(1):32–35 21. Short DB (2015) Use of 3D printing by museums: educational exhibits, artifact education, and artifact restoration. 3D Print Addit Manuf 2(4):209–215 22. Colorado HA, Echeverri-Lopera GI (2020) The solid waste in Colombia analyzed via gross domestic product: towards a sustainable economy. Revista Facultad de Ingeniería Universidad de Antioquia 96:51–63

540

H. A. Colorado et al.

23. de Azevedo AR, Teixeira Marvila M, Barbosa de Oliveira L, Macario Ferreira W, Colorado H, Rainho Teixeira S, Mauricio Fontes Vieira C (2021) Circular economy and durability in geopolymers ceramics pieces obtained from glass polishing waste. Int J Appl Ceram Technol 24. Marvila MT, de Azevedo AR, Alexandre J, Colorado H, Pereira Antunes ML, Vieira CM (2021) Circular economy in cementitious ceramics: replacement of hydrated lime with a stoichiometric balanced combination of clay and marble waste. Int J Appl Ceram Technol 18(1):192–202 25. Cheng L, Antonenko PD, Ritzhaupt AD, Dawson K, Miller D, MacFadden BJ, Ziegler M (2020) Exploring the influence of teachers’ beliefs and 3D printing integrated STEM instruction on students’ STEM motivation. Comput Educ 158:103983 26. Farrell G (1997) Thinking, saying, playing: children learning the tabla. Bull Counc Res Music Educ 14–19

Part XV

Advances in Multi-principal Elements Alloys X

Development of a High Entropy Alloy AlX (CoCrCuFeNi)1-X for Diverse Security Applications D. Butcher, J. C. T. Cullen, N. Barron, S. Mehraban, M. Calvo-Dahlborg, S. G. R. Brown, and N. P. Lavery

Abstract High Entropy Alloys are often associated with good corrosion resistance and high hardness properties. These are of interest in demanding and premium high-security applications where high hardness (>700HV) and high tensile strength (>1200 MPa) materials are required. In transport applications lightweight materials ( 1.88). HCP phases have also been noted to exist in Domain I and III if rmean < 1.365 and > 1.387, respectively. Another atomic parameter is δ which is a measure of the atomic radii mismatch and is given by Eq. (1), [14].   N N   ci (1 − r i /( ci r i ))2 δ= i=1

(1)

i1

The purpose of this paper is to investigate the suitability of the Alx (CoCrCuFeNi)1-x alloy for diverse security applications requiring high hardness (>700HV) and high tensile strength (>1200 MPa). Tensile behaviour was prioritised, looking for the best properties possible from the 5-element HEA system, and assessed using punch disk testing to rapidly identify trends in tensile ductility with varying alloy composition.

Materials and Methodology Each composition was prepared by weighing out elementals in solid form and melting using induction heating under an argon (>99.9% pure) atmosphere. Each sample weighed approximately 20 g and was held for 5 min in the molten state to ensure homogeneity. The samples were solidified as cylinders (15 mm diameter). After sectioning, some material was retained for SEM, XRD, and hardness evaluation, and some material was processed into an 8 mm diameter cylinder which was sectioned into small punch specimens 8 mm diameter × 0.5 mm thick. A Wilson VH3300 Automatic Hardness Tester was used to determine the Vickers Hardness (HV) of each alloy composition. Using a Vickers diamond indenter with a force of 1 kgf, an array of 5 × 2 indents was produced and measured at 50 × magnification. A Brucker D8 Discovery X-Ray Diffraction (XRD) with a Cu Kα source was used to examine crystal structures found in each alloy. The compositional analysis was performed on a JEOL JSM-6010 Secondary Electron Microscopy with Energy Dispersive X-ray Spectrometry SEM–EDS. The chemistries were verified using EDS and closely agreed with the defined target chemistries shown in Table 1. Small punch testing was selected for Mechanical Properties testing as it only requires small quantities of different alloy variants to be made. The small punch test allowed multiple tests to be performed per 20 g casting. The tests were conducted on an in-house test rig manufactured in accordance with European Standards for small punch testing of metallic materials [15] using a Tinius Olsen H25KS tensile test machine. The disc is placed in the test section where it is held above a hole 4 mm in diameter. A hemispherical punch with a 1.25 mm radius is then pressed through the disc at a rate of 0.5 mm/min whilst the displacement of the opposing face of the disc is recorded. To evaluate the results of the test the maximum force and the failure displacement were chosen as indicators of ductility and strength. Each was

546

D. Butcher et al.

divided by specimen thickness to account for any variation in thickness between tests. A regression analysis was used to get an empirical relationship between the test parameters and the Ultimate Tensile Strength given in Eq. (2) [16]: σU T S =

LU t (0.14D − 0.82Cl + 2.17d F + 0.6

(2)

where: L U is the ultimate load, t is the thickness, D is punch diameter, d F is the failure displacement, and Cl is the die clearance. 316L stainless steel was selected as a benchmark for a high ductility steel (with a failure elongation >60%) and 17-4PH was selected as a high strength steel comparator (UTS ~1300 MPa). Two types of alloys were investigated (i) ‘equimolar’ Alx (CoCrCuFeNi)1-x alloys with different Al additions and (ii) non-equimolar alloys (HEA12 to HEA17). The non-equimolar alloys arose directly from a search of multiple possible combinations of elements screened by the Hume-Rothery approach [4, 10] aimed at identifying systems with same elements consisting of FCC phases only.

Results XRD Analysis Figure 1a shows the XRD scan for x = 0–3. When x = 0, two FCC phase crystal plains (111) were detected at 43.5 and 43.8°, in agreement with literature [8]. The two FCC phase peaks overlap at x = 0.45. The formation of the first BCC phase was observed to occur when x = 0.75 and the second BCC phase appeared at x = 1.0, in agreement with Tong et al. [7]. The alloy system becomes BCC fully when x = 3. Figure 1 (b) shows the XRD scans for samples 12 to 17 which have FCC structure. The measured phases are in coherence with the predicted domains as presented in Table 1 and can be seen in Fig. 2, where alloy domains on a e/a versus atomic radius diagram are compared between the current work on to a AlxCoCrFeNi system, Kao

Fig. 1 a XRD scans of Alx (CoCrCuFeNi)1-x where x = 0.0–3.0. b XRD scans of samples HEA12 to HEA17

Ni

Cu

6.0

HEA 17

37.5

HEA11 (x = 3.00)

6.0

33.3

HEA10 (x = 2.50)

7.0

28.6

HEA09 (x = 2.00)

HEA 16

23.1

HEA08 (x = 1.50)

HEA 15

19.4

HEA07 (x = 1.20)

7.0

18.0

HEA06 (x = 1.10)

HEA 14

16.7

HEA05 (x = 1.00)

6.5

15.3

HEA04 (x = 0.90)

6.5

13.0

HEA03 (x = 0.75)

HEA 13

8.3

HEA 12

0.0

HEA02 (x = 0.45)

20.0

23.0

23.0

23.0

23.0

23.0

12.5

13.3

14.3

15.4

16.1

16.4

16.7

16.9

17.4

18.3

20.0

34.0

31.0

33.0

33.5

31.5

34.5

12.5

13.3

14.3

15.4

16.1

16.4

16.7

16.9

17.4

18.3

20.0

5.0

5.0

5.0

5.0

8.0

5.0

12.5

13.3

14.3

15.4

16.1

16.4

16.7

16.9

17.4

18.3

20.0

29.0

29.0

26.0

26.0

26.0

26.0

12.5

13.3

14.3

15.4

16.1

16.4

16.7

16.9

17.4

18.3

20.0

6.0

6.0

6.0

5.5

5.0

5.0

12.5

13.3

14.3

15.4

16.1

16.4

16.7

16.9

17.4

18.3

20.0

1.510

1.480

1.520

1.525

1.525

1.525

2.000

1.933

1.857

1.769

1.710

1.689

1.667

1.644

1.609

1.532

1.400

1.371

1.371

1.371

1.371

1.371

1.372

1.325

1.330

1.336

1.342

1.342

1.347

1.348

1.350

1.352

1.354

1.360

Rad

2.83

2.83

2.97

2.97

2.90

4.67

4.62

4.51

4.35

4.09

3.88

3.80

3.70

3.60

3.43

2.98

1.79

Delta

Alloy predicted properties Co

E/A

Fe

Al

Cr

at.%

HEA01 (x = 0.00)

Composition Alx (CoCrCuFeNi)1-x

Table 1 Alloy compositions and predicted properties

7.83

7.81

7.73

7.72

7.77

7.74

6.24

6.47

6.74

7.05

7.26

7.34

7.41

7.49

7.62

7.89

8.36

ρ (g/cc)

I (FCC)

I (FCC)

I (FCC)

I (FCC)

I (FCC)

I (FCC)

III (BCC)

III (BCC)

II (FCC + BCC)

II (FCC + BCC)

II (FCC + BCC)

II (FCC + BCC)

II (FCC + BCC)

II (FCC + BCC)

II (FCC + BCC)

I (FCC)

I (FCC)

Domain

100

100

100

100

100

100

0

7.4

20.2

20

49.9

45.8

61.3

81.7

78.2

100

100

0

0

0

0

0

0

100

92.6

79.8

80

50.1

54.2

38.7

18.3

21.8

0

0

BCC

Measured XRD FCC

Development of a High Entropy Alloy AlX (CoCrCuFeNi)1-X … 547

548

D. Butcher et al.

Fig. 2 Plot showing alloy domains on a e/a versus atomic radius diagram for the current work on Alx (CoCrCuFeNi)1-x in comparison with a Alx CoCrFeNi system, Kao et al. [17], and a Alx CrCuFeNi2 system, Guo et al. [18]. The labels are measured phases

et al. [17], and a AlxCrCuFeNi2 system, Guo et al. [18]. The labels are measured phases, giving a good match to the three domains theory (I-FCC, II-Mixed phases, and III-BCC), particularly for the system studied in this work.

SEM–EDS Analysis In order of increasing molar content of aluminium, the following can be observed from the SEM analysis, and generally correlate with the observations of [10] for the 6-element system on Alx (CoCrCuFeNi)1-x , and with [19] for the 5-element system Alx (CoCrFeNi)1-x . • (x = 0) With no aluminium, the microstructure consists of Cu rich dendrites in a Fe–Cr–Co rich matrix. • (x = 0.75) With increasing Al an Al-rich BCC phase appears next to the Cu rich dendrites Fig. 3a. • (x = 1.0) Further increasing Al promotes the FCC matrix to become a side plate structure Fig. 3b matching the observation of Wang et al. [19]. The Al-rich phase is observed to have a spinodal deposition dendritic structure [10]. • (x > 1.5) At higher Al additions, Al appears throughout the matrix with the FCC being replaced by BCC Fig. 3c. The Cu rich phase is the only remaining FCC phase. The second BCC phase (Al-Cu rich) forms around the FCC phase [10]. • (x = 3) At the highest Al level, the Cu rich phase has disappeared leaving only the Al-Cu rich phase and a high Al content matrix and the alloy is now fully BCC. The non-equimolar AlCoCrCuFeNi samples were found to contain dendritic grain structures shown in (Fig. 4).

Development of a High Entropy Alloy AlX (CoCrCuFeNi)1-X …

549

Fig. 3 Microstructural analysis of with Al molar contents of a x = 0, b x = 0.75, c x = 1 and d x = 1.5

Hardness The hardness testing of the Alx (CoCrCuFeNi)1-x ‘equimolar’ system from x = 0 to 3 shows a strong correlation between the aluminium content and the hardness of the alloy. The measured hardness data is entirely consistent with reported data [5] as shown in Fig. 5 which adds confidence to the specimen production methods used. In contrast, the hardness of the six non-equimolar alloys selected using the HumeRothery approach [12] did not vary dramatically with a range of only 20 HV, from 139 to 159 HV (Table 2).

Small Punch Disk Testing The small punch test showed that in the ‘equimolar’ Alx (CoCrCuFeNi)1-x system, for compositions varying from x = 0 to 3, there is a rapid drop in ductility with increasing Al content. Except for the 0.45Al composition, none of these ‘equimolar’ alloys were comparable with the properties of the 316L or 17-4PH (Figs. 6 and 7). However, for the non-equimolar alloys strength values close to 316L were found and higher ductility (Figs. 6 and 8).

550

D. Butcher et al.

Fig. 4 Cr distribution (top) and Fe distribution (bottom), for HEA12 (left) and HEA17 (right)

Fig. 5 Hardness of Alx(CoCrCuFeNi)1-x compared to Yeh et al. [5] Table 2 Hardness of non-equimolar AlCoCrCuFeNi HEAs HEA

HEA 12

HEA 13

HEA 14

HEA 15

HEA 16

HEA 17

Hardness (HV)

147

155

159

159

153

139

SD

3

2

5

10

5

5

Development of a High Entropy Alloy AlX (CoCrCuFeNi)1-X …

551

Discussion Eleven samples were produced to investigate the effect of Al content on tensile properties in the Alx (CoCrCuFeNi)1-x alloy system, with Al varied from x = 0 to x = 3 mol while the CoCrCuFeNi additions were kept equimolar. Microstructures and phase samples closely matched previous work [5], with observed and measured phases agreeing with predictions from an empirical domain model [4, 10]. The small punch test showed a transition from ductile to brittle behaviour when Al content was increased from x = 0.45 to x = 0.75, as might be expected from the empirical model showing a move from domain I (FCC) and domain II (mixed FCC + BCC), as shown in Fig. 2. It can be seen in Fig. 6 that several alloys are very brittle (e.g. 1.1Al) and alloys that were more ductile had relatively low strengths (e.g. 0.75Al) compared to the two steel benchmark alloys. The exception to this was alloy 0.45Al. This rapid transition from domain I (FCC) to domain II (FCC + BCC) is also matched with an expected increase in hardness, with values closely match those from previous work by Yeh et al. [5]. The Hume-Rothery approach [4, 10] was also used to provide six potential compositions (HEA12 to HEA17) each predicted to contain FCC phases only, and these are shown in Fig. 2. The hardness values for these alloys were found to be 149 ± 10 HV in agreement with the model [10]. In agreement with model predictions these alloys all consisted of only FCC phases. It is noticeable from Fig. 8 that these alloys consistently display higher strength and ductility properties compared to the ‘equimolar’ group. Indeed, they are also more ductile than the ‘high ductility’ benchmark 316L steel, using the biaxial punch disk test, with what looks to be an equivalent upper tensile strength, even though the hardness is significantly lower. This high ductility

Fig. 6 Summary of small punch test

552

D. Butcher et al.

Fig. 7 Test data for equimolar Alx (CoCrCuFeNi)1-x compared to 316 stainless steel and 17-4PH

Fig. 8 Small punch test data for non-equimolar AlCoCrCuFeNi compared to 316 stainless steel and 17-4PH

will probably give high elongation even in comparison with 316L but uniaxial tensile testing will be needed to verify this and also the Yield and Upper Tensile Strengths. From the perspective of the intended application, no single system has yet been found which jointly meets the required high hardness, tensile strength, and ductility, for the equimolar AlCoCrCuFeNi HEA. While it seems difficult to overcome the brittle behaviour without rapidly softening the material, it is possible to find nonequimolar combinations which result in more ductile alloys (especially HEA12 to HEA17) which all possess low values of the delta parameter (Table 1). This calculated parameter could provide a potential method to further filter HEAs via relative ductility in each phase Domain, [20].

Development of a High Entropy Alloy AlX (CoCrCuFeNi)1-X …

553

Conclusions • Alloys produced through a rapid alloy prototyping method matched reported literature in terms of hardness properties and microstructures observed. • The observed crystal structures of these alloys also matched the Hume-Rothery approach predictions [4]. • Using the model, it was possible to predict some non-equimolar alloy compositions which should consist of FCC phases only, and this was verified in the experiments. • Several variants of the AlCoCrCuFeNi system were found possessing comparable properties to two benchmark steel grades (316L and 17-4PH) in terms of strength but particularly in terms of ductility. • The ductility of the HEAs produced showed an inverse type relationship to the delta parameter. Acknowledgements The authors would like to thank Zeal and the UK Research and Innovation for the UK Smart Grant 2019-46371 which made this work possible. The authors would also like to thank the Welsh Government, European Regional Development Fund (ERDF), and SMART Expertise Wales for funding the Materials Advanced Characterisation Centre (MACH1) where the work was carried out, and specifically for the Smart Expertise Funded project COMET (Combinatorial Metallurgy).

References 1. Cantor B, Chang ITH, Knight P, Vincent AJB (2004) Microstructural development in equiatomic multicomponent alloys. Mater Sci Eng A 375–377(1–2):213–218. https://doi.org/ 10.1016/j.msea.2003.10.257 2. Yeh JW et al (2004) Nanostructured high-entropy alloys with multiple principal elements: Novel alloy design concepts and outcomes. Adv Eng Mater 6(5):299–303. https://doi.org/10. 1002/adem.200300567 3. Steurer W (2020) Single-phase high-entropy alloys—a critical update. Mater Charact 162:110179. https://doi.org/10.1016/j.matchar.2020.110179 4. Calvo-Dahlborg M et al (2021) “Prediction of phase, hardness and density of high entropy alloys based on their electronic structure and average radius. J Alloys Compd 865. https://doi. org/10.1016/j.jallcom.2021.158799 5. Tsai MH, Yeh JW (2014) High-entropy alloys: a critical review. Mater Res Lett 2(3):107–123. https://doi.org/10.1080/21663831.2014.912690 6. Kuznetsov AV, Shaysultanov DG, Stepanov ND, Salishchev GA, Senkov ON (2012) Tensile properties of an AlCrCuNiFeCo high-entropy alloy in as-cast and wrought conditions. Mater Sci Eng A 533:107–118. https://doi.org/10.1016/j.msea.2011.11.045 7. Manzoni AM et al (2016) On the path to optimizing the AlCoCrCuFeNiTi high entropy alloy family for high temperature applications. Entropy 18(4). https://doi.org/10.3390/e18040104 8. Daoud HM, Manzoni A, Völkl R, Wanderka N, Glatzel U (2013) Microstructure and tensile behavior of Al8Co17Cr 17Cu8Fe17Ni33 (at.%) high-entropy alloy. JOM 65(12):1805–1814. https://doi.org/10.1007/s11837-013-0756-3

554

D. Butcher et al.

9. Daoud HM, Manzoni AM, Wanderka N, Glatzel U (2015) High-temperature tensile strength of Al10Co25Cr8Fe15Ni36Ti6 compositionally complex alloy (high-entropy alloy). JOM 67(10):2271–2277. https://doi.org/10.1007/s11837-015-1484-7 10. Tong CJ et al (2005) Microstructure characterization of AlxCoCrCuFeNi high-entropy alloy system with multiprincipal elements. Metall Mater Trans A 36(4):881–893. https://doi.org/10. 1007/s11661-005-0283-0 11. Wu B et al (2018) Microstructures and thermodynamic properties of high-entropy alloys CoCrCuFeNi. Intermetallics 93:40–46. https://doi.org/10.1016/j.intermet.2017.10.018 12. Calvo-Dahlborg M, Brown SGR (2017) Hume-Rothery for HEA classification and selforganizing map for phases and properties prediction. J Alloys Compd 724:353–364. https:// doi.org/10.1016/J.JALLCOM.2017.07.074 13. Calvo-Dahlborg M, Dahlborg U, Brown SGR, Juraszek J (2020) Influence of the electronic polymorphism of Ni on the classification and design of high entropy alloys. J Alloys Compd 824. https://doi.org/10.1016/j.jallcom.2020.153895 14. Miracle DB, Miller JD, Senkov ON, Woodward C, Uchic MD, Tiley J (2014) Exploration and development of high entropy alloys for structural applications. Entropy 16(1):494–525. https:// doi.org/10.3390/e16010494 15. Bruchhausen M et al (2018) European standard on small punch testing of metallic materials. Ubiquity Proc 1(S1):11. https://doi.org/10.5334/uproc.11 16. Norris SD, Parker JD (1996) Deformation processes during disc bend loading. Mater Sci Technol 12(2):163–170. https://doi.org/10.1179/mst.1996.12.2.163 17. Kao YF, Chen TJ, Chen SK, Yeh JW (2009) Microstructure and mechanical property of as-cast, -homogenized, and -deformed AlxCoCrFeNi (0 ≤ x ≤ 2) high-entropy alloys. J Alloys Compd 488(1):57–64. https://doi.org/10.1016/j.jallcom.2009.08.090 18. Guo S, Ng C, Liu CT (2013) Anomalous solidification microstructures in Co-free Al xCrCuFeNi2 high-entropy alloys. J Alloys Compd 557:77–81. https://doi.org/10.1016/j.jal lcom.2013.01.007 19. Wang WR, Wang WL, Yeh JW (2014) Phases, microstructure and mechanical properties of AlxCoCrFeNi high-entropy alloys at elevated temperatures. J Alloys Compd 589:143–152. https://doi.org/10.1016/j.jallcom.2013.11.084 20. Keil T, Utt D, Bruder E, Stukowski A, Albe K, Durst K (2021) Solid solution hardening in CrMnFeCoNi-based high entropy alloy systems studied by a combinatorial approach. J Mater Res 36(12):2558–2570. https://doi.org/10.1557/s43578-021-00205-6

Part XVI

Advances in Powder and Ceramic Materials Science

Catalytic Pyrolysis of Polyethylene and Polypropylene Over Y Zeolite Xunrui Wang, Chengdong Wang, Xiang Wang, and Jinhong Li

Abstract Polyethylene (PE) and polypropylene (PP) are typical plastic waste. At present, there are mainly four methods of treatment: landfill, incineration, recycling, and thermal pyrolysis. Compared with thermal cracking, the advantages of catalytic cracking are mainly reflected in lower reaction temperature, faster reaction rate, and higher yield of pyrolysis target products. In this study, HDPE/PP was catalyzed by three kinds of catalysts. The high-temperature treatment and acid–alkali leaching were used to change textural properties and acid sites attribution of zeolite. It was found that the acidity density affected the gas and liquid yield, but the existing form of coke deposits changed due to the difference in adsorption. Besides, the pore structure has an obvious effect on acidity retention and catalytic cycling stability. Compared with pore structure, the distribution of acidity obviously has a larger impact on the distribution of product yield. Keywords Waste plastic · Pore structure · Catalytic · Pyrolysis

Introduction According to the different molecular compositions and structures, plastic components can be divided into the following categories: polyolefin (polyethylene PE, polypropylene PP), polyvinyl chloride (PVC), polystyrene (PS), and poly terephthalic acid ethylene glycol (PET). In the production of waste plastics, polyolefins account for more than 60% of the total [1–3]. X. Wang · C. Wang · J. Li (B) Beijing Key Laboratory of Materials Utilization of Nonmetallic Minerals and Solid Wastes, National Laboratory of Mineral Materials, School of Materials Science and Technology, China University of Geosciences, Beijing 100083, People’s Republic of China e-mail: [email protected] X. Wang State Key Laboratory of Multiphase Complex Systems, Institute of Process Engineering, Chinese Academy of Sciences, Beijing 100190, People’s Republic of China © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_51

557

558

X. Wang et al.

The landfill method has a short operation process and low treatment cost, and it is currently one of the main methods for solid waste treatment [4]. However, the good chemical stability of plastic makes it difficult to degrade for a long time after being landfilled, and the land is seriously polluted, and it may even enter the ecological food chain, seriously threatening the good operation of the ecological environment. In addition, landfill occupies a large amount of land, which is undoubtedly a waste of land resources [5]. The incineration method can reduce the amount of plastic waste and generate a lot of heat energy, and the heat recovery efficiency is about 60% [6]. However, due to the large amount of waste gas produced by incomplete combustion of macromolecules, such as CO, HCN, HCl, and NOX, it will cause air pollution [7]. Using this method to treat solid waste requires a set of advanced equipment to deal with the pollution caused by these gases, and the decomposed fly ash still contains microplastics, which will adsorb a large amount of heavy metal elements and still need to be further processed [8]. The mechanical recycling method [9, 10] requires special personnel to sort the garbage, and the plastic obtained after mechanical recycling and reprocessing is difficult to restore the previous performance [11]. The chemical recycling method degrades waste plastics and high-molecularweight polymers into reusable low-molecular-weight compounds [12, 13]. The C– C bond and C-H bond in the polymer molecule are broken at the higher temperature (350–900 °C) of waste plastics. The cracking of waste plastics includes thermal cracking, catalytic cracking, and stepwise cracking. Compared with thermal cracking, the advantages of catalytic cracking include lowering the cracking temperature of waste plastics; accelerating the rate of cracking reaction; and selectively increasing the yield of cracking target products. There are many kinds of catalysts used for cracking plastics. At present, most researchers hope to obtain zeolite catalysts with better effects in the cracking of polyolefins. Therefore, scientists have carried out different work to determine the influence of zeolite pore structure on the cracking effect of polyolefins [14]. Adjusting the structure and acidity of the catalyst was regarded as a great method to control product distribution [15, 16]. The effect of the porous structure of zeolite on the reaction is realized by the reaction and diffusion of intermediates in the pores [17].

Experimental Material Hydrochloric acid (HCl, 37.2 wt.%, Analytical reagent (A.R.)) and sodium hydroxide (NaOH, > 96 wt.%, A.R.) were purchased from Beijing Chemical Reagent Co., Ltd. China, ammonium phosphate ((NH4 )3 PO·3H2 O, > 98 wt.%, A.R.) was bought by Tianjin Guangfu Fine Chemical Research Institute. The Na-Y faujasite (zeolite) was purchased from Yuanli Chemical Engineering Company, Tianjin, China. HDPE and PP were purchased from SINOPEC Beijing Yanshan Company.

Catalytic Pyrolysis of Polyethylene and Polypropylene …

559

Preparation of Modified Zeolite First, put the Na-Y zeolite in a water bath of ammonium phosphate solution for one hour at a temperature of 50 °C, and then filter and dry with deionized water. After repeated three times, an ion-exchanged zeolite (denoted as HY) is obtained. The sample was then placed in a hydrochloric acid solution in a water bath at 80 °C for 3 h, filtered and washed. After completion, the collected sample was dried at 90 °C for 12 h. The sample name was recorded as DA. Then put the sample in a sodium hydroxide solution in a water bath at 65 °C for one hour. After filtering and washing, the sample was collected and dried at 90 °C for 12 h. The sample was denoted as DY.

Catalytic Pyrolysis of HDPE and PP Three different reactants (HDPP, HDPE, and mixing the two together at a ratio of 1:1) are thermally cracked until they are complete, and different rates of carrier gas are applied to the reaction site to observe the reaction products. Separately mix HDPE and catalyst at a mass ratio of 10:1 for catalytic cracking until complete; mix HDPP and catalyst at a mass ratio of 10:1 for catalytic cracking until complete; and catalytically crack HDPP with no insulation cotton (method same as above), heated in the power of 500 W for 1 h.

Results and Discussion Pyrolysis Product Distribution Three different reactants (HDPP, HDPE and mixing the two in a ratio of 1:1) are directly thermally cracked without using a catalyst until the reaction is complete, and different rates of carrier gas are applied to the reaction sites. After cooling, the products were collected and the yields were calculated separately. The ratio is shown in Table 1. Different types of polyolefins undergo thermal cracking at different N2 flow rates, and the product composition varies greatly. HDPP and HDPP/PE mixed samples have good pyrolysis conversion rate and remain stable (the sum of gas product and liquid product yield), and the conversion rate can reach 100% at various flow rates. However, the pyrolysis conversion rate of HDPE is worse than that of the first two samples, and the pyrolysis is incomplete at a lower flow rate. Mix different zeolite catalysts and catalysts in different mass ratios and put them into the heating reaction device, and pass N2 . After purging for 20 min, start the heating jacket, and the heating power is constant at 500 W. After cooling, record the mass of remaining solids and produced fluid, and calculate the yield of each phase after collecting the produced fluid and mixed gas.

560

X. Wang et al.

Table 1 Proportion of products produced by thermal cracking of polyolefin Type of reactant

N2 flow rate (mL/min)

Liquid (%)

Gas (%)

Solid (%)

HDPP

60

86.2

13.8

0

80

85.3

14.7

0

100

83.9

16.1

0

60

85

15

0

80

82.1

17.9

0

100

78.3

21.7

0

60

81.9

13.6

4.5

80

77.4

16.4

6.2

100

84.4

15.6

0

HDPP/HDPE

HDPE

Figure 1a shows the results of different treatment to crack HDPP. The first and second on the left are the liquid products obtained by catalyzing HDPP by DY zeolite, the third and fourth on the left are the liquid products obtained by catalyzing HDPP by HY zeolite, and the first on the right is the liquid product obtained by pyrolyzing HDPP. Catalytic cracking of HDPP was carried out with constant power under atmospheric pressure. The experimental temperature range was approximately 380–430 °C, and compared with thermal cracking, the temperature has dropped by 40 °C. Table 2 is obtained by calculating the yield under the N2 flow rate of 150 ml/min. It is not difficult to see that the three treatment methods have good

Fig. 1 a Different treatments to crack HDPP. b Different treatments to crack HDPE. c Cracking HDPP without thermal insulation cotton. d Liquid products obtained from the catalytic cracking of HDPE with different treatments

Catalytic Pyrolysis of Polyethylene and Polypropylene …

561

Table 2 Proportion of products produced by cracking HDPP Processing method

N2 flow rate (mL/min)

Liquid (%)

Gas (%)

Without catalyst

150

82.1

17.9

DY catalyst

70.8

30.2

HY catalyst

73.4

28.6

catalytic pyrolysis conversion rates for HDPP. Compared with the three catalysts, the modified zeolite has better selectivity to volatile products. Among them, DY is the best, which can reach 30.56%, and with the increase of gas phase products, the color of pyrolysis products gradually becomes lighter. Figure 1b, d show the cracking results of HDPE by different treatment methods. The first on the left is the liquid product obtained by catalyzing HDPE with DY zeolite, the second and third on the left are the liquid products obtained by reusing the catalyst, and the first on the right is the liquid product obtained by the thermal cracking of HDPE. After the modified zeolite is catalytically cracked HDPE, the catalytic conversion rate of the obtained product. It is better than thermal cracking. Compared with thermal cracking, HDPE modified zeolite catalytic cracking products have higher catalytic conversion rate and better absorption of volatile products. The result of the obtained product is consistent with Table 4. With the increase of the number of catalytic cracking cycles and the increase of the conversion rate, the liquid phase products catalyzed by DY zeolite gradually increase, and DA zeolite is also consistent with it. Figure 1c shows the different catalytic methods of cracking HDPP without thermal insulation cotton. The first on the left is the liquid product obtained by thermal cracking of HDPP, the second and third on the left are the liquid products obtained by catalyzing HDPP by DY zeolite, and the first on the right is the solid wax obtained by pyrolyzing HDPP. The result of the obtained product is consistent with Table 3. The solid-phase residue rate of HDPP pyrolyzed without thermal insulation cotton is as high as 46.8%; therefore, the main body obtained by cracking HDPP is solid wax. After adding DY zeolite catalysis, the catalytic conversion rate has been greatly improved, and the solid-phase residues of different catalysts and polyolefin ratios are basically the same, but as the proportion of polyolefin increases, the liquid phase decreases and the volatile products increase. Table 3 Proportion of products produced by cracking HDPP without thermal insulation cotton Processing method

N2 flow rate (mL/min)

Liquid (%)

Gas (%)

Solid (%)

Without catalyst

60

41.2

12.0

46.8

DY catalyst (1:20)

47.0

28.5

24.5

DY catalyst (1:10)

50.9

24.8

24.3

562

X. Wang et al.

Table 4 Proportion of products produced by catalytic cracking of HDPE N2 flow rate (mL/min)

Liquid (%)

Gas (%)

Solid (%)

60

74.1

23

2.9

DY catalyst -2

80

18.3

1.7

DY catalyst -3

82.3

17.4

0.3

DA catalyst -1

74.2

22.9

2.9

DA catalyst -2

81.6

18.4

0

DA catalyst -3

84.7

15.3

0

DY catalyst -1

Conclusion In this study, three different reactants (HDPP, HDPE, and a mixture of the two in a 1:1 ratio) are catalyzed by three catalysts. It was found that without adding any catalyst, the conversion rate of HDPE was significantly lower than the thermal cracking of the other two different types of polyolefins; at the same time, whether to add insulation cotton in the catalytic cracking has a greater impact on the catalytic efficiency. After adding modified zeolite, the catalytic conversion rate of polyolefin is significantly improved, and as the number of catalytic cycles increases, the conversion rate also increases.

References 1. Canopoli L, Coulon F, Wagland ST (2020) Degradation of excavated polyethylene and polypropylene waste from landfill. Sci Total Environ 698:134125.1–134125.8 2. Achillas DS, Roupakias C, Megalokonomos P et al (2007) Chemical recycling of plastic wastes made from polyethylene (LDPE and HDPE) and polypropylene (PP). J Hazard Mater 149(3):536–542 3. Kiran Ciliz N, Ekinci E, Snape CE (2004) Pyrolysis of virgin and waste polypropylene and its mixtures with waste polyethylene and polystyrene. Waste Manage 24(2):173–181 4. Liu J et al (2006) Assessment of the hazards of domestic plastic waste and its management policies. J Environ Health 23(4):348–50 5. Zhou C, Fang W, Xu W et al (2014) Characteristics and the recovery potential of plastic wastes obtained from landfill mining. J Clean Prod 80(10):80–86 6. Aguado J, Serrano DP, Escola JM (2008) Fuels from waste plastics by thermal and catalytic processes: a review. Ind Eng Chem Res 47(21):7982–7992 7. Yang Z, Fan L, Zhang H et al (2020) Is incineration the terminator of plastics and microplastics?. J Hazard Mater, 123429 8. Shen M, Hu T, Huang W et al (2021) Can incineration completely eliminate plastic wastes? An investigation of microplastics and heavy metals in the bottom ash and fly ash from an incineration plant. Sci Total Environ 779:146528 9. Jeswani H, Krüger C, Russ M et al (2021) Life cycle environmental impacts of chemical recycling via pyrolysis of mixed plastic waste in comparison with mechanical recycling and energy recovery. Sci Total Environ 769(1):144483 10. Soto JM et al (2018) A real case study of mechanical recycling as an alternative for managing of polyethylene plastic film presented in mixed municipal solid waste. J Cleaner Prod 203:777–87

Catalytic Pyrolysis of Polyethylene and Polypropylene …

563

11. Ragaert K, Delva L, Van Geem K (2017) Mechanical and chemical recycling of solid plastic waste. Waste Manag 69:24–58 12. Goto M (2009) Chemical recycling of plastics using sub- and supercritical fluids. J Supercrit Fluids 47(3):500–507 13. Oliveux G, Bailleul JL, Salle E (2012) Chemical recycling of glass fibre reinforced composites using subcritical water. Compos A Appl Sci Manuf 43(11):1809–1818 14. Songip AR, Masuda T, Kuwahara H et al (1993) Test to screen catalysts for reforming heavy oil from waste plastics. Appl Catalysis b-Environ 2(2–3):153–164 15. Almustapha N, Farooq M, Mohammed ML et al (2019) Modification of acidic and textural properties of a sulphated zirconia catalyst for efficient conversion of high-density polyethylene into liquid fuel. Environ Sci Pollut Res 27(11) 16. Zhao Y, Wang W, Jing X et al (2020) Catalytic cracking of polypropylene by using Fe-SBA-15 synthesized in an acid-free medium for production of light hydrocarbon oils. J Analyt Appl Pyrolys 146(1–2):104755 17. Park JW, Kim J-H, Seo G (2002) The effect of pore shape on the catalytic performance of zeolites in the liquid-phase degradation of HDPE. Polymer Degrad Stab 76:495–501

Cold Sintering of Iron Powdered Metal Compacts and Their Performance Linsea Paradis, Ramakrishnan Rajagopalan, Austin Fairman, Kyle Robertson, Daudi R. Waryoba, and Clive Randall

Abstract Iron powder metallurgy is a well-established field in the powder metal (PM) industry due to its ease of processing and appreciable mechanical properties. One area that can contribute significantly to increased production and warrants more study is green machining. Additionally, our current global environmental state calls for more energy efficient processing methods. Cold sintering process (CSP) of metals may provide a means for decreasing required sintering temperature for metal via liquid phase sintering. Our application of CSP utilizes surface modification of iron particles to form an ultrathin hydrated phosphate layer (~10 nm). The hydrated layer promotes driving force for rearrangement and densification under warm compaction to yield compacts with significantly increased green strength up to 70 MPa. This method is currently under investigation for the impact of alloyed iron as well. Implementation of CSP for iron may result in increased mechanical properties, decreased sintering temperature requirements, and decreased energy consumption. Keywords Iron · Cold sintering · Surface modification · Warm compaction

Introduction The cold sintering process is a method developed by Dr. Clive Randall and his group of researchers for sintering ceramics at unconventionally low temperatures, allowing for less demanding sintering conditions. The CSP for ceramics utilizes an aqueous solution of powders to induce what is believed to be a dissolution–precipitation mechanism from hydrated powders via liquid phase sintering at particles boundaries. This results in densification at significantly lower temperatures than conventional sintering [1–4]. For the study discussed in this paper, the focus is on powder metal systems rather than ceramic systems. The process involves surface modification of iron particles through treatment with very small amount of phosphoric acid to form hydrated phosphate layers [5–7]. Furthermore, samples undergo warm L. Paradis (B) · R. Rajagopalan · A. Fairman · K. Robertson · D. R. Waryoba · C. Randall Pennsylvania State University, PA, USA e-mail: [email protected] © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_52

564

Cold Sintering of Iron Powdered Metal Compacts …

565

compaction which heats the coated powders while simultaneously applying pressure via hot die [8]. The hydrated phosphates formed from deposition combined with the warm compaction allow for liquid phase sintering of particle boundaries. That is, the phosphate mobility of the particles increased. Herein, the motion and densification may occur more efficiently than during normal iron compaction. Ideally, the enhanced strength of this system should be sufficient to allow for machining of the green phase compact. One may choose from two sintering conditions or some intermediate form, depending on the application and goals for the samples. If the intent is to reduce energy and cost requirements, a person may choose to sinter modified powders at a temperature lower than that of a traditional iron compact, while maintaining a designated sintered strength. If instead the goal is to enhance the mechanical properties of compacts, a person may choose to maintain similar sintering conditions as traditional iron compacts, while instead using the modified iron-phosphate. This should result in increased strength. The increased densification prior to sintering increases the freedom for manipulating sintering temperature and strength.

Methodology Phosphoric acid was measured to 0.15wt% against 250 g of unlubricated iron powder. The phosphoric acid was mixed with the 500 mL of water, then the 250 g of powder added to the mixture. The mixture was stirred for 10 min. Powders were rinsed and filtered with distilled water until the pH reached that of the unmixed water (~7). Then the powders were rinsed and filtered with 300 mL of ethanol. The powder was placed in a fume hood for 3 days to dry, then placed under vacuum for 1 day to evaporate excess hydration left in the powder. Later, some iron-phosphate samples were combined with copper and carbon to produce carbon steel and is better discussed under the steel section. Two studies were performed, one where the applied phosphoric acid content was varied, and one where the applied compact temperature was varied. First, both green phase and sintered compact strength were investigated as a function of the applied phosphoric acid content, with samples produced with 0.05wt%, 0.50wt%, and 1.0wt% applied content, in addition to the original 0.15wt%. Here it is important to note that this is the applied content. Beyond the point of saturation of phosphoric acid to the system, it is presumed that most excess acid is washed away during vacuum filtration. Coated powders were mixed with 0.75%wt acrawax binder in a turbula mixer for 30 min. In steel samples, the iron-phosphate powders were mixed with copper and carbon graphite prior to the addition of the binder. Approximately 18.5 g samples were pressed in a Carver hot compaction press to dimensions about 0.25 × 1.25 × 0.50 to produce transverse rupture bars for strength testing. Actual dimensions and post pressing mass were measured to calculate individual density. In most cases, powders were pressed for 10 s at various temperatures. Some samples were pressed to

566

L. Paradis et al.

either 70, 100, 200, 620, or 700 MPa to study the relative density as a function of time and pressure. To remain consistent with literature for warm compacted transverse rupture bars, a compaction pressure of 620 MPa was chosen for strength testing [9]. Samples were initially pressed at 140 °C to allow for evaporation of excess phosphates. Later, the applied compact temperature was varied to observe whether there was actually significant evaporation of phosphates at this temperature. For the compact temperature study, temperatures applied ranged from 60 °C to 140 °C in 20 °C increments. Following compaction, some samples were immediately tested in the green phase for rupture strength, while other samples sintered prior to strength testing. Sintered samples were subject to 90%N2 –10%H2 atmosphere in an industrial belt furnace. The samples were heated at ~2 °C/min to 1175 °C and held for 30 min, then cooled at about the same rate.

Results Qualitative Data SEM–EDS analysis suggested the presence of phosphate in the compact iron system. To have a greater degree of certainty, the sample was also examined in TEM. In control samples, there was an iron oxide layer approximately 20 nm thick surrounding the iron particles. In the coated samples, the iron oxide layer is replaced by a hydrated iron-phosphate bonding the iron particles. These iron-phosphate layers are presumably what contributes to the increased densification and strength of the modified iron sample. TEM imaging also suggests the presence of both crystalline and amorphous regions in the compact. XRF further substantiates the presence of phosphates in the system, with a relative increase with increasing applied phosphoric acid content. DRIFTS analysis was performed on powders with the 0.15wt% applied phosphoric acid content. The IR graph shows O–H water stretching and bending at ~3300 cm−1 , ~1600 cm−1 , and ~1300 cm−1 as well as tetrahedral phosphate anions at ~1100 cm−1 (Fig. 1). Additionally, there is a small peak at approximately ~800 cm−1 which represents vibration and stretching across bridged oxygen ions from phosphate groups. In addition to these peaks, there was a change in intensity as a function of temperature. As temperature increased, the intensity from O–H stretching and bending decreased, suggesting hydration was lost upon heating.

Quantitative Data Upon compaction, coated powders reached a density of 7.0–7.3 g/cc. Dilatometry studies showed approximately a 2% increase in relative density compared to the controls after about 300 s. This was present in all the samples of the various applied

Cold Sintering of Iron Powdered Metal Compacts …

567

Fig. 1 DRIFTS spectrum as function of temperature

compaction pressure. Meanwhile, the control samples showed minimal change in density over the same period of time. Strength testing was performed using a universal Instron machine for 3-point bending tests of the transverse rupture bars. The effects of applied phosphoric acid content and applied compact temperature on transverse rupture strength (TRS) were studied for green samples, and a phosphoric acid content study was performed for sintered samples. In green phase samples, relative strength of compacts appeared to work as a function of applied phosphoric acid content. Samples with the lowest applied content (0.05wt%) achieved the greatest green strength, and strength decreased with increased phosphoric acid content. As it appears, the lower the phosphoric acid content, the greater the green strength (Table 1). This implies that if the iron sample is oversaturated with phosphoric acid at the iron interfaces, it may be detrimental to the strength. In temperature studies, increased compact heating appears to have increased the green strength up to about 80 °C, then plateaued beyond this temperature (Table 2). Here, there is no evidence of significant evaporation around 140 °C. Ordinarily, green phase samples do not survive machining. It has been suggested that to successfully machine a green phase sample, compacts must have a transverse rupture strength of 20–30 MPa [10]. Untreated iron compacts in this study only Table.1 Green strength as a function of applied phosphoric acid content 0.05wt% Ave. green density Ave. green TRS (MPa)

7.22 66

0.15wt% 7.22 63

0.5wt% 7.18 53

1.0wt% 7.16 51

568

L. Paradis et al.

Table.2 Green strength as a function of compact temperature Ave. green density (g/cc) Ave. green TRS (MPa)

20 °C

60 °C

80 °C

100 °C

120 °C

140 °C

7.00

7.12

7.23

7.22

7.18

7.17

19

54

60

63

57

62

achieved about half that strength. In the warm compacted iron-phosphate samples, the green strength was sufficiently increased to allow for machining in the green phase (Fig. 2). All samples heated to 60 °C and above (under 620 MPa of pressure) far exceeded the proposed 20–30 MPa rupture strength for machining. The application of machining under the green phase rather than the sintered phase may significantly decrease energy requirements and cost for product processes in an industrial setting [10]. For sintered compacts, there was a significant increase in strength for untreated iron sintered samples that were warm compacted compared to conventional PM products (Fig. 3). Samples that were both coated and warm compacted experienced even greater strength; warm compacted phosphate sample strength was increased by Fig. 2 Green machined iron-phosphate warm compact

Fig. 3 TRS measurements of sintered conventional, warm compacted, and phosphate coated plus warm compacted PM

Cold Sintering of Iron Powdered Metal Compacts …

569

Table.3 Transverse rupture strength of samples of various phosphoric acid content warm compacted at 100 °C and sintered at 1175 °C 0.05wt% Sintered density (g/cc) Sintered TRS (MPa)

Table.4 Transverse rupture strength of modified sintered steels

7.13 735

TRS (MPa)

0.15wt% 7.13 740

0.5wt% 7.03

1.0wt% 7.03

692

721

FCP-0205

FCP-0208

FCP-0212

1011

1194

1257

about 130 MPa compared to conventional PM when sintered. The strength of sintered samples with varied applied phosphoric acid content yielded similar results for transverse rupture strength, about 700 MPa, regardless of the amount of applied acid (Table 3). This may be due iron absorption of the phosphate layer during sintering. The coated iron powders were also applied to form steel systems. The 0.05wt% iron-phosphate coated powders were combined with copper and carbon to produce FCP-0205, FCP-0208, and FCP-0212. These samples were also pressed at 100 °C. The addition of 2% copper and 0.5% graphite into the system increased the rupture strength by more than 500 MPa compared to the iron-phosphate samples prepared under the same conditions (Table 4). The increase in graphite to 0.8 and 1.2% increased the strength even further.

Conclusions and Future Work The phosphorous coated iron powder method shows promising results with increased strength at extremely low temperatures which would be attractive to PM companies. With the increased strength, pressed samples were successfully machined in the green phase, which may be beneficial in the PM industry, as this means products can be machined before sintering and reduce energy required for processing. Additionally, the adoption of cold sintering of metals in the PM industry could lead the way for modern methods of PM production, with significantly improved mechanical properties. Acknowledgements The authors acknowledge PA Manufacturing innovation program for funding this project; Penn State’s Materials Characterization Lab for access to excellent equipment and services; Penn State’s Multi-Campus Research Experience for Undergraduates for undergraduate and research funding; Advantage Powdered Metals for supplying iron powders

570

L. Paradis et al.

References 1. Ndayishimiye A, Sengul MY, Sada T, Dursun S, Hwi Bang S, Grady ZA, Tsuji K, Funahashi S, van Duin ACT, Randall CA (2020) Roadmap for densification in cold sintering: chemical pathways. Open Ceram 2 2. Guo J, Floyd R, Lowum S, Maria J-P, Herisson de Beauvoir T, Seo J-H, Randall CA (2019) Cold sintering: progress, challenges, and future opportunities. Ann Rev Mater Res 49(1):275–295 3. Herisson de Beauvoir T, Dursun S, Gao L, Randall C (2019) New opportunities in metallization integration in Cofired Electroceramic multilayers by the cold sintering process. ACS Appl Electron Mater 1(7):1198–1207 4. Guo J, Guo H, Baker AL, Lanagan MT, Kupp ER, Messing GL, Randall CA (2016) Cold Sintering: a paradigm shift for processing and integration of ceramics. Angew Chem Int Ed Engl 55(38):11457–11461 5. El-Lateef H, Touny A, Saleh M (2018) Synthesis of crystalline and amorphous iron phosphate nanoparticles by simple low-temperature method. Mater Res Express 6. https://doi.org/10. 1088/2053-1591/aaf82b 6. Hsiang H-I, Fan L-F, Hung J-J (2018) Phosphoric acid addition effect on the microstructure and magnetic properties of iron-based soft magnetic composites. J Magn Magn Mater 447:1–8 7. Ma Y, Shen W, Yao Y (2019) Preparation of Nanoscale Iron (III) Phosphate by using ferro-phosphorus as raw material. IOP conference series: earth and environmental science 252:022032. https://doi.org/10.1088/1755-1315/252/2/022032 8. Simchi A, Nojoomi AA (2013) Warm compaction of metallic powders, woodhead publishing series in metals and surface engineering, advances in powder metallurgy, pp 86–108 9. Gagné M, Thomas Y, Lefebvre L (2001) Effect of compaction temperature on the lubricant distribution in powder metal parts 10. Kulkarni H, Dabhade VV (2019) Green machining of powder-metallurgy-steels (PMS): an overview. J Manuf Process 44:1–18

Design of New High Entropy Ceramics in the Pseudo-Binary System RGaO3 -R2 Ti2 O7 Victor Emmanuel Alvarez-Montano, Francisco Brown, Jorge Mata Ramírez, Subhash Sharma, Ofelia Hernández Negrete, Javier Hernández Paredes, and Alejandro Durán Abstract In the last years, high entropy ceramics (HECs) compounds have attracted significant attention due to their unique chemical compositions and crystal structure which make them potentially useful functional materials. One of them is the RGa1/3 Ti2/3 O10/3 (R: rare earth element) ceramic layered compound, which comes from the pseudo-binary system RGaO3 –R2 Ti2 O7 partial solid solution. In this work, we design a single phase of (Lu0.2 Yb0.2 Tm0.2 Er0.2 Ho0.2 )Ga1/3 Ti2/3 O10/3 high entropic ceramic compound. This compound was synthesized by the solid-state reaction method and employing several thermal treatments at high temperatures. The phase stability was determined using X-ray powder diffractometry analysis (XRD). The morphology and cation distribution in the samples were identified using scanning electron microscopy (SEM) and elemental mapping. In addition, the dielectric behavior of samples exposed under several heating treatments is presented.

The original version of this chapter was revised: The co-author name has been updated from “V.E. Alejandro Durán” to “Alejandro Durán”. A correction to this chapter is available at https://doi.org/10.1007/978-3-030-92381-5_150 V. E. Alvarez-Montano (B) · F. Brown · O. Hernández Negrete · J. Hernández Paredes Departamento de Ingeniería Química y Metalurgia, Departamento de Investigación en Polímeros y Materiales, Departamento de Física, Universidad de Sonora, Rosales y Luis Encinas s/n col. Centro, 83000 Hermosillo, Sonora, México e-mail: [email protected] J. Mata Ramírez Universidad Autónoma de Baja California, Facultad de Ingeniería, Arquitectura y Diseño, Ensenada, Baja California, México S. Sharma CONACYT- Centro de Nanociencias Y Nanotecnología, Universidad Nacional Autónoma de México, km. 107 Carretera Tijuana-Ensenada, Apartado Postal 14, 22860 Ensenada, Baja California, México A. Durán Centro de Nanociencias Y Nanotecnología, Universidad Nacional Autónoma de México, km. 107 Carretera Tijuana-Ensenada, 22860 Ensenada, Baja California, México © The Minerals, Metals & Materials Society 2022, corrected publication 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_53

571

572

V. E. Alvarez-Montano et al.

Keywords High entropy ceramics · RGa1 /3 Ti2 /3 O10/3 · Microstructure · Phase stability

Introduction Research in advanced ceramics engineering is focused on the development of specific materials for solving important problems in modern technology, such as looking for new resources of energy, the care of the environment, as well as the optimization of particular processes in electrical, magnetic and, optical applications. There are trends in ceramic science to recognize and systematize the behavior of defects in functional materials, together with a good understanding of the structure-properties relation [1]. One way to continue with this important work is through the synthesis and study of high entropy ceramics (HECs), which have been an important topic in the last six years since new possibilities in applications can be observed in multicomponent ceramics systems [2–4]. The idea of study of the HECs comes from the well-known high entropy alloys (HEAs) in the metallic systems, and they are defined as materials with at least four cations (or anions) in their composition [3]. The initial studies have started from the more basic systems such as FCC (rock salt) and hexagonal (metal borides) [5, 6]. Later, more complex structures such as perovskite, bixbyite, and spinel have been tasted [7–9]. From many recent results, HECs have attracted attention because of promising applications in several fields like ceramic catalysts, thermal materials for protection, and electrochemical devices [2]. In this research report, we describe the design of a new HEC in the pseudo-binary system RGaO3 –R2 Ti2 O7 . There exist compounds with composition RGa1-x Tix O3+x/2 with a novel layered crystal structure [10]. The phase stability is studied by XRD and the microstructure behavior after thermal treatments of compressed pellets is studied by SEM and electron mapping. An initial description of the dielectric behavior of sintered samples is included.

Materials and Methods High purity oxides (99.9%), R2 O3 (R: Lu, Yb, Tm, Er, Ho), TiO2 , and Ga2 O3 , were weighted in stoichiometric relations (R2 O3 :TiO2 :Ga2 O3 = 3:4:1) in an agate mortar and mixed under ethanol to form RGa1/3 Ti2/3 O10/3 (R: Lu, Yb, Tm, Er, Ho) compounds. The solid-state reaction method was successfully used to prepare each RGa1/3 Ti2/3 O10/3 compound. Individual samples were compressed in a stainless-steel die (φ = 13 mm), applying 2.5 tons of pressure using a uniaxial hydraulic press (Carver Model C). The reaction temperature of heating cycles was 1300 °C and applying periodic grinding every two days. All samples were quenched (1300 °C to room temperature in air). X-ray diffraction was used to identify the phases, according

Design of New High Entropy Ceramics in the Pseudo-Binary System …

573

to reference [10]. For the X-ray diffraction analysis, a Bruker D2 Phaser facility, with Cu-Kα radiation (1.5406 Å) was used. The X-ray tube was operated to 30 kV and 10 mA, with a scanning rate of 1°/min in 2θ in the range from 10 to 80. For the synthesis of HEC (Lu0.2 Yb0.2 Tm0.2 Er0.2 Ho0.2 )Ga1/3 Ti2/3 O10/3 , now labeled HECRGTO, equimolar amounts samples RGa1/3 Ti2/3 O10/3 (R: Lu, Yb, Tm, Er, Ho) were weighted and mixed under ethanol and the same procedure applied to starting materials was used. Three periods of heating were applied and later the characterization was done. We calculated lattice parameters employing the least-squares method. After HEC-RGTO single phase was successfully obtained, samples of 120 mg were pelletized in a stainless-steel die (6 mm diameter) applying a pressure of 0.5 tons. Several pellets were sintered at 1300, 1400, 1500, and 1600 °C in air by one day and then quenched. Microstructure, EDS, and elemental mapping were done in a JEOL 5400LV operated at 20 kV. The dielectric behavior of sintered samples was verified by using an LCR bridge (HP-4284A) from 25 to 500 °C with frequencies of 10, 50, and 100 kHz.

Results The solid-state reaction was monitored after each heating cycle during the synthesis process by SEM–EDS. We found that almost a single phase was obtained after the first two days of the chemical reaction, although two more grinding and heating cycles were applied. In Fig. 1 we can see SEM image and linear elemental mapping for the HEC-RGTO sample after finishing the first heating cycle. All the elements

Fig. 1 Linear elemental mapping of powder sample of HEC-RGTO after the first reaction cycle at 1300 °C

574

V. E. Alvarez-Montano et al.

are well distributed along the different particles in this powder sample. Figure 2 presents the indexed XRD pattern of the HEC-RGTO as a single phase. All peaks were indexed in space group R3m (#160), with lattice parameters a(Å) = 18.018(1), c(Å) = 16.828(2), and V(Å3 ) = 4731.1 (hexagonal system). Comparing these lattice parameters with those of Lu(Ga1/3 Ti2/3 )O10/3 (a(Å) = 17.924(3), c(Å) = 16.883 (2), V(Å3 ) = 4697.3), it is reasonable that they have an increase in their value since the 80% of the present cations in the octahedral sites are higher in their ionic radii [11]. It is possible that the effect of the several cations in the hexagonal unit cell volume is that of an average ionic radio, which is close to that of Tm(III) in coordination 6 (0.88 Å), and that is between the five rare earth cations present. The crystal structure of HECRGTO can be described as four kinds of sheets stacked in the 001 direction, O-I, R (R: Lu, Yb, Tm, Er, Ho), O-II, and (Ga/Ti)O. It can also be described as layers of polyhedral, RO6 (octahedral), and (Ga/Ti)O5 (distorted trigonal bipyramid) touched by their vertices. An extensive crystal structure analysis is well described for the isostructural Yb(Fe1−x Tix )O3+x/2 [10]. Figure 3 shows the XRD patterns of three pellet samples after being submitted to three different heating treatments (1 day) of 1300, 1400, and 1500 °C. It is observed that the solubility of the rare-earth cations is maintained in the RGa1/3 Ti2/3 O10/3 phase without the precipitation of other phases, indicating that the stability of the main phase is maintained up to 1400 °C. Contrary to this, the XRD shows splitting peaks of the (003), (052) and (504) planes at 1500 °C. This fact strongly suggests two scenarios; a) the freezing of a new phase by quenching and that is stable at 1500 °C and b) the coexistence of two phases in the solid solution. However, more studies are required to resolve this issue. In Fig. 4, SEM images are presented for pellet samples after one day of thermal treatment at 1300, 1400, 1500, and 1600 °C. Sintered pellet at 1300 °C shows high porosity, and it is gradually reduced with the increase of the sintering temperature. The bulk density was calculated by the conventional weight and volume ratio of cylindrical pellets, and it is presented in Table 1. Figure 5 presents SEM image and

052

3000

Fig. 2 Indexed XRD pattern of HEC-RGTO

003

2000

10

20

30

40

50

807

550 058 553

0

055

500

234

501

1000

006 504

1500

303 204

Intensity (a. u.)

2500

60

70

80

Design of New High Entropy Ceramics in the Pseudo-Binary System …

575

Fig. 3 XRD of HEC-RGTO pellets after heating treatments for 1 day (1300, 1400, and 1500 °C)

Fig. 4 Scanning electron images of HEC-RGTO after one day at four different thermal treatments, a 1300, b 1400, c 1500, and d 1600 °C

576

V. E. Alvarez-Montano et al.

Table 1 Density (g/cm3 ) of HEC-RGTO after 1-day thermal treatments HEC

Compound

1300 °C 1400 °C 1500 °C 1600 °C

RGTO (Lu0.2 Yb0.2 Tm0.2 Er0.2 Ho0.2 )Ga1/3 Ti2/3 O10/3 4.08

5.27

6.56

6.68

Fig. 5 Area elemental mapping of HEC-RGTO single-phase pellet sample sintered at 1300 °C

area elemental mapping for HEC-RGTO single-phase pellet sample. Homogeneous distribution of all cations can be observed. Figure 6 shows the dielectric constant, (left side) and dielectric loss, Tan (right side) as a function of temperature for the HECs samples heat-treated at 1300, 1400 and 1500 °C. It is interesting to observe that the values of the dielectric constant (less than 15) and the loss tangent (0.01) are small with slight increasing from room temperature to 300 °C, for all sintered samples, which is indicative of good dielectric material. It is noted a dispersion and crossover in both, and Tan with frequencies around 300 °C. Similar behavior is observed for 1400 °C sintered sample. For 1500 °C sintered sample an interesting feature is observed like a decrease in dielectric constant value up to nearly 150 °C for 50 and 100 kHz. The same behavior of dispersion has been observed in dielectric loss plots in co-relation to dielectric constant. This fact could be related to the formation of a new phase or the coexistence of two phases in the solid solution, as we have indicated in the structural characterization. However, more studies are necessary to clarify the origin of this behavior in the 1500 °C heat-treated sample.

Design of New High Entropy Ceramics in the Pseudo-Binary System …

577

Fig. 6 Dielectric constant (εr ) and loss tangent (tanδ) dependence with the temperature at several frequencies for HEC-RGTO single-phase pellet samples sintered at 1300, 1400, and 1500 °C

Conclusion The design of a new High Entropy Ceramic Compound (HEC) with formula (Lu0.2 Yb0.2 Tm0.2 Er0.2 Ho0.2 ) Ga1/3 Ti2/3 O10/3 come from the pseudobinary system RGaO3 -R2 Ti2 O7 is presented. By the solid-state reaction method, (Lu0.2 Yb0.2 Tm0.2 Er0.2 Ho0.2 )Ga1/3 Ti2/3 O10/3 (HEC-RGTO) was successfully formed at 1300 °C. The XRD data showed that the hexagonal unit cell is stable up to 1400 °C. Presumably, a phase transformation or the coexistence of two phases occurs at 1500 °C. The grain size and apparent densification increase with increasing the sintering temperature. Microstructural analysis shows an expected reduction of porosity after heating treatments at higher temperatures. Elemental mapping showed homogeneity in the presence of all cations in the sintered samples. Dielectric behavior of HECsRGTO shows small values of dielectric constant (750

>750

3.33 × 10–7

1.02 × 10–8

and cast iron coupons appeared glaring in the potentiodynamic polarization scans shown in Fig. 3c. Corrosion resistance of cast aluminum significantly increased in 1000 ppm chloride containing coolants with a large passive range and approx. an order of magnitude lower passive currents when compared with equivalent cast iron coupons. A summary of selected electrochemical parameters extracted out of EIS and polarization scans for cast iron and cast aluminum is shown in Table 1. With exposure to elevated temperatures cast aluminum surface appears to have a formation of corrosion inhibition layer which is protective even at 1000 ppm chloride levels as opposed to what was observed earlier at room temperature shown in Fig. 1b. These are currently ongoing studies and investigations to gather deeper understanding of this reversal in the corrosion behavior of cast aluminum material. Metal-chelate formation has been observed over cast iron’s surface that provided inhibition to the base material when exposed to engine coolant with organic additives up to 100 ppm chloride additions [12]. In a parallel study, electrochemical tests were performed to understand the room temperature corrosion behavior of brass material in an engine coolant infused with organic additives and a mixture of potential corrosive species, i.e. chlorides and sulfates. This study is in conjunction with the earlier published data on cast iron and cast aluminum corrosion behavior in coolants with organic inhibitors [10–12]. Figure 4a and b show potentiodynamic scans on brass material in engine coolants containing varying concentrations of chlorides and sulfates, respectively. In the organic additive infused engine coolants, brass material showed no difference in polarization profiles up to 500 ppm chlorides which was the threshold upper limit observed on cast iron coupons [11, 12]. Similarly, sulfates also demonstrated no discernable change in the polarization scans up to 1500 ppm levels. And even at 3000 ppm sulfate levels, the coupons showed a large passivation behavior up to ~350 mV versus Ag/AgCl beyond which a breakdown was noticed. These are preliminary results, and currently, investigations are ongoing to further understand if the same corrosion behavior persists at elevated temperatures for brass materials in engine coolants containing either organic or inorganic corrosion inhibitors. However, these initial results on brass seemed promising when compared with cast iron corrosion behavior in the similar engine coolant environments.

An Electrochemical Study of Ferrous …

627

Fig. 4 Room temperature potentiodynamic polarization scans on brass coupons in engine coolants containing organic inhibitor additives and varying concentrations of a chlorides and b sulfates

Conclusions Engine coolant corrosion studies on three different metallurgies, viz. cast aluminum, cast iron, and brass, were conducted using electrochemical tests. Following are the key outcomes: 1.

2.

3.

The room temperature potentiodynamic testing in chloride contaminated engine coolants containing inorganic inhibitor additives showed that increasing chloride additions reduced the breakdown potentials and increased the anodic current densities. Cast aluminum showed the smallest passive range and broke down at a relatively smaller overpotentials compared to cast iron and brass in 1000 ppm chloride added coolants. Brass showed a much lower drop in breakdown potential of ~100 mV in coolants with inorganic inhibitor additives and chloride additions. On the other hand, brass did not show any decrease in current densities or breakdown potentials in chloride added coolants containing organic inhibitor additives. Sulfate contaminated coolants with organic inhibitors showed a drop in breakdown potentials only at very high concentrations of 3000 ppm sulfate. High temperature exposure of cast aluminum and cast iron in chloride doped coolants containing inorganic additives showed that cast aluminum showed improved protection after 14 days of exposure in spite of 1000 ppm chloride additions. On the other hand, cast iron showed decreasing trend in corrosion resistance with longer exposure times and higher chloride additions.

628

G. Argade et al.

References 1. Lee JA (2003) Cast aluminum alloy for high temperature applications. In: Automotive alloys, 132nd TMS annual meeting and exhibition 2. Garat M, Laslaz G (2007) Improved aluminium alloys for common rail diesel cylinder heads. AFS Trans 115:89–96 3. Fan KL et al (2013) Tensile and fatigue properties of gravity casting aluminum alloys for engine cylinder heads. Mats Sci Eng A 586:78–85 4. Weir TW, Ven PV (1996) Review of organic acids as inhibitors in engine coolants, SAE Technical Paper, pp 37–47 5. Pellet RJ, Bartley LS, Hunsicker DP (2001) The role of carboxylate-based coolants in cast iron corrosion protection. SAE Trans 1342–1348 6. Hudgens RD (1999) Comparison of conventional and organic acid technology (OAT) coolants in heavy duty diesel engine service. SAE Trans 82–91 7. Zheludkevich M, Yasakau KA, Poznyak SK, Ferreira MGS, Thiazole and Triazole additives as corrosion inhibitors for AA 2024 alloy. Corros Sci 47(2005):3368–3383 8. Maes JP, VanNeste WA (1994) Corrosion inhibited antifreeze formulations having monocarboxylic, triazole, and imidazole compounds, U.S. Patent No. 5, 366,651 9. Chilukuri A (2012) Corrosion inhibition by inorganic cationic inhibitors on the high strength aluminum alloy, 2024-T3, The Ohio State University, Columbus, PhD dissertation 10. Chilukuri A, Argade GR, Perry J, Trobaugh C, Schafer R, Raisor E, Steenhoek J, Lugo GP (2019) A corrosion study of light metal cylinder head material in chloride containing engine coolant environment. Mater Sci Technol 1255–1259. https://doi.org/10.7449/2019/ MST_2019_737_741 11. Argade GR, Chilukuri A, Perry J, Trobaugh C, Schafer R, Raisor E, Steenhoek J, Lugo GP (2019) An investigation of cast iron corrosion behavior in chloride containing engine coolant environment. Mater Sci Technol 737–741 12. Argade G, Chilukuri A, Perry J, Viers M, Steenhoek J, Debusk J,Wang C, Trobaugh C (2021) Corrosion behavior of alloyed cast iron in ethylene glycol-based engine coolants at elevated temperature. Coatings 11:357. https://doi.org/10.3390/coatings11030357

An Investigation of the Microstructure and Oil Retention of Electrolyte Jet Plasma Oxidation (EJPO) Coating Nasim Bahramian, Sina Kianfar, Joshua Stroh, Dimitry Sediako, and Jimi Tjong

Abstract Next-generation automotive engines demand new materials and technologies to meet efficiency mandates regarding energy conservation and emission reduction. Replacing heavy ferrous metals with lightweight aluminum (Al) alloys can considerably improve fuel efficiencies in the transportation sector. However, most Al alloys suffer from poor tribological characteristics, which urged original equipment manufacturers to develop advanced materials for high friction applications like cylinder blocks. Recently, there has been a growing trend to produce coatings with high-temperature strength and wear resistance. These coatings are deposited on an Al substrate of linerless engine blocks using various technologies, including Electrolytic Jet Plasma Oxidation (EJPO). The present study investigates the coating’s microstructure, surface roughness, and oil retention and compares the latter with the oil retention of the Plasma Transferred Wire Arc (PTWA) coating which is currently used in cylinder block applications. The results revealed higher roughness (Ra = 0.86 μm) and oil retention (V0 = 0.356 μm3 /μm2 ) of the EJPO coating than the PTWA one (Ra = 0.34 μm, V0 = 0.041 μm3 /μm2 ). Keywords Surface modification and coating · High-temperature material · Characterization · Linerless Al engine block

Introduction To combat climate change and global warming, many governments have introduced energy conservation and fuel consumption mandates within various industries. For instance, in the automotive sector, weight reduction has been one of the most promising approaches to address the new demands [1, 2]. Replacing heavy ferrous N. Bahramian (B) · S. Kianfar · J. Stroh · D. Sediako The University of British Columbia Okanagn Campus, 1137 Alumni Ave, Kelowna, BC V1V 1V7, Canada e-mail: [email protected] J. Tjong University of Windsor, 401 Sunset Ave, Windsor, ON N9B 3P4, Canada © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_60

629

630

N. Bahramian et al.

metals with lightweight aluminum (Al) alloys, particularly in powertrain components, is an effective solution toward this goal. However, Al alloys suffer from poor tribological characteristics and insufficient high-temperature strength, which are both required for applications such as engine blocks. To tackle this challenge, a common approach is to embed cast-in iron (Fe) liners in the cylinder bores to protect them from the harsh conditions that are present inside the combustion chamber. However, during the manufacturing of Al engine blocks with cast-in Fe liners, the mismatch of thermal expansion/contraction coefficients between the Al cylinder wall and the Fe liner leads to the development of unwanted residual stresses [3, 4]. These stresses can cause distortion and premature failure of engine blocks while under in-service loadings [5–7]. Another approach is to use press-fit Fe liner, as opposed to cast-in Fe liner, which lowers the maximum magnitude of tensile stress and changes the stress mode from purely tensile to mostly compressive [8]. However, the reduced interface area between the liner and the cylinder wall results in a lower thermal conductivity, adversely affecting the engine’s efficiency [9, 10]. Nowadays, there has been a growing trend toward manufacturing linerless engine blocks with advanced coating materials as alternatives to the Fe liners [11]. These coatings are to possess specific characteristics, protecting the Al part from the severe conditions in the combustion chamber. Other than high-temperature strength, as well as wear and corrosion resistance, the cylinder wall surface must also have sufficient oil retention. Oil retention is the ability of a surface to retain enough oil to prevent scuffing of the poor-lubricated area, such as the exposed surface of the cylinder wall when the piston is at Top Dead Center (TDC). Among various coating technologies, Plasma Transferred Wire Arc (PTWA) is a steel-based coating process that offers improved thermal conductivity and lower mass as compared to Fe liners [12, 13]. Previous studies showed that the tribological characteristics and strength of the PTWA coating are sufficient for engine block applications [14]. However, due to its dense structure (i.e., low open-porosity level), the PTWA coated cylinder wall requires post-processing honing to provide the surface with sufficient oil retention. The additional post-processing time, accompanied by the mechanical bonding between the coating and substrate (i.e., Al alloy cylinder wall), urged OEMs to find a more promising coating technique for next-generation internal combustion (IC) engines. Plasma Electrolyte Oxidation (PEO) is an environmentally friendly and costeffective coating technique that demonstrates considerable promise in applications subjected to severe wear, corrosion, and elevated temperatures [15–17]. PEO is a plasma-assisted coating generated by electrochemical conversion of the lightweight metal surface to a hard and well-bonded ceramic film. For instance, during PEO coating on an Al component, the Al part is immersed in an electrolyte bath while being subjected to high voltages, leading to numerous local plasma discharges on the metal surface and gradual formation of an alumina coating [18–20]. While the PEO technique is highly applicable for coating the external surfaces of a metallic component, its application is limited if certain areas of the external surface must remain uncoated. In these cases, the uncoated area would be covered from electrolyte, giving rise to additional post-processing cost and time. To overcome this challenge, a novel

An Investigation of the Microstructure and Oil Retention …

631

coating method called Electrolyte Jet Plasma Oxidation (EJPO) is being developed, with a similar coating mechanism as PEO but a different configuration in applying the electrolyte [21]. In the EJPO coating process, the electrolyte is directly sprayed onto the locations of interest for the coating, allowing it to coat only the selected regions of the complex component. Therefore, EJPO coating has a great potential to provide a forever solution for using lightweight metals in broader applications such as aerospace, railway, and automotive powertrain. Despite numerous studies on the application of PEO on simple geometries, there is a lack of microstructural investigation on the EJPO coating applied to complex geometries. This study conducts a microstructural characterization on the cylinder wall of an EJPO coated engine block. The coatings’ morphology and phase composition are investigated using Scanning Electron Microscopy (SEM) and Electron Disperse Xray Spectroscopy (EDS). In addition, surface roughness profiles were measured for an EJPO and a PTWA coated engine blocks. The roughness data were, then, used to perform a comparative study between two coatings to evaluate their oil retention characteristics.

Experimental Procedure and Results Microstructural characterization was performed on a cylinder cut obtained from a V8 engine block coated with the novel EJPO technique. Field Emission Scanning Electron Microscope (FE-SEM, Tescan Mira3 XMU) operating at an accelerating voltage of 15 kV was utilized to evaluate the surface and cross-section morphologies of the coatings. Before SEM examination, the EJPO samples were sputter-coated with gold due to the low electrical conductivity of aluminum oxide. Oxford Instruments Aztec data acquisition and processing software with an 80 mm2 Oxford EDS detector were used for the composition analysis of the coatings and corresponding substrates. As mentioned in the previous section, oil retention is one of the crucial surface characteristics to be considered in cylinder block applications as it is a direct measure for preventing scuffing in poor-lubricated areas. In this study, surface roughness parameters were first measured using a Mitutoyo SJ-410 Series surface roughness tester, as shown in Fig. 1. Then, the roughness parameters were used in the following formula to calculate the oil retention. V0 = Rvk

100 − Mr 2 200

(1)

where V0 is the oil retention parameter (μm3 /μm2 ), Rvk is the proportion of profile valleys below core roughness, and M r2 is the lower intersection point of the core roughness datum line with bearing ratio curve. The terminology and parameters are based on ASME B46.1–2009 [22].

632

N. Bahramian et al.

Fig. 1 Surface roughness measurement set-up

Microstructure of EJPO Coating SEM observations of the EJPO cross-section indicate that the coating consists of three layers, as shown in Fig. 2. The outer layer is exposed to the electrolyte and is subjected to a high cooling rate and rapid solidification, leading to having a porous amorphous structure [23]. A relatively high amount of silicon (Si) content (20%) can also be observed in this layer due to electrolyte constituent interaction with the coating during plasma discharge. Being under a complex cooling rate during coating formation, the intermediate layer is dense and more likely to have a nanocrystalline structure. Due to its dense structure, this layer plays a vital role in improving the coating’s thermal conductivity. Furthermore, compared to the outer region, a dramatic reduction in Fig. 2 EDS phase analysis on EJPO coating’s three layers across the cross-section

An Investigation of the Microstructure and Oil Retention …

633

Fig. 3 Intermetallic phase at the Al substrate boundary reaching to coating/substrate interface

Si content can be seen toward this layer. Finally, there is a thin alumina film, 1– 2 μm, where a sharp drop toward zero occurs in the oxygen level, accompanied by a simultaneous growth in the Al content. Figure 3a depicts the SEM micrographs of the coating cross-section where the substrate intermetallics (such as β-AlFeSi) were occasionally observed to be present at the coating/substrate interface. A phase composition analysis on the coating surface (Fig. 3b) also reveals the presence of intermetallic elements (like Fe) in the outer layer of the coating. These observations suggest that the intermetallic elements can participate in conversion of the substrate into the oxide layer during plasma discharges. The previous study on PEO coating also confirmed the impact of substrate phases on the coating microstructural composition [24]. In addition, the elevated levels of oxygen diffusion into the Al substrate were observed occasionally in the vicinity of the coating/substrate interface (Fig. 4). This can adversely affect the substrate hardness, which shows the necessity for further investigation of the mechanical properties of the Al component.

Roughness Measurements and Oil Retention Figures 5 and 6 depict the roughness profiles obtained from EJPO and PTWA coated cylinder cuts using a Mitutoyo profilometer. The results indicate that the EJPO coating has a high roughness value (i.e., Ra = 0.86 μm, average from ten measurement), which is more than twice the roughness value achieved from the honed surface of the PTWA one (Ra = 0.34 μm). The SEM micrographs of the surfaces of both coating also reveal a much greater level of porosity in the EJPO coating.

634

N. Bahramian et al.

Fig. 4 SEM micrograph showing the oxygen diffusion in the Al substrate

Fig. 5 Roughness profile and surface micrograph of EJPO coating representing the coating’s porosity level

Fig. 6 Roughness profile and surface micrograph of PTWA coating representing the coating’s porosity level

90.568

Mr2 (%)

0.033

0.693

Rvk (μm)

V0 (μm3 /μm2 )

0.318

Ra (μm)

1

0.703

0.322

0.041

88.201

2

Measurement points

0.632

0.309

0.033

89.592

3

Table 1 Roughness and oil retention data for PTWA coating

0.683

0.311

0.045

86.753

4

0.070

82.272

0.791

0.402

5

0.030

88.336

0.517

0.303

6

0.038

89.936

0.754

0.327

7

0.042

90.528

0.878

0.424

8

0.039

88.580

0.687

0.319

9

0.040

87.640

0.652

0.310

10

An Investigation of the Microstructure and Oil Retention … 635

76.216

Mr2 (%)

0.386

3.245

Rvk (μm)

V0 (μm3 /μm2 )

0.892

Ra (μm)

1

4.355

0.996

0.441

79.752

2

Measurement points

2.906

0.907

0.382

73.707

3

Table 2 Roughness and oil retention data for EJPO coating

2.858

0.884

0.384

73.160

4

0.291

77.347

2.571

0.703

5

0.271

77.813

2.446

0.709

6

0.426

80.344

4.339

0.992

7

0.372

73.840

2.842

0.849

8

0.283

80.536

2.909

0.817

9

0.320

78.667

2.997

0.894

10

636 N. Bahramian et al.

An Investigation of the Microstructure and Oil Retention …

637

Tables 1 and 2 represent the roughness data that are used to calculate (see Eq. 1) the oil retention (V0 ) of both surfaces. The results indicate that the average oil retention of the EJPO coating is 0.356 μm3 /μm2 , which is about nine times more than for the PTWA coating (0.041 μm3 /μm2 ). Since the oil retention has a crucial impact on the coating’s ability to prevent severe wear and friction at TDC, there is a potential that the EJPO coating may perform superiorly, as compared to the PTWA coating.

Conclusion In this study, a microstructural analysis was performed on a novel coating technique, known as Electrolytic Jet Plasma Oxidation that was applied to the cylinder bore of a V8 engine block. Then, the surface roughness and oil retention of the coating were measured and compared to the values obtained from a PTWA coated block. The following conclusions may be drawn from this study: (1)

(2)

(3)

(4)

The EJPO coating consists of three layers, each with distinct features. The outer layer, as shown in the SEM micrograph, is primarily amorphous since it is cooled and solidified rapidly; however, the dense intermediate layer is exposed to a complex cooling rate, increasing the possibility of having nanocrystaline structure. The thin inner layer has no porosity and show consistent thickness (~2 μm) along the coating/substrate interface. The surface roughness measurements of the EJPO and PTWA coatings revealed that the EJPO coating had a surface roughness that was nearly 3 times greater (i.e., 0.86 versus 0.34 μm, respectively) than for PTWA. As a result, the EJPO coating has a calculated oil retention that is approximately 9 times higher (i.e., 0.356 μm3 /μm2 versus 0.041 μm3 /μm2 , respectively). The SEM micrograph of the EJPO coating cross-section indicated the occasional elevated oxygen diffusion in Al substrate. Therefore, further research on the effect of the coating on the Al substrate mechanical properties such as hardness is recommended. The present study provides a better understand of the microstructural characteristics of the novel EJPO coating. In addition, the notable increase in oil retention provides good evidence that the EJPO coating may perform superiorly in engine applications, as compared to PTWA. Thus, it is important to continue examining the EJPO coating, focusing on its thermal conductivity and wear resistance.

638

N. Bahramian et al.

References 1. Klier T, Linn J (2011) Corporate average fuel economy standards and the market for new vehicles. Ann Rev Resour Econ 3(1):445–462 2. Aghaie E, Stroh J, Sediako D, Smith M (2018) In-situ fitness-for-service assessment of aluminum alloys developed for automotive powertrain lightweighting. In: Martin O (ed) Light metals 2018. Springer, Cham. pp 397–400 3. Sediako D, Stroh J, Kianfar S (2021) Residual stress in automotive powertrains: methods and analyses. Mater Sci Forum 1016:1291–1298 4. Kianfar S, Aghaie E, Stroh J, Sediako D, Tjong J (2021) Residual stress, microstructure, and mechanical properties analysis of HPDC aluminum engine block with cast-in iron liners. Mater Today Commun 26:101814 5. Sediako D et al (2011) Analysis of residual stress profiles in the cylinder web region of an as-cast V6 Al engine block with cast-in fe liners using neutron diffraction. SAE Int J Mater Manuf 4(1):138–151 6. Lombardi A, Ravindran C, Sediako D, MacKay R (2014) Determining the mechanism of inservice cylinder distortion in aluminum engine blocks with cast-in gray iron liners. Metall Mater Trans A 45(13):6291–6303. https://doi.org/10.1007/s11661-014-2580-y 7. Lombardi A, Sediako D, Machin A, Ravindran C, MacKay R (2017) Effect of solution heat treatment on residual stress in Al alloy engine blocks using neutron diffraction. Mater Sci Eng A 697:238–247 8. Stroh J (2012) Development of precipitation-strengthened aluminum alloys and manufacturing processes for next generation automotive powertrains. University of British Columbia 9. Stroh J, Sediako D, Byczynski G, Lombardi A, Paradowska A (2020) Stress characterization of bore-chilled sand cast aluminum engine blocks in as-cast and T7 condition with application of neutron diffraction. In: Light metals 2020. Springer, Cham, pp 153–157 10. Kianfar S et al (2021) Residual stress prediction in the casting process of automotive powertrain components. In: Light metals 2021: 50th anniversary edition, pp 858–864 11. Bobzin K et al (2008) Coating bores of light metal engine blocks with a nanocomposite material using the plasma transferred wire arc thermal spray process. J Therm Spray Technol 17(3):344– 351 12. Heinig K-P, Stephenson DA, Beyer TG (2017) Thermal response of aluminum engine block during thermal spraying of bores: comparison of FEA and thermocouple results. SAE Int J Mater Manuf 10(3):360–365 13. Bobzin K, Ernst F, Richardt K, Schlaefer T, Verpoort C, Flores G (2008) Thermal spraying of cylinder bores with the plasma transferred wire arc process. Surf Coat Technol 202(18):4438– 4443 14. Morawitz U, Mehring J, Schramm L (2013) Benefits of thermal spray coatings in internal combustion engines, with specific view on friction reduction and thermal management. No. 2013-01-0292. SAE technical paper 15. Bosch D et al (2015) Secondary Al-Si-Mg high-pressure die casting alloys with enhanced ductility. Metall Mater Trans A 46:1035–1045. https://doi.org/10.1007/s11661-014-2700-8 16. Shen X, Nie X, Hu H, Tjong J (2012) Effects of coating thickness on thermal conductivities of alumina coatings and alumina/aluminum hybrid materials prepared using plasma electrolytic oxidation. Surf Coat Technol 207:96–101 17. Hussein RO, Northwood DO, Nie X (2014) Processing-microstructure relationships in the plasma electrolytic oxidation (PEO) coating of a magnesium alloy. Mater Sci Appl 5:124–139 18. Wang G, Nie X, Tjong J (2015) Surface effect of a PEO coating on friction at different sliding velocities. No. 2015-01-0687. SAE technical paper 19. Feng Su J, Nie X, Hu H, Tjong J (2012) Friction and counterface wear influenced by surface profiles of plasma electrolytic oxidation coatings on an aluminum A356 alloy. J Vac Sci Technol A Vacuum, Surfaces, Film 30(6):61402

An Investigation of the Microstructure and Oil Retention …

639

20. Hussein RO, Nie X, Northwood DO, Yerokhin A, Matthews A (2010) Spectroscopic study of electrolytic plasma and discharging behaviour during the plasma electrolytic oxidation (PEO) process. J Phys D Appl Phys 43(10):105203 21. Shen X, Nie X, Tjong J (2019) Effects of electrolytic jet plasma oxidation (EJPO) coatings on thermal behavior of engine cylinders. Heat Mass Transf 55(9):2503–2515 22. American National Standards Institute, American Society of Mechanical Engineers (2009) Surface texture: surface roughness, waviness, and lay: ASME B46.1-2009 (revision of ANSI/ASME B46.1-2002). American Society of Mechanical Engineers, New York 23. Datta A, Carpenter JD, Ott RD, Blau PJ (2002) Tribological characteristics of electrolytic coatings for aluminum engine cylinder lining applications. SAE Trans 111:272–278 24. Hussein RO, Nie X, Northwood DO (2013) An investigation of ceramic coating growth mechanisms in plasma electrolytic oxidation (PEO) processing. Electrochim Acta 112:111–119

Electrochemical Corrosion Tests of Aluminum 1100 Alloy Coupons in Acid Condensate Environment Vasundhara Shinde, Gaurav Argade, Anusha Chilukuri, Monica Gehrich, and Chirag Parikh

Abstract In this study, product validation tests were compared with rapid electrochemical tests in diesel exhaust condensate environments. Al 1100 alloy coupons were exposed to elevated temperature cycles in an acid condensate environment (simulating diesel exhaust conditions) for 36 days, followed by room temperature electrochemical tests. Day 10 coupons showed the lowest corrosion rates which are attributed to formation of protective aluminum hydroxide layer. In the potentiodynamic test, day 10 coupons showed least anodic current densities at +200 mV versus Ag/AgCl and nobler corrosion potentials as compared to day 36 coupons. A similar trend of lower frequency impedance was observed in electrochemical impedance spectroscopy (EIS) scans between 10–2 and 105 Hz. In the post-corrosion examination, pitting on coupons was confirmed along with higher wt.% oxygen measured through energy-dispersive X-ray spectroscopy, and the corrosion rate increase after 10 days is attributed to eventual degradation of the aluminum hydroxide layer. Keywords Aluminum corrosion · Electrochemical impedance spectroscopy · Potentiodynamic polarization · Acid condensate · Diesel exhaust environment

V. Shinde (B) Media & IP, Cummins Filtration Inc., 1801 US-51, Stoughton, WI 53589, USA e-mail: [email protected] G. Argade · A. Chilukuri · M. Gehrich Materials Science and Technology, Cummins Technical Center, Cummins Inc., 1900 McKinley Avenue, Columbus, IN 47201, USA C. Parikh Research and Product Technology, Cummins Filtration Inc., 1801 US-51, Stoughton, WI 53589, USA © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_61

640

Electrochemical Corrosion Tests of Aluminum 1100 Alloy …

641

Introduction Aluminum and its alloys have a unique mix of qualities that make them one of the most versatile, cost-effective, and appealing metallic materials for a wide range of applications, especially in the automotive industry. Aluminum 1100 is a commercially pure aluminum grade and is highly resistant to chemical attack and weathering [1]. With increased worries about global energy consumption and pollution, there is a noticeable increase in expectations and laws for improving vehicle fuel economy and reducing emissions. Automotive manufacturers all over the world are looking at alternative lightweight materials, such as aluminum, to meet these restrictions [2]. There are case studies on the corrosion performance of an automotive component with steel design and an alternate aluminum design, and corrosion rates for aluminum alloy in acidic condensate conditions are found to be much lower than some steel grades [2]. Corrosion is a natural process that occurs when metals react with oxygen to return to their stable oxidized state. Aluminum’s exposed surface reacts with oxygen to generate an inert aluminum oxide film which is a few ten-millionths of an inch thick that prevents further oxidation. This thin oxide film can be thought of as a colorless and clear coating that adheres to the metal strongly [1]. Chemically, this film is relatively inert, but aluminum’s corrosion resistance is based on the inactivity of this oxide layer [1]. In the pH range of around 4.5–8, this surface film is often stable. But most strong acids and bases can dissolve the oxide layer, resulting in faster aluminum corrosion. The aluminum oxide layer can also be attacked by chlorides or sulfides in the atmosphere [3]. As a result, studying the rate and mechanism of degradation of oxide layer in highly acidic environment and in presence of chlorides or sulfides is important. The purpose of this study was to study aluminum 1100 alloy corrosion in an acid condensate environment as a part of a material characterization process. The material under investigation is exposed to diesel blowby gases which can condensate to form strong acidic residues during application. Studies report that the primary constituents of blowby gases generally comprise of hydrocarbons, carbon-monoxide, nitrogen oxides, and particulate matter including sulphate particulates which can act as corrosive medium [4, 5]. Wolfgang Tillmann et.al have demonstrated the influence of condensate corrosion using well-established VDA 230–214 procedure for testing the resistance of metallic materials to condensate corrosion in exhaust gas-carrying components [6, 7]. With this information, the process utilized for corrosion testing of aluminum 1100 material that is intended for use in exhaust gas-carrying components is based on VDA 230–214 test specification [8] followed by electrochemical tests in an acid condensate environment.

642

V. Shinde et al.

Experimental VDA Test Description VDA tests consist of daily cyclic corrosive environment exposure alternating between direct liquid acidic solution exposure and acidic condensate exposure. The corrosive environment used for in this testing was a solution of nitric acid, sulfuric acid, and sodium chloride mentioned in aggressive system with pH 1.2 in VDA 230–214 [8]. The test set-up comprised a glass vessel to hold liquid test solution and a glass rod for hanging samples. The samples were hung to the rod using zip ties. Both the glass vessel and hanging rod were enclosed in a test container with lid and provision for inlet and exhaust air (referred as test chamber). Heaters were used to maintain test temperatures. The entire set-up was then placed inside chemical ventilation hood for safety. The test duration was 36 daily test cycles to simulate material conditioning equivalent to duration of the application the material will have to sustain for. Samples were collected after 6, 10, 15, 23, and 36 cycles, and post-corrosion analyses were performed using scanning electron microscopy (SEM) coupled with energy dispersive spectrometer (EDS). In addition to this, electrochemical tests were also conducted to further understand the corrosion behavior. During the course of cyclic corrosion test, the pH and temperature of the corrosive medium were maintained. Al 1100 foil material of 50-micron thickness was cut into 2.5 inch by 5-inch size for each sample. A total of 20 samples were tested which consisted of 4 replicates each for cycles 6, 10, 15, 23, and 36. Figure 1 illustrates the test procedure of cyclic corrosion test.

Electrochemical Testing Potentiodynamic polarization of the Al 1100 coupons were carried out using a Gamry™ Ref 600 potentiostat in conjunction with Gamry Framework software. This was done using a standard three electrode flat cell with a silver-silver chloride (Ag/AgCl) as the reference electrode, a graphite block as the counter electrode, and the Al 1100 as the working electrode with an exposed area of 2.84 cm2 . Polarization studies were done in acid condensate solution of 1000 ppm sulfuric acid (H2 SO4 ), 100 ppm nitric acid (HNO3 ), and 10 ppm hydrochloric acid (HCl). The open circuit potential (OCP) was monitored for an hour prior to polarization. The potentiodynamic scan was started at about −0.75 V below OCP and was stopped at about 1.5 V above OCP in the anodic region. The scan rate used was 0.5 mV/s for all the experiments. Corrosion potential and corrosion current data were plotted for each sample, and following equation was used to convert corrosion current values to corrosion rate in mills per year for each sample respectively.

Electrochemical Corrosion Tests of Aluminum 1100 Alloy …

643

Fig. 1 Illustration of cyclic corrosion test set-up

Electrochemical impedance spectroscopy (EIS) scans were performed with a frequency range of 10–2 –105 Hz at OCP with a sinusoidal voltage of 10 mV in acid condensate solution of 1000 ppm sulfuric acid (H2 SO4 ), 100 ppm nitric acid (HNO3 ), and 10 ppm hydrochloric acid (HCl).

644

V. Shinde et al.

Results and Discussion VDA Test Results Corrosion behavior of Al 1100 samples was qualitatively and quantitatively analyzed after test cycle 6, 10, 15, 23, and 36. An increase in surface discoloration was noticed visually in the regions where coupons were in direct contact with liquid test solution in semi-immersion phase (step 2) after each cycle. Due to proprietary reasons, images of post-tested coupons are not shown here. SEM was used to observe the surface of corrosion on a different sample replicate from the respective cycles. For local chemistry characterization, EDS was performed to estimate oxygen and aluminum weight percentages. Figure 2 shows SEM micrographs of samples after 6, 10, 15, 23, and 36 cycles compared to an unexposed baseline Al 1100 material. Increasing surface degradation was observed with increment in time of exposure of these coupons. The corrosion on the foil surface appeared to be localized possibly due to non-uniform coverage of the acid condensate droplets occurring during steam aging step. These micrographs were taken in back scattered mode which clearly revealed the elemental contrast with darker regions measuring higher oxygen content and brighter regions aluminum content. EDS results from spot 3 (darker regions) in Fig. 2 are summarized in Table 1, and they showed an increasing trend in oxygen concentration with a concurrent decrease in aluminum content till cycle 23. No significant difference in oxygen and aluminum levels was observed between cycles 23 and 36. Although through SEM–EDS a discernable surface degradation was noticed on these coupons as a function exposure time, to further investigate the corrosion behavior electrochemical tests, EIS and potentiodynamic polarization were also performed in acid condensate environment.

Fig. 2 SEM images of baseline aluminum sample and aluminum samples after respective acid condensate exposure cycles

Electrochemical Corrosion Tests of Aluminum 1100 Alloy …

645

Table 1 Elemental composition in weight % of samples after specified number of acid condensate cycle identified using EDS analysis Samples

Data point

Weight % C

O

Al

Fe

S

Baseline

3

4.25

1.44

94.15

0.76

Cycle 6

3

6.80

38.50

52.63

0.47

1.60

Cycle 10

3

2.06

50.15

46.44

0.49

0.86

Cycle 15

3

10.62

53.41

33.16

0.63

2.18

Cycle 23

3

2.11

63.68

34.21

Cycle 36

3

1.24

61.96

36.79

Electrochemical Test Results Potentiodynamic polarization scans of Al 1100 coupons from various exposure cycles were performed in acid condensate solutions and are shown in Fig. 3a, b. Day 6 and

Fig. 3 a and b Potentiodynamic polarization scans of Al 1100 coupons in acid condensate solution from various exposure cycles. c Summary plot of the open circuit potentials for the Al 1100 coupons. d Summary plot of the corrosion rates of the coupons

646

V. Shinde et al.

day 10 sample curves shifted to the left and had more positive potentials indicating a reduction in both anodic corrosion currents and an increase in corrosion resistance. Day 15, day 23, and day 36 samples on the other hand showed a reversing trend with an increase in anodic corrosion densities and a decrease in open circuit potentials (OCP), indicating an overall decrease in corrosion resistance from 15 to 36 days of acid condensate exposure cycles. This is better captured in the summary plot of the OCPs or zero current potentials and the corrosion currents of the samples in Fig. 3c, d. The OCP increases by about ~300 mV from the baseline sample with no exposure to 10 days of exposure cycles in the acid condensate and then sees a reverse trend of decreasing OCP by ~300 mV from day 10 to day 36 samples. The corrosion currents follow the same trend with a decrease by more than an order of magnitude from the baseline sample (5.56 × 10–6 A/cm2 ) to the day 10 exposure sample (2.81 × 10–7 A/cm2 ) and an increase by more than an order of magnitude from day 10 to day 23 samples (4.82 × 10–6 A/cm2 ). Bode plots (|Z| and phase angle) from EIS scans of the samples exposed to different cycles are shown in Fig. 4. The |Z| vs frequency plot clearly shows an increase in low frequency impedance (at 10–2 Hz) from the baseline to the day 10 samples and then a decrease by more than an order of magnitude from day 10 to day 23 samples. An evolution of the second time constant was seen on the phase angle vs frequency plots for all the exposed samples indicating the corrosion at the interface and possible formation of corrosion products such as aluminum hydroxide layer. An increase in the low frequency phase angle (at 10–2 Hz) up to 15 days of exposure followed by a decrease in phase angle was seen from 15 to 36 days of exposure which indicates an initial increase followed by a decrease in the corrosion resistance. Overall, the EIS shows a similar trend as the potentiodynamic scans. The EDS analysis and the SEM images showed formation of corrosion product likely in the form of aluminum oxide/hydroxide layer with increasing exposure cycles on the coupons. This layer formed from the aluminum alloy corrosion could have

Fig. 4 Bode plots of (a) |Z| and (b) phase angle of Al 1100 coupons in acid condensate solution from various exposure cycles

Electrochemical Corrosion Tests of Aluminum 1100 Alloy …

647

provided good barrier protection, thereby decreasing the corrosion rate of the coupons up to 10 days of exposure. However, with increase in time, the voluminous corrosion layer would tend to thicken and crack, exposing fresh base metal to the corrosive solution. This phenomenon is clearly seen in the SEM images from day 10 onwards in Fig. 2 and explains the high corrosion response in the electrochemical tests.

Conclusions Corrosion evaluation of Al 1100 was demonstrated through combination of cyclic immersion tests followed by electrochemistry. The main highlights are: 1.

2.

Non-uniform corrosion on Al 1100 coupons was attributed to formation of acid condensate droplets during the steam-aging step of VDA230-214. The increasing trend of oxygen concentration and corresponding decrease in aluminum concentration suggest the corrosion product layer thickness increased with an increase in exposure times. Corrosion product formed on the surface demonstrated a greater degree of nonadherence with increased exposure times. This led to an order of magnitude increase in corrosion rates beyond 10 exposure cycles.

References 1. Davis JR (2001) Alloying: understanding the basics. ASM International. https://doi.org/10. 1361/autb2001p351 2. Giri A, Anugula G, Srivastava V, Adhikari S (2012) Corrosion performance evaluation of aluminum alloys for automotive applications. Conference: 16th national congress on Corrosion control 3. Duchesne D (2018) Aluminum corrosion: why it happens. https://www.wileymetal.com/cat egory/engineering-services/ 4. Clark N et al (2006) Evaluation of crankcase emissions abatement device. Center for Alternative Fuels Engines & Emissions, West Virginia University report to New Condensator, Inc. 5. Addy Majewski W, Jääskeläinen H (2019) Exhaust particulate matter. DieselNet 6. Tillmann W, Walther F, Manka M, Schmiedt A, Wojarski L, Eilers A, Wilhelm Reker D (2019) Investigations of the corrosion damage process of the brazed joint AISI 304L/BNi-2. Results Phys 1245–1252 7. Schmiedt A, Lingnau L, Manka M, Tillmann W, Walther F (2018) Effect of condensate corrosion on tensile and fatigue properties of brazed AISI 304L stainless steel joints using gold-base filler metal. Proc Struct Integr 22–27 8. German Association of the Automotive Industry (VDA) (2018) VDA 230–214 resistance of metallic materials to condensate corrosion in exhaust-gas-carrying components

648

V. Shinde et al.

9. Usman BJ, Scenini F, Curioni M (2020) Corrosion testing of anodized aerospace alloys: comparison between immersion and salt spray testing using electrochemical impedance spectroscopy, 167. The Electrochemical Society by IOP Publishing Limited 10. ASTM G 102-89(2015)e1. Standard practice for calculation of corrosion rates and related information from electrochemical mesaurements

Electrochemical Study of Stainless Steels in Diesel Exhaust Fluid (DEF) and Simulated Exhaust Acid Condensate Environments Anusha Chilukuri, Michael Warwick, and Gaurav Argade

Abstract Electrochemical testing at room temperature was carried out in DEF (Diesel Exhaust Fluid) and simulated acid condensate environments on austenitic (AISI 304) and ferritic stainless steels (AISI 409 and 439). It was found that all the stainless steels are spontaneously passive in DEF solution at open circuit potential (OCP) with a large passive range of ~1000 mV above OCP. No evidence of pitting of the stainless steels was seen in this environment at higher anodic potentials. However, 409 stainless steel showed active dissolution in acidic condensate solutions with corrosion currents higher than two orders of magnitude when compared to 439 and 304 stainless steels. Each of the stainless steels was subjected to thermal exposure at 500, 600, and 700 °C for several hours to simulate the long-term high temperature exposure of the aftertreatment systems. Subsequent electrochemical tests indicated that the corrosion currents and the passivation currents in acidic condensate solutions increased with an increase in exposure times at lower temperature of 500 °C. Similarly, the currents also increased when exposed to a higher temperature of 700 °C with a relatively shorter exposure time of 5 h. Keywords Stainless steels · Potentiodynamic polarization · Diesel Exhaust Fluid (DEF) · Open circuit potential · Corrosion current

Introduction The high combustion temperatures of the IC engine and the introduction of different systems for exhaust after treatment affect the choice of material for those systems. A selective catalytic reduction system is employed in heavy duty engines to reduce the NOx level from the exhaust gases. A solution of urea in water is used as an ammonia A. Chilukuri (B) · G. Argade Materials Science and Technology, Cummins Technical Center, Cummins Inc., 1900 McKinley Avenue, Columbus, IN 47201, USA e-mail: [email protected] M. Warwick Cummins Emissions Solutions, 1801 US Highway 51-138, Stoughton, WI 53589, USA © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_62

649

650

A. Chilukuri et al.

source for the reduction of NOx gases to N2 and H2 O. A complex environment consisting of high temperature exhaust gases along with urea and its decomposition products restricts the material choice to stainless steel grades. Several researches conducted exposure testing that took hundreds of hours to induce damage in stainless steels [1–10]. Saedlou et al. injected DEF drop by drop on the sample surface subjected to a heat cycle up to 700 °C, and the total amount of microstructural degradation was measured in the cross-section [1]. The authors concluded that a minimum Cr content of 16 wt% was necessary for resistance to hot urea corrosion. Nockert et al. used exhaust pipes subjected to diesel exhaust gases and urea spray where temperature was about 550 °C with intermittent low temperatures [2]. All candidate materials are assessed after a long exposure to hot moist air. The authors concluded that 304L suffered severe mass loss compared to 904L. They also found that austenitic stainless steel 304L performed considerably better than the different ferritic alloys considered in the study [3]. In this paper, we describe a different testing scenario. The assessment is not weight loss, or pitting depth, but electrochemical response. These properties will indicate whether the material after a long exposure is passive and unlikely to corrode further, or active and prone to corrode more. The measured corrosion current can be used to estimate how quickly the corrosion will advance. A second advantage to this approach is that it separates the inevitable exposure to exhaust gas itself from special conditions such as DEF wall wetting and exhaust gas acid condensation, which are very design and duty cycle dependent.

Experimental Electrochemical Testing Three different kinds of stainless steel were used to test their corrosion resistance in various electrolytes such as DEF solution, an acid cocktail solution, and a mixture of DEF and acid cocktail solutions. Table 1 shows the chemical composition of the stainless steels used for the electrochemical tests in this study. Potentiodynamic polarization of the stainless steel coupons were carried out using a GamryTM Ref 600 potentiostat in conjunction with the Gamry Framework software. All the samples were polished to 1200 grit before any electrochemical or exposure testing. This was Table 1 Chemical composition of the stainless steels used in this work Chemistry

C

Si

Mn

Cr

Ni

Ti + V

Fe

AISI 409

0.01

0.39

0.41

11.29

0.19

0.30

Balance

AISI 439

0.02

0.29

0.23

17.49

0.19

0.42

Balance

AISI 304L

0.02

0.40

1.41

18.17

8.01

0.08

Balance

Electrochemical Study of Stainless Steels in Diesel Exhaust Fluid (DEF) …

651

done using a standard three electrode flat cell with a silver-silver chloride (Ag/AgCl) as the reference electrode, a graphite block as the counter electrode, and the stainless steel as the working electrode with an exposed area of 2.84 cm2 . The open circuit potential (OCP) was monitored for an hour prior to polarization. The cyclic polarization scan was started at about −0.3 V below OCP, and the scan was reversed at 1.5 V versus Ag/AgCl reference electrode and stopped when it reaches down to −0.2 V versus OCP. DEF solution is alkaline, and the pH is measured to be in the range of 9–10. The acid cocktail solution is measured to have a pH between 1.5–2.0.

Results and Discussion The cyclic polarization scans performed in the DEF solution are shown in Fig. 1a. All the three stainless steels are spontaneously passive in DEF solution at open circuit potential and have both similar corrosion currents, passivation, or anodic currents. The passivation is seen from the open circuit potential of about −200 mV up to 830 mV versus Ag/AgCl reference electrode. The stainless steels did not show any

Fig. 1 Polarization plots of stainless steels in (a) DEF solution, (b) simulated acid condensate solution, (c) summary plot showing the current densities at OCP and 800 mV versus Ag/AgCl and the Ecorr or OCP for the three alloys. AC stands for acid condensate solution

652

A. Chilukuri et al.

evidence of pitting at higher anodic potentials from visual inspection post testing. They all show similar characteristic transpassive potentials. The increase in current beyond the transpassive region is from the oxygen evolution and dissolution of the oxide layer. It seems that the stainless steel protected by chromium oxide is not attacked by DEF solution. The reverse scan however showed a difference in the zero current potentials, with 409 showing a slightly lower reverse zero current potential indicating that the passive film formed upon anodic polarization isn’t as noble as that of 304L and 439. The cyclic polarization scans performed in the acid condensate solution are shown in Fig. 1b. 409 showed about two orders of magnitude higher corrosion currents (~675 µA) than 304 and 439 stainless steels (~2–3 µA). Also, the OCP of 409 in the acidic condensate solution is about 200 mV lower than that of the 304 and 439 stainless steels. This indicates that 409 stainless steel has an unstable passive layer in acidic condensate solutions and is subject to uniform dissolution. This is expected with the low chromium content of 409 stainless steel. However, 304 and 439 stainless steels seem to still be passivated at open circuit potential. This indicates that 409 may perform poorly in real life conditions and is best suited to short warranty periods where useful life is reached before any corrosion damage can progress to failure. Figure 1c summarizes the corrosion parameters for the three alloys comparing the OCP and corrosion currents at OCP and at an anodic potential of 800 mV versus Ag.AgCl. Stainless steels 439 and 304L were subjected to oxidation in a high temperature furnace at 500, 600, and 700 °C for a certain exposure time, and the oxidized samples were then subject to potentiodynamic polarization in DEF and acidic condensate solutions. Figure 2a, b show the potentiodynamic polarization curves of oxidized 304L and 439 SS respectively in acidic condensate solutions. 304L oxidation at 500 °C for 5 h significantly reducing the anodic corrosion currents and increased the OCP to noble potentials. An increase in oxidation time at 500 °C for 24 h increased the anodic current densities relative to the 5 h oxidation time at the same temperature. At 500 °C, 304L remained to be spontaneously passive in acidic condensate solutions. However, with increase in oxidation temperatures to 600 °C for 24 h and 700 °C for 5 h increased the corrosion currents and the anodic currents where 304L was subject to active dissolution. On the other hand, oxidized 439 was undergoing active dissolution in acidic condensate solutions at all the three temperatures. This shows that the thermal oxidation in the furnace for the stainless steels modifies the protective chrome oxide layer to a defective oxide layer allowing the acidic solution to attack the surface. All the oxidized steels remained spontaneously passive in DEF solution. Figure 2c shows the potentiodynamic polarization curves in DEF solution of bare 304L and 439 and pre-exposed to furnace at 600 °C for 24 h. Oxidation of the samples did not make the steels actively corroding in this environment. The result is expected as the environment in this test is alkaline and should not affect the passivity of stainless steels per the Pourbaix diagram. Oxidation of 304L did not seem to have any impact on the anodic kinetics in DEF solution. However, oxidized 439 seemed to have lesser passivation current in DEF solution, indicating that the oxide layer generated from 439 oxidation has provided more protection in alkaline solution than that of the bare

Electrochemical Study of Stainless Steels in Diesel Exhaust Fluid (DEF) …

653

Fig. 2 Potentiodynamic polarization curves of (a) 304L and (b) 439 SS in acidic condensate (AC) solutions of the samples pre-exposed to thermal oxidation at 500 °C for 5 h and 24 h, 600 °C for 24 h, and 700 °C for 5 h. Potentiodynamic polarization curves in (c) DEF solution of bare 304L and 439 at room temperature (RT) and pre-exposed to thermal oxidation at 600 °C for 24 h

metal in the same environment. The data presented in Fig. 2 demonstrates how the behavior of steels varies in acidic condensates from alkaline DEF solutions. Figure 3a summarizes the effect of oxidation of the two stainless steels 304L and 439 at 500 and 700 °C for 5 h. The plot shows that 304L exhibited two orders of magnitude increase in corrosion current and anodic current at 800 mV versus Ag/AgCl with an increase in oxidation temperature. This is in conjunction with the fall of the open circuit potential from 600 mV for the 500 °C oxidized sample to -400 mV to the 700 °C oxidized coupon. Although the stainless steel used is an low carbon grade, the oxidation temperature falls in the sensitization range and the data show that the protective performance of the oxide layer in acidic solutions degrades with an increase in the oxidation temperature exposed. 439 shows poor oxide layer protection or much higher corrosion kinetics even at a lower temperature exposure of 500 °C for 5 h when compared to 304L that was exposed to similar conditions. 439 also shows a slight drop in OCP of about ~ 75 mV and about half an order of magnitude in corrosion current with an increase in oxidation temperature. This could be attributed to the absence of Ni in the less passive oxide layer for the ferritic stainless steels.

654

A. Chilukuri et al.

Fig. 3 Parameter a Column chart—5-h oxidation time and b line chart—24-h oxidation time for 500 and 600 °C, whereas 5-h oxidation time for 700 °C for alloys 304L and 439 in acidic condensate solutions. Both the plots have corrosion current, anodic current at 800 mV versus Ag/AgCl, and the OCP or Ecorr at 800 mV versus Ag/AgCl being compared for the alloys

Figure 3b shows the parameters as a function of temperature and 24-h oxidation time for 500 ºC and 600 ºC, whereas 5-h oxidation time for 700 ºC. 439 shows a decreasing trend in Ecorr or OCP with increase in temperature. 304L shows a sharp decrease in OCP from 500 to 600 ºC and continues to decrease at 700 ºC. Similarly, the corrosion currents increase sharply by three orders of magnitude for 304L from 500 ºC to 600 ºC oxidation temperature. Also, 439 showed two and a half orders of magnitude higher corrosion current in acidic condensate solutions when compared to 304L at 500 ºC–24-h oxidation condition. Overall, the corrosion current and anodic current at 800 mV versus Ag/AgCl are approximately similar for 600 ºC—24-h oxidation and 700 ºC 5-h exposure for both the alloys.

Conclusions (a)

(b)

(c)

All the three bare stainless steels were spontaneously passive in DEF solution at open circuit potential and exhibited much similar corrosion kinetics. None of the steels tested showed any evidence of pitting at higher anodic potentials. Bare 304L and 439 stainless steels showed spontaneous passivation in acidic condensate solutions. However, 409 showed about two orders of magnitude higher corrosion currents than 304 and 439 stainless steels and showed uniform dissolution. This is due to the low chromium content of 409 stainless steel (~11 wt%) when compared to 17–19 wt% chromium containing 304L and 439 stainless steels. 304L showed an increase in passivity in acidic condensate solutions with an increase in OCP to noble potential and more than an order of magnitude reduction in corrosion current upon oxidation at 500 ºC for 5 h. However increasing the oxidation time to 24 h at 500 ºC or increasing the oxidation temperature

Electrochemical Study of Stainless Steels in Diesel Exhaust Fluid (DEF) …

(d)

(e)

655

to 600 and 700 ºC showed a sharp reduction in OCP to more active potentials and a steep increase in corrosion current densities by almost three orders of magnitude when compared to the 500 ºC–5 h kinetics in acidic condensate solutions. This could be attributed to the partial loss of corrosion resistance of the austenitic stainless steels when exposed to the sensitization temperatures where the loss is known to increase with an increase in exposure temperature or exposure time at a certain temperature. 439 stainless steel lost its spontaneous passivity upon oxidation at any of the three temperatures and showed active dissolution in acidic condensate solutions. This indicates that the ferritic stainless steel could have a less protective oxide layer form upon oxidation which is no longer passive in acidic solutions. Oxidation of stainless steels does not affect DEF corrosion kinetics, and the stainless steels remain to be spontaneously passive in these alkaline solutions.

References 1. Saedlou S, Santacreu P, Leseux J Suitable stainless steel selection for exhaust line containing a selective catalytic reduction (SCR) system. SAE 2011 World Congress & Exhibition 2. Nockert J, Norell M Corrosion at the urea injection in SCR-system during component test. Mater Corros 64(1):34–42 3. Nockert J, Nyborg L, Norell M Corrosion of stainless steels in simulated diesel exhaust environment with urea. Mater Corros 63(5):388–395 4. Cao Y, Norell M (2013) Role of nitrogen uptake during the oxidation of 304L and 904L austenitic stainless steels. Oxid Met 80(5–6):479–491 5. Nichols DE et al (1991) The effect of conditioning agents on the corrosive properties of molten urea. In: 202nd National Meeting of the American Chemical Society, New York 6. Floyd R, Kotrba A, Martin S, Prodin K (2009) Material corrosion investigations for urea scr diesel exhaust systems. SAE Technical Paper 2009-01-2883. https://doi.org/10.4271/2009-012883 7. Bergamo T Effect of nitridation on high temperature corrosion of ferritic stainless steel, Diploma work No. 107/2013, Department of Materials and Manufacturing Technology, Chalmers University of Technology, Sweden 8. Patterson W (1978) Materials, design and corrosion effects on exhaust-system life. SAE Technical Paper 780921. https://doi.org/10.4271/780921 9. Wei Z (2013) Characterization of materials for exhaust systems under combined mechanical and corrosive environment. SAE Technical Paper 2013-01-2420. https://doi.org/10.4271/201301-2420 10. Hirasawa J, Ujiro T, Satoh S, Furukimi O (2001) Development of high corrosion resistant stainless steels for automotive mufflers based on condensate corrosion test and field investigation. SAE Technical Paper 2001-01-0640

Part XVIII

Advances in Titanium Technology

A Review on Impact Resistance of Partially Filled 3D Printed Titanium Matrix Composite Designed Aircraft Turbine Engine Fan Blade Shade Rouxzeta Van Der Merwe, Daniel Ogochukwu Okanigbe, Dawood Ahmed Desai, and Glen Snedden Abstract Price competition forces airline operators to save costs through lowering operating costs. One possibility of lowering these operating costs is reduced fuel consumption of aircrafts. The use of titanium to design lightweight aircraft components can help achieve this goal while the use of 3D printing technology to fabricate titanium components can further help to achieve this goal, because of its speed, flexibility, and cost advantage. However, to achieve maximum strength of 3D printed aircraft components, the infill is often 100% filled, thus, implying higher costs (i.e. in terms of time and material) and heavier components, which defeats the goal of maximizing profit in aerospace industry. Hence, demand for novel scientific idea in this area of research (AR) is required, for positive impact on overall economics of airline operations. Prompting a critical review of past and current publications under the following sub-headings: effect of infill density on mechanical properties of 3D printed titanium matrix composite; effect of material selection on impact tolerance of 3D printed components; additive powder-based technology and sophisticated mixed-material composites; simulation study for impact tolerance of 3D fabricated components from titanium, its alloys and composite. Based on outcome of review, it was concluded that a gap of knowledge exists in the area of determining effect of infill density on impact tolerance of 3D printed titanium components. Aligning with this thought, it was recommended that future research work should focus on predicting

S. R. Van Der Merwe (B) · D. A. Desai Faculty of Engineering and the Built Environment, Department of Mechanical Engineering, Tshwane University of Technology, Mechatronics and Industrial Design, Pretoria 0183, Republic of South Africa e-mail: [email protected] G. Snedden University of Kwa-Zulu-Natal, Aerospace Systems Research Group, Durban, Republic of South Africa D. O. Okanigbe Faculty of Engineering and the Built Environment, Department of Chemical Metallurgical and Materials Engineering, Tshwane University of Technology, Mechatronics and Industrial Design, Pretoria 0183, Republic of South Africa © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_63

659

660

S. R. Van Der Merwe et al.

the impact tolerance of a partially filled 3D printed titanium matrix composite for aerospace application. Keywords Titanium matrix composite · Fused Deposition Modelling (FDM) · Additive manufacturing · 3D printing · Raster angle · Infill density

Introduction Modern manufacturing methods necessitate the development of products with quick turnaround time to maintain product quality. This has sparked the development of new production techniques and materials, [1–3] one of these methods is rapid prototyping (RP). Additive manufacturing, 3D printing used to define technologies that manufacture parts without finishing being required by layering the material, to manufacture the final product, like in traditional manufacturing procedures. Additive manufacturing permits the fabrication of complicated shaped parts by uploading the computer assisted design (CAD) models to the fabricating machine. The key benefits of this approach include cheap maintenance costs, easy material replacement, a low operating temperature, supervision-free operation, and small size [4]. 3D printing technology is being embraced by manufacturers as the industry standard for product development [5, 6]. In a computer-integrated industrial setting, 3D printing simplifies production. A significant improvement in production processes can be obtained by integrating 3D printing with concurrent engineering [7]. Fused Deposition Modelling is the most widely used 3D printing process (FDM). During this method, objects are formed by layering melted filament (generally usually Acrylonitrile Butadiene Styrene [ABS] or poly lactic acid [PLA] that cools to form the solidified final product. The plastic filament is melted and deposited on the bed, which moves in the z-direction to produce the final 3D object, by an extruder capable of moving in the x–y plane. FDM models reduce waste and make the process more environmentally friendly due to its capacity to be recycled [8]. The FDM technique has the unique ability to adjust the mechanical characteristics, density, and porosity of the created object at a local level [9]. FDM can be used to make working parts as well as prototypes. A number of enhancements are required to fully develop the FDM into a manufacturing tool. Dimensional control, surface quality, and accurate tolerances are only a few of the requirements. Additionally, a wider range of polymers should be provided, and the mechanical qualities of prototyped parts should be improved to ensure their integrity throughout operation [10]. The values of process parameters in additive manufacturing processes can be more significant as compared to the properties of the part material, which is contrary to most of the manufacturing processes. Parts having same geometry but fabricated using different sets of process parameters will have entirely different properties, e.g. strength [11, 12] or accuracy [13]. Each combination of process parameters, namely, bed temperature, layer thickness and infill pattern, and infill density, will produce a

A Review on Impact Resistance of Partially Filled 3D …

661

Fig. 1 Aircraft turbine engine fan blade. Source Google image

different part structure, ultimately resulting in different values of mechanical properties. Hence, this paper will present reports on review of publications in literature connected with this scientific idea, in order to establish where the gap of knowledge lies and propose future research focus on the basis of the identified gap of knowledge.

Theoretical Background Aircraft Turbine Engine Fan Blade The fan blades (Fig. 1) are connected to the fan disk, which is rotated by a shaft power-driven by a gas turbine [14]. The majority of propulsive thrust in modern passenger aircraft is generated by fans powered by gas turbines.

3D Print Infill Density The “fullness” of a component’s interior is defined as its infill density in slicers, this is typically defined as a percentage between 0 and 100 (Fig. 2), with 0 indicating a hollow part and 100 indicating a completely solid part. As you can imagine, this greatly impacts a part’s weight: The fuller the interior of a part, the heavier it is. Besides weight, print time, material consumption, and buoyancy are also impacted by infill density. So, too, is strength, albeit in combination with many other elements such as material and layer height [15].

662

S. R. Van Der Merwe et al.

Fig. 2 Infill density of 3D. Source Google image

Impact Resistance of 3D Printed Components Foreign object impact (FOI) has been one of the most significant concerns in the design of engine fan blades [16]. To manage the risk related to FOI, impact resistance of materials used for 3D print of engineering components is evaluated in accordance with safe regulation for FOI by turbine engine fan blade as shown in Fig. 3 [17].

Review of Past and Current Publications: Impact Resistance of Partially Filled 3D Printed Titanium Matrix Composite Effect of Infill Density on Mechanical Properties of 3D Printed Titanium Matrix Composite There is limited literature published on 3D printed components made of titanium matrix composite that focuses on the effect of infill density on mechanical characteristics [18]. However, it has been noted that infill density has a significant impact on the material properties of printed samples, among other parameters [19, 20]. The sintering step in the 3D printing process involves heating the parts uniformly to just below their melting point, resulting in fully dense parts without the residual stresses that laser-based systems create [18]. The influence of fluidity on the deposition of Ti6Al-4 V rod for the fabrication of scaffolds was investigated [21]. A lower viscosity of the melt causes distortion of the deposited rod, whereas a higher viscosity of the material creates a higher resistance to flowing, both of which affect the quality of components produced. As illustrated in Fig. 4, where Ti-6Al-4 V powder was used as the solid loading for the construction of scaffolds, the powder concentration is a

A Review on Impact Resistance of Partially Filled 3D …

663

Fig. 3 Impact resistance process of 3D braided composite fan blade taken by high-speed camera [17]

Fig. 4 Effects of infill density of powder on the stability of scaffolds. a Low infill density reveals that the fiber is deformed 64% (volume fraction) Ti-6Al-4 V powder used, b Optimal infill density 66% (volume fraction), c when infill density of Ti6Al4V powder is at 68% volume fraction Ti-6Al-4 V, this results in no adhesion and attachment between the layers [21]

664

S. R. Van Der Merwe et al.

significant component that impacts the viscosity of the melt. If the powder concentration is too low, gravity induced flow occurs after deposition, as shown in Fig. 4a, b, c.

Effect of Material Selection on Impact Tolerance of 3D Printed Components In this report, strength of materials refers to materials with capacity to withstand high impact and repeated use, and are durable. Some filaments are more impact resistant, while others are more fatigue resistant or durable, and still others are a combination of the three. The following are Common Impact Resistant 3D Printing Materials and these filaments differ on the basis that some are more impact resistant, while others are more fatigue resistant.

Acrylonitrile Butadiene Styrene (ABS) ABS is a stiff and impact-resistant thermoplastic material. It’s a common material that’s primarily utilized for engineering and technical prints. ABS is relatively inexpensive when compared to other fiber filaments, making it an excellent alternative if you’re on a tight budget yet still require high strength. ABS is not only impact and fatigue resistant, it is also heat and water resistant. ABS is the ideal 3D printing material for a moving item with high mechanical stress or for functionality that requires high stress. ABS is also a suitable option when the print will be an end-use product, as it has a smooth surface finish [6].

Thermoplastic Polyurethane (TPU) TPU famous for its plasticity, this characteristic makes it durable material 3D printing. This material is also resistance to impact, wear and tear, chemicals, and abrasion. TPU is an ideal 3D printing material to use for shock absorption. With the ability to extend up to 4.5 times its original size without breaking, TPU is your go-to choice for any prints that require high levels of flexibility and strength. TPU is commonly used to print parts like wheels, springs, shock absorbers, and other flexible objects [22, 23].

Polyethylene Terephthalate Glycol-modified (PET-G) Polyethylene terephthalate glycol-modified (PET-G) material is best for 3D printing beginners to practice with due to that PET-G is both strong and easy to print. PET-G

A Review on Impact Resistance of Partially Filled 3D …

665

advantages are that it’s easier to print than ABS, yet stronger and more technical than ABS. Additionally, it possesses better impact resistant, abrasion resistant, and also features some flexibility compared to ABS. Featuring a lustrous appearance and high strength, PET-G is an ideal option for end-use parts. It’s commonly utilized to make prototypes, low-stress components, and protective casings [24].

Polyamide (PA) PA is a highly durable and heat-resistant 3D printing material. This semicrystalline material is ideal for prints that will be subjected to high usage. PA is suitable material to manufacture moving components reason been it has high strength, excellent impact resistance, and is fatigue resistant. PA is widely used to make washers, gears, jigs, sliding components, and high-fatigue item. It is one of the most durable 3D printing materials on the market which is resistant to abrasion, impact, and heat. PA outperforms ABS in terms of impact strength and flexibility [25].

High-Temperature Polyamide Carbon Fiber Reinforced (PAHT CF15) PAHT CF15 is another popular 3D printing material in the automobile sector. This filament delivers a powerful punch and is the most durable 3D printing material on our list. PAHT CF15 has exceptional strength, heat resistance, stiffness, and impact resistance. Although it might be tempting to choose the strongest material you can find straight once, this isn’t always the best option. As PAHT CF15 being consider as strong 3D printing filament that is ideal for tough objects or parts. This material is suitable to substitute metal and manufacture parts that can be exposed to high temperature also high stresses [8].

Polypropylene (PP) Polypropylene (PP) is one of the lightweight and flexible polymers. The advantage of PP is that it is durable 3D printing material that is both impact and fatigue resistant. Its sturdiness and impact resistance, along with its airy nature, make it an excellent choice for items that will be used often, such as packing, pipes, and joints. The most well-known property of PP is its high resistance to most chemicals, including alkalis, acids, and organic solvents [7].

Polypropylene 30% Glass Fiber (PP GF30) PP GF30 is similar to PP, with small improvements. The glass fiber in PP GF30 permits this 3D printing material to have excellent levels of strength and stiffness while remaining light and chemically resistant. This is one of the most durable 3D

666

S. R. Van Der Merwe et al.

printing materials available, and it’s widely utilized in the automobile and aerospace sectors. One of PP GF30’s most outstanding features, aside from being lightweight, robust, and stiff, is its great resistance to adverse environments, such as chemicals and weather conditions. If a 3D print will be exposed to outside environment or must endure extreme temperatures while retaining structural integrity, PP GF30 is the best material option [14].

Additive Powder-Based Technology and Sophisticated Mixed-Material Composites More complicated and sophisticated mixed-material composites can now be produced because to advancements in additive powder-based technologies. There is already a plethora of literature on this subject, covering the usual design of cellular systems. (strut diameter and pore size, cell architecture, topology optimization), mechanical properties (orientation/location dependence, microstructure/phase investigation, fracture toughness, tensile strength, fatigue, etc.), metal powders or alloys (thermal deformation, solidification, etc.), printing parameters and industrial applications, [26–31], however, the literature on the behavior of these materials under static or dynamic impact, on the other hand, is still relatively limited. Dekhtyar et al. [32] investigated both additively manufactured and heat-treated Al-Si-10 Mg positioning in a horizontal and vertical axis as well as trans-track and inter-layer fracture being impacted. Large-scale aerospace components’ porosity and mechanical characteristics are also studied under impact [33]. The microanalysis of the form and composition of Ti-6Al-4 V was examined using ultrasonic impact treatment in the air for 30–150 s, which results in the development of an amorphous oxidation layer mixed with Fe, Al, and V separation [34]. Using ultrasonic impact treatment, several sliding impacts of Ti-6Al-4 V are produced, resulting in increased fatigue strength and a longer lifespan [32].

Impact Tolerance of 3D Printed Component Using Titanium Matrix Composite Impact-resistant (IR) composites, of several heavy objects with a lower penetration effect, have been developed in recent years with emphasis on lighter constructions, comprised of metals, metal–metal hybrids, ceramics, ceramic–metal (cermet), and fiber composites. Most components, including as helmets, body armor, and ballistic vests, are comprised of synthetic fibers like Kevlar and metals like steel and titanium. These materials’ primary function is to disperse and absorb the kinetic energy of the penetrating item upon impact. Ceramic and non-metallic composites or hybrids incorporating various materials, such as ceramic tiles, may be easily incorporated into

A Review on Impact Resistance of Partially Filled 3D …

667

the shield. In a setup like this, ceramics help achieve efficiency in such a construction due to their great hardness, compressive strength, and low density. Ceramics that are commonly known are Alumina (Al2 O3 ), Boron Carbide (B4 C), Silicon Carbide (SiC), and Boron-Silicon Carbide. However, they have a number of disadvantages, including poor tensile strength (low bending strength) and a high melting point (high-temperature sintering equipment is required). They also have lesser ductility than metals like Ti and Al, although this can be compensated for by combining beneficial characteristics in a composite structure. Metal-ceramic plates in ballistic armors provide state-of-the-art armor piercing (AP) protection by shattering projectile tips, dampening impact energy, and interrupting the propagation of penetrating fragments. In Zheng et al. [35] ballistic impact tests are performed on Ti-6Al-4 V titanium alloy plates that have been rolled and annealed using various heat treatments (12.70 mmAP bullet). The findings of a microscale failure mechanism study revealed brittle fragmentation to ductile hole development dependent on the heat treatment settings. Whereas Holmen et al. [36] investigated when area mass is taken into account, AA6070 aluminum alloy, heat treated in various tempers, suggests that aluminum alloys have equivalent perforation resistance (7.62 mm AP bullet) and strength of the steels. The present study adopts this concept and uses Ti and Al alloys in tandem to increase the lightness and strength of metal-based structures, provided that the metals consolidate, embed each other, and assemble in a layer-by-layer manner.

Simulation Study for Impact Tolerance of 3D Fabricated Components from Titanium, Its Alloys and Composite The requirement for accurate modeling and simulation tools has become critical since the cost of experiments to evaluate the collision, explosion, and destruction resistance of novel materials is generally rather costly. As a result, finite element analysis (FEA) is convenient approved for modeling and simulation that is best employed to determine the potential, and ability of lattice structure when it is subjected to armor piercing (AP). Thus, suitable validate stress, plastic deformation, and impact absorption properties of materials with particular reference to titanium matrix composite-based lattice. Furthermore, simulations have been found in some studies to be beneficial in the development of a process. Consider [37], where an explicit dynamic FEA was developed in ABAQUS to compare numerical predictions with experimental findings. In [38] utilized a software (LS-DYNA) to evaluate penetration depth and construct thin, light, and affordable metal-ceramic armor. The monolithic and multilayer plate simulations against different shank diameters in [39] revealed that increasing projectile diameter increased structural resistance. This significant finding indicated that monolithic systems were more resistant. In the study conducted by [18], titled “Lightweight 3D printed Ti-6Al-4 V mixed Al-Si-10 Mg hybrid composite for impact resistance

668

S. R. Van Der Merwe et al.

Fig. 5 Samples subjected to a penetrating object (6.4 g) simulated in ANSYS (cross-sectional view): A Solid Al-Si-10 Mg (4.7 g), B Solid Ti-6Al-4 V (7.8 g), C Uniform 3 mm cell size Ti6Al-4 V embedded inside Al-Si-10 Mg (5.7 g), D Uniform 1.5 mm cell size Ti-6Al-4 V embedded inside Al-Si-10 Mg (7.6 g), E Longitudinal gradient Ti-6Al-4 V embedded inside AlSi10Mg (4.9 g), F Radial gradient lattice Ti-6Al-4 V embedded inside Al-Si-10 Mg (4.8 g) structure. Depth of penetration for samples A to F are approximately ≈7.0, 4.5, 6.5, 4.0, 5.5 and 6.0 mm, respectively (adapted from Rahmani, Antonov and Brojan [18])

and armor piercing shielding”. According to the authors, it is cheaper for local component reinforcing, but it is not appropriate for minimizing lattice displacement or distortion of the outer-most layer of the cylinder. As a result, while the finer (smaller cell size) uniform lattice (Fig. 5D) is optimal for achieving the lowest penetration depth, the longitudinally graded lattice (Fig. 5E) is better for preventing distortion and deformation. The depth of piercing and weight values for samples as shown in Fig. 5a–f are approximately (7.0 mm, 4.7 g), (4.5 mm, 7.8 g), (6.5 mm, 5.7 g), (4.0 mm, 7.6 g), (5.5 mm, 4.9 g), and (6.0 mm, 4.8 g), respectively.

Conclusions At this moment in time, the progress that has been made using 3D print has led to a new way of building engineering components. In the fabrication of engineering components using 3D printing, it is often a common practice to employ a variety of

A Review on Impact Resistance of Partially Filled 3D …

669

infill densities with the goal of reducing printing cost which entails time and material consumption. However, it is not well understood how infill densities affect energy absorption properties of the engineering component fabricated. Hence, appraising the energy absorption property of the engineering component fabricated employing this technology is considered important. In line with this, it is recommended that future investigation should focus on determining the impact tolerance of partially filled 3D printed titanium matrix composite for the design of aircraft turbine engine fan blade. Acknowledgements The following institutions are acknowledged for their contributions to the success of this review paper: • Tshwane University of Technology, Faculty of Engineering and the Built environment, Department of Mechanical Engineering, Mechatronics and Industrial Design, Pretoria, Republic of South Africa. 0183. • Tshwane University of Technology, Faculty of Engineering and the Built environment, Department of Chemical Metallurgical and Materials Engineering, Pretoria, Republic of South Africa. 0183. • University of Kwa-Zulu-Natal, Aerospace Systems Research Group, Durban, Republic of South Africa.

References 1. Geethapriyan T, Kalaichelvan K, Muthuramalingam T (2016) Multi performance optimization of electrochemical micro-machining process surface related parameters on machining Inconel 718 using Taguchi-grey relational analysis. La Metallurgia Italiana 4:13–19 2. Muthuramalingam T, Mohan B (2014) Application of Taguchi-grey multi responses optimization on process parameters in electro erosion. Measurement 58:495–502 3. Singh N, Mir IUH, Raina A, Anand A, Kumar V, Sharma SM (2018) Synthesis and tribological investigation of Al-SiC based nano hybrid composite. Alex Eng J 57(3):1323–1330 4. Bernard A, Fischer A (2002) New trends in rapid product development. CIRP Ann 51(2):635– 652 5. Abdullah MIHC, Abdollah MFB, Amiruddin H, Tamaldin N, Nuri NRM (2013) Optimization of tribological performance of hBN/AL2O3 nanoparticles as engine oil additives. Proc Eng 68:313–319 6. Chua CK, Leong KF, Lim CS (2010) Rapid prototyping: principles and applications (with companion CD-ROM). World Scientific Publishing Company 7. Vasylyev MA, Chenakin SP, Yatsenko LF (2016) Ultrasonic impact treatment induced oxidation of Ti6Al4V alloy. Acta Mater 103:761–774 8. Górski F, Wichniarek R, Kuczko W, Zawadzki P, Bu´n P (2015) Strength of ABS parts produced by Fused Deposition Modelling technology–a critical orientation problem. Adv Sci Technol Res J 9(26):12–19 9. Li L, Sun Q, Bellehumeur C, Gu P (2002) Composite modeling and analysis for fabrication of FDM prototypes with locally controlled properties. J Manuf Process 4(2):129–141 10. Gajdoš I, Slota J (2013) Influence of printing conditions on structure in FDM prototypes. Tehniˇcki vjesnik 20(2):231–236 11. Ahn SH, Baek C, Lee S, Ahn IS (2003) Anisotropic tensile failure model of rapid prototyping parts-fused deposition modeling (FDM). Int J Modern Phys B 17(08n09):1510–1516 12. Bellini A, Güçeri S (2003) Mechanical characterization of parts fabricated using fused deposition modeling. Rapid Prototyp J

670

S. R. Van Der Merwe et al.

13. Górski F, Kuczko W, Wichniarek R (2013) Influence of process parameters on dimensional accuracy of parts manufactured using fused deposition modelling technology. Adv Sci Technol Res J 7(19):27–35 14. Roy PA, Meguid SA (2021) Containment of blade shedding in gas turbine engines: part I— design and development of a scaled down test rig. Int J Mech Mater Des 17(1):3–12 15. Alkebsi EAA, Ameddah H, Outtas T, Almutawakel A (2021) Design of graded lattice structures in turbine blades using topology optimization. Int J Comput Integr Manuf 34(4):370–384 16. Melentiev R, Fang F (2021) Prediction of crater depth, surface roughness and erosion rate during abrasive jet machining of glass. Wear 468:203596 17. Liu L, Luo G, Chen W, Zhao Z, Huang X (2018) Dynamic behavior and damage mechanism of 3D braided composite fan blade under bird impact. Int J Aerospace Eng 18. Rahmani R, Antonov M, Brojan M (2020) Lightweight 3D printed Ti6Al4V-AlSi10Mg hybrid composite for impact resistance and armor piercing shielding. J Market Res 9(6):13842–13854 19. Fernandez-Vicente M, Calle W, Ferrandiz S, Conejero A (2016) Effect of infill parameters on tensile mechanical behavior in desktop 3D printing. 3D Print Additive Manuf 3(3):183–192 20. Patel DM (2017) Effects of infill patterns on time, surface roughness and tensile strength in 3D printing. Int J Eng Dev Res 5(3):566–569 21. Li JP, de Wijn JR, Van Blitterswijk CA, de Groot K (2006) Porous Ti-6Al-4V scaffold directly fabricating by rapid prototyping: preparation and in vitro experiment. Biomaterials 27(8):1223– 1235 22. Allami T, Alamiery A, Nassir MH, Kadhum AH (2021) Investigating physio-thermomechanical properties of polyurethane and thermoplastics nanocomposite in various applications. Polymers 13(15):2467 23. Wu H, Fahy WP, Kim S, Kim H, Zhao N, Pilato L, Kafi A, Bateman S, Koo JH (2020) Recent developments in polymers/polymer nanocomposites for additive manufacturing. Progr Mater Sci 111:100638 24. Mikula K, Skrzypczak D, Izydorczyk G, Warchoł J, Moustakas K, Chojnacka K, WitekKrowiak A (2021) 3D printing filament as a second life of waste plastics—a review. Environ Sci Pollut Res 28(10):12321–12333 25. Beckman IP, Lozano C, Freeman E, Riveros G (2021) Fiber selection for reinforced additive manufacturing. Polymers 13(14):2231 26. Alsalla H, Hao L, Smith C (2016) Fracture toughness and tensile strength of 316L stainless steel cellular lattice structures manufactured using the selective laser melting technique. Mater Sci Eng, A 669:1–6 27. Lewandowski JJ, Seifi M (2016) Metal additive manufacturing: a review of mechanical properties. Annu Rev Mater Res 46:151–186 28. Liu J, Gaynor AT, Chen S, Kang Z, Suresh K, Takezawa A, Li L, Kato J, Tang J, Wang CC, Cheng L (2018) Current and future trends in topology optimization for additive manufacturing. Struct Multidiscip Optim 57(6):2457–2483 29. Liu S, Shin YC (2019) Additive manufacturing of Ti6Al4V alloy: a review. Mater Des 164:107552 30. Ngo TD, Kashani A, Imbalzano G, Nguyen KT, Hui D (2018) Additive manufacturing (3D printing): a review of materials, methods, applications and challenges. Compos B Eng 143:172– 196 31. Zhang S, Lane B, Whiting J, Chou K (2019) On thermal properties of metallic powder in laser powder bed fusion additive manufacturing. J Manuf Process 47:382–392 32. Dekhtyar AI, Mordyuk BN, Savvakin DG, Bondarchuk VI, Moiseeva IV, Khripta NI (2015) Enhanced fatigue behavior of powder metallurgy Ti–6Al–4V alloy by applying ultrasonic impact treatment. Mater Sci Eng A 641:348–359 33. Fiegl T, Franke M, Körner C (2019) Impact of build envelope on the properties of additive manufactured parts from AlSi10Mg. Opt Laser Technol 111:51–57 34. Xu MK, Liu J, Zhang HB, Zhang Y, Wu X, Deng Z, Yu ZZ (2021) Electrically conductive Ti3C2T x MXene/polypropylene nanocomposites with an ultralow percolation threshold for efficient electromagnetic interference shielding. Ind Eng Chem Res 60(11):4342–4350

A Review on Impact Resistance of Partially Filled 3D …

671

35. Zheng C, Wang F, Cheng X, Fu K, Liu J, Wang Y, Liu T, Zhu Z (2014) Effect of microstructures on ballistic impact property of Ti–6Al–4V targets. Mater Sci Eng, A 608:53–62 36. Holmen JK, Johnsen J, Jupp S, Hopperstad OS, Børvik T (2013) Effects of heat treatment on the ballistic properties ofAA6070 aluminium alloy. Int J Impact Eng 57:119–133 37. Bürger D, De Faria AR, De Almeida SF, De Melo FC, Donadon MV (2012) Ballistic impact simulation of an armour-piercing projectile on hybrid ceramic/fiber reinforced composite armours. Int J Impact Eng 43:63–77 38. Krishnan K, Sockalingam S, Bansal S, Rajan SD (2010) Numerical simulation of ceramic composite armor subjected to ballistic impact. Compos B Eng 41(8):583–593 39. Senthil K, Iqbal MA (2013) Effect of projectile diameter on ballistic resistance and failure mechanism of single and layered aluminum plates. Theoret Appl Fract Mech 67:53–64 40. Crawford RJ, Martin PJ (2020) Plastics engineering. Butterworth-Heinemann 41. Girelli L, Giovagnoli M, Tocci M, Pola A, Fortini A, Merlin M, La Vecchia GM (2019) Evaluation of the impact behaviour of AlSi10Mg alloy produced using laser additive manufacturing. Mater Sci Eng A 748:38–51 42. He X, Xiao B, Cai G, Takahashi J (2021) Morphological, physicochemical, and flexural characterization of carbon fiber paper–reinforced polyamide 6 for long-term application in aqueous environments. J Polym Res 28(9):1–12 43. Rotman D (2017) The 3-D printer that could finally change manufacturing 44. Schirmeister CG, Schächtele S, Keßler Y, Hees T, Köhler R, Schmitz K, Licht EH, Muelhaupt R (2021) Low warpage nanophase-separated polypropylene/olefinic elastomer reactor blend composites with digitally tuned glass fiber orientation by extrusion-based additive manufacturing. ACS Appl Polym Mater 3(4):2070–2081 45. Vaezi M, Chua CK (2011) Effects of layer thickness and binder saturation level parameters on 3D printing process. Int J Adv Manuf Technol 53(1):275–284

Microstructural Evolution and Mechanical Properties of Additively Manufactured Commercially-Pure Grade 2 Titanium After Post-process Heat Treatment Ralf D. Fischer, Greyson Harvill, Hossein Talebinezhad, and Barton C. Prorok Abstract Additive manufacturing is becoming a preferred manufacturing method for small-batch manufacturing. Titanium-based AM is often chosen as a material for many applications. Much of the ongoing titanium-based AM research has focused on Ti6Al4V due to its exceptional strength and heat treatability. However, commerciallypure titanium can offer enhanced corrosion resistance as well as improved ductility. This alloy has not received a lot of attention with regard to AM processing. This work investigates the processability of Grade 2 Titanium through the Laser-Powder Bed Fusion process. The microstructural evolution of the alloy was characterized for three typical heat treatment processes as well as the corresponding mechanical response. Keywords Additive manufacturing · Laser powder bed fusion · Titanium · Heat treatment · Microstructure

Introduction Over the past 30 years, additive manufacturing (AM) has undergone a rapid transformation from a rapid prototyping technology to an advanced manufacturing technique. It is being widely employed in various industries, such as aerospace, energy, automotive, and medical [1, 2]. Conventional manufacturing methods are based on the removal of material. This comes with limitations, such as a need for fixtures or limited access to internal machined faces and channels [3]. AM, by contrast, adds material to a substrate or underlying layer to create a near-net-shape part, providing benefits such as a high degree of complexity and low-cost, one-of-a-kind parts [1, 4]. This freedom of design is one, if not the defining feature AM offers. Because AM

R. D. Fischer · G. Harvill · H. Talebinezhad · B. C. Prorok (B) Materials Research and Education Center, Auburn University, 275 Wilmore Laboratories, Auburn, AL 36849, USA e-mail: [email protected] © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_64

672

Microstructural Evolution and Mechanical Properties of Additively Manufactured …

673

fabricates near net shape parts strengthening the material through plastic deformation, i.e. cold rolling, is out of the question. Therefore, it is important to investigate possible pathways that can be taken to enhance the microstructure obtained through the AM process. Laser Powder Bed Fusion (LPBF) is an AM technology that has widely been accepted due to its high resolution and exceptional surface finishes [1]. LPBF utilizes a high-power laser to selectively melt a thin layer of powder (20–50 μm) onto a substrate or the underlying layer. This step is repeated hundreds, sometimes thousands of times until the final part is finished. Because of the narrow temporal and spatial resolution of the process, the generated thermal gradients are very high and cause unique microstructural characteristics, such as anisotropy, chemical separation, residual stresses, and/or porosity [5, 6]. Many different materials, such as ferrous alloys, nickel superalloys, aluminum, and titanium alloys, have been investigated extensively [1]. Out of these, titanium is especially interesting because of its exceptional strength to weight ratio, high corrosion resistance, and biocompatibility [7]. While extensive research efforts focus on the employment of Ti6Al4V (Grade 5 or Grade 23) because of its high strength to weight ratio for load-bearing applications [8, 9], commercially-pure titanium is gaining more traction due to its enhanced corrosion resistance, higher elasticity, and lower elastic modulus. Aside from these aspects being beneficial in preventing stress shielding in biomedical applications, it also lacks the potentially hazardous/toxic alloying elements such as Al and V [10, 11]. Therefore, this study is aimed to investigate the processability of commercially-pure titanium (Grade 2). Furthermore, the effect of heat treatment on the microstructural evolution and it corresponding mechanical response is investigated.

Materials and Methods Materials Gas-atomized Grade 2 Titanium (Concept Laser, Germany) with a composition defined in Table 1 and possessing a powder distribution of 15–45 μm was used for the LPBF process. An SEM image of the as-received powder is shown in Fig. 1a along with its Particle Size Distribution with a comparison of that provided by the manufacturer is shown in Fig. 1b. Table 1 Elemental distribution of CP-Titanium (grade 2) as reported by the manufacturer Element

Ti

Al

Fe

C

O

N

H

Powder [mass-%]

bal

0.011

0.23

0.008

0.163

0.004

0.0024

Stock [mass-%]

bal

n.a

0.09

0.01

0.12

0.007

n.a

674

R. D. Fischer et al.

Fig. 1 a SEM image of CP-Titanium powder particles showing the powder morphology and b Particle Size Distribution of CP-Titanium as reported by the manufacturer

All L-PBF specimens in this work were fabricated on a Concept Laser MLab 100 R (Concept Laser, Germany) in one print job. The oxygen concentration within the process chamber during the print was kept below 0.2 vol. % at room temperature. The machine has a 100 W ytterbium fiber laser, which selectively melts thin layers of metal powder to create complex metal parts directly from a CAD file [1, 12].

Process Parameters A parameter study was conducted to develop a set of parameters that allow to produce fully dense parts for the succeeding characterization. Twenty-five different cuboid samples with dimensions of 8 × 8 × 8 mm3 were fabricated in a single print job. The laser power and layer thickness were kept constant at 90 W and 25 μm, respectively. The scan speed was increased from 400 to 1000 mm/s in 200 mm/s intervals. The hatch distance was increased from 40 to 120 μm in 20 μm steps. A continuous scanning strategy was utilized that rotated the exposure strategy by 90° between succeeding layers. The samples were printed on a titanium substrate plate with a 3 mm support structure. The density of the parts was measured using an optical microscope (OM). All samples were sectioned in half to obtain access to the interior microstructure. Four images of the polished cross-sections were taken, and a threshold measurement was applied, which separates the deposited material and pores by their color value. The ratio between number of pixels determined as material and total number of pixels gives the density of the specimen. The average of the analyzed images was chosen for the determination of the density.

Microstructural Evolution and Mechanical Properties of Additively Manufactured …

675

Microstructure and Heat Treatment Three different heat treatments were conducted: (1) a stress relief at 650 °C for 1 h, (2) an anneal at 700 °C for 1 h, and (3) a solution treatment at 850 °C for 1 h. After the heat treatment process, the samples were air cooled back to room temperature. The heat-treated specimens were sectioned and then polished using 320 grit sandpaper, 9 μm and 3 μm diamond suspension, and a final polishing step using a mixture of 20% hydrogen peroxide and 80% OP-S solution. Images for the microstructure were taken using a polarized light microscope. The microstructure was characterized using X-ray diffraction with a Bruker D8 Diffractometer with a Cu X-ray source set at a working current of 40 mA and a voltage of 40 kV.

Mechanical Properties The hardness measurements were performed on polished cross-sections. Vickers microhardness was measured using a LECO DM-400 Hardness Tester with a 1000 g load for 30 s. An average hardness was calculated by taking the mean average of 6 measurements taken at random locations.

Results and Discussion Density The density of as-fabricated CP-Titanium was analyzed to determine the optimum printing parameters. Figure 2 shows the pore sizes and distributions for the selected parameter ranges. The density ranged from very high values of 99.98% to rather low values of 95.97%. Three main regions with different pore characteristics are observed. The first region is dominated by large, spherical pores and are found at low velocities, below 600 mm/s. This so-called “keyhole porosity” develops at high energy intensities, where the evaporation of metal leads to a strong recoil pressure exerted on the melt pool, which in turn forms a deep, narrow depression (keyhole) [13]. Inside the keyhole, the laser beam is reflected and absorbed multiple times causing a very high absorptivity. When the keyhole becomes unstable due to thermocapillary forces, recoil pressure, Marangoni convection, or the appearance of plasma, the keyhole collapses on itself and leaves behind a pore of entrapped gas [13]. Because the energy density of the low-velocity samples is very high compared to other samples, the developed porosity is assumed to be associated with keyhole instabilities.

676

R. D. Fischer et al.

Velocity (mm/s)

1000 99.95 %

99.98 %

99.97 %

99.25 %

95.97 %

99.96 %

99.87 %

99.91 %

99.76 %

98.17 %

97.78 %

98.87 %

98.67 %

99.23 %

98.95 %

98.96 % 40

97.43 % 97.65 % 97.63 % 60 80 100 Hatch Spacing (μm)

97.83 % 120

800

600

400 500 μm

Fig. 2 Graph showing the cross-section and density of printed CP-Titanium samples for different process parameters. The parameters of the marked specimen were used to fabricate specimen for the remainder of this work. The laser power was constant at 90 W. The build direction is along the arrow on the left. The red marked square shows the cross-section of the samples used for the succeeding characterizations

The second region can be found at high velocities (>800 mm/s) and large hatch spacing (>100 μm). The pores present here are aligned parallel to the building direction, and their shape is irregular. The spacing between the lines of pores is consistent and increased with increasing hatch spacing. The formation of these types of pores can be attributed to an insufficient energy input, which prevents the applied powder layer from completely melting [14]. The pores originate in the locations between the individual laser tracks and become larger with increasing hatch distance, confirming the assumption. Additionally, the sample with 600 mm/s velocity and 120 μm hatch distance appears to show both the lack of fusion vertical lines and random spherical pores, suggesting that both lack of fusion and keyhole pores may be present. The highest densities in this parameter study were achieved at velocities of 800– 1000 mm/s and a hatch distance of 40–80 μm. The cross-sections reveal a lack of major pores and a maximum density of 99.98%, meaning that the energy input into the applied powder layer was sufficient enough to completely melt the material without causing the development of keyhole pores or lack of fusion. For the succeeding characterization, the parameters of the cross-section marked with the red outline were chosen because they were within the exceptional density range as well as promised high productivity (90 W laser power; 25 μm layer thickness; 1000 mm/s velocity; and 60 μm hatch spacing).

Microstructural Evolution and Mechanical Properties of Additively Manufactured …

677

Microstructural Evolution The polarized light images in Fig. 3a show the obtained microstructures of the analyzed additively manufactured samples. The as-fabricated additively manufactured samples showed a similar microstructure of LPBF solidified pure titanium similar to that reported in the literature [15–17]. Upon rapid cooling, inherent to the LPBF process, columnar β-phase grains formed along the build direction and transformed into the martensitic α’-phase. These grains retained their columnar structure but formed a fine lathe morphology within them, which is caused by the rapid β → α phase transition occurring at ca. 890 °C [18]. The 650 °C samples mostly retained the as-printed columnar grain structure, including the lathe morphology. However, some regions appear to have evolved somewhat with the nucleation of new stress free grains. Although this temperature is typically applied for only stress relief, the high thermal gradients of the LPBF process produced heterogeneous distortions in the microstructure capable of new grain nucleation, even at this low temperature [15]. The microstructure of the 700 °C specimens shows a significant evolution over the previous microstructures. Here, some of the lathe morphology is still retained, but a significant number of recrystallized α grains have formed. Furthermore, the grain size is nonuniform with clusters of larger and smaller grains. A more fully recrystallized and equiaxed microstructure was obtained after the 850 °C heat treatment. Some of lathe morphology is present within some of the grains. While the grain size is more uniform than the 700 °C samples, a slight difference in grain size is still observable. The XRD spectra within a 2θ range between 30° and 90° for the AM powder and the AM-fabricated and heat-treated specimen are depicted in Fig. 3b. The corre-

700 °C 1 hr

850 °C 1 hr

(112) (202)

(103)

(110)

(102)

850 °C 1hr

Intensity (a.u.)

50 µm

50 µm

(002)

(101)

(b)

650 °C 1 hr

As Printed

(100)

(a)

700 °C 1hr

650 °C 1hr

As Printed Powder

30 50 µm

50 µm

40

50

60

70

80

Angle (2 )

Fig. 3 a Polarized light images of the microstructure of AM Grade 2 Titanium after different heat treatments and from wrought material. The arrow indicates the build direction. b XRD patterns of Titanium Grade 2 samples

678

R. D. Fischer et al.

Fig. 4 Vickers hardness of commercially-pure Titanium after various heat treatments

Micro-hardness (HV)

300

283.1 260.8 212.3

201.8

200

100

0

As built

650 1hr

700 1hr

850 1hr

sponding diffraction peaks for hexagonal close-packed (hcp) titanium are present for all samples, indicating that the microstructure is composed of the α/α’ phase.

Mechanical Properties The results of the microhardness measurements for the AM Titanium before and after heat treatments are presented in Fig. 4. The as-printed samples displayed the highest hardness of 283.1 HV and is attributed to the martensitic microstructure obtained from the LPBF process. The microstructures of the heat-treated AM samples showed a reduction in hardness with increasing annealing temperature down to 201.8 HV. It should be noted that the major reduction in hardness occurred at annealing temperatures below 700 °C, or before the morphology transformation from the lathe structure to the equiaxed grain structure occurred. This decrease in hardness for the AM Grade 2 Titanium can be attributed to an increase in grain size, which can be observed in our sample set [15]. This finding will contribute to developing heat treatment strategies that can tailor the hardness of the material to be higher or lower, which in turn enables the use of LBPF Titanium for new applications.

Conclusion Commercially-pure Grade 2 Titanium specimens were manufactured using LPBF. A process parameter map was developed to fabricate a fully dense specimen. High

Microstructural Evolution and Mechanical Properties of Additively Manufactured …

679

energy and low energy regions were identified, in which either keyhole or lack of fusion was the dominating mechanism governing the creation of porosity. Heat treatment was employed on the titanium specimen to alter the microstructure and mechanical properties. It was found that the microstructure changed from a martensitic α’phase to a recrystallized equiaxed α-grain morphology. Recrystallization started at a temperature of 650 °C and was completed at a temperature of 700 °C, but some of the lathe morphology is still retained within the grains. At 850 °C, the grains are fully recrystallized and appear equiaxed. This change from a martensitic α’-phase to a recrystallized one caused a reduction in the hardness of the titanium; however, most of the hardness reduction was observed at a temperature below 700 °C. These findings will help to develop heat treatment strategies for commercially-pure titanium and allow for novel applications to be enabled through LPBF. Acknowledgements This work was sponsored by the United States National Institute of Standards and Technology under contracts NIST-70NANB16H272, NIST-70NANB17H295, and NIST-70NANB18H220.

References 1. DebRoy T et al (2018) Additive manufacturing of metallic components—process, structure and properties. Progress Mater Sci 92:112–224. https://doi.org/10.1016/j.pmatsci.2017.10.001 2. Ngo TD, Kashani A, Imbalzano G, Nguyen KTQ, Hui D (2018) Additive manufacturing (3D printing): a review of materials, methods, applications and challenges. Compos Part B Eng 143:172–196. https://doi.org/10.1016/j.compositesb.2018.02.012 3. Gao W et al (2015) The status, challenges, and future of additive manufacturing in engineering. Comput Aided Des 69:65–89 4. Poprawe R, Hinke C, Meiners W, Schrage J, Bremen S, Merkt S (2015) SLM production systems: recent developments in process development, machine concepts and component design. In: Brecher C (eds) Advances in Production Technology. Lecture Notes in Production Engineering. Springer, Cham. https://doi.org/10.1007/978-3-319-12304-2_5 5. Simson T, Emmel A, Dwars A, Böhm J (2017) Residual stress measurements on AISI 316L samples manufactured by selective laser melting. Addit Manuf 17:183–189 (2017). https://doi. org/10.1016/j.addma.2017.07.007 6. Guo C et al (2021) A comparing study of defect generation in IN738LC superalloy fabricated by laser powder bed fusion: continuous-wave mode versus pulsed-wave mode. J Mater Sci Technol 90:45–57. https://doi.org/10.1016/j.jmst.2021.03.006 7. Grover T et al (2020) Role of titanium in bio implants and additive manufacturing: an overview. Mater Today Proc 26:3071–3080. https://doi.org/10.1016/j.matpr.2020.02.636 8. Carrion PE, Soltani-Tehrani A, Phan N, Shamsaei N (2019) Powder recycling effects on the tensile and fatigue behavior of additively manufactured Ti-6Al-4V parts. JOM 71(3):963–973. https://doi.org/10.1007/s11837-018-3248-7 9. Vrancken B, Thijs L, Kruth JP, Van Humbeeck J (2012) Heat treatment of Ti6Al4V produced by selective laser melting: microstructure and mechanical properties. J Alloys Comp 541:177–185. https://doi.org/10.1016/j.jallcom.2012.07.022 10. Wauthle R et al (2015) Revival of pure titanium for dynamically loaded porous implants using additive manufacturing. Mater Sci Eng C 54:94–100. https://doi.org/10.1016/j.msec. 2015.05.001

680

R. D. Fischer et al.

11. Attar H, Calin M, Zhang LC, Scudino S, Eckert J (2014) Manufacture by selective laser melting and mechanical behavior of commercially pure titanium. Mater Sci Eng A 593:170–177. https:// doi.org/10.1016/j.msea.2013.11.038 12. Gu DD, Meiners W, Wissenbach K, Poprawe R (2012) Laser additive manufacturing of metallic components: materials, processes and mechanisms. Int Mater Rev 57(3):133–164. https://doi. org/10.1179/1743280411Y.0000000014 13. Zhao C et al (2020) Critical instability at moving keyhole tip generates porosity in laser melting. Science 370(6520):1080–1086. https://doi.org/10.1126/science.abd1587 14. Bayoumy D, Schliephake D, Dietrich S, Wu XH, Zhu YM, Huang AJ (2021) Intensive processing optimization for achieving strong and ductile Al-Mn-Mg-Sc-Zr alloy produced by selective laser melting. Mater Des 198:109317. https://doi.org/10.1016/j.matdes.2020.109317 15. Li C-L et al (2019) Simultaneous achievement of equiaxed grain structure and weak texture in pure titanium via selective laser melting and subsequent heat treatment. J Alloys Comp 803:407–412. https://doi.org/10.1016/j.jallcom.2019.06.305 16. Pehlivan E et al (2020) Effects of build orientation and sample geometry on the mechanical response of miniature CP-Ti grade 2 strut samples manufactured by laser powder bed fusion. Addit Manuf 35:101403. https://doi.org/10.1016/j.addma.2020.101403 17. Gu D et al (2012) Densification behavior, microstructure evolution, and wear performance of selective laser melting processed commercially pure titanium. Acta Mater 60(9):3849–3860. https://doi.org/10.1016/j.actamat.2012.04.006 18. Kim SK, Park JK (2002) In-situ measurement of continuous cooling β → α transformation behavior of CP-Ti. Metall Mater Trans A 33(4):1051–1056. https://doi.org/10.1007/s11661002-0206-2

Preparation of TiAl Alloy by Magnesium Aluminum Synergistic Reduction of TiO2 in Molten Salt Medium Jialong Kang, Zhenyun Tian, Guibao Qiu, and Yaoran Cui

Abstract The traditional method of preparing TiAl alloy needs titanium and aluminum as raw materials for alloy preparation. The process of preparing pure metal is complex, the production process is too long, and high production causes energy consumption and serious pollution. In this study, a new method of preparing TiAl alloy by magnesium thermal reaction reduction of TiO2 in mixed molten salt was proposed. The experimental data and results show that the temperature of 750 °C is conducive to the reduction of TiO2 . The mixed molten salt of KCl and MgCl2 is conducive to improving the mass transfer rate and promoting the reaction. The mixed molten salt of KCl and MgCl2 is conducive to improving the mass transfer rate and promoting the reaction. The titanium oxide is completely reduced. At the same time, the feasibility of this method is proved by experiments. Keywords TiAl alloy · Molten salt · Magnesium thermal reduction

Introduction Titanium has excellent properties, such as high strength, high hardness, high temperature resistance, corrosion resistance, etc. Due to these excellent properties, titanium is widely used in aerospace, petroleum, chemical industry, shipbuilding, metallurgy, biomedical, and other important fields [1]. However, pure titanium has insufficient oxidation resistance at high temperature, and oxygen absorption will have an adverse effect on its mechanical properties. Fortunately, these shortcomings can be overcome by TiAl master alloy [2]. Therefore, TiAl master alloy has become the focus and hotspot in the development of titanium alloys all over the world [3]. Titanium aluminum alloy has higher service temperature and high-temperature strength, which can make the engine work at higher temperature (above 600 °C), and the density of TiAl alloy is small. The smaller density can reduce the weight of the engine and prolong the service life of the mechanical components. J. Kang · Z. Tian · G. Qiu (B) · Y. Cui College of Materials Science and Engineering, Chongqing University, Chongqing 400044, China e-mail: [email protected] © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_65

681

682

J. Kang et al.

As mentioned above, at present, the main methods for preparing TiAl alloys are high-temperature melting with sponge titanium and metal aluminum as raw materials [4]. The most successful and main way to prepare sponge titanium is Kroll method [5]. However, this process has some defects, such as long process, low efficiency, discontinuous production, high reaction temperature, and serious corrosion. At present, the research on the production of titanium by this method mainly focuses on updating the equipment, optimizing and reasonably designing the process parameters, how to reduce the equipment maintenance cost, etc. But there is no overall improvement in the preparation method of titanium, and its unreasonable production problems still exist. It is inevitable to transform the traditional preparation process of TiAl master alloy and develop the preparation process of TiAl master alloy with low cost and low energy consumption. TiAl master alloy was prepared after sponge titanium production. These processes have long processes, low efficiency, discontinuous production, high energy consumption, and pollution. How to directly prepare TiAl alloy with TiO2 is the key problem to be solved. In order to brief the process flow, pollution, and energy consumption, this experimental researcher thought about the traditional methods of producing titanium aluminum alloy and creatively put forward the “preparation of titanium aluminum alloy by reduction of semi-molten magnesium aluminum alloy”.

Experimental Materials TiO2 (purity: 99.5wt%, Granularity: 0.1–0.3 µm), Mg (purity: 99wt%, Granularity: 0.16 mm), Al (purity: 99wt%, Granularity: 74 µm), and MgCl2 and KCl (purity: 99wt%, Granularity: 74 µm) were obtained through Aladdin reagent company. The raw material ratio is shown in Table 1.

Experimental Process The experimental process is shown in Fig. 1. After the sample is prepared in proportion, it is placed on the molybdenum crucible, and then heated to 700–1000 °C in a tubular furnace to obtain the reduction slag. Then, it is acid soaked with 5% dilute salt to remove MgCl2 , KCl, MgO, and Al2 O3 , and finally TiAl alloy is obtained. Table 1 Proportion of main experimental raw materials (%)

KCl-MgCl2

Mg

Al

TiO2

45–50

12–15

12–25

20–25

Preparation of TiAl Alloy by Magnesium Aluminum Synergistic …

683

Fig. 1 Flowchart of the experimental process

Analysis The reduced slag and final powder were analyzed by X-Ray Diffraction (XRD), and the final powder was analyzed by Scanning Electron Microscope (SEM) equipped with Energy Disperse Spectroscopy (EDS). The principle of the reaction process was deduced by thermodynamic calculation.

Results and Discussion Phase Analysis The XRD patterns at different reaction times are shown in Fig. 2. The reaction products are mainly TiAl and TiAl3 when the reaction time is within 2–4 h. This shows that TiAl3 alloy powder can be effectively produced by semi-molten magnesium aluminum alloy reduction method. It can be clearly seen from the XRD spectrum that TiAl3 alloy phase increases and TiAl alloy phase decreases with the extension of time. However, the mechanism of the effect of reaction time on the reaction products will be discussed in the next section.

Analysis of TiAl Alloy Products The prepared metal powder is analyzed by SEM surface scanning. Figure 3 shows the SEM surface scanning spectrum of the metal powder. From the results, it can be seen that the metal powder is Ti and Al, which is confirmed to be TiAl alloy.

684

J. Kang et al.

Fig. 2 XRD of products with different reaction time

Fig. 3 SEM and EDS of alloy powder

The results show that titanium aluminum alloy powder is obtained by semi-molten magnesium aluminum cooperative reduction. At the same time, EDS analysis was carried out on Ti–Al alloy powder. The content of Ti in the alloy is 36.4%, the content of Al is 62.4%, and the oxygen content of the alloy is 1.2%. After surface scan analysis, Ti and Al are evenly distributed.

Preparation of TiAl Alloy by Magnesium Aluminum Synergistic …

685

Thermodynamic Analysis of Magnesium Reduction of TiO2 The reduction of TiO2 by magnesium is a step-by-step reduction process. According to the step-by-step transformation principle, the high valence metal oxides will undergo the transformation process of each intermediate valence state step by step in the process of reducing to metal. According to the relevant research, the reduction of TiO2 by magnesium can be divided into the following steps: 4TiO2 + Mg = Ti4 O7 + MgO

(1)

3Ti4 O7 + Mg = 4Ti3 O5 + MgO

(2)

2Ti3 O5 + Mg = 3Ti2 O3 + MgO

(3)

Ti2 O3 + Mg = 2TiO + MgO

(4)

TiO + Mg = Ti + MgO

(5)

It can be seen from the above formulas that when titanium is prepared by magnesium thermal reduction, TiO2 will undergo the process of intermediate products such as Ti4 O7 , Ti3 O5 , Ti2 O3 , and TiO in the process of reducing to metal Ti. It can be seen from Fig. 4 that magnesium has a relatively negative Gibbs free.

Fig. 4 Gibbs free energy change of titanium oxide reduced by magnesium

686

J. Kang et al.

Fig. 5 Schematic diagram of TiAl3 alloy reaction process

Al

TiAl

Energy change when reducing Ti4 O7 , Ti3 O5 , and Ti2 O3 , so magnesium reduction of Ti4 O7 , Ti3 O5 , and Ti2 O3 is not the key influencing factor of magnesium reduction of titanium dioxide. Whether the metal titanium can be obtained finally depends on whether the low valence oxide of titanium can be reduced. When magnesium reduces TiO, if the temperature is greater than 1681 K, the Gibbs free energy becomes positive and the reaction cannot be carried out. Therefore, the reduction reaction is carried out at a relatively low temperature of 700–1000 °C in this experiment, so that the thermodynamics of the reduction reaction can proceed smoothly. However, at a relatively low temperature, the melting point of titanium oxide is higher than 1000 °C. At this time, it is all solid, the mass transfer efficiency is very low, and the kinetics of reduction reaction is not smooth. Therefore, the mixed molten salt system is used for the reaction at the same time. KCl and MgCl2 are added, and the melting points are 770 °C and 714 °C, respectively. The reduction reaction is liquid, which can improve the mass transfer rate, promote the progress of the reaction, and fully reduce the titanium oxide. Aluminum is liquid in the reaction temperature range. At this time, Al reacts with Ti as shown in Fig. 5. The reaction process is controlled by diffusion. At the initial stage, the liquid Al contacts Ti and forms TiAl on the surface of Ti. Then, with the extension of reaction time, a large amount of Al gradually reacts with Ti to form TiAl3 , and finally TiAl3 is formed after 6 h.

Conclusion (1)

(2)

The thermodynamics of the step-by-step reduction process of titanium oxide by magnesium is calculated. The results show that TiO2 will be effectively reduced by magnesium when the reaction temperature is 700–1000 °C. The experimental study on the one-step preparation of titanium aluminum alloy by magnesium thermal reduction in KCl-MgCl2 mixed molten salt reaction medium is carried out. The results show that TiO2 can be reduced step by step by metal magnesium in molten salt medium, and the reduced metal Ti will react with metal aluminum to form TiAl3 alloy, and the oxygen content of the alloy is 1.2%.

Preparation of TiAl Alloy by Magnesium Aluminum Synergistic …

687

References 1. Mo W (1998) Titanium metallurgy. Metallurgical Industry Press, China 2. Tan XQ, Chen J, Zhi W, Brown J (2010) Electronic structures and transformations of intermediate phases and intermetallic compounds of binary Ti-Al alloys. Phys B Phys Conden Matter 405(17):3543–3546 3. Li J, Lu X, Yang S, Wu E, Hou J, Li C, Lai Q, Ma L (2017) Theoretical and experimental research on preparation of Ti-Al alloy by electric aluminothermic reduction. Iron Steel Vanadium Titanium 38(5):46–52 4. Zhang F, Lu L, Lai MO (1997) Structural evolution of Ti-Al via mechanical alloying. Int Soc Opt Photon. https://doi.org/10.1117/12.269805 5. Capus J (2016) Titanium powder metallurgy at POWDERMET 2015: past, present and future. Met Powder Rep 71(1):25–27

Part XIX

AI/Data Informatics: Computational Model Development, Validation, and Uncertainty Quantification

Investigating the Suitability of Tableau Dashboards and Decision Trees for Particulate Materials Science and Engineering Data Analysis Bryer C. Sousa, Richard Valente, Aaron Krueger, Eric Schmid, Danielle L. Cote, and Rodica Neamtu Abstract Informed integration of data-driven models for materials processing has yet to be fully realized due to data science knowledge gaps, incomplete materials and processing datasets, and a lack of data-driven tools designed explicitly for classically trained engineers. On the other hand, modern particle size distribution analyzers enable hundreds of thousands of particle-to-particle size, shape, and morphological properties to be easily gathered. Accordingly, we present suitable data analysis, sharing, and visualization approaches for developing a powder particle classification based upon powder morphology and size metrics for flowability on demand (FoD). We demonstrate the utility of Tableau Dashboards connected to a live powder database for making data-driven integration convenient to assess, visualize, and analyze particulate data, thus making comparisons between the features of individual powders and micro-particulate constituents accessible for traditional materials scientists and engineers. The FoD framework reduced the time taken for common workflows for FoD-based tasks. Keywords Data-driven materials science · Powder metallurgy · Particulate materials

Introduction When considering the properties-structure-processing-performance paradigm of materials science and engineering [1], the long-standing Edisonian nature of studying and developing new materials is slow and painstaking [2]. Accordingly, researchers have historically implemented low-throughput trail-and-error schemes wherein one must test each candidate material and do so over a range of processing histories, thus B. C. Sousa (B) · D. L. Cote Department of Mechanical Engineering, Materials Science and Engineering Program, Worcester Polytechnic Institute, Worcester, MA, USA e-mail: [email protected] R. Valente · A. Krueger · E. Schmid · R. Neamtu Department of Computer Science, Worcester Polytechnic Institute, Worcester, MA, USA © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_66

691

692

B. C. Sousa et al.

making application-driven materials research and development expensive and laborious for many modern industries [3, 4]. Fortunately, advancements in computer technologies and concurrently decreased cost surrounding access to increased computing power, and materials scientists and engineers have started to leverage novel in-silico techniques and sophisticated machine learning algorithms to drive materials innovation at accelerated rates [5, 6]. In turn, one may improve the work efficiency of materials researchers as well as advanced materials manufacturing and processing engineers through data science and business intelligence methods, which creates more commercial and business value for the materials sector at large [7, 8]. With the aforementioned in mind, we present suitable data analysis, sharing, and visualization approaches to developing a powder particle classification based upon powder morphology and size metrics, which serves as an illustrative and demonstrative case study for the powder-based metal additive manufacturing (AM) community in particular. More specifically, we demonstrate the utility of Tableau Dashboards connected to a live powder database for making data-driven integration convenient to assess, visualize, and analyze particulate data, thus making comparisons between the features of individual powders and micro-particulate constituents accessible for traditional materials scientists and engineers.

Background Powder Flowability Many modern and emerging metal AM technologies rely on metallic powder particles as feedstock material [9]. Since metallurgical powders are widely and commonly utilized in metal AM methods, powder metallurgists and particulate-focused materials researchers have relied upon statistical descriptors and metrics to quantify the general characteristics of a given feedstock before consumption during materials processing [10]. Moreover, the metrics commonly ascribed to a given feedstock powder include the particle size/shape distributions (PSDs), the span of said distributions [11], and additional metrics associated with powder rheology [12], which is concerned with the properties, behaviors, and characteristics of a collection of granular particles (i.e., powder) [13]. Many researchers have considered particle size the dominant “critical quality attribute” ascribed to a given micro-particulate powder feedstock [13]; however, studies have demonstrated that individual attributes alone are insufficient to predict how the powder will behave during AM consumption and processing [14]. Accordingly, powder has also been quantified and analyzed in terms of parameters such as bulk density, Hausner ratio, Hall flow, angle of repose, Carney flow, and more to infer powder behavior and processability in AM conditions. However, such inferences have resulted in incremental success when approached through the lens of physical principles, relations, and empirically-related phenomena. On the other hand, machine learning and data science tools have recently unveiled data-driven

Investigating the Suitability of Tableau Dashboards and Decision Trees …

693

linkages between powder particle size distributions and Hall flow rates [15]. With such success in mind, continued research and development centers upon utilizing machine learning techniques to predict in-process powder behavior and flowability [16]. In addition to a discussion centered upon the continued use of data-driven tools to predict in-process flowability of powder, consideration is also given herein to Tableau Dashboards (Tableau Software, Seattle, WA), an industrial or business intelligence friendly data-centric software platform, as a means of allowing a range of personnel with variable levels of domain knowledge to intuit the implications of associated datasets readily.

Metal Additive Manufacturing With the above passage in mind, AM relies upon creating components or parts with previously unachievable complex geometries through a layer-by-layer build process [17]. Powder-based metal AM has witnessed notable and continual growth [18], adoption and innovation [19], and diversity in the technique-dependent in-process conditions that powder particles experience [20]. Powder-based metal AM techniques include laser powder bed fusion or selective laser melting [21], electron beam powder bed fusion [22], and cold spray additive manufacturing [23]. For each of the respective processes listed above, powder feedstocks are directly introduced to the AM system and subsequently undergo spreading, melting, fluidization, deposition, or spraying, among other in-process conditions. Nevertheless, direct introduction (let alone in-process conditional response) of metallurgical powder into said AM technologies and respective manufacturing systems depends upon appropriate powder flowability [24] and suitable powder rheology [25]. Consequently, metal AM relies heavily on powder feedstock properties and behaviors to target and achieve optimal and desirable part performance by designing processing parameters around a given feedstock [26, 27].

Flowability on Demand (FoD) Framework While the present work primarly focused upon the implications and suitability of Tableau Dashboards for live dataset visualization, exploration, and data interactivity/accessibility for interdisciplinary use by project managers, data scientists, and materials engineers, one ought to note that the Tableau-driven solutions considered compliment the work of Valente et al. [15]. Stated otherwise, Valente et al. presented a data-driven classification model for classifying flowability, in terms of Hall flow rates, for a given metallic particulate feedstock powder for AM processing when considering particle-level geometrical properties and characteristics. Accordingly, a machine learning model known as flowability on demand (FOD) was developed for the automatic classification of powder particle constituents in terms of being well-suited or ill-suited for cold spray additive manufacturing needs. Underpinning

694

B. C. Sousa et al.

the 2020 formulation of FoD as a classifaction model was the use of a generalizable, supervised as well as non-parametric decision tree formulation. Achieving notable classification accuracy in [15], the FoD pipeline may be itemized as follows: (1) data acquisition using Microtrac-based PSD and PSD-like data, (2) preprocessing of data obtained as part of item (1) through Yeo–Johnson transformations, followed by dataset balancing and concluded via (3) classification through algorithmic implementation of decision trees. For further information, the reader may consult [15].

Methods Research Team Structure The collaborative research team consisted of two groups that work in conjunction to develop data science solutions for materials science applications. The materials science subdivision of the data-driven materials science (DDMS) team consisted of five students and an advisor, and the data science subdivision of the DDMS team consisted of eight students and an advisor. Overall, this presents three distinct domains in terms of access to data: advisors should be able to see a high-level overview of the data, materials scientists should have access to specific data on each powder and particle within a given batch of powder, and data scientists need information on the distribution of data and summary statistics.

Dataset Consideration The presently considered dataset was based upon output data from a Microtrac system (MicrotracRetsch GmbH, Haan/Duesseldorf, Germany) particle analyzer as described in [15]. The particle-based dataset consisted of just under 2,000,000 rows, and each particle belonged to 1 of 31 powders. Particle counts varied greatly among powders, with some powders consisting of as few as 10,000 particles and others having well over 100,000. The Microtrac system measured various features of each particle through image-based and laser-diffraction data recorded, deriving information on particle-level morphology characteristics underpinning each powder. In total, 29 different features were measured, including raw size/shape measurements as well as calculated metrics like the ellipticity of a particle, for instance.

Investigating the Suitability of Tableau Dashboards and Decision Trees …

695

On-Cloud Data Storage Data was uploaded and stored on the AWS Master Data Management solution Athena (Amazon, Seattle, WA), which is a tool similar to a traditional relational database, therefore, allowing users to store data in a row-column format. The pipeline for building and curating the on-cloud data storage framework is noted hereafter. First, members of the materials science subdivision of WPI’s DDMS team uploaded data gathered from the Microtrac system in the form of .xlsx, csv, and .json files, which were subsequently cleaned and made available to the data science DDMS subdivision. Next, data science DDMS members administered the tabulated schemas using AWS Glue, which helps clean and move data between different data stores. Finally, SQL and the Athena endpoint accessed the cleaned data, so DDMS team members could analytically and intelligently consider the datasets when developing prospective powder-based and AM-inspired solutions.

On-Cloud Data Cleaning Characterization output files obtained through the Microtrac system can contain missing data points and non-UTF-8 formatted characters in data. As a result, many machine learning, business analytics/intelligence, and data science algorithms cannot handle the raw data, thus resulting in the need for data files to be pre-processed or cleaned before data-driven processing. Unclean data was initially uploaded and stored in S3 buckets. Then, it is pre-processed and cleaned by removing non-UTF-8 characters to enable compatibility with Python libraries like Pandas. Metadata was also added at the particle level during the preprocessing step, such as the particle’s UUID and the corresponding powder ID. The newly cleaned data was stored in another S3 bucket, separate from the raw, unclean data. Once the uploaded data was cleaned, data analysis methods were applied, additional analysis via the Tableau Dashboard and data visualization was offered within the Athena system.

Live Connection with Tableau Tableau is a graphical data visualization tool leveraged in industry and academic research [28]. The functionality allows users to quickly create high-quality, dynamic, and interactive visualizations without requiring the user to have expertise in programming or specific dataset details. Using Tableau, a live server connection was established with the AWS Athena endpoint. Said connectivity allowed users to create and view visualizations with live data. Members of different subdivisions of the DDMS team could then interact with and visualize different aspects of the data as well. In

696

B. C. Sousa et al.

turn, the Tableau Dashboards formulated were found to be notably useful for making data-driven decisions.

Visualization Dashboards As noted earlier, the DDMS research team consisted of three significant stakeholders: advisors, materials scientists, and data scientists. Each group needed to regularly view the powder dataset and perform various tasks utilizing respective datasets as well as apply analyses upon the data. Still, each group maintained notably different needs and goals whilst considering analytics and visualizations. Therefore, three different Tableau Dashboards were created using Tableau and the AWS infrastructure described above, with each Tableau Dashboard tailored per group. The advisor Dashboard focuses on a high-level overview of the composition of the dataset, including the type of the powder and the flowability values for different samples, as illustrated in Fig. 1. Additionally, it shows the growth of the dataset over time, enabling advisors to keep track of the frequency in which data is added to the dataset. In addition to a high-level overview of the types of powders in the dataset and the number of particles in the powders, the materials scientist Dashboard is more detail-oriented, allowing materials scientists to visualize the features of a selected powder dynamically by selecting a powder and a set of features to view, as shown in Fig. 2. The Dashboard then updates some visualization elements to show data distribution within the selected features and powder. The data scientist Dashboard focuses more on the distribution of the data and other more detailed summary statistics, which is presented in Fig. 3 hereafter. For

Fig. 1 Tableau dashboard visualization for advisor/managerial DDMS stakeholders

Investigating the Suitability of Tableau Dashboards and Decision Trees …

697

Fig. 2 Tableau dashboard visulation for DDMS materials science stakeholders

Fig. 3 Tableau dashboard visulation for DDMS data science stakeholders

example, data scientists can view the distribution of flowability values and particle counts concerning the three powder types, the distribution of powder types, and the total count of powders in the dataset.

698

B. C. Sousa et al.

Experimental Validation Integration was tested on both the data science and materials science teams to validate the performance and usability of the Tableau tool. The purpose of testing on the data science team was to examine the speed improvement compared to manually obtaining the same data using Python, which was identified as an everyday use case for data scientists. The purpose of testing on the materials science team was to examine the general usability of the tool and the ease of accessing accurate information. All team members were evaluated using the materials science Dashboard to foster easier comparison between the two groups.

Approach to Validation Six data science team members were tasked with obtaining information on a specific powder using the Tableau Dashboards and were subsequently asked to find similar information by manually writing Python 3 code to search through a local copy of the dataset. Participants were also required to provide plots of the dataset, such as a histogram of one of the powder’s features. These tasks were common among data science team members, as information retrieval and data visualization are critical aspects of their duties and represent a common use case of the Dashboard. The members were timed while performing these tests, enabling comparisons in the speed, and efficiency of each method tested. In addition, three of the six members tested were given a brief tutorial on how to use the Dashboard, and the other two were not. The purpose of this was to examine the Dashboard’s general usability and gauge effectiveness on an audience that was or was not familiar with such a service. Five members of the materials science team were also tasked with obtaining information on specific powders using the Tableau Dashboards. However, these tests were generally aimed at ensuring that the team members could obtain accurate information quickly. The first task was the same as that of the data science team; participants were asked to find information on the flowability of a specific powder and to provide a histogram of that feature. The task was timed to compare each team’s ability to use the Dashboard. Then, in the second task, the participants were asked to compare information between different powders. Again, both tasks were designed to mimic everyday tasks performed by materials science researchers and the general use cases of the Tableau Dashboard. Participants were evaluated based on the accuracy of their responses, which would allow for an assessment of the usefulness of the Dashboard from a materials science point of view. Data was recorded anonymously and only consisted of the time taken to complete tasks, the user’s responses to questions found in each task, and the user-created histograms of feature data.

Investigating the Suitability of Tableau Dashboards and Decision Trees …

699

Anecdotal Performance The members of the data science team that were given a brief tutorial on the Dashboard experienced a significant reduction in time taken to obtain data compared to the manual method. On average, this group experienced a time reduction of 15.33 min or approximately a 75.41% decrease in time taken to obtain data. A significantly smaller reduction in time was examined for the members who did not receive a tutorial on using the Dashboard, with an average time reduction of only 48.5 s. This translates to a 6.43% reduction in the time taken to obtain data. It is worthwhile to note that one participant who did not receive a tutorial had a portion of their Dashboard covered by a popup, which likely slowed them down when attempting to create the histogram. Users performed slightly better for the materials science participants who did not receive a tutorial than those on the data science team; on average, the first task took 8 min and 59 s rather than 11 min and 46 s. If this performance is factored into the average time taken among all users who did not receive a tutorial, an improvement of 1.21 min or 17.53% is seen compared to manual use of Python code. As for accuracy among the six data science team participants, only one of the six experienced an error in manually reporting the flowability of a powder. In addition, all participants successfully created a histogram using Python. On the other hand, all participants (both those that received a tutorial and those that did not receive a tutorial) successfully reported the flowability of a powder using the Tableau Dashboard. In contrast, one participant created a histogram with incorrect axis limits. All materials science team participants successfully reported the correct flowability value using the Dashboard, and all but one participant successfully created a histogram. However, the participant who failed to create the correct histogram experienced a similar error to the data science participant who failed to create the correct histogram, as the materials science participant also created a histogram with incorrect axis limits. One possible reason for these erroneous histograms is that specific histograms were created to model features with values between 0 and 1, whereas others were created to model features with no limit on their value. So, when the participants tried to model the histograms required in the first task, they used the incorrect model.

Concluding Remarks, Implications, and Discussion The use of the Tableau Dashboard allows teams to facilitate challenging workflows that involve time-consuming data retrieval and visualization. As demonstrated in the section on experimental validation, participants who received a tutorial could cut back on time spent obtaining and visualizing data significantly. Even users who did not receive a tutorial demonstrated a slight improvement in completing tasks

700

B. C. Sousa et al.

compared to manually writing Python code. Besides improvements in time demonstrated by the Dashboard, most users (with or without tutorials) could accurately retrieve information from the Dashboard. The Tableau Dashboard is also useful for those unskilled in manually creating data visualizations using various programming languages. For materials scientists, a critical use case is to visualize the vast amount of information created by machines like the Microtrac particle analyzer. Instead of learning the necessary programming language or asking a data scientist to visualize or retrieve data, materials science team members can rapidly access the Dashboard and create the visualization in minutes. Materials science team members demonstrated a solid capability to use the Dashboard even when they did not receive a tutorial, and members were also able to find the necessary data efficiently and accurately. In a professional environment, Tableau Dashboards would significantly improve workflows related to data visualization and retrieval. Without the hurdle of utilizing manual code, cross-functional teams would likely obtain this data more efficiently. The advantages of this system are clear; by offloading a portion of work related to data visualization and retrieval to a tool created for that purpose, teams will be able to save time and convert their focus to less arduous tasks. Acknowledgements This work is in part funded by the United States Army Research Laboratory under grant number W911NF-10-2-0098. We would also like to thank the original creators of the Tableau Dashboards for their contributions to this work. In addition, this work used the Extreme Science and Engineering Discovery Environment (XSEDE), which is supported by National Science Foundation grant number DMR200035. The authors wish to thank Min Huang, Liangyu Liu, and Gaayathri Sankar for their time and assistance. Finally, we also thank WPI undergraduate students Christopher Vieira and Ashley Schuliger.

References 1. Flemings MC, Suresh S (2001) Materials education for the new century. MRS Bull 26(11):918– 924 2. Waldrop MM (1990) The reign of trial and error draws to a close. Science 247(4938):28–30 3. Fuchs ERH et al (2008) Strategic materials selection in the automobile body: economic opportunities for polymer composite design. Compos Sci Technol 68(9):1989–2002 4. Busch JV (1987) Technical cost modelling of plastics fabrication processes. Dissertation, Institute of Technology, Massachusetts 5. Horstemeyer MF (2012) Integrated Computational Materials Engineering (ICME) for metals: using multiscale modeling to invigorate engineering design with science. Wiley, Hoboken 6. Zhou T, Song Z, Sundmacher K (2019) Big data creates new opportunities for materials research: a review on methods and applications of machine learning for materials design. Engineering 5(6):1017–1026 7. Rodrigues JF et al (2021) Big data and machine learning for materials science. Discov Mater 1(1):1–27 8. National Science and Technology Council (US) (2011) Materials genome initiative for global competitiveness. Executive Office of the President, National Science and Technology Council 9. Popovich A, Sufiiarov V (2016) Metal powder additive manufacturing. New trends in 3D printing. IntechOpen, Philadelphia

Investigating the Suitability of Tableau Dashboards and Decision Trees …

701

10. Cooke A, Slotwinski J (2012) Properties of metal powders for additive manufacturing: a review of the state of the art of metal powder property testing. US Department of Commerce, National Institute of Standards and Technology 11. Kiani P et al (2020) A statistical analysis of powder flowability in metal additive manufacturing. Adv Eng Mater 22(10):2000022 12. Touzé S, Rauch M, Hascoët J-Y (2020) Flowability characterization and enhancement of aluminium powders for additive manufacturing. Addit Manuf 36:101462 13. Clayton J, Millington-Smith D, Armstrong B (2015) The application of powder rheology in additive manufacturing. JOM 67(3):544–548 14. Lee Y, Simunovic S, Kate Gurnon A (2019) Quantification of powder spreading process for metal additive manufacturing. No. ORNL/TM-2019/1382; CRADA/NFE-17–06812. Oak Ridge National Lab. (ORNL), Oak Ridge, TN (United States) 15. Valente R et al (2020) Classifying powder flowability for cold spray additive manufacturing using machine learning. In 2020 IEEE international conference on big data (big data). 2919– 2928. https://doi.org/10.1109/BigData50022.2020.9377948 16. Zhang J, Habibnejad-korayem M, Liu Z et al (2021) A computer vision approach to evaluate powder flowability for metal additive manufacturing. Integr Mater Manuf Innov 10:429–443. https://doi.org/10.1007/s40192-021-00226-3 17. Druzgalski CL et al (2020) Process optimization of complex geometries using feed forward control for laser powder bed fusion additive manufacturing. Addit Manuf 34:101169 18. Tofail SAM et al (2018) Additive manufacturing: scientific and technological challenges, market uptake and opportunities. Mater Today 21(1):22–37 19. Cotteleer M, Joyce J (2014) 3D opportunity: additive manufacturing paths to performance, innovation, and growth. Deloitte Rev 14:5–19 20. Herzog D et al (2016) Additive manufacturing of metals. Acta Materialia 117:371–392 21. Yap CY et al (2015) Review of selective laser melting: Materials and applications. Appl Phys Rev 2(4):041101 22. Lee YS et al (2018) Role of scan strategies on thermal gradient and solidification rate in electron beam powder bed fusion. Addit Manuf 22:516–527 23. Jafarlou DM et al (2021) Solid-state additive manufacturing of tantalum using high-pressure cold gas-dynamic spray. Addit Manuf 47:102243 24. Tan Y et al (2021) Comprehensive evaluation of powder flowability for additive manufacturing using principal component analysis. Powder Technol 393:154–164 25. Snow Z, Martukanitz R, Joshi S (2019) On the development of powder spreadability metrics and feedstock requirements for powder bed fusion additive manufacturing. Addit Manuf 28:78–86 26. Balbaa MA et al (2021) Role of powder particle size on laser powder bed fusion processability of AlSi10mg alloy. Addit Manuf 37:101630 27. Heelan J et al (2020) Effect of WC-Ni powder composition and preparation on cold spray performance. Coatings 10(12):1196 28. Ko I, Chang H (2017) Interactive visualization of healthcare data using tableau. Healthcare Inform Res 23(4):349–354

Part XX

Algorithm Development in Materials Science and Engineering

A Finite Difference Analysis of the Effect of Graphene Additions on the Electrical Conductivity of Polycrystalline Copper William Frazier, Bharat Gwalani, Julian Escobar, Joshua Silverstein, and Keerti S. Kappagantula

Abstract A finite difference method was used to explore the effect of graphene on the bulk electrical conductivity of copper-graphene composites. In this capacity, grain orientation information from pure copper and copper-graphene composites was used to generate synthetic 3D microstructures. The electrical conductivity of these microstructures was calculated using the finite difference method assuming different average grain sizes. From these calculations, we demonstrate that when high-conductivity grain boundaries are present within the microstructure arising from the presence of graphene, an increase in the bulk electrical conductivity is observed. On the other hand, the difference in textures between copper and copper-graphene composites may not account for a significant difference in bulk electrical conductivity. In comparison, the copper grain size has a considerably larger effect on electrical conductivity as previously anticipated. This is one of the first demonstrations of a physical basis for enhanced conductivity composites and presents pathways for further investigations on the effects of composite microstructural features, material interfaces, and graphene content on electrical performance. Keywords Microstructure · Materials simulation · Electronic materials

Introduction The addition of graphene (Gr) to metallic materials for the purpose of enhancing metal conductivity has been an attractive goal due to the high electrical conductivity and electron velocity of graphene [1]. Recently, Cao et al. reported the fabrication of copper (Cu)–Gr nanocomposite materials with an electrical conductivity of 117% IACS or 67.86 MS/m using only 0.008% Gr by volume, with the Gr able to channel perhaps 3,000 times the amount of electrical current as bulk Cu [2]. In electrodeposited films, attempts to create Cu–Gr composites have led to improvements in electrical conductivity of 20–30% [3, 4]. However, Hidalgo–Manrique et al. note W. Frazier (B) · B. Gwalani · J. Escobar · J. Silverstein · K. S. Kappagantula Pacific Northwest National Laboratory, 902 Battelle Boulevard, Richland, WA 99354, USA e-mail: [email protected] © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_67

705

706

W. Frazier et al.

that other attempts to do so have resulted in decrease in net conductivity, which may be related to the presence of defects within the fabricated microstructure or anisotropic effects due to processing [5]. Cao et al. note that such increases are made achievable through “elaborate interface design and morphology control” [2]. The addition of Gr to Cu to form composite materials generally results in a decrease in grain size [6, 7]. Notably, this means that any increase in electrical conductivity yielded from the presence of Gr in Cu is counteracted by the presence of additional grain boundaries, which is generally understood to increase the electrical resistivity of polycrystalline materials [8]. Experimental studies have shown that refinement of Cu from micron-scale grains to nanoscale grains can reduce its electrical conductivity by approximately an order of magnitude [9, 10]. Additionally, the grain boundary character plays a significant role in the overall Cu conductivity. The presence of dense amounts of twin grain boundaries, for example, has a significantly smaller effect on conductivity than high-angle grain boundaries [8, 11]. This means that considerable grain refinement can be achieved in Cu while maintaining high conductivity by creating a nano-twinned material [10]. The interplay between these discussed microstructural factors and the addition of Gr on the overall conductivity of a Cu–Gr nanocomposite is therefore a complex function of this microstructural information, which has not been analyzed in a substantial way, and the effect the fabrication process on the distribution of Gr. This work explored the effects of microstructural features on Cu and Cu–Gr composite electrical properties, namely varying Cu grain sizes and the fraction of grain boundaries housing Gr in the composite.

Methods In this work, the finite difference code was set up to work in the following way. First, a cubic 150 × 150 × 150 cell microstructure containing 875 grains was generated using a Potts model algorithm detailed elsewhere [12–14]. Each grain was assigned an orientation sampled at random from the respective EBSD maps of samples manufactured using shear-assisted processing and extrusion technology shown in Fig. 1. In this way, it was possible to represent the ODF of the two samples in the synthetic 3D microstructure. Next, in each case, the microstructure was divided into regions containing grain boundary material and those not containing grain boundary material, which was accomplished by simply checking each cell in the synthetic microstructure for neighboring cells with different orientations. In this context, the closest 26 neighbors were checked. At this point, the cells in the bulk were assigned a single conductivity as reported by Zeng et al. [15], and the cells containing grain boundary material were assigned conductivity σ as a function of the angle of disorientation θ between the abutting grains. The conductivity of the grain boundary material was assumed to scale approximately with the grain boundary energy γ as calculated using the Read–Shockley equation [16].

A Finite Difference Analysis of the Effect of Graphene Additions …

707

Fig. 1 Inverse pole figures of Cu–Gr composite wire (a), pure Cu wire (b), and a histogram of grain boundaries by misorientation angle (c). Black regions in these figures represent areas where a pattern could not be obtained (Color figure online)

σ (θ ) = σBulk − (σBulk − σG B )γ (θ )

(1)

Therefore, assuming no other defects present, the conductivity of each cell containing grain boundary material was calculated as a function of the inclination of the grain boundary with respect to the x, y, and z directions, ϕx , ϕy , and ϕz . σx,y,z (ϕ) =

1 ρParallel

 + sin(ϕ)

1 ρPerpendicular

−

1 ρParallel

 (2)

Here, ρParallel and ρPerpendicular, the extreme possibilities in the overall conductivity of the cell, were calculated based on σ and the thickness d reported by Islamgaliev et al. [17] ρParallel =

  d l −d 1 − l 2 σ (θ ) σBulk

ρPerpendicular =

1 σ (θ )d + σBulk (l − d)

(3) (4)

For the Cu–Gr microstructure, Gr was assumed to be in the form of monoatomic thick flakes with micron-scale lateral dimensions located at the Cu grain boundaries in the polycrystalline material with coherent interfaces with the surrounding Cu atoms. The Gr was modeled to facilitate electron transport across the Cu grain boundaries such that those housing Gr demonstrate very high electrical conductivity. In each calculation, a constant potential difference of 1 V was applied across the synthetic microstructure in order to determine the flow of current through the microstructure. This was accomplished by solving the Poisson equation for the microstructure using finite difference methods, as previously described by Jin [18], who was able to use similar methods to solve for the local current density and charge density as a function of polycrystalline microstructure. At a steady state, the net electrical conductivity for the microstructure was calculated by obtaining the flow of current through a cross section of the microstructure [19]. Note that for the purpose

708

W. Frazier et al.

of these calculations, the application of the potential difference was assumed not to result in any phase transformations or changes in the microstructure that may normally occur.  I =

σ (r )n∇ϕ(r )d A

(5)

Variations of the synthetic microstructures generated were evaluated using this method. First, for both the pure Cu and Cu-Gr microstructures, net conductivity was calculated for cell spacings of 10, 100 nm, 1, and 10 μm, for average Cu grain sizes of approximately 97 nm, 970 nm, 9.7 μm, and 97 μm, respectively. Second, for a Cu–Gr microstructure with the pure Cu ODF and an average grain size 9.7 μm, the net electrical conductivity was determined for composites where Gr was housed at zero percent, twenty percent, fifty percent, eighty percent, and one hundred percent of grain boundaries. In this case, grain boundaries without Gr were assumed to demonstrate a lower conductivity determined by Eqs. (1)–(4) while those with Gr were assumed to have an electrical conductivity approximately that of the bulk Cu material.

Results The electrical conductivity with respect to time across the pure Cu and Cu–Gr microstructures is shown in Fig. 2a, b, for average grain diameters of 97, 970 nm, 9.7 μm, and 97 μm. It can be seen that the conductivity of the microstructure at the steady state decreases as grain size decreases, which is an expected result of the associated increase in grain boundary surface area. In all of these calculations, the

Fig. 2 Conductivity of pure Cu (a) and the Cu–Gr (b) synthetic microstructures, as calculated assuming average grain sizes of 917, 970 nm, 9.7, and 97 μm

A Finite Difference Analysis of the Effect of Graphene Additions …

709

Fig. 3 Conductivity of 9.7 μm Cu microstructures with 0, 20, 50, 80, and 100% of grain boundaries assigned bulk conductivity signifying the presence of graphene at these locations

Cu–Gr microstructure is only slightly more conductive than the pure Cu microstructure. This suggests that any changes in conductivity that occur with the addition of Gr to Cu do not occur as the result of changes in the ODF. The second set of simulations, in which random grain boundaries were assigned an elevated conductivity equivalent to the bulk material signifying the presence of Gr at those locations, show that increasing the conductivity for additional grains leads to an increase in the net electrical conductivity. As a result, the conductivity increases from that calculated for the 9.7 μm microstructure to the bulk conductivity of Cu, 5.81 × 107 S/m. An inspection of the current density maps for each of these microstructures at the steady state demonstrates that the increase in conductivity originates from the elevated conductivity of these grain boundaries, as shown in Fig. 4. When no grain boundaries are assigned the elevated conductivity, that is for the Cu-only sample, nearly all of the grain boundaries are visible as regions in which a relatively small amount of current is present. However, as additional grain boundaries are assigned the bulk conductivity, signifying the presence of Gr at those locations, these regions disappear, until current becomes nearly uniform when 100% of boundaries have the bulk conductivity. Note that the regions with the lowest current density tend to be grain boundaries inclined orthogonally to the z-axis, along which the potential difference is applied. This is consistent with our model’s assumption that the regions containing grain boundary material should have increased or decreased conductivity based on inclination.

Discussion The electrical conductivity of materials is generally understood to be determined by the “mean free path” of electrons they pass through the material [20]. In polycrystalline materials, defects such as grain boundaries, dislocations, second-phase

710

W. Frazier et al.

Fig. 4 Current present in a pure, 9.7 μm Cu microstructures under a 1 V potential difference, as a function of the number of grain boundaries with “special” grain boundaries assigned the bulk conductivity

particles, and impurity atoms create a scattering effect that increases the overall resistivity or alternately decreases the conductivity of the material. For this reason, it is possible to measure dislocation density in a material by measuring its conductivity [21], as is the practice in industry. In the case of Gr, however, due to its highly conductive properties, the opposite effect can occur in that the addition of Gr can decrease the overall resistivity of Cu. However, this improvement is predicated upon the preservation of the Gr within the microstructure and its organization within the bulk material. Another important factor is the development of coherent interfaces between the Cu lattice and Gr to minimize the scattering and facilitate the transport of electron from the Cu into the Gr. In this work, we explore one possible way in which this can occur, which is that the Gr is located at the grain boundaries. Our calculations show that even a small amount of Gr located at the grain boundaries, selected at random, could feasibly lead to appreciable changes in the net conductivity for a micron-scale Cu–Gr microstructure. If it were possible for these grain boundaries to be arranged in a connected network, additional gains in conductivity could be achieved for small additions of Gr to the Cu microstructure. The change in grain boundary conductivity selected is not significantly different from that measured for twin boundaries, which are often present within Cu in frequencies at or exceeding 50% of all boundaries. In these cases, our simulations are consistent with the previous experimental observations that nano-twinned Cu materials have conductivity comparable to coarse-grained Cu. This effect becomes larger as the overall grain structure becomes more refined, because the twinning creates less resistance to the flow of electrons than high-angle grain boundaries.

A Finite Difference Analysis of the Effect of Graphene Additions …

711

The model conservatively notates that the presence of Gr at the Cu grain boundaries would enhance the conductivity locally to that of bulk metal behavior. However, ideally if the interfaces between the Cu and Gr are coherent to minimize scattering, it is entirely possible that the grain boundary conductivity may be higher than that of the bulk metal. In such a situation, the Cu–Gr composite could become considerably more conductive than a coarse-grained Cu microstructure. In the event that Gr is able to improve grain boundary conductivity so that it is higher than bulk Cu, then the addition of grain boundary surface area would yield significant increases in Cu conductivity. The extent to which such an effect could occur and how such a microstructure could be fabricated are not obvious and are subject of further investigations.

Conclusions A finite difference algorithm was applied in order to evaluate the influence of grain size, texture, and the presence of Gr on the electric conductivity of Cu and Cu–Gr nanocomposite microstructures. Simulations on synthetic microstructures demonstrated that while the addition of Gr to the Cu wires only slightly affected electric conductivity by the virtue of changes to its texture, potential changes in the grain boundary character or the arrangement of Gr within the nanocomposite could result in significant increases in net conductivity for even micron-scale grains. Future work will explore the extent to which these changes could foreseeably improve the electronic conductivity in other scenarios. Acknowledgements The Pacific Northwest National Laboratory is operated by the Battelle Memorial Institute for the United States Department of Energy under contract DE-AC06-76LO1830.The authors would like the acknowledge the support from the U.S. Department of Energy Advanced Manufacturing Office to conduct this work. The authors are also grateful for the dedication of Woongjo Choi, Xiao Li, and Aditya Nittala for sample synthesis and Anthony Guzman for preparation of specimens for microstructural analysis.

References 1. Zourdos MC, Sanchez-Gonzalez MA, Mahoney SE (2017) Nanotwinned copper graphene foils—a brief review. Rev Adv Mater Sci 51:160–176. https://doi.org/10.1519/jsc.000000000 0000636 2. Cao M, Xiong DB, Yang L, Li S, Xie Y, Guo Q, Li Z, Adams H, Gu J, Fan T, Zhang X, Zhang D (2019) Ultrahigh electrical conductivity of graphene embedded in metals. Adv Funct Mater 29:1–8. https://doi.org/10.1002/adfm.201806792 3. Jagannadham K (2013) Volume fraction of graphene platelets in copper-graphene composites. Metall Mater Trans A 44:552–559. https://doi.org/10.1007/s11661-012-1387-y 4. Kim Y, Lee J, Yeom MS, Shin JW, Kim H, Cui Y, Kysar JW, Hone J, Jung Y, Jeon S, Han SM (2013) Strengthening effect of single-atomic-layer graphene in metal-graphene nanolayered composites, Nat Commun 4. https://doi.org/10.1038/ncomms3114

712

W. Frazier et al.

5. Hidalgo-Manrique P, Lei X, Xu R, Zhou M, Kinloch IA, Young RJ (2019) Copper-graphene composites: a review. J Mater Sci 54:12236–12289 6. Nasibulin AG, Koltsova T, Nasibulina LI, Anoshkin IV, Semencha A, Tolochko OV, Kauppinen EI (2013) A novel approach to composite preparation by direct synthesis of carbon nanomaterial on matrix or filler particles. Acta Mater 61:1862–1871. https://doi.org/10.1016/j.actamat.2012. 12.007 7. Saboori A, Pavese M, Badini C, Fino P (2018) A novel approach to enhance the mechanical strength and electrical and thermal conductivity of Cu-GNP nanocomposites. Metall Mater Trans A 49:333–345. https://doi.org/10.1007/s11661-017-4409-y 8. Andrews PV, West MB, Robeson CR (1969) The effect of grain boundaries on the electrical resistivity of polycrystalline copper and aluminium. Philos Mag 19:887–898. https://doi.org/ 10.1080/14786436908225855 9. Chen XH, Lu L, Lu K (2007) Electrical resistivity of ultrafine-grained copper with nanoscale growth twins. J Appl Phys 102. https://doi.org/10.1063/1.2799087 10. Sataloff RT, Johns MM, Kost KM (2004) Ultrahigh strength and high electrical conductivity in copper. Science (80–):304 11. Cesar M, Lie D, Gall D, Guo H (2014) Calculated resistances of singe grain boundaries in copper.pdf. Phys Rev Appl 2:044007 12. Sahni PS, Grest GS, Anderson MP, Srolovitz DJ (1983) Kinetics of the Q-state potts model in two dimensions. Phys Rev Lett 50:263–266. https://doi.org/10.1103/PhysRevLett.50.263 13. Zöllner D (2014) A new point of view to determine the simulation temperature for the Potts model simulation of grain growth. Comput Mater Sci 86:99–107. https://doi.org/10.1016/j.com matsci.2014.01.044 14. Donegan SP, Tucker JC, Rollett AD, Barmak K, Groeber M (2013) Extreme value analysis of tail departure from log-normality in experimental and simulated grain size distributions, Acta Materialia Inc. https://doi.org/10.1016/j.actamat.2013.06.001 15. Zeng W, Xie J, Zhou D, Fu Z, Zhang D, Lavernia EJ (2018) Bulk Cu-NbC nanocomposites with high strength and high electrical conductivity. J Alloys Compd 745:55–62. https://doi. org/10.1016/j.jallcom.2018.02.215 16. Read WT, Shockley W (1950) Dislocation models of crystal grain boundaries. Phy Rev 78:275– 289 17. Islamgaliev R, Pekala K, Pekala M, Valiev RZ (1997) The determination of the grain boundary width of ultrafine grained copper and nickel from electrical resistivity measurements. Phys Stat Sol 162(2):559–566 18. Jin YM (2013) Phase field modeling of current density distribution and effective electrical conductivity in complex microstructures. Appl Phys Lett 103. https://doi.org/10.1063/1.481 3392 19. Lei Y, Le Cheng T, Wen YH (2017) Phase field modeling of microstructure evolution and concomitant effective conductivity change in solid oxide fuel cell electrodes. J Power Sources 345:275–289. https://doi.org/10.1016/j.jpowsour.2017.02.007 20. Tian L, Anderson I, Riedemann T, Russell A (2014) Modeling the electrical resistivity of deformation processed metal-metal composites. Acta Mater 77:151–161. https://doi.org/10. 1016/j.actamat.2014.06.013 21. Hähnert I, Wienecke M (1993) Relation between dislocation density, bulk electrical properties and Ohmic contacts of CdTe. Mater Sci Eng B 16:168–171. https://doi.org/10.1016/0921-510 7(93)90036-M

Clustering Algorithms for Nanomechanical Property Mapping and Resultant Microstructural Constituent and Phase Quantification Bryer C. Sousa, Chris Viera, Rodica Neamtu, and Danielle L. Cote

Abstract Tacit assumptions have been made about the suitability of two primary data-driven deconvolution algorithms concerning large (10,000+) data sets captured using nanoindentation grid array measurements, including (1) probability density function determination and (2) k-means clustering and deconvolution. Recent works have found k-means clustering and probability density function fitting and deconvolution to be applicable; however, little forethought was afforded to algorithmic compatibility for nanoindentation mapping data. The present work highlights how said approaches can be applied, their limitations, the need for data pre-processing before clustering and statistical analysis, and alternatively appropriate clustering algorithms. Equally spaced apart indents (and therefore measured properties) at each recorded nanoindentation location are collectively processed via high-resolution mechanical property mapping algorithms. Clustering and mapping algorithms also explored include k-medoids, agglomerative clustering, spectral clustering, BIRCH clustering, OPTICS clustering, and DBSCAN clustering. Methods for ranking the performance of said clustering approaches against one another are also considered herein. Keywords Nanoindentation · Clustering algorithms · Microstructures

Introduction Advances in nanoindentation testing systems’ application, understanding, and functionality have continued with regularity since the formalization of the original OliverPharr (OP) in the late 1980s and early 1990s [1]. Such advancements include the in-situ integration of nanoindentation systems with scanning electron as well as transmission electron microscopes [2], the development of high-strain rate impact testing methods via nanoindentation [3], the ability to quantify stress–strain relations [4], and the ability to perform nanoindentation testing of materials at notably elevated and B. C. Sousa (B) · C. Viera · R. Neamtu · D. L. Cote Worcester Polytechnic Institute, Worcester, MA, USA e-mail: [email protected] © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_68

713

714

B. C. Sousa et al.

cryogenic temperatures [5]. In addition to the advancements, considerable research and development have also been dedicated to the formalization of statistically significant and high-throughput mechanical property mapping at a rate of an indent per second [6, 7]. With the latter in mind, the present work aims to build upon the current state of nanoindentation-based mechanical property mapping and the analysis of the data obtained through such experimental protocols. That said, consideration of prior work related to the data-driven analysis of nanomechanically mapped datasets is considered first.

Background When consideration was initially being given to the potential value of nanoindentation grid arrays for mechanical property mapping, Randall et al. noted that a grid array of 2500 preprogrammed and automated nanoindentation measurements could be successfully obtained over three to four days [8]. However, by 2012, nanoindentation “tomography” remained relatively limited in high-throughput functionality (relative to modern systems), which can be shown by way of considering the work of Tromas et al. via [9]. That is not to say that the work of Randall et al. and Tromas et al. was any less valuable; rather, detailing the history of nanomechanical mapping or grid array protocol implementation with a nanoindenter enables one to contextualize better the degree of advancement achieved since that period. Specifically, as nanoindentation technologies advanced in the 2010s, the rate at which individual indents could be measured continued to the point of an indent per second in the case of Nanomechanics, Inc. (now KLA), via a method named NanoBlitz3D [10]. Such a revolution in the high-throughput nature of nanoindentation mapping implementation can be exemplified by comparing the three-to-four-day timeframe encountered by Randall et al. for 2500 indents to the amount of time required to measure the same number of indents via NanoBlitz3D, which would only be 0.0289 days or just shy of 42 min. With such remarkable testing speeds, nanoindentation grid arrays and nanomechanical property mapping quickly enabled relatively massive datasets to be obtained in realistic timeframes and therefore enabled big data or data-driven techniques to be suitably applied for analyzing the results. For context, Fig. 1 presents a nanoindentation array measured using NanoBlitz3D and the iMicro Pro from KLA that houses 160,000 indentation measurements within one array measured on a 4xxx series steel.

Clustering Algorithms for Nanomechanical Property Mapping …

715

Fig. 1 a A NanoBlitz 3D 4140 steel hardness histogram. b The hardness versus depth data for the same material system displayed in the hardness histogram (a). c The hardness contour plot for the 160,000 indentation measurements within one array measured on a 4xxx series steel. d The hardness versus array’s x-axis position within the nanomechanical property map shown in (c)

Methods and Materials Data-Driven Details The data analysis techniques applied herein include: (1) probability density function (PDF) and deconvolution, (2) k-means clustering and deconvolution, (3) k-medoids clustering and deconvolution, (4) agglomerative clustering and deconvolution, (5) spectral clustering and deconvolution, (6) balanced iterative reducing and clustering using hierarchies (BIRCH) and deconvolution, (7) ordering points to identify the clustering structure (OPTICS) and deconvolution, and (8) density-based spatial clustering of applications with noise (DBSCAN) and deconvolution. (1)

The probability density function determination and deconvolution method rely on the idea of fitting a variable number of normal curves to the PDF of a dataset. A normal curve represents each cluster. If a data point is in each normal curve, then that point and all other points in that normal curve will make up a cluster. The deconvolution method will iteratively fit new normal curves to the data set, keeping track of the best result so far as it does this. The new curves are generated by assigning them to random sections in the probability density function. The best so far is the combination of normal curves that best account for the probability density function of the original data. Combining these normal curves should result in a similar probability density function as

716

(2)

(3)

(4)

(5)

B. C. Sousa et al.

the original, assuming the number of curves is appropriately chosen for the data set. A limit can also be applied to determine an acceptable combination of normal curves to use, as shown by the decon 3.0 program [11], which the deconvolution process in this paper is based on. K-means clustering focuses on generating k number of centroids to represent k different clusters in a data set. A centroid is an arbitrary point in the range of data values representing a cluster’s center. The distance between the point and each centroid must be calculated to determine what cluster a point is affiliated with. The data point will belong to the cluster paired with the centroid it is closest to. The locations of these centroids are determined through a random and iterative process. To start, k centroids are distributed randomly throughout the data set. The algorithm then moves these centroids to minimize the within-cluster sum-of-squares. These centroids are moved iteratively to locations where the within-cluster sum-of-squares is decreased. Once the within-cluster sum-ofsquares has reached its minimum value, the centroids are in their final locations and returned as the clustering result. K-medoids clustering focuses on generating k number of medoids representing k different clusters in a data set. K-medoids is incredibly like that of k-means, differing in the use of medoids instead of centroids. A medoid differs from a centroid as it must be one of the data points themselves. This makes the algorithm more robust when dealing with outliers and noise as it is less likely to have its clusters’ centers closer to undesirable points. Despite this difference, the method performs similarly to k-means, iteratively moving around its medoids until it finds the smallest within-cluster sum-of-squares. Agglomerative clustering is a form of hierarchical clustering. Agglomerative clustering involves creating a dendrogram with all the data points by pairing them together interactively. To start, each data point is represented as its own cluster or group. With each iteration, a metric is computed to determine the absolute difference between each data point and every other data point, making every possible pair of clusters. It will then take the two points with the best metric value for that iteration and combine them into their own cluster. The cluster of the two points now has a new value that represents it to compare with other data points. This first iteration forms the first step of the dendrogram. This process is then done iteratively until only the number of clusters left is specified at the start of the algorithm. Spectral clustering is a form of clustering which performs a low-dimensional embedding of the affinity matrix between samples and then clusters the result using k-means. Spectral clustering takes in all data points and then computes a similarity graph using either a radius (epsilon-neighborhood) or k-nearest neighbors. Once this is completed, it will create a Laplacian matrix. After this, it will compute the first k eigenvectors of its Laplacian matrix to define a feature vector for each object. Finally, after the original data points have been represented in this way, the algorithm runs k-means on these features to separate objects into k classes.

Clustering Algorithms for Nanomechanical Property Mapping …

(6)

(7)

(8)

717

Birch clustering is a form of clustering that builds a Clustering Feature Tree to create Cluster Feature Nodes (CF Nodes) to reduce data. These nodes represent several subclusters called Clustering Feature subclusters. Each of these subclusters stores information involving the data points, allowing them to represent them accurately. This algorithm reduces the amount of data by creating a tree of the data and then clustering the resulting CF nodes in the tree. The clustering algorithm used after this point to further cluster the data is arbitrary. The library used by this paper runs agglomerative clustering. In terms of how the respective tree is formulated, a new sample is inserted into the root of the CF Tree, which is a CF Node. It is then merged with the subcluster of the root that has the smallest radius after merging, constrained by the threshold and branching factor conditions. If the subcluster has any child node, then this is done repeatedly till it reaches a leaf. After finding the nearest subcluster in the leaf, the properties of this subcluster and the parent subclusters are recursively updated. If the radius of the subcluster is obtained by merging the new sample and the nearest subcluster is greater than the square of the threshold, and if the number of subclusters is greater than the branching factor, then a space is temporarily allocated to this new sample. The two farthest subclusters are taken, and the subclusters are divided into two groups based on the distance between these subclusters. OPTICS is a form of clustering incredibly like DBSCAN with a few additions. Along with the fundamental properties of DBSCAN, OPTICS also has two additional metrics. The first metric is a minimum distance to make a given point a core point. The second metric OPTICS uses are known as the reachability distance or the distance between density-reachable points. This reachability metric can then be used to separate clusters, separating clusters every time there are peaks in the reachability metric. DBSCAN is a form of clustering which focuses on the idea of a dataset being separated into areas of high-density data points and low-density data points. The goal of this algorithm is to identify the sections of high-density points into separate clusters. This algorithm requires that points have a minimum number of points in that cluster to be classified as a cluster. The process completes its clustering by creating a circle around each data point and classifying each data point based on the number of points within a radius. Once this is done, clusters can be formed from these points by forming cores. Iteratively, it goes through the process of joining points as follows. First, X is density-reachable from Y when X is in the radius of Y and Y is a core point. Next, X is density-connected to Y when there is a point O where both X and Y are density-reachable from O. All density-connected points become a separate cluster. Once this process is complete, there will be several clusters from the density-connected point sets and several outliers that did not fit into the requirements of being densityconnected with a minimum number of points. Moreover, HDBSCAN is a form of clustering that combines DBSCAN and hierarchical clustering. It is like DBSCAN but does not use a fixed cutoff as the radius around a point to group points with. It instead handles any offshoots in the dendrogram by discarding

718

B. C. Sousa et al.

them using the minimum cluster size parameter. This creates a denser dendrogram and reduces the number of small extra clusters often present in DBSCAN. This method also relies on generating an estimated probability density function of the data through sampling, varying the number of samples to find a balance between a noisy PDF and one that may be too smooth. HBSCAN also has another parameter that specifies the minimum cluster size, which must be balanced to prevent too many extra clusters from forming from being too low or merging too many clusters together from being too high.

Data Pre-processing Once several clustering methods had been explored, it became apparent that the data needed to be transformed for the clustering methods to create optimal models. This was done through interpolation, correcting outliers, and separating outliers from data sets to be added on later. While plotting the original data set, it became apparent that nulls were a constant issue. A single null value can prevent the software from displaying a map. To counter this, any nulls in the original data are interpolated by basing them off their neighbors. To do this, the project uses SciPy’s interpolate library. It can find all points which are null and then interpolate them based on their neighbors. To interpolate the data, it runs through a two-step processing using two different interpolation algorithms. The first algorithm generates the most accurate guess of a data point possible based on the surrounding neighbors. However, this is not guaranteed to fill all null values in with a numerical value. If the null data point has many null neighbors, it will be unable to generate a value. To compensate for this, the data is run through another interpolation method which is guaranteed to fill in every null value regardless of the number of null neighbors at the cost of accuracy. Instead of using a calculation like the first algorithm, it picks the neighbor closest to it and uses that value. Once this process has been completed, all null values have been corrected. Due to the focus on clustering, some required the data to be cleaned before clustering. K-means and agglomerative clustering, for example, is very prone to be skewed by outliers. If there is a small group of outliers far away from the data points, then it is very likely that a cluster will only be composed of outliers and take away from the analysis of the substance, especially when these outliers are defects. To counter this, the data were cleaned to remove all outliers. Due to the lack of a standardized metric, existing statistical methods were explored. The first method involved taking a sample’s mean and standard deviation and defining any point 3 standard deviations or more away from the mean for an outlier. The second method involved calculating the first quartile, the first quartile, and the interquartile range. After this, any value outside the range of the first quartile −1.5 * IQR and the first quartile +1.5 * IQR were defined as outliers. The method involving the mean and standard deviation was much more successful in accurately removing outliers. Once outliers had been removed, the resulting null values would be interpolated just as

Clustering Algorithms for Nanomechanical Property Mapping …

719

they were in the previous section. This allowed an entire grid of data for clustering and mapping. After exploring removing outliers, a sample was chosen to be mapped where the outliers were essential to identify. These involved a small section of a material whose hardness value was more considerable to an unexpected material. Another process was developed to prevent outliers from causing the clustering methods to fail but still consider outliers. This process involved defining the outliers the same as above. Instead of removing and interpolating the values, the data values would be separated into two sections, one containing outliers with higher-than-expected values and another containing outliers with lower-than-expected values. The expected data values, which were not classified as outliers, would then be clustered. After this, they would be recombined with the outliers. When recombining the outliers, they would be identified as being in their own clusters. All outliers with lower-than-expected values would exist in a cluster, and those with higher-than-expected values would exist. If outliers existed on both sides of a dataset and k-means clustering was used without outliers into three clusters, the resulting contour plot would have five clusters. This allowed the clustering algorithms to perform as intended while still marking off anomalies. After developing a framework to generate clustering models on more optimal data and exploring the evaluation of these models as possible, it became necessary to use a standardized metric to compare them after implementing the clustering methods above. The following metrics were explored to solve this issue. After generating the clustering configurations used in this paper, it became practical to compare how similarly two clustering methods performed. Unlike the other evaluation techniques, this would not determine how well a clustering model performs, only how similar it is to another clustering model. The rand index score is used to compare the results of two clustering outputs and determine their similarity. If two sets were the same, it would produce 1.0; if they were completely different, it would produce 0.0. SciPy’s adjusted rand index score was used for this project, which allowed for values to be lower after taking chance into account. This allowed the quantitative measurement of how similar clustering configurations were to one another after running them on the same data set. Most importantly, if the original data set had each data point with a labeled phase fraction, the metric could then be run between each clustering configuration and the labeled data. This would produce a metric for how well the clustering configuration scored to correctly identify the material phase at each (x, y) location.

Engineering-Driven Details and Initial Performance The primary material considered during the present work was a Pb–Sn soldering alloy. The Pb–Sn soldering alloy utilized was formulated with 60% Sn and 40% Pb with a 2% (by weight) leaded rosin-activated flux core. The 60/40 soldering alloy system was selected due to the solders’ near eutectic nature. Given the near eutectic nature

720

B. C. Sousa et al.

of the selected alloy system, along with consultation of prior literature of relevance on 60/40 solder solidification microstructures, Pb40/Sn60 microstructures house two microstructural phases. Specifically, the two microstructural phases housed within Pb40/Sn60 solders include an alpha-Pb dendritic constituent in addition to an alphaPb/beta-Sn near eutectic constituent, which has previously been shown to have separable micromechanical properties (such as mean contact pressure, that is, indentation hardness); thus offering an economically viable and easily processable material for as-solidified microstructural property mapping and subsequent data analysis for dualphase fraction quantification. Due to the thermodynamically unstable nature of the soldered microstructure obtained via a soldering iron, experimental methods were applied shortly after solidification and metallurgical preparation. In terms of metallurgical preparation, the as-solidified 60/40 soldering alloy was compression mounted in black phenolic resin using mounting materials and a compression mounting system from Buehler (Lake Bluff, IL USA). Upon compression mounting in phenolic resin, the sample was mechanically polished to a mirror finish wherein a 0.05 um colloidal silica suspension-based final polishing step was employed using an automatic mechanical polishing suite sourced from Buehler. Buehler’s automatic polisher and compression mounting system, which was utilized in the present study, are housed and maintained at Worcester Polytechnic Institute (Worcester, MA, USA) as part of the Buehler Center of Excellence affiliated with the Metals Processing Institute. Following the mounting, grinding, and polishing procedures implemented, scanning electron microscopy (SEM), digital microscopy, and nanoindentation-based mechanical mapping was performed for a dual-phase fraction or phase area percentage determination benchmarking. Regarding SEM analysis, a tabletop Zeiss (Oberkochen, Baden-Württemberg, Germany) Evo MA-10 series scanning electron microscope was employed. An accelerating voltage between 5 and 10 kV was used during SEM analysis alongside a working distance of 10.5 mm and a secondary electron detector. An example SEM-captured micrograph, which was measured after nanomechanical mapping was performed, is presented in Fig. 2. As noted above, nanomechanical mapping was performed before secondary electron-based SEM analysis. Consistent with the dual-phase microstructure discussed for the Pb40/Sn60 soldering alloy, the light gray constituents captured in Fig. 2 represent the near eutectic alpha-Pb/beta-Sn phase while the dark gray constituents represent the dendritic alpha-Pb phase. The relatively spherical and dark alpha-Pb features can also be observed in the near eutectic primary phase of the light gray beta-Sn. As for the 50-by-75 array of indents shown in the SEM micrograph of the soldering alloy presented in Fig. 2 followed from nanoindentation mapping using an InForce 1000 mN actuator, diamond Berkovich nanoindenter tip from Micro Star Technologies, Inc. (Huntsville, TX, USA), which has since been obtained by Bruker (Billerica, MA, USA), and KLA’s (Milpitas, CA, USA) iMicro Pro, which was manufactured by Nanomechanics, Inc. (Oak Ridge, TN, USA), before KLA acquired Nanomechanics, Inc. Rapid nanomechanical mapping with the iMicro Pro system was achieved through the use of the NanoBlitz 3D test method.

Clustering Algorithms for Nanomechanical Property Mapping …

721

Fig. 2 SEM-captured micrograph, measured after nanomechanical mapping was performed on the as-solidified 60/40 soldering alloy

That said, dynamic or continuous stiffness measurement (CSM) nanoindentation testing was performed before applying the NanoBltiz 3D test method to identify an average target load associated with the desired target nanoindentation mapping depth of approximately 100 nm. After that, the nanomechanical map was obtained by defining the target applied load at each location within the grid array per the dynamically/CSM-determined load associated with a 100 nm nanoindentation depth. As a result, the relatively recently refined nanoindentation spacing of ten times the depth rule-of-thumb demonstrated in [10] was able to be applied herein as well for enhanced nanomechanical property contour plotting/mapping. Furthermore, implementing the k-means clustering protocol built into the commercially available nanoindentation data analysis software known as InView, which is associated with commercial nanoindentation systems from KLA, basic benchmarking initialization of clustered and deconvoluted phase fractions was achieved. Further details surrounding benchmarking nanoindentation insights obtained are subsequently presented too. Additional benchmarking was procured by way of image analysis using ImageJ and computational thermodynamic analysis via Thermo-Calc. In terms of computational thermodynamic analysis, the commercial software used was Thermo-Calc

722

B. C. Sousa et al.

2021b, which enabled equilibrium-based volume fractions of the dendritic (alphaPb) phase as well as the primary constituent of the near eutectic (beta-Sn) phase such that the results could be compared as a benchmark against the various clustering techniques described in the Data-driven Details subsection of the Methods and Materials section of the present work. Furthermore, equilibrium-based computational analysis via Thermo-Calc is achieved through the “CALculation of PHAse Diagrams,” or the CALPHAD technique. Moreover, the computationally assessed system was defined within Thermo-Calc 2021b via the soldering alloy-based database, denoted within Thermo-Calc as TCSLD3: Solder Alloys v3.3. Accordingly, the mass percent of Pb was set to 40%, while the mass percent of Sn within the system was set to 60%, given the Pb40/Sn60 composition of the soldering alloy experimentally considered herein. Furthermore, beyond the use of the database, the conditional temperature and pressure defined for the single-point equilibrium calculation were set to 294.15 °K and 100,000 Pa, respectively. At the same time, the system size was set to 1.0 mol. Finally, one may consider the details surrounding the use of image analysis via ImageJ herein. First, a JPEG formatted digital micrograph was obtained using the digital microscope accompanying the iMicro Pro nanoindenter for image analysis via ImageJ. Then, said JPEG-based digital micrograph was initially opened within ImageJ 1.53e and cropped to remove regions containing shadowing and edge effects. After that, the JPEG-based image file was converted to a TIFF-based file format. Upon TIFF reformatting of the cropped JPEG-based image, thresholding was applied to the 8-bit TIFF-based image such that binarization of the alpha-Pb (light constituents shown in Fig. 3a) and the beta-Sn (darker constituents shown in Fig. 3a) was achieved. After that, the area of the binarized micrograph shown in Fig. 3b associated with alpha-Pb (the black features in Fig. 3b) and the beta-Sn (the white features in Fig. 3b) could be quantified as an area percentage relative to the total surface microstructural area shown in Fig. 3b post-thresholding. To establish a ground truth for comparison of clustering and deconvolution algorithm results with one another and with independent methods of phase fraction determination, nanoindentation mapping via KLA’s respective software and test method gave 29.1% of the Beta-Sn dominant phase, ImageJ suggested 31.79% of the same phase, and the Thermo-Calc volume phase fraction computed at ambient equilibrium for the 60–40 alloy gave 30.36%. Follow-on work based upon the framework detailed herein and using the methods detailed in the Data-driven Details subsection of the Materials and Methods section of the present work will enable identification of the optimal data clustering techniques for nanomechanical property and phase mapping.

Clustering Algorithms for Nanomechanical Property Mapping …

723

Fig. 3 a Digital micrograph of soldering alloy experimentally considered; b The binarized digital micrograph from ImageJ pre-processing; c Nanoindentation hardness contour plot obtained using NanoBlitz 3D and iMicro Pro; and d The k-means clustered phase map

References 1. Oliver WC, Pharr GM (1992) An improved technique for determining hardness and elastic modulus using load and displacement sensing indentation experiments. J Mater Res 7(6):1564– 1583 2. Minor AM, Morris JW Jr, Stach EA (2001) Quantitative in situ nanoindentation in an electron microscope. Appl Phys Lett 79(11):1625–1627 3. Sudharshan Phani P, Oliver WC (2017) Ultra-high strain rate nanoindentation testing. Materials 10(6):663 4. Hay J (2019) U.S. Patent No. 10,288,540. U.S. Patent and Trademark Office, Washington, DC 5. Wheeler JM, Armstrong DEJ, Heinz W, Schwaiger R (2015) High-temperature nanoindentation: the state of the art and future challenges. Curr Opin Solid State Mater Sci 19(6):354–366

724

B. C. Sousa et al.

6. Vignesh B, Oliver WC, Kumar GS, Phani PS (2019) Critical assessment of high-speed nanoindentation mapping technique and data deconvolution on thermal barrier coatings. Mater Des 181:108084 7. Yang M, Sousa B, Smith R, Sabarou H, Cote D, Zhong Y, Sisson RD (2021) Bainite percentage determination and effect of Bainite percentage on mechanical properties in austempered AISI 5160 steel. Mater Perform Character 10(1):110–125 8. Randall NX, Vandamme M, Ulm F-J (2009) Nanoindentation analysis as a two-dimensional tool for mapping the mechanical properties of complex surfaces. J Mater Res 24(3):679–690 9. Tromas C et al (2012) Hardness and elastic modulus gradients in plasma-nitrided 316L polycrystalline stainless steel investigated by nanoindentation tomography. Acta Materialia 60(5):1965–1973 10. Phani PS, Oliver WC (2019) A critical assessment of the effect of indentation spacing on the measurement of hardness and modulus using instrumented indentation testing. Mater Des 164:107563 11. Nˇemeˇcek (2009) Nanoindentation of heterogeneous structural materials. PhD diss, Czech Technical University in Prague

Part XXI

Biological Materials Science

Biodegradable Superabsorbent Polymers Kaylon Draney and Jeffrey Bates

Abstract In addressing issues of plastic accumulation in the environment and negative impacts of plastic degradation, the development of biobased alternatives is crucial in solving these hazards. Single-use, disposable hygiene products, such as diapers and feminine hygiene, significantly contribute to plastic waste. These products often contain non-biodegradable, synthetic, and superabsorbent polymers. In this research, biobased superabsorbent polymers have been designed and synthesized, using biological crosslinker and backbone components to create a hydrogel system, which absorbs water into the polymeric matrix. The hydrogels are synthesized using chitosan and sodium alginate as the backbone foundation and genipin as the crosslinker, which are all commonly found in nature. Through chemical ratio alteration, including the crosslinker to backbone ratio, the superabsorbent polymers successfully absorb and retain water. The characterization of the hydrogels, including the absorbency capacity, absorbency retention, performance under a load, and performance with ion presence, has proven that fully biological superabsorbent polymers are possible. Keywords Absorbency · Biodegradability · Biopolymer · Superabsorbent polymer

Introduction Plastic waste has been a highly discussed issue in public discourse, largely due to the urgency of this crisis. Approximately 300 million pounds of plastic are produced every year, half of which is single-use plastic [1]. Municipal landfills contain household waste products that contain large amounts of synthetic polymers in various forms such as packaging films, storage containers, carpet fibers, and absorbent hygienic products. Only a small variety of polymers found in consumer products are bio- or photo-degradable, but most public landfills are not properly equipped with facilities or technologies to biodegrade or photodegrade solid waste components. Nearly 25% K. Draney · J. Bates (B) Materials Science and Engineering, University of Utah, Salt Lake City, UT 84112, USA e-mail: [email protected] © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_69

727

728

K. Draney and J. Bates

of the anticipated 24 million tons of plastic products produced in the United States reach landfills every year [2]. The entire life cycle of current plastic, from production to disposal, is creating a significant environmental problem, due to carbon emissions upon production, how these plastics degrade in nature, and plastic waste accumulation. While high carbon emissions from production directly contribute to climate change, these plastics often will break down into smaller pieces that are ingested by animals, creating a significant impact on the environment and wildlife. Furthermore, as these plastics degrade, toxins are released into the water, leading to contamination [3]. While there are many solutions that show potential, a significant factor in reversing these issues will be creating biobased alternatives to current synthetic and petroleum-based plastics. The differences between synthetic and biological polymers are important in addressing the problems of plastic accumulation and their effect on the environment. While there are various differences between the two types of polymers, biological polymers typically have many kinds of monomers. As opposed to biological polymers, there are only a few different monomers in synthetic polymers. Both synthetic and biological polymers are used in many applications. The purpose of this project is to analyze the effects of each type within single-use plastics [4]. A more comprehensive comparison of synthetic and biological polymers in single-use products can be seen in Table 1. The general structures of synthetic and biological polymers are essentially the same. Polymers are made up of many monomers which are small chemical molecules that can build on each other to create a polymer. A large contributor to single-use plastic waste is superabsorbent polymers (SAPs). SAPs are a specific type of polymer known as a hydrogel, which are polymer networks able to retain water without compromising the structural integrity of the network. These features of the polymer matrix within hydrogel structures are necessary for the development of products whose primary function is the absorption of liquids [5]. This type of polymer is often able to absorb up to 300 times their weight in liquids, making them useful in various applications [6]. As the polymer network absorbs water, the structure begins to expand, often resulting in the swelling of the material. These polymers have various commercial applications, including personal care products such as diapers, feminine hygiene products, and incontinence products. There are also many other applications Table 1 Comparison of drawbacks of synthetic polymers to the benefits of biological polymers in single-use products [4]

Synthetic polymer

Biological polymer

Unstable intermediate molecules

Renewable

Long-lasting environmental impact

Biological materials

Effects in soil, landfills, and water Biodegradable Not a candidate for circular economy

Short lifespan of environmental impact

Specialized degradation

Nutrients for organisms

Hundreds of years to degrade

Cradle-to-cradle

Biodegradable Superabsorbent Polymers

729

such as enzyme immobilization, preparative chromatography, agricultural mulches, and controlled release devices [6]. The backbone of the polymer can alter the properties of the hydrogel significantly, depending on the functional groups and the associated properties. The crosslinker provides resistance to the hydrogel structure when water is being absorbed. There are two types of crosslinking in polymer structures, chemical crosslinks, and physical crosslinks. Chemical crosslinks create permanent covalently bonded networks, while physical crosslinks typically involve crystalline domains. The polymerization and crosslinking process starts with thermal- or photo-initiation [5]. The purpose of this study is to examine the absorbency capacity of biological polymer materials that have the potential for use as superabsorbent polymers.

Superabsorbent Polymers Materials and Methods In both initial studies and the actual synthesis of the biological SAP, similar synthesis methods were followed. Each study and test required the use of a backbone monomer and a crosslinking monomer. Chitosan, sodium alginate, and genipin were used as received from Sigma Aldrich. In order to create the biological SAP, the backbone of the polymeric structure was first synthesized. To begin the process, a mixture of chitosan and sodium alginate, or 100% of either compound, was synthesized through suspension. Using a beaker, the compound, or mixture of the compounds, was placed into isopropyl alcohol and stirred at room temperature, with a magnetic stir bar and stirring machine, until the powder completely dissolved. Sodium hydroxide was then added incrementally over a 25 min period and continuously stirred using the stir bar. After the mixture reacted, monochloroacetic acid was added while stirring and heating the solution to 60 °C for approximately three hours. The solution was monitored to ensure that it did not exceed 60 °C in order to prevent boiling. Using a filtering flask setup, the solid precipitate was filtered out of the solution and then dried at room temperature [7]. Once the solid completely dried, the solid was then added to deionized (DI) water and mixed with genipin. This mixture was left at room temperature and stirred, in order to remove air bubbles, for 15 h. After the designated time passed, the mixture was rinsed using DI water and dried at room temperature. This process created an interpenetrating polymer network where genipin is the crosslinker. The polymer is typically in a large polymeric slab, which can be broken down to proceed with absorbency testing and other characterization techniques [7].

730

K. Draney and J. Bates

Testing Methods The synthesized SAP must meet various industrial standards in order to transition into industry-grade applications. Those standards include a high absorbency, performance under a load, various liquid retention, and have high absorbency capacity. Each of these requirements will be tested in the lab. The polymer composition can also be altered accordingly to best fulfill these standards.

Absorbency Testing As the SAP will be used in applications where high amounts of liquid will be absorbed, the polymer must have a high absorbency capacity. In order to properly measure how much liquid is absorbed, two absorbency tests will be performed, using DI as a baseline. The two testing methods being used to measure the performance of the polymer are an oversaturation test and a simple water addition test. The oversaturation test places the synthesized polymer into a beaker containing an excess of water and left to saturate for a long period of time, typically overnight. The polymer is weighed using a basic scale both before and after the oversaturation process. Weighing the polymer at both of these instances will allow for the percent mass increase to be calculated. Similar to the oversaturation test, the simple water addition method will measure the mass of the polymer both before and after the addition of DI water. In the simple water addition method, the polymer will be placed on a dry surface, and using a pipette, DI water will be placed onto the polymer until the matrix no longer absorbs anything. The percent mass increase of both methods can be compared in order to determine a relative time of absorbency. Since the applications of superabsorbent polymers vary greatly and often absorb various liquids, the effect of the presence of ions must be assessed. The absorbency tests use DI water as a baseline; however, to test the effect of ions, DI water can be replaced with ionized water in all methods. By comparing the results of DI water and ionized water, the effect of ion presence can be determined. Similarly, any liquid can be used in any method in order to observe the effect of that liquid.

Absorbency Under a Load Being that many applications of superabsorbent polymers perform under a load, such as diapers, pet pads, and feminine hygiene products, the polymer must perform properly and still meet standards under these conditions. In order to test performance under a load, the polymer will be synthesized and saturated, using the previously described methods. After saturation of the SAP, various amounts of pressure will be applied to the polymer using a weight. The polymer mass will be measured before and after the load to determine the amount of liquid lost during performance. Weight

Biodegradable Superabsorbent Polymers

731

versus mass loss graphs can be deducted from this process, which allows for an assessment of the polymer’s success. Similarly to performance under a load, superabsorbent polymers need to retain the water during movement or use to prevent leaking of any liquid, which is important in many applications. The water retention capability of the synthesized SAP will be tested by saturating the polymer with DI water to maximum capacity, using one of the saturation methods, and then placing the sample in a centrifuge and spun. Using the centrifuge will remove any excess water that is not retained in the SAP structure and will thus indicate how much water is retained during use.

Results Using similar testing and characterization techniques, the absorbency of synthetic and biological polymers was successfully assessed. It was found that synthetic, industrygrade SAPs absorb 300–700 g of water for every 1 g of SAP [8, 9]. In these tests, sodium polyacrylate was used, being the most common SAP. Fossil-fuel-based SAPs, such as sodium polyacrylate, were found to have a very negative impact on the environment. During production, sodium polyacrylate generates various toxins which lead to water and air pollution. Sodium polyacrylate was also found to be highly absorbed in landfill soils and wastewater, which poses a serious problem due to the high toxicity of the SAP. In addition to the harsh environmental impact, sodium polyacrylate and other synthetic SAPs were found to not be biodegradable, taking hundreds of years to degrade [8–10]. Similar to that of synthetic SAPs, tests have also been performed using biological polymers. There is not a lot of research on fully biological polymer networks, which includes a biological backbone and crosslinker monomer; however, many studies have been performed using synthetic monomers in conjunction with biological monomers. While the ideal biological SAP would have a biological backbone monomer and biological crosslinker monomer, this research can provide a foundation in developing a fully biological SAP. As seen in Table 2, the absorbency was tested for these hydrogels. In current research, completely biological backbone and crosslinker monomers are being used, specifically chitosan, sodium alginate, and genipin. Using the synthesis and absorbency techniques outlined, hydrogels have been synthesized with these biological components to serve as a baseline for future research. Five main hydrogel compositions have been synthesized, with various ratios of chitosan and sodium alginate: 100% chitosan, 100% sodium alginate, 25% sodium alginate with 75% chitosan, 50% sodium alginate with 50% chitosan, and 75% sodium alginate with 25% chitosan. After the synthesis of each hydrogel was complete, absorbency testing was performed, using both the oversaturation and simple water addition methods. An absorbency of 8.1 g of water absorbed for every 1 g of polymer (g/g) was found for the sodium alginate hydrogel in both the oversaturation and simple water addition absorbency testing methods, while an absorbency of 11.4 g/g was found

732

K. Draney and J. Bates

Table 2 Absorbency results of tests using biological backbone monomers with varying crosslinker monomers [8, 9] Backbone monomer

Synthetic or biological

Absorbency (g/g)

Crosslinker monomer

Synthetic or biological

Chitosan

Biological

39.5

Acrylic acid

Synthetic

Chitosan

Biological

700

N, N -methylenebis acrylamide

Synthetic

Carboxymethyl cellulose-PEG

Biological

100

Citric acid

Biological

K-carrageenan

Biological

135

N, N -methylenebis acrylamide

Synthetic

Polysaccharides

Biological

None

for the 100% chitosan hydrogel in both absorbency testing methods. Absorbency capacities were also measured for each of the ratios containing a mixture of sodium alginate and chitosan with absorbency capacities of 9.88 g/g for the 25% sodium alginate hydrogel, 7.38 g/g for the 50% sodium alginate hydrogel, and 1.64 g/g for the 75% sodium alginate hydrogel. In order to relate the hydrogels to industrial standards, the same testing was performed for sodium polyacrylate. An absorbency capacity of 59.6 g/g was found for the sodium polyacrylate sample. A comparison of each of these polymers and their associated absorbency capacity can be seen in Fig. 1.

Fig. 1 Comparison of each hydrogel with varying backbone ratios of sodium alginate to chitosan and industrial-grade sodium polyacrylate

Biodegradable Superabsorbent Polymers

733

The research previously conducted in various studies in relation to biological polymer potential is very promising. While these tests used synthetic components and were not aiming to develop a fully biodegradable hydrogel, the results show that the use of biological backbones is possible while maintaining a high level of absorbency. In some cases, the biological backbone, specifically chitosan, performed just as well as current industry-grade synthetic polymers. Chitosan performed extremely well when crosslinked with a synthetic crosslinker, with absorbency of up to 700 g/g, which is at the high end of current technology. While there were SAPs in this study that performed well, there were also SAPs that performed very poorly. While this is not ideal, the results are still important. Despite the performance of the SAP, this study proved that biological-based SAPs can be successfully synthesized. With these initial studies establishing a foundation, five fully biological SAPs were synthesized, with ratios of 100% chitosan-based backbone, 100% sodium alginatebased backbone, a backbone of 25% sodium alginate, a backbone of 50% sodium alginate, and a backbone of 75% sodium alginate. Based on absorbency characterization, the chitosan hydrogel performed best of all the hydrogel compositions, with the absorbency capacity significantly decreasing as the ratio of sodium alginate increased. While each SAP was successful, an ideal SAP will have a better absorbency capacity than any of the baselines created in order to meet the industrial standards of sodium polyacrylate, which may be addressed by altering the polymer composition. The absorbency capacity test results remained consistent across all absorbency capacity testing, the oversaturation, and simple water addition test methods, for each composition. Because the simple water addition test is more time effective, provides consistent results with the oversaturation method, and is more accurate due to the ability to control the polymer, the simple water addition test will continue to be used in future testing.

Conclusions With the urgency of the speed at which plastic accumulation is increasing, the need for biodegradable and biobased alternatives to plastic is at an all-time high. SAPs negatively impact the environment through the production and degradation processes and plastic accumulation. In order to address these issues, biobased polymers are being designed and developed. Previous studies successfully created SAPs with biological backbone monomers. These studies used synthetic crosslinkers; however, the results were promising. This research proved that a biological SAP was possible and that there is potential to develop a fully biological SAP. Using these studies as a guide and using standard hydrogel synthesis techniques involving backbone and crosslinking components, five biological SAPs were designed and synthesized. Each SAP used a varying ratio of chitosan and sodium alginate as the backbone, all compositions using genipin as the crosslinking component. Each polymer network successfully absorbed water and further proved that biological SAPs are possible. The absorbency capacity of the

734

K. Draney and J. Bates

hydrogels needs to be significantly increased in order to meet the industrial standard of sodium polyacrylate; however, the initial results are promising and will be explored in future testing. In order to optimize and further develop these SAPs in order to maximize absorbency, more testing is needed. Water and liquid retention testing will be performed in order to determine how much liquid the hydrogel network can retain under a load, as well as measuring the amount of time it takes for the SAP to absorb a liquid. The ion concentration of the liquid absorbed into the matrix will also be tested. This will be done by absorbing liquids with varying pH in order to determine the effect of ion presence on the performance of the SAP. In order to optimize each of these hydrogel characteristics, alteration of the chemical composition will be performed, specifically alteration of the ratio of crosslinker to backbone. In addition to these future directions, optimization of the drying and synthesis techniques and processes will need to occur in order to better standardize and optimize the overall SAP synthesis, which will ideally aid in the performance of the hydrogel. The construction of bioreactors, including the environments and microbes that samples will be placed in, is a challenge for measuring how degradable these materials are. The systems needed to achieve acceptable results largely depend on maintaining optimal conditions. Failure to do so may render inconclusive results. Characterization standards for these materials are dependent on constraints including the expected time frame of degradation, sample composition, and intended use. As polymers derived from natural, renewable resources become increasingly available for single-use consumer applications, and removal of these polymers via composting or repurposing through recycling is the next emerging trend. Non-biodegradable polymeric materials mostly accumulate in landfills and remain intact without any degradation for hundreds of years. Developing fully biodegradable plastics is a crucial step in alleviating environmental stresses induced by municipal waste products.

References 1. Lindwall C (2020) Single-use plastics 101, NRDC (2020) 2. Hamilton J, Reinert K, Hagan J, Lord W (1995) Polymers as solid waste in municipal landfills. J Air Waste Manag Assoc 45(4):247–251 3. Prata JC, Silva ALP, da Costa JP, Mouneyrac C, Walker TR, Duarte AC, Rocha-Santos T (2019) Solutions and integrated strategies for the control and mitigation of plastic and microplastic pollution. Int J Environ Res Public Health 453–500 4. Martin JE (1998) Environmental impact studies of the disposal of polyacrylate polymers used in consumer products. Sci Total Environ 191(3):225–226 5. Ahmed EM (2013) Hydrogel: preparation, characterization, and applications: a review. J Adv Res 6(2):105–121 6. Buchholtz FL, Peppas NA (1993) Superabsorbent polymers: science and technology. HathiTrust 573(7):88–99 7. Kijchavengkul T, Auras R, Rubino M, Ngouajio M, Fernandez R (2006) Development of an automatic laboratory-scale respirometric system to measure polymer biodegradability. Polym Test 25(8):1006–1016

Biodegradable Superabsorbent Polymers

735

8. Chen S-C, Wu Y-C, Lin Y-H, Yu L-C, Sung H-W (2017) A novel pH-sensitive Hydrogel Composed of N,O-carboxymethyl Chitosan and Alginate cross-linked by Genipin for protein drug delivery. J Controlled Release: Off J Controlled Release Soc 96(2):285–300 9. Geissdoerfer M, Savaget P, Bocken N, Hultink E (2017) The circular economy—a new sustainability paradigm? J Clean Prod 143:757–768 10. Essawy HA, Ghazy MBM, El-Hai FA, Mohamed MF (2016) Superabsorbent hydrogels via graft polymerization of acrylic acid from chitosan-cellulose hybrid and their potential in controlled release of soil nutrients. Int J Biol Macromol 89:144–151

Characterization of Multi-walled Carbon Nanotube Reinforced into Poly(3Hydroxybutyrate-Co-3-Hydroxyvalerate) (PHBV)-Epoxidized Natural Rubber 50 (ENR50) Biofilms A. Turner, S. Zainuddin, D. Kodali, and S. Jeelani Abstract Biopolymers such as poly(3-hydroxybutyrate-co-3-hydroxyvalerate) (PHBV) can be tailored potentially by adding carbon nanoparticles and natural toughening agents to improve their multifunctional properties for a wide variety of applications. In this study, PHBV biopolymer was blended with epoxidized natural rubber (ENR50) to improve the ductility of the biofilm. The inclusion of ENR50 made the composite an amorphous material which decreased crystallization percentage and caused crystallization enthalpy to be non-existent. Furthermore, the addition of ENR50 caused a 14.62% decrease in the elastic modulus, while the storage modulus decreased approximately 29.41%. To improve the elastic modulus of the blended biofilm, 0.3, 0.5, and 1.0% multi-walled carbon nanotubes (MWCNTs) were inserted into the biofilm. The addition of MWCNT to PHBV-ENR50 resulted in a noticeable improvement in the tensile strength compared to the PHBV-ENR50. When analyzed for tensile strength, incorporating MWCNT into the blended polymer resulted in an improvement in the elastic modulus and storage modulus when compared to the PHBV-ENR50 polymer. The 0.3% and 0.5% MWCNT-PHBV-ENR50 biocomposites resulted in a 7.0 and 7.78% increase in the elastic modulus, respectively, in comparison with the PHBV-ENR50. When the MWCNT-PHBV-ENR50 biocomposites were analyzed for storage modulus by dynamic mechanical analysis, 0.3% had a 6.25% decrease, while the 0.5% had a 17.24% increase compared to the PHBV-ENR50. These findings show that when adding ENR50 into a semi-crystalline polymer, the crystallinity will decrease the crystallization of the material which causes a decrease in elastic modulus. The biofilm can be used as a replacement for synthetic plastic food containers. Keywords PHBV · MWCNT · ENR50: chloroform · Solvent casting · FTIR · Mechanical properties · Thermal properties

A. Turner (B) · S. Zainuddin · D. Kodali · S. Jeelani Department of Material Science, Tuskegee University, Tuskegee, AL 36088, USA e-mail: [email protected] © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_70

736

Characterization of Multi-walled Carbon Nanotube Reinforced …

737

Introduction Petroleum-based polymers are causing air pollution and creating biohazards that are negatively affecting the environment. To fight global warming, numerous companies are looking for ways to reduce their use of plastic materials, in addition to improving the ways they dispose of these plastic materials. To limit the amount of waste created on the planet, environmental-conscious companies are looking at more ways to incorporate the use of biodegradable polymers in their businesses. The importance of employing biodegradable polymers in manufacturing is mainly to reduce the usage of petroleum-based polymers. In turn, biodegradable polymers improve the environment by helping to fight against increases in global warming and by improving people’s health [1, 2]. A good substitute for petroleum polymer is a poly(3-hydroxybutyrate-co-3hydroxyvalerate) (PHBV), a semi-crystalline thermoplastic polyester, which is derived from the polyhydroxyalkanoates (PHA) family. Members of the PHA family are derived from lineal biopolyester mixed with hydroxyalkanoates [3–5]. PHBV is a microbial biopolymer with excellent biocompatible and biodegradable properties such as absorption capacity, biological origin, low cytotoxicity, and piezoelectricity. These properties make PHBV an ideal candidate for targeted biomaterial-application research and a thermoplastic polymer [3, 4]. Although PHBV has several outstanding positive characteristics, its beneficial usages are not without some reservations. Some of the concerns associated with it include its lack of mechanical strength, thermal properties, wettability, porosity, and water sorption. Currently, there is an increase in research that specifically looks at ways to improve or enhance the properties of existing biodegradable materials [3]. The structure of the PHBV polymer can be modified by adding other polymers, nanoparticles, natural fibers, etc. to improve its mechanical and thermal properties. To improve its toughness, epoxidized natural rubber (ENR) was blended with PHBV at a 50-mol percentage (ENR50). ENR50 has been used to enhance the ductility of a brittle polymer. ENR is a material that has high toughness properties, and it is related to the original natural rubber (NR) with an addition of the epoxy groups on the backbone chain [6–8]. NR is also a biopolymer that is made up of cis-1,4-isoprenyl repeating units in its molecular chains. The epoxidized group has been shown to improve the tensile strength and fatigue properties of a polymer. However, the presence of the epoxy group reduces other mechanical and thermal properties like elastic modulus and crystallization temperature [6]. The addition of NR to a biodegradable material can improve the biomaterial’s properties, especially its mechanical properties. ENR can be used in fields like automotive, aerospace, thermal, and electrical insulating systems. ENR has been blended with plastics to make closed mold forms like sponges, curing tubes, carpet underlays, connectors, curing flaps, bumpers, heavy-duty pads, seals, gaskets, and wheels. This natural toughening agent has been used with thermoplastic and thermoset composites [7, 8].

738

A. Turner et al.

To further improve the structure of PHBV, multi-walled carbon nanotubes (MWCNTs) were added to the polymer mixture. Long employed for its electronic, thermal, and mechanical properties, the addition of carbon nanotubes helps improve the mechanical properties in polymers, metal, and ceramic matrix composites. MWCNTs, the cylindrical large molecules consisting of a hexagonal arrangement of hybridized carbon atoms, are made by rolling up a single sheet or multiple sheets of graphene [9–11]. MWCNTs are more commonly used because of their cost advantage over single-walled carbon nanotubes (SWCNTs) [12–15]. The trust of this study focuses on blending ENR50 with PHBV-MWCNTs to see if the mixture improves the mechanical and thermal properties.

Experimental Materials and Methods Materials PHBV polymer, ENR50 natural rubber form, MWCNT nanopowder, and the solvent, chloroform, were purchased from Sigma Aldrich (St. Louis, MO). The bacterial grade PHBV, which contains ~12% mol of hydroxyvalerate and ~88% mol of hydroxybutyrate, was supplied by Goodfellow Corporation (Coraopolis, PA). The average material (pallet form) size was ~0.5 mm. MWCNT was purchased from Nanostructured & Amorphous Material Inc. (Katy, TX). MWCNT is a carboxylic (COOH) functionalized material with a purity of ~95%, and various diameter lengths of 10– 20 nm and 10–30 µmENR50 are commercially manufactured in Muang Mai Guthrie PCL (Muang, Phuket Thailand). And, various diameter lengths of 10–20 nm and 10– 30 µmENR50 are commercially manufactured in Muang Mai Guthrie PCL (Muang, Phuket Thailand).

Sample Preparation As shown in Fig. 1, MWCNT was dispersed into 110 ml of chloroform in a 400 ml glass beaker and sonicated for 1 h and 30 min at an amplitude of 65% with 40/20 s on/off pulses. After the sonication, 10 g of PHBV and 531 mgENR50 were added into the MWCNT/chloroform solution and mixed with a magnetic stirrer for a minimum of 10 h. The resulting solution was then vacuum mixed using 100.2 kPa pressure for 15 min at a speed of 1500 rpm to remove any entrapped air bubbles. Finally, the solution was vacuum mixed, poured into a baking pan, and allowed to cure for 24 h at room temperature.

Characterization of Multi-walled Carbon Nanotube Reinforced …

739

Fig. 1 Fabrication flow chart of the biofilms Left (PHBV) and Right (1% MWCNT-PHBV-ENR50)

Fourier Transform Infrared Spectroscopy Structural characterization was conducted on samples using Thermo Scientific Nicolet iS5 equipped with iD7 AtR, and samples were scanned from 400–4000 cm−1 with a resolution of 4 cm−1 .

Thermal Testing Differential Scan Calorimetry Crystallization characteristics and melting temperature were analyzed using differential scanning calorimetry using TA instruments Q1000. Measurements were carried out in a nitrogen atmosphere. Samples were heated from 25 to 250 °C with a heating rate of 10 °C /min.

740

A. Turner et al.

Thermogravimetric Analysis Thermogravimetric analysis (TGA) was carried out using a TA instrument Q500. The prepared samples were heated up to 600 °C from 30 °C at a heating rate of 10 °C/min. The gas that was used was nitrogen gas at a rate of 40 ml/min. The temperature at the onset of degradation, maximum degradation temperature (from the peak of the DTG curve), and the inorganic content (residue left at 600 °C) were calculated.

Mechanical Test Tensile Test The tensile test was running under the ASTM D882-02 standard. Biocomposites films were cut according to the dimensions specified by the standard. The average sample had a width of 23 mm, a thickness around 0.25 mm, and the length was 117.5 mm. The tensile test was done on a Zwick Roell equipment at room temperature 25 °C, load cell of 2.5 kN, and a crosshead speed of 3 mm/min. The test includes five samples of each material. Tensile modulus, tensile strength, and strain at break were calculated.

Dynamic Mechanical Analysis The dynamic mechanical analysis was done by using TA instrument Q80. The test ran an isothermal temperature at 30 °C for 5 min before doing a frequency sweep from 1–100 Hz. The gas that was used was air at 65 psi. An average of four samples was tested. The average width was 5.5 mm, the thickness was about 0.2 mm, and the length was 40 mm.

Results and Discussion Fourier Transform Infrared Spectroscopy The purpose of using Fourier transform infrared spectroscopy (FTIR) was to observe the chemical structure and bonding between PHBV, ENR50, and MWCNTs. Figure 2 represents the FTIR spectra of MWCNTs ENR50, PHBV, and PHBV/MWCNTs samples. The peak detected at 1560 cm−1 correlates to the carboxylate anion stretch mode. At 2960 cm−1 , the asymmetric methyl stretching bands occurred at 2923 and 2853 cm −1 , and asymmetric/symmetric methylene stretching bands occurred. The occurs align with the PHBV 2870 peak in which was corresponds with the symmetric

Characterization of Multi-walled Carbon Nanotube Reinforced …

741

Fig. 2 FTIR graph PHBV, MWCNT, ENR50, and MWCNT-PHBV-ENR50

and asymmetric vibration of CH. For PHBV, ENR50, and PHBV/MWCNTs, it has a sharp peak at 1700 cm−1 that is resulted because of the stretching vibration of the -C = O group in PHBV. The weak peak at 1450 cm−1 comes from the vibration of the methyl group, and the peak at both the 1350 and 1380 cm−1 comes from the symmetric and asymmetric vibration of CH. The MWCNT did not show any reaction during the FTIR graph mainly because the amount of

Thermal Test Results TGA Results Table 1 and Fig. 2 showed an increase during the 5% onset degradation temperature, but it decreases during the maximum decomposition temperature. With respect to neat, PHBV-ENR50 had a 2.32% increase when adding ENR50 and adding MWCNTs, and 0.3% MWCNT-PHBV-ENR50 had a 3.41%, while the 0.5% Table 1 TGA temperature at 5 and 10% weight loss and residue % as the MWCNT content increases in the PHBV-ENR50 biofilms Materials

PHBV (neat)

T 5%

265.47 271.64 °C °C

274.53 °C

275.5 °C

262.65 °C

T 10%

282.39 277.89 °C °C

281.28 °C

281.31 °C

275.53 °C

0.99%

1.13%

1.71%

Residue % 0.80%

PHBV-ENR50 0.3% MWCNT 0.5% MWCNT 1% MWCNT w/PHBV-ENR50 w/PHBV-ENR50 w/PHBV-ENR50

0.85%

742

A. Turner et al.

Fig. 3 TGA graph of neat (PHBV), PHBV-ENR50, and MWCNT-PHBV-ENR50 when increasing the amount of MWCNTs

MWCNT-PHBV-ENR50 had 3.77%. Adding MWCNTs and ENR50 did not improve the onset and maximum degradation temperature. Since MWCNTs are nucleation agents, it did not help improve the degradation temperature. ENR50 did not affect how degradation on these graphs either since a small amount of it blended into the PHBV polymer. Figure 4 shows that the second degradation temperature and the weight percentage were higher when blending ENR50 into PHBV compared to the PHBV (neat). ENR50 improved the stability with the main block of PHBV and that allowed the composite to decompose at a higher temperature (Fig. 3).

DSC Results Adding ENR50 into the MWCNT-PHBV did not affect the heat enthalpy, crystallization, and melting point. However, as shown in Table 2, the ENR50 decreases the crystallization for the PHBV-ENR50 and MWCNT-PHBV-ENR50 compared to the regular PHBV (neat). The PHBV (neat) crystallization temperature was around 45.12, but when blending ENR50 into the polymer, the composite became an amorphous material. In previous literature, blending PHBV-ENR50 has caused two glass transition temperatures because adding more than 30% of ENR50 into the polymer can cause the polymer to be immiscible. Since the PHBV-ENR50 biofilm has only one glass transition temperature, then it means that it is miscible with each other when lowering the amount of ENR50 to about 5%.

Characterization of Multi-walled Carbon Nanotube Reinforced …

743

Table 2 DSC results from values of melting enthalpy, crystallization percentage, cooling enthalpy, and cooling crystallization as MWCNT content increases in the PHBV-ENR50 biofilms Materials

PHBV (neat)

PHBV-ENR50 0.3% MWCNT 0.5% w/PHBV MWCNT w/PHBV

1% MWCNT w/PHBV

Melting enthalpy (J/g)

27.25 ± 1.96

28.28 ± 1.05

27.69 ± 1.08

27.75 ± 1.69

26.14 ± 0.88

Crystallization (%)

30.51 ± 1.20

30.40 ± 1.14

29.81 ± 1.11

29.84 ± 1.82

28.11 ± 0.95

Melting point (°C)

160.69 ± 1.01 160.88 ± 0.54 161.74 ± 1.23

160.10 ± 0.53 161.28 ± 0.60

Cooling enthalpy (J/g)

35.47 ± 2.88 N/A

1.21 ± 0.55

3.34 ± 1.06

5.79 ± 0.22

Crystallization temperature (°C)

45.12 ± 0.60 N/A

26.49 ± 0.315

27.44 ± 0.63

30.28 ± 1.00

Crystallization (%)

38.14 ± 3.10 N/A

1.30 ± 0.59

2.88 ± 0.64

6.23 ± 0.24

Mechanical Test Results Tensile Results Table 3 shows that adding ENR50 into the PHBV-MWCNTs decreased the elastic modulus compared to the PHBV and PHBV-ENR50. The tensile strength decreased compared to the PHBV and PHBV-ENR50, but the displacement breakage increased compared to the PHBV and PHBV-ENR50. Since integrating ENR50 into PHBV, the biofilm crystallization decreased in caused the elastic modulus to decrease compared to the neat, but ENR50 did improve the elongation of the film at breakage. The ENR50 improved the tensile strength of the biofilm compared to the PHBV. The addition of MWCNTs did not improve the elastic modulus nor tensile strength for the film because of the dispersion of MWCNTs. The MWCNTs did not disperse completely throughout the film in which caused the MWCNTs to agglomerate together that led to decreasing the elastic modulus and tensile strength.

Dynamic Mechanical Analysis As shown in Fig. 4, PHBV had a higher modulus compared to the PHBV-ENR50 and MWCNT-PHBV-ENR50 because of the degree of crystallinity in the material. The crystalline regions play as physical crosslinks and as filler particles because of their finite size, which would increase the modulus substantially [4]. Just like the DSC results, blending the ENR50 into the PHBV, the crystallinity within the

257 ± 0.03 14.42 ± 1.46 2.44 ± 0.36

301 ± 0.07

13.34 ± 1.00

0.997 ± 0.20

Tensile strength (MPa)

Displacement break (mm)

PHBV -ENR50

PHBV (neat)

Elastic modulus (MPa)

Materials

2.40 ± 1.31

11.64 ± 0.49

275 ± 0.03

0.3% MWCNT w/PHBV-ENR50

2.20 ± 1.30

12.44 ± 0.75

277 ± 0.04

0.5% MWCNT w/PHBV-ENR50

Table 3 Elastic modulus, tensile strength, and displacement break of small increment amount MWCNT into PHBV-ENR50

2.17 ± 0.55

11.35 ± 0.57

259 ± 0.03

1% MWCNT w/PHBV-ENR50

744 A. Turner et al.

Characterization of Multi-walled Carbon Nanotube Reinforced …

745

Fig. 4 DSC graph of neat (PHBV), PHBV-ENR50, and MWCNT-PHBV-ENR50 during heating and cooling stages

polymer decreases compared to the PHBV in which led to the storage modulus decrease compared to the PHBV. When adding a different percentage of MWCNT into the blended polymer, that is when storage modulus increase compared to the PHBV-ENR50 biopolymer. An increase in the storage modulus can be ascribed to the increase in stiffness when increasing the amount of MWCNT in which allows efficient stress transfer between the filler and polymer chain (Fig. 5). Fig. 5 DMA graph (storage modulus vs. frequency) of neat (PHBV), PHBV-ENR50, and MWCNT-PHBV-ENR50

746

A. Turner et al.

Conclusion In this work, good dispersion when adding MWCNTs into a PHBV will lead to improvement of crystallization which enhances the mechanical properties. However, when inserting MWCNTs into PHBV-ENR blended polymer, it slightly enhanced the crystallinity compared to the PHBV-ENR50, but the crystallinity dramatically decreased compared to the neat. Blending the PHBV and ENR50 enhanced the displacement of the biofilm before breakage compared to the neat PHBV. The Young’s modulus decreased approximately 14.62% compared to the neat PHBV. The storage modulus decreased approximately 29.41%. When adding MWCNT into the blended polymer, the elastic modulus and storage modulus showed improvement compared to the PHBV-ENR50 polymer. About 0.3% MWCNT-PHBV-ENR50 had a 7% increase in the elastic modulus compared to the PHBV-ENR50, while 0.5% MWCNT-PHBV-ENR50 had an elastic modulus increase of 7.78%. For the storage modulus, 0.3% had a 6.25% decrease, while the 0.5% had a 17.24% increase compared to the PHBV-ENR50.

References 1. Rwahwire S, Tomkova B, Periyasamy AP, Kale BM (2019) Green thermoset reinforced biocomposites. Elsevier Ltd 2. Vroman, Tighzert L (2009) Biodegradable polymers. Materials 2(2):307–344 3. Rivera-Briso L, Serrano-Aroca Á (2018) Poly(3-Hydroxybutyrate-co-3-Hydroxyvalerate): enhancement strategies for advanced applications. Polymers (Basel) 10(7):1–28 4. Sridhar V, Lee I, Chun HH, Park H (2013) Graphene reinforced biodegradable poly(3hydroxybutyrate-co-4-hydroxybutyrate) nano-composites. Express Polym Lett 7(4):320–328 5. Zhao X, Venoor V, Koelling K, Cornish K, Vodovotz Y (2019) Bio-based blends from poly(3hydroxybutyrate-co-3-hydroxyvalerate) and natural rubber for packaging applications. J Appl Polym Sci 136(15):1–14 6. Anon (1994) Epoxyprene-a speciality polymer. Rubber Dev 47(3–4): 48–51 7. Tanjung FA, Hassan A, Hasan M (2015) Use of epoxidized natural rubber as a toughening agent in plastics. J Appl Polym Sci 132(29):1–9 8. Masek A, Diakowska K, Zaborski M (2016) Physico-mechanical and thermal properties of epoxidized natural rubber/polylactide (ENR/PLA) composites reinforced with lignocellulose. J Therm Anal Calorim 125(3):1467–1476. https://doi.org/10.1007/s10973-016-5682-5 9. Faibunchan P, Pichaiyut S, Kummerlöwe C, Vennemann N, Nakason C (2016) Green biodegradable thermoplastic natural rubber based on epoxidized natural rubber and poly(butylene succinate) blends: influence of blend proportions. J Polym Environ 28(3):1050–1067, 2020. Therm Anal Calorim 125(3):1467–1476 10. Agboola O, Sadiku ER, Mokrani T (2016) Carbon containing nanostructured polymer blends. Elsevier Inc 11. Montanheiro TLDA, Cristóvan FH, Machado JPB, Tada DB, Durán N, Lemes AP (2014) Effect of MWCNT functionalization on thermal and electrical properties of PHBV/MWCNT nanocomposites. J Mater Res 30(1):55–65 12. Lemes P, do TL, Montanheiro A, da Silva AP, Durán N (2019) PHBV/MWCNT films: hydrophobicity, thermal and mechanical properties as a function of MWCNT concentration. J Compos Sci 3(1):12

Characterization of Multi-walled Carbon Nanotube Reinforced …

747

13. Rahman G et al (2019) An overview of the recent progress in the synthesis and applications of carbon nanotubes. J Carbon Res 5(1):3 14. Deborah M, Jawahar A, Mathavan T, Dhas MK, Benial AMF (2015) Spectroscopic studies on sidewall carboxylic acid functionalization of multi-walled carbon nanotubes with valine. Spectrochim Acta - Part A Mol Biomol Spectrosc 139(1):138–144 15. Mertens J, Senthilvelan S (2018) Mechanical and tribological properties of carbon nanotube reinforced polypropylene composites. Proc Inst Mech Eng Part L J Mater Des Appl 232(8): 669–680

Deep Learning and Finite Element Method Towards the Application of Microfracture Analysis for Prevention of Fatigue Fractures in Bones Gerardo Presbítero-Espinosa, José Quiroga-Arias, Inés Hernández-Ferruzca, Bibiana González-Pérez, Carlos Mora-Núñez, Eduardo Macías-Ávila, Álvaro Gómez-Ovalle, Christian Mendoza-Buenrostro, and Marco A. L. Hernandez-Rodriguez Abstract Our current work aims to confirm the identification and growth modalities of microfractures on X-ray computed tomography images. We worked in the automatic identification by deep learning modalities and finite element analysis in microfractures developed in sections of cortical bone tissue. We achieved a modality for detecting microfractures through image processing with a convolutional neural network. Additionally, it was possible to create a meshwork of the microstructure and develop finite element models by differentiating the phases. The studies presented will enable us to define trends in the development of fatigue fractures based on the growth of microfractures. We previously described the distribution of microfracture lengths using the two-parameter Weibull equation towards developing theoretical models to prevent fractures and bone injuries due to fatigue. These affirmations G. Presbítero-Espinosa (B) · E. Macías-Ávila Centro de Ingeniería y Desarrollo Industrial (CIDESI), Sede Nuevo León, 66629 Apodaca, Mexico e-mail: [email protected] E. Macías-Ávila e-mail: [email protected] J. Quiroga-Arias Universidad Aeronáutica en Querétaro, 76278 Santiago de Querétaro, Mexico I. Hernández-Ferruzca · B. González-Pérez · C. Mora-Núñez División Industrial, Universidad Tecnológica de Querétaro, 76148 Santiago de Querétaro, México Á. Gómez-Ovalle Centro de Ingeniería y Desarrollo Industrial (CIDESI), 76125 Santiago de Querétaro, Mexico e-mail: [email protected] C. Mendoza-Buenrostro Centro de Innovación en Diseño y Tecnología (CIDyT), Tecnológico de Monterrey, ITESM, 66629 Apodaca, Mexico e-mail: [email protected] M. A. L. Hernandez-Rodriguez Facultad de Ingeniería Mecánica y Eléctrica de la Universidad Autónoma de Nuevo León, Av. Universidad S/N Ciudad Universitaria, 66451 San Nicolás de los Garza, CP, Mexico © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_71

748

Deep Learning and Finite Element Method Towards the Application …

749

will converge soon with the results currently obtained, confirming a methodology towards precise prediction procedures for the prevention of fatigue fractures in bone and biomedical materials through the development of microfractures and the use of non-destructive tests. Keywords Bone microdamage · Weibull distribution · Fatigue fractures · Deep learning · Finite element analysis

Introduction In bone tissue, fatigue fractures occur due to cyclical stresses, leading to the formation and growth of micro-cracks under compression stress states [1]. The development of micro-cracks is affected by the microstructure of the bone. For example, diffuse damage has been observed in the tensile stress areas, while linear micro-cracks are dominant in the compressive stress areas [1, 2]. Therefore, there is an interaction between external stresses, pre-existing micro-cracks, and the bone microstructure. However, at the experimenting level, access to human bone samples is restricted, limiting these studies. Analyses of microfractures in materials include metals, alloys, ceramics, rocks, and polymers [3, 4]. In this regard, a large field of research is still open for studies to clarify how microfractures generate, grow, and cause fracture of the material in each case. We implemented an analysis methodology of observation of microfractures, which we consider can be quantified and compared with other equal or similar materials concerning their generation and growth due to the external fatigue efforts applied to the material in each specific case to study and analyze. Although our studies have been implemented primarily in cortical bone structures, the methodology is applicable in studying the development of microfractures in other materials. The process is based on a concept called characteristic length, obtained according to the modality in which microfractures grow at the time of fatigue fracture, according to the Weibull equation of two parameters [5]. Integrating Weibull constants into the theoretical model, in the case of bone, it was possible to predict fatigue failure according to how characteristic lengths increased by reducing the intrinsic mechanical properties, such as low BMD and osteoporosis [6]. We aim to confirm our methodology towards its use for the accurate prediction and prevention in human bones in vivo and different industrial and biomedical materials. We analyzed in the previous studies each of the images and marked manually the existing microfractures, consisting of a long and tedious procedure. Our current study aimed to use modern tools, such as image processing and usage of neural networks. Image processing techniques may include computational vision by deep learning in neural networks and gradient detection, which varies and is analyzed to determine the most appropriate value. A recent study [7] compared both methods for the detection of cracks in concrete walls. The author found that deep learning, using the TernausNet

750

G. Presbítero-Espinosa et al.

network, was more effective, obtaining a precision of 81.9% compared to the method by threshold change. Our current work focused on distinguishing microfractures through image processing through a convolutional neural network. The network classified images with and without microfractures, attaining microfracture detection, characterization, and fracture classification. Using image processing techniques and deep learning, we sought the characteristics that described microfracture categorization. Additionally, bone tissue geometry is difficult to generate using finite element software CAD tools. An alternative way to develop geometric parts based on the microstructure is from authentic images captured using experimental characterization. For example, scanning electron microscopy micrographs or computed tomography images can be digitized and later used as geometric parts [8]. Object-oriented finite element analysis (OOF2) allows forming a finite element model using the inherent microstructure into the model [9]. Therefore, it is possible to model the effect of bone tissue phases (i.e., lamellae and osteons) on the mechanical behavior of bone from its exact geometry and particular mechanical properties [10]. In this research work, we developed a fatigue fracture computational mesoscale model of bone tissue, considering the growth, coalescence, and interaction with the microstructure of pre-existing microfractures. The microstructure of bone tissue was modeled from computed tomography and scanning electron microscopy images using OOF2 software.

Materials and Methods Deep Learning for Automatic Identification of Microfractures The detection of microfractures was done jointly by image analysis and processing by semantic segmentation. Pre-processing, Segmentation, Object Detection, and classification and Image Analysis [11] conformed to Digital Image Processing. These steps detected microfractures on computed tomography images from bone samples from the previous investigations [12]. Following these steps, it was possible to obtain a final and more suitable image for a specific application by improving certain characteristics to carry out processing operations. Additionally, MATLAB software image processing tools were the working environment for microfractures detection, giving the possibility of creating a neural network trained through deep learning. The pre-processing of images was given, in turn, by several points, which ranged from the conversion of the image to black and white (binarization), noise elimination filters, elimination of isolated regions, and filling gaps among pixels, to the elimination of porosities in the material, smaller than a determined size. Image segmentation consisted of dividing the sets of pixels that represented a shape into regions, being each of the porosities, microfractures, and elements appearing in the image separated. The segmenting methodology was the following:

Deep Learning and Finite Element Method Towards the Application …

751

Fig. 1 Examples of image variations with data augmentation

1. Detect edges through a Canny filter. 2. Characterize the regions looking for inner and outer edges. 3. Extract data. Applying a Canny filter allows detecting the change in intensity of the pixels through the first Gaussian derivative to reduce noise, followed by the image gradient found through four filters. Finally, these parameters allowed to obtain the gradient and the direction of the edge. Using the procedure for MATLAB enabled to train the neural network using 301 images with fracture and 290 images without fracture. To synthetically increase the number of images, the data augmentation technique was used (Fig. 1). We decided to create a new neural network designed for the detection of microfractures. The neural network could read as an input image, a resolution of 480 × 480 pixels at three color channels; translated to MATLAB, a vector [480 480 3] as input layer. The easy use of the software allowed to train the neural network with the image folders containing or not microfractures. The results of the neural network allowed us to classify the images containing microfractures extracting the data automatically. Specifically, we applied the use of semantic segmentation, being a relatively new methodology for image classification. We aimed to associate a label or category to each pixel present in an image, recognizing a set of pixels making up different categories. We used the following five steps: (a) Tag images; (b) Training (deep learning); (c) Convolutional neural network; (d) Automatic detection; (e) Extract data.

752

G. Presbítero-Espinosa et al.

Fig. 2 Digitized image of the microstructure of the cortical bone together with the respective mesh. The boundary conditions applied to the model are also shown

Microstructure Modelling The mesoscale microstructure OOF2 model was developed from an image of a real microstructure cross-section obtained using X-ray computed tomography. The X-ray computed tomography image was previously digitalized using Image J software in order to differentiate the phases present in the cortical bone tissue. The cortical bone microstructure was separated into the osteonal bone and interstitial lamellae. Later, using OOF2, mechanical properties were assigned to the previously differentiated phases. Finally, a finite element model was implemented to determine the effect of the deformation mismatch generated by the differences in the mechanical properties of the phases of the cortical bone. The differences in mechanical properties were based on the main constituents of cortical bone, these being collagen (elastic modulus ~800 MPa) and hydroxyapatite (elastic modulus ~22100 MPa). Figure 2 shows the digitized image of the microstructure of the cortical bone together with the respective mesh. In the finite element model, the movement was restricted at the bottom, and tensile stress of 100 MPa was applied.

Results and Discussion Deep Learning for Automatic Identification of Microfractures Deep learning training was applied for the corresponding automatic identification after labeling manually in the previous procedure. As shown in Fig. 3, within MATLAB’s Image Labeler tool, the images were labeled with one-pixel precision, indicating manually detected microfractures with a red label and the relatively

Deep Learning and Finite Element Method Towards the Application …

753

Fig. 3 Use of image labeler for image labeling

Table 1 Results of the iterations for the training of the net network Iteration Microfracture Input Resolution Processing Theoretical Data Approximate entry imaging imaging in pixels time precision loss true without (%) (%) accuracy (%) microfracture 81

2

81

36

480 × 480 8 min 8 s

95.65

0

80

3

301a

290a

480 × 480 48 min 14 s

49.15

1

95

a The

36

640 × 640 18 min 3 s 69.57

1

4

70

number of images was synthetically increased by five when using data augmentation

medium and large porosities within the material with a cyan label. The rest of the image was not tagged to allow for less loaded processing. It was decided to create a new neural network designed for the detection of microfractures. The neural network could read as an input image, a resolution of 480 × 480 pixels at three color channels; translated to MATLAB, a vector [480 480 3] as input layer. In the first iteration for network training, an image size of 640 × 640 pixels was configured as an input, reconfiguring the network subsequently for 480 × 480 pixels, being a size with which it was possible to work better, as shown in Table 1.

Semantic Segmentation A neural network net was developed to carry out automatic identification to be trained with computed tomography images, whether or not they contained microfractures.

754

G. Presbítero-Espinosa et al.

Fig. 4 Training statistics for net

Designing the neural network let to have as input data a 480 × 480 pixel image and three color channels. This network is comprised of 15 layers and 14 vector connections, which followed the SGDM criteria: Stochastic gradient descent with momentum to optimize the direction in which the vectors move through the neurons, thus reducing the time of neural training. There were images of different resolutions and higher than the input layer of the network. Thus, the generation of a previous code allowed to carry out the resolution adjustment, which rescaled the set of input images for training. The net network included a PC precision of up to 49.15% with an epoch of 20, having an F1 data loss that approaches 1% (Fig. 4). It was necessary to manually determine scores based on the results obtained, as these numbers thrown by MATLAB did not truly reflect the network capacity. This appeared to be more satisfactory than the theoretical data thrown by the software. The network automatically classified images according to whether they included (Fig. 5a) or not included (Fig. 5b) microfractures. The training results presented by MATLAB software are not conclusive and must be analyzed manually to determine the reliable precision of the neural network. Semantic segmentation allows presenting the characteristics sought within computed tomography images. However, it requires a lot of time to label the images prior to deep learning training.

Microstructure Modelling Figure 6 shows the image analysis process performed for the cortical bone microstruc-

Deep Learning and Finite Element Method Towards the Application …

755

Fig. 5 a Image correctly classified as microfracture; and b Image correctly classified as no microfracture

Fig. 6 a–g Original images of bone tissue generated by means of X-ray computed tomography. b–h The digitization of the images for the separation of the phases is also observed. c–i Finally, the meshes of the microstructures are obtained

756

G. Presbítero-Espinosa et al.

Fig. 7 Displacement field generated in the microstructure of cortical bone

ture. Figure 6a represents an X-ray computational tomography image of a cortical bone segment. This image makes it possible to differentiate the osteons (dark phase) from the matrix (light phase). Figure 6b shows the digitization of the image using Image J software. Finally, Fig. 6c shows the phases’ meshing and the properties’ assignment (red for the osteons and white for the matrix). Figure 6d–f shows a similar image treatment, but in this case, it corresponds to an image of trabecular bone. Figure 6g–i represents a cross-section of the cortical bone. In this image, it is possible to observe a pre-existing crack. Figure 7 shows the results of the simulation formulated in the methodology section. In the microstructure of cortical bone, there is a rigid phase (the hydroxyapatite phase) and a ductile phase (the collagen phase). This difference in mechanical behavior produces a mismatch in their deformation. Figure 7a, b shows the deformation field of the composite microstructure. In this case, collagen is more ductile than hydroxyapatite; the cracks nucleate in the matrix and stop their growth when they reach the osteon. As described in the previous section, the neural network went through iterations to find the most optimal result (Table 1). In the first stage, the network read a 640 × 640 pixel input image in three color channels; this network was trained with 86 images with microfractures and 31 without microfractures. For the second stage, the network was modified to read a 480 × 480 pixel input image in three color channels; the training was carried out with the same 117 images of the previous stage. Finally, the network was kept with the same input resolution; however, the number of training images was increased to 301 with microfractures and 290 without microfractures, adding the data augmentation technique to multiply the number of these by five times. Hence, it is planned for the near future to focus on using the neural network created for, after identifying microfractures, to count characteristics (porosities, area, perimeter, and characteristic length). It is of utmost importance to have a choice of methods to search for the target. Semantic segmentation is a tool that is currently used and allows to increase the precision of object detection. Regarding the microstructure modeling analyses, it was possible to generate a methodology for digitizing images of cortical and trabecular bone. This allowed

Deep Learning and Finite Element Method Towards the Application …

757

differentiating the current phases in the microstructure (osteons, interstitial zone, pre-cracks, and pores). In addition, it was possible to generate meshwork of the microstructure, which served to give rise to finite element models. In the following publication, the cohesive zone model and the extended finite element method will be implemented to predict nucleation and crack growth in this type of microstructures. By using deep learning, future work will involve applying finite element simulations for automatically identifying microfractures’ growth tendencies. Experimental results concerning fatigue fracture and analysis of their generation and growth under X-ray computed tomography will complement these affirmations. Therefore, it will be possible to establish and verify a methodology towards precise prediction procedures for preventing fatigue fractures in bone and biomedical materials through the development of microfractures and the use of non-destructive tests.

Conclusions We worked in the automatic identification of microfractures by deep learning, according to studies previously carried out concerning prediction modalities towards preventing fatigue fractures in bone tissue. Other analyses included modeling with the finite element method. These studies developed on images obtained from cortical sections of bone tissue by X-ray computed tomography. It was possible to verify the correct application of processing by semantic segmentation, in which soon it will be required a choice of methods to count characteristics, such as porosities, area, perimeter, and characteristic length. Additionally, we achieved a procedure for the generation and differentiation of bone phases to generate finite element models, in which nucleation and crack growth prediction will be essential to implementing soon. The comparison with experimental results in under fatigue tests and analysis of the development of microfractures will confirm a methodology based on the prediction and prevention of fracture formation in bone tissue and may be useful in developing new clinical approaches to the problem of osteoporosis. Likewise, the correct development of biomedical materials will be possible in terms of optimal reinforcements on conditions of resistance to fatigue fracture. Acknowledgements We are grateful to the following researchers: José Brayan Linares Gutiérrez and Alejandra Hernández Partida, of the Faculty of Mechanical and Electrical Engineering, Autonomous University of Nuevo León, for the help received towards the correct obtaining of the research results.

References 1. Currey JD (2012) The structure and mechanics of bone. J Mater Sci. https://doi.org/10.1007/ s10853-011-5914-9 2. Diab T, Vashishth D (2005) Effects of damage morphology on cortical bone fragility. Bone 37(1):96–102

758

G. Presbítero-Espinosa et al.

3. Pang HT, Reed PAS (2008) Effects of microstructure on room temperature fatigue crack initiation and short crack propagation in Udimet 720Li Ni base superalloy. Int J Fatigue 30:2009–2020 4. Saini SK, Dubey AK (2019) Study of material characteristics in laser trepan drilling of ZTA. J Manuf Process 44:349–358 5. Presbitero G, O’Brien FJ, Lee TC, Taylor D (2012) Distribution of microcrack lengths in bone in vivo and in vitro. J Theor Biol. https://doi.org/10.1016/j.jtbi.2012.03.027 6. Presbítero G, Gutiérrez D, Taylor D (2017) Osteoporosis and Fatigue fracture prevention by analysis of bone microdamage. In: TMS 2017 146th annual meeting & exhibition supplemental proceedings. https://doi.org/10.1007/978-3-319-51493-2_30 7. Rezaie A, Achanta R, Godio M, Beyer K (2020) Comparison of crack segmentation using digital image correlation measurements and deep learning. Construct Build Mater 261:120474 8. Kim JJ, Nam J, Jang IG (2018) Computational study of estimating 3D trabecular bone microstructure for the volume of interest from CT scan data. Int J Numer Methods Biomed Eng. https://doi.org/10.1002/cnm.2950 9. Reid ACE, Lua RC, Garcia RE, Coffman VR, Langer SA (2009) Modelling microstructures with OOF2. Int J Mater Prod Technol. https://doi.org/10.1504/ijmpt.2009.025687 10. Wang M, Li S, Scheidt A, Qwamizadeh M, Busse B, Silberschmidt VV (2020) Numerical study of crack initiation and growth in human cortical bone: effect of micro-morphology. Eng Fract Mech. https://doi.org/10.1016/j.engfracmech.2020.107051 11. Ramírez J, Chacón M (2011) Redes neuronales artificiales para el procesamiento de imágenes, una revisión de la última década. Revista de ingeniería eléctrica, electrónica y computación 9:1 12. Presbítero G, Hernández M, Contreras Susarrey O, Gutiérrez D (2017) Microdamage distribution in fatigue fractures of bone allografts following gamma-ray Exposure. Acta Bioeng Biomech 19:4

Effect of Suppressing Pressure on the Properties of AZ91 Foamed Magnesium Alloy Hanghang Zhou, Guibao Qiu, Zhenyun Tian, and Qingjuan Li

Abstract Foamed magnesium alloy has good mechanical properties and porosity and is suitable for biological materials, but the disadvantage is that it has poor corrosion resistance. Relevant studies have shown that AZ91 magnesium alloy has good corrosion resistance. Therefore, this paper uses AZ91 magnesium alloy powder as raw material and urea as a pore former. It is hoped that a foamed magnesium alloy material with better corrosion resistance can be developed. Under the conditions of sintering temperature of 550 °C, holding time of 1.5 h, and urea content of 60%, the influence of different powder pressing pressures on the porosity and mechanical properties of the sample was studied. After many experiments, the results show that when the pressing pressure is 11 MPa, the pore distribution of the sample is relatively uniform, and the mechanical properties are better, which meet the performance requirements as a biological material. Keywords AZ91 magnesium alloy · Suppressing pressure · Mechanical properties · Porosity

Introduction Magnesium foam is a new material with three-dimensional pore structure, which is composed of magnesium matrix and a large number of pores. Compared with solid magnesium alloy, foamed magnesium alloy has lower density, better damping, sound absorption, energy absorption, and thermophysical properties [1–3]. At present, the preparation methods of foam metal have been developed, such as casting seepage method, powder metallurgy method, and melt foaming method. As a structural material, magnesium foam has better damping and shock absorption ability, excellent electromagnetic interference resistance, and good thermal conductivity and can also be used as a biological material [4, 5]. In addition, magnesium alloy is also easy H. Zhou · G. Qiu (B) · Z. Tian · Q. Li College of Materials Science and Engineering, Chongqing University, Chongqing 400044, China e-mail: [email protected] © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_72

759

760

H. Zhou et al.

to recycle. However, the research on its properties mainly focuses on the preparation process and mechanical properties and bionic properties. Foamed magnesium is an emerging material with excellent performance. It has a special structure and a large number of connected or unconnected pores in the matrix. It not only has good mechanical properties but also can be used as a functional material [6]. As a kind of foam metal, compared with physical metal magnesium, foam magnesium has low density, good energy absorption rate, sound absorption, and sound insulation performance [7, 8]. It has good biocompatibility and bioabsorption. The pore structure of foamed magnesium is similar to that of human bone. The pore structure of foamed magnesium can be controlled to match the human bone tissue and implanted into the human body instead of human bone [9]. Magnesium is one of the lightest materials in industrial metals, with a density of 1.74 g/cm3. This value is about 2/3 that of aluminum, 2/5 of titanium, 1/4.5 of steel, and 1/4 of iron. The damping performance is far incomparable to other metals [10]. In addition, the foam material itself has the characteristics of many pores, so that it has a smaller density, greater specific strength, specific stiffness, and good sound absorption performance [11, 12]. In the same pore condition, the foamed magnesium alloy has the same or even better mechanical properties than the widely used aluminum foam by virtue of its lighter weight [13–15]. At the same time, the foamed magnesium alloy material has very high recyclability and reusability. Because of this, it is also called “21st Century Green Engineering Material”. Although the unique properties of foamed magnesium alloy make it have a very wide development space in many fields, it has certain defects, such as it is easy to be oxidized, high activity, poor corrosion resistance, and other defects which severely limit it for large-scale preparation and commercial use [16, 17]. In addition, people’s research on foam metal is relatively late, and the research is not sufficient. Therefore, there is still no complete and mature system for preparing foamed magnesium alloy. It is currently almost in the laboratory stage, and magnesium has a porous structure. After that, its easy to be oxidized and easy to corrode defects will be further expanded, which makes its application greatly restricted. It is because of the emergence of these problems that it is more urgent and important to develop a simple and safe product technology method with good corrosion resistance, lower oxidation rate, high production efficiency, and excellent product performance.

Experimental The pressing pressure is an extremely important process parameter in the preparation process of the foamed magnesium alloy. If the pressing pressure is too high, the prepared magnesium foam will be too dense, and it is impossible to prepare a metal material with a reasonable pore structure. In order to study the effects of different pressing pressures on the pore structure and properties of foamed magnesium alloys, AZ91 magnesium alloy powder with better corrosion resistance among magnesium alloys was selected as the matrix material for the experiment, and urea was used as

Effect of Suppressing Pressure on the Properties of AZ91 Foamed …

761

the pore former. The specific raw material components and specifications are shown in Table 1. In order to obtain a foamed magnesium alloy with better corrosion resistance, AZ91 magnesium alloy powder is selected as the research object in this paper. AZ91 magnesium alloy is a relatively mature magnesium alloy powder with relatively good corrosion resistance. Its main components and properties are shown in Tables 2 and 3 as follows. Table 4 shows the experimental scheme for preparing foamed magnesium alloy. The experimental variables were the pressure, which was set as 8 MPa, 9 MPa, 10 MPa, 11 MPa, and 12 MPa, respectively; sintering temperature is 550 °C. Holding time is 1.5 h; urea volume ratio is 60%. According to the thermogravimetric analysis curve of urea, the temperature rise curves of raw pressed billets under different pressing pressures were set in Fig. 1 for Table 1 Ingredients and specifications of raw materials Raw material Purity

Average granularity (μm) Purchase platform

AZ91

99.99% 40–50

Xin Hao alloy wear-resistant welding shop

Urea

99.5%

100–500

Aladdin reagent net

Zinc stearate

99.5%

1–2

Aladdin reagent net

Sn

99.99% 0.5–0.7

Aladdin reagent net

Table 2 Main components of AZ91D (%) Mg

Al

Zn

Mn

Sn

Fe

Ni

Cu

Margin

8.5–9.5

0.45–0.90

0.17–0.4

1 [26]. The MA+ cation has an ionic radius of 1.18 Å and methylammonium lead iodide (MAPbI3 ) perovskite has t = 0.83 which just touches the cubic range of perovskites [27]. However, at around 57 °C, MAPbI3 undergoes a cubic-to-tetragonal phase transition and gradually becomes white (MAPbI3 ·H2 O) or yellow (MA4 PbI6 ·2H2 O) when exposed to humid conditions [28]. The FA+ cation has a larger ionic radius of 2.54 Å as compared to the MA+ cation. As a result, formamidinium lead iodide (FAPbI3 ) perovskite has t = 0.987, approximating the FAPbI3 structure with t ~ 1, and making it more cubic compared to MAPbI3 [29]. However, FAPbI3 perovskite exists in two different phases, i.e. cubic photoactive phase (α-FAPbI3 ) and hexagonal non-photoactive phase (δ-FAPbI3 ). Owing to the greater size and asymmetric nature of the FA+ groups, the FAPbI3 perovskite suffers from a critical issue due to its phase instability, which results in a α-FAPbI3 -to-δ-FAPbI3 phase transition at room temperature. The dynamic mobility of the FA+ cations in the high temperature zone (i.e. >130 °C) ensures that they have isotropic orientation, which helps to sustain the symmetry of the required cubic form. Conversely, the FA+ groups obtain significant preferred orientations in the low temperature zone (i.e. 98%), and indium-doped tin oxide (ITO)-coated glass substrates (sheet resistance: 8–12 /sq) were purchased from Sigma-Aldrich. Finally, the tin (IV) oxide (15% in H2 O colloidal dispersion) was purchased from Alfa-Aesar. All chemicals were used as received.

Preparation of Perovskite Inks and Fabrication of Perovskite Films For the MAPbI3 perovskite absorber layer, a perovskite precursor solution was prepared with MAI (1.5 M) and PbI2 (1.5 M) dissolved in a fixed solvent formulation of DMF:DMSO (4:1). For the mixed MA+ /FA+ cation-based perovskite (i.e. MA1 − x FAx PbI3 ; x = 0.4), FAI (0.6 M), MAI (0.9 M), and PbI2 (1.5 M) were all dissolved in a fixed solvent formulation of DMF:DMSO (4:1); all solutions were stirred at 65 °C overnight. Both MAPbI3 and MA0.6 FA0.4 PbI3 films were fabricated on indium-doped tin oxide (ITO)-coated glass substrates. First, ITO substrates were cleaned rigorously using soapy water, deionized water (DI), ethanol, and isopropanol (IPA) for 20 min., each under sonication. After cleaning, the substrates were N2 blow-dried to remove the residual solvents from the substrates and then thermally treated in a convection

1380

M. Parashar et al.

oven at 70 °C for 30 min. Cleaned substrates were further treated in a UV-ozone plasma for 20 min. before starting the thin film deposition of the perovskite layers. The MAPbI3 and MA0.6 FA0.4 PbI3 perovskite films were deposited by spin coating their respective precursor solutions on top of glass/ITO substrates via a spin-coating program comprising of 10 s at 1200 rpm, with a subsequent ramp to 4500 rpm for 20 s. Toluene, used as the anti-solvent, was dripped dynamically onto the underlying spinning substrate, during the last 10 s of the second stage ramp to 4500 rpm. The MAPbI3 layer was annealed at 120 °C for 15 min., while the MA0.6 FA0.4 PbI3 layer was annealed at the 130 °C for the same duration.

Morphology Characterization FE-SEM was performed using a FEI-Quanta environmental SEM. 1D-XRD measurements were conducted using a Rigaku Ultima III X-ray diffractometer.

Optical Characterization UV–vis optical absorption spectroscopy of the synthesized films was conducted using the Agilent CARY 5000 spectrophotometer. Steady-state PL spectra were measured using a LabRAM HR Evolution spectrometer from HORIBA Scientific, equipped with a ∼532 nm laser source for optical excitation.

Results and Discussion The MAPbI3 and MA0.6 FA0.4 PbI3 perovskite films were fabricated as described in Sect. “Preparation of Perovskite Inks and Fabrication of Perovskite Films” and characterized using scanning FE-SEM, XRD, UV–vis and steady-state PL techniques. The SEM was used to analyze the morphological features of the produced perovskite films, as shown in Fig. 1. The MAPbI3 film showed a higher density of pinholes (Fig. 1a) but after the inclusion of FA+ cation, the morphology of the perovskite structure changes significantly, showing densely packed perovskite grains, largely devoid of pin holes (Fig. 1a). The phase purity of mixed cation perovskites and the integration of the FA+ cation into the MAPbI3 perovskite lattice were investigated using XRD measurements, as shown in Fig. 2. The major diffraction peaks in pure MAPbI3 were observed at ~14.1°, 19.98°, 23.42°, 24.5°, 28.46°, and 31.92°, which correspond to the (110), (200), (211), (202), (220), and (310) crystal planes of the tetragonal perovskite structure (Fig. 2a). After introduction of FA+ , the peaks at 14.1°, 19.98°, 24.5°, 28.46°, and 31.92° shift towards lower diffraction angles to attain a new position at 14.06°, 19.86°, 24.46°,

Solution-Processed Perovskite Photoabsorbers …

1381

Fig. 1 SEM images of the perovskite films; a MAPbI3 and b MA0.6 FA0.4 PbI3

Fig. 2 XRD spectra of the perovskite films for a MAPbI3 and b MA0.6 FA0.4 PbI3

28.28°, and 31.7°, respectively. The greater size of the FA+ cation causes lattice expansion, and this causes a gradual shift to the lower diffraction angles implying the presence of a mixed phase, in which both cations (MA+ and FA+ ) are integrated in the same lattice. Also, the absence of any diffraction peak at ~11.7° implies that the mixed cation perovskite is not affected by the undesirable δ-FAPbI3 phase [33]. The absorbance spectra of the perovskite films were observed using UV–vis spectroscopy as shown in Fig. 3a. The inclusion of FA+ results in the increased absorbance of the perovskite film with its absorption tail approaching in the NIR region; this is evident from the absorption onset of MAPbI3 film (at ~ 780 nm) which gets redshifted to ~813 nm after the addition of FA+ cation in the MA0.6 FA0.4 PbI3 film, suggesting that the inclusion of FA+ cation decreases the bandgap of the perovskite structure which agrees with prior studies [33]. These results were further confirmed by steadystate PL measurements (Fig. 3b), where the PL emission peaks of MAPbI3 and MA0.6 FA0.4 PbI3 films were observed at ~ 768.77 nm and ~779.51 nm, respectively.

1382

M. Parashar et al.

Fig. 3 a UV–vis spectra gathered for MAPbI3 and MA0.6 FA0.4 PbI3 perovskite films and b the PL spectra of the perovskite films for MAPbI3 and MA0.6 FA0.4 PbI3 perovskite films, films were fabricated on glass/ITO substrates

Conclusion To summarize, we have studied the effect of FA+ ion addition into the MAPbI3 perovskite crystal structure to form the mixed cation MA0.6 FA0.4 PbI3 perovskite film using a one-step solvent engineering approach. We compared the MAPbI3 and MA0.6 FA0.4 PbI3 films and found that the addition of FA+ cation in MAPbI3 results in the densely packed morphology with larger grains and fewer pinholes. Also, the effect of FA+ cation on the absorption spectra of the perovskite film is quite significant, increasing the absorbance and bringing down the bandgap value of the mixed cation perovskite absorber closer to the optimum theoretical value needed for a single-junction solar cell according to the Shockley–Queisser limit. The tail of the absorbance spectra also extends into the NIR region after the FA+ addition in the perovskite structure, which is beneficial for the generation of more photogenerated charge carriers in solar cell platforms. Acknowledgements We thank the Office of Naval Research (grant number ONR N00014-20-12597) that enabled us to pursue this work. A.B.K. is also grateful to the support from the PACCAR Technology Institute at UNT and the Endowed Professorship support.

References 1. Chugh S, Adhikari N, Lee JH, Berman D, Echegoyen L, Kaul AB (2019) Dramatic enhancement of optoelectronic properties of electrophoretically deposited C60-graphene hybrids. ACS Appl Mater Interf 11(27). https://doi.org/10.1021/acsami.9b00603

Solution-Processed Perovskite Photoabsorbers …

1383

2. Jayanand K, Chugh S, Adhikari N, Min M, Echegoyen L, Kaul A (2020) B. Sc3N@C80 and La@C82 doped graphene for a new class of optoelectronic devices. J Mater Chem C 8(12). https://doi.org/10.1039/c9tc06145b 3. Min M, Sakri S, Saenz GA, Kaul AB (2021) Photophysical Dynamics in semiconducting graphene quantum dots integrated with 2D MoS2for optical enhancement in the near UV. ACS Appl Mater Interf 13(4). https://doi.org/10.1021/acsami.0c18615 4. Mehta RK, Kaul AB (2021) Black phosphorus-molybdenum disulfide hetero-junctions formed with ink-jet printing for potential solar cell applications with indium-tin-oxide. Crystals 11(5). https://doi.org/10.3390/cryst11050560 5. Mehta R, Min M, Kaul AB (2020) Sol-gel synthesized indium tin oxide as a transparent conducting oxide with solution-processed black phosphorus for its integration into solar-cells. J Vacuum Sci Technol B 38(6). https://doi.org/10.1116/6.0000471 6. Huang H, Pradhan B, Hofkens J, Roeffaers MBJ, Steele JA (2020) Solar-driven metal halide perovskite photocatalysis: design, stability, and performance. ACS Energy Lett. https://doi.org/ 10.1021/acsenergylett.0c00058 7. Lin K, Xing J, Quan LN, de Arquer FPG, Gong X, Lu J, Xie L, Zhao W, Zhang D, Yan C, Li W, Liu X, Lu Y, Kirman J, Sargent EH, Xiong Q, Wei Z (2018) Perovskite light-emitting diodes with external quantum efficiency exceeding 20%. Nature 562(7726). https://doi.org/10. 1038/s41586-018-0575-3 8. Liu P, He X, Ren J, Liao Q, Yao J, Fu H (2017) Organic–Inorganic hybrid perovskite nanowire laser arrays. ACS Nano 11(6). https://doi.org/10.1021/acsnano.7b01351 9. Yang Y, Dai H, Yang F, Zhang Y, Luo D, Zhang X, Wang K, Sun XW, Yao J (2019) Allperovskite photodetector with fast response. Nanosc Res Lett 14(1). https://doi.org/10.1186/ s11671-019-3082-z 10. Hossain RF, Min M, Ma LC, Sakri SR, Kaul AB (2021) Carrier Photodynamics in 2D perovskites with solution-processed silver and graphene contacts for bendable optoelectronics. NPJ 2D Mater Appl 5(1). https://doi.org/10.1038/s41699-021-00214-3 11. Min M, Hossain RF, Adhikari N, Kaul AB (2020) Inkjet-printed organohalide 2D layered perovskites for high-speed photodetectors on flexible polyimide substrates. ACS Appl Mater Interf 12(9). https://doi.org/10.1021/acsami.9b21053 12. Parashar M, Singh R, Shukla VK (2019) Fabrication of perovskite solar cells in ambient conditions. In Materials today: proceedings, vol 34. https://doi.org/10.1016/j.matpr.2020. 03.182 13. Dong Q, Fang Y, Shao Y, Mulligan P, Qiu J, Cao L, Huang J (2015) Electron-hole diffusion lengths > 175 Mm in solution-grown CH3NH3PbI3 single crystals. Science 347(6225). https:// doi.org/10.1126/science.aaa5760 14. Kojima A, Teshima K, Shirai Y, Miyasaka T (2009) Organometal halide perovskites as visiblelight sensitizers for photovoltaic cells. J Am Chem Soc 131(17). https://doi.org/10.1021/ja8 09598r 15. Liang H, Yuan F, Johnston A, Gao C, Choubisa H, Gao Y, Wang YK, Sagar LK, Sun B, Li P, Bappi G, Chen B, Li J, Wang Y, Dong Y, Ma D, Gao Y, Liu Y, Yuan M, Saidaminov MI, Hoogland S, Lu ZH, Sargent EH (2020) High color purity lead-free perovskite light-emitting diodes via sn stabilization. Adv Sci 7(8). https://doi.org/10.1002/advs.201903213 16. Nazeeruddin MK, Snaith H (2015) Methylammonium Lead triiodide perovskite solar cells: a new paradigm in photovoltaics. MRS Bull 40(8). https://doi.org/10.1557/mrs.2015.169 17. Savenije TJ, Ponseca CS, Kunneman L, Abdellah M, Zheng K, Tian Y, Zhu Q, Canton SE, Scheblykin IG, Pullerits T, Yartsev A, Sundström V (2014) Thermally activated exciton dissociation and recombination control the carrier dynamics in organometal halide perovskite. J Phys Chem Lett 5(13). https://doi.org/10.1021/jz500858a 18. Singh R, Suranagi SR, Yang SJ, Cho K (2018) Enhancing the power conversion efficiency of perovskite solar cells via the controlled growth of perovskite nanowires. Nano Energy 51. https://doi.org/10.1016/j.nanoen.2018.06.054 19. Steirer KX, Schulz P, Teeter G, Stevanovic V, Yang M, Zhu K, Berry JJ (2016) Defect tolerance in methylammonium lead triiodide perovskite. ACS Energy Lett 1(2). https://doi.org/10.1021/ acsenergylett.6b00196

1384

M. Parashar et al.

20. Stranks SD, Eperon GE, Grancini G, Menelaou C, Alcocer MJP, Leijtens T, Herz LM, Petrozza A, Snaith HJ (2013) Electron-hole diffusion lengths exceeding 1 micrometer in an organometal trihalide perovskite absorber. Science 342(6156). https://doi.org/10.1126/science.1243982 21. Wetzelaer GJAH, Scheepers M, Sempere AM, Momblona C, Ávila J, Bolink HJ (2015) Trapassisted non-radiative recombination in organic-inorganic perovskite solar cells. Adv Mater 27(11). https://doi.org/10.1002/adma.201405372 22. Younis A, Lin CH, Guan X, Shahrokhi S, Huang CY, Wang Y, He T, Singh S, Hu L, Retamal JRD, He JH, Wu T (2021) Halide perovskites: a new era of solution-processed electronics. Adv Mater. https://doi.org/10.1002/adma.202005000 23. Goldschmidt VM (1927) Krystallbau Und Chemische Zusammensetzung. Berichte der deutschen chemischen Gesellschaft (A and B Series) 60(5). https://doi.org/10.1002/cber.192 70600550 24. Li C, Lu X, Ding W, Feng L, Gao Y, Guo Z (2008) Formability of ABX 3 (X = F, Cl, Br, I) Halide perovskites. Acta Crystallographica Section B: Struct Sci 64(6). https://doi.org/10. 1107/S0108768108032734 25. Goldschmidt VM (1926) Die Gesetze der Krystallochemie. Naturwissenschaften 14(21). https://doi.org/10.1007/BF01507527 26. Li Z, Yang M, Park J-S, Wei S-H, Berry JJ, Zhu K (2016) Stabilizing perovskite structures by tuning tolerance factor: formation of formamidinium and cesium lead iodide solid-state alloys. Chem Mater 28(1). https://doi.org/10.1021/acs.chemmater.5b04107 27. Park N-G (2015) Perovskite solar cells: an emerging photovoltaic technology. Mater Today 18(2). https://doi.org/10.1016/j.mattod.2014.07.007 28. Muscarella LA, Petrova D, Jorge Cervasio R, Farawar A, Lugier O, McLure C, Slaman MJ, Wang J, Ehrler B, von Hauff E, Williams RM (2019) Air-stable and oriented mixed lead halide perovskite (FA/MA) by the one-step deposition method using zinc iodide and an alkylammonium additive. ACS Appl Mater Interfaces 11(19). https://doi.org/10.1021/acsami.9b0 3810 29. Gu L, Zhang D, Kam M, Zhang Q, Poddar S, Fu Y, Mo X, Fan Z (2018) Significantly improved black phase stability of FAPbI3 nanowires via spatially confined vapor phase growth in nanoporous templates. Nanoscale 10(32). https://doi.org/10.1039/C8NR03058H 30. Zheng X, Wu C, Jha SK, Li Z, Zhu K, Priya S (2016) Improved phase stability of formamidinium lead triiodide perovskite by strain relaxation. ACS Energy Lett 1(5). https://doi.org/10.1021/ acsenergylett.6b00457 31. Chen T, Foley BJ, Park C (2016) Entropy-driven structural transition and kinetic trapping in formamidinium lead iodide perovskite. Sci Adv 2(10). https://doi.org/10.1126/sciadv.1601650 32. Mahmud MA, Elumalai NK, Upama MB, Wang D, Wright M, Chan KH, Xu C, Haque F, Uddin A (2016) Single versus mixed organic cation for low temperature processed perovskite solar cells. Electr Acta 222. https://doi.org/10.1016/j.electacta.2016.11.132 33. Salado M, Calio L, Berger R, Kazim S, Ahmad S (2016) Influence of the mixed organic cation ratio in lead iodide based perovskite on the performance of solar cells. Phys Chem Chem Phys 18(39). https://doi.org/10.1039/C6CP03851D

Investigation of Thermal Properties and Thermal Reliability of Ga-based Low Melting Temperature Alloys as Thermal Interface Materials (TIMs) Yifan Wu, Rajath Kantharaj, Albraa Alsaati, Amy Marconnet, and Carol Handwerker Abstract Gallium-based low melting temperature alloys have been proposed as candidates for next generation thermal interface materials (TIMs) due to their high thermal conductivity (~30 W/m*K) and liquidity. However, poor wettability as well as embrittling and corroding effect of Ga on metals have limited their use by the electronics industry. Studies on the relationship between the evolution of thermal properties and interfacial reactions between Ga-based TIMs and metal substrates are thus vital for creating a path forward. We measured thermal conductivity and thermal interface resistance of eutectic Ga-In alloy (EGaIn) sandwiched between two Ni-plated Cu substrates following simulated assembly and accelerated aging. The rapid interfacial reaction between EGaIn and both Ni and Cu at elevated temperatures led to an increase in the thermal conductivity. Further study showed the change in thermal properties was due to the depletion of Ga in the system through intermetallic formation, creating a higher conductivity In-rich alloy. Keywords Ga alloys · Low melting temperature alloys (LTAs) · Thermal interface materials (TIMs)

Introduction For electronic devices, heat dissipation is becoming increasingly challenging as a result of both the increase in power density and the decrease in dimensions of products. Effective thermal management is thus critical to the design and development of electronic devices. One aspect of thermal management is the use of thermal interface materials (TIMs) at interfaces between different components to reduce thermal contact resistance and temperature rises in the system. Thermal contact resistance arises from the inevitable surface roughness of mating surfaces. At a solid–solid interface, apart from a few point-point contacts, most of the interfacial volume is occupied by air, which is a poor conductor of heat. Thermal interface materials Y. Wu · R. Kantharaj · A. Alsaati · A. Marconnet · C. Handwerker (B) Purdue University, West Lafayette, IN, USA e-mail: [email protected] © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_132

1385

1386

Y. Wu et al.

improve heat conduction across interfaces by filling in the air gap and reducing the thermal interface resistances. An ideal thermal interface material has high thermal conductivity, high conformability, and good wetting with the target surfaces, as well as being nontoxic and environmentally friendly. The most extensively used TIMs are thermal greases which combinations of silicone and thermally conductive particle fillers [1]. Due to the intrinsically low conductivity of silicone, the thermal conductivity of even high-performance thermal greases is generally less than 10 W/m K [2]. Gallium-based low melting temperature alloys have recently emerged as candidates for next generation TIMs due to several factors. First, Ga offers a high thermal conductivity of 30 W/m K [3] and a relatively low melting temperature (for a metal) of only 29.8 °C. Further, unlike its near-room temperature liquid metal counterparts, Ga is unique in the way that it is not radioactive (like Rb), explosively reactive (like Fr and Cs), nor toxic (like Hg). By alloying Ga with other elements (e.g., In, Sn, or Zn), a melting temperature well below room temperature can be easily achieved. For instance, the eutectic Ga-In alloy (EGaIn with 24.5 wt.% In) has a melting temperature of 15.5 °C, and the commercial product Galinstan (an alloy of Ga, In, and Sn) has a reported melting temperature of −19 °C [4]. In addition to its low melting temperature, Ga and its alloys are known experience significant undercooling due to the difficulty in nucleation upon cooling, making it possible for Ga-based alloys to maintain its liquid state for an extended temperature range. As a low viscosity liquid, Ga-based alloys can easily flow in the air gaps formed at solid–solid interfaces, offering improved heat flow [5]. Furthermore, in stark contrast to mercury, the most commonly used room temperature liquid metal, Ga, has a very low vapor pressure (10–12 mmHg) [6], making the dry-out issue commonly found in TIMs less of a concern for Ga and its alloys. Despite all these advantages, the wetting behavior and reactivity with other metals have prevented Ga-based low melting temperature alloys from being widely adopted as TIMs in commercial products. In terms of wettability, pure Ga is a high surface tension metal. The surface tension of liquid gallium is 0.707 N/m [7], which means Ga simply does not wet most metal surfaces. Furthermore, Ga is highly susceptible to oxidation and rapidly forms an oxide layer (Ga2 O3 ) when exposed to oxygen, which lowers its surface energy [6]. The low viscosity, high surface tension liquid and its elastic, and high adhesion oxide result in a viscoelastic behavior of Ga-based liquid metals. This complicates their wetting behavior in oxygen-containing environments. The high reactivity of Ga with other metals also presents a challenge. On certain metal substrates, for example Al, Ga wets the grain boundaries and percolates the entire structure causing subsequent structural failure, known as liquid metal embrittlement [8]. On metals like Cu, Ga forms intermetallic compounds (IMCs) rapidly with the solid metal substrate [9]. Since these metals are used extensively in the electronic industry, reaction between Ga and them can lead to premature failure of the device and thus must be mitigated. Considerable work has been done on Ga-based liquid metal alloys, most of which target the wettability issue. Because of the low surface tension and high adhesion offered by the naturally formed Ga2 O3 , a lot of the work tackling the poor wettability of Ga center themselves around the manipulation of the oxide layer [10].

Investigation of Thermal Properties and Thermal …

1387

A common and intuitive method is to incorporate high thermal conductivity solid particle fillers into Ga-based alloys to alter the rheology of the liquid metal [11–17]. Tang et al. [11] synthesized a series of CuGa2 -EGaIn amalgams by doping copper particles into EGaIn matrix with different packaging ratios with the CuGa2 particles forming as a reaction product between Ga and Cu. With increasing packing fractions, the amalgams showed a liquid-like to paste-like transition, an increase in thermal conductivity, and improved wetting on metal surfaces. Similar studies based on the principle of incorporating solid metal particles into Ga-based liquid metal matrices via intermetallic compound formation can also be found in Ga-Ag and GaMg systems [12, 13]. Similar liquid metal-solid particle pastes were also achieved using an oxide-mediated method rather than an IMC formation route in Ga-W, GaDiamond, and Ga-quartz systems [14–16]. Alternatively, Gao et al. [17] devised a method of micro-oxidation to improve wetting by increasing the percentage of Ga2 O3 in the liquid metal matrix through stirring in an oxygen-containing environment. Despite its simplicity, this method suffers from the low thermal conductivity due to the oxide. Instead of modifying the Ga-based liquid metal itself, researchers have also explored the potential of using Ga-based alloys in liquid metal-polymer composites [12, 18, 19]. Although the end products show lower thermal conductivity compared to liquid metal itself, this method has the advantages of preventing leakage of the liquid metal while greatly improve the thermal conductivity of the polymer composite. For example, Tutika et al. [19] achieved a thermal conductivity of 11 W/m K by dispersing EGaIn in a silicone elastomer. Most previous studies on liquid metal TIMs have focused on addressing the wettability issue while preserving or improving the thermal conductivity. However, little work has been done to quantify the thermal performance of such materials in contact with metal substrates. Therefore, in this study, we created Cu-LM-Cu joints using an oxide-mediated spontaneous wetting technique. The thermal performance of the Ga-based alloy as a TIM on a Ni-coated Cu substrate was investigated as a function of thermal history, including aging at 125 °C. The microstructural evolution of the TIM and the substrate due to interfacial reactions, intermetallic formation, and corrosion has been related to changes in thermal properties and composition of the liquid metal. The thermodynamics aspect of Ga-based alloys was also discussed to facilitate the understanding of interfacial reactions and the resulting thermal properties.

Experimental Materials and Sample Preparation The liquid metal eutectic Ga-In (EGaIn) was prepared by combining 75.5 wt.% Ga and 24.5 wt. In (Rotometal). Ni-coated (5 microns, electroplated) Cu substrates were cut into 10 mm × 10 mm pieces and used as the metal substrate to sandwich the liquid metal TIM, as shown in Fig. 1. To facilitate wetting, a layer of Ga2 O3 was deposited

1388

Y. Wu et al.

Fig. 1 a Image of one of the 2 sides of a Cu-LM-Cu joint illustrating the EGaIn within a pocket created on the Ni-coated Cu substrate with double-sided tape; b SEM image of the cross-section of the Ni-plated Cu substrate; c overview of the IR temperature mapping setup; d example temperature map of a Cu-LM-Cu joint with the different regions for analysis outlined

on the surface by scrubbing a small droplet of EGaIn with a cotton swab prior to the application of the bulk liquid. Accelerated aging was carried out to study the effect of processing and thermal aging on the stability of the interface. To mimic the reflow process during assembly, several specimens were reflowed in a DDM Novastar GF12HC-HT 3-zone reflow oven for one and five cycles using a reflow profile for SAC solders, which has a peak temperature of 260 °C. For thermal aging, specimens were annealed at 125 °C for various amounts of time (0 day, 5 days, 10 days, and 15 days) in a Fisher Scientific 725F annealing furnace. A commercial TIM, ARCTIC APT 2560 Thermal Pad, was also tested for comparison.

Thermal Testing and Microstructural Characterization Two specimens that underwent the same thermal treatment were joined together using heat resistant tapes, exposing one edge for IR temperature imaging using a QFI InfraScope MWIR Temperature Mapping Microscope. Calculations of thermal conductivity and thermal interface resistance were based on the principles based on

Investigation of Thermal Properties and Thermal …

1389

the ASTM D-5470 standard adapted for use with infrared thermal mapping [20, 21]. In this setup, the Cu-LM-Cu joint was sandwiched between two aluminum bars. During measurements, the aluminum bar with cartridge heaters served as the heat source, while the one connected to a fixed temperature water chiller served as the heat sink, as shown in Fig. 1. The top of the sample was lightly sprayed with graphite to increase its emissivity. A total of four temperature maps with different heat flux were taken for each sample by changing the power output of the cartridge heater. At each power level, the sample was allowed to reach steady state prior to recording the temperature profile. For analysis, a 1D temperature profile along the heat flow direction was calculated from each 2D map by averaging the temperature normal to the heat flow direction (each column in the image in Fig. 1d). Assuming 1D heat transfer and minimal losses, the heat flux was constant through each layer and across the interfaces. Thus, by measuring the slope in temperature of each layer and the temperature jump at interfaces, the thermal conductivity and thermal interface resistance of the Ga alloy TIM was measured based on: 

q  = −k1

  dT  dT  Tint = −k = ref d x 1 d x r e f Rint

(1)



where q is the heat flux through the sample stack, k1 is the thermal conductivity of the TIM, kr e f is the thermal conductivity of the reference material, Tint is the temperature jump at the interface, and Rint is the thermal contact resistance (in m2 K/W). In this study, Cu was the reference material, thus kr e f has a value of 386 W/m K. After IR imaging, samples were taken apart. Excess liquid metal was brushed off for differential scanning calorimetry (DSC) testing using Thermal Analysis Q2000 in copper pans. The substrates were then rinsed with 5% HCl solutions to remove as much LM as possible to reveal interfacial reaction products. The interfacial reaction products were imaged using SEM (Quanta 650). Energy-dispersive spectroscopy (EDS) analysis was performed to identify intermetallic phases formed.

Results and Discussion Thermal Properties Figure 2 shows the evolution of the thermal conductivity with annealing and reflow. The EGaIn prior to any aging had a thermal conductivity of 29.9 W/m K, in good agreement with existing literature. During annealing at 125 °C, it shows an initial decrease in thermal conductivity after 5 days of aging. However, the thermal conductivity dramatically increased to 37.6 W/m K after 10 days of aging. After 15 days, the thermal conductivity decreased to 32.7 W/m K, which is still higher than its original

1390

Y. Wu et al.

Fig. 2 Thermal conductivity of Ga-based TIMs after a annealing at 125 °C and b reflow

value. Changes in thermal conductivity as a function of reflow cycles show a similar trend. After one cycle of reflow using a SAC alloy reflow profile, the thermal conductivity of the TIM in the joint increased to 43.3 W/m K but decreased to 36.7 W/m K upon further reflow. One advantage of using liquid metal as TIM is the high conformability it offers and henceforth the low thermal boundary resistance. As shown in Fig. 3, prior to aging, the EGaIn-Cu interface is measured to have a remarkably low thermal boundary resistance of 3.4 × 10–8 m2 K/W. In comparison, the ARCTIC APT2560 thermal pad’s thermal boundary resistance was measured to be 2.2 × 10–7 m2 K/W, which is an order of magnitude higher than that of the liquid metal. After 5 days of annealing at 125 °C, the thermal boundary resistance slightly increased to 3.5 × 10–8 m2 K/W, before decreasing to 2.8 × 10–8 m2 K/W after 10 days. The thermal boundary resistance after 15 days of annealing showed a slight increase from the 10 days sample but still remained lower than the original value. A similar trend is also seen in reflowed samples. The increase in thermal conductivity and decrease in thermal boundary resistance suggest an improved thermal performance of EGaIn after accelerated aging on Ni-Cu

Fig. 3 Thermal boundary resistance of Ga-based TIMs after (a) annealing at 125 °C and (b) reflow

Investigation of Thermal Properties and Thermal …

1391

substrates. This is counterintuitive as generally a decrease in thermal performance would be expected from an interface that is deteriorating as the interfacial reaction takes place. One possible explanation for this observation could be the change in alloy composition as a result of rapid interfacial reaction at elevated temperatures, as discussed in Sect. “Interfacial Reaction Products”.

Interfacial Reaction Products The EGaIn/Ni-Cu interface after annealing at 125 °C and reflow is shown in Fig. 4. After annealing for 5 days, intermetallic compound (Ni,Cu)3 Ga7 formed as a result of

Fig. 4 EGaIn/Ni-Cu interface with EGaIn partially removed to reveal reaction products after a 5 days annealing at 125 °C; b 10 days annealing at 125 °C; c 15 days annealing at 125 °C; d 1 × reflow;e 5 × reflow; and f 5 × reflow + 5 days annealing at 125 °C

1392

Y. Wu et al.

Fig. 5 Cross-section of EGaIn/Ni-Cu interface after 15 days of annealing at 125 °C showing spalling of Ni coating as a result of Ga corrosion

Ga and Ni/Cu reaction, occurring as cuboids of varying sizes. Most of the liquid metal on this sample was successfully removed through etching, revealing the underlying substrate. The interfaces of 10 days and 15 days annealed samples show similar features, as shown in Fig. 4b, c. The Ga-based liquid metal became much more difficult to remove as a result of increasing amounts of IMCs. Other than (Ni,Cu)3 Ga7 cuboids, large plates of CuGa2 can been seen on the surface, indicating the breach of the Ni coating by Ga. Further evidence of the breach of the Ni coating can be found in the SEM image of the cross-section of a sample annealed for 15 days. As shown in Fig. 5, after the Ni coating was breached, Ga came into contact with Cu and wet the Ni/Cu interface, causing the spalling of the Ni layer. IMCs formed during the reflow process, as shown in Fig. 4d, e, with different morphologies than annealed samples. Colonies of large sheets of Ni3 Ga7 along with relatively small (Ni,Cu)3 Ga7 cuboids can be seen in both samples. However, further annealing at 125 °C for 5 days following the reflow process resulted in the disappearance of sheet-shaped IMCs and growth of faceted (Ni,Cu)3 Ga7 prisms. SEM images of the interface demonstrate that IMC formation between Ga and Ni is rapid at elevated temperatures. EGaIn used in this study was able to breach the 5 µm Ni coating within 10 days of annealing at 125 °C, leading to even greater extent of corrosion as Ga reacts with Cu more readily. This offers a potential explanation to the observed increase in thermal conductivity after aging. As the Ga/Ni and Ga/Cu take place, the percentage of Ga in the liquid metal TIM decreases, resulting in an enrichment of In. Since In has a higher thermal conductivity than Ga, it is thus reasonable to assume that a more In-rich alloy would show higher thermal conductivity. The decrease in thermal boundary resistance can be explained by the decrease in the Ni layer thickness and the spalling of the Ni coating.

Investigation of Thermal Properties and Thermal …

1393

Thermodynamics of EGaIn Ga-based room temperature liquid metal alloys are known for their large undercooling during solidification as it is difficult for Ga to nucleate. Figure 6 shows a DSC curve of an off-eutectic Ga-In alloy. The undercooling of the alloy is measured to be 33 °C. The loop, which is also the major solidification peak, is an increase in temperature upon solidification as a result of supercooling of the liquid, known as recalescence. DSC testing of EGaIn before and after accelerated aging on Ni-Cu substrates shows a further retardation of the onset temperature of solidification as a result of aging. As shown in Table 1, after 15 days of annealing, the onset temperature of solidification decreased by 29 °C to −47 °C, accompanied by a decrease in heat of fusion from 69.6 J/g to 31.8 J/g. Even one cycle of reflow decreases the onset

Fig. 6 DSC curve of near eutectic Ga-In alloy. Note that the equilibrium melting and solidification temperature for this alloy is 15 °C. Substantial undercooling is observed on cooling

Table 1 Summary of DSC results

Onset temperature of Heat of fusion (J/g) solidification/°C Time 0

−18

69.6

5 days annealing

−28

44.4

10 days annealing −30

32.0

15 days annealing −47

31.8

1 × reflow

−32

67.8

5 × reflow

−32

66.4

1394

Y. Wu et al.

temperature of solidification to −32 °C, while little changes to the heat of fusion are observed in reflowed samples.

Conclusion Ga-based low melting temperature alloys are promising candidates for next generation TIMs. In this study, we focus on the direct measurement of thermal properties of EGaIn on Ni-plated Cu substrates using an IR temperature mapping setup and relate the measurements to intermetallic formation, interfacial reactions, and corrosion. Thermal property measurements confirm that Ga-based liquid metal TIM exhibits excellent thermal performance compared with conventional TIMs. Counterintuitively, an increase in thermal conductivity and a decrease in thermal boundary resistance were observed after accelerated aging either by annealing at 125 °C or reflowing. Interfacial reaction study using SEM shows rapid IMC formation between Ga and Ni at elevated temperatures. The 5-micron Ni coating on the Cu substrate was breached within 10 days of annealing at 125 °C, leading to spalling of the Ni layer. The enrichment of In in the liquid metal system as a result of Ga consumption is the likeliest explanation for the observed increase in thermal conductivity after aging. At the same time, the decrease in Ni layer thickness and direct contact between Cu and Ga explain the decreased contact resistance as a result of aging. The rapid interfacial reaction observed in this study also suggests the necessity of a protective layer to be used on metal substrates in the presence of Ga-based TIMs. Acknowledgements Research funding from Purdue University’s Cooling Technologies Research Center (CTRC), an industry-funded, graduated National Science Foundation Industry/University Cooperative Research Center, is gratefully acknowledged. Alsaati acknowledges the support of a Saudi Arabia Cultural Mission (SACM) fellowship, sponsored by the Saudi Arabian Ministry of Education.

References 1. Gwinn JP, Webb RL (2003) Performance and testing of thermal interface materials. Microelectron J 34(3):215–222 2. Yu W et al (2015) Exceptionally high thermal conductivity of thermal grease: synergistic effects of graphene and alumina. Int J Therm Sci 91:76–82 3. Duggin MJ (1969) The thermal conductivity of liquid gallium. Phys Lett A 29(8):470–471 4. Roy CK et al (2015) Investigation into the application of low melting temperature alloys as wet thermal interface materials. Int J Heat Mass Transf 85:996–1002 5. Dickey MD et al (2008) Eutectic gallium-indium (EGaIn): a liquid metal alloy for the formation of stable structures in microchannels at room temperature. Adv Funct Mater 18(7):1097–1104

Investigation of Thermal Properties and Thermal …

1395

6. Cochran CN, Foster LM (1962) Vapor pressure of gallium, stability of gallium suboxide vapor, and equilibria of some reactions producing gallium suboxide vapor. J Electrochem Soc 109(2):144 7. Hardy SC (1985) The surface tension of liquid gallium. J Cryst Growth 71(3):602–606 8. Kolman DG (2019) A review of recent advances in the understanding of liquid metal embrittlement. Corrosion 75(1):42–57 9. Deng Y-G, Liu J (2009) Corrosion development between liquid gallium and four typical metal substrates used in chip cooling device. Appl Phys A 95(3):907–915 10. Doudrick K et al (2014) Different shades of oxide: from nanoscale wetting mechanisms to contact printing of gallium-based liquid metals. Langmuir 30(23):6867–6877 11. Tang J et al (2017) Gallium-based liquid metal amalgams: transitional-state metallic mixtures (TransM2ixes) with enhanced and tunable electrical, thermal, and mechanical properties. ACS Appl Mater interfaces 9(41):35977–35987 12. Mei S et al (2014) Thermally conductive and highly electrically resistive grease through homogeneously dispersing liquid metal droplets inside methyl silicone oil. J Electron Packaging 136(1):011009 13. Wang X et al (2018) Soft and moldable Mg-Doped liquid metal for conformable skin tumor photothermal therapy. Adv Healthc Mater 7(14):1800318 14. Kong W et al (2019) Oxide-mediated formation of chemically stable tungsten–liquid metal mixtures for enhanced thermal interfaces. Adv Mater 31(44):1904309 15. Chang H et al (2020) Recoverable liquid metal paste with reversible rheological characteristic for electronics printing. ACS Appl Mater Interfaces 12(12):14125–14135 16. Wei S et al (2019) Investigation on enhancing the thermal conductance of gallium-based thermal interface materials using chromium-coated diamond particles. J Mater Sci Mater Electron 30(7):7194–7202 17. Gao Y et al (2017) Investigation on the optimized binary and ternary gallium alloy as thermal interface materials. J Electron Packaging 139(1):011002 18. Uppal A et al (2019) Pressure-activated thermal transport via oxide shell rupture in liquid metal capsule beds. ACS Appl Mater Interfaces 12(2):2625–2633 19. Tutika R et al (2018) Mechanical and functional tradeoffs in multiphase liquid metal, solid particle soft composites. Adv Funct Mater 28(45):1804336 20. Gaitonde A, Nimmagadda A, Marconnet A (2017) Measurement of interfacial thermal conductance in Lithium ion batteries. J Power Sources 343:431–436 21. Marconnet AM et al (2011) Thermal conduction in aligned carbon nanotube–polymer nanocomposites with high packing density. ACS Nano 5(6):4818–4825

Part XXXIX

Phase Transformations and Microstructural Evolution

Dissolution of Carbides in HAZ During NG-TIG Welding for Alloy 617/9%Cr Dissimilar Welded Joint Kai Ding, Guanzhi Wu, Wufeng Dong, and Yulai Gao

Abstract The carbide evolution plays a significant effect on the mechanical property of welded joint. The microstructure in each characteristic zone and the carbide evolution of Alloy 617/9%Cr dissimilar welded joint are analyzed. The austenitic grains with an average size of about 150 μm are found in Alloy 617-BM. The columnar grain can be observed in WM. The microstructure of 9%Cr-BM is determined as lath martensite. The carbides in the base metal (BM) of Alloy 617 are mainly distributed along the grain boundaries and within the grains, while the carbides adjacent to the fusion line appeared an obvious dissolution behavior due to the heat input during narrow-gap tungsten inert gas (NG-TIG) welding process. Thus, the growth and also the coarsening of the carbides can be avoided attributing to the carbide dissolution, favorable to enhance the service performance of Alloy 617/9%Cr dissimilar welded joint. Keywords NG-TIG · Carbide dissolution · Dissimilar welded joint · Heat affected zone

Introduction Advanced ultra-supercritical (A-USC) power plant requires the heat resistance materials to produce a combination of resistance to hot corrosion at elevated temperature, resistance to creep, and stable properties for prolonged application [1]. Alloy 617/617B, a solid solution strengthened and carbide hardened Ni-based superalloy, is regarded as one of the candidate materials for elevated temperature applications [2, 3]. The addition of Co and Mo in Alloy 617 can generate solution strengthening K. Ding · G. Wu · W. Dong · Y. Gao (B) State Key Laboratory of Advanced Special Steel, School of Materials Science and Engineering, Shanghai University, Shanghai 200444, P.R. China e-mail: [email protected] Y. Gao Shanghai Engineering Research Center for Metal Parts Green Remanufacture, Shanghai 200444, P.R. China © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_133

1399

1400

K. Ding et al.

[4], meanwhile the presence of elements Al and Ti is beneficial to the precipitation strengthening through the formation of γ (Ni3 (Al, Ti)) phase [5]. Oxidation and hot corrosion resistance are superior attributing to the addition of elements Cr and Al, which can form a dense oxide layer [6, 7]. In addition, 9–12%Cr martensite steels are regarded as the ideal candidate materials to replace the traditional CrMoV heatresistant steels at the further increased working temperature over 600 °C. Knezevic et al. [8] found that the excellent properties of 9–12%Cr steels could be achieved through apparently minor compositional changes. Dissimilar welded joints are extensively put into applications and generally regarded as a great industrial significance in the field like automobile manufacture [9] and power plant [10–13] attributing to their characteristics to make full use of the advantages for materials with different mechanical properties [14, 15]. However, due to the different composition and physical properties of various base metals (BMs), several issues including inhomogeneous microstructure, macro-segregation, and formation of intermetallic compounds would seriously affect the mechanical properties of the dissimilar welded joints. The heat affected zone (HAZ) is generally the most concerning part across the welded joint, and the HAZ liquation cracking can be found in the welding of Alloy 617 by some researchers. Fink et al. [16] found that liquation cracking can be located predominantly at or near the fusion lines in HAZ in the cold metal transfer (CMT) welded joint of Alloy 617, and the HAZ liquation cracking is connected with the carbide stringers of the base metal microstructure. Ren et al. [17] pointed that one of the features of the liquation cracking is the presence of grain boundary resolidified phase around the liquation cracking. Thus, the carbide evolution plays an important role in the process of welding. However, little information is available for the carbide evolution behavior and its corresponding mechanism in HAZ of Alloy 617 welded joint. The present investigation focuses on the Alloy 617–9%Cr dissimilar welded joint by the narrow-gap tungsten inert gas welding (NG-TIG). The elemental distribution of the carbides, microstructural evolution especially the dissolution behavior of carbides in the region adjacent to the fusion line are studied in details. Basing on the analysis and experimental results, the dissolution mechanism of carbides in Alloy 617-HAZ is discussed in detail to clarify the effect of carbide dissolution on the performance of the Alloy 617–9%Cr dissimilar welded joint.

Experimental Procedures Alloy 617 nickel-based alloy and 9%Cr steel are used as the base metals (BMs) to fabricate the Alloy 617/9%Cr dissimilar welded joint. To obtain considerable high temperature creep rupture strength as well as excellent match with Alloy 617B, the ENiCrCoMo-1 Thermanit 617 is selected as filler material. The composition of the BMs and filler metal is listed in Table 1. The BMs are joint by narrow-gap tungsten inert gas welding (NG-TIG) technique. The welding current and voltage are 250

Dissolution of Carbides in HAZ During NG-TIG Welding …

1401

Table 1 Chemical composition of Alloy 617, 9%Cr steel, and the filler metal (wt. %) Elements

C

Si

Mn

Cr

Mo

Ni

V

Al

S

P

Fe

9% Cr

0.11

0.06

0.45

10.39

1.04

0.73

0.18

0.004

0.001

0.008

Bal

Alloy 617

0.052

0.04

0.01

22.38

9.02

Bal

–

1.03

0.001

0.001

0.33

Filler metal

0.06

0.05

0.05

22.5

8.9

Bal

–

1.3

0.001

0.001

0.32

Fig. 1 Schematic of the microstructure observation for Alloy 617–9%Cr dissimilar welded joint

A and 11.5 V, respectively. The welding speed is 8.0 cm/min. The post weld heat treatment (PWHT) at a temperature of 980 °C for 10 h with subsequent air cooling is carried out in order to relieve the residual stress occurred during welding. The schematic of the welded joint is shown in Fig. 1. The specimen for the microstructure observation containing all the characteristic zones is grinded and polished after being cut from the whole Alloy 617–9%Cr dissimilar welded joint. Owing to the different corrosion resistance between the Alloy 617 and 9%Cr steel, the etchant of HCl + HNO3 + H2 O with the volume proportion of 3:3:5 is applied to reveal the microstructure of 9%Cr steel. The Alloy 617 and the WM are etched with Kalling’s reagent (100 ml HCl + 100 ml C2 H5 OH + 5 g CuCl2 ) to reveal the microstructure and distribution of second phase particles. Optical microscopy (OM, Zeiss Imager A2m) and scanning electron microscopy (SEM, JSM-6700F) are used to analyze the microstructure evolution. The elemental distribution of the carbides is obtained by electro-probe microanalyzer (EPMA, EPMA-8050G).

Results and Discussion The optical images of the microstructure for each characteristic zone are shown in Fig. 2. The overall macrostructure of the Alloy 617–9%Cr dissimilar welded joint is shown in Fig. 2a, and the results show that there existed five characteristic zones containing base metal of 9%Cr steel (9%Cr-BM), heat affected zone of 9%Cr steel (9%Cr-HAZ), weld metal (WM), Alloy 617-HAZ, and Alloy 617-BM. The width of WM is about 13 mm, and columnar grains form owing to the high cooling rate during the TIG welding. In addition, no liquation cracking can be found in Alloy 617-HAZ, implying that NG-TIG is an appropriate welding method to join Alloy 617B and 9%Cr steel. Microhardness distribution along the axial direction of specimens in the

1402

K. Ding et al.

Fig. 2 Microstructure of the characteristic zones for Alloy 617-CrMoV dissimilar welded joint: a the overall macrostructure of the welded joint, b the microstructure details in Alloy 617-BM, c the microstructure details in WM, and d the microstructure details in 9%Cr-BM

whole welded joint is already measured by Vickers hardness tester (MH-5L) in our previous study [11]. The microhardness across the joint exhibited an inhomogeneous distribution with the lowest microhardness of ~230 HV in the heat affected zone (HAZ) of 9%Cr steel yet the highest one of ~360 HV in the weld metal (WM). The microstructure details of Alloy 617-BM, WM, and 9%Cr-BM are shown in Fig. 2b–d, respectively. Figure 2b displays the microstructure of Alloy 617B-BM. The austenitic grains with the average size of about 150 μm are detected, and several annealing twins are apparent in the interior of the grains. A large amount of dispersed globular particles with the size around 2–8 μm can be found, distributing along the grain boundaries and within the grains. In addition, the precipitate in orange color traced by OM, which is generally regarded as Ti(C, N) phase, can also be observed. Figure 2c displays the microstructure of WM, and the columnar grain can also be observed. Figure 2d shows the microstructure of 9%Cr-BM is lath martensite. Figure 3a, b shows the second electron (SE) images of the precipitates in Alloy 617-BM. The elemental distribution results are obtained and displayed in Fig. 3c–f. It can be confirmed that the globular precipitates are Cr-rich carbides and the particles in orange traced by OM are Ti- and N-rich particles, and the particles are mainly distributed along the grain boundaries. Li et al. [18] investigated the microstructure and high temperature fracture toughness of NG-TIG welded Alloy 617B superalloy and found that the Cr-rich particles in Alloy 617B are M23 C6 type carbides and Ti(C, N) for Ti–rich particles. Similar results are found by Xiang et al. [19] and Kang et al. [20]. Xu et al. [21] considered that the presence of these discrete second particles on the grain boundaries in Alloy 617 can prevent the boundary migration and contribute to the creep resistance. Unfortunately, an adverse effect exists between the creep life and material ductility when the precipitates form a continuous network along grain boundaries [22].

Dissolution of Carbides in HAZ During NG-TIG Welding …

1403

Fig. 3 Microstructure of Alloy 617B-BM: a, b the second electron (SE) images of the precipitates in Alloy 617-BM and c–f the elemental mapping results including the elements C, Cr, Ti, and N, respectively

Owing to the high temperature microstructure stability of Alloy 617, the microstructure cannot be clearly distinguished based on the overall macrostructure shown in Fig. 2a. To shed light on the microstructure evolution in the region adjacent to the fusion line of Alloy 617, the corresponding microstructure details are obtained by OM and SEM, which is displayed in Fig. 4. No obvious grain growth and microstructure evolution is found adjacent to the fusion line except for the region with the width around 150 μm in Alloy 617 away from the fusion line according to the carbide variation. The carbides, as shown in Fig. 4a, c, present local variation in the region near the fusion line. The pre-existing carbides in Alloy 617-BM are transformed to an irregular phase with the distance decreased from the fusion line (see Fig. 4d). Figure 5 shows the elemental mapping results of the second phase particle in Alloy 617-BM by EPMA, and it can be found that the particles are mainly rich in C, Cr, and Mo. Figure 6 displays the elemental mapping results of the irregular phase in the region near the fusion line. The irregular phase is rich in C, Cr, and Mo. Distribution of element Cr is displayed in Fig. 6b. It is clear that the element Cr is distributed homogeneously in the matrix but rich in the region around the irregular carbide, indicating the dissolution of the carbides. Figure 7a exhibits the overall microstructure from WM across HAZ to Alloy 617BM, the carbides near the fusion line experienced an obvious evolution. More details about the carbide evolution with different distance to the fusion line are shown in Fig. 7b, and the carbide dissolution can be obviously detected. The schematic of the carbide dissolution during welding is illustrated in Fig. 7c. In our previous study, the macrostructure evolution in the HAZ is largely related to the temperature experienced during welding process [23]. The region closer to the fusion line will undergo higher heating peak temperatures above A3 (namely the transformation temperature from

1404

K. Ding et al.

Fig. 4 Microstructure of Alloy 617 in the region adjacent to the fusion line: a, b the metallographic structure, and c, d the second electron (SE) images of the microstructure

Fig. 5 Elemental distribution of the second phase particles in Alloy 617-BM: a the SE image of the second phase particle, b element Cr, c element Mo, d element C, e element Ti, and f element N

Dissolution of Carbides in HAZ During NG-TIG Welding …

1405

Fig. 6 Elemental distribution of the second phase particles in the region near the fusion line: a the SE image of the second phase particle, b element Cr, c element Mo, d element C, e element Ti, and f element N

ferrite to austenite). As the temperature reaches to that of carbide dissolution in Alloy 617, the diffusion of the alloy elements is strongly enhanced, leading to the carbide dissolution. As a result, the solute atoms of Cr released from carbide diffuse into the adjacent γ matrix, causing the increasing concentration of Cr in the surrounding matrix. With the distance decreased from the fusion line, the extent of dissolution is enhanced, refining the carbides to a certain extent. That is to say, the growth and also the coarsening of the carbides can be avoided attributing to the carbide dissolution, favorite to enhance the service performance of Alloy 617/9%Cr dissimilar welded joint to some extent.

Conclusions Five characteristic zones are detected in the whole Alloy 617–9%Cr dissimilar welded joint. The austenitic grains with an average size of about 150 μm are found in Alloy 617-BM. A large amount of dispersed globular particles with the size around 2–8 μm can be found, distributing along the grain boundaries and within the grains. The columnar grain can be observed in WM. The microstructure of 9%Cr-BM is determined as lath martensite. During TIG welding, the region closer to the fusion line will undergo higher heating peak temperatures above A3 . The carbides of Alloy 617 in the region near the fusion line experienced an obvious dissolution. The solute atoms of Cr and Mo released from carbide diffuse into the adjacent γ matrix, causing the increasing concentration of Cr in surrounding matrix. With the distance decreased from the fusion line, the extent of dissolution is enhanced. The growth and also the coarsening of the carbides can be avoided attributing to the carbide dissolution,

1406

K. Ding et al.

Fig. 7 SE images of the carbides in the region near the fusion line and schematic for the corresponding carbide evolution process: a, b the SE images of the carbides, and c schematic of the carbide dissolution

Dissolution of Carbides in HAZ During NG-TIG Welding …

1407

favorable to enhance the service performance of Alloy 617/9%Cr dissimilar welded joint. Acknowledgements This work is supported by the National Natural Science Foundation of China (Grant no. 52101042) and Open Project of State Key Laboratory of Advanced Special Steel and Shanghai Key Laboratory of Advanced Ferrometallurgy, Shanghai University (Grant no. 19DZ2270200).

References 1. Schlegel S, Hopkins S, Young E, Cile J, Lillo J, Frary M (2009) Precipitate redistribution during creep of alloy 617. Metall Mater Trans A 40(12):2812–2823 2. Wankins WL, Hosier JC, Bassford TH (1974) Microstructure and phase stability of Inconel alloy 617. Metall Mater Trans A 4(12):2579–2590 3. Rahman MS, Priyadarshan G, Raja KS, Nesbitt C, Misra M (2009) Characterization of high temperature deformation behavior of Inconel 617. Mech Mater 41(3):261–270 4. Gariboldi E, Cabibbo M, Spigarelli S, Ripamonti D (2008) Investigation on precipitation phenomena of Ni-22Cr-12Co-9Mo alloy aged and crept at high temperature. Int J Pres Ves Pip 85(1–2):63–71 5. Di Martino SF, Faulkner RG, Hogg SC, Vujic S, Tassa O (2014) Characterisation of microstructure and creep properties of alloy 617 for high-temperature applications. Mater Sci Eng A 619:57–86 6. Speicher M, Kauffmann F, Shim JH, Chandran M (2018) Microstructure evolution in Alloy 617B after a long-term creep and thermal aging at 700 °C. Mater Sci Eng A 711:165–174 7. Wu QY, Song H, Swindeman RW, Shingledecker JP, Vasudevan VK (2008) Microstructure of long-term aged IN617 Ni-base superalloy. Metall Mater Trans A 39(11):2569–2585 8. Knezevic V, Balun J, Sauthoff G, Inden G, Schneider A (2008) Design of martensitic/ferritic heat-resistant steels for application at 650 °C with supporting thermodynamic modelling. Mater Sci Eng A 477(1):334–343 9. Gupta SK, Pandey KN, Kumar R (2015) Comparison of ANN and regression modeling for predicting the responses of friction stir welded dissimilar AA5083-AA6063 aluminum alloys joint. Appl Mech Mater 813–814:415–419 10. Shang YB, Shi HJ, Wang ZX, Zhang GD (2015) In-situ SEM study of short fatigue crack propagation behavior in a dissimilar metal welded joint of nuclear power plant. Mater Des 88:598–609 11. Ding K, Wang P, Liu X, Li XH, Zhao BG, Gao YL (2018) Formation of lamellar carbides in Alloy 617-HAZ and their role in the impact toughness of Alloy 617/9%Cr dissimilar welded joint. J Mater Eng Perform 27(11):6027–6039 12. Ding K, Qiao SF, Liu SP, Zhao BG, Huo X, Li XH, Gao YL (2019) Failure transition mechanism of stress rupture performance of the Inconel 625/9 pct Cr steel dissimilar welded joint. Metall Mater Trans A 50(10):4652–4664 13. Dehaghi EM, Moshayedi H, Sattari-Far I, Arezoodar A (2017) Residual stresses due to cladding, buttering and dissimilar welding of the main feed water nozzle in a power plant reactor. Int J Pres Ves Pip 152:56–64 14. Guo Q, Lu FG, Liu X, Yang RJ, Cui HC, Gao YL (2015) Correlation of microstructure and fracture toughness of advanced 9Cr/CrMoV dissimilarly welded joint. Mater Sci Eng A 638:240–250 15. Rahmani M, Eghlimi A, Shamanian M (2014) Evaluation of microstructure and mechanical properties in dissimilar austenitic/super duplex stainless steel joint. J Mater Eng Perform 23(10):3746–3753

1408

K. Ding et al.

16. Fink C, Zinke M (2013) Welding of nickel-based alloy 617 using modified dip arc processes. Weld World 47(3):323–333 17. Ren WJ, Lu FG, Yang RJ, Liu X, Li ZG (2015) Liquation cracking in fiber laser welded joints of inconel 617. J Mater Process Tech 226:214–220 18. Li XG, Li KJ, Li SL, Wu Y, Cai ZP, Pan JL (2020) Microstructure and high temperature fracture toughness of NG-TIG welded Inconel 617B superalloy. J Mater Sci Technol 39:173–182 19. Xiang XM, Yao ZH, Dong JX, Sun LG (2019) Dissolution behavior of intragranular M23 C6 carbide in 617B Ni-base superalloy during long-term aging. J Alloys Compd 787:216–228 20. Kang SH, Lee SJ, Suh JY, Kim HJ, Lee YK (2019) Self-healing behavior of Inconel 617B superalloy. J Alloys Compd 805:1217–1223 21. Xu HL, Liu W, Lu FG, Wang P, Ding YM (2017) Evolution of carbides and its characterization in HAZ during NG-TIG welding of Alloy 617B. Mater Charact 130:270–277 22. Tan L, Ren X, Sridharan K, Allen TR (2008) Corrosion behavior of Ni-base alloys for advanced high temperature water-cooled nuclear plants. Corros Sci 40(11):3056–3062 23. Liu W, Liu X, Lu FG, Tang XH, Cui HC, Gao YL (2015) Creep behavior and microstructure evaluation of welded joint in dissimilar modified 9Cr-1Mo steels. Mater Sci Eng A 644:337–346

Effect of Changes in Phase and Grain Interface on Physical Properties During Aging of Ultra-Deformation Cu-Ag Alloy Wires Zhang Yuan-wang, Wang Shu-sen, and Yao Da-wei

Abstract In order to obtain high strength and high conductivity and simultaneous keeping enough elongation copper alloy wires, the physical properties of ultradeformation Cu-Ag alloy wires were studied versus aging parameters and interface changes. The experimental results show that the phase interface and grain boundary of ultra-deformation Cu-Ag alloy wire has massive amorphous structure and subgrain structure. It can be mentioned that the physical properties of Cu-Ag alloy wires were affected dominantly by these substructures. The present comparative study suggests that there is an obviously increasing on elongation for ultra-deformation Cu-Ag alloy wires after recrystallization, and the strength is compatible with high conductivity in Cu alloys due to the solute atoms stable existence in the matrix. Keywords Low-temperature aging · Cu-Ag alloy wires · Orientation · Grain boundary · Recrystallization

Introduction Severe cold drawing (SCD), one of the severe plastic deformations (SPD) which could obtain grain refinement in polycrystalline metals and alloys, is widely applied in the preparation of high strength and high conductivity of cooper alloy wires, such as Cu-Ni, Cu-Cr-Zr, Cu-Ag, and Cu-Ti [1, 2]. The application of high strength and high conductivity Cu alloys using variously in the medical diagnostic and electrical equipment industry requires a compromise of mechanical properties (especially elongation) and electrical resistivity. However, the SCD materials produced by the different SPD methods exhibit very high strengths but the tensile ductility is often

Z. Yuan-wang · W. Shu-sen · Y. Da-wei (B) Shanghai Electric Cable Research Institute Co., Ltd, Shanghai 200093, China e-mail: [email protected] The State Key Laboratory Of Special Cable Technology, Shanghai 200093, China © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_134

1409

1410

Z. Yuan-wang et al.

very low (the uniform elongation is //ND fiber textures were observed simultaneous [11]. The relative proportion of these two components depends on the true strain, and the < 001> //ND fiber textures increase as the severe cold drawing. The change depends on the stacking fault energy (SFE) which is the main mechanism of metal deformation and it means for copper //ND fiber is the most stable deformation texture. However, silver is the low SFE crystal which tended //ND after severe plastic deformation [12]. So, it is worth studying that the change of crystallographic orientation of the low-silver copper alloy during severe drawing and low-temperature aging. The solid solubility of silver in copper matrix is only 0.1wt% at room temperature, for which the alloy formed by copper and silver has excellent high strength and high conductivity. Therefore, the copper-silver series alloy is the most promising one in the development of high strength and high conductivity alloy materials [13]. The aim of this study is to investigate and study the recrystallization mechanisms of SCD Cu-0.7Ag wires during LTA. Low-temperature aging behaviors of recrystallization texture evolution characterized by transmission electron microscopy (TEM) and electron backscatter diffraction (EBSD) and the subsequent mechanical and electrical properties were investigated to prepare high strength and high conductivity and high elongation Cu-Ag alloy.

Experimental Procedures The investigated SCD Cu-0.7Ag alloy wires were produced on the state key laboratory at Shanghai, China. First, chemical composition of copper and silver was melt in vacuum. After melting, the bar was casting with the diameter 30 mm, and then

Effect of Changes in Phase and Grain Interface …

1411

the bar was solid solution treated at 760 °C for 4 h and then water-quenched. After that, the bar was aged at 400 °C for 4 h and then turning into 27.0 mm in diameter to remove the surface defects. Finally, the bar was rolled and severe cold drawn into 0.089 mm with reduction no more than 10% each pass die in diameter. To analyse the SCD and static recovery mechanisms, TEM observations were carried out after drawing and aging. For each parameter specimen, two to three thin sections have been characterized, and the results are very reproducible. Thin foils were cut in direction along the drawing axis. The foils were mechanically thinned down to a thickness of 30 µm and then pasted on copper grids in 3 mm diameter discs and then ion-beam milled using Gatan PIPS II 695. TEM analyses were carried out with a JEOL 2000 EX electron microscope operating at 200 kV. Samples were treated by Triple Ion-Beam Cutter, and a field emission gun scanning electron microscope (SEM, MIRA3) equipped with an electron backscatter diffraction (EBSD, AZtec Nordlys Max3) system (HKL Channel 5 System) was used to examine the microstructure and texture evolutions. The recrystallization kinetics were followed by tensile strength measurements of the wires and taken the average of six samples per parameter.

Results and Discussion Evolution of the Microstructure In order to describe the crystallographic texture evolution, the standard convention with crystallographic directions parallel to the wire axis has been used. EBSD images and corresponding pole and inverse pole figures of Cu-0.7wt.%Ag 0.089 mm wires with different states were shown in Figs. 1, 2 and 3. The analysis of the initial microstructure of SCD wire, measured by EBSD in the section along the wire, showed fiber grains in Fig. 1a. After severe plastic deformation, there are more noise points in EBSD image analysis results shown in Fig. 1a. The red position of the pole and inverse polar maps indicates the orientation of the of the {111} face. This is a preferred orientation that is gradually formed during the drawn process. The analysis of the microstructure of SCD wire with 1h LTA, measured by EBSD in longitudinal section of the wire, showed fiber and approximate equiaxed grains in Fig. 2(a). The equiaxed grains appeared in the fibrous grains boundary, some of them were even 5 times bigger than those around. This indicates that after a 1 h LTA, the Cu-Ag alloy wire has begun to recrystallization. Although the area of the equiaxed grains is already close to that of the fibrous grains, the polar map and inverse polar show that its orientation is still . The analysis of the microstructure of SCD wire with 8 h LTA, measured by EBSD in the section along the wire, showed approximate equiaxed grains in Fig. 3, the fibrous grains disappeared. This indicates that after a 8 h LTA, the Cu-Ag alloy

1412

Z. Yuan-wang et al.

Fig. 1 Representative EBSD images a corresponding pole, b inverse pole, and c figures of SCD wires

Fig. 2 Representative EBSD images a, corresponding pole b and inverse pole c figures of LTA wires in 1 h

Effect of Changes in Phase and Grain Interface …

1413

Fig. 3 Representative EBSD images a corresponding pole, b inverse pole, and c figures of LTA wires for 8 h

wire has been completely recrystallized. The polar map and inverse polar shows that its orientation has been changed from of the {111} face to < 001> of the {100} face. The evolution in grain shape related to the SCD and LTA wires is definitely accompanied with the time of LTA process. With the increasing of LTA time to the initial SCD wire, the equiaxed grains appeared and the fibrous grains disappeared. With further increasing LTA time, the fibrous grains disappear and are replaced by equiaxed grains that will grow up as the fibrous grains disappear. Due to the large deformation energy (stacking fault energy, SFE), the mutation of the crystal orientation occurs during LTA, which is unstable for the crystal structure is unstable. Figure 4 showed the representative TEM (a, b) and HRTEM (c) images image of LTA wires for 1 h. In the subsequent LTA process, the unstable for the crystal

Fig. 4 Representative TEM (a, b) and HRTEM (c) images image of LTA wires for 1 h

1414

Z. Yuan-wang et al.

structure gradually replaced by stable structures, until it completely disappeared. In this process, the LTA reduces the time of the recrystalstallization process, which controlling the microstructure of the alloy wire accurately and flexibly, and which provides a reliable method for accurately controlling the physical properties of the wires.

Evolution of the Electrical and Mechanical Properties Accompanying with the evolution in microstructure induced by the LTA process, there are significant changes in both the electrical and mechanical properties. Figure 5 shows the electrical conductivity of Cu-0.7 wt% Ag alloy in SCD and LTA states. The electrical conductivity of sample in SCD state was 92%IACS and increased to 100.2%IACS after LTA for 8 h. The result demonstrated that after 8 hof LTA, the recrystallization process has been completed, the same as EBSD results. In order to study the LTA process, tensile tests were also carried out for the Cu-0.7 wt% Ag alloy in different states, with the result shown in Fig. 6. The tensile strength and elongation of SCD wire were 624 MPa and 1.2%, respectively. The tensile strength of the wires increased monotonically with the deformation ratio while the electrical conductivity present a contrary trend. Tensile strength shows an decrease while elongation decreased during LTA process. The commonly observed reduction in strength is mostly related to a drop in the dislocation density and the grain coarsening motivated by LTA. Generally, the reduction in strength is accompanied with 104

Electrical conductivity, /%IACS

102 100 98 96 94 92 90 88 86

SCD wire

LTA LTA 1h 2hrs

LTA 4hrs Different states

Fig. 5 Electrical conductivity of Cu-0.7Ag wires in different states

LTA 8hrs

Effect of Changes in Phase and Grain Interface …

1415

800

40

Tensile Strength Elongation

700

25

500

20 400 15 300

Elongation, %

30

600

Tensile Strength, MPa

35

10

200

5

100

0 -5

0

SCD wire

LTA LTA 2hrs 1h

LTA 4hrs

LTA 8hrs

Different states

Fig. 6 Tensile strength and elongation of Cu-0.7Ag wires in different states

an increase in electrical conductivity, showed in Fig. 5, and vice versa; furthermore, the elongation rate is related to the deformation rate and the annealing state [14].

Conclusions In this work, the change of crystallographic orientation of the low-silver copper alloy during LTA process has been discussed. The changes in the electrical and mechanical properties during the LTA process are then discussed, and the results show that the decrease in strength and the increase in electrical conductivity and elongation during the process can be controlled. This is because these physical properties change relatively slowly during the LTA process. The present study suggests that there is an obviously increasing on elongation for SCD Cu-Ag alloy wires after LTA process, and the strength is compatible with high conductivity in Cu-Ag alloys due to the solute atoms stable existence in the matrix. Acknowledgements The authors thank the National Key R&D Program of China: Science and Technology for China’s Economy 2020 (No. 2020YFB0408000ZL).

References 1. Raju KS, Sarma VS, Kauffmann A et al (2013) High strength and ductile ultrafine-grained Cu– Ag alloy through bimodal grain size, dislocation density and solute distribution. Acta Mater 61(1):228–238

1416

Z. Yuan-wang et al.

2. Satoshi S, Yasuyuki K, Takayuki T et al (2018) High strength and high electrical conductivity Cu-Ti alloy wires fabricated by aging and severe drawing. Metall Mater Trans A 49(10):4956– 4965 3. Blank E , Kaspar M , Rappaz M (1985) Work hardening due to internal stresses in dendritic microstructures—science direct. Strength Met Alloy (ICSMA 7) 87–92 4. Jen˝o G (2013) Defect structure in ultra ne-grained silver and a copper-silver alloy processed by severe plastic deformation 5. Jakani S, Baudin T, Novion CHD et al (2007) Effect of impurities on the recrystallization texture in commercially pure copper-ETP wires. Mater Sci Eng A 456(1–2):261–269 6. Jakani S, Baudin T, Benyoucef M et al (2012) Impurities effects on the mechanisms of recrystallization of wiredrawn copper. Seg Tech Program Expanded Abs 4(22):4609 7. Ateba BY , Helbert AL, Brisset F et al (2015) Effect of microalloying elements on the Cube texture formation of Fe48%Ni alloy tapes. IOP Conf 82:012036 8. Saha J , Bhattacharjee PP (2021) Influences of thermomechanical processing by severe cold and warm rolling on the microstructure, texture, and mechanical properties of an equiatomic CoCrNi medium-entropy Alloy. J Mater Eng Perform (5) 9. Pornrat S, Sumate T, Rommanee S, Sumolaya K, Kerr WL (2007) Changes in the ultrastructure and texture of prawn muscle (Macrobrachuim rosenbergii) during cold storage. Food Sci Technol Zurich 10. Chen J, Ma X, Yan W et al (2014) Effect of transverse grain boundary on microstructure, texture and mechanical properties of drawn copper wires. J Mater Sci Technol 30(2):184–191 11. Wu YZ, Ma YQ, Gao HT et al (2006) Microstructure and mechanical properties of cladding copper-steel wire by drawing at room temperature 12. Loretto MH, Clarebrough LM, Segall RL (1964) The stacking-fault energy of silver. Phil Mag 10(106):731–732 13. Ruff AW, Ives LK (1973) On the temperature dependence of stacking fault energy in cubic and hexagonal silver-tin alloys. Phys Status Solidi A 16(1):133–149 14. Kawecki A, Knych T, Sieja-Smaga E et al (2012) Fabrication, properties and microstructures of high strength and high conductivity copper-silver Wires/Otrzymywanie Oraz Wasnoci I Mikrostruktura Wysokowytrzymaych I Wysoko Przewodzcych Drutów Ze Stopów Cu-Ag. Arch Metall Mater 57(4)

Effect of Electron Spin Fluctuation on the Magnetism and Elastic Properties of the Slab Matrix Phase Songyuan Ai, Chenxi Yang, Mujun Long, Haohao Zhang, Dengfu Chen, and Huamei Duan

Abstract The effect of electron spin fluctuation on the intrinsic properties of Fe and its alloys cannot be ignored compared to the changes driven merely by volume expansion. Based on the electron spin fluctuation density distribution calculation, the evolution of magnetism and elastic properties on paramagnetic (PM) Fe has been studied. The electron spin fluctuation density distribution is established using Boltzmann statistics that accurately represents the spin fluctuation state as a function of temperature. The results show that compared with α-Fe, electron spin fluctuations have a stronger influence on local magnetic moment of γ-Fe. It is a necessary treatment for the electron spin fluctuation to consider the the influence of the direction and intensity in three dimensions, which omitting the Jacobian weight significantly increases the elastic modulus C’ (by ~16.5 GPa) and C 44 (by ~12.8 GPa) of γ-Fe. As for the γ-Fe, the evolution rates of the elastic modulus C 44 and C’ reach −3.55 and −1.48 × 10–2 GPa K−1 , respectively, while the elastic modulus C’ of α-Fe is relatively stable. S. Ai · C. Yang · M. Long (B) · H. Zhang · D. Chen · H. Duan Laboratory of Metallurgy and Materials, College of Materials Science and Engineering, Chongqing University, Chongqing 400030, P. R. China e-mail: [email protected] S. Ai e-mail: [email protected] C. Yang e-mail: [email protected] H. Zhang e-mail: [email protected] D. Chen e-mail: [email protected] H. Duan e-mail: [email protected] Chongqing Key Laboratory of Vanadium-Titanium Metallurgy and New Materials, Chongqing University, Chongqing 400044, PR China © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_135

1417

1418

S. Ai et al.

Keywords Spin fluctuation · Density distributions · Magnetic moment · Elastic properties · Evolution rate

Introduction Steel materials are widely used in many branches of industry due to the excellent mechanical properties. Current, steel products are almost produced by continuous casting, and the quality directly depends on the microstructure and ultimately on the properties of the matrix phase. The matrix phase of the slab has undergone a series of complex structural and magnetic changes during the continuous casting process [1, 2]. Compared with the crystal structure, the change of magnetism which shows a high sensitivity to alloying and temperature has a non-negligible effect on the thermodynamic and mechanical properties of Fe and its alloys [3]. Actually, the fluctuation of the direction and density of the electron spin motion at high temperature is the essence of the thermal-magnetic excitation of the material, which the spin magnetic moment of the electron largely determines the magnetism of the material [4]. Therefore, clarifying the influence of electron spin fluctuations on the properties of the slab matrix phase at high temperatures has theoretical guidance for the accurate calculation of the inherent properties of Fe and its alloys. The magnetic and elastic properties of the matrix phase have an important influence on the quality of the final material. Due to the difficulty of accurately describing the magnetic state, some studies [5–8] have reported the magnetic and elastic properties of Fe and its alloys, which ignoring the influence of electron spin fluctuations on the material properties. However, due to the effect of thermal activation energy, the position of electrons with different spin states in the energy band changes rapidly, causing the local magnetic moment μ to fluctuate around the statistical average at high temperatures. As an important input parameter for the calculation of the intrinsic properties of the slab matrix phase, the accurate average magnetic moment is of great significance to the final calculation results. Actually, the influence of electron spin fluctuation on the properties of Fe, Co, Ni, and ferromagnetic hexagonal Co has been confirmed [9–11]. The electron spin fluctuation distributions at certain temperatures are studied for PM bcc Fe and fcc Ni by Ruban et al. [12], while massive ab initio calculations are required to map out the parameters in this Hamiltonian formalism. Therefore, in order to accurately calculate the intrinsic properties of the slab matrix phase at high temperatures, a simple and effective method of electron spin fluctuation distribution is very critical. In the present paper, the electron spin fluctuation distributions at various temperatures are described efficiently within the classical statistical thermodynamics via the Boltzmann distribution. Moreover, the average local magnetic moment is introduced in a quadratic form, which is shown to accurately represent the spin fluctuation distribution as a function of temperature to reduce the computational load. Then, the influence of thermal spin fluctuations on the elastic modulus of both PM α-Fe and γ-Fe is introduced using a series of constrained disordered local magnetic moment

Effect of Electron Spin Fluctuation on the Magnetism …

1419

calculations. Due to its simplicity and efficiency, the electron spin fluctuation distributions method can be easily extended to multi-component alloys containing multiple magnetic substances.

Model and Method Electron Spin Fluctuation Model for the DLM State The spin density Pi is used to represent the probability of the occurrence of any local magnetic moment μi , in order to accurately and efficiently describe the spin fluctuation distribution of the electrons in the slab matrix phase at high temperatures. Therefore, the continuous probability distribution of the random variable μ can be replaced by P1 , P2 , …, Pi , …, Pn , and at each temperature: n 

Pi = 1

(1)

i=1

The PM state of matrix phase is described by the disordered local magnetic moment (DLM) approach [13] in combination with the coherent-potential approximation (CPA) [14, 15]: ↑μ

↓μ

↑μ

↓μ

↑μ

↓μ

1 1 n n Fe P1 /2 · · · Fe Pi /2i Fe Pi /2i · · · Fe Pn /2 Fe Pn /2 Fe P1 /2

(2)

There are the same randomly distributed spin density up (↑) and down (↓) as the different local magnetic moments μi , and the system has no net magnetic moment. Since the energy of the system is basically determined by the quadratic (μ2 ) of the local magnetic moment, a quadratic form of the average local magnetic moment μ Pi is constructed to describe the electron spin fluctuation distribution, in order to simplify the calculation and the application of the model in the multi-component alloy system. Thus, the multi-element alloy model can be reduced to the following binary alloy model: ↑μ P

↓μ P

Fe0.5 i Fe0.5

i

(3)

The Jacobian weight as originally introduced in Ref. [16], which all states of spin fluctuations are considered equivalent, has been used to describe the spin density in the three-dimensional space. Thus, the spin density Pi involves the Jacobian weight μ2 is expressed as x i J . The spin fluctuation distribution at each temperature is formulated using Boltzmann’s factor:

1420

S. Ai et al.

xiJ

1 = J Z



μi μB

2



−E i ex p kB T

 (4)

The Z J is the partition function including the Jacobian weight μ2 : ZJ =

 n   μi 2 i=1

μB

 ex p

−E i kB T

 (5)

where the μB is the Bohr magneton, and k B is the Boltzmann constant. Here, ΔE i = E i - E 0 is the changes of system energy. E i represents the energy of embedding an atom with a local magnetic moment of μi into a medium with mean magnetic moment , and E 0 is the internal energy of the reference state. This energy is directly accessible from a series of multi-component DLM calculations. In this calculations, an atomic moment scale ranging from 0 to 3.0 μB with an interval of 0.5 μB (i.e., n = 7) is adopted to map E i . On the basis of a series of fixed magnetic moment calculations, the spin density distribution of the matrix phase at different temperatures can be obtained by Eqs. (4) and (5). The mean magnetic moment μxiJ is proposed in the quadratic form of the atomic moment as μxiJ

  n  = x J μ2 i

i

(6)

i=1

For comparison, the case that excluding the Jacobian weight is also considered. The corresponding probabilities of spin density x i and mean moment μxi are shown as follows:   −E i 1 (7) xi = · ex p Z kB T   n  −E i Z= (8) ex p kB T i=1   n  μxi =  xi μi2 (9) i=1

It is worth noting that the model corresponding to omitting the Jacobian weight considers the longitudinal fluctuations in each direction of the moment separately, and thus, the statistical weight of the fluctuation depends on the energy change ΔE i instead of the magnitude of the moment.

Effect of Electron Spin Fluctuation on the Magnetism …

1421

Ab Initio Computational Methods Based on the accurate local magnetic moment of the system, the exact muffintin orbitals (EMTO) method [15], which combined with CPA, is used to calculate the magnetism and elastic properties of the slab matrix phase. The total energy is calculated applying the full charge-density technique employing the quasinon-uniform approximation, which turned out to be superior for Fe as compared to more standard approximations. For the exchange–correlation functional, the Perdew-Burke-Ernzerhof (PBE) [17] generalized-gradient approximation is used. According to Hooke’s law, the cubic crystal system has three independent elastic constants C 11 , C 12 , and C 44 , which is related to the tetragonal shear elastic constants C’ = (C 11 -C 12 )/2. Based on the volume-conserving orthorhombic (δ 0 ) and monoclinic deformations (δ m ) are applied on the conventional cubic cell, two cubic shear modulus C’ and C 44 are obtained through the relationship between the total energy and the amount of deformation [15]. For the tetragonal shear modulus C’, the following orthorhombic deformation is used: ⎛

⎞ 0 1+δ0 0 ⎜ 0 1−δ 0 ⎟ 0 ⎝ ⎠ 1 0 0 1−δ 2

(10)

0

which leads to the energy change: E(δ0 ) = 2V C  δ02 + O(δ04 ) The shear modulus C 44 is determined from a monoclinic distortion: ⎞ ⎛ 1 δm 0 ⎜δ 1 0 ⎟ ⎠ ⎝ m 1 0 0 1−δ 2

(11)

(12)

m

yielding: E(δm ) = 2V C44 δm2 + O(δm4 )

(13)

where the δ is the strain parameter, and six orthorhombic and monoclinic distortions (with δ = 0, 0.01, 0.02, 0.03, 0.04, 0.05) are considered in the calculation. The O(δ 4 ) is the error term, which can be ignored. The results in this paper are obtained under the condition of a fixed magnetic moment, ignoring the influence of the small deformation process on the local magnetic moment of the system.

1422

S. Ai et al.

Results and Discussion Volume of PM Fe Versus Temperature The thermal lattice expansion of crystals is an important physical phenomenon emerging from lattice anharmonicity. The theoretical volumes (represented here by the Wigner–Seitz radius wtheor ) at various temperatures are obtained by rescaling the calculated equilibrium Wigner–Seitz radius obtained for PM Fe (corresponding to static conditions) using the experimental linear thermal expansion coefficient. Figure 1 shows the variation of Fe energy and local magnetic moment with crystal volume calculated in the static conditions. The local magnetic moment of α-Fe increases linearly with the increase in volume, and the local magnetic moment remains at a high level (>1.8 μB ) as the Fig. 1a. Moreover, the energy of the system changes with the volume showing a “parabolic” evolution law. Then, the Morse [15] fitting is performed on the energy curve of α-Fe, and the equilibrium Wigner– Seitz radius of PM α-Fe is determined to be 2.6513 Bohr. However, there are two energy troughs in the energy curve of γ-Fe (as shown in Fig. 1b), which correspond to two magnetic states. When the Wigner–Seitz radius is less than 2.57 Bohr, the local magnetic moment is basically 0, that is, non-magnetic; while the Wigner–Seitz radius exceeds 2.57 Bohr, the local magnetic moment increases rapidly, showing a PM state. In comparison, the energy valley in the PM state has lower energy of γ-Fe. Therefore, the energies in the range higher than 2.57 Bohr are fitted, and the equilibrium Wigner–Seitz radius of PM γ-Fe is 2.6090 Bohr. Thus, the equilibrium crystal volumes of α-Fe and γ-Fe at different temperatures are obtained (as shown in Table 1), which combined with the experimental linear thermal expansion coefficient of 15.02 and 23.56 × 6 K−1 for PM α-Fe and γ-Fe, respectively [18].

Fig. 1 Dependence of the energy and local magnetic moment on the Wigner–Seitz radius for PM Fe: a α-Fe; b γ-Fe

Effect of Electron Spin Fluctuation on the Magnetism … Table 1 Relationship between the theoretical Wigner–Seitz radii wtheor of PM Fe and temperature

System α-Fe

γ-Fe

1423

T/K

wtheor /Bohr

0

2.6513

1000

2.6911

1100

2.6951

1200

2.6991

0

2.6090

1200

2.6828

1300

2.6889

1400

2.6951

1500

2.7013

1600

2.7074

Electron Spin Density and Local Magnetic Moments of PM Fe Accurate average magnetic moment, which is one of the important input parameters, is of great significance to the calculation results of elastic modulus. Boltzmann’s formula is a simple and efficient method to obtain the distribution of electron spin fluctuations without performing time-consuming calculations of the parameters of the magnetic Hamiltonian. The input is the energy versus configuration, which is accessible from constrained DLM calculations. The energies ΔE of α-Fe and γ-Fe at different temperatures are calculated by fixing the magnetic moment μ, as shown in Fig. 2. In the spirit of Landau theory, the energy has been expressed as ΔE = aμ2 Fig. 2 Internal energy of PM α-Fe and γ-Fe as a function of local magnetic moment for various temperatures

1424

S. Ai et al.

+ bμ4 + cμ6 + dμ8 , where a, b, and c are the usual Landau coefficients [9]. The fitting results are shown in Table 2. The result shows that, compared with α-Fe, the energy curve of γ-Fe is shallower overall, and the absolute value of the coefficient a is significantly higher, that is, γ-Fe is easier to complete magnetic excitation at high temperature. Moreover, the absolute value of the coefficient a increases and the internal energy curves of α-Fe and γ-Fe gradually deepen with the temperature increasing. This means that the intensity of magnetic excitation decreases with the temperature increasing under the high-temperature PM state. For comparison, the electron spin density has been constructed by Eqs. (4) and (7) according to the energy curves. The variation of the energy and spin density of α-Fe and γ-Fe with the local magnetic moment at two typical temperatures (1100 K and 1400 K) are shown in Fig. 3. The purple and red transparent areas in the figure indicate the continuous distributions of spin density with (x i J ) and without (x i ) involving the Table 2 Fitting results of PM α-Fe and γ-Fe internal energy curve System

T/K

a

b

c

d

R2

α-Fe

1000

−8.77900

0.81070

−0.00237

0.00517

0.99999

1100

−8.83690

0.80493

−0.00270

0.00514

0.99998

γ-Fe

1200

−8.89050

0.79746

−0.00273

0.00510

0.99999

1200

−3.16039

0.44028

−0.01975

0.00730

0.99999

1300

−3.31232

0.43456

−0.01974

0.00721

0.99999

1400

−3.46081

0.42977

−0.01985

0.00712

0.99999

1500

−3.57875

0.42429

−0.02031

0.00700

0.99989

1600

−3.75060

0.41915

−0.01985

0.00695

0.99999

Fig. 3 Internal energy and corresponding spin density distributions of PM Fe as a function of the local magnetic moment: a α-Fe, 1100 K; b γ-Fe, 1400 K

Effect of Electron Spin Fluctuation on the Magnetism …

1425

Jacobian weight, respectively. The electron spin density distributions x i J and x i are asymmetrical, which the peak appears near the minimum energy. The x i has a wide distribution range and tends to be distributed in low magnetic moments, especially in PM γ-Fe. However, the distribution range of the x i J is relatively narrow, and the excitation probability of a high magnetic moment is greater under the same energy condition. At high temperatures, the evolution of the average magnetic moment of the slab matrix phase with temperature is closely related to the electron spin density. The relationship between the average magnetic moment of the system with the temperature under the two spin density distributions is shown in Fig. 4 according to the Eqs. (6) and (9). The red and black dotted lines in the Fig. 4 indicate the average magnetic moment (μxiJ and μxi ) calculated from the spin density (x i J and x i ). The high-temperature evolution law of the average magnetic moment μxiJ and μxi is similar, while the μxiJ is always higher than μxi . As for the PM α-Fe, the average magnetic moment μxiJ increases from 2.131 μB to 2.159 μB with the temperature rising. And the μxi is basically maintained at about 1.943 μB , which is because the spin density x i is mainly distributed in the low local magnetic moment region. The difference between the average magnetic moments μxiJ and μxi is more obvious in γ-Fe, which the discrepancy of average magnetic moment reaches ~0.47 μB between 1200 ~ 1600 K. Compared with α-Fe, γ-Fe has a smaller local magnetic moment and higher temperature sensitivity, which the temperature coefficients of μxiJ and μxi have reached 3.00 and 2.70 × 10–4 μB K−1 . Obviously, the system magnetic moment μxiJ calculated on the basis of the spin density x i J can better reflect the true magnetic excitation state of the matrix phase at high temperature. The maximum error (near the Curie temperature) between μxiJ and the reported value is only 9.7%, which is mainly caused by the transition between the short-range ordered PM structure and the long-range ordered ferromagnetic structure near the Curie temperature. Fig. 4 Variation of average magnetic moment with temperature. The values of present results are plotted as symbols connected with lines, and the available data from Refs. [5, 19] are represented as open symbols

1426

S. Ai et al.

Elastic Modulus of PM Fe Under the Spin Fluctuation The elastic properties are a measure of the resistance against stretching of the lattice and related to the second derivative of the interatomic potential [20]. Two typical cubic shear elastic modulus C’ and C 44 describe the ability of the crystal to resist orthorhombic and monoclinic deformation, respectively. Figure 5 shows the elastic modulus of PM Fe under the spin fluctuation when two average magnetic moments μxiJ and μxi are used as the basic input parameters. For comparison, the results of some experimental tests and calculations are also plotted in the figure. The two elastic modulus show softening behavior characteristics with increasing temperature, and the elastic parameters of γ-Fe are always higher than that of α-Fe, which is consistent with the results of the literature. The elastic modulus of PM Fe which considering the influence involving the Jacobian weight is basically consistent with the reported results. When ignoring the influence of the Jacobian weight, the C’ of α-Fe is underestimated by 6.8 ~ 7.6 GPa (43.6 ~ 48.1%), while the C’ and C 44 of γFe are overestimated by ~16.5 GPa (~62%) and ~12.8 GPa (~10.7%), respectively. Therefore, in order to accurately calculate the elastic modulus of the slab matrix phase, the local magnetic moment fluctuation should be treated as a vector, and the fluctuation distribution in the space should be considered. In addition, when considering the spatial variation of the local magnetic moment, the evolution rate of the elastic modulus C 44 of α-Fe and γ-Fe is as high as −1.85 and −3.55 × 10–2 GPa K−1 , while the C’ of γ -Fe changes at a rate of −1.48 × 10–2 GPa K−1 . Fig. 5 Calculated temperature dependence of elastic modulus of PM Fe under the spin fluctuation in comparison with published data of Refs. [6, 21]

Effect of Electron Spin Fluctuation on the Magnetism …

1427

Conclusions The EMTO Ab Initio method has been used to construct the electron spin fluctuation distribution model of the slab matrix phase, and the influence of the electron spin fluctuation at high temperature on the magnetic properties and elastic constants of α-Fe and γ-Fe has been discussed. The result shows: (1)

(2)

(3)

(4)

The electron spin fluctuation model which is formulated using the Boltzmann factor including the Jacobian weight has been constructed and its accuracy and usability have been verified. The electron spin fluctuation has a great influence on the average magnetic moment of γ-Fe, the discrepancy of average magnetic moment μxiJ and μxi reaches ~0.47 μB between 1200 ~ 1600 K. The spin fluctuation of electrons at high temperature has a non-negligible effect on the magnetism and elastic properties of the slab matrix phase. When ignoring the spatial variation of the local magnetic moment, the C’ of α-Fe is underestimated by 6.8 ~ 7.6 GPa (43.6 ~ 48.1%), while the C’ and C 44 of γ-Fe are overestimated by ~16.5 GPa (~62%) and ~12.8 GPa (~10.7%), respectively. The evolution rate of the elastic modulus C 44 of α-Fe and γ-Fe is as high as − 1.85 and −3.55 × 10–2 GPa K−1 , while the C’ of γ -Fe changes at a rate of − 1.48 × 10–2 GPa K−1 .

Acknowledgements This study is financially supported by the Natural Science Foundation of China (NSFC, Project No. 51874059 and U1960113).

References 1. Chen Q, Sundman B (2001) Modeling of thermodynamic properties for bcc, fcc, liquid, and amorphous iron. J Phase Equilib 22(6):631–644 2. Körmann F, Breidi AAH, Dudarev SL, Dupin N, Ghosh G, Hickel T, Korzhavyi P, Muñoz JA, Ohnuma I (2014) Lambda transitions in materials science: recent advances in CALPHAD and first-principles modelling. Phys Status Solidi B 251(1):53–80 3. Ai S, Long M, Zhang S, Chen D, Dong Z, Liu P, Zhang Y, Duan H (2020) Ab initio calculations on elastic properties of if steel matrix phase at high temperature based on lattice expansion theory. Metals 10(2):283 4. Dong ZH (2016) Evolution of steel property in continuous casting and its ab initio investigations for Fe matrix at elevated temperatures. Ph.D. thesis, Chongqing University 5. Ehteshami H, Korzhavyi PA (2017) Thermophysical properties of paramagnetic Fe from first principles. Phys Rev B 96(22):224406 6. Hutchinson B, Malmström M, Lönnqvist J, Bate P, Ehteshami H, Korzhavyi PA (2018) Elasticity and wave velocity in fcc iron (austenite) at elevated temperatures–experimental verification of ab-initio calculations. Ultrasonics 87:44–47 7. Ai S, Long M, Guo W, Liu P, Chen D, Dong Z, Zhang Y, Duan H (2020) Ab initio study on continuous evolution of mechanical properties in phase-transition region of low-carbon steel. Steel Res Int 91(8):2000070

1428

S. Ai et al.

8. Huang S, Li X, Huang H, Holmström E, Vitos L (2018) Mechanical performance of FeCrCoMnAlx high-entropy alloys from first-principle. Mater Chem Phys 210:37–42 9. Dong Z, Li W, Schönecker S, Lu S, Chen D, Vitos L (2015) Thermal spin fluctuation effect on the elastic constants of paramagnetic Fe from first principles. Phys Rev B 92(22):224420 10. Sandratskii LM, Kuvaldin EN (1991) Band-theory approach to the study of local magnetization fluctuations in Fe and Ni. J Phys Condens Matter 3(39):7663 11. Dietermann F, Sandratskii LM, Fähnle M (2012) On the energetics of transversal and longitudinal fluctuations of atomic magnetic moments. J Magn Magn Mater 324(18):2693–2695 12. Ruban AV, Khmelevskyi S, Mohn P, Johansson B (2007) Temperature-induced longitudinal spin fluctuations in Fe and Ni. Phys Rev B 75(5):054402 13. Staunton J, Györffy BL, Pindor AJ, Stocks GM, Winter H (1984) The “disordered local moment” picture of itinerant magnetism at finite temperatures. J Magn Mater 45(1):15–22 14. Gyorffy BL (1972) Coherent-potential approximation for a nonoverlapping-muffin-tinpotential model of random substitutional alloys. Phys Rev B 5(6):2382 15. Vitos L (2007) Computational quantum mechanics for materials engineers: the EMTO method and applications. Springer, London, UK 16. Sandratskii LM (2008) Thermal magnetic properties of the Ni sublattice in half-metallic NiMnSb: a theoretical study based on first-principles calculations. Phys Rev B 78(9):094425 17. Perdew JP, Burke K, Ernzerhof M (1996) Generalized gradient approximation made simple. Phys Rev Lett 77(18):3865 18. Basinski ZS, Hume-Rothery W, Sutton A (1955) The lattice expansion of iron. Proc R Soc London Ser A 229(1179):459–467 19. Pan F, Chico J, Delin A, Bergman A, Bergqvist L (2017) Extended spin model in atomistic simulations of alloys. Phys Rev B 95(18):184432 20. Körmann F, Dick A, Grabowski B, Hickel T, Neugebauer J (2012) Atomic forces at finite magnetic temperatures: phonons in paramagnetic iron. Phys Rev B 85(12):125104 21. Dever DJ (1972) Temperature dependence of the elastic constants in α-iron single crystals: relationship to spin order and diffusion anomalies. J Appl Phys 43(8):3293–3301

In Situ Observation of Coupled Growth Morphologies in Organic Peritectics Under Pure Diffusion Conditions Johann Mogeritsch, Wim Sillekens, and Andreas Ludwig

Abstract Isothermally coupled peritectic solidification for a hyper-peritectic alloy under pure diffusive conditions is presented. For this purpose, directional solidification experiments were performed aboard the International Space Station using a model system for peritectic layer solidification patterns. At constant temperature gradient of 3.0 K/mm and for pulling velocities ranging from 0.12 μm/s to 0.09 μm/s, coupled peritectic growth was observed. At lower pulling velocities, contrary to expectations, only a planar solidification front of the pro-peritectic phase was detected. Two effects were noticed: a significant effect of the pro-peritectic interface on the capability of the peritectic phase to nucleate and, in the further course of the experiments, a dynamic change of the coupled peritectic growth microstructure. Keywords Peritectic solidification · International space station · Microgravity · TRIS-NPG

Introduction A peritectic reaction is the solidification of a liquid  and the transformation of an already existing primary solid phase α into a second solid peritectic phase β( + α → β) at the invariant peritectic temperature T P . The interesting characteristic in peritectic systems is the possibility of forming peritectic phase β above the peritectic temperature. From a thermodynamic point of view, an alloy with a concentration C 0 within the peritectic plateau C α ≤ C 0 ≤ C l should start to solidify with the primary pro-peritectic α phase which transforms to the peritectic β phase when the interface temperature goes below the peritectic temperature. Under conditions where the solidification morphology for one or both phases is planar, it was found J. Mogeritsch (B) · A. Ludwig Department Metallurgy, University of Leoben, Franz-Josef-Strasse 18, 8700 Leoben, Austria e-mail: [email protected] W. Sillekens European Space Agency ESA, ESTEC, Keplerlaan 1, 2201 AZ Noordwijk, Netherlands © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_136

1429

1430

J. Mogeritsch et al.

that within the peritectic plateau two microstructures occur: (i) oscillations of the concentration in liquid close by the solid/liquid (s/l) interface forming an alternating growth sequence in form of bands of both phases parallel to the solidification front, or (ii) simultaneous growth of both phases in the form of fibers or lamellae, called peritectic coupled growth (PCG), similar to a regular eutectic growth front, whereby, peritectic layer structures are highly sensitive to convection ahead of the solidification front. In fact, depending on the material properties and the intensity of convection, peritectic alloys showed in directional solidification experiments a variety of complex microstructures like isothermal peritectic coupled growth (IPCG), cellular peritectic coupled growth, discrete bands, island bands, or oscillatory tree-like structures [1–6]. To explain the experimentally obtained microstructures, several models were published. Trivedi [7] issued a model which explains cyclic nucleation and overgrowth under purely diffusive growth conditions in the hypo-peritectic region (C α ≤ C 0 ≤ C p) . The theory predicts that the pro-peritectic and peritectic phases would grow independently and alternately as planar fronts below (pro-peritectic phase) and above (peritectic phase) the peritectic temperature. Since the model cannot explain the huge variation of experimentally obtained peritectic solidification patterns, and convection is nearly always present on Earth, advanced models were published. Hunzinger’s model [8] was based on the nucleation and constitutional undercooling criterion under the assumption of infinitely high nuclei density and steady-state growth and Lo et al. [9] showed by simulation that bands are formed only for approximately equal volume fraction of the two phases; otherwise, islands bands are formed. Trivedi [10] revised and updated his theory as related to the role of heterogeneous nucleation on the microstructure evolution. The authors analyzed layered peritectic structures by using an organic model system instead of metallic alloys. Organic components with a high-temperature transparent non-faceted phase, called plastic phase, solidify metal-like. The transparency of the phases enables the real-time observation of the solidification dynamics with a light microscope. The binary peritectic system TRIS–NPG [11] was selected for directional solidification experiments with compositions within the peritectic plateau. The studies confirmed the peritectic microstructures found in metal alloys, but also provided insight into the dynamics leading to peritectic layered structures [12–15] as mentioned above. It was also found that in the case where both phases grew separately the competing growth resulted in alternating oscillating interfaces of the two phases [15]. In this article, we describe the experimental findings for directional solidification experiments under purely diffusive conditions. For this, a hyper-peritectic concentration was processed aboard the International Space Station (ISS).

In Situ Observation of Coupled Growth Morphologies …

1431

Experimental Procedure In this section, we describe (i) the equipment used aboard the ISS, (ii) the alloy preparation, and (iii) the selected process conditions. (i)

The directional solidification experiments were carried out aboard the ISS. For this purpose, the TRANSPARENT ALLOYS (TA) instrument was used. This device was specially developed by the European Space Agency (ESA) and QinetiQ Space for microgravity (μg) experiments with organic transparent alloys and was installed in the Microgravity Science Glovebox (MSG). The main part of the TA experiment unit is the Bridgman assembly, a thermal unit with two hot clamps, called hot zone, on one side, and two cold clamps, called cold zone, on the other side, as shown in Fig. 1. The Bridgman furnace had a distinct temperature gradient created by the hot and cold clamps temperature. The hot clamps had temperatures higher than the melting point, and the cold clamps had temperatures lower than the melting point. A 7 mm width gap between the hot and cold clamps acts as an adiabatic zone. A glass cartridge filled with the organic alloy and set in between the clamps was slowly pulled from the hot zone into the cold zone (cartridge travel), whereby the organic material in the cartridge was completely molten in the hot zone and the s/l interface was positioned within the adiabatic gap. The special designed Transparent Alloy Cartridge (TAC) consisted of a quartz window clamped in between 2 stainless steel parts. The TAC had a solidification volume of 100 mm (length) × 6 mm (width) × 1 mm (depth).

Fig. 1 Sketch of the Bridgman furnace assembly within TRANSPARENT ALLOYS device. Source QinetiQ Space, Kruibeke, Belgium [16]

1432

(ii)

(iii)

J. Mogeritsch et al.

Due to the necessary minimum contact areas with the hot and cold clamps, an effective solidification length of 66 mm was available. The optical set-up includes an LED light source illumination system and a CCD camera centered on the adiabatic zone with a Field of View (FOV) of 6.1 × 5.1 mm2 . During the experiments, a set of 3 images were taken with a time step of 3 s at 3 different focal depths, foc1 = 0 mm, directly on the inside of the front glass wall, foc2 = 0.5 mm, in the middle of the cartridge, and foc3 = 0.8 mm, close to the rear glass wall. A set of new images were shot each 30 s. The execution of the tests on board was supervised by the authors. For this purpose, only one single image set was transmitted from the ISS to the authors every hour. The findings presented in this paper are based on these images. The organic component TRIS (tris(hydroxymethyl)aminomethane) was employed in combination with the organic component NPG (Neopentylglycol) as a model system for metal-like peritectic solidification under μg-conditions aboard the ISS. Both organic substances have at low temperature faceted phases and at high-temperature transparent orientationally disordered crystals (short form ODIC), usually called plastic crystal, which forms in an intermediate concentration range a peritectic equilibrium. It should be noted that there are several publications of the TRIS-NPG phase diagram. These are based either on experimental investigations or on thermodynamic calculations, showing an enlarged concentration range with respect to the peritectic plateau. The phase diagram [11] used by the authors to define the process conditions is based on experimental data. Details on the peritectic plateau and the used alloy concentration at x = 0.53 mol fraction NPG are shown in Fig. 2. TRIS and NPG were delivered as powder from Sigma Aldrich [17] with an indicated purity of 99.9 + % and 99%, respectively. Both substances are highly hydroscopic [18], whereby, the water content of the organic substance NPG was reduced by a drying process at 375 K for 24 h. TRIS, sensitive to long time annealing at temperatures above the faceted transformation temperature, and delivered with high-purity, was used without further purification. The alloy manufacturing was performed by the authors, while the TAC filling was done by QinetiQ Space and their on-orbit processing operated from ground by E-USOC Spanish User Support and Operation Centre under the coordination of the ESA. Peritectic layered structures are expected for process conditions where both phases grow planar that occurs at solidification rates below the critical velocity [19]. For this purpose, a temperature gradient of 3.0 K/mm was set in the adiabatic zone and the TAC was drawn with a pulling velocity V p from the hot zone to the cold zone in discrete steps of 0.01 μm in the range of 0.08 ≤ V p ≤ 0.12 μm/s. The 66 mm observable part of the cartridge was virtual evenly divided into 6 starting points that created 6 equal lengths of 11 mm, called in this paper segments. This allowed 6 different solidification experiments to be performed with a fresh segment that has never been molten before. Due to the lack of

In Situ Observation of Coupled Growth Morphologies … 420

Temperature [K]

416

hypo-peritectic region [L]

1433

hyper-peritectic region

[L+ ]

412

TP [L+ ]

408 [ ]

[ ]

[ + ]

404

400 0.40 0.42 0.44 0.46 0.48 0.50 0.52 0.54 0.56 0.58 0.60

TRIS

x

NPG

Fig. 2 Peritectic region of the system TRIS-NPG. The squares show the liquidus (black/white rectangles) temperature, the solidus (white/black rectangles) temperature, and the peritectic temperature T p for the TRIS-rich side published by [11]. The triangles show the corresponding situation for the NPG-rich side. The straight lines are placed by the authors to approach the published measured points. The dashed line shows the extension of the solidus line of the pro-peritectic phase into the metastable region (Color figure online)

convection in the melt under μg-conditions, segments that had been processed already and that are now located in the hot zone and are thus liquid do not interact with the new fresh segment. The experiments started with segment 1 that was at the beginning in the hot zone and were then continued in ascending order. This guaranteed that all the segments already processed were in the hot zone. Each experiment consisted of the same sub-sequences. First, the temperature on all clamps was set on 440 K and the alloy within the TAC was annealed for 2 h. Afterwards, the temperature in the cold zone was set to 403 K and the temperature in the hot zone was set to 503 K. The TAC was kept in rest for one hour to perform thermal equilibrium within the device. Finally, the actually solidification experiment was activated and the TAC was processed between 8 and 30 h.

Results The results presented in this paper are from the author’s μg-experiments aboard the ISS, in particular, the solidification morphologies for a hyper-peritectic organic alloy with x = 0.53 mol fraction NPG under pure diffusion conditions. The results pointed

1434

J. Mogeritsch et al.

out nucleation events that resulted in IPCG for pulling velocities in the range of 0.09 μm/s ≥ V p ≥ 0.11 μm/s. For V p = 0.12 μm/s, a nucleation event was observed, but the initial planar s/l liquid interface transformed into a cellular/dendritic one. In contrast, for a pulling velocity of V p = 0.08 μm/s, the initial planar pro-peritectic phase grew, but no nucleation event was detected. Initially after thermal equilibration, the plastic phase showed a polycrystalline structure. It should be noted that according to the phase diagram, the polycrystalline plastic phase consisted of the pro-peritectic phase and the peritectic phase. Due to the identical optical properties of both phases, it was not possible to determine whether both phases were really present. When the sample was moved, the pro-peritectic phase started to grow until for the pulling velocities mentioned above the nucleation event occurred. In this section, we describe, as examples, (i) the microstructure of a coupled peritectic growth and its main features, and (ii) we exhibit the dynamics of peritectic coupled growth for the pulling velocities mentioned above by presenting partial cuttings of the s/l interface. (i)

In Fig. 3, the main features of the transparent alloy during peritectic coupled growth for V p = 0.11 μm/s after an experiment duration of 17 h are presented. The image was taken where the camera focus was set on foc1 = 0 mm. Hence, all structures close to the inner side of the front glass wall showed sharp lines, whereas all others pattern exhibited a more or less a blurred form. The matrix phase grew planar while the peritectic phase solidified as fibers within the matrix. The fibers were 27 ± 7 μm diameter, and the distance between the

Fig. 3 Peritectic coupled growth of the hyper-peritectic alloy for a pulling velocity of V p = 0.11 μm/s. The β phase grows fibers-like within the α phase matrix (width of the image = 2.55 mm)

In Situ Observation of Coupled Growth Morphologies …

(ii)

1435

rods was λ = 230 ± 40 μm. The diameter of the fibers revealed a slightly oscillating behavior within the plastic phase. It has to be mentioned that the s/l interface was wavy in such a way that the interface was inclined with the lowest position at the rear glass wall/material interface. This can be seen from the shadow effect in the plastic phase at the interface with the melt. A result due to slight temperature deviations in the control of the clamps. The effect proved useful in retrospect because it made the dynamics of growth more observable. Grain boundaries were recognized as curved lines close to and/or in contact with the s/l interface, as well as impurities in the melt which accumulated in front of the solidification front. The s/l interface showed few perturbations with a diameter of up to 55 μm. In some perturbations, the growth of the peritectic phase was visible. Since the pro-peritectic and the peritectic phases were transparent, the simultaneous growth of both phases was only recognizable by the formation of phase boundaries, exhibited as dark lines, within the plastic phases. Figures 4, 5, 6 and 7 show the solidification pattern shortly after a nucleation event of the peritectic phase and the dynamics of coupled growth for four different pulling velocities, namely V p = 0.12, 0.11, 0.10, and 0.09 μm/s. All experiments have in common that the peritectic patches failed to spread along

Fig. 4 Solidification pattern evolution after a nucleation event that happened at a pulling velocity of V p = 0.12 μm/s within the TAC center (foc2 = 0.5 mm). (a) The initial s/l interface grew planar. (b) The β phase grows fiber-like within the α matrix before (c) bath phases grew in a cellular/dendritic manner (width of the image = 2.55 mm)

1436

J. Mogeritsch et al.

the α interface and grew together, a precondition for banded growth; instead, IPCG was formed. The results for V p = 0.12 μm/s are explained in Fig. 4. Figure 4a shows the solidification pattern shortly after a nucleation event has taken place with a focus setting foc2 = 0.5 mm (TAC center). Few perturbations were visible at the s/l interface in the form of small rings. The peritectic phase can be recognized by the pancake-like structures. However, the β phase did not begin to spread out over the α phase to form a band. On the contrary, it can be seen that the number of local regions exhibiting β-phase increases but decreases in size, compare Fig. 4a, b. Additionally, fine blurry lines can be detected within the plastic phase. These blurry lines were the indication for solid–solid interfaces within the plastic phase occurring during coupled growth (Fig. 4b). After nucleation, the peritectic phase grew at the same temperature level as the α phase, or slightly below, whereas, in Fig. 4c, the interface of the peritectic phase grew at a higher temperature level as the surrounding pro-peritectic phase. This indicated the transition from a planar growth to a cellular/dendritic growth. In the upper part of each image (Fig. 4a–c), the s/l interface close to the glass wall can be detected as blurred line since it was not in focus. This blurry interface, which was still planar in Fig. 4a, formed already a cellular solidification structure in Fig. 4b and grew cellular/dendritic in Fig. 4c. The peritectic coupled growth observed for V p = 0.11 μm/s is shown in Fig. 5. Note that the images were brightened to lighten the pattern just after nucleation. As a result, the coupled growth is no longer clearly visible as in Fig. 3. Initially, the peritectic phase appeared in a pancake-like shape, as shown in Fig. 5a. It seems to be that the nucleation density was higher than in the previous experiment, see Figs. 4a and 5a. Initially, the β phase grew in a lamellar-like manner, recognizable by the zig-zag growth pattern at the

Fig. 5 Peritectic coupled growth of the hyper-peritectic alloy for a pulling velocity of V p = 0.11 μm/s. (a) taken just after the nucleation event, (b) after a while the peritectic phase grew lamellae-like, and (c) later even fibers-like within the pro-peritectic phase (width of the image = 2.55 mm)

In Situ Observation of Coupled Growth Morphologies …

1437

s/l interface (Fig. 5c). This zig-zag pattern leads to an overlap of the phase boundaries in the plastic phase and thus prevented the evaluation of the lateral phase growth. As the experiment progressed, the proportion of the β phase decreased, evident by the transition from lamellar growth to a fibrous structure. After 10 h of coupled growth, the β phase disappeared and only the α phase grew planar. The results for a pulling velocity of V p = 0.10 μm/s were similar to the findings of the experiment described previously. The nucleation events were followed by lamellar growth which transformed into a fibers-like growth (Fig. 6a–c). Nonetheless, differences can be detected. The nucleation density and the proportion of the ß phase were higher as for V p = 0.11 μm/s. The coupled growth could be observed for 18 h, when the experiment was aborted. The findings for V p = 0.09 μm/s were isothermal peritectic coupled growth after nucleation of the β phase. The nucleation density and phase fraction were lower than in the previously described experiments. In Fig. 7a, the peritectic phase nucleated at different positions at the s/l interface and spread out in a circle form. The peritectic coupled growth morphology shows, as described in the experiments before, first a lamellar growth structure which transforms in the further course of the solidification experiments into a fiber-like one (Fig. 7b, c). The proportion of the new phase decreased as the experiment progresses until only the planar growth of the α phase was observed. For pulling velocities V p ≤ 0.08 μm/s, no nucleation event was observed and only the initial existing phase grew.

Fig. 6 Peritectic coupled growth of the hyper-peritectic alloy for a pulling velocity of V p = 0.10 μm/s. (a) taken just after the nucleation event, (b) peritectic coupled growth that shows a lamellar zig-zag structure, and (c) transformation of the zig-zag structure into a fibers-like coupled growth (width of the image = 2.55 mm)

1438

J. Mogeritsch et al.

Fig. 7 Peritectic coupled growth of the hyper-peritectic alloy for a pulling velocity of V p = 0.09 μm/s. (a) circular spreading of the peritectic phase after the nucleation event, (b) lamellar growth that transitions into (c) fibrous growth (width of the image = 2.55 mm)

Discussion The results of μg solidification experiments for a hyper-peritectic alloy that showed a nucleation event are presented and details with respect to (i) the temperature range for the nucleation event, (ii) the temperature level of the s/l interface, and (iii) the change in microstructure during coupled growth are now discussed. For each subject, we compared the experimental results with Trivedi’s models [7, 10]. (i)

The onset of the nucleation events was observed at the s/l interface at a temperature of T nuc = 408.5 ± 1.0 K. That the nucleation event took place below the peritectic temperature agrees with expectations. Only when the concentration in the melt before the solidification front reaches the supercooling necessary for the formation of the peritectic phase (here determined to be T u = 2.2 K), the β phase nucleates. However, it should be noted that the published temperature accuracy is in the order of ± 2 K. After nucleation, the peritectic phase grew circular from several nucleation points. Since all nucleated regions showed a minor depression within the center, we assume that the peritectic phase nucleated in perturbations at the planar s/l interface of the α phase. Immediately after the first embryos were formed, the β phase grew radially along the interface between the melt and the α phase. This happened until the β phase almost filled the depression within the matrix, but did not exceed the temperature level of the α-liquid interface. It can be noted that the proportion of the peritectic phase increased when the pulling velocity was reduced. At a pulling velocity of V P = 0.10 μm/s, the largest proportion of peritectic phase was observed. The peritectic phase fraction decreased by a further decrease of the pulling velocity. Finally, no nucleation event was observed for a pulling velocity of V p = 0.08 μm/s. Thus, there seems to be a pulling velocity that creates the optimal wetting conditions for the peritectic phase to nucleate at the pro-peritectic matrix.

In Situ Observation of Coupled Growth Morphologies …

(ii)

(iii)

1439

Trivedi predicted that the developing microstructure depends on nucleation and growth competition between the primary and peritectic phases. Single nucleation at the wall and subsequent spreading of the nucleating phase across the interface result in discrete bands or sub-bands. If nucleation occurs at the s/l interface, particulate bands or peritectic coupled growth can form. In the present case, the nucleation event was governed by the wetting conditions of the peritectic phase on depressions of the primary phase and the nucleation event at the s/l interface led to peritectic coupled growth. Thereby, the μgexperiments confirmed Trivedi’s forecast. Initially, the temperatures of the s/l interface were at the liquidus temperature. Once the TAC was moved from the hot zone to the cold zone, the s/l interface attempted to compensate this movement by planar growth. Simultaneously, an NPG-enriched boundary layer forms in front of the interface and the temperature level of the s/l interface decreases continuously. In theory, the nucleation of the peritectic phase is possible as soon as the NPG-enriched liquid interface layer dropped below the peritectic temperature. In the absence of any nucleation events, a steady state for planar growth will be reached at the solidus temperature of the pro-peritectic phase. It should be noted that at no time and in all experiments, the growth rate of the organic alloy was sufficient to a reach steady-state condition. Here, nucleation occurred below the peritectic temperature and above the solidus temperature. The solidus temperature was determined by the authors by a simple linear extension in the metastable region. In the further course of the experiment, the solidification front did not grow at a constant temperature as expected. Instead, the s/l interface temperature level continuously decreased and was finally below the presumed α solidus temperature. This is in contrast to the fact that the peritectic temperature in a binary diagram is an invariant temperature. Nonetheless, it is true that at the s/l interface the melt and both solid phases were in equilibrium (L → α + β) and that at a constant decreasing temperature level. If the growth of both phases is considered separately from each other, then both phases would try to grow planar at their own solidus line. If both solidus temperatures are in lower temperature ranges, the coupled growth and the simultaneous reduction of the s/l interface temperature could be explained. However, this is a speculative assumption and an answer to this question must be postponed pending further investigation. Regarding the shape of the peritectic phase at the s/l interface, lamellar-like structures formed after the nucleation process. In the further course of the solidification experiments, the lamellae became smaller and transformed into fiber-like structures. This indicates that the amount of peritectic phase was not constant and the fraction of growing peritectic phase decreased as well as the temperature level at the s/l interface. When the interfacial temperature fell below 401 ± 2 K, the peritectic phase, and thus the coupled growth, disappeared. It should be noted that this corresponds to the range of temperature at which, according to the phase diagram (Fig. 2), the concentration line intersects the phase boundary α + β/β.

1440

J. Mogeritsch et al.

Summary and Conclusions Directional solidification experiments were performed under μg-conditions aboard the ISS. The experimental findings presented above can be summarized as follows: • Trivedi predicted peritectic layered structures under pure diffusion conditions. He forecasts that depending on the nucleation conditions band or coupled growth formed. Our directional solidification experiments showed nucleation events at the solid/liquid interface of the pro-peritectic phase. These nucleation events led to isothermal peritectic growth. Thus, the experimental investigations confirmed the theoretically predicted morphology at least qualitatively. • After the nucleation events, the peritectic phase grew lamellar in a zig-zag pattern at first. The proportion of peritectic phase did not remain constant, but decreased. This reduction changed the growth from lamellar to fibrous. The temperature level of the solid/liquid interface dropped constantly to lower temperatures during the coupled growth. When the interface fell below the temperature range of 401 ± 2 K, the peritectic phase completely disappeared and the coupled growth turned into a planar growth of the pro-peritectic phase. • The nucleation event of the peritectic phase occurred in the case where the planar solid/liquid interface of the primary phase shows small perturbations, otherwise not. It looks like the necessary surface roughness of the pro-peritectic phase was not present at lower pulling velocities. • During the coupled growth, the solid/liquid interface does not grow at the invariant peritectic temperature, or with constant undercooling but decreases to lower and lower temperature levels. Apart from speculative assumptions, this finding could not be explained with the available experimental data, and therefore, further investigations are needed. Acknowledgements This work was supported in part by the European Space Agency ESA and in part by the Austrian Space Agency ASA through means of the ESA MAP project METCOMP (AO-1999-114), the hardware was developed by QinetiQ Space, and the μg-experiment operations were carried out by the team at E-USOC.

References 1. Park JS, Trivedi R (1998) Convection-induced novel oscillating microstructure formation in peritectic systems. J Cryst Growth 511–515 2. Yasuda H, Ohnaka I, Tokieda K (1997) In-situ observation of peritectic solidification in Sn-Cd and Fe-C alloys, Solidification processing, University of Sheffield, pp 44–48 3. Boettinger WJ (1974) The structure of directionally solidified two-phase Sn-Cd peritectic alloys. Metall Trans 5:2023–2031 4. Su YQ, Luo LS, Li XZ, Guo JJ, Yang HM, Fu HZ (2006) Well-aligned in situ composites in directionally solidified Fe-Ni peritectic system. Appl Phys Lett 89:2319181–2319183

In Situ Observation of Coupled Growth Morphologies …

1441

5. Luo LS, Su YQ, Guo JJ, Li HZ, Yang HM, Fu HZ (2008) Producing well aligned in situ composites in peritectic systems by directional solidification. Appl Phys Lett 92:0619031– 0619033 6. Dobler S, Lo TS, Plapp M, Karma A, Kurz W (2004) Peritectic coupled growth. Acta Mater 52:2795–2808 7. Trivedi R (1995) Theory of layered structure formation in peritectic system. Metall Mater Trans A 26A:583–590 8. Hunziker O, Vandyoussefi M, Kurz W (1998) Phase and microstructure selection in peritectic alloys close to the limit of constitutional undercooling. Acta Mater 46:6325–6336 9. Lo TS, Karma A, Plapp M (2001) Phase-field modeling of microstructural pattern formation during directional solidification of peritectic alloys without morphological instability. Phys Rev E 63:031504 10. Trivedi R (2005) The role of heterogeneous nucleation on microstructure evolution in peritectic systems. Scripta Mat 53:47–52 11. Barrio M, Lopez DO, Tamarit JL, Negrier P, Haget Y (1995) Degree of miscibility between non-isomorphous plastic phases: binary system NPG (Neopentyl-glycol)-TRIS tris(hydroxymethyl)aminomethane. J Mater Chem 5(3):431–439 12. Ludwig A, Mogeritsch JP, Grasser M (2009) In situ observation of unsteady peritectic growth modes. Trans Indian Inst Met 62:433–6 13. Mogeritsch JP, Eck S, Grasser M, Ludwig A (2010) In situ observation of solidification in an organic peritectic alloy system. Mater Sci Forum 649:159–164 14. Ludwig A, Mogeritsch JP (2011) In situ observation of coupled peritectic growth. In: John Hunt international symposium london, pp 233–242 15. Mogeritsch JP, Ludwig A, Pfeifer T (2017) In-situ observation of coupled peritectic growth in a binary organic model alloy. Acta Mater 126:329–335 16. https://www.qinetiq.com/en/sectors/space 17. http://www.sigmaaldrich.com. Accessed 23 Aug 2021 18. NPG CAS-No 126–30–7 and TRIS CAS-No 77–86–1 19. Kurz W, Fischer DJ (1998) Fundamentals of solidification. Trans Tech Publications Ltd.

Microstructure Evolution of HP40 and HK40 Steels After Isothermal Aging Victor M. Lopez-Hirata, Maribel L. Saucedo-Muñoz, Hector J. Dorantes-Rosales, Carlos Ferreira Palma, Eduardo Pérez-Badillo, and Diego I. Rivas-Lopez

Abstract Thermo-Calc and experimental analyses were carried out for the as-cast and heat-treated HK40 and HP40 steels. The as-cast steel specimens of about 1 × 1 × 1 cm were aged at 900 °C for times between 100 and 1000 h. As-cast and heat-treated specimens were characterized by X-ray diffraction, SEM, and Vickers hardness. Thermo-Calc Scheil analysis indicated an austenite matrix with the presence of Crrich M7 C3 , and M7 C3 and (Nb,Ti)C carbides in the as-cast HK40 and HP40 steels, respectively, which is in agreement with X-ray diffraction, optical microscope, and scanning electron microscope results of present work. The aging treatment at 900 °C produces the precipitation of M23 C6 carbides, and M23 C6 and (Nb,Ti)C carbides in the austenite matrix of the as-cast HK40 and HP40 steels, respectively. This precipitation process promotes an increase in hardness. Thermo-Calc Prisma also predicted the precipitation of the same phases, and the fastest growth kinetics of precipitation was at 900 °C for the as-cast HK40 and 960 °C for the as-cast HP40 steel. In general, the aging response is better for the as-cast HK40 steel. Keywords Heat-resistant steels · HK40 steel · HP40 steel aging · Precipitation · Mechanical properties

Introduction Heat-resistant cast steels are designed to be used at high temperatures and thus the microstructure degradation may occur after its prolonged service exposure above 800 °C [1]. For instance, the HK40 and HP 40 steels have been applied on manufacture of different components for oil, gas, thermal power, chemical, and petrochemical industries [2–7]. These steels are basically Fe–Ni-Cr alloys, which are used in the ascast condition. The HK40 and HP 40 steels have good creep strength and resistance to oxidizing and carburizing atmospheres. The mechanical strength of this steel comes V. M. Lopez-Hirata (B) · M. L. Saucedo-Muñoz · H. J. Dorantes-Rosales · C. F. Palma · E. Pérez-Badillo · D. I. Rivas-Lopez Instituto Politécnico Nacional (ESIQIE-ESFM), UPALM Edif. 7, Mexico, D.F. 07300, Mexico e-mail: [email protected] © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_137

1442

Microstructure Evolution of HP40 and HK40 Steels …

1443

from a tough austenitic matrix strengthened with different carbides such as M7 C3 and M23 C6 . The aging process of these steels is usually conducted during the service operation. These components are sensitive to failure because of creep damage, creep fatigue, σ phase embrittlement, carburization, thermal shock, and high-temperature corrosion [4]. The microstructural changes during heating of this steel are the main responsible for most of these failures [2–5]. Therefore, several studies have been conducted to describe the microstructural changes in this type of steel [8]. ThermoCalc [9] has been applied successfully to analyze the microstructural changes in different alloys at the service temperatures. Besides, Thermo-Calc enables us to calculate the expected phases and its composition for an alloy system not only in the equilibrium state, but also in the no equilibrium condition. That is, the as-cast HK40 and HP 40 steels can also be analyzed using this software. Thus, the purpose of the present work is to analyze the microstructure evolution during aging at 800 and 900 °C, and its effect on the mechanical behavior for the as-cast HK40 and HP 40 steels.

Experimental Procedure The chemical composition of the HK40 and HP40 steels is shown in Table 1. This composition is according to the nominal one in the ASTM A-351 standard [4]. Specimens of about 10 × 10x10 mm were cut from the cast ingot. These were heated (aged) at 900 °C for up to 2000 h in an electric tubular furnace. The as-cast and heattreated specimens were metallographically prepared, etched with Kalling reactant, 5 g CuCl2 , 100 ml HCl, and 100 ml CH3 -CH2 -OH, and then observed with both optical (OM) and scanning electron (SEM) microscopes. Energy-Dispersive X-ray (EDX) analysis was used to determine the chemical composition of some precipitates in the steel specimens. Some specimens were also analyzed by X-Ray Diffraction (XRD) with Cu K α radiation. According to the standard procedure, the Rockwell “C” hardness was determined in all the specimens using 150 kgf with a diamond indenter. Table 1 Chemical composition of HK40 and HP40 steels Steel

C

Mn

Si

Nb

Ti

Cr

Ni

Fe

HK40

0.40

1.5

1.6

0

0

25

20

Bal

HP40

0.40

0.73

1.11

1.27

0.13

27.9

33.2

Bal

1444

V. M. Lopez-Hirata et al.

Numerical Procedure Thermo-Calc, TC, software was used to analyze the stability of phases in the as-cast HK40 and HP40 steels. Likewise, the analysis of precipitation of the aged steels was pursued employing the Thermo-Calc (TC) PRISMA software. The kinetic and thermodynamic data was determined from the Thermo-Calc databases for TCFe7 and MobFe2 [9]. Homogeneous nucleation was assumed for the precipitation simulation in the austenite matrix, named bulk nucleation in TC-PRISMA.

Results and Discussion Microstructure Evolution of the As-Cast Steels Figures 1 (a and b) show the TC calculated Scheil diagram of temperature vs. mole fraction of solid for the as-cast HK40 and HP40 steels, respectively. In the case of the HK40 steel, the γ austenite phase is the first solid phase formed during solidification, and subsequently the M7 C3 and M23 C6 carbides. The following eutectic reaction, L → γ + M7 C3 , caused the formation of M7 C3 carbide. The presence of M23 C6 carbide takes place by a precipitation reaction from the austenite dendrites, as will be shown later. The TC calculated chemical composition of M7 C3 carbide indicates a Cr-rich carbide, containing Fe, Ni, and Mn. In contrast, the Scheil diagram of the HP40 steel, Fig. 1b, shows that the first formed solid phase corresponds to a MC-type carbide and then the M7 C3 carbide, which is produced also by a eutectic reaction, L → γ + M7 C3 . The TC calculated chemical composition of MC and M7 C3 carbides illustrates that the MC-type carbide is mainly composed of Nb and Ti, while the M7 C is a Cr-rich carbide. The separation between the equilibrium line in black and non-equilibrium line in different colors is wider for the as-cast HK40 steel than that of the as-cast HP40 steel, as shown in Fig. 1a and b. This fact suggests that the

Fig. 1 TC calculated plot of temperature versus mole fraction of solid for the a HK40 and b steels

Microstructure Evolution of HP40 and HK40 Steels …

1445

Fig. 2 TC calculated plot of C composition in austenite vs. mole fraction of solid for the as-cast a HK40 and b steels

microsegregation of the former steel is higher than the latter. Besides, the chemical composition of M7 C3 carbide presents a higher Cr content for the HP40 steel than that of the HK40 steel, which seems to be attributable to a higher Cr content of the former steel. Figure 2a and b present the TC calculated plot of the C content in austenite versus mole fraction of solid for the HK40 and HP40 steels, respectively. These figures indicate that the microsegregation corresponds to an alloy system with a partition coefficient k < 1. That is, the dendrite presents a solute enrichment from the center to the outer part of dendrite. Both Fig. 2a and b show an increase in the C content from the first solidified solid to the last one. Likewise, the C microsegregation is higher for the HK40 steel than that of the HP40 steel, which can be associated with its higher Si content. In both cases, the formation of carbides takes place after the maximum C content. Figure 3a–d illustrate the SEM micrograph and EDX elemental mapping for the as-cast HK40 steel. The eutectic microconstituent of M7 C3 and austenite, and the austenite dendrites are clearly observed in the micrograph. Besides, the EDX elemental mapping confirms the presence of Cr-rich M7 C3 carbides. Mn is mainly located in the eutectic microconstituent, Fig. 3d. In contrast, the SEM micrograph of the as-cast HP40 presents, in addition to the eutectic microconstituent of M7 C3 and austenite dendrites, the MC carbide, Fig. 4a. The EDX elemental mapping, Fig. 9c–f, reveals that the M7 C3 is rich in carbide and the MC carbide is Nb-rich. Mn is not located in the eutectic microconstituent for this case. This can be also associated with the Si content. Additionally, the XRD pattern of the as-cast HK40 and HP40 steels, as shown in Fig. 5a and b, respectively, confirm the presence of austenite and M7 C3 carbide, and austenite, M7 C3 carbide, and MC carbide, respectively. All the above results are consistent with the TC calculated results of Figs. 1 and 2. The main difference between these two steels is that the C microsegregation is higher for the HK40 steel and the formation of the MC carbide in the case of HP40 steel.

1446

V. M. Lopez-Hirata et al.

Fig. 3 a SEM micrograph and EDX elemental mapping of b Cr, c Ni, and d Mn for the as-cast HK40 steel

Fig. 4 a SEM micrograph and EDX elemental mapping of b Cr, c Ni, d Mn, and e Nb for the as-cast HP40 steel

Microstructure Evolution of HP40 and HK40 Steels …

1447

Fig. 5 XRD pattern of the as-cast a HK40 and b HP40 steels

Microstructure Evolution of the Aged Steels Figure 6a and b show the TC calculated plot of the amount of all phases in equilibrium as a function of the temperature for the HK40 and HP40 steel, respectively. These figures also show the presence of the eutectic reaction L →γ + M7 C3 . Besides, the following precipitation reaction γ → γ + M23 C6 can occur at temperatures lower than 1100 °C for the HK40 steel. In contrast, there is a competition between these two reactions γ → γ + M23 C6 , and γ → γ + MC for the HP steel. The Time–Temperature-Precipitation diagram for these steels was calculated with TC-Prisma. The TC calculated chemical composition of dendrites is shown in Table 2. The TTP diagrams are shown in Fig. 7a and b for intragranular precipitation of the HK40 and HP40 steels, respectively. Figure 7a confirms that the main precipitation

Fig. 6 TC calculated plot of the total amount of phases in equilibrium versus temperature the as-cast a HK40 and b HP40 steels

1448

V. M. Lopez-Hirata et al.

Table 2 TC calculated chemical composition of austenite dendrites for HK40 and HP40 steels Steel

C

Mn

Si

Nb

Ti

Cr

Ni

Fe

HK40

0.10

1.4

1.6

0

0

22.3

24.5

Bal

HP40

0.08

0.07

0.95

0.09

0.03

25.3

34.4

Bal

Fig. 7 TC-Prisma calculated TTP diagram of the as-cast a HK40 and b HP40 steels

is γ → γ + M23 C6 for the former steel. Figure 8b shows the SEM micrograph of the HK40 steel aged at 900 °C for 500 h. This micrograph indicates the presence of the austenite dendrites with dispersed precipitate plates of M23 C6 carbide. Besides, the interdendritic microconstituent, formed by M7 C3 and austenite, can also be noted. Furthermore, the eutectic microconstituent is gradually dissolved into the austenite to promote more precipitation of M23 C6 . In the case of the TTP of the HP 40 steel, the precipitation of reactions γ → γ + M23 C6 , and γ → γ + MC is verified. This fact is also noted in the SEM micrograph, as shown in Fig. 8b. The MC precipitation is shown by small cuboid precipitates, while the precipitation M23 C6 exhibits a plate morphology. This micrograph also permits observing the presence of the eutectic microconstituent.

Fig. 8 SEM micrographs of the as-cast a HK40 and b HP40 steels aged at 900 °C for 500 h

Microstructure Evolution of HP40 and HK40 Steels … Table 3 Rockwell “C” hardness of the steels

1449

Steel

As-cast condition

Aged condition

HK40

14

17

HP40

12

13

The precipitation process of the HP40 steel shows faster growth kinetic as shown in Fig. 8b, compared to that of the HK40 steel, Fig. 8a, which is attributable to its higher Cr solute content, as shown in Table 2. This fact causes a higher driving for the precipitation [10]. Nevertheless, the volume fraction was higher for the latter steel than that of the former, which seems to be attributable to the highest C content, and the precipitation competition of M23 C6 and MC carbides during aging of the HP40 steel.

Mechanical Properties of the Aged Steels Table 3 presents the hardness values for the as-cast HK40 and HP40 steels, and for aging at 900 °C for 500 h. The Rockwell “C “hardness of the as-cast HK40 steel is slightly higher than that reported is marginally higher than that of the as-cast HP40 steel. This behavior seems to be attributable to its higher content of solutes such as Si and Mn. Additionally, the hardness for the aged condition is higher for the former steel than the latter due to its higher volume fraction of M23 C6 precipitates. These results are in good agreement with results reported in the literature [1–9].

Conclusions A study of the precipitation process in two resistant steels, HK40 and HP40, was conducted, and the conclusions are summarized as follows: 1. 2. 3. 4.

The C microsegregation was higher for the as-cast HK40 than that of the as-cast HP40 steel, which can be attributable to its higher Si content. Mn is present in the eutectic microconstituent for the as-cast HK40 steel because of its higher microsegregation behavior than the other steel. The as-cast HK40 steel responses better to the precipitation hardening treatment than the other steel, forming a higher volume fraction of precipitates. The growth kinetics of M23C6 precipitation was faster for the HP40 steel due to its higher Cr content.

Acknowledgements The authors wish to thank the financial support from SIP-COFAA-IPN and CONACYT A1-S-9682.

1450

V. M. Lopez-Hirata et al.

References 1. Davis JR (1997) ASM specialty handbook: heat-resistant materials. ASM International, USA 2. Kaya AA (2002) Microstructure of HK40 alloy after high temperature service in oxidizing/carburizing environment, I. Oxidation phenomena and propagation of a crack. Mater Charact 49:11–21 3. Soares GDA, Almeida LH, Silveira TL, May IL (1992) Niobium additions in HP heat-resistant cast stainless steels. Mater Charact 29:387–396 4. ASTM Standard A351/A351M–03 (2004) Standard specification for casting, austenitic, austenitic-ferritic (duplex), for pressure-containing parts, ASTM 5. Alvino A (2014) Influence of chemical composition on microstructure and phase evolution of two HP heat resistant stainless steels after long term plant-service aging. Mater High Temp 31:1–11 6. Buchanan KG, Kral MV (2012) Crystallography and Morphology of niobium carbide in as-cast HP-niobium reformer tubes. Met Mater Trans A 43:1760–1765 7. Ribeiro EAAG, Papaleo R, Guimaraes JRC (1986) Microstructure and creep behavior of a niobium alloyed cast heat-resistant 26 pct Cr steel. Met Trans A 17:691–696 8. Thomas CW, Stevens KJ, Ryan MJ (1996) Microstructure and properties of alloy HP50–Nb: comparison of as cast and service exposed materials. Mater Sci Tech 12:469–475 9. Thermo-Calc software, TCFe7 and MoFe2, Version 2021 a, 2021 10. Kostorz G (2001) Phase transformations in materials. Wiley-VCH, Germany

Precipitation Process During Isothermal Aging of an Austenitic Stainless Fe-12Cr-10Mn-12Ni-5Mo-0.24 N-0.03C Steel and Its Effect on the Mechanical Properties Maribel L. Saucedo-Muñoz, Victor M. Lopez-Hirata, Erika O. Avila-Davila, Felipe Hernandez-Santiago, and Jose D. Villegas-Cardenas Abstract The precipitation process was studied during the isothermal aging of a Fe12Cr-10Mn-12Ni-5Mo-0.24 N-0.03C steel at temperatures of 700, 800, and 900 °C for times up to 1000 min. The aged specimens were characterized by SEM, XRD analysis, and CVN impact test at –196 °C. Thermo-Calc analysis indicated that the Mo-rich M6 C carbide precedes the precipitation of Cr-rich M2 N nitrides in the austenite matrix at aging temperatures lower than 700 °C. The fastest growth kinetics of intragranular takes place at about 800 °C for the M6 C carbide and 850 °C for the M2 N nitride. The CVN impact energy decreases with aging time and temperature. The presence of brittle fracture also increases with aging time and temperature in the tested specimens. The intergranular precipitation was the main responsible for the reduction of cryogenic toughness. Keywords Austenitic stainless steels · Aging · Precipitation · Cryogenic toughness

Introduction The austenitic stainless steels are usually exposed to high temperature for very prolonged times during in-service operation of different industrial components or in continuous cooling after the welding process while manufacturing industrial parts. Thus, these steels may be subject to the intergranular and intragranular precipitation of different phases on the austenite matrix phase because of the heating process [1– 6]. This precipitation could cause the sensitization and the reduction of mechanical properties of stainless steel. The precipitation of austenitic steels involves different phases [1, 4] such as carbides, nitrides, Laves phases, among other phases. The precipitation of phases with chromium is associated with a decrease in corrosion resistance [5]. However, M. L. Saucedo-Muñoz · V. M. Lopez-Hirata (B) · E. O. Avila-Davila · F. Hernandez-Santiago · J. D. Villegas-Cardenas Instituto Politécnico Nacional (ESIQIE-ESFM), UPALM Edif. 7, 07300 Mexico, D.F, Mexico e-mail: [email protected] © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_138

1451

1452

M. L. Saucedo-Muñoz et al.

the presence of precipitates is essential to give this type of steel creep strength when they are used in applications at high temperatures [5]. Besides, these steels are used for cryogenic applications at very low temperatures. In this case, the presence of Crrich precipitates, originated by previous heating, may reduce the toughness property, which compromises its structural integrity. In the case of cryogenic austenitic stainless steels, N-containing steels have been developed [1–4] to reach higher yield strength and fracture toughness KIC at cryogenic temperatures, 1200 MPa, and 200 MPam−1/2 , respectively, than those of conventional austenitic stainless steels without decreasing toughness. It is also important to note that, in contrast to carbon, nitrogen-containing austenitic stainless steels maintain high fracture toughness at low temperatures. Thus, these mechanical properties of nitrogen-containing austenitic stainless have made its application very attractive in the power-generation industry, shipbuilding, railway, cryogenic process, chemical equipment, pressure vessels, and nuclear sectors. Thus, it is important to evaluate the degree of microstructural degradation due to the precipitation phenomenon which may affect the cryogenic toughness in the austenitic stainless steels [1–6]. Thermo-Calc and Prisma software has been used to analyze the phase formation, reaction precipitation, and precipitation growth kinetics in the steels [7]. Hence, the purpose of this work is to carry out the numerical and experimental analyses of the precipitation process for the isothermally aged Fe-12Cr-10Mn-12Ni5Mo-0.24 N-0.03C steel to determine its effect on the cryogenic toughness.

Experimental Procedure The chemical composition of Fe-12Cr-10Mn-12Ni-5Mo-0.24 N-0.03C steel is shown in Table 1. Specimens for Charpy-V-Notch test of about 10 mm × 10 mm × 55 mm were cut from plates of about 250 mm thick in a spark-erosion cutting machine. These specimens were solution treated at 1050 °C for 50 min in an electric furnace and subsequently quenched in iced water. Then, to cause precipitation in these two steels, the solution treated samples were artificially aged at temperatures of 700 °C, 800 °C, and 900 °C for times from 10 to 1000 min. The CVN impact test was carried out at −196 °C (liquid nitrogen temperature). The testing temperature was kept constant for 10 min before testing. Three specimens were tested for each heat-treating condition. The fracture surface of the tested CVN test specimens was analyzed at 15 kV in a SEM. Specimens of 10 × 10 × 10 mm were extracted from the opposite side of the fracture surface to carry out its corresponding Table 1 Chemical composition of the steel C

Si

Mn

Ni

Cr

N

Mo

0.025

0.48

10.13

11.79

12.01

0.236

4.94

Precipitation Process During Isothermal Aging of an Austenitic …

1453

metallographic analysis. The solution treated and the aged specimens were metallographically prepared using emery-paper up to 1500 grit number and polished with alumina 0.05 μm. After that, polished specimens were chemically etched in a solution of Villella’s reagent (1 g of picric acid and 5 ml of hydrochloric acid in 100 ml of ethylic alcohol) for about 2 min. Finally, the etched specimens were characterized with an optical microscope (OM) and scanning electron microscope (SEM) at 15 kV equipped with an EDX microanalysis system. The precipitates were extracted by electrochemical dissolution at room temperature and 6 V (D.C.) of the austenite matrix. The used chemical solution was an electrolyte of 5 vol. % hydrochloric acid in methanol. The extracted precipitates were analyzed with an X-ray diffractometer using monochromatic Kα copper radiation.

Numerical Procedure Thermo-Calc software was used to analyze the stability of phases. Likewise, the analysis of precipitation was pursued in the aged steel employing the Thermo-Calc (TC) PRISMA software. The kinetic and thermodynamic data were determined from the Thermo-Calc databases for TCFe7 and MobFe2 [7]. Homogeneous nucleation and grain boundary assumed for the precipitation simulation in the austenite matrix, named bulk nucleation in TC-PRISMA.

Results and Discussion Thermo-Calc and TC-Prisma Calculations Figure 1 shows the Thermo-Calc TC calculated pseudobinary Fe–C equilibrium phase diagram. This diagram clearly illustrates the solvus curve, and the austenite phase is completely present at temperatures above approximately 930 °C. This diagram also indicates the presence of austenite γ, M2 N nitride, and M6 C carbide. This figure also suggests that the following precipitation reactions, γ → γ + M2 N and γ → γ + M6 C, may occur during aging at temperatures lower than 930 °C. Table 2 presents the TC calculated volume % of the equilibrium phase, Fe γ, M6 C, and M2 N, at 700, 800, and 900 °C. The M2 N nitride precipitate is present in a higher volume fraction than that of the M6 C carbide at all the temperatures. This fact suggests that the following precipitation reaction, γ → γ + M2 N, is predominant at all temperatures. Besides, Table 3 shows the chemical composition of the phases in Table II, for instance, at 800 °C. This composition clearly illustrates that the M2 N is Cr-rich, while the M6 C carbide is Mo-rich.

1454

M. L. Saucedo-Muñoz et al.

Fig. 1 TC calculated pseudobinary Fe–C equilibrium phase diagram of the steel Table 2 TC calculated volume percent of equilibrium phases at different temperatures

Temperature (°C)

Phase

Vol. %

700

Fe γ

93.12

M2 N

2.74

M6 C 800

900

Table 3 TC calculated chemical composition of equilibrium phases at 800 °C

0.74

Fe γ

95.2

M2 N

2.56

M6 C

0.62

Fe γ

97.67

M2 N

2.04

M6 C

0.29

wt. % element/phase

Fe γ

M2 C

M6 C

Fe

62.34

0.66

25.99

Ni

12.35

0.06

0.02

Cr

10.90

69.61

6.75

Mn

10.56

2.21

0.0

Mo

3.32

16.85

64.24

Si

0.50

0

0

N

0.03

10.46

0

C

0.004

0.13

2.56

Precipitation Process During Isothermal Aging of an Austenitic …

1455

Fig. 2 TC-Prisma calculated TTP diagram for the intergranular precipitation of the steel

The TC-Prisma calculated Time–Temperature-Precipitation (TTP) diagram is shown in Figs. 2 and 3 for the intergranular and intragranular precipitation, respectively. In the case of intergranular precipitation, TC-Prisma results indicate that the following precipitation reaction γ → γ + M2 N occurs at aging temperatures from 600 to 900 °C, Fig. 2. No M6 C intergranular precipitation is detected for this case. On the other hand, Both M2 N and M6 C intragranular precipitations can occur during aging at temperatures between 600 and 950 °C. Besides, the precipitation reaction, γ → γ + M2 N, takes place initially at aging temperatures higher than 750 °C. In contrast, the precipitation reaction γ → γ + M6 C occurs firstly at temperatures lower than 750 °C. The fastest intergranular and intragranular precipitation of the M2N nitrides takes place at approximately 825 and 850 °C, respectively. This fact seems to be reasonable since the grain boundaries are more favorable nucleation sites [8].

Microstructural Characterization of Precipitation Figure 4a–d show the SEM micrographs of the steel specimens in the solution treated condition and after aging at 700, 800, and 900 °C for 1000 min. No precipitation at all can be noted for the solution treated specimen, Fig. 4a, The intergranular precipitation is the only observed at 700 °C, Fig. 4b, while both intergranular and intragranular

1456

M. L. Saucedo-Muñoz et al.

Fig. 3 TC-Prisma calculated TTP diagram for the intragranular precipitation of the steel

precipitations can be noted at 800 and 900 °C, Fig. 4c and d. The higher the aging temperature, the more abundant the precipitation, which is in good agreement with the fastest precipitation kinetics in the TTP diagrams for intergranular and intragranular precipitations in Figs. 2 and 3, respectively. The XRD pattern of the residues electrolytically extracted from the steel aged at 800 °C for 1000 min indicates mainly the presence of M2 N and M6 C, Fig. 5. Some amount of austenite was dragged into the residues. As expected, the XRD peaks with higher intensity correspond to the M2 N nitrides, which suggests a higher volume fraction of this precipitate, in comparison with the M6C carbide. Similar results were found for the specimens aged at 700 and 900 °C for 1000 min. This fact is consistent with the TC calculated results. The EDX-SEM analysis of the 1000 min-900°C aged steel revealed that some intragranular precipitates are Mo-rich, while most intergranular precipitates are Crrich. This fact confirms that the former corresponds to M6 C carbides and the latter to M2 N, which is in good agreement with TC and TC-Prisma calculated results.

Precipitation Process During Isothermal Aging of an Austenitic …

1457

Fig. 4 SEM micrographs of the steel in the a solution treated condition and after aging at b 700, c 800, and d 900 °C for 1000 min. of the steel

Effect of Precipitation in Cryogenic Toughness The energy variation of the CVN impact test at −196 °C with aging time is shown in Fig. 6 for the steel aged at 700, 800, and 900 °C. This plot presents a decrease in energy with aging time. The higher the aging temperature, the higher the reduction in CVN impact test energy. This fact seems to be caused by the highest volume of precipitation at 900 °C. Figure 7 shows the plot of CVN impact test energy as a function of the volume fraction of precipitates, measured using SEM micrographs. This figure clearly shows the decrease in energy with the increase in volume fraction of precipitates. Fractography of the tested specimen reveals an increase in the volume fraction of brittle fracture with aging time and temperature.

1458

M. L. Saucedo-Muñoz et al.

Fig. 5 XRD pattern of the residues extracted from the steel aged at 800 °C for 1000 min

Fig. 6 Plot of CVN impact test energy versus aging time for the steel aged at 700, 800, and 900 °C

Precipitation Process During Isothermal Aging of an Austenitic …

1459

Fig. 7 Plot of CVN impact test energy versus volume fraction of precipitates for the steel aged at 700, 800, and 900 °C

Conclusions The numerical and experimental analyses of the precipitation process of Fe-12Cr10Mn-12Ni-5Mo-0.24 N-0.03C steel during aging at 700, 700, and 800 °C permitted to state the following conclusions: 1. 2. 3. 4.

The volume fraction of M2 N precipitates was higher than that of M6 C carbides, and it dominated at aging temperatures higher than 750 °C. The M6 C carbides precipitated intragranularly in competition with M2 N nitrides. The results of experimental characterization are in good agreement with the corresponding TC calculated ones. The CVN impact test energy at -196 °C is critically affected by the precipitation of M2N nitrides and M6 C carbides. Likewise, this energy decreases with aging time and the increase in aging temperature because of the increase in volume fraction of precipitates.

Acknowledgements The authors wish to thank the financial support from SIP-COFAA-IPN and CONACYT A1-S-9682.

1460

M. L. Saucedo-Muñoz et al.

References 1. Suemune K, Sakamoto T, Ogawa T, Ozaki T, Maehara S (1998) Manufacture and properties of nitrogen-containing Cr-Mn and Cr-Ni austenitic stainless steels for cryogenic use. Adv Cryog Eng 34A:123–129 2. Nakajima H, Nunoya Y, Nozawa M, Ivano O, Takano K, Ando T, Ohkita S (1996) Development of high strength austenitic stainless steel for conduit of Nb3 Al conductor. Adv Cryog Eng 42A:323–330 3. Stoter LP (1981) Thermal ageing effects in AISI type 316 stainless steel. J Mat Sci 16:1039–1051 4. Summers LT, Walsh RP, Miller JR (1996) The cryogenic tensile and fracture toughness properties of austenitic steels given low temperatures, short-time Nb3 Sn reaction treatments. Adv Cryog Eng 42A:339–344 5. Simmons JW, Covino BS, Hawk JA, Dunning JS (1996) Effect of nitride (Cr2 N) precipitation on the mechanical, corrosion and wear properties of austenitic stainless steels. ISIJ Int 36:846–854 6. Sauced-Muñoz ML, Lopez-Hirata VM, Avila-Davila EO, Villegas-Cardenads D, GonzalezVelazquez JL (2017) Effect of precipitation on cryogenic toughness in isothermally aged austenitic stainless steels. Metals Sci Heat Treat 58:707–711 7. Thermo-Calc and Prisma software, TCFe7 and MobFe2, Version 2021a, 2021 8. Kostorz G (2001) Phase transformations in materials. Wiley-VCH, Germany

Part XL

Powder Materials Processing and Fundamental Understanding

Characterization of Gas Atomized Nickel Silicide Powder for Additive Manufacturing with Varying Silicon Content Mohammad Ibrahim, Tor Oskar Sætre, and Ragnhild E. Aune

Abstract The industry is presently moving towards an integration of advanced production systems, where additive manufacturing (AM) is an important contributor that provides a way to build complex lightweight and strong structures. Quality consistency is a challenge in AM, and it is therefore essential that the properties of the raw materials are adequately understood to allow process optimization. In the present study, a comprehensive review of the characteristics of gas atomized NiSi16, NiSi32, and NiSi38 has been performed using scanning electron microscope (SEM), X-ray diffraction (XRD), particle size distribution (PSD), X-ray photoelectron spectroscopy (XPS), and differential scanning calorimetry (DSC). The flow behavior of the powders has also been studied using a Hall flowmeter. The results clearly show the effect of a varying silicon content on the microstructure and physical properties of the powders, and thereby also on their suitability with respect to the fabrication of complex geometrical shapes by AM. Keywords Additive manufacturing · Gas atomization · Nickel silicide · Silicon · Parameters

Introduction Additive manufacturing (AM) is a process of integrating materials in a tight specification to form 3D components by gradually building up layers that are then selectively fused together [1, 2]. The process is also referred to as additive fabrication or additive layer manufacturing. Although the process has been around and used for materials like polymers and plastics for three decades, the commercial aspect of the technology M. Ibrahim (B) · T. O. Sætre · R. E. Aune Department of Engineering and Science, Faculty of Engineering Sciences, University of Agder, Grimstad, Norway e-mail: [email protected] R. E. Aune Department of Materials Science and Engineering, Norwegian University of Science and Technology (NTNU), Trondheim, Norway © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_139

1463

1464

M. Ibrahim et al.

has come to light in the last 4–5 years. Finally, AM has evolved from the class of manufacturing techniques focused on production of prototypes to those of end-use components [3]. Even though many names are coined for various AM techniques, they basically have the same approach. It involves starting with a 3D CAD model on a computer and virtually slicing this model into layers. The thickness of these layers varies from 1 mm to 20 μm depending on which AM process is being used [4]. A physical part is then derived from these data by repeatedly depositing layer by layer of the material. This includes highly localized melting of the raw materials by a heat source (laser, electron beam, etc.) and solidification of the micro-weld bead. In fact, a region may undergo numerous melting and solidification rounds during an AM process due to parameters like depth of diffusion of the heat source, scan patterns, and speed [5]. An important component of metal AM is the raw material used for it; this raw material can be different depending on the type of technology used for its production. Like any other process, it is important that the performance of the raw material can be controlled so that process efficiency and quality of the printed product can be ensured [6]. Some of the most widely used technologies like selective laser melting and electron beam melting use powdered raw material. In principle, powdered raw material needs to have good flow properties so that it can be spread uniformly resulting in good packing characteristics. It allows the formation of high-density powder layers. These powder characteristics have an impact on the properties of the printed part such as porosity and density [7]. Keeping the above-mentioned characteristics in mind, spherical shape for the particles is preferred for the feedstock powder particularly in powder bed types. Direct energy deposition or laser melting deposition techniques are much more tolerant of powder in fragmented shapes as long as a stable feed-rate can be maintained through the nozzle [5]. Atomization (gas and water) processes are the most common method for production of spherical metal powders [8]. Gas atomization is preferred over water atomization in AM processes because water atomization produces powders with higher oxygen content [9]. This is undesirable as oxide layers on the powders can have an impact on powder flow, melt pool, and characteristics of the printed parts [10]. Other major problems arising from feedstock powder include cracking or microstructure defects in printed parts that remain after processing. Some of these defects are related to the build parameters of the AM processes and can be minimized using heat treatments, but some defects in porosity are attributed to the initial feedstock material and cannot be fixed by post-processing [5]. One such example is the porosity arising from the satellites which prohibit the smooth flow and packing of powders [11]. Porosity in an AM build can also be attributed to adsorbed water vapors on the surface of AM powders which promotes agglomeration. It is usually because of poor atmosphere control during powder storage [12]. In AM, the approach towards optimization of powders is usually not taken. Instead, if a powder is available that seems suitable when it comes to size and chemistry, it is chosen, and the process parameters are changed accordingly to yield parts with acceptable quality. However, if AM has to rival conventional manufacturing techniques, fine tuning of powder properties to obtain optimum results is necessary.

Characterization of Gas Atomized Nickel Silicide Powder …

1465

Industries that want to develop more complex parts from high performance metals are attracted towards additive manufacturing as it provides them that ability while also maintaining the structural integrity of such parts in demanding environments. Nickel-based alloys are known for their exceptional mechanical properties and high temperature oxidation and wear resistance [13]. They share these properties with nickel silicides that have excellent corrosion resistance as well, but they also have a high melting point and hardness which makes it difficult to machine and cast them. Silicon can impart strong corrosion resistance to nickel and iron alloys. It also increases the oxidation resistance of the materials. High-silicon nickel was first introduced in 1929 for the first time. High-silicon nickel alloys become unworkable at a certain stage of silicon content. Although these alloys are still used due to their exquisite resistance to abrasion, mechanical wear, and corrosion, they are subject to failure by mechanical or thermal shock. Nickel silicon alloys containing up to 10% silicon were found to be quite strong, but they were not resistant to corrosion. Whereas if the silicon content was increased to 15%, their resistance rises sharply but they also become extremely brittle [14]. In this study, NiSi powders produced using gas atomization were thoroughly analyzed to determine the effects of varying silicon content in them. Microstructure of these powders was observed using scanning electron microscopy. Their composition was determined by the use of X-ray diffraction and X-ray photoelectron spectroscopy. Differential scanning calorimetry helped determine their behavior under elevated temperature, whereas their physical properties were determined using their particle size distribution and powder flow abilities.

Materials and Characterization The scope of this study was to investigate the properties of gas atomized nickel silicide and to see the effect of silicon concentration in these powders. Hence, 16, 32, and 38% by weight concentration of silicon were chosen. NiSi16 and NiSi32 were provided by Phoenix Scientific Industries Ltd (PSI/MPP), and NiSi38 was produced by Indutherm GmbH. Parameters disclosed by the manufacturer are given in Table 1. Scanning electron microscopy (SEM) was done at Electron Microscopy Laboratory of the Materials Department at Norwegian University of Science and Technology (NTNU), Trondheim, using a Low Voltage Field Emission Scanning Electron Table 1 Processing parameters of gas atomized powders

No Sample Atomizing gas pressure Atomizing temperature ( bar) (°C) 1

NiSi16 45–50

1450

2

NiSi32 45–50

1200

3

NiSi38 21.5

1130

1466

M. Ibrahim et al.

Microscope (LVFESEM), Zeiss Supra 55VP. Images were acquired using secondary electron detector to provide information on morphology of atomized powders. The accelerating voltage was set to 10 keV, and the working distance was set to 10 mm. Elemental analysis was performed via energy dispersive X-ray spectroscopy (EDS) using an EDAX detector. SEM analysis is accepted throughout the AM industry as it provides a general overview of the powder particles when it comes to size, shape, and surface roughness. Chemical verification of present phases was done using X-ray diffraction (XRD). For the purpose of the present work, XRD patterns were obtained at an angle range of 10º-100º and recorded using a D8 focus diffractometer using Cu Kα radiation. The obtained scans were later matched with the database provided by Bruker AXS DIFFRAC SUITE EVA. Since powders were manufactured by different producers, they arrived in different sieve cuts. A size range of 25–100 μm was created to make them comparable for analysis. Particle size distribution was carried out using Partica LA-960 from Horiba which makes use of laser diffraction to measure size distributions. Wet analysis method (LD-W) was used in which particles are dispersed in liquid. To qualify a metal powder suitable for additive manufacturing processes, flowability of the powder remains an important property. Particle flow behavior was observed using a hall flow meter from SLM Solutions GmbH. All three powders were dried in vacuum at a 100 °C for 12 h to eliminate the moisture content, and then 50 g of powder was made to flow from a funnel with – μm hole at the bottom and its time to flow out was recorded. Further surface characterization was done using X-ray photoelectron spectroscopy as it provides credible information about the surface ( 30 vol. % precipitates in Pt-Al-Cr-Ni alloys [25] and ~40 vol. % in Pt-Al-Cr-Ru [39, 40]. L12 Pt3 Al has no maximum flow stress anomaly with increasing temperature like L12 -Ni3 Al [41], although it was predicted by first principles calculations [42], and related to the transformation between L12 and D0 c [43]. ~Pt3 Al is stronger at any temperature [25], and the nanohardness of γ was ~12 ± 2.5 GPa compared to ~8 ± 2.5 GPa for γ [44].

1530

L. A. Cornish

There are two conflicting phase diagrams regarding the transformation temperatures of γ Pt3 Al. McAlister & Kahan [45] have the transformation of high temperature L12 Pt3 Al to low temperature tetragonal D0’c at ~1280 °C, whereas Oya et al. [46] have two structures at much lower temperatures: γ → γ 1 at ~340 °C and γ 1 → γ 2 at 127 °C. The high temperature L12 Pt3 Al structure can be stabilised to lower temperatures for good mechanical properties by alloying with Ti, Ta, and Cr [12, 47, 48] or Zr, Hf, Mn, Fe, and Co [24]. The lowest temperature form [46] was confirmed, and is impurity-dependent [24] since different researchers using different source materials achieved different, but reproducible, results. Douglas et al. [49] identified a differently ordered form by TEM, and a similar form was calculated from first principles [50]. In Pt86 :Al10 :Z4 (at.%) alloys (where Z = Ru, Re, W, Mo, Ni, Ti, Ta, and Cr), hardnesses > 400 HV1 were achieved after annealing at 1350 °C for 96 h, with Cr stabilising L12 ~Pt3 Al phase, and Ru being a solid solution strengthener [47, 48, 51]. Both W- and Ni-containing alloys formed lower temperature Pt3 Al. Coarse microstructures were produced in Mo-containing alloys and Mo substituted for Pt in Pt3 Al. However, none of these ternary alloys had high proportions of precipitates, so new alloys were targeted to increase the γ volume fraction, and the compositions were informed by results of Pt-Al-Cr and Pt-Al-Ru [21–23]. The Pt84 :Al11 :Ru2 :Cr3 alloy had a fine two-phase structure, without a primary phase, with better oxidation resistance than the original ternary alloys [52]. Other additions were made to improve properties, and decrease the alloy cost and density. Nickel was added to improve matrix solution strengthening, but with less effect than expected [53]. Cobalt was added for solid solution strengthening and also increased the formability, and extensive phase diagram work was done on Pt-Al-Co [54, 55] and Pt-Al-Ni [53]. The Co, Nb, and V additions were to decrease the Pt content and hence density and price, while simultaneously increasing the melting point [32, 33].

Assessment of Mechanical Properties Compression Testing The best potential alloys were subjected to compressive testing and compared to MAR-M247 (a NBSA) and PM2000 (Fe–Cr–Al alloy with a ferritic matrix mechanically alloyed with Y2 O3 ). Compressive strengths of Pt-Al-Z alloys (where Z = Ti, Cr, Ru, Ta, and Re) [35, 47] were higher than MAR-M247. Creep tests were done at 1300 °C [56] on Pt86 :Al10 :Z4 (at.%) alloys (where Z = Ti, Cr, Ru, Ta, and Ir), and compared to PM2000. Although PM2000 had the highest strength, the shallow slope of the stress-rupture curve indicated high stress sensitivity and brittle creep behaviour, and Pt86 :Al10 :Cr4 had the highest strength of the tested Pt-based alloys. On creep curves obtained at 30 MPa [56], Pt-based alloys had no primary

Platinum-Based Superalloys: Combating High Temperatures …

1531

Table 1 Room temperature mechanical properties of selected Pt-based alloys [22, 59] Alloy composition (at.%)

Hardness (HV10 )

Maximum ultimate tensile strength (MPa)

Elongation (%)

Pt86 :Al10 :Cr4

317 ± 13

836

~4

Pt86 :Al10 :Ru4

278 ± 14

814

~9

Pt84 :Al11 :Ru2 :Cr3

361 ± 10

722

~1

creep, and after secondary creep, there was substantial tertiary creep, with fracture strains between 10 and 50%, whereas PM2000 had very low creep rates and fracture strains < 1%. Creep strengths of the Pt-based alloys at 1300 °C were higher than for Ni- and Co-based superalloys where the precipitates dissolved. The results of Pt86 :Al10 :Cr4 were encouraging, since the precipitate volume was only ~40%. After annealing at 1300 °C for 96 h and helium-quenching, the compression strength of Pt80 :Al14 :Ru3 :Cr3 was higher than Pt86 :Al10 :Cr4 and Pt86 :Al10 :Ru4 because there is a higher proportion of γ , even though it was coarser [57, 58].

Tensile Testing The first tensile tests were done at room temperature on alloys that were not optimised by heat treatment [22, 59], although the samples had been homogenised for twophase microstructures. Due to the high material cost, mini-tensile specimens were used after Lucas et al. [60]. Each flat specimen was cut from a ~50 g ingot by wire spark erosion, after ageing in air at 1250 °C for 100 h and water quenching. Tensile tests were done at 5 mm/min cross-head speeds, and elongations was measured from the specimens (Table 1). TEM showed that the precipitate volume fractions varied between specimens and the properties correlated with the precipitate distributions and morphologies [23]. The spread in the results was due to the low number of specimens. However, the Pt-based alloys were similar to other high temperature alloys: PM2000 with room temperature 720 MPa UTS and 14% elongation [61]; γ-TiAl with 950 MPa UTS and ~1% elongation [62]; and CMSX-4 with 870 MPa UTS [63], despite the unoptimised microstructures.

Cold Rolling Pt-Al-Co and Pt-Al-Co-Cr-Ru alloys were cold rolled [22, 54]; alloys with 400 HV10 hardness had good cold formability (>75% total thickness reduction), while alloys with > 450 HV10 hardness had poor cold formability (0

(2)

Enhancing Mechanical Performance of a Commercial …

1573

VPFFT Method The VPFFT method for simulating has been described in detail [8]. The local constitutive equation (power law relationship) for stress and strain-rate for every Fourier point is solved iteratively. ˙ = γ˙o ε(x)

 s

 m is j

m skl σ (x) τs

n (3)

 where m is j = 21 n is bsj + n sj bis is the Schmid tensor for a particular slip system s, where n s and bs are the normal and Burgers vector of such slip (or twinning) system, τos (x) is the critical resolved stress. The hardening is described by Voce equation: τ = s

τ0s



  s  θ0s s + τ1 + θ1  1 − exp − s τ1

(4)

where the evolution of the slip resistance for each slip system, s, is represented by phenomenological parameters reflecting a purely curve fitting approach: θ0s , τ0s ,θ1s , τ1s , The  term is the accumulated shear strain in each grid-point.

Results and Discussion Microstructural Inhomogeneity Figure 1a, b shows digital generated microstructures that comprise an equiaxedgrain structure (soft domain) that is characterized with high tensile ductility and rolled grained structures (hard domain) that usually has higher strength than their coarse-grained counterparts. The site saturated nucleation is considered in modeling the microstructure evolution of the heterogeneous structure [9]. The recrystallized nuclei are provided to the structure by selecting sites in the rolled layers at early simulation time [9], as indicated by high local GND density. In the undeformed structure layer, the evolution of the structure is controlled by the grain boundary energy [9]. Figure 1c, d obviously produce two-layered structures of coarse grains and fine grains extracted. As a result, the recrystallized structure with a large reduction tends to comprise finer grains because of the high number of nuclei and coarse grain as a grain growth in the undeformed layer [9].

1574

A. M. Mohamed et al. (a)

(b)

(b)

(c)

Annealin

Fine

Annealin

Coarse Fine rain la ers

Coarse Fine

Coarse rain la ers

Fig. 1 Snapshots from 3D microstructural evolution during recrystallization and grain growth of the digital heterogenous structure materials a Rolled in both sides b Rolled in the middle at 25 MCS, c coarse grains bounded by fine grains layers, d fine grains bounded by coarse grains layer

Effect of Grain Size on the Flow Strength Figure 2 represents the flow stress curves of the heterostructures that comprises fine and coarse grains layers simultaneously, and the red stress–strain depicts the heterostructure sample that comprise coarse grain layer bounded by two fine grains layers in both sides. Whereas the black line represents the structure that comprise a fine grain layer bounded by two coarse grain structure. VPFFT does not capture grain effect nevertheless still shows that the heterostructure with two fine grain layers has higher flow stress in comparison with the heterostructure comprises large grain layer it accommodates more strain.

Enhancing Mechanical Performance of a Commercial …

1575

Fig. 2 Engineering stress–strain curves of heterogeneous structure

Conclusions The results show that introducing partial deformation can be useful to control grain size is essential during annealing recrystallization can be of great importance for the aluminum industry. On top of that the present work provides a microstructure design approach that leads to producing microstructures with limited recrystallization fractions. Also, it was found that both coarse and fine grains regions have a direct effect on mechanical properties of heterostructure materials. This way of simulation has the benefit of being able to control the fraction and even spatial distribution of the recrystallized grains. Furthermore, the disparities in mechanical properties can be triggered by presence of different grain size layers.

References 1. Adam K, Belgasam T (2021) A parametric study of grain size and its volume fraction effect on heterogeneous materials mechanical properties. In: TMS 2021 150th Annual meeting & exhibition supplemental proceedings, pp 565–570. https://doi.org/10.1007/978-3-030-652616_51 2. Adam K, Field DP (2021) Developing novel heterogenous microstructures to balance between strength and ductility without restoration processes in commercial Al alloy. Mech Adv Mater Struct 3. Adam K, Root JR, Long Z, Field DP (2016) Modeling the controlled recrystallization of particlecontaining aluminum alloys. J Mater Eng Perform 26(1):207–213

1576

A. M. Mohamed et al.

4. Wu X et al (2015) Heterogeneous lamella structure unites ultrafine grain strength with coarsegrain ductility. Proc Natl Acad Sci USA 112(47):14501–14505. https://doi.org/10.1073/pnas. 1517193112 5. Adam KF, Zöllner D, David P (2018) Field “3D microstructural evolution of primary recrystallization and grain growth in cold rolled single-phase aluminum alloys. Modell Simul Mater Sci Eng 26:035011 6. Adam KF, Long Z, Field DP (2017) Analysis of Particle-Stimulated Nucleation (PSN)dominated recrystallization for hot-rolled 7050 aluminum alloy. Metall Mater Trans A 48:2062–2076. https://doi.org/10.1007/s11661-017-3967-3 7. Rollett AD, Manohar P (2004) “The Monte Carlo Method” Trans. array continuum scale simulation of engineering materials fundamentals microstructures process applications. U. Krieg, Berlin: Wiley-Vch Verlag, Weinheim 8. Lebensohn RA (2001) N-Site modeling of a 3D viscoplastic polycrystal using Fast Fourier transform. Acta Mater 49:2723–2737 9. Adam KF, Field DP (2021) Analyzing recrystallization behavior of heterogeneous structures single-phase Al alloys. Materialia 19:1–13. https://doi.org/10.1016/j.mtla.2021.101190

The Role of the Heterogenous Structure on the Mechanical Properties of Additively Manufactured AlSi10Mg Alloys Haoxiu Chen, Sagar Patel, Mihaela Vlasea, and Yu Zou

Abstract Additive manufacturing (AM) has been widely used to produce AlSi10Mg samples, which have heterogeneous structures. They have coarse cellular dendrites along the melting pools, fine cellular dendrites inside the melting pools, and heataffected zone with broken Si networks beneath the melting pools. The role of the heterogeneous structures on the mechanical properties of AMed AlSi10Mg is rarely studied. Results show that melting pool boundaries deform more than melting pool interiors during plastic deformation. Gradients of plastic deformation build up across melting pool boundaries. Accommodation of such plastic gradients requires the storage of geometrically necessary dislocations, which contribute to work hardening. Correspondingly there will be a gradient in the density of geometrically necessary dislocations. Dislocation density depends on the gradient dendrite size and resulting gradient plastic strains. Keywords Laser metal deposition · AlSi10Mg · Heterogenous structure · Mechanical property

Introduction There are three modes in fabricating AlSi10Mg alloys using LPBF: conduction, transition, and keyhole. The significant differences in absorptivity characteristics for aluminium alloys in the conduction and transition melting modes [1] contribute towards the difficulties in obtaining defect-free Al parts [2–4]. In LPBFed AlSi10Mg samples, the melt pool track boundary can be seen obviously [5–7]. A fine microstructure composed of primary Al phase and eutectic Al-Si H. Chen · Y. Zou (B) Department of Materials Science and Engineering, University of Toronto, 184 College St, Toronto, ON M5S 34, Canada e-mail: [email protected] S. Patel · M. Vlasea Department of Mechanical and Mechatronics Engineering, University of Waterloo, Waterloo, ON N2L 3G1, Canada © The Minerals, Metals & Materials Society 2022 TMS 2022 151st Annual Meeting & Exhibition Supplemental Proceedings, The Minerals, Metals & Materials Series, https://doi.org/10.1007/978-3-030-92381-5_149

1577

1578

H. Chen et al.

network could be observed. Across the melt pool, three zones could be seen according to the size of the cellular dendrites: the fine zones, the coarse zones, and heat-affected zones. The fine zones are inside the melting pools and are made up of small cells. The coarse zones exist at the border of the melt pool with big cells. A heat-affected zone is right behind the coarse zones with broken fibrous Si network. Liu et al. [5] pointed out that the formation of coarse zones might be caused by the semi-solid state during the melting of next hatch or layer. During the melting of the next hatch or layer, the temperature decreases from the contact point of laser to the basic metal. The formation of different zones corresponds to the temperature ranges. Furthermore, Noriko et al. [8] and Chen et al. [9] studied the role of Si on the mechanical properties of LPBFed AlSi10Mg. Xiong et al. [10] focus on melting pools and found that crack initiated and progressed along the melting pools. Stress partitioning behavior between Al matrix and Si precipitates was reported [11, 12], but the strain partitioning behavior between different sized cells is ignored. The role of the heterogeneous structures on the mechanical properties (especially yield strength and work hardening ability) of LPBFed AlSi10Mg alloys needs to be explored.

Materials and Methods Sample Preparation and Microstructure Characterization AlSi10Mg samples were printed on the reduced build volume (RBV) of a modulated LPBF system (AM 400, Renishaw, UK). The AlSi10Mg cylinders were further analyzed for porosity characteristics by a 3D X-ray computed tomography (XCT) scanner (ZEISS Xradia 520 Versa) using a 6 μm voxel size. To visualize the defect distribution within the AlSi10Mg sample, the CT scanned files were analyzed using an image processing software (Dragonfly 3.0, Object Research Systems Inc., Montreal, QC). Both OM (model: Leica DM2500) and SEM (model: Hitachi SU3500 and JSM-6610 V) were used to analyze the microstructure.

Mechanical Test Nanoindenter (KLA iMicro) with a Berkovich diamond tip (Synton-MDP, Switzerland) was used to measure the nanoindentation properties of the sample. Highspeed nanoindentation mapping was performed in selected areas. Tensile tests were conducted in accordance with the ASTM E8/E8M-11 standard with an Instron® 3365 universal testing machine. The load will be measured by a load cell with 5 kN capacity and the strain will be measured by a digital image correlation machine. Tensile tests will be conducted at a strain rate of 2.5 × 10–4 /s to extract the general

The Role of the Heterogenous Structure on the Mechanical …

1579

mechanical properties. At least 3 samples will be tested for each testing condition for each material. In-situ tensile tests were carried out using a Kammrath & Weiss tensile stage with a 5 kN loadcell in a Hitachi SU3500 scanning electron microscope. 2 mm thick dogbone-shaped specimens with a gage length of 10 mm and a gage width of 2 mm were used. Incremental loading of the specimen was performed using a displacement rate of 0.02 mm/min. SEM images captured before and after tensile deformation at given strains were used to calculate the plastic strain via the Matlab Ncorr-2D digital image correlation software.

Results and Discussion Three sets of processing parameters were investigated in this work. We successfully obtained high density (>99.98%) samples with varying melting pools. Examples of OM images and SEM images of the same sample of a XZ section of SLMed AlSi10Mg are shown in Fig. 1. The pile up of melting pools can be seen from OM images. It can be observed that the melt track is characterized with cellular-dendritic microstructures. The grey cellular features are primary α-Al matrix decorated with white Si network. The fine dispersion of Si networks in the Al matrix has a positive effect on the mechanical properties of the as-built LPBF AlSi10Mg samples [9]. Across the melt track, three distinctive regions with different cellular microstructure in size can be distinguished, namely, coarse cellular zone, heat-affected zone, and fine cellular zone. These three regions have experienced different thermal histories

Fig. 1 OM images showing melting pools and SEM images showing big cells along melting pool boundaries and small cells inside melting pools. Z: building direction

1580

H. Chen et al.

Table 1 The sizes of different cells and the width of different zones based on SEM images Size of small cells inside melting pools (μm)

Size of big cells along Coarse zone width melting pools (μm) (μm)

Transitional zone width (μm)

A

0.557

1.049

6.909

1.936

B

0.472

0.823

6.34

1.860

C

0.465

0.804

4.80

1.169

[11]. The sizes of different cells as well as the width of coarse zone and heat-affected zone of three samples were calculated and shown in Table 1. Nanoindentation mapping results together with a corresponding SEM image are shown in Fig. 2. Coarse cellular zones have lower nanohardness compared with fine cellular zones. The relatively uniform elastic modulus distribution means that different sized cells have same elastic modulus. Yellow color in the red matrix could stand for the Si precipitates in the Al matrix. The value of nanohardnes/elastic modulus represents for elastic limit, which is significant for determining the onset of yielding. From the distinct nanohardnes/elastic modulus value between coarse cellular and fine cellular zones, it can be concluded that these two zones will deform differently.

Fig. 2 a Nanohardness map. b Elastic modulus map. c Nanohardness/elastic modulus map. d Corresponding SEM image

The Role of the Heterogenous Structure on the Mechanical … Table 2 LPBFed AlSi10Mg tensile property data

1581

Sample

Yield strength (MPa)

Ultimate tensile strength (MPa)

Elongation (%)

A

179

339

6.17

B

180

352

5.92

C

237

419

6.87

From the tensile deformation performance in Table 2, Sample C, which lies in the transition mode, has higher strength and ductility than sample A and B. In addition, both ultimate tensile strength (UTS) and elongation of the additively manufactured samples are higher than those of cast counterpart alloy (for mechanical properties of die cast A360 from see [13, 14]), as shown in Fig. 3. Such an enhancement in the mechanical behavior is believed due to the very fine microstructure of the LPBFed alloy (cell size of