258 31 24MB
English Pages 768 [727] Year 2021
Heterogeneous Catalysts
Heterogeneous Catalysts Advanced Design, Characterization and Applications
Edited by Wey Yang Teoh, Atsushi Urakawa, Yun Hau Ng, and Patrick Sit
Volume 1
Heterogeneous Catalysts Advanced Design, Characterization and Applications
Edited by Wey Yang Teoh, Atsushi Urakawa, Yun Hau Ng, and Patrick Sit
Volume 2
Editors Prof. Wey Yang Teoh
University of Malaya Department of Chemical Engineering 50603 Kuala Lumpur Malaysia Prof. Atsushi Urakawa
Delft University of Technology Faculty of Applied Sciences Building 58 E2 100 Van der Maasweg 9 2629 Delft The Netherlands Prof. Yun Hau Ng
City University of Hong Kong School of Energy and Environment Tat Chee Avenue Kowloon Hong Kong, S.A.R. Prof. Patrick Sit
City University of Hong Kong School of Energy and Environment Tat Chee Avenue Kowloon Hong Kong, S.A.R. Cover
Cover image: Courtesy of Wey Yang Teoh and Nat Phongprueksathat
All books published by Wiley-VCH are carefully produced. Nevertheless, authors, editors, and publisher do not warrant the information contained in these books, including this book, to be free of errors. Readers are advised to keep in mind that statements, data, illustrations, procedural details or other items may inadvertently be inaccurate. Library of Congress Card No.:
applied for British Library Cataloguing-in-Publication Data
A catalogue record for this book is available from the British Library. Bibliographic information published by the Deutsche Nationalbibliothek
The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available on the Internet at . © 2021 WILEY-VCH GmbH, Boschstr. 12, 69469 Weinheim, Germany All rights reserved (including those of translation into other languages). No part of this book may be reproduced in any form – by photoprinting, microfilm, or any other means – nor transmitted or translated into a machine language without written permission from the publishers. Registered names, trademarks, etc. used in this book, even when not specifically marked as such, are not to be considered unprotected by law. Print ISBN: 978-3-527-34415-4 ePDF ISBN: 978-3-527-81356-8 ePub ISBN: 978-3-527-81358-2 oBook ISBN: 978-3-527-81359-9 Typesetting SPi Global, Chennai, India Printing and Binding
Printed on acid-free paper 10 9 8 7 6 5 4 3 2 1
v
Contents
Volume 1 Preface xv Section I
Heterogeneous Catalysts Design and Synthesis 1
1
Evolution of Catalysts Design and Synthesis: From Bulk Metal Catalysts to Fine Wires and Gauzes, and that to Nanoparticle Deposits, Metal Clusters, and Single Atoms 3 Wey Yang Teoh
1.1 1.2 1.3 1.4
The Cradle of Modern Heterogeneous Catalysts 3 The Game Changer: High-Pressure Catalytic Reactions 5 Catalytic Cracking and Porous Catalysts 8 Miniaturization of Metal Catalysts: From Supported Catalysts to Single-Atom Sites 12 Perspectives and Opportunities 15 References 16
1.5
2
Facets Engineering on Catalysts 21 Jian (Jeffery) Pan
2.1 2.2 2.3 2.3.1 2.3.2 2.3.3 2.4 2.4.1 2.4.2 2.5
Introduction 21 Mechanisms of Facets Engineering 22 Anisotropic Properties of Crystal Facets Anisotropic Adsorption 27 Surface Electronic Structure 28 Surface Electric Field 29 Effects of Facets Engineering 32 Optical Properties 32 Activity and Selectivity 33 Outlook 34 References 35
27
vi
Contents
3
Electrochemical Synthesis of Nanostructured Catalytic Thin Films 39 Hoi Ying Chung and Yun Hau Ng
3.1 3.2 3.2.1 3.2.1.1 3.2.2 3.2.2.1 3.2.3 3.2.4 3.2.4.1
Introduction 39 Principle of Electrochemical Method in Fabricating Thin Film 40 Anodization 42 Pulse or Step Anodization 45 Cathodic Electrodeposition 46 Pulse Electrodeposition 47 Electrophoretic Deposition 48 Combinatory Methods Involving Electrochemical Process 50 Combined Electrophoretic Deposition–Anodization (CEPDA) Approach 51 Conclusions and Perspective 52 References 53
3.3
4
Synthesis and Design of Carbon-Supported Highly Dispersed Metal Catalysts 57 Enrique García-Bordejé
4.1 4.2 4.2.1 4.2.2 4.2.2.1 4.2.2.2
Introduction 57 Preparation of Catalysts on New Carbon Supports 58 Catalyst on Graphene Oxide 59 Catalyst on Graphene 60 Graphene or rGO as Starting Material 60 Graphene Oxide as Precursor of Graphene-Supported Catalyst 61 Graphene Derivatives: Doped Graphene and Synthetic Derivatives 62 Catalyst on Nanodiamonds and Onion-Like Carbon 63 SACs on Carbon Nitrides and Covalent Triazine Frameworks 67 Catalyst on Carbon Material from Hydrothermal Carbonization of Biomolecules 68 Emerging Techniques for Carbon-Based Catalyst Synthesis 69 Deposition of Colloidal Nanoparticles 70 Single-Metal Atom Deposition by Wet Chemistry 71 Immobilization of Metal Clusters and SACs by Organometallic Approach 71 Chemical Vapor Deposition Techniques on Carbon Supports 72 Simultaneous Formation of Metallic Catalyst and Porous Carbon Support by Pyrolysis 73 Dry Mechanical Methods 73 Electrodeposition 73 Photodeposition 74 Conclusions and Outlook 74 References 75
4.2.2.3 4.2.3 4.2.4 4.2.5 4.3 4.3.1 4.3.2 4.3.3 4.3.4 4.3.5 4.3.6 4.3.7 4.3.8 4.4
Contents
5
Metal Cluster-Based Catalysts 79 Vladimir B. Golovko
5.1 5.2
Introduction 79 Catalysts Made by Deposition of Clusters from the Gas Phase Under Ultrahigh Vacuum 81 Chemically Synthesized Metal Clusters 85 Catalysis Using the Chemically Synthesized Metal Clusters 88 Conclusion 95 References 96
5.3 5.4 5.5
103
6
Single-Atom Heterogeneous Catalysts Yaxin Chen, Zhen Ma, and Xingfu Tang
6.1 6.2 6.2.1 6.2.2 6.2.2.1 6.2.2.2 6.2.2.3 6.2.2.4 6.3 6.3.1 6.3.2 6.3.2.1 6.3.2.2 6.4
Introduction 103 Concept and Advantages of SACs 104 Concept of SACs 104 Advantages of SACs 105 Maximum Atom Efficiency 105 Unique Catalytic Properties 105 Identification of Catalytically Active Sites 105 Establishment of Intrinsic Reaction Mechanisms 106 Synthesis of SACs 107 Physical Methods 108 Chemical Methods 108 Bottom-Up Synthetic Methods 109 Top-Down Synthetic Methods 112 Challenges and Perspective 113 References 114
7
Synthesis Strategies for Hierarchical Zeolites 119 Xicheng Jia, Changbum Jo, and Alex C.K. Yip
7.1 7.2 7.2.1 7.2.2 7.2.3 7.2.4 7.3 7.3.1 7.3.1.1 7.3.1.2 7.3.1.3 7.3.2 7.3.2.1 7.3.2.2
Introduction 119 Hierarchical Zeolites 122 Increased Intracrystalline Diffusion 123 Reduced Steric Limitation 123 Changed Product Selectivity 124 Decreased Coke Formation 124 Modern Strategies for the Synthesis of Hierarchical Zeolites 124 Hard Templates 124 Confined-Space Method 125 Carbon Nanotubes and Nanofibers 127 Ordered Mesoporous Carbons 128 Soft Templates 130 Templating with Surfactants 130 Silanization Templating Methods 135
vii
viii
Contents
7.3.3 7.3.4 7.4
Dealumination 136 Desilication 138 Conclusion 140 References 141
8
Design of Molecular Heterogeneous Catalysts with Metal–Organic Frameworks 147 Marco Ranocchiari
8.1 8.2
Secondary Building Units (SBUs) and Isoreticular MOFs 151 The Tools to Build Molecular Active Sites: Reticular Chemistry and Beyond 152 Pre-synthetic Methodologies 153 Post-synthetic Methodologies 155 Post-synthetic Modification (PSM) 155 Post-synthetic Exchange (PSE) 156 MOFs in Catalysis 156 The Difference Between MOFs and Standard Heterogeneous and Homogeneous Catalysts 157 Conclusion: Where to Go from Here 158 References 158
8.2.1 8.2.2 8.2.2.1 8.2.2.2 8.3 8.3.1 8.4
9
Hierarchical and Anisotropic Nanostructured Catalysts 161 Hamidreza Arandiyan, Yuan Wang, Christopher M.A. Parlett, and Adam Lee
9.1 9.2 9.3 9.4 9.5 9.6 9.7
Introduction 161 Top-Down vs. Bottom-Up Approaches 162 Shape Anisotropy and Nanostructured Assemblies 162 Janus Nanostructures 165 Hierarchical Porous Catalysts 169 Functionalization of Porous/Anisotropic Substrates 170 Perspective 174 References 176
10
Flame Synthesis of Simple and Multielemental Oxide Catalysts 183 Wey Yang Teoh
10.1 10.2 10.3 10.4 10.4.1 10.4.2 10.4.3 10.5
From Natural Aerosols Formation to Engineered Nanoparticles 183 Flame Aerosol Synthesis and Reactors 185 Simple Metal Oxide-Based Catalysts 189 Multielemental Oxide-Based Catalysts 192 Solid Solution Metal Oxide Catalysts 192 Composite Metal Oxide Catalysts 192 Complex Metal Oxide Catalysts 197 Perspective and Outlook 197 References 199
Contents
11
Band Engineering of Semiconductors Toward Visible-Light-Responsive Photocatalysts 203 Akihide Iwase
11.1 11.2 11.2.1 11.2.2 11.3 11.3.1 11.3.2 11.3.3 11.4 11.4.1 11.4.2 11.4.3
Basis of Photocatalyst Materials 203 Photocatalyst Material Groups 204 Variety of Photocatalyst Materials 204 Main Constituent Metal Elements in Photocatalyst Materials 205 Design of Band Structures of Photocatalyst Materials 206 Doped Photocatalysts 206 Valence-Band-Controlled Photocatalysts 208 Solid Solution Photocatalysts 209 Preparation of Photocatalysts 210 Solid-State Reaction Method 211 Flux Method 211 Hydrothermal Synthesis Method/Solvothermal Synthesis Method 211 Polymerized (Polymerizable) Complex Method 211 Precipitation Method 212 Loading of Cocatalysts 212 References 212
11.4.4 11.4.5 11.4.6
Section II Surface Studies and Operando Spectroscopies in Heterogeneous Catalysis 215 12
Toward Precise Understanding of Catalytic Events and Materials Under Working Conditions 217 Atsushi Urakawa
References 220 13
Pressure Gaps in Heterogeneous Catalysis 225 Lars Österlund
13.1 13.2 13.3 13.3.1 13.3.2 13.3.3 13.4
Introduction 225 High-Pressure Studies of Catalysts 226 Adsorption on Solid Surfaces at Low and High Pressures 229 Kinetically Restricted Adsorbate Structures 229 Thermodynamically Driven Reactions on Solid Surfaces 234 Reactions on Supported Nanoparticle Catalysts 244 Conclusions and Outlook 246 Acknowledgments 247 References 247
14
In Situ Transmission Electron Microscopy Observation of Gas/Solid and Liquid/Solid Interfaces 253 Ayako Hashimoto
14.1 14.2
Introduction 253 Observation in Gas and Liquid Phases 254
ix
x
Contents
14.2.1 14.2.2 14.2.3 14.3
Window-Type System 254 Differential Pumping-Type System 256 Other Systems 257 Applications and Outlook 259 References 261
15
Tomography in Catalyst Design 263 Dorota Matras, Jay Pritchard, Antonios Vamvakeros, Simon D.M. Jacques, and Andrew M. Beale
15.1 15.2 15.3 15.4 15.5 15.6 15.7 15.8 15.9
Introduction 263 Imaging with X-Rays 264 Conventional Absorption CT to Study Catalytic Materials 265 X-Ray Diffraction Computed Tomography (XRD-CT) 267 Pair Distribution Function CT 269 Multimodal XANES-CT, XRD-CT, and XRF-CT 270 Atom Probe Tomography 272 Ptychographic X-Ray CT 273 Conclusions 274 References 275
16
Resolving Catalyst Performance at Nanoscale via Fluorescence Microscopy 279 Alexey Kubarev and Maarten Roeffaers
16.1 16.2 16.3
Fluorescence Microscopy as Catalyst Characterization Tool 279 Basics of Fluorescence and Fluorescence Microscopy 280 Strategies to Resolve Catalytic Processes in a Fluorescence Microscope 283 Wide-Field and Confocal Fluorescence Microscopy 284 Super-resolution Fluorescence Microscopy 285 What Can We Learn About Catalysts from (Super-resolution) Fluorescence Microscopy: Case Studies 286 Conclusions and Outlook 291 References 292
16.4 16.5 16.6 16.7
17
In Situ Electron Paramagnetic Resonance Spectroscopy in Catalysis 295 Yiyun Liu and Ryan Wang
17.1 17.2 17.3 17.4 17.4.1 17.4.2 17.5
Introduction 295 Basic Principles of Electron Paramagnetic Resonance (EPR) Experimental Methods and Setup for In Situ cw-EPR 298 Applications of In Situ EPR Spectroscopy 302 Cu-Zeolite Systems 303 Radicals and Radical Ions 305 Conclusions 306 References 307
296
Contents
18
Toward Operando Infrared Spectroscopy of Heterogeneous Catalysts 311 Davide Ferri
18.1 18.2 18.3 18.4 18.5 18.6 18.6.1
Brief Theory on Infrared Spectroscopy 311 Different Modes of IR Measurements 314 Measuring the “Background” 318 Using Probe Molecules to Identify Heterogeneous Sites 320 IR Measurements Under Operando Conditions 325 Case Studies of Operando IR Spectroscopy 328 Selective Catalytic Reduction of NO by NH3 Measured Using Operando Transmission IR 328 Methanation of CO2 Measured Using Operando DRIFTS 329 Selective Oxidation of Alcohols Measured Using Operando ATR-IR 331 Perspective and Outlook 333 References 334
18.6.2 18.6.3 18.7
19
Operando X-Ray Spectroscopies on Catalysts in Action Olga V. Safonova and Maarten Nachtegaal
19.1 19.2 19.3
Fundamentals of X-Ray Spectroscopy 339 X-Ray Absorption Spectroscopy Methods 342 High-Energy-Resolution (Resonant) X-Ray Emission Spectroscopy 347 In Situ and Operando Cells 351 Application of Time-Resolved Methods 353 Limitations and Challenges 356 Concluding Remarks 357 References 358
19.4 19.5 19.6 19.7
339
20
Methodologies to Hunt Active Sites and Active Species 363 Atsushi Urakawa
20.1 20.2 20.3 20.4 20.5
Introduction 363 Modulation Excitation Technique 365 Steady-State Isotopic Transient Kinetic Analysis (SSITKA) 369 Multivariate Analysis 371 Outlook 373 References 373
21
Ultrafast Spectroscopic Techniques in Photocatalysis 377 Chun Hong Mak, Rugeng Liu, and Hsien-Yi Hsu
21.1 21.1.1 21.1.2 21.1.3 21.2 21.2.1
Transient Absorption Spectroscopy 377 Introduction 377 Conventional Heterogeneous Photocatalyst 380 Dye-Sensitized Heterogeneous Photocatalyst 384 Time-Resolved Photoluminescence 386 Introduction 386
xi
xii
Contents
21.2.2 21.3 21.3.1 21.3.2
Applications of TRPL in Heterogeneous Catalysis 387 Time-Resolved Microwave Conductivity 389 Introduction 389 Applications of TRMC in Heterogeneous Catalysis 391 References 393 Volume 2 Preface xv Section III Ab Initio Techniques in Heterogeneous Catalysis 399
22
Quantum Approaches to Predicting Molecular Reactions on Catalytic Surfaces 401 Patrick Sit
23
Density Functional Theory in Heterogeneous Catalysis 405 Patrick Sit and Linghai Zhang
24
Ab Initio Molecular Dynamics in Heterogeneous Catalysis Ye-Fei Li
25
First Principles Simulations of Electrified Interfaces in Electrochemistry 439 Stephen E. Weitzner and Ismaila Dabo
26
Time-Dependent Density Functional Theory for Excited-State Calculations 471 Chi Yung Yam
27
The GW Method for Excited States Calculations 483 Paolo Umari
28
High-Throughput Computational Design of Novel Catalytic Materials 497 Chenxi Guo, Jinfan Chen, and Jianping Xiao
419
Section IV Advancement in Energy and Environmental Catalysis 525 29
Embracing the Energy and Environmental Challenges of the Twenty-First Century Through Heterogeneous Catalysis 527 Yun Hau Ng
Contents
30
Electrochemical Water Splitting 533 Guang Liu, Kamran Dastafkan, and Chuan Zhao
31
New Visible-Light-Responsive Photocatalysts for Water Splitting Based on Mixed Anions 557 Kazuhiko Maeda
32
Electrocatalysts in Polymer Electrolyte Membrane Fuel Cells 571 Stephen M. Lyth and Albert Mufundirwa
33
Conversion of Lignocellulosic Biomass to Biofuels 593 Cristina García-Sancho, Juan A. Cecilia, and Rafael Luque
34
Conversion of Carbohydrates to High Value Products 617 Isao Ogino
35
Enhancing Sustainability Through Heterogeneous Catalytic Conversions at High Pressure 633 Nat Phongprueksathat and Atsushi Urakawa
36
Electro-, Photo-, and Photoelectro-chemical Reduction of CO2 649 Jonathan Albo, Manuel Alvarez-Guerra, and Angel Irabien
37
Photocatalytic Abatement of Emerging Micropollutants in Water and Wastewater 671 Lan Yuan, Zi-Rong Tang, and Yi-Jun Xu
38
Catalytic Abatement of NOx Emissions over the Zeolite Catalysts 685 Runduo Zhang, Peixin Li, and Hao Wang Index 699
xiii
v
Contents
Volume 1 Preface xv Section I
Heterogeneous Catalysts Design and Synthesis 1
1
Evolution of Catalysts Design and Synthesis: From Bulk Metal Catalysts to Fine Wires and Gauzes, and that to Nanoparticle Deposits, Metal Clusters, and Single Atoms 3 Wey Yang Teoh
2
Facets Engineering on Catalysts 21 Jian (Jeffery) Pan
3
Electrochemical Synthesis of Nanostructured Catalytic Thin Films 39 Hoi Ying Chung and Yun Hau Ng
4
Synthesis and Design of Carbon-Supported Highly Dispersed Metal Catalysts 57 Enrique García-Bordejé
5
Metal Cluster-Based Catalysts 79 Vladimir B. Golovko
6
Single-Atom Heterogeneous Catalysts Yaxin Chen, Zhen Ma, and Xingfu Tang
7
Synthesis Strategies for Hierarchical Zeolites 119 Xicheng Jia, Changbum Jo, and Alex C.K. Yip
8
Design of Molecular Heterogeneous Catalysts with Metal–Organic Frameworks 147 Marco Ranocchiari
103
vi
Contents
9
Hierarchical and Anisotropic Nanostructured Catalysts 161 Hamidreza Arandiyan, Yuan Wang, Christopher M.A. Parlett, and Adam Lee
10
Flame Synthesis of Simple and Multielemental Oxide Catalysts 183 Wey Yang Teoh
11
Band Engineering of Semiconductors Toward Visible-Light-Responsive Photocatalysts 203 Akihide Iwase Section II Surface Studies and Operando Spectroscopies in Heterogeneous Catalysis 215
12
Toward Precise Understanding of Catalytic Events and Materials Under Working Conditions 217 Atsushi Urakawa
13
Pressure Gaps in Heterogeneous Catalysis 225 Lars Österlund
14
In Situ Transmission Electron Microscopy Observation of Gas/Solid and Liquid/Solid Interfaces 253 Ayako Hashimoto
15
Tomography in Catalyst Design 263 Dorota Matras, Jay Pritchard, Antonios Vamvakeros, Simon D.M. Jacques, and Andrew M. Beale
16
Resolving Catalyst Performance at Nanoscale via Fluorescence Microscopy 279 Alexey Kubarev and Maarten Roeffaers
17
In Situ Electron Paramagnetic Resonance Spectroscopy in Catalysis 295 Yiyun Liu and Ryan Wang
18
Toward Operando Infrared Spectroscopy of Heterogeneous Catalysts 311 Davide Ferri
19
Operando X-Ray Spectroscopies on Catalysts in Action Olga V. Safonova and Maarten Nachtegaal
20
Methodologies to Hunt Active Sites and Active Species 363 Atsushi Urakawa
339
Contents
21
Ultrafast Spectroscopic Techniques in Photocatalysis 377 Chun Hong Mak, Rugeng Liu, and Hsien-Yi Hsu Volume 2 Preface xv Section III Ab Initio Techniques in Heterogeneous Catalysis 399
22
Quantum Approaches to Predicting Molecular Reactions on Catalytic Surfaces 401 Patrick Sit
22.1 22.2 22.3
Heterogeneous Catalysis and Computer Simulations 401 Theory of Quantum Mechanics 403 Quantum Mechanical Techniques in the Study of Heterogeneous Catalysis 403 References 404
23
Density Functional Theory in Heterogeneous Catalysis 405 Patrick Sit and Linghai Zhang
23.1 23.2 23.2.1 23.2.2 23.2.3 23.2.4 23.3 23.3.1 23.3.2 23.4 23.5
Introduction 405 Basics of Density Functional Theory Calculations 406 Born–Oppenheimer Approximation 406 The Hohenberg–Kohn Theorems and the Kohn–Sham Approach Basis Sets 409 Forces on the Ions 409 The Search for Better Energy Functionals 410 Energy Functional Development 410 Other Corrections and Approaches 411 DFT Applications in Heterogeneous Catalysis 412 Conclusions and Perspective 414 References 416
24
Ab Initio Molecular Dynamics in Heterogeneous Catalysis Ye-Fei Li
24.1 24.2 24.2.1 24.2.2 24.2.3 24.3 24.4 24.5 24.5.1 24.5.2
Introduction 419 Basic Algorithm of Molecular Dynamics 420 Verlet Algorithm 421 Velocity Verlet Algorithm 421 Conserved Quantity 422 Molecular Dynamics in Canonical Ensembles 422 Transition State Theory 423 Free Energy Calculations 425 Thermodynamic Integration and Constrained MD 425 Umbrella Sampling 428
419
406
vii
viii
Contents
24.5.3 24.6 24.7 24.8
Metadynamics 430 Accelerating MD Simulations by Neural Network 432 Examples for MD Simulations 433 Conclusions 437 References 437
25
First Principles Simulations of Electrified Interfaces in Electrochemistry 439 Stephen E. Weitzner and Ismaila Dabo
25.1 25.2 25.2.1 25.2.2 25.3 25.3.1 25.3.2 25.3.3 25.3.4 25.4 25.4.1 25.4.2 25.4.3
Toward Stable and High-Performance Electrocatalysts 439 A Brief Thermodynamic Detour 441 The Fundamental Relation 441 Alternative Forms of the Fundamental Relation 445 Statistical Mechanics 447 Preliminaries 447 The Electrochemical Canonical (N, V , T, Φ) Ensemble 448 The Electrochemical Grand Canonical (𝜇, V , T, Φ) Ensemble 451 Computational Methods 452 The Quantum-Continuum Approach 455 Overview 455 Electric Double Layer (EDL) Models 456 Example: Silver Monolayer Stripping on Au(100) 461 Acknowledgments 468 References 469
26
Time-Dependent Density Functional Theory for Excited-State Calculations 471 Chi Yung Yam
26.1 26.2 26.3 26.4 26.5
Introduction 471 Theoretical Foundation of TDDFT 473 Linear Response Theory 474 Real-Time TDDFT 476 Nonadiabatic Mixed Quantum/Classical Dynamics 477 References 479
27
The GW Method for Excited States Calculations 483 Paolo Umari
27.1 27.2 27.3 27.4 27.5 27.6 27.7 27.8
Introduction 483 Excitations in Many-Electron Systems 485 Green’s Functions 486 Many-Body Perturbation Theory 487 GW in Practice 489 The Bethe–Salpeter Equation 491 BSE in Practice 493 Conclusions and Perspectives 493 References 494
Contents
28
High-Throughput Computational Design of Novel Catalytic Materials 497 Chenxi Guo, Jinfan Chen, and Jianping Xiao
28.1 28.2 28.2.1 28.2.2 28.2.3 28.2.4 28.2.5 28.2.6 28.3 28.3.1 28.3.2 28.3.3
Introduction 497 The Framework of Computational Catalyst Design 498 Elementary Reactions and Material Selection 498 The Scaling Relation and the Reaction Energy 502 The BEP Relation and the Activation Barrier 503 The Activity Volcano Curve 507 Explicit Kinetic Simulations Based on DFT Calculations 507 Data Mining and Machine Learning in Catalyst Design 509 Examples for Rational Catalyst Design 509 Synthesis of Higher Alcohols from Syngas on Alloys 509 HT Screening for Hydrogen Evolution Reactions (HERs) 511 Rational Design for CO Oxidation on Multicomponent Alloy Surfaces 516 Adsorbate–Adsorbate Interactions for CO Methanation 516 RhAu Alloy Nanoparticles for NO Decomposition by Machine Learning 518 Summary and Prospects of HT Catalytic Material Design 520 References 520
28.3.4 28.3.5 28.4
Section IV Advancement in Energy and Environmental Catalysis 525 29
Embracing the Energy and Environmental Challenges of the Twenty-First Century Through Heterogeneous Catalysis 527 Yun Hau Ng
References 530 30
Electrochemical Water Splitting 533 Guang Liu, Kamran Dastafkan, and Chuan Zhao
30.1 30.2 30.2.1 30.2.1.1 30.2.1.2 30.2.1.3 30.2.1.4 30.2.2 30.2.3 30.3 30.3.1 30.3.1.1 30.3.1.2 30.3.1.3 30.3.1.4
Fundamentals of Electrochemical Water Splitting 533 Technological and Practical Considerations 535 Liquid Electrolyte Water Electrolysis 535 Overall Water Electrolysis (OWE) 537 Doubled Water Electrolysis (DWE) 538 Hybrid Water Electrolysis (HWE) 538 Tandem Water Electrolysis (TWE) 539 Polymer Electrolyte Membrane Water Electrolysis 539 Solid Oxide Electrolyte Water Electrolysis 540 Electrocatalyst Materials in Liquid Electrolyte Water Splitting 541 Oxygen Evolution Reaction Electrocatalysts 541 Metal Oxides 541 Metal Chalcogenides 545 Metal Pnictides, Carbides, and Borides 546 Metal–Organic Frameworks and Related Materials 546
ix
x
Contents
30.3.2 30.3.2.1 30.3.2.2 30.3.2.3 30.4
Hydrogen Evolution Reaction Electrocatalysts 547 Non-noble Metals and Noble Metal–Free Alloys 547 Non-precious Metal Composites 549 Metal-Free Electrocatalysts 551 Conclusions and Outlook 551 References 552
31
New Visible-Light-Responsive Photocatalysts for Water Splitting Based on Mixed Anions 557 Kazuhiko Maeda
31.1 31.2
Introduction 557 New Doped Rutile TiO2 Photocatalysts for Efficient Water Oxidation 559 Unprecedented Narrow-Gap Oxyfluoride 564 Conclusion and Future Perspective 567 References 568
31.3 31.4
32
Electrocatalysts in Polymer Electrolyte Membrane Fuel Cells 571 Stephen M. Lyth and Albert Mufundirwa
32.1 32.2 32.3 32.4 32.5 32.6 32.7 32.8 32.9 32.10 32.11
Introduction 571 Platinum Electrocatalysts 574 Voltammetry 576 Cyclic Voltammetry 577 Linear Sweep Voltammetry 578 Electron Transfer Number 580 Durability Measurements in a Three-Electrode Cell 581 Membrane Electrode Assembly (MEA) Fabrication 582 MEA Measurements 583 Recent Electrocatalyst Research 584 Future Perspectives 587 Acknowledgments 588 References 588
33
Conversion of Lignocellulosic Biomass to Biofuels 593 Cristina García-Sancho, Juan A. Cecilia, and Rafael Luque
33.1 33.2 33.3 33.3.1 33.3.2 33.3.3 33.3.4 33.3.5 33.4
Introduction 593 Lignocellulosic Biomass: Composition and Resources 595 Biofuel Production from Lignocellulosic Biomass 596 Ethoxymethylfurfural (EMF) 597 Ethyl Levulinate (EL) 602 2,5-Dimethylfuran (DMF) 604 2-Methylfuran (MF) 606 γ-Valerolactone (GVL) 607 Outlook and Conclusions 610 References 611
Contents
34
Conversion of Carbohydrates to High Value Products 617 Isao Ogino
34.1 34.2
Introduction 617 Overview of Strategy for Catalyst Development and Routes for Conversion of Carbohydrates 619 Synthesis of Value-Added Chemicals from Carbohydrates 623 Dihydrolevoglucosenone 623 1,6-Hexanediol 623 Furandicarboxylic Acid (FDCA) 623 Terephthalic Acid 626 Lactic Acid 627 Lactide 628 Perspective 630 References 630
34.3 34.3.1 34.3.2 34.3.3 34.3.4 34.3.5 34.3.6 34.4
35
Enhancing Sustainability Through Heterogeneous Catalytic Conversions at High Pressure 633 Nat Phongprueksathat and Atsushi Urakawa
35.1 35.1.1 35.1.2 35.1.2.1 35.1.2.2 35.1.3 35.1.3.1 35.1.3.2 35.1.4 35.2
Importance of High-Pressure Reaction Condition 633 Chemical Equilibrium (One Phase) 634 Phase Behavior (Multiphase) 634 Phase Separation 634 Supercritical State 635 Mass Transfer and Kinetics 635 Molecular Diffusion 635 Multiphase Reaction 636 Process Efficiency and Economy 636 State-of-the-Art Application of High Pressure in Heterogeneous Catalysis 637 Boosting CO2 Conversion and Surpassing One-Phase Chemical Equilibrium by In situ Phase Separation at High-Pressure Reaction Condition 637 Exploitation of Supercritical Fluid Properties for Catalytic Reactions 639 A Greener High-Pressure System Using Microchannel Reactor 641 Surpassing Chemical Equilibrium by In situ Product Separation Using a High-Pressure Membrane Reactor 643 Concluding Remark 645 References 645
35.2.1
35.2.2 35.2.3 35.2.4 35.3
36
Electro-, Photo-, and Photoelectro-chemical Reduction of CO2 649 Jonathan Albo, Manuel Alvarez-Guerra, and Angel Irabien
36.1 36.2 36.2.1
Introduction 649 Fundamentals 651 Redox Processes 651
xi
xii
Contents
36.2.2 36.2.3 36.3 36.3.1 36.3.1.1 36.3.1.2 36.3.1.3 36.3.2 36.3.2.1 36.3.2.2 36.3.2.3 36.3.3 36.3.3.1 36.3.3.2 36.3.3.3 36.4
Assessment of Reaction Performance: Figures of Merit 652 Role of Heterogeneous Catalysts 653 Innovative Technologies for CO2 Reduction 654 Electrochemical Reduction 654 How Does the Technology Work? 654 Key Factors Influencing Reaction Performance 655 Main Challenges 657 Photochemical Reduction 658 How Does the Technology Work? 658 Key Factors 660 Main Challenges 661 Photoelectrochemical Reduction 662 How Does the Technology Work? 662 Key Factors 663 Main Challenges 665 Concluding Remarks 666 Acknowledgments 666 References 666
37
Photocatalytic Abatement of Emerging Micropollutants in Water and Wastewater 671 Lan Yuan, Zi-Rong Tang, and Yi-Jun Xu
37.1 37.2
Introduction 671 Main Processes for Photocatalytic Abatement of Micropollutants in Water and Wastewater 672 Advancements in Photocatalysts for Photocatalytic Abatement of Micropollutants in Water and Wastewater 674 Photocatalysts Components Optimization 674 Semiconductor Doping 674 Surface Sensitization 676 Metal Deposition 676 Carbon Materials Combination 677 Combining Semiconductors 677 Photocatalysts Configuration Optimization 678 Freestanding Particulate 678 Film, Immobilized, and Aerogel-Based Catalysts 679 Reaction System Optimization 680 Reaction Conditions 680 Solar Reactors 680 Future Challenges and Prospects 681 Acknowledgments 682 References 682
37.3 37.3.1 37.3.1.1 37.3.1.2 37.3.1.3 37.3.1.4 37.3.1.5 37.3.2 37.3.2.1 37.3.2.2 37.4 37.4.1 37.4.2 37.5
38
Catalytic Abatement of NOx Emissions over the Zeolite Catalysts 685 Runduo Zhang, Peixin Li, and Hao Wang
38.1
Zeolite Catalysts with Different Topologies 686
Contents
38.2 38.2.1 38.2.2 38.2.3 38.2.4 38.3 38.4
Essential Nature of Novel Cu–CHA catalyst 688 Shape Selectivity 688 Cation Location 689 Copper Status 691 CHA-Type Silicoaluminophosphate 691 SCR Reaction Mechanism 692 Conclusions and Perspectives 694 References 695 Index 699
xiii
xv
Preface Heterogeneous catalyst to a chemical reaction is akin to a microscope to microbiology or a sail to a yacht. It is a necessary tool that helps to speed up reactions and at the same time steers the reactions in such a way that maximum selectivity of the desirable product can be attained. In that sense, heterogeneous catalysts will always be relevant as far as chemical reactions are of interest, whether at large industrial scales (e.g. commodity chemicals, petrochemical refineries), small scales (e.g. devices with catalytic functions such as automobile catalytic converters), or even microscales (e.g. microfluidic devices, catalytic nanomachines). The dedication of scientists and engineers working in the field of heterogeneous catalysis throughout the twentieth century was instrumental in solving of some of the most important problems facing humanity, including the nitrogen food chain issue (the so-called Nitrogen problem), production of high-quality automobile fuels (e.g. gasoline and diesel), abatement of noxious airborne pollutants, and the manufacturing of methanol as well as other building block chemicals. Entering the twenty-first century, the two immediate and overarching challenges in heterogeneous catalysis are (i) to address the issues related to global warming and climate change and (ii) to align with the United Nations sustainable development goals. Catalytic reactions such as water splitting, reduction of carbon dioxide, waste biomass conversion, removal of emerging aqueous micropollutants, and the abatement of NOx that are focused on enhancing renewable energy security and environmental sustainability will take the center stage against a backdrop of swelling population and increasing urbanization. In adapting to these grand challenges, new and sophisticated emerging techniques in heterogeneous catalysis are continuously being developed to overcome the various limitations in catalyst design and to understand the mechanism of the molecular reactions occurring on the catalytic surface (information that can feed back to the catalyst design). At the same time, catalysts with different modes of activation are increasingly being appreciated, which besides the conventional thermal catalysts now include electrocatalysts and photocatalysts and their underlying physics. For newcomers entering the field, acquiring such vast amount of knowledge, although essential, can be overwhelming. That is not to mention the tenacity in mastering the basic fundamentals in heterogeneous catalysis, itself a century worth of knowledge, prior to the appreciation of these state-of-the-art advancements. With this in hindsight, the book is geared
xvi
Preface
toward attracting and assisting beginners who wish for a head start and quick overview on some of the most important emerging tools for catalyst design, techniques for operando/in situ characterization and ab initio computation, as well as a glimpse of the advancements in heterogeneous catalysis toward some of the grand challenges. Undergraduates with some prior exposure to reaction engineering/heterogeneous catalysis and analytical chemistry/spectroscopy or early postgraduates pursuing research on heterogeneous catalysis but only with some primitive background of the field shall find the book useful. Because the aim is to bridge the gap between amateur readers and expert knowledge, each chapter provides a brief description of the required basic fundamentals that lead to the appreciation of state-of-the-art advancements. It should be mentioned that the Contributors of the different chapters in the book are themselves among the most promising Emerging and Pioneering Researchers in the field of heterogeneous catalysis. We capitalized on that point in our book design to allow each Contributor to articulate the advancement of his/her own technique in a semitutorial manner that can be appreciated by the target readers. We strive to preserve a delicate balance between readability and articulating the complexity of the advanced techniques. In that sense, we present the content in a less mathematical (in a semiquantitative form, as much as we could) but comprehensible setting as the first step to inculcate interest and inspiration among beginners. With some patience, self-learning is highly possible, following which readers should have the ability to pursue more quantitative references of specific techniques. With the heightened expectation of “cross skills” among the new generation of catalyst researchers, this book shall come in handy for readers to gain appreciation on some of the most advanced techniques before deciding to specialize in some of them. In fact, we hope that the book would serve as a platform to inspire readers to potentially develop their own original or hybrid techniques in a wider effort to tackling the grand challenges using heterogeneous catalysts. Finally, we take the opportunity to thank Emerging and Pioneering Researchers who have contributed to this book, its vision and purpose. It has been a massive effort that took us more than three years to put together this book, and we thank the Contributors for their patience. Wey Yang Teoh University of Malaya, Malaysia 23 November 2020 Atsushi Urakawa Delft University of Technology, The Netherlands 23 November 2020 Yun Hau Ng City University of Hong Kong, S.A.R. 23 November 2020 Patrick Sit City University of Hong Kong, S.A.R. 23 November 2020
1
Section I Heterogeneous Catalysts Design and Synthesis
3
1 Evolution of Catalysts Design and Synthesis: From Bulk Metal Catalysts to Fine Wires and Gauzes, and that to Nanoparticle Deposits, Metal Clusters, and Single Atoms Wey Yang Teoh 1,2 1 University of Malaya, Centre for Separation Science and Technology, Department of Chemical Engineering, 50603 Kuala Lumpur, Malaysia 2 The University of New South Wales, School of Chemical Engineering, Sydney 2052, Australia
1.1 The Cradle of Modern Heterogeneous Catalysts The modern discovery of heterogeneous catalysts stretches as far back as 1800 when Joseph Priestley and Martinus van Marum reported the dehydrogenation of alcohol over a heated metal catalyst, although not too much was thought about the role of the metal catalyst at that time except as a heating source. Then in 1813, Louis Jacques Thénard of École Polytechnique in Paris discovered the decomposition of ammonia to nitrogen and hydrogen over “red-hot metals” and recognized that the phenomenon was due to some catalytic reaction [1, 2]. The concept was followed up by Humphry Davy and Michael Faraday at the Royal Institution of London who, in 1817, reported the flameless catalytic combustion of coal gas and air over heated platinum wire producing bright white ignition. Their results were reproducible when using palladium, but not on copper, silver, iron, gold, and zinc [1, 3]. These experiments made clear that there was some form of catalytic role associated with the different metals. The discovery soon became the basis for the invention of the coal mine safety lamp, also known as the Davy lamp – although mysteriously but rather practically, the use of inefficient steel iron rather than platinum gauze became the standard for Davy lamps. At around the same time, Thénard and Pierre Dulong found that the catalytic ammonia decomposition rates decrease in the following order: iron, copper, silver, gold, and platinum, marking the first recognition of the kinetics of different metal catalysts. The importance of catalytic surface area, as we now know to be one of the most important governing factors in heterogeneous catalysis, was discovered by Edmund Davy (cousin to Humphry Davy) at the University College Cork in the 1820s, who found that finely divided platinum could catalyze the oxidation of alcohol as well as the oxidation of hydrogen at room temperature [4]. In 1831, a little-known gentleman by the name of Peregrine Phillips, Jr., patented sulfuric acid production by oxidizing sulfur dioxide in air over platinum packed in porcelain tubes heated to “strong yellow heat”. The resultant sulfur trioxide forms sulfuric acid fume upon contact with water, hence earning Heterogeneous Catalysts: Advanced Design, Characterization and Applications, First Edition. Edited by Wey Yang Teoh, Atsushi Urakawa, Yun Hau Ng, and Patrick Sit. © 2021 WILEY-VCH GmbH. Published 2021 by WILEY-VCH GmbH.
4
1 Evolution of Catalysts Design and Synthesis
its name as the Contact Process [5]. Ironically, despite the high importance of this catalytic process, not much is known about Phillips except that he was son of a tailor and was born in Bristol [1]. A large-scale manufacturing of sulfuric acid using the Contact Process and platinum catalyst was realized many years later in 1875 by Rudolph Messel, a German-born and naturalized English industrial chemist. Messel himself was very much involved in the studies of the kinetics as well as the problematic poisoning of platinum catalysts by arsenic trioxide. In 1913, BASF was granted patents on a new catalyst based on the more versatile supported vanadium pentoxide and alkali oxide on porous silica [6, 7]. The first manufacturing plant based on this new catalyst was commissioned in 1915. Improvement in the activity of the supported vanadium pentoxide catalyst through the addition of potassium sulfate promoter was invented in Germany and the United States between 1916 and 1919. It was only in 1988 that Haldor Topsoe and Anders Nielsen revealed that the addition of cesium or rubidium promoter, rather than potassium, was more efficient in enhancing the activity of sulfur dioxide oxidation. With a typical lifetime of up to 10 years, the industrial catalyst composition for the Contact Process has been largely unchanged even to this day [8]. Going back to 1838, just a few years after the discovery of the Contact Process, Frédéric Kuhlmann discovered the production of nitric acid from the oxidation of ammonia in air over platinum sponge at 300 ∘ C and filed a patent on this [9]. Based on the discovery, he later founded the Etablissements Kuhlmann company, which still exists to this day as part of the Pechiney SA. Despite being an important chemical commodity for the use in fertilizers and explosives manufacturing, the interest in Kuhlmann reaction was not immediately of interest since Chile saltpetre (a naturally occurring mineral of alkali metal nitrate precursor found at the Atacama desert repository) was widely available. In his vision, Kuhlmann stated that “If in fact the transformation of ammonia to nitric acid in the presence of platinum and air is not economical, the time may come when this process will constitute a profitable industry.” Indeed, the Kuhlmann reaction picked up interest toward the end of the century as part of the solution to “The Nitrogen Problem.” In 1901 and building on Kuhlmann’s earlier findings, Wilhelm Ostwald of the University of Leipzig investigated the production of nitric acid using supported platinum on asbestos before moving to coiled platinum strips that gave higher conversion [9]. A large-scale nitric acid manufacturing plant went into operation at Gerthe in 1908 with an output of 3 tons nitric acid per day using 50 g of corrugated platinum catalyst of 2 cm wide. Given the short catalyst lifetime of no more than six weeks, it was soon realized to be a costly operation. To tackle the problem, Karl Kaiser of Technische Hochschule, Charlottenburg, developed the platinum gauze catalyst in 1909, consisting of 0.06 mm diameter wires woven to 1050 mesh/cm2 , that gave a higher surface-to-bulk ratio and uninterrupted production of nitric acid of up to six months [9]. But because the source of ammonia at that time was derived from gas works liquors containing impurities such as arsenic and sulfur that deactivate the platinum catalyst, the really large industrial-scale production was only possible after the implementation of the Haber–Bosch process that provided clean ammonia. The present-day nitric acid catalyst is based on rhodium–platinum gauze (5–10% Rh) [10].
1.2 The Game Changer: High-Pressure Catalytic Reactions
Further advancement in the design of bulk metal catalysts was evident from the work of Murray Raney on the synthesis of skeletal nickel, which was granted US patent in 1925 [11]. The Raney catalyst was prepared by first forming a Ni–Al alloy and ground into small particles, followed by the selective leaching of Al in caustic brine (such as NaOH) to yield the skeletal structure. The resultant Raney catalyst is composed of finely divided nickel so fine that it is pyrophoric and hence requiring storage under deionized water. Initially, the Raney Ni was used as an industrial catalyst for the hydrogenation of vegetable oil (to make butter substitutes) but later proved to be useful for a range of other hydrogenation reactions. Other forms of Raney catalysts including those of metallic cobalt, copper, palladium, silver, and ruthenium were later developed and found applications in methanol synthesis, conversion of furfural into furfural alcohol, and the hydrogenation of acrolein to allyl alcohol, among others [12, 13].
1.2 The Game Changer: High-Pressure Catalytic Reactions The implementation of high-pressure reactor technologies pioneered by Robert Le Rossignol (assistant to Fritz Haber) [14] and later by Carl Bosch [15] was one of the most important milestones in the advancement of heterogeneous catalysis. Their breakthroughs enabled a series of high-pressure catalytic reactions that include the ammonia synthesis and methanol synthesis, which to this day rank among the most important industrial catalytic reactions. High-pressure conditions are particularly useful in overcoming reaction dilemma that under ambient pressure could obtain high selectivity but at extremely sluggish rates and vice versa at high temperatures. By carrying out the same reaction under high-pressure conditions, one can shift the equilibrium line to higher selectivity even at high temperatures, thus allowing high yield of the desired product. Chapter 35 is devoted to this topic. Haber in one of his earlier efforts in synthesizing ammonia by N2 fixation (through reaction with H2 ) under ambient pressure could only obtain 0.005% yield when using iron catalysts at 1000 ∘ C [16]. A year later, in 1906, Walther Nernst at the University of Berlin reported favorable conversion at 1000 ∘ C when using iron catalysts in a ceramic apparatus that allowed him to perform the reaction at 75 bar. Unfortunately, the reactor and the extreme condition were far too impractical for industrial-scale implementation. Haber, who became professor at the Karlsruhe Technische Hochschule, used a steel-based reactor but this time working with Le Rossignol (who actually built the bench-scale high-pressure reactor, equipped with a high-pressure and high-temperature valve, now known as the Le Rossignol valve). With the new reactor, they were able to screen a number of catalytic materials ranging from iron, chromium, nickel, manganese, osmium, and uranium (as uranium carbide) at 200 atm and in excess of 700 ∘ C. Osmium and uranium catalysts were found to be active, with the former achieving a 6% conversion. Realizing that the N2 fixation reaction is limited by its kinetics rather than equilibrium, Haber further developed the feed
5
6
1 Evolution of Catalysts Design and Synthesis
recycle system for which he received a patent [17]. BASF AG acquired Haber’s patents on ammonia synthesis and, interestingly, also the total world supply of osmium at that time (100 kg) [2]! The amount of osmium was estimated to be capable of producing 750 tons of ammonia per year, although that amount would still be insufficient to cope with the total ammonia demand. Alwin Mittasch, who was tasked by BASF to look for more commercially feasible alternatives, together with his colleague, George Stern, screened more than 2500 catalysts and found that a magnetite (Fe3 O4 ) sample taken from a Swedish mine gave very high yield. Mittasch soon realized that the presence of impurities in the sample was critical before arriving at an optimized synthetic Fe3 O4 catalysts promoted with 2.5–4% Al2 O3 , 0.5–1.2% K2 O, 2.0–3.5% CaO, and 0.0–1.0% MgO (together with 0.2–0.5% Si present as impurity in the metal) [18, 19]. The catalyst formulation was so robust that it has not significantly changed until now. Meanwhile, the major challenge in high-pressure reactor design shall be described. The diffusion of hydrogen through the standard carbon steel reactor under high pressure and temperature can result in the decarbonization and formation of brittle iron hydride, thus reducing the pressure rating of the reactor [20]. As such, using such reactors would limit the standard operation of ammonia synthesis (200 atm, 500 ∘ C) to a mere 80 hours [17]. The groundbreaking work by Bosch arrived in 1909 when he, after observing Le Rossignol’s reactor design, came up with an ingenious design of using a concentric tube consisting of an inner soft (low-carbon) steel tube encased in a pressure-bearing carbon steel outer jacket [16]. Narrow grooves were machined on the outer wall of the inner tube to create small pockets in between the tube and the jacket. During operation, high-pressure and high-temperature hydrogen from the reaction in the inner tube would diffuse out through the soft steel into the pockets while experiencing rapid loss of pressure and temperature. Small holes were drilled on the outer jacket to allow continuous release of the diffused hydrogen from the pockets [21]. With the catalyst formulation and reactor design in place, a pilot test on a 4 m reactor was carried out in 1911, subsequently leading to the commissioning of a full-scale manufacturing plant at Oppau consisting of an 8 m high reactor to produce 20 tons of ammonia per day [16], which is known now as the Haber–Bosch process. The triumph in ammonia synthesis in Germany caught on with the industrial production of methanol (from syngas). As early as 1921, George Patas in the neighboring France patented a high-pressure process for the synthesis of methanol using copper as well as nickel, silver, and iron catalysts [22]. BASF has again sought the help of Mittasch to search for suitable catalysts. This resulted in the discovery of zinc chromite (Cr2 O3 –ZnO) catalyst that was used in its industrial methanol production plant at Leuna in 1923. The catalytic reactor operated at 300 atm and 300–400 ∘ C [23, 24]. Although iron-containing (as well as nickel) catalysts also show methanol synthesis activity, they were later excluded from the catalysts screening due to the formation of iron carbonyl (from the reaction with carbon monoxide in the syngas) during the reaction that further decomposes to metallic iron (or iron carbide) [25]. Instead of catalyzing
1.2 The Game Changer: High-Pressure Catalytic Reactions
the methanol synthesis, these iron phases are more efficient at producing hydrocarbons (the basis for Fischer–Tropsch synthesis!), which is a more exothermic reaction. For the same reason, high-pressure steel reactors were lined with copper, silver, or aluminum [26]. In 1947, Polish chemist Eugeniusz Błasiak patented a highly active methanol synthesis catalyst containing mixed copper, zinc, and aluminum prepared by coprecipitation [27]. Using the same catalyst, the Imperial Chemical Industries (ICI) developed a low-pressure methanol synthesis process that only required operation at 30–120 atm with sufficient kinetics at 200–300 ∘ C and selectivity of over 99.5%. The process along with the upstream high-pressure steam reformer was patented in 1965 [28], followed closely by another landmark patent on the synthesis of mixed oxide of copper–zinc catalyst with promoter element from groups II–IV [29]. The catalytic process and catalyst formulation have remained largely unchanged. Using Bosch’s high-pressure reactor, Franz Fischer and Hans Tropsch of Kaiser Wilhelm Institute for Coal Research (now known as Max Planck Institute of Coal Research) found the formation of high-molecular-weight hydrocarbons when using iron filings at 100 atm and 400 ∘ C. As mentioned earlier, this was an undesirable reaction during the methanol synthesis, but Fischer understood the importance of this reaction. While continuing to work on this direction, they routinely assessed a range of metal oxides, hydroxides, and carbonates and in 1926 reported that reduced iron and cobalt catalysts yielded gasoline fuels from coal-derived syngas [30, 31]. The reaction is known as the Fischer–Tropsch synthesis (FTS), which in 1935 marked the first FTS plant commissioned by Ruhrchemie using the cobalt catalyst. By 1938, there were nine such facilities within Germany with a manufacturing capacity of 600 000 tons/annum. The cobalt catalyst (100 Co/100 SiO2 /18 ThO2 ) used by Ruhrchemie was developed by Fischer with Meyer and later with Koch by rapidly coprecipitating hot solutions of cobalt and thorium nitrate on SiO2 (Kieselguhr diatomaceous earth) suspended in an ammonia-containing solution [32, 33]. The irreducible thorium oxide restricts the crystallization of the cobalt metal to maintain a high dispersion. The slightly radioactive thoria has been replaced by zirconia, titania, or manganese oxide in the present-day catalysts. While cobalt is known to produce a large fraction of diesel and paraffin wax, the iron catalyst results in higher content of short-chain olefins when carried out at high reaction temperatures (∼340 ∘ C) or paraffin wax at much lower temperatures. As the reaction proceeds, the iron metal is gradually converted into iron carbide, which is an even more active phase [24, 34]. Compared with crude oil–derived fuels, the FTS-derived diesel and gasolines are characterized by their exceptionally high cetane and octane ratings due to the high yields of straight-chain paraffins for cobalt-derived diesel and olefins/isomers in the iron-derived gasoline, respectively. Although nickel and ruthenium catalysts are also active in FTS, they are rarely used as stand-alone catalysts. Nickel, which forms carbonyl and decomposes to the metallic phase (like iron), has a high tendency to form methane instead of liquid fuels. Ruthenium, which is the most active FTS catalyst, is far more expensive than cobalt and iron to justify its bulk
7
8
1 Evolution of Catalysts Design and Synthesis
usage except as a promoter to cobalt catalysts. Incipient wetness impregnation is by far the most common technique for the synthesis of FTS catalysts [35].
1.3 Catalytic Cracking and Porous Catalysts One of the earliest applications of heterogeneous catalysts in the modern petrochemical industries (crude oil refineries) can perhaps be traced to the catalytic cracking process. In the early 1920s, French engineer Eugene Jules Houdry, E. A. Prudhomme (the pharmacist who discovered the reaction) and their team developed the catalytic lignite-to-gasoline process, whereby lignite was first pyrolyzed to high-boiling-point liquid hydrocarbons, followed by vaporization and catalytic conversion to the gasoline fractions [36]. The latter step is similar to noncatalytic, high-temperature, and high-pressure cracking of the heavier fractions of the crude oil to produce (low octane rating) gasoline developed by Standard Oil Company in the United States a few years earlier. Efforts were made to boost the octane rating of the synthetic gasoline including trial using aluminum chloride as the cracking catalyst but was found to be economically unfeasible. Thomas Midgley and Charles Kettering of General Motors patented the addition of tetraethyl lead to gasoline to improve its octane rating substantially, which was rather successful commercially but was banned worldwide many years later due to the release of toxic exhaust fumes [37]. Houdry discovered a more environmentally benign solution, that is, use of Fuller’s earth, a naturally occurring aluminosilicate layered clay, as a cracking catalyst to produce extremely high-quality gasoline from heavy crude. Despite not having found much success in France, where the process was deemed not commercially viable, Houdry brought his catalytic cracking process to the United States in the 1930s for further development with Sonoco Vacuum Oil Company (later Mobil Oil Corporation and now ExxonMobil) and adapting the technology to the petrochemical processing. Upon overcoming various reactor engineering challenges to cope with the rapid catalyst coking during the cracking reaction, the Houdry process became a phenomenal success that revolutionized the petrochemical industry. His inventions paved the way for the development of the modern fluidized catalytic cracking (FCC) process, where catalysts were fluidized for continuous looping between the catalytic cracking reactor and adjacent regenerator unit (to remove coke by air oxidation). The Houdry process was so successful that the production of synthetic silica–alumina and magnesia–silica catalysts was commenced in the 1940s to meet the needs for catalytic cracking reaction [38]. In fact, the silica–alumina catalyst is still used to this day in industrial FCC, but in the form of synthetic zeolites, which have a much higher surface area than the clay minerals. Synthetic zeolites, which constitute crystalline microporous (0.3–2.0 nm pores) aluminosilicates, have been actively developed since the late 1950s by the Union Carbide and Mobil Oil Corporation, resulting in the discovery of zeolites A (Linde Type A) and X (Linde Type X) in 1959 [39], zeolite Y (Linde Type Y) in 1964 [40], and ZSM-5 in 1972 [41, 42]. These landmark catalysts continue to find important applications not only in FCC but also in the isomerization
1.3 Catalytic Cracking and Porous Catalysts
of hydrocarbons, synthesis of specialty chemicals, methanol-to-hydrocarbon conversions, and catalytic deNOx , with a great deal of advancement achieved in the last decade in the conversion of biomass, among many others. Excellent accounts on the fundamentals as well as the state-of-the-art progress in some of these topics are highlighted in Chapter 33 (on the conversion of lignocellulose to biofuels), Chapter 34 (on the conversion of carbohydrates to high-value products), and Chapter 38 (on the abatement of NOx ). In fact, the discovery of new zeolites has been thriving since the 1980s, with a unique set of material compositions, frameworks, and pore dimensions being discovered annually. A large database of zeolites is maintained by the International Zeolite Association since 1977 through the Atlas of Zeolite Structure Types [43]. While silicate and aluminosilicate zeolites dominate a large extent of the database, other zeolites based on aluminophosphates, metallosilicates, germanosilicates, aluminoborates, and so on also exist. Among them, some of the most widely used zeolites in industrial catalysis besides zeolite Y and ZSM-5 include zeolite X, MCM-22 (Mobil Composition of Matter No. 22), MCM-49, SAPO-34, Beta zeolite, and SSZ-13. The most common approach to the synthesis of zeolites involves interfacing sol–gel chemistry with organic structure-directing agents (SDAs) as soft templates. In a classical sol–gel process, precursors especially those of alkoxides such as tetraethyl orthosilicate (TEOS) are first hydrolyzed to form alkoxysilanols and/or orthosilicic acid. Subsequent cross-linking reaction through the dehydration of the hydroxyl moieties results in the formation of nuclei, and further polymerization yields amorphous silica particles that appear either as sol (well-dispersed particles in solution medium) or gel (continuous network formed by particles throughout the solution medium). The physical sizes of these amorphous particles are strongly influenced by concentration, pH, and temperature of the reaction medium. In the presence of SDAs, typically amines or quaternary ammonium surfactants but in some cases inorganic ions, the cationic head of SDAs will bind strongly to the silicate anions. Under such situations, there exist concerted interactions between (i) the silicate and surfactant (functioning as structural stabilization and blocking agents), (ii) surfactant and surfactant (functioning as structural template for the micropores), and (iii) silicate and silicate (assembly of silicate network) during the self-assembly of the crystalline zeolites. The term “crystalline” refers to the repeated assembly of the basic unit cells of the microporous silicate network. Studies have shown that the slow crystallization process takes place during the hydrothermal aging after the formation of the amorphous silica particles. The surfactant SDAs can be removed by simple calcination, leaving behind well-ordered micropore channels within which catalytic reaction can take place. These micropores range from 8-membered ring (8-MR) (ultrasmall pore ∼4 Å), 10-MR (∼5 Å) to 12- (∼7 Å) and 14-MR (ultralarge pore, ∼8 Å) or above. Channels of 6-MR or less are too narrow to allow molecules to pass through and hence considered nonporous. The signature strong acidity of silicate-based zeolites originates from the partial substitution of the silicate (SiO4 4− ) building block with that of the aluminate (AlO4 5− ). The additional charge deficiency brought about by the latter can be readily neutralized by a labile proton, i.e., Brønsted acid. The Brønsted acid site can be conveniently used as an ion-exchange site to immobilize other cations for
9
10
1 Evolution of Catalysts Design and Synthesis
single-atom catalysis (discussed below). Interestingly, ion-exchanged Ca2+ , Y3+ , and La3+ sites are efficient catalytic sites for the pyrolytic carbonization of ethylene and acetylene. This produces homogeneous graphene-like layers within the micropores that upon the removal of the zeolite template produce faithful carbon replica of the microporous framework [44]. Such zeolite-templated carbon (ZTC) is interesting not only because of the electrically conductive and well-ordered microporous framework that can now be utilized for electrochemical and fuel cell-related reactions but also because the carbon, which can be easily removed by calcination, can potentially serve as secondary templates to synthesize other nonzeolite microporous catalysts. Care should be taken not to confuse zeolites with well-ordered mesoporous catalysts (e.g., MCM-41, SBA-15, KIT-6), which belong to a different class of porous materials and, by definition, consist of pores in the range of 2–50 nm. The MCM-41 (tunable pore size of 2–9 nm) and SBA-15 (tunable pore size of 5–10 nm), discovered by Charles T. Kresge et al. at the Mobil Oil Corporation in 1992 [45] and Galen D. Stucky and coworkers at the University of Santa Barbara in 1998 [46], respectively, are arguably the gold standards for this class of catalytic materials. These mesoporous catalysts are templated through the addition of bulky micelles such as those formed by cetyltrimethylammonium bromide (CTAB) surfactant and Pluronic P123 triblock copolymer, and sol–gel silica particles will precipitate in between these self-assembled soft templates. Because the micelles serve as long-range structural templates (and none at short range like those used for the synthesis of zeolites), well-ordered mesopores can be obtained, but the silica walls are basically amorphous. These glassy walls are catalytically inactive, in stark contrast with the crystalline walls of zeolites. Nevertheless, the mesoporous materials are attractive as high-surface-area supports with mesoporous channels large enough for the deposition of a wide range of active metals without pore blocking and at the same time accessible to bulky reactant molecules that otherwise could not penetrate the zeolite micropores. Because there is no requirement for short-range ordering, these surfactant templates can be flexibly used to fabricate a plethora of other mesoporous metal oxides including TiO2 , WO3 , and Al2 O3 . Furthermore, the mesoporous silica can be used as hard templates for the synthesis of mesoporous carbon and metal oxide nanorods [47]. An area that is actively being pursued is the synthesis of hierarchical zeolites, where mesoporous channels are introduced in zeolites, in such a way that the wall of the mesoporous catalyst is no longer amorphous silica but that of catalytically active, microporous crystalline silicate. This allows the accessibility of acid sites by large reactant molecules while overcoming the mass diffusion limitation associated with the narrow micropores of zeolites during catalytic reactions. More details on the design and synthesis of such hybrid micro-/mesoporous catalysts are presented in Chapter 7. Metal–organic framework (MOF) is a term first coined by Omar Yaghi in 1995 to describe a class of crystalline porous solids formed by a continuous network of multivalent metal cations/clusters and organic linkers of at least two coordination positions [48]. It is analogous to the zeolites, except with different set of building blocks. The elegance of MOFs arises from the simplicity of the template-free synthesis, and the micropore size can be easily tunable by adjusting the length of the organic linker. A classic example is the fabrication
1.3 Catalytic Cracking and Porous Catalysts
of UiO-66 that involves the simple hydrothermal reaction between zirconyl chloride and 1,4-benzenedicarboxylic acid (BDC) linker. By replacing the BDC with a longer 1,4-biphenyldicarboxylic acid (BPDC), one can obtain UiO-67 and an extension of the pore size from 7.5 and 12 Å to 12 and 16 Å, respectively. In fact, the design of MOFs is so flexible that it can be extended to fabricate mesoporous catalysts by manipulation of the linkers or using SDAs [49]. The catalytic active sites of MOFs may originate from the active metal atoms or compounds covalently functionalized on the linkers or the framework metal cation centers if made coordinatively unsaturated (without affecting the rigidity of the MOF structure). An elegant account on the different strategies in designing MOF catalysts can be found in Chapter 8. To date, MOFs find wide applications in organic synthesis, biomass conversion, photocatalysis, and electrocatalysis, among others. Because of their organic frameworks, MOFs are normally used in mid- to low-temperature applications below 500 ∘ C. A more recent sister class of compound is the covalent organic frameworks (COFs), first discovered by Yaghi in 2005, that are built entirely based on nonmetal centers [50]. In their pristine forms, some COFs are effective in catalyzing photocatalytic and electrocatalytic reactions, while their tunable porous structures can also be functionalized with the desired metal catalysts similar that of the mesoporous silica structure to catalyze a wider range of reactions, e.g., the Suzuki–Miyaura coupling reaction when deposited with the Pd2+ single-atom catalyst. The synthesis of porous anisotropic catalysts received significant interests since 2005 or so, especially for photocatalytic reactions such as solar water splitting, abatement of environmental pollutants, and CO2 reduction. Photocatalysts are composed of semiconductor materials, that is, they can photoexcited with photons equal to or larger than their bandgaps to produce usable charges for surface redox reactions. Photocatalytic reactions can be carried out in two ways: particulate photocatalysis where the redox reactions as mediated by the electron–hole pairs take place on the same photocatalyst particle/aggregate (see Chapter 11 on the art of photocatalysts design) and photoelectrocatalysis where the photocatalyst is made into a photoelectrode and connected with a counter electrode in such a way that the electron–hole pairs are separated across the two electrodes (see Chapter 36 on the basics of photoelectrocatalysis) [51]. One-dimensional (1D) photocatalysts such as nanorod and nanotube arrays are particularly attractive to capitalize on the high surface-to-bulk ratio as well as the much sought-after vectorial charge transport for efficient photocharge separation during photoelectrocatalytic reactions. A variety of synthesis techniques to obtain such structures have been developed, ranging from chemical vapor deposition, spray pyrolysis, and hydro/solvothermal synthesis to electrochemical anodization, producing efficient anisotropic photocatalysts of TiO2 nanotubes, WO3 nanosheets, Nb2 O5 nanorods, Ta2 O5 nanotubes, 𝛼-Fe2 O3 nanotubes, etc. The electrochemical synthesis of these fascinating array photocatalysts can be found in Chapter 3. In recent years, the interest has expanded to two-dimensional (2D) photocatalysts such as the graphitic carbon nitride, molybdenum disulfide, tungsten disulfide, and MXenes. Besides maximizing the surface-to-bulk ratio, these materials exhibit unique quantum electronic properties seen only when made into atomic-thin layers [52].
11
12
1 Evolution of Catalysts Design and Synthesis
1.4 Miniaturization of Metal Catalysts: From Supported Catalysts to Single-Atom Sites A core criterion in the design of catalysts is to maximize the active metal dispersion (the ratio of surface atom to bulk), such that the highest reactivity per amount of metal loading on the catalyst can be achieved. This is especially relevant when precious metals are used, which is indeed the case for a large number of catalytic reactions. In the abovementioned historical overview, the strategies for maximizing dispersions include making metal sheets into metal sponges and gauzes of fine wires, as well as depositing active metals onto high-surface-area supports to make very fine deposits or thin atomic layers. The synthesis of supported catalysts capitalizes on the strong interfacial interactions between the active metal and the (usually oxide) support to allow the former to exist as stable and small size deposits. Without the strong interfacial interactions, the initially small deposits tend to diffuse on the support surface and coalesce with another deposit until its surface energy (i.e., a function of surface area) decreases to that of the interfacial energy. Incipient wetness impregnation is by far the most common procedure for the preparation of supported catalysts, where metal precursor solution is drawn into the pores of the support by means of capillary effect. To prevent overflowing of the solution to the external surface of the support, the solution volume introduced in each impregnation step should not exceed that of the pore volume (typically maintained at 80–90%). More liquid solution can be introduced repeatedly upon complete drying of the liquid solvent, leaving behind more metal salt within the pore during each repetition. The advantage of the incipient wetness impregnation is that it does not require very strong interactions between the oxide support surface and the coordinated metal cation from the precursor to reach the desired loading amount. On the contrary, wet impregnation is when the porous support is immersed in the metal precursor solution and the amount that penetrates the pores depends on the metal precursor–support interactions. If the interaction is strong, the impregnated concentration would be higher than that of the bulk, and vice versa. Further drying to remove the solvent from the pores and calcination yield the supported catalysts in both cases of impregnation. To minimize coalescence between the metal deposits during the calcination step, it is essential to remove moisture and oxygen by flowing inert gas and introducing a small amount of NO, respectively [53]. Other techniques such as deposition–precipitation, chemical vapor deposition, and the one-step flame synthesis (see Chapter 10) have also become popular alternatives for producing supported metal catalysts. The ability to produce small Pt deposits on carbon support has been one of the major breakthroughs that led to popularity of low-temperature H2 -polymer electrolyte membrane (PEM) fuel cell. In fact, the amount of the ∼3.5 nm Pt used is so small (0.2 mg/cm2 of fuel cell, compared with 28 mg/cm2 in the early days) that it significantly reduced the device cost and thus popularising the H2 -PEM fuel cell [54]. Chapter 32 introduces the design of electrocatalysts for PEM fuel cell applications, while Chapter 4 complements nicely the strategies of using carbon supports for such purpose.
1.4 Miniaturization of Metal Catalysts: From Supported Catalysts to Single-Atom Sites
Besides maximizing the metal dispersions, further miniaturization of metal deposits to or approaching the quantum-related level can result in altered electronic properties not otherwise seen in larger particles. Gold catalysis is an intriguing example of such a phenomenon, which was led notably by the independent efforts of Graham J. Hutchings and Masatake Haruta since the mid-1980s. They showed that gold, which was classically believed to be almost inactive, can be made extremely active in the hydrochlorination of acetylene [55] and the oxidation of carbon monoxide (at −77 ∘ C!) [56], respectively, when made less than 25 nm. The latter, which gold size was 4.5 ± 1.6 nm, was first prepared by the coprecipitation technique but was later superseded by the deposition–precipitation technique in which dissolved gold precursor was precipitated by raising the pH of the medium in the presence of suspended oxide support. Over time, the commercial flame-synthesized P25 TiO2 became the preferred support. Many new reactions by gold catalysis followed in the next three decades, ranging from the oxidation of aqueous polyalcohols to carboxylic acids, selective oxidation of cyclohexane to cyclohexanol and cyclohexanone, epoxidation of propylene, water-gas shift, to the selective hydrogenation of 3-nitrobenzene and the hydrogenation of alkynes to alkenes. Size-dependent turnover frequencies (i.e., conversion rate per active site) is typically observed due in part to the variation of electronic interactions, with the optimum gold deposit size for CO oxidation in the range of 2–4 nm [57, 58]. The size-dependent activity is a general phenomenon as observed readily on different metal deposits including cobalt for FTS [59], palladium for Suzuki coupling [60], and platinum for propane dehydrogenation [61]. Synthesizing ultrasmall size deposits of less than 2 nm ( 𝛾 {100} > 𝛾 {111} . Considering the surface energy order is {110} > {100} > {111}, the final crystal has a tendency to form an octahedron-shaped crystal that is dominated with {111} facet, rather than a cube enclosed by {100} facet (as shown in Figure 2.1). However, the octahedral shape has a larger surface area than the cube of the same volume. As a consequence, the shape turns to be a truncated octahedron with a mix of {100} and {111} facets [1]. Another example is anatase TiO2 . According to the Wulff construction and surface energy calculation, the equilibrium shape of anatase TiO2 crystal (as shown in Figure 2.1) is a slightly truncated bipyramid enclosed with 94% {101} facet and 6% {001} facet [21], although the order of the surface energy of low-index facets is {001} (0.90 J/m2 ) > {010} (0.53 J/m2 ) > {101} (0.44 J/m2 ) [22]. In practice, the product often shows a different shape from that predicted by equilibrium. The possible reasons may be that (i) the equilibrium condition was not fully satisfied during the synthesis, and/or (ii) the anisotropic surface energies of different facets were interfered by impurities or other factors. In other words, this allows the manipulation of the nucleation and crystal growth by intentional addition of impurities and tuning of synthesis conditions to
2.2 Mechanisms of Facets Engineering
{111}
{111} {100} {100}
(a)
Octahedron
Truncated octahedron
{001}
Cube
{001} {101}
{001} {101}
{100}
{101}
(b)
Equilibrium shape of anatase TiO2
Figure 2.1 (a) Octahedron, truncated octahedron, and cube with the same volume. (b) The equilibrium shape of anatase TiO2 (middle) and two variants.
achieve product particles with desired shape and exposed facets. This is the core concept of facets engineering. Selectively controlling the nucleation and anisotropic growth rate is known as the bottom-up route. The most common method is to use a selective capping agent to reduce the surface energies of the adsorbed facets, or to change the order of surface energies of different facets, or to terminate the crystal growth of a selective facet. Figure 2.2 indicates how the solvents and capping agents can be used to tune the morphologies during the crystal growth [23]. The capping agents can be atomic or molecular species originating from a gas or liquid environment. As early as in 1986, it was found that H2 S could cause drastic morphological changes of Pt nanocrystals [24]. Pt{100} facet had a stronger interaction with sulfur than Pt{111} facet, resulting in the formation of Pt nanocubes rather than Pt nanospheres. More capping agents are generally used in the solution-phase synthesis. For example, inorganic species such as bromides and organic species such as poly(vinylpyrrolidone) (PVP) are very popular for tailoring both metal [25–27] and semiconductor crystals [28–30]. Selectivity and adsorption capability of the capping agents, regardless of organic or inorganic capping agents, are basically controlled by the density and arrangement of undercoordinated atoms on different surfaces. The capping
25
26
2 Facets Engineering on Catalysts
Solvent 1
A A
A B
B
B
A
A
A
Solvent 2
A A
A
B B A A
A, B: Crystal facet
: Capping agents
A
B
: Preferential growth direction
Figure 2.2 Schematic of the effect of solvent and capping agents on the morphology control of crystal facets. Source: Adapted from Liu et al. 2011 [23].
agents stabilize these high-energy surfaces by covering the undercoordinated atoms. This also means that the reactive sites on the surfaces are also likely to be covered. For instance, fluorine is the most frequently used capping agent for faceted TiO2 crystals. Fluorine always exists in the surface of as-prepared faceted TiO2 crystals. Pan et al. demonstrated that the fluorine-terminated anatase TiO2 crystals with different percentages of {001}/{101}/{010} facets have similar low photocatalytic performance. After the removal of fluorine by calcination, all TiO2 crystals exhibited much higher and diverse performance depending on the ratio of different facets. Although most studies suggest that the removal of the surface fluorine improves the performance in photocatalytic hydrogen evolution from water splitting with sacrificial agents, in some cases, the capping agents may also tune the activity and selectivity of the catalysts by involving in the catalytic process. Another strategy to tailor the crystal morphology is via the top-down route, where a starting particle sample is selectively etched to remove the undesirable facets. This method is more often used in the synthesis of semiconductors. The protective capping agents can be used to shield the desirable facets, leaving the uncapped and unwanted facets to be dissolved in the etching process. For instance, truncated octahedral Cu2 O crystals can be synthesized via the hydrothermal method by using PVP as a capping agent. PVP is preferentially adsorbed on the Cu2 O{111} facet. And then, in the following oxidative etching process, PVP can protect the Cu2 O{111} facet. As a result, the truncated octahedral Cu2 O crystals turned into a hollow structure with six {100} facets
2.3 Anisotropic Properties of Crystal Facets
absent [31]. In some cases, even without any protective agents, selective etching can be used to prepare hollow structures, due to anisotropic corrosion of different facets. For example, a rectangular rutile TiO2 nanorod can be selectively etched along the [001] direction in hydrochloric acid, turning into a hollow rutile TiO2 tube. This is because the rutile TiO2 {001} facets have a higher dissolution rate than the {110} facets [32].
2.3 Anisotropic Properties of Crystal Facets The properties of crystal surface differ from the ones inside crystal bulk, due to the termination of the periodic arrangement of the atoms. The anisotropy of the crystal lattice determines the anisotropy of the atomic arrangement of the facets exposed, leading to anisotropic properties of the crystal surface. 2.3.1
Anisotropic Adsorption
Heterogeneous reaction takes place on a catalyst surface (most commonly as solid) where phase differs from that of the reactants [33]. In a typical case, the process consists of three steps, adsorption, reaction, and desorption occurring at the catalyst surface. The effective adsorption, which is an essential step during heterogeneous reaction, is largely related to the exposed atomic structures of the catalyst surface. Because the surface atomic arrangement and coordination vary in different crystal facets, adsorption energy and states of the reactant molecules are different on various exposed facets. This variation of adsorbing states has a profound impact on the subsequent catalytic steps. Various surface science studies have been performed to investigate the interactions between the different surface facets of metals or semiconductors, and the reactant molecules. Platinum, being one of the most versatile metal catalysts, has an fcc crystal structure and is one of the most studied surfaces. On the flat surface, such as Pt{100} and Pt{111}, each platinum atom is surrounded by four and six adjacent neighbors, respectively. These two low-index facets have highest atomic density. High-index facets, such as Pt{530} and Pt{730}, contain many terrace structure and steps with low-coordinated atoms [34]. The atomic arrangement of Pt{730} facet is periodically overlapped with one Pt{310} subfacet and two Pt{210} subfacets, leading to a multiple-height terrace structure. Furthermore, the faceted metal catalysts have more atoms located at the edges and corners, where they have much lower atomic density and coordination, which can have a profound impact on surface adsorption. Titanium dioxide (TiO2 ) has been the most intensively investigated binary transition metal oxide as a photocatalyst, where many theoretical calculations have been conducted to investigate the interactions between water molecules and the TiO2 surface [35]. Because the water molecule is one of the most important adsorbates present in many photocatalytic reactions, such as water splitting, photodegradation of organic pollutants, and CO2 reduction, many experimental and theoretical studies have focused on this aspect [36–40]. Anatase {101}, {001}, and {100} surfaces are three typical low-index surfaces of
27
2 Facets Engineering on Catalysts
[010]
02
2.28
1.12 1.31
1.
[100]
H2O molecularly adsorbed on anatase TiO2{101} facets
8 1.7
3
96
1.9
1.
4
– [101] 1.85
(001)
0
1.88
1.0
[010] (101)
1.7
28
1.83
H2O dissociatively adsorbed on anatase TiO2{001} facets
Figure 2.3 Side view of anatase TiO2 {101} and {001} facets. Top view for adsorbed water molecules on anatase {101} surface and side view of adsorbed water molecules on anatase {001} surface. Source: Vittadini et al. 1998 [36]. Reproduced with permission of American Physical Society.
anatase TiO2 . Anatase {101} contains 50 % sixfold coordinated (Ti6c ) Ti atoms (saturated) and 50 % fivefold coordinated (Ti5c ) Ti atoms (unsaturated), whereas anatase {001} and {100} surface contains 100% Ti5c atoms. Most theoretical and experimental results suggest molecular adsorption of water on the defect-free {101} surface [41] but dissociative adsorption of water molecules on the {001} surface at low coverages (shown in Figure 2.3) [36]. The dissociative adsorption of water is also favorable on the anatase {100} surface [12]. 2.3.2
Surface Electronic Structure
Anisotropic electrical properties of surfaces are logically attributed to the anisotropy of the crystal lattice, which determines the different atomic arrangements and configurations depending on exposed facets. Work function is the minimum energy for valence electrons in the solid to overcome in order to exit into the vacuum [42], which can be defined as W = −e𝜙 − EF where for elemental metals, EF is the potential energy of the electrons at the top of the valence band (VB) called the Fermi level, −e is the charge of an electron, and 𝜙 is the electrostatic potential in vacuum. For elemental metals, work function highly depends on the crystallographic orientation of the surface, as atomic density and electron charge density vary across different facets. The difference of work function between two surfaces of the same metal in different orientation can reach up to 1 eV [43]. For example, work functions of tungsten W{001} facet, W{112} facet, and W{111} facet are 4.56 eV, 4.69 eV, and 4.39 eV, respectively [44].
2.3 Anisotropic Properties of Crystal Facets
Work function of polycrystalline Ag is 4.26 eV, but work functions of Ag{100} facet, Ag{110} facet, and Ag{111} facet are 4.64 eV, 4.52 eV, and 4.74 eV, respectively [45]. Furthermore, surface roughness and particle size also have a profound impact on work function. The electronic structure of semiconductors is different from that of metals because of a bandgap between the conduction band (CB) and the VB. According to the band theory, when a large number of identical atoms assemble to form a solid, the atomic orbitals with discrete energy levels will overlap. Each atomic orbital will split into discrete molecular orbitals with different energies, due to the Pauli exclusion principles stating that it is impossible for two electrons in the solid to have the same values of the four quantum numbers. For example, the VB of TiO2 is composed of O 2p orbitals, while the CB is composed of Ti 3d orbitals [46]. In the bulk of TiO2 crystals, no matter in anatase or rutile, there are numerous TiO6 octahedron units connected to their neighbors by sharing corners and edges in different ways. But at the surface, this periodic arrangement terminates, leading to variation in the coordination of Ti and O atoms. It is reasonable to deduce that the band structure at the surface is more or less different from the band structure in the bulk. A blueshift of light absorption edge was found when comparing the nanosized anatase TiO2 crystals with 82% {101} and 18% {001} facets with the micrometer-sized anatase TiO2 crystals with 28% {101} and 72% {001} facets [47]. The 9 nm blueshift of absorption edge means a larger bandgap, which is attributed to the different dominant facets exposed. The dependencies of bandgap and exposed surface were also found in other materials [48–50]. 2.3.3
Surface Electric Field
For a metal, the surface electric field is oscillating when the light strikes the surface. Light, an electromagnetic wave, oscillates the electric field in a plane perpendicular magnetic field. The electric field’s oscillatory patterns would cause a rippling wave pattern in the distribution of electrons, where the resonant oscillation of conduction electrons is called surface plasmon resonance (SPR). The SPR only exists in metals or other electrically conductive materials containing conduction electrons. When the size of the metal crystals shrinks to the nanoscale, which is smaller than the wavelength of the incident light, the surface plasmon is confined to a very small surface rather than the bulk material, known as localized surface plasmon resonance (LSPR). The LSPR frequency affects the light absorption and scattering of metal nanoparticles. How the LSPR frequency is affected by facets (shapes) of a nanoparticle is explained in a later section, but as a consequence, the color of metal nanoparticles will be changed, and it is sensitive to the shape of nanoparticles (with different facets exposed). Based on the Mie theory, it is possible to tune the LSPR spectra of Ag nanocrystals of different shapes, as shown in Figure 2.4 [51]. For a faceted semiconductor, the surface electric field is a disparate situation. Besides the variation of the bandgap, the band edge position shifts as a function of different facets due to surface band bending. The different band edge positions provide varying redox potentials of the photogenerated electrons and holes, resulting in spatial separation of charge carriers and the built-in electric
29
Sphere 4
Cube 4
12
600 Wavelength (nm)
400
800 (b)
600 Wavelength (nm)
Octahedron 4
Optical coefficient
8
400
40 nm
15 Triangular plate
10 5 0
400
600 Wavelength (nm)
800
600 Wavelength (nm)
800
40 nm
15 5 nm 10
Disc
5 0
400 (e)
4
20 5 nm
0
Tetrahedron
(c)
20
40 nm
8
0
800
Optical coefficient
(a)
Optical coefficient
8
0 400
40 nm
12 Optical coefficient
8
0
(d)
40 nm
12
40 nm
Optical coefficient
Optical coefficient
12
600 800 Wavelength (nm)
1000
400 (f)
600 800 Wavelength (nm)
1000
Figure 2.4 Calculated UV–visible extinction (black), absorption (red), and scattering spectra (blue) of Ag nanocrystals, illustrating the effect of shape on its spectral characteristics, including isotropic sphere (a), anisotropic cubes (b), tetrahedra (c), and octahedra (d), triangular plate (e) and circular disc (f ). Source: Wiley et al. 2006 [51]. Reproduced with permission of American Chemical Society. (See online version for color figure).
2.3 Anisotropic Properties of Crystal Facets
field. In addition, selectively depositing a noble metal as cocatalysts on the surface facets can further enhance the strength of the built-in electric field. When a single BiVO4 crystal enclosed by {010} and {011} facets was characterized by spatially resolved surface photovoltage spectroscopy (SRSPS), {011} facets exhibited a much higher signal intensity of surface photovoltage than {010} facet [52]. This phenomenon indicated a significant difference in surface band bending between BiVO4 {011} and {010} facets. As a consequence, the different band bending will lead to the variation in the spatial distribution of the charge carriers and build an electric field between different facets. By changing the area ratio of (011)/(010) facets of BiVO4 crystal, the surface built-in electric field varied as well. Such an intrinsic difference in the surface photovoltage between different facets can be further enhanced by selectively depositing cocatalysts, such as MnOx and Pt deposited on faceted BiVO4 crystal, as shown in Figure 2.5 [53]. 100 mV
SEM
h+
MnOx
Pt
e–
2.0 μm (a)
–50 mV (b)
011
010 Built-in electric field
7.2 mV
CB SCR
011 95 mV 65 mV
010 Built-in electric field CB
3.4 mV SCR
SCR VB
Pt
VB
~100 nm
~1000 nm MnOx
(c)
–35 mV –79 mV
Pt
Vector directions (d)
Figure 2.5 (a) Scanning electron microscopy (SEM) image of a BiVO4 single crystal with Pt photodeposited on {010} facet and MnOx photodeposited on {011} facet. (b) Spatial distribution of the surface photovoltage signals. Pink and green colors correspond to holes and electrons separated toward the external surface, respectively. Schematic band diagrams across the border between the {011} and {010} facets of (c) a bare single BiVO4 photocatalyst particle and of (d) a single BiVO4 photocatalyst particle with MnOx cocatalyst selectively deposited at {011} facets (green line) and with MnOx and Pt nanoparticles selectively deposited at {011} and {010} facets, respectively (dashed pink line). Source: Zhu et al. 2017 [53]. Reproduced with permission of American Chemical Society. (See online version for color figure.)
31
32
2 Facets Engineering on Catalysts
2.4 Effects of Facets Engineering Each facet in a single crystal has different properties. However, combining anisotropic surface properties could dramatically alter the properties of the crystal, especially when the particle size is reduced to the nanoscale and the ratio of surface atoms/bulk atoms is no longer negligible. 2.4.1
Optical Properties
As mentioned previously, the surface electric field of a metal oscillates when the light strikes the surface. The oscillating electric field causes a rippling wave pattern in the spatial distribution of electrons. According to Lenz’s law, the wave created by the surface plasmon opposes the electromagnetic wave of the incident light. The oscillating electrons absorb the energy of light and reemit the energy as the reflected light, due to which metals have shiny and reflective surfaces. However, when the particle size becomes very small, the surface plasmon is confined to a very small surface (i.e. LSPR). When the electron cloud is excited at one of the resonance frequencies, light absorption will become stronger. This is how LSPR frequency affects the light absorption of metal nanoparticles. The plasmon frequency is determined by electron density, dielectric constant, and effective mass of an electron. The well-defined facets of a crystal form different shapes with more symmetries compared with spherical particles [54]. The surface charges tend to accumulate at edges and corners, which further promote surface polarization, i.e. the charge separation between mobile electrons and immobile atoms. Surface polarization determines the frequency and intensity of LSPR as it provides the main restoring force for electron oscillation. Large surface polarization reduces the restoring force, resulting in a redshift of resonance peak, and multiple distinct symmetries may induce several light absorption peaks [55]. Therefore, the same metal nanoparticles with different size and shape may exhibit different colors, indicating diverse light absorption. The light absorption of semiconductors is quite different from that in metals due to the electronic band structure in semiconductors. Between the VB and CB of semiconductors, no electron states exist in this energy range called the bandgap. In some semiconductors, the minimal energy state of the CB (conduction band minimum, CBM) and the maximal energy state of the VB (valence band maximum, VBM) are situated in the same crystal momentum in the Brillouin zone (direct gap); in other semiconductors, they are not (indirect gap). There is a slight difference in light absorption between these two types of bandgap structure. But it is not necessary to discuss in this chapter. In general, light absorption of a semiconductor is associated with its bandgap. Semiconductors only absorb photons with the energy equal to or greater than the bandgap. As a result of facets-dependent anisotropic surface electronic properties that in turn influence the band positions, semiconductor crystals with different dominant facets show shifting in light absorption edges. In applications
2.4 Effects of Facets Engineering
such as photoelectrochemical catalysis, faceted semiconductors can enhance the light harvesting of photoelectrodes [56–58]. The combination of plasmonic metals and semiconductors with facets engineering has the great potential to adjust the light harvesting for photoelectrodes. For example, the LSPR absorption of the faceted plasmonic metal nanoparticles, such as Au and Pd, can be tuned by embedding them in Cu2 O to form core–shell heterostructure [59–61]. 2.4.2
Activity and Selectivity
Numerous studies have shown that catalysts with facets engineering exhibit greater catalytic performance. This exploits one or more unique properties of the well-defined crystal facets to tune the overall catalytic activity and/or selectivity. For metal catalysts, activity and selectivity are related to the surface atomic structure of the catalysts, where they can be tuned to enhance effective adsorption and/or promote favorable coordination of adsorbates. These processes are strongly influenced by the arrangement and coordination of surface atoms as well as by the corresponding surface density of states of the different facets. For example, there are two types of flat surfaces of Pt, namely, the Pt{111} facet (hexagonal surface) where each surface atom has six nearest neighbors, and the Pt{100} facet (square surface) where each surface atom has four nearest neighbors. The hexagonal surface is up to seven times more active than the square surface in the aromatization reaction of n-heptane to toluene, but the square surface is seven times more active than the hexagonal surface in the alkane isomerization reaction of isobutane to n-butane [62]. Also in electrocatalysis, the Pt{210} facet has high activity for electrooxidation of formic acid and electroreduction of CO2 [63]; the Pt{410} facet exhibits high performance in NO decomposition [64]; the Pt{730} shows superior activity in electrooxidation of formic acid and methanol [34]. For the semiconductors in photo-related catalytic processes, the performance of the faceted semiconductor crystals is affected by the synergetic combination of the intrinsic properties of the bulk and surfaces. A typical photocatalytic process includes the following steps: light absorption of the semiconductor catalyst, excited charges (electrons and holes) generation, excited electrons and holes recombination (bulk and surface), charge migration to the surface, charge trapping at the surface, and transfer to reactants. As such, an ideal semiconductor with good reactivity and selectivity is expected to have a suitable bandgap to absorb the light as much as possible, good separation of charge carriers to avoid the recombination at the bulk and surface, good adsorption of reactants at the surface, fast surface transfer of charge carriers from catalysts to reactants, and suitable CBM and VBM position to provide electron–hole pairs with sufficient redox potentials to catalyze the reactions. Each aspect is affected by one or more properties of the surface and bulk. And these effects
33
34
2 Facets Engineering on Catalysts
Facet engineering
Surface atomic structure
Surface electronic structure
Anisotropic adsorption
Surface band position
Built-in electric field Adsorption states of reactants
Surface tansfer of charge carriers
Selectivity
Redox power
Charge carriers separation
Reactivity
Figure 2.6 Diagram of how the facets engineering affects the selectivity and reactivity of a semiconductor photocatalyst.
are not isolated and may affect one another. Figure 2.6 shows how the properties synergistically affect the reactivity and selectivity of semiconductor catalysts and the significance of facets engineering.
2.5 Outlook Reviewing the published studies of facets engineering, from the faceted crystal synthesis to the investigation of the intrinsic property of the exposed surface and to the overall performance of catalysts in various reactions, it is not difficult to find that the benefits of facets engineering are attributed to crystal anisotropy. Facets engineering itself cannot solve all the issues of catalysts. Although catalytic reactions occur mainly on the surface, there are important inherent bulk properties of the catalysts that play significant roles in concert with the engineered facets to maximize the catalytic performance. Just as in single-atom catalysis, the substrate has a direct influence on the catalytic atoms. Based on the characteristics
References
of the original catalyst, the morphology control brings more possibilities to the performance of the catalysts. Although we have attained a good understanding on the mechanisms of facets engineering, we are still facing many challenges, uncertainties, and uncontrollability, such as surface reconstruction. Surface reconstructions always take place on most crystals as a natural phenomenon to minimize surface energy. In many cases where the active facets are the high surface energy facets composed of a high percentage of unsaturated atoms, they become highly unstable upon the removal of capping agents. It inevitably led to the rearrangement of surface atoms or the formation of surface defects, which would have a profound impact on the catalytic properties of the surface. More efforts need to be dedicated to developing new synthesis methods for new active facets and structure to reveal new facets-dependent properties and the synergy between these properties and to control surface defects and identify the substantial differences between the ideal and real surface. There is no doubt that facets engineering is an important strategy to promote the performance of catalysts and provides a valuable platform for the rational design and fabrication of efficient material system.
References 1 Xia, Y.N., Xiong, Y.J., Lim, B., and Skrabalak, S.E. (2009). Angew. Chem. Int.
Ed. 48: 60–103. 2 Liu, G., Yang, H.G., Pan, J. et al. (2014). Chem. Rev. 114: 9559–9612. 3 Burda, C., Chen, X.B., Narayanan, R., and El-Sayed, M.A. (2005). Chem. Rev. 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
105: 1025–1102. Pan, J. and Liu, G. (2017). Semicond. Semimet. 97: 349–391. Xia, Y.N. and Halas, N.J. (2005). MRS Bull. 30: 338–344. Wang, A.Q., Li, J., and Zhang, T. (2018). Nat. Rev. Chem. 2: 65–81. Bohme, D.K. and Schwarz, H. (2005). Angew. Chem. Int. Ed. 44: 2336–2354. Jiang, H.B., Cuan, Q.A., Wen, C.Z. et al. (2011). Angew. Chem. Int. Ed. 50: 3764–3768. Wang, Z.Y., Huang, B.B., Dai, Y. et al. (2012). CrystEngComm 14: 4578–4581. Wang, J.H., Yang, T.H., Wu, W.W. et al. (2006). Nanotechnology 17: 719–722. Kim, C., Gu, W.H., Briceno, M. et al. (2008). Adv. Mater. 20: 1859–1863. Pan, J., Liu, G., Lu, G.M., and Cheng, H.M. (2011). Angew. Chem. Int. Ed. 50: 2133–2137. Pan, J., Wu, X., Wang, L.Z. et al. (2011). Chem. Commun. 47: 8361–8363. Zhao, X., Jin, W., Cai, J. et al. (2011). Adv. Funct. Mater. 21: 3554–3563. Mallikarjuna, N.N. and Varma, R.S. (2007). Cryst. Growth Des. 7: 686–690. Praneeth, N.V.S. and Paria, S. (2018). CrystEngComm 20: 4297–4304. Yu, Y.Y., Chang, S.S., Lee, C.L., and Wang, C.R.C. (1997). J. Phys. Chem. B 101: 6661–6664. Kim, F., Song, J.H., and Yang, P.D. (2002). J. Am. Chem. Soc. 124: 14316–14317.
35
36
2 Facets Engineering on Catalysts
19 Placido, T., Comparelli, R., Giannici, F. et al. (2009). Chem. Mater. 21:
4192–4202. 20 Lovette, M.A., Browning, A.R., Griffin, D.W. et al. (2008). Ind. Eng. Chem. Res.
47: 9812–9833. 21 Gong, X.-Q. and Selloni, A. (2007). Phys. Rev. B 76: 235307-1–235307-11. 22 Lazzeri, M., Vittadini, A., and Selloni, A. (2002). Phys. Rev. B 65: 155409. 23 Liu, G., Yu, J.C., Lu, G.Q., and Cheng, H.M. (2011). Chem. Commun. 47: 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51
6763–6783. Harris, P.J.F. (1986). Nature 323: 792–794. Cha, S.I., Mo, C.B., Kim, K.T. et al. (2006). J. Mater. Res. 21: 2371–2378. Sun, Y.G. and Xia, Y.N. (2002). Science 298: 2176–2179. Xiong, Y.J., Cai, H.G., Yin, Y.D., and Xia, Y.N. (2007). Chem. Phys. Lett. 440: 273–278. Yang, H.G., Sun, C.H., Qiao, S.Z. et al. (2008). Nature 453: 638–641. Huang, W.C., Lyu, L.M., Yang, Y.C., and Huang, M.H. (2012). J. Am. Chem. Soc. 134: 1261–1267. Zhang, D.F., Zhang, H., Guo, L. et al. (2009). J. Mater. Chem. 19: 5220–5225. Sui, Y.M., Fu, W.Y., Zeng, Y. et al. (2010). Angew. Chem. Int. Ed. 49: 4282–4285. Liu, L., Qian, J.S., Li, B. et al. (2010). Chem. Commun. 46: 2402–2404. Schlogl, R. (2015). Angew. Chem. 54: 3465–3520. Tian, N., Zhou, Z.Y., Sun, S.G. et al. (2007). Science 316: 732–735. Sun, C.H., Liu, L.M., Selloni, A. et al. (2010). J. Mater. Chem. 20: 10319–10334. Vittadini, A., Selloni, A., Rotzinger, F.P., and Gratzel, M. (1998). Phys. Rev. Lett. 81: 2954–2957. Herman, G.S., Dohnalek, Z., Ruzycki, N., and Diebold, U. (2003). J. Phys. Chem. B 107: 2788–2795. Diebold, U. (2003). Surf. Sci. Rep. 48: 53–229. Onda, K., Li, B., Zhao, J. et al. (2005). Science 308: 1154–1158. He, Y., Tilocca, A., Dulub, O. et al. (2009). Nat. Mater. 8: 585–589. Selloni, A., Vittadini, A., and Gratzel, M. (1998). Surf. Sci. 402: 219–222. Kittel, C. (2004). Introduction to Solid State Physicis, 8e. Wiley. Yamamoto, Y. and Miyokawa, T. (1998). J. Vac. Sci. Technol., B 16: 2871–2875. Kaminsky, M. (1965). Atomic and Ionic Impact Phenomena on Metal Surfaces. Academic Press. Dweydari, A.W. and Mee, C.H.B. (1975). Phys. Status Solidi A 27: 223–230. Asahi, R., Taga, Y., Mannstadt, W., and Freeman, A.J. (2000). Phys. Rev. B 61: 7459–7465. Liu, G., Sun, C.H., Yang, H.G. et al. (2010). Chem. Commun. 46: 755–757. Bi, Y., Ouyang, S., Umezawa, N. et al. (2011). J. Am. Chem. Soc. 133: 6490–6492. Wang, X., Yin, L., Liu, G. et al. (2011). Energy Environ. Sci. 4: 3976–3979. Xie, Y.P., Liu, G., Yin, L., and Cheng, H.-M. (2012). J. Mater. Chem. 22: 6746–6751. Wiley, B.J., Im, S.H., Li, Z.Y. et al. (2006). J. Phys. Chem. B 110: 15666–15675.
References
52 Zhu, J., Fan, F.T., Chen, R.T. et al. (2015). Angew. Chem. Int. Ed. 54:
9111–9114. 53 Zhu, J., Pang, S., Dittrich, T. et al. (2017). Nano Lett. 17: 6735–6741. 54 Huh, J.H., Lee, J., and Lee, S. (2018). ACS Photonics 5: 413–421. 55 Tao, A., Sinsermsuksakul, P., and Yang, P.D. (2006). Angew. Chem. Int. Ed. 45:
4597–4601. Zhang, Y.N., Qin, N., Li, J.Y. et al. (2017). Appl. Catal., B 216: 30–40. Tao, J.G., Luttrell, T., and Batzill, M. (2011). Nat. Chem. 3: 296–300. Wang, L., Nguyen, N.T., Shen, Z.Q. et al. (2018). Nano Energy 50: 331–338. Huang, M.H., Rej, S., and Chiu, C.Y. (2015). Small 11: 2716–2726. Rej, S., Wang, H.J., Huang, M.X. et al. (2015). Nanoscale 7: 11135–11141. Yang, K.H., Hsu, S.C., and Huang, M.H. (2016). Chem. Mater. 28: 5140–5146. Somorjai, G.A. (1994). Introduction to Surface Chemistry and Catalysis. Wiley. 63 Hoshi, N., Kawatani, S., Kudo, M., and Hori, Y. (1999). J. Electroanal. Chem. 467: 67–73. 64 Banholzer, W.F. and Masel, R.I. (1984). J. Catal. 85: 127–134. 56 57 58 59 60 61 62
37
39
3 Electrochemical Synthesis of Nanostructured Catalytic Thin Films Hoi Ying Chung and Yun Hau Ng City University of Hong Kong, School of Energy and Environment, Kowloon, Hong Kong, Special Administrative Region (S.A.R.)
3.1 Introduction Catalytically active materials in the physical form of thin film (in the range of nanometer to micrometer) have found wide applications in reactions involving thermal catalysis, electrocatalysis, and photocatalysis [1]. Although depending on the targeted applications, usage of catalytic thin films offers a few advantages from the operational viewpoint over the powder or homogeneous catalyst counterpart. For instance, the elimination of catalyst separation process upon completion of reactions is helpful in simplifying processes. Improved robustness against sintering at elevated operating temperature is another crucial benefits offered by thin films to prolong the stability of catalyst because the heat-induced sintering always results in the loss of activity. Furthermore, catalytic reactions involving electrical circuit such as electrochemical hydrogen evolution reaction (HER), oxygen evolution reaction (OER), and other electrocatalytic reactions must at least require the active materials to be immobilized on the electrodes. Thin film is one of the most common forms of active electrodes – see, for example, their detailed applications in electrochemical water splitting, polymer electrolyte membrane fuel cells, and photo/electrochemical CO2 reduction in Chapters 30, 32, and 36, respectively. These catalytic thin films are prepared either by direct growth of catalytic materials on thin substrates (e.g. glass or metal sheets) or they can be pre-synthesized as powder materials followed by an immobilization process on the thin films [2, 3]. Traditionally, flat thin films can be prepared using thermal/chemical/physical/vapor deposition, sputtering, spin/ dip/doctor-blade coating, electroplating, etc. [4–10] Principles used in guiding the formation of thin films are vastly different. For example, in vapor deposition methods, usually low pressures and high temperatures are needed for generating the vapor of precursors. In particle coating techniques, particle size, binder, and viscosity modifier play important roles in ensuring good adhesion with the substrates. Control over the uniformity of thickness, composition of materials, and strength of adhesion are the typical aspects considered in the flat thin film synthesis. Heterogeneous Catalysts: Advanced Design, Characterization and Applications, First Edition. Edited by Wey Yang Teoh, Atsushi Urakawa, Yun Hau Ng, and Patrick Sit. © 2021 WILEY-VCH GmbH. Published 2021 by WILEY-VCH GmbH.
40
3 Electrochemical Synthesis of Nanostructured Catalytic Thin Films
Evolved from these flat thin films, catalytic thin films with nanostructure (whether with or without ordered nanostructure; with or without regular pattern) are emerging as a new class of functional materials. Nanostructures of catalytic components on thin films can be generally grouped into 1D (e.g. nanotubes, nanorods), 2D (e.g. nanosheets), and 3D (hierarchical nanostructures, e.g. tetrapods, nanoflowers, sea urchin-like structures) configurations, with unique properties found in each nanostructure. Modulation of these anisotropic nanostructures, where the shapes of the nanostructures are formed as a result of preferential growth (or leaching) in certain directions, on thin films is another domain that needs to be addressed with precise control. As the conventional flat thin films are typically made of bulk materials (metal, metal oxides, metal-based semiconductors, polymeric structures, etc.), the introduction of nanostructures offers additional properties (e.g. physical, electronic, and optical) [11–13] to the thin film, in which new applications are found. Among all methods capable of fabricating traditional flat catalytic thin film, a handful of existing methods can be adopted with modification to find usefulness in the preparation of nanostructured thin films. In this chapter, electrochemical method, which has been used in both the fabrication of flat single-component thin films and the emerging complicated nanostructured multicomponent thin films, will be discussed in detail. Upon reading this chapter, the readers will understand the core principles shared within all electrochemical synthetic methods and at the same time keep up with the latest progress in the evolution of these techniques in meeting the renewed requirements in designing functional nanostructured catalytic thin films.
3.2 Principle of Electrochemical Method in Fabricating Thin Film Electrochemical processes have been extensively used for preparing thin films with their unique advantages in scalable production and ability to form films with precise control of thickness and its homogeneity [14]. Based on the principle of electrochemical processes, thin films made of metal, simple metal oxides, or polymerized organic film can be formed using anodization, cathodic electrolytic deposition, electrophoretic deposition, electro-oxidative polymerization, and combinatory methods. All these mentioned methods are operating based on the manipulation of electrons induced by a simple power supply or potentiostat with various functionalities. In principle, electrochemical method is a versatile technique because it involves only cost-effective and basic apparatus setup as shown in Figure 3.1. A simple electrochemical setup consists of two electrodes (i.e. anode and cathode, with an optional separate reference electrode), a power supply or potentiostat, and a single or separated electrolyte medium. In a typical electrochemical process, driven by the applying voltage, electrons flow from anode to cathode through an external circuit. Oxidation–reduction (redox) reactions occur when electrons are withdrawn from anodic site (oxidation happens at anode) to reach
3.2 Principle of Electrochemical Method in Fabricating Thin Film
Potentiostat
A
Anode
Cathode
V
Reference electrode
Figure 3.1 Schematic drawing of a general electrochemical setup with basic components used for electrochemical synthesis of nanostructured thin films. In this specific configuration, the cathode (on which the reduction reaction takes place) is the working electrode, with potential applied with respect to the reference electrode. The anode (on which the oxidation reaction takes place) in this case is the counter electrode. (See online version for color figure).
Electrolyte
the cathodic site to reduce substances (reduction happens at cathode). The potential can be manipulated with respect to the reference electrode in the case of a three-electrode configuration (as shown in Figure 3.1) or simply across the anode and cathode in a two-electrode configuration to control the extent or vigor of such redox reactions. As it is essential to have both reduction and oxidation to occur simultaneously to complete the electron-flow circle, a redox reaction can also be called as two half-reactions, i.e. one representing the oxidation process and the other reduction reaction, respectively. Basically, the foundation of electrochemical fabrication of thin films lies in the manipulation of such redox reactions at either anode or cathode of electrochemical setup. The techniques can be grouped into anodization (when the substrate of interest undergoes oxidation reaction), cathodic electrodeposition (when the substrate of interest undergoes reduction reaction), and electrophoretic deposition (attractions of oppositely charged particles onto the substrate) [15–17]. Recent developments also see the adoption of combinatory methods integrating electrochemical methods with other chemical or physical ways in preparing complex material thin film with binary, ternary, and multicomponent structures [18]. Sections below will introduce the working principle of anodization, cathodic electrodeposition (sometimes it is called electroplating), electrophoretic deposition, and
41
42
3 Electrochemical Synthesis of Nanostructured Catalytic Thin Films
some examples of combinatory methods. The resultant catalytic films with properties unique to the method will be discussed. 3.2.1
Anodization
Anodization is usually referred to the electrochemical oxidation of metallic thin films. It is a useful tool for the creation of catalytic oxide films with the focus in recent years on array nanostructures of nanotubes and nanorods. When a metallic foil is used as an anode in an electrolyte containing oxide etching agent, nanostructured metal oxide is systematically formed when anodic voltage is applied between the metallic anode and the cathode (as counter electrode). During anodization, the positive voltage applied from the potentiostat drives the electrons away from the metallic foil results in the oxidation of metallic foil (Eq. 3.1). The formation of this anodic oxide layer of metal, however, faces the competition of chemical/electric-assisted dissolution of the formed oxide layer. This dissolution of metal oxide is promoted by the presence of oxide etching agent in the electrolyte (Eq. 3.2), in which fluoride ions are the typical etching dissolution agent: Met (Metal) + 2H2 O → Met-O2 (Metal oxide, assuming 4 + valence state) + 4H+ + 4e− Met-O2 + 6F− + 4H+ → Met-F2− 6 + 2H2 O
(3.1) (3.2)
The growth of the anisotropic nanostructures of metal oxide is driven by the competition between anodic oxidation and chemical/electric field-assisted dissolution [19]. As an example, Figure 3.2 shows the growth mechanism of titanium dioxide (TiO2 ) nanotube arrays during anodization. When anodization of titanium (Ti) foil starts, dense and thin TiO2 layer are quickly formed (Figure 3.2a). Subsequently, this thin and dense oxide layer will undergo localized dissolution (induced by the oppositely charged fluoride anions in the electrolyte and the weakened outer layer of surface metal–oxygen bond) to initialize the small pores formation (Figure 3.2b). The continuous thinning of barrier oxide layer through the localized dissolution leads to an increasing electric field intensity across the barrier oxide layer to deepen the pores further (Figure 3.2c). The barrier oxide layer serves as the resistance to the flow of reactive ions that are needed to be transported through the anodic oxide layer to sustain the oxidation. Formation of anodic oxide layer ceases when the resistance is increased too high (accompanied by thick barrier oxide layer). However, this issue can be overcome by increasing the applied potentials for a higher electric field or by using a higher concentration of fluoride ions in the electrolyte to allow the creation of internal channels. These channels facilitate the uneven internal resistance to maintain the ongoing oxidation process. The thickness of the barrier oxide layers underneath the pores is kept in an equilibrated state between the two competitive reactions, i.e. the thickness is reduced by dissolution but simultaneously regenerated by oxidation (Figure 3.2d). The electric field distribution at the pores boundaries causes anisotropic widening and deepening
3.2 Principle of Electrochemical Method in Fabricating Thin Film
barrier layer (b)
(a) void
pore
void
(c) tube
Ti + 2H2O → TiO2 + 4H+ + 4e– Anodic oxidation formation TiO2 + 6F – + 4H+ → TiF62– + 2H2O
(d)
(e)
Field-assisted dissolution
Figure 3.2 Schematic diagram of the growth mechanism for anodized metal foil: (a) growth of thin and compact oxide layer, (b) the initial formation of pores, (c) formation of patterned pores within oxide layer, (d, e) continue growth of the pores to form nanostructures.
of pores. As the pores formed at deeper region, the electric field closer to the underlayer metallic regions increases, thus promoting the anodic oxide growth. Afterward, the formation of anisotropic nanostructures of oxide continues to grow in equilibrium. The growth of such nanostructure will continue until the oxidation rate at the metal–oxide interface equals the chemical dissolution rate at the oxide/electrolyte interface. As the schematic formation of nanostructure during anodization is explained in Figure 3.2, the presence of those processes can be observed in the measurement of current transient pattern during anodization. Figure 3.3 shows a typical transient current profile measured during the reaction under constant applied anodization voltage. The transient current consists of three main stages. In the first stage, the current decays rapidly during the first few minutes of the anodization. At the start of the reaction, electrolyte is in direct contact with the conductive metallic surface, and thus the starting current level is usually high. Under the applied anodization, the oxidation of metallic foil occurs almost instantaneously to form a compact thin oxide surface layer. As the initial oxide layer increases in thickness, so does the electrical resistance of the substrate. As a result, the oxidation of metallic foil slows down, which is reflected by the rapid drop of current density. Sequentially, chemical dissolution of oxide layer by fluoride ion in the electrolyte starts to take place and render more underlayer metal to oxidation; a temporary bounce-back in current is therefore observed in the second stage. The third stage is the equilibrium stage in which the formation and dissolution of oxide happen at the same competitive rates, reflected by the steady-state current generation. As the growth of anodic nanostructure is regulated by the formation and dissolution of oxides, the readers can logically expect the important synthetic conditions to be related to the magnitude of applied voltage (strength of oxidation process) and the type or concentration of etching agent (e.g. fluoride ions, which directly influence the strength of dissolution process). Indeed, most of the morphological controls over the anodic oxide thin films are a result of optimal modulation of these two parameters (voltage and electrolyte composition).
43
Third stage
Second stage
First stage
3 Electrochemical Synthesis of Nanostructured Catalytic Thin Films
Current density
44
Anodizing time
Figure 3.3 Typical current profile under a constant applied anodization voltage of metal foil in organic electrolyte containing fluoride ions.
300 nm (a)
500 nm (b)
600 nm (c)
Figure 3.4 SEM images of the simple metal oxides obtained through anodization: (a) titanium dioxide (TiO2 ), (b) molybdenum trioxide (MoO3 ), and (c) tungsten trioxide (WO3 ). Source: (a) Reprinted with permission from Yun et al. [20]. Copyright 2011, American Chemical Society. (c) Reprinted with permission from Ng et al. [21]. Copyright 2010, American Chemical Society.
A number of industrially important metal oxide thin films have been prepared through anodization method such as alumina, silica, and titania. Many more academically interesting simple metal oxide films can be afforded by this method. Figure 3.4 shows scanning electron microscopy (SEM) images of metal oxides with anisotropic nanostructures obtained through anodization [20–22]. Some of the most important ones as relevant to catalytic applications include TiO2 and α-Fe2 O3 nanotubes, Nb2 O5 nanorods, WO3 nanoflowers, and MoO3 truncated rhombohedra. These oxide materials are intrinsic semiconductors with or without nanostructured morphologies, but in the case of the former, their efficiencies as (photo)electrodes can be further augmented. The effect stems from (i) the increased ratio of surface area to volume for electrolyte contact and where reaction takes place, (ii) enhanced light absorption due to trapping of photons within pores, and (iii) the vectorial charge transport.
3.2 Principle of Electrochemical Method in Fabricating Thin Film
The vectorial charge transport is an intriguing phenomenon that relies on the vertically oriented 1D array of the oxide semiconductor and hence deserves special mention here. Under photoexcitation, i.e. when the semiconductor photoanode is exposed to photons with energy equal to or greater than its band gap (see Chapters 11, 31, and 36 on the basics of photocatalysis), the generated photoelectrons would need to diffuse to the back of the electrode within its charge carrier lifetime, or they will recombine with the photoholes, hence the loss of photocharge for surface reaction. The photocharge transport can be described by the following equation: √ Lc = Dc 𝜏c where Lc is the diffusion length or distance traveled by the charge carrier (electrons or holes) before recombination, Dc is the diffusion coefficient of the charge carrier, and 𝜏 c is the lifetime of the charge carrier. It should be noted that the Dc (and hence Lc ) is different for both electron and hole even on the same semiconductor material. For a photoanode (or photocathode) that is composed of irregular-shaped or randomly packed particles, the photoelectrons (or photoholes) undergo the “random walk motion” that are rarely the most straightforward path to the back of the electrode. With the creation of 1D array of the oxide semiconductor, the diffusion of the photoelectrons is restricted to the shortest vertical path to the back of the electrode. This enables a large fraction of photoelectrons (or photoholes in the case of photocathode) to be collected within their 𝜏 c . At the same time, it is important to restrict the wall thickness/diameter to not more than twice the Lc of the photoholes (or photoelectrons) such that majority of them could diffuse to the semiconductor surface to catalyze the oxidation (or reduction) reaction. Owing to their advantages, aligned nanotube and nanorod arrays have been used for water photoelectrolysis and the reduction of CO2 . 3.2.1.1
Pulse or Step Anodization
To control the growth of internal channel as mentioned above, a number of strategies external to anodization (dominant by lithography-based method) have been explored. This includes electron beam lithography, focused ion beam lithography, colloid sphere lithography, and direct laser writing lithography. To search within electrochemical means, pulse or step anodization can offer alternative solutions. This approach is built on the understanding that the development of pore channel can be guided by surface texturization at the initial stage of anodization. Periodic nanostructures are known to grow along the internal channel under anodization. When pulse or step is introduced periodically during anodization, acid anions (such as fluoride ions) are intermittently transported or attracted to the layer near the top of pores. Therefore, compensation of such anions would be established from top to the channel bottom in a periodical manner, which results in the morphological variance of the formed anodic oxide thin film. In literature, it is reported that a low frequency voltage (period 𝜏 in minutes) permits the growth of bamboo-type nanotubes and 2D nanolace sheets of alumina [23], while high frequency current oscillation (period 𝜏 in seconds) yields the irregular nanoporous morphology. [24] A more advanced anodization setup involving multiple steps
45
46
3 Electrochemical Synthesis of Nanostructured Catalytic Thin Films
has been explored to precisely control pore channels. Although not yet popular, approaches of changing the type of electrolyte (or concentration) after each step, modulating reaction temperature, and varying the applied voltage or current during anodization have also been explored [25–27]. Although overall aim is to achieve rapid reaction to shorten the catalyst preparation time (industrial’s preference), an over-reactive anodization with ultrafast growth rate can cause concerns, i.e. the physical breaking of nanostructures. Debris of nanotubes has been reported physically lying on the thin films due to the breakage. Unwanted delamination of metal oxide nanostructures from the substrate can happen. Therefore, stabilizing the anodization reaction at ultrafast metal oxide growth rate condition is one of the main aspects in improving the method. Depending on the selected catalytic reaction, amorphous or crystalline structure is desired. As-anodization thin films usually possess lower degree of crystallinity. Additional crystallization process for the anodized thin film is sometimes needed to suit the targeted catalytic reactions. 3.2.2
Cathodic Electrodeposition
Cathodic electrodeposition is another well-established electrochemical method in fabricating thin films of nanocrystalline metals, alloys, and composite catalysts, where instead of forming the desired oxide layer on the anode as demonstrated in the previous section of anodization method, the targeted materials are now deposited on the cathode through electroreduction of metal cations dissolved in the electrochemical bath. The electrochemical configuration is as shown in Figure 3.1, where precise control over the thickness and composition of the catalyst layer can be achieved either through the control of applied voltage (potentiostatic method) or current (galvanostatic method). It is a low-temperature synthesis that is typically performed at ambient temperature. The potentials needed to drive the reduction of metallic cations depend on the redox potential of such targeted metal. In electrodeposition of more than mono-component, the element with positive reduction potentials will be reduced first as it can receive electrons relatively easier, although the reduction kinetics can be different. However, in principle, a convenient scanning of cathodic voltage over a window of voltage would be sufficient to identify the required potentials. Electrodeposition is considered a strong method of coating because of their abrasion resistance, hardness, coating adhesion, and corrosion resistance. Furthermore, modulation of the abovementioned properties can be performed through a number of experimental variations including temperature of medium, precursors’ concentration, pH of electrolyte, and current density. The higher thickness of coating can also be achieved by simply lengthening the duration of deposition. Because it follows the Faraday’s law for electrolysis, a deposited metal (Met) started from its cations (Metz+ ) under constant current density (I) should follow this relationship: dx∕dt = (MI (𝜑)∕𝜌AzF)
(3.3)
where x is the film thickness, M is the molar weight of materials, (𝜑) is current efficiency, 𝜌 is density of the layer, A is deposited space, z is the number of transferred
3.2 Principle of Electrochemical Method in Fabricating Thin Film
electrons, and F is the Faraday constant. Assuming current efficiency to be 100% with a known current density and knowing the electrode ambience, the film thickness can be established by deposition time. From an engineering perspective, cathodic electrodeposition is extremely versatile and flexible because there are many different metals, alloys, and composites that can be readily deposited in a variety of forms of thin film (e.g. coatings and freestanding sheet). 3.2.2.1
Pulse Electrodeposition
Besides the conventional direct current (DC) electrodeposition as discussed above, pulse current (PC) and pulse reverse current (PRC) methods are other modes of electrodeposition that are found useful in yielding nanostructured thin films with enhanced features. Figure 3.5 depicts the modulation of pulse condition in conventional direct current electrodeposition (non-pulse), PC, and PRC electrodeposition methods. Empowered by modern electronics and microprocessor, applied current waveform can be conveniently introduced. PC and PRC are among the most typical waveforms generated. In PC, cathodic pulse is followed by a period of time without current, while cathodic pulses followed by anodic pulses are defined as PRC. In terms of advantages in the resultant deposits, in brief, PC method affords a more compact deposit with fewer defects of pores or cracks, while PRC method can decrease internal stresses of the deposits. The technical details and advantages PC and PRC over direct current and the relationships of the deposit properties to the pulsing parameters such as frequency and current distribution have been reported previously [28]. Example of catalytic thin films made by pulse technique includes a low loading and Figure 3.5 Different modes of current density for electrodeposition (a) direct current (DC), (b) pulse current (PC), and (c) pulse reverse current (PRC).
Current, I +
– (c)
+
(b)
+
(a)
Time, t
47
48
3 Electrochemical Synthesis of Nanostructured Catalytic Thin Films
evenly distributed Pt–Co cathode catalyst on a Nafion-bound carbon layer [29]. The pulse-deposited Pt–Co catalysts are homogeneously coated on the carbon electrode at a thickness of around 6 μm, four times thinner than conventional Pt–C, and were found to improve proton-exchange membrane fuel cell (PEMFC) performance. Another noticeable feature of pulse electrodeposition is its ability to yield a conformal coating on nanostructured substrate. The emerging use of 1D nanotube or nanorod thin film has introduced new challenge in the coating of such an uneven surface at microscopic scale. Conventional direct current method always results in a solid layer electrodeposited on top of the nanotubular layer without penetration into the bottom region of the nanotubes. The limitation in conformal coating by normal direct current is due to the rapid reduction of the metallic cations at the surface (or entrance) of the nanotube/nanorod arrays. The resultant deposit formed initially on the top region continues to grow and eventually blocks the subsequent reduction of cations. Introduction of PC in the deposition involving nanotube/nanorods overcomes this weakness. During cathodic pulse, metallic cations are reduced on the surface of nanotubular substrate, while the following zero-current pulse allows the relaxation of the deposition that allows the fresh precursor to diffuse into the deeper region of such nanostructure for the subsequent cathodic pulse. The repeating deposition and relaxation cycles eventually allow the entire surface of nanotube/nanorod to be deposited by the target metals from the top to the bottom region [30–32]. Conformal coating of nanostructured catalytic thin film is a critical aspect. When the catalytic active material is not stable in the reaction medium (e.g. the catalyst will go through chemical or photo-dissolution during reaction), the conformal coating helps to suppress or avoid this deactivation challenge by providing a conformal protective layer. Cuprous oxide and zinc oxide are examples of unstable catalytic materials. Pulse electrodeposition of protective layers on the nanostructures of cuprous oxide and zinc oxide has been found useful in extending their usable lifetime. 3.2.3
Electrophoretic Deposition
The term “electrodeposition” is sometimes misused in representing the electrophoretic deposition as both processes involve the use of applied direct current. Electrodeposition is based on solution of ionic species, and the deposition is achieved when ionic reduction occurs. Electrophoretic deposition, on the other hand, is a colloidal processing method in which a suspension of particles in a solvent is used. The deposition is driven by the attraction (or repulsion) of charged particles to the charged electrodes using galvanostatic or potentiostatic condition. Both positively and negatively charged colloidal particles can be deposited using electrophoretic method. For positively charged particles, they are attracted to the negatively charged electrode (cathode), and there the deposition occurs on the cathode. This process is called cathodic electrophoretic deposition. Vice versa, when deposition of negatively charged particles takes place at positively charge electrode (anode), anodic electrophoretic deposition is the term. As the readers can imagine, if the surface charge on the particles can
3.2 Principle of Electrochemical Method in Fabricating Thin Film
–
+
–
+
+
– + –
+
+
–
–
+ – + Cathode (a)
– Anode
Cathode (b)
Anode
Figure 3.6 Schematic illustration of electrophoretic deposition process: (a) cathodic electrophoretic deposition and (b) anodic electrophoretic deposition.
be tuned into positive or negative region, which is feasible by suitable surface modification, the particles can be deposited using either cathodic or anodic electrophoretic deposition. Figure 3.6 shows the schematic illustration of the two electrophoretic deposition processes. Unlike electrodeposition, the electrophoretic deposition does not require the liquid medium to be conductance, and in fact organic solvent is a preferred medium. Water is not an ideal medium in electrophoretic deposition because the applied voltage may cause the water splitting reaction that produces hydrogen and oxygen gases at the cathode and anode, respectively. The rigorous evolution of gases could adversely affect the formation of deposit on the electrodes. Because the deposition is greatly determined by the surface charge of particles, this technique offers versatility in depositing complex compounds and composites as long as surface charge can be modulated. One potential weakness, however, is the absence of particle and electrode reactions in electrophoretic deposition. The deposited materials do not lose their charge upon deposition, and therefore a reversal of electric field can result in delamination of the deposited layer. It is important to carefully select similarly charged particles and similar solvent–binder–dispersant systems for having good control over layer thickness. As the mechanism of electrophoretic deposition involves movement of charged particles in liquid suspension driven by an applied voltage, the important parameters governing the deposition process are therefore related to the regulation of suspension and the physical parameters of the process (e.g. voltage, deposition time, etc.). For the suspension containing the targeted particles, many physicochemical natures of both particle and the liquid medium must be considered. Apparently, particle size suitable for electrophoretic deposition is an important factor. Although there is no general rule of thumb to specify ideal particle size for electrophoretic deposition, it is crucial to have a high dispersion and good colloidal stability to ensure even deposition. Colloidal particles with size smaller
49
50
3 Electrochemical Synthesis of Nanostructured Catalytic Thin Films
than 1 μm in diameter tend to remain in suspension stably owing to the Brownian motion. Therefore, in general, it is the main problem for larger particles as they tend to settle by gravity over shorter period. When the sedimentation of particles occur faster than the electrophoresis-induced particles mobility, uniform deposition is unlikely: thinner coating on the top region while thicker deposit at the bottom of the substrate. However, this does not necessarily mean that large particles could not be deposited. Colloidal stability of large particle can be improved by having a very strong surface charge, and the electrophoresis movement can be increased with larger electrical double layer region. Selection of liquid medium is another factor governing the quality of electrophoretic deposition. Dielectric constant of the liquid medium is highly correlated with the deposition performance. Deposition fails in the liquid medium with too low dielectric constant because of weak dissociative power, while electrophoretic mobility is significantly reduced when a medium with high dielectric constant is used due to the reduced size of double layer region. Consequently, liquids of low dielectric constant (but not too low) are a favorable condition. As mentioned earlier, stabilization of suspension, i.e. colloidal stability, is critical in facilitating smooth deposition, and this stability can be achieved by having high and uniform surface charge on the particle. The ζ potential of particle becomes one of the key factors in the electrophoretic deposition. The magnitude of the ζ potential determines the stability of the suspension (repulsive interaction between particles), while the polarity of the ζ potential (positively or negatively charge) decides the direction and migration velocity of the particles during electrophoretic process. The probability of coagulation of the suspended particle must be minimized by modulating the two forces with opposite effects: electrostatic and van der Waals forces. High electrostatic repulsion attributed to the high ζ potential (high particle charge) is desired to suppress agglomeration. Conveniently, the ζ potential can be tuned by carefully selecting pH and additives to the suspension. 3.2.4
Combinatory Methods Involving Electrochemical Process
Although the electrochemical principles discussed in Section 3.2 are sufficient to guide a variety of thin film synthesis, recently there are more examples adopting a combinatory approach to prepare new functional thin films. The combinatory approaches can integrate two electrochemical methods or combine an electrochemical method with other technique to synthesize thin film with unique properties. As discussed in the earlier section, anodization is an effective method in producing nanostructured thin films made of simple metal oxide. A wide variety of simple metal oxides such as Al2 O3 , TiO2 , WO3 , SiO2 , MoO3 , and NiO fabricated by anodization process has been reported. Relatively rare is the fabrication of ternary or complex oxide using anodization. Composites of metal oxide decorated with other functional materials are also generally not being reported using anodization process alone. In this section, the design of metal oxide-based composite and the preparation of ternary oxide thin film using a combinatory electrochemical method will be discussed.
3.2 Principle of Electrochemical Method in Fabricating Thin Film
3.2.4.1
Combined Electrophoretic Deposition–Anodization (CEPDA) Approach
Combined electrophoretic deposition–anodization (CEPDA) is a combined electrochemical process that developed to allow the simultaneous anodic growth of metal oxide and the electrophoretic-driven deposition of reduced graphene oxide (RGO) on the resulting anodic oxide [18]. Metallic foil is employed as the anode in a typical anodization setup as shown in Figure 3.7. Formation of TiO2 nanotube (TNT) arrays following the typical anodization route is expected. However, the organic electrolyte of this anodizing setup is suspended with particles/flakes of RGO with negative surface charge. Within the anodization duration, electrophoretic forces drive the negatively charged RGO to the TiO2 anode to form TiO2 –RGO composite that find uses in photocatalytic reactions. The RGO deposited on TNTs using CEPDA method has a stronger physical interaction between host and guest compared with the typical coating techniques such as spin, drop, and dip coating. This is a demonstration of preparing composite materials by combining two electrochemical working principles in a single experimental process. Otherwise, this kind of composite thin film can be prepared by multistep process. In this combinatory method, both the growth of oxide and its integration with secondary component can take place simultaneously. –
Pt counter electrode (Cathode) – –
DC power supply (20–60 V)
–
Negatively charged RGO
– – –
–
– –
– –
–
–
– – – – – – – – – – – – +
Ethylene glycol-based electrolyte containing RGO sheets and NaF
+ + + + + TI working electrode (Anode)
– – – – – – – – – – – – – – – – – – – +
+ +
+ + +
+
+
+
Growth of RGO-deposited TNT array
Figure 3.7 Schematic illustration of the apparatus for combined electrophoretic deposition–anodization (CEPDA) approach in the preparation of TiO2 nanotube (TNT)–reduced graphene oxide (RGO) composite. Source: Yun et al. 2012 [18]. Reproduced with permission of Royal Society of Chemistry. (See online version for color figure).
51
52
3 Electrochemical Synthesis of Nanostructured Catalytic Thin Films
Although anodization is, in general, limited to the synthesis of simple oxide thin film, by combining the as-anodized thin film (before rigid crystallization) with hydrothermal/solvothermal treatment, transformation of the simple metal oxide into complex ternary oxide is possible [18, 33, 34]. Depending on the metallic substrate, bare anodization process usually yields the metal oxide with amorphous phase or partially crystalline structure. Subsequent annealing at elevated temperature is needed to transform the as-anodized films into well-crystallized metal oxide that find applications in various reactions. This intermediate state of as-anodized film offers the possibility to further engineering their composition for final product. For example, upon anodization of tungsten foil, the formed layer is hydrated WO3 (WO3 ⋅2H2 O) with loosely packed structure. Subjecting this as-anodized WO3 ⋅2H2 O film into a hydrothermal treatment containing bismuth species (Bi2 O2 )2+ forces the interstitial water molecules to be replaced by (Bi2 O2 )2+ . Crystallization process takes place during the hydrothermal treatment, and bismuth tungstate (Bi2 WO6 ) thin film is obtained. With proper control, the rate and the depth for the insertion of Bi cations into host material (example is WO3 ) can be tuned. As a result, catalytic thin film with two Bi as dopant is made. More interestingly, the concentration of dopant (not limited to Bi) indicates gradient across interfaces. This gradient concentration has important and constructive impact on charge transfer. The combination of anodization with hydrothermal treatment has been proven applicable to prepare different types of ternary oxide films. However, detailed procedures should be strictly considered in this combinatory method. The successful transformation of simple metal oxide into ternary or complex metal oxide is based on the relatively amorphous nature of the as-anodized metal oxide film. Such transformation is not likely when the highly rigid or crystalline anodized films are employed due to the limited rearrangement of the lattice structure.
3.3 Conclusions and Perspective Electrochemical route is probably one of the most cost-effective synthesis methods of functional and nanostructured thin films because it does not require expensive or complicated apparatus. It also offers great versatility in fabricating thin film with various desired properties by simply modulating a few operating parameters. All electrochemical methods discussed in this chapter are based on the same principle of electrochemistry that the deposition is a result of redox reactions and the migration behavior of charged elements given an applied voltage or current (i.e. anodization, cathodic electrodeposition, and electrophoretic deposition). Changing the ways of voltage/current being applied to the system (e.g. through the introduction of pulse) can also have great influence in the quality of the deposition process. Current status of electrochemical thin film synthesis has seen the domination of catalytic materials with relatively simple composition. In many cases, metal oxides in the form of binary oxides are produced. Modulation of their morphology using electrochemical methods (as described in this chapter) and
References
the subsequent investigation on the impact of such morphological control offer new opportunity in improving the overall reaction kinetics. Recent trend in nanostructuring the catalytic materials in thin film configuration will continue. Electrochemical synthesis of nanostructured catalytic thin film can also offer the opportunities in exploring new catalytic materials beyond the currently dominant simple oxide (or other simple composition). Introduction of secondary component in the making of composite materials and the transformation of simple oxide into ternary/quaternary counterparts may become an apparent alternative in the near future to prepare complex functional catalytic thin films. The matrix of composition afforded by electrochemical synthesis of thin films might also be helpful in discovering new catalytic reactions.
References 1 Mehla, S., Das, J., Jampaiah, D. et al. (2019). Recent advances in preparation
2
3 4
5
6
7
8
9
10 11
methods for catalytic thin films and coatings. Catal. Sci. Technol. 9: 3582–3602. Lee, W. and Park, S.-J. (2014). Porous anodic aluminum oxide: anodization and template synthesis of functional nanostructures. Chem. Rev. 114: 7487–7556. Uchiyama, H. (2015). Evaporation-driven self-organization of sol–gel dip-coating films. J. Ceram. Soc. Jpn. 123: 457–464. Hunt, H.K., Lew, C.M., Sun, M. et al. (2010). Pure-silica zeolite thin films by vapor phase transport of fluoride for low-k applications. Microporous Mesoporous Mater. 128: 12–18. Karches, M., Morstein, M., von Rohr, P.R. et al. (2002). Plasma CVD-coated glass beads as photocatalyst for water decontamination. Catal. Today 72: 267–279. Singh, J. and Wolfe, D.E. (2005). Nano and macro-structured component fabrication by electron beam-physical vapor deposition (EB-PVD). J. Mater. Sci. 40: 1–26. Sarakinos, K., Alami, J., and Konstantinidis, S. (2010). High power pulsed magnetron sputtering: a review on scientific and engineering state of the art. Surf. Coat. Technol. 204: 1661–1684. Nistico, R., Scalarone, D., and Magnacca, G. (2017). Sol–gel chemistry, templating and spin-coating deposition: a combined approach to control in a simple way the porosity of inorganic thin films/coatings. Microporous Mesoporous Mater. 248: 18–29. Mechiakh, R., Sedrine, N.B., Chtourou, R., and Bensaha, R. (2010). Correlation between microstructure and optical properties of nano-crystalline TiO2 thin films prepared by sol–gel dip coating. Appl. Surf. Sci. 257: 670–676. Ojo, A.A. and Dharmadasa, I.M. (2018). Electroplating of semiconductor materials for applications in large area electronics: a review. Coatings 8: 262. Kay, A., Cesar, I., and Gratzel, M. (2006). New benchmark for water photooxidation by nanostructured Fe2 O3 films. J. Am. Chem. Soc. 128: 15714–15731.
53
54
3 Electrochemical Synthesis of Nanostructured Catalytic Thin Films
12 Brust, M., Bethell, D., Kiely, C.J., and Schiffrin, D.J. (1998). Self-assembled
13 14 15
16
17 18
19
20
21
22 23
24
25
26 27 28
29
gold nanoparticle thin films with non-metallic optical and electronic properties. Langmuir 14: 5425–5429. Ung, T., Liz-Marzan, L.M., and Mulvaney, P. (2001). Optical properties of thin films of Au@SiO2 particles. J. Phys. Chem. B 105: 3441–3452. Osaka, T. (1997). Electrochemical formation and microstructure in thin films for high functional devices. Electrochim. Acta 42: 3015–3022. Wang, K., Liu, G., Hoivik, N. et al. (2014). Electrochemical engineering of hollow nanoarchitectures: pulse/step anodization (Si, Al, Ti) and their applications. Chem. Soc. Rev. 43: 1476–1500. Xiao, F., Hangarter, C., Yoo, B. et al. (2008). Recent progress in electrodeposition of thermoelectric thin films and nanostructures. Electrochim. Acta 53: 8103–8117. Besra, L. and Liu, M. (2007). A review on fundamentals and applications of electrophoretic deposition (EPD). Prog. Mater. Sci. 52: 1–61. Yun, J.-Y., Wong, R.J., Ng, Y.H. et al. (2012). Combined electrophoretic deposition–anodization method to fabricate reduced graphene oxide-TiO2 nanotube films. RSC Adv. 2: 8164–8171. Prakasam, H.E., Shankar, K., Paulose, M. et al. (2007). A new benchmark for TiO2 nanotube array growth by anodization. J. Phys. Chem. C 111: 7235–7241. Yun, J.-H., Ng, Y.H., Ye, C. et al. (2011). Sodium fluoride-assisted modulation of anodized TiO2 nanotube for dye-sensitized solar cells application. ACS Appl. Mater. Interfaces 3: 1585–1593. Ng, C., Ye, C., Ng, Y.H., and Amal, R. (2010). Flower-shaped tungsten oxide with inorganic fullerene-like structure: synthesis and characterization. Cryst. Growth Des. 10: 3794–3801. Lou, S.N., Yap, N., Scott, J. et al. (2014). Influence of MoO3 (110) crystalline plane on its self-charging photoelectrochemical properties. Sci. Rep. 4: 7428. Albu, S.P., Kim, D., and Schmuki, P. (2008). Growth of aligned TiO2 bamboo-type nanotubes and highly ordered nanolace. Angew. Chem. Int. Ed. 47: 1916–1919. Lee, W., Kim, J.-C., and Gosele, U. (2010). Spontaneous current oscillations during hard anodization of aluminium under potentiostatic conditions. Adv. Funct. Mater. 20: 21–27. Sulka, G.D. (2008). Highly Ordered Anodic Porous Alumina Formation by Self-Organized Anodizing. Weinheim: Wiley-VCH Verlag GmbH & Co. KGaA. ISBN: 978-3-527-31876-6. Korotcenkov, G. and Cho, B.K. (2010). Silicon porosification: state of the art. Crit. Rev. Solid State Mater. Sci. 35: 153–260. Roy, P., Berger, S., and Schumuki, P. (2011). TiO2 nanotubes: synthesis and applications. Angew. Chem. Int. Ed. 50: 2904–2939. Chandrasekar, M.S. and Pushpavanam, M. (2008). Pulse and pulse reverse plating – conceptual, advantages and applications. Electrochim. Acta 53: 3313–3322. Saji, V.S. (2018). Electrodeposition in bulk metallic glasses. Materialia 3: 1–11.
References
30 Yun, J.-H., Ng, Y.H., Huang, S. et al. (2011). Wrapping the walls of n-TiO2
31
32
33
34
nanotubes with p-CuInS2 nanoparticles using pulsed-electrodeposition for improved heterojunction photoelectrodes. Chem. Commun. 47: 11288–11290. Tang, Y., Traveerungroj, P., Tan, H.L. et al. (2015). Scaffolding an ultrathin CdS layer on a ZnO nanorod array using pulsed electrodeposition for improved photocharge transport under visible light illumination. J. Mater. Chem. A 3: 19582–19587. Tang, Y., Wang, P., Yun, J.H. et al. (2015). Frequency-regulated pulsed electrodeposition of CuInS2 on ZnO nanorod arrays as visible light photoanodes. J. Mater. Chem. A 3: 15876–15881. Ng, C., Iwase, A., Ng, Y.H., and Amal, R. (2012). Transforming anodized WO3 films into visible-light active Bi2 WO6 photoelectrodes by hydrothermal treatment. J. Phys. Chem. Lett. 3: 913–918. Lou, S.N., Scott, J., Iwase, A. et al. (2016). Photoelectrochemical water oxidation using a Bi2 MoO6 /MoO3 heterojunction photoanode synthesised by hydrothermal treatment of an anodised MoO3 thin film. J. Mater. Chem. A 4: 6964–6971.
55
57
4 Synthesis and Design of Carbon-Supported Highly Dispersed Metal Catalysts Enrique García-Bordejé Instituto de Carboquímica (ICB-CSIC), Department of Chemical Processes and Nanotechnology, Miguel Luesma Castán 4, E-50018 Zaragoza, Spain
4.1 Introduction Activated carbon is typically the choice of industry as catalyst supports for noble metals for hydrogenation and oxidation in liquid phase due to its outstanding properties such as resistance to basic/acidic media, chemical inertness, high surface area, and the easy recovery of the noble metals by calcination. Carbon black, consisting of microporous and graphitic carbon, is the benchmark support for noble metals used in electrocatalysis due to its excellent conductivity (see for examples, Chapter 30 for electrocatalytic water splitting, and Chapter 32 for fuel-cell-related electrocatalytic reactions). More recently, development of carbon-based catalysts have been identified as key technology in enabling the future biobased economy due to its resistance to hydrothermal conditions typical of biomass conversion processes (see Chapter 33 for the conversion of lignocellulose to fuel and Chapter 34 for the valorization of carbohydrates) [1]. The inertness of some carbon materials endows them with certain advantages over oxidized supports such as keeping metal nanoparticles in more reduced state compared with other supports [2], low metal–support interaction, and preventing the deactivation by coke deposition. Despite the aforementioned advantages of carbon over other metal oxide supports, the preparation of supported catalysts on carbon is more challenging than on metal oxide supports due to the complex surface chemistry of the former. Owing to their rich surface chemistry, variations of carbon-based materials can be designed as metal-free catalysts. There are excellent tutorials and reviews about the preparation and use of carbon-based materials as metal-free catalysts or “carbocatalysts” [3, 4]. Therefore, this chapter does not cover the preparation of carbocatalysts and instead focuses on the design and preparation of metal catalysts supported on carbon. Supported catalysts on carbon have been commonly prepared based on the adsorption of metal precursor on carbon surface via equilibrium adsorption, incipient wetness impregnation, or dry impregnation. The adsorption of the precursor takes place on the oxygenated surface groups or within micropores. Heterogeneous Catalysts: Advanced Design, Characterization and Applications, First Edition. Edited by Wey Yang Teoh, Atsushi Urakawa, Yun Hau Ng, and Patrick Sit. © 2021 WILEY-VCH GmbH. Published 2021 by WILEY-VCH GmbH.
58
4 Synthesis and Design of Carbon-Supported Highly Dispersed Metal Catalysts
The latter one takes place due to the high adsorption potential of the pore walls in confined spaces. The process needs a subsequent reduction of the metal precursor either by thermal treatment in H2 gas or using a reductant in liquid phase (hydrazine, sodium borohydride, formaldehyde, etc.). However, these methods impart limited control over the particle size and interparticle distance. Recently, the interparticle distance has been confirmed as a key factor for the selectivity of certain reactions, and it is also crucial to prevent sintering [5]. New emerging techniques have been receiving intense interest for the preparation of single-site catalysts or metal clusters of defined particle size and interparticle distance. Single-atom metal catalysts (SACs) with full utilization of the metal centers have the potential to bridge the gap between molecular and solid-state heterogeneous catalysis. While the preparation of stable SACs with high loadings on conventional carbon supports is still challenging, the advent of new carbon allotropes is providing better alternative supports for single-atom catalysts. Section 4.2 deals with these novel carbon materials. The catalyst design involves the optimization of not only the catalytic performance, but also the physicochemical properties (surface area, acidity) and mechanical properties (strength, attrition, 3D structures) as well as the catalyst distribution within 3D structures. Along with the use of conventional techniques for the preparation of SACs or nanoclusters on novel carbon materials, new techniques are emerging as a result of new technical developments as described in Section 4.3.
4.2 Preparation of Catalysts on New Carbon Supports The method of deposition of catalyst precursors must be different for sp2 carbon than for sp3 carbon due to the different chemical nature of both carbon atoms. Conventional carbon materials such as activated carbon have a turbostratic structure that consists of a random mixture of sp2 and sp3 , which is difficult to control and quantify. Nowadays, new carbon materials have emerged with more defined structures and controlled sp2 or sp3 character. Carbon nanodiamonds are ideally pure sp3 carbon and are sometimes coated by a layer of sp2 carbon. By treating at high temperatures, nanodiamonds are converted into nano-onions with sp2 character. Activated carbon usually contains a high proportion of sp3 carbon, whereas carbon nanotubes (CNTs) and graphene are mainly sp2 , containing sp3 carbon atoms on defects and edges. Therefore, the proportion of sp3 /sp2 is variable depending on the amount of point defects and the size of the basal plane. The creation of well-defined carbon structures can enhance our precision control over the catalytic site location, interparticle distances, and metal catalyst size. In fact, the size of the carbon-supported catalysts can be controlled down to clusters or even SACs, which in turn influences the catalytic selectivities and efficiencies. In general, the anchoring of catalyst precursor is more difficult on sp2 carbon such as graphene than on sp3 carbon because the interaction is weaker in the former due to the “smooth” and “inert” surface of the basal plane. The interaction is stronger on the prismatic edges of basal planes (sp3 ), which accounts for only a minor fraction of the carbon material. The density of anchoring sites on graphene can be increased by doping, functionalizing, and creating defects or
4.2 Preparation of Catalysts on New Carbon Supports
dangling bonds on graphitic lattices. Another option is to use the highly oxidized form of graphene, called graphene oxide (GO). The metal precursor can be directly attached to these defects in graphitic lattices with dangling bonds or to heteroatoms (hydrogen, oxygen, nitrogen, sulfur, etc.). These heteroatoms can function as ligands to attach the single-metal catalyst. Furthermore, the good attachment of the metal to carbon or dopant atoms guarantees the stability of the catalyst under reaction conditions. The techniques of deposition of metal catalyst on different carbon materials are discussed in detail in the following sections. 4.2.1
Catalyst on Graphene Oxide
GO offers unique properties such as near atomic thickness, scalable syntheses, the possibility of processing using solution-based techniques, and facile deposition as nanothin films. Moreover, the primary properties such as flake size and quantity of oxygenated groups are tunable. The GO flakes are amphiphilic, containing variable quantities of aromatic domains besides the hydrophilic oxygenated groups. The most commonly accepted structure of graphene is based on the Lerf–Klinowski model (Figure 4.1) [6]. The model concludes that GO consists of two different randomly distributed domains: (i) pure graphene with sp2 -hybridized carbon atoms and (ii) sp3 -hybridized and oxidized carbon domains. In this model, the oxidized GO areas contain mostly epoxy and hydroxyl functional groups on the basal planes with carboxyl groups at the edges. The rich surface chemistry of GO is favorable as active sites for the design of metal-free catalysts (carbocatalysts) as well as for the preparation of composite catalysts with metals. GO surface interacts with other GO flakes or other nanomaterials by chemical interactions or physical interactions, i.e. hydrogen bonds, van der Waals forces, and dispersive forces (π–π interactions). This makes the preparation of catalysts more straightforward and versatile than for other carbon materials, although a high degree of control has not yet been achieved on GO. The other advantage is that GO is highly hydrophilic, thus allowing the use of aqueous media for impregnation. To prepare supported metal catalysts on GO while preserving properties of the latter (i.e. high oxygenated group content), one should ensure the reduction OH
HO2C OH
HO2C O HO2C
O
O HO
HO2C
O HO
OH OH OH O HO O OH O OH O OH OH O OH OH OH O HO HO C 2
CO2H
O CO2H
O
HO OH O OH HO HO HO
O
O
O OH CO2H
Figure 4.1 Structure of graphene oxide based on the Lerf–Klinowski model. Source: Dreyer et al. 2010 [6]. Reproduced with permission of Royal Society of Chemistry.
59
60
4 Synthesis and Design of Carbon-Supported Highly Dispersed Metal Catalysts
of the metal precursor but avoid the overreduction of GO. Overreduction of GO leads to reduced graphene oxide (rGO) support. Therefore, mild reduction methods should be used to preserve GO surface chemistry. The nucleation of noble metal nanoparticles can be carried out by hydrothermal treatment without any reductant, microwave heating, or using mild reductants such as glucose. Depending on the reduction temperature, the resulting substrate may be retained as GO (at 25 ∘ C) or converted to rGO (at 60 ∘ C) [7]. Alternatively, the reduction of pre-impregnated metal precursors can be carried out in the gas phase under flowing H2 at low temperatures (250 ∘ C). The main advantage of using GO as metal catalyst supports is that the oxygenated surface groups of GO naturally act as nucleation sites for nanoparticles, thus minimizing the aggregation. In that sense, the use of a capping agent that stabilizes bare metal nanoparticles can be mitigated. Moreover, GO can be functionalized with specific ligands such as dopamine that allow subsequent binding with metal nanoparticles [8]. The rich surface chemistry of GO makes it hydrophilic, and it can even catalyze side reactions. These reactions can be beneficial as a bifunctional catalyst in tandem reactions such as coupling followed by hydrogenation [9]. In the event that the reduction potential of the reductant is greater than that of pristine GO, another type of support, i.e. rGO, results, as will be explained in Section 4.2.2.1. To the best of our knowledge, the formation of single-metal sites on GO has not been described yet, probably because the excessive density of nucleation sites prevents the isolation of single sites and favors the aggregation to metal nanoparticles. 4.2.2 4.2.2.1
Catalyst on Graphene Graphene or rGO as Starting Material
Graphene has a number of favorable properties as catalyst support compared to other carbon materials. Graphene has a theoretical specific surface area as high as ∼2600 m2 /g, which is twice that of single-walled CNTs and much higher than those of most carbon blacks and activated carbons. This structure makes graphene highly desirable for potential applications as a 2D support for loading metal catalysts. The absence of micropores favors accessibility of reactants and desorption of products. Moreover, the locally conjugated structure endows graphene with enhanced adsorption capacities toward nonpolar aromatic substances, which can be beneficial for catalytic reactions involving such compounds. In contrast, graphene has low wettability in polar solvents, rendering aqueous-phase impregnation forbidden. Graphene materials can be obtained at a relatively low cost on a large scale by using graphite or graphite oxide and its derivatives as starting materials. The catalyst preparation can start from graphene or from GO that is subsequently reduced to rGO. The graphene materials are free from the metallic impurities that are almost inevitably present in CNTs, which is one of the drawbacks of the latter as catalyst supports. The superior electron mobility of graphene can facilitate efficient electron transfer during the catalytic reactions, thereby improving its catalytic activity. Finally, graphene has also high chemical, thermal, optical, and electrochemical stabilities, which can possibly improve the durability of the catalysts.
4.2 Preparation of Catalysts on New Carbon Supports
A wide variety of methods, such as hydrothermal procedure and microwaveassisted heating, have been developed for the synthesis of metal nanoparticles supported on graphene sheets. To prepare the supported noble metal nanoparticles, hydrazine, NaBH4 , and ethylene glycol are generally used as reductants. For transition-metal nanoparticles (Fe, Co, Mn), NaOH or aqueous ammonia is often used to hydrolyze metal salts (to metal hydroxide precipitates). Sometimes, reduction is carried out under high pressures using supercritical water or CO2 where the graphene support would act as a reductant (carbothermal reduction). Although small noble metal clusters (< 2 nm) could be formed directly on rGO without adding any reductant or capping agent [10], the alternative is to presynthesize the fine nanoparticles using the organometallic approach, and followed by the deposition on rGO and removal of capping agents. TrGO tends to have higher concentration of defective (heteronuclear) sites, which turns to be beneficial as the adsorption sites for metal cations. With its mildly reducing behavior, the rGO sheets subsequently act as electron donors for metal clusters to grow on its surface. The formation of small clusters only occurs when using rGO with low density of heteronuclear sites, where nucleation sites are distanced adequately. If the GO substrate is not sufficiently reduced, the rGO contains excessive nucleation sites and results in large nanoparticles. In a recent work [11], the reducing potential of graphite intercalation compounds (GICs) as precursors for graphenide solutions is used to deposit transition-metal nanoparticles and metal oxides on graphene under mild conditions and without the use of other reductants. Small Fe nanoparticles (2–5 nm) were prepared by this method. In general, the main problem encountered when using graphene as a catalyst support is that the basal planes lack the anchoring points for the metal and the metal sinters in the subsequent treatments of removal of capping agents or reduction. A nonpolar solvent or a mildly polar solvent is preferred for impregnation of the metal precursor. The advantage is that graphene is intrinsically reductive (albeit mildly) to circumvent the use of reductants. Sometimes, sacrificial stabilizing agents such as metal oxides are used to prevent the sintering of nanoparticles during thermal treatment [12]. Another option is to use GO to anchor a metal precursor and reduce both GO and the metal precursor simultaneously, i.e. GO reduced to rGO and metal precursor to metal nanoparticles. This approach is explained in Section 4.2.2.2. 4.2.2.2
Graphene Oxide as Precursor of Graphene-Supported Catalyst
This method is derived from the preparation described in Section 4.2.1, but the reduction is stronger and leads to the simultaneous reduction of both the metal precursor and GO. This is the most commonly used method to prepare metal/graphene catalysts due to the strong interactions of GO with the metal precursor and the possibility of carrying out impregnation in aqueous phase. A composite based on metal ions and GO is prepared first, followed by a harsh reduction treatment using a reducing agent either in liquid or gas phase. The final product consists of metal nanoparticles on rGO. When the reduction is carried out in liquid phase, nucleation and reduction of the metal take place simultaneously. This is the case of solvothermal reduction in liquid phase at moderate temperatures. Generally, the mixture of GO, metal precursor, and
61
62
4 Synthesis and Design of Carbon-Supported Highly Dispersed Metal Catalysts
reducing agent (e.g., hydrazine, ethylene glycol, NaBH4 ) is heated (100–150 ∘ C) by conventional heating or microwave, leading to the simultaneous reduction of GO that becomes rGO, resulting in a metal supported on graphene material. The simultaneous reduction of GO and metal in solution usually leads to a broad metal distribution or to relatively large nanoparticles (50–100 nm). The reduction can also be carried out in gas phase (e.g., H2 gas) after drying the pre-impregnated metal precursor on GO. The method involves several steps of impregnation, drying, and reduction, leading to nanometric particles, i.e. Pd and Pt nanoparticles of around 2 nm, whereas the size of Ni and Mn nanoparticles is between 4 and 6 nm [13]. The first step is the intercalation of a chosen metal salt into GO via impregnation, forming the metal salt anchored on GO oxygenated groups. In the second step, this composite undergoes an explosive reaction or “popping” at around 200 ∘ C, leading to the formation of large-surface-area graphene with well-dispersed, partially decomposed metal precursors strongly anchored on the graphene sheets. A special case is the preparation of catalyst nanoparticles on 3D macrostructures consisting of 2D graphene nanosheets with a foam-like structure, namely, graphene aerogels. When GO is reduced under solvothermal mild conditions, the graphene nanosheets self-assemble into 3D hydrogels, which can be dried subsequently by freeze-drying or supercritical CO2 to produce graphene aerogels [14]. These 3D structures are favorable as structured catalytic reactors due to the ease in handling and separation. It is very interesting that the formation of both metal nanoparticles and the graphene aerogel support can take place in a one-pot solvothermal synthesis using hot water as a mild reducing agent. The addition of divalent and trivalent ions (e.g. Ca2+ , Mg2+ , Cu2+ , Pb2+ , Cr3+ , Fe3+ ) in GO dispersion promotes the formation of GO hydrogels [15]. Fe2+ ions are anchored onto the oxygenated functional groups of GO and hydrolyzed at acid pH to give Fe oxide nanorods (60 nm) or precipitated as oxide nanoparticles (30 nm) at basic pH [16]. By adding a reducing agent, 10 nm Fe oxide nanoparticles wrapped by graphene aerogels could be prepared [17]. By using also the one-pot hydrothermal reduction method but adding some mild reductants such as sodium citrate and sodium acetate, the size of Fe3 O4 clusters was reduced to 5 nm on graphene aerogel [18]. Adding ascorbic acid as reductant and 100 ∘ C solvothermal treatment, 5–13 nm noble metal nanoparticles have been prepared on graphene aerogels[19]. Instead of using aqueous dispersion of GO, a GO dispersion in ethylene glycol and subsequent hydrothermal method was used to prepare 4–6 nm Ru nanoparticles on graphene aerogel [20]. It is possible to decouple the formation of the rGO hydrogel and the deposition of the metal using a two-step approach. By capitalizing on the advantage of high water content within the rGO hydrogel, an aqueous solution of the metal is infiltrated into the hydrogel. This approach has been used to introduce MoS2 in the hydrogel, which is subsequently freeze-dried [21]. 4.2.2.3
Graphene Derivatives: Doped Graphene and Synthetic Derivatives
Graphyne and graphdiyne are synthetic derivatives of graphene. They are synthetic flat single atomic layers consisting of carbon hexagons connected by linear carbon chains instead of only carbon hexagons as in graphene. In graphyne
4.2 Preparation of Catalysts on New Carbon Supports
structure, the hexagons are bonded by linear acetylenic chains, whereas in graphdiyne the hexagons are bonded by two acetylenic chains (Figure 4.2A–C). Graphdiyne is a new man-made carbon allotrope prepared from a molecular precursor (hexaethynylbenzene) that possesses uniform 18 C-hexagonal pores formed by three butadiyne linkages (-C between the benzene rings), which can provide ideal anchoring sites for SACs with high stability as demonstrated by the results of theoretical calculations and in experiments in hydrogen evolution reaction (HER) [22]. The research about these graphene derivatives as catalyst support is almost unexplored due to the novelty of the material. It is foreseen that these derivatives are more amenable to functionalization with SACs than their parent material graphene. Therefore, a research field remains open for new researchers.
4.2.3
Catalyst on Nanodiamonds and Onion-Like Carbon
Nanodiamonds are diamonds (carbon allotrope with sp3 hybridization) of 2–5 nm found in meteorites and interstellar dust [23]. Recently, nanodiamonds have been produced synthetically in the form of films and powders. A large quantity of nanodiamond powder has been successfully synthesized by the detonation of explosive carbonaceous mixture. In fact, carbon atoms in nanodiamonds do not have a purely diamond structure, rather they have an intermediary structure with sp2 and sp3 character, with a diamond-like core covered by an outer shell of graphitic/amorphous carbon (Figure 4.3). High-temperature annealing (>1750 ∘ C) of nanodiamonds could transform them into “carbon nano-onions” (Figure 4.3a). Annealing at lower temperatures gives rise to intermediate sp3 @sp2 core–shell structures. The outer graphitic layer is amenable to functionalization and doping, which has been exploited to prepare metal-free catalysts in several reactions. Moreover, nanodiamonds have been used as metal catalyst support, and the following paragraphs describe more in detail some of the few preparations of catalysts supported on nanodiamonds. As explained, annealing of nanodiamonds at increasing temperatures increases the thickness of the graphitic shell, ultimately transforming them into onion-like carbon (OLC). By exploiting this concept, some researchers prepared intermediate nanodiamond core/graphitic shells (ND@G) and OLC. Pt nanoparticles (1700 °C
Peroxide
Progressive polygonization
Anhdride O
R O
O O
Complete transformation to carbon onion
2 nm
1300–1500 °C Desorption of functional groups
sp2 shells
Ether O
R
Carbon onion shell formation
O
2 nm
700–800 °C
O OH Alcohol
ND core
Ester
900–1100 °C
R Ketone
Imine N R O
Desorption of functional groups
O O
N+
Nitro
Carboxylic acid O H2N Amine H 2N
Reconstruction of dangling bonds and amorphous carbon
Desorption of water and functional groups
Nitroso
N O–
Amide
2 nm
R
Lactone O
Continued sp3–sp2 phase transformation from the outside to the inside
2 nm
Alkene O R
O
O OH O O O
O
O H
H Hydroperoxide
Aldehyde
O Carbonate ester
0.21 nm
sp3 core
(a)
Functional groups
(b)
Figure 4.3 (a) Gradual transformation of nanodiamond to onion-like carbon at increasing annealing temperatures. (b) Schematic representation of nanodiamond. Source: Zeiger et al. 2016 [24]. Reprinted with permission of Royal Society of Chemistry.
4.2 Preparation of Catalysts on New Carbon Supports
4.2.4
SACs on Carbon Nitrides and Covalent Triazine Frameworks
In graphene doped with heteroatoms (N, O, S, P, etc.), the doped element can function as an anchoring site for metal or metal oxide, leading to the synthesis of hybrid organic–inorganic materials with high stability as a result of the strong binding between the metal species and dopant atoms. Graphitic carbon nitride (g-C3 N4 ) belongs to the family of carbon nitride compounds with a general formula near to C3 N4 (albeit typically with nonzero amounts of hydrogen) and two major substructures based on heptazine and poly(triazine imide) units (Figure 4.2D). Graphitic carbon nitride is usually prepared by polymerization of cyanamide, dicyandiamide, or melamine. Depending on reaction conditions, carbon nitride exhibits different degrees of condensation, properties, and reactivities. On the other hand, covalent triazine frameworks (CTFs) are structurally related to polymeric carbon nitride (Figure 4.2E). CTF is a high-performance polymer framework based on triazine with regular porosity and high surface area. It can be obtained by dynamic trimerization reaction of simple, economical, and abundant aromatic nitriles in ionothermal conditions. Primarily, these materials are large bandgap semiconductors, but their bandgaps are tailorable. For this reason, they are being investigated for photocatalysis. In addition, the loading of metal catalyst should be feasible, due to the presence of the abundant nitrogen atoms and voids within the structure. In fact, they are outstanding supports for SACs because they can stabilize metal ions or small metal nanoparticles even under harsh reaction conditions. Graphitic carbon nitride (g-C3 N4 ) has been proposed as support to coordinate metal (M–N2 ). This single-site catalyst has shown excellent performance for the oxygen reduction reaction (ORR)/oxygen evolution reaction (OER) electrocatalytic reaction and hydrogenation reactions. While the atoms on the alumina support are unstable and tend to aggregate, forming a Pd cluster, this does not occur for Pd SACs on mpg-C3 N4 , which exhibited more stable performance. Besides SACs, atomically precise clusters with two Fe atoms (Fe2 ) have been stabilized on g-C3 N4 [27]. The preselected metal precursor bis(dicarbonylcyclopentadienyliron) (Fe2 O4 C14 H10 ), containing two Fe atoms, ensures the formation of diatomic clusters, whereas mpg-C3 N4 provides abundant anchoring sites to stabilize the metallic species. A mild reduction process was selected (300 ∘ C in 5% H2 ), leading to a complete removal of organic ligands from the precursors and, at the same time, prevent agglomeration of the Fe2 clusters. For the sake of comparison, Fe SACs from iron porphyrin precursor and Fe nanoparticles were supported on g-C3 N4 following the same methodology but with different precursors. Biatomic Fe2 species exhibit highest activity in epoxidation, possibly promoted by the formation of reactive oxygen species. On the other hand, CTF also coordinated noble metal atoms by simply hydrothermally treating (60 ∘ C) the mixture of precursors. The 2 N atoms of bipyridinic regions can coordinate the molecular catalyst [28] and also metal atoms after the corresponding reduction [29]. Pt atoms coordinated to CTF have close similarities to the molecular Periana catalyst Pt(bpym)Cl2 , which is active and selective for the partial oxidation of methane via C–H activation in fuming sulfuric acid [30]. Therefore, Pt/CTF combines the advantages of homogeneous
67
68
4 Synthesis and Design of Carbon-Supported Highly Dispersed Metal Catalysts
catalyst, CTFs acting as a ligand, and the robustness of heterogeneous catalysts supplied by the rigid CTF network. The porous structure of CTFs can be also tailored. CTFs with varying pore size, specific surface area, and N content could be prepared varying the monomers, the linker, and the synthesis time [31]. Ru clusters on CTFs with a mesoporous structure provides highest conversion in the selective oxidation of hydroxylmethylfurfural (HMF) compared to other support materials such as activated carbon, g-Al2 O3 , hydrotalcite, or MgO. Moreover, CTFs have been shaped into spheres of a few hundreds of microns in diameter to increase their robustness [32]. CTFs can be used as support for SACs in electrocatalysis despite their modest conductivity. To increase their conductivity, CTFs have been hybridized with carbon nanoparticles [33]. SACs on CTFs are more stable for electrooxidation and electroreduction than homogeneous catalysts and immobilized organometallic catalysts, respectively, due to the rigid cross-linked structure of covalent bonds in CTFs [34]. Pt SACs have been also supported on CTFs leading to a performance comparable to commercial catalysts but with a reduction of Pt loading by one order of magnitude [35]. In summary, g-C3 N4 and CTFs are ideal to disperse single-atom catalysts in a stable manner for catalysis and electrocatalysis combining the ligand effect found in homogeneous catalysis and the robustness typical of heterogeneous catalysis. 4.2.5 Catalyst on Carbon Material from Hydrothermal Carbonization of Biomolecules Hydrothermal carbon is synthesized by the treatment of biomass or carbohydrate molecules under high pressure hydrothermal conditions [36]. The carbonaceous material produced has a spherical shape and diameters of a few hundreds of nanometers and contains high oxygen content (30–40%). It is characterized by hydrophilic external layers, while being more hydrophobic (bearing less oxygen) at the core. Metal catalysts have been prepared using the one-pot approach by introducing a metal precursor during the hydrothermal treatment. Depending on the metal precursor, the metal is deposited on the external hydrophilic surface or on the internal hydrophobic surface. Hydrophobic metal precursors such as Pd acetylacetonate tend to be reduced on the internal hydrophobic core of the carbon spheres, leading to the metal core–carbon–shell structure (Figure 4.4) [37]. The catalyst shows higher selectivity in the hydrogenation of phenol to cyclohexanone compared to the charcoal-supported catalyst. The enhanced performance of the former was attributed to hydrophilicity of the carbon shell. On the other hand, transition-metal ions (Fe3+ , Ni2+ , Co2+ , Ce4+ , Mg2+ , and Cu2+ ), which are less reducible, tend to bound to the hydrophilic shell of the carbon particles, leading to carbon core–metal shell structures upon calcination [38]. To the best of our knowledge, there are only a handful of examples in the literature on the use of hydrothermal carbon for catalytic applications, either as metal-free or hybrid catalysts with supported metals or metal oxides. For instance, hybrid inorganic–organic niobia–carbon catalyst has been prepared in one pot by hydrothermal carbonization of glucose, ammonium niobium oxalate,
4.3 Emerging Techniques for Carbon-Based Catalyst Synthesis
(a)
(b)
2 µm
200 nm
Figure 4.4 Hydrothermal carbon spheres (a) and noble metal@carbon core–shell structures (b). Source: Makowski et al. 2008 [37]. Reprinted with permission of Royal Society of Chemistry.
and urea [39]. Improved hydrothermal stability for aqueous-phase reactions was demonstrated for the highly dispersed niobia particles embedded within carbon. Besides the addition of active phase in one pot, metal active phase has been supported on previously prepared hydrothermal carbons in a second step. In some cases, the hydrothermal carbon materials are further carbonized in an inert atmosphere at high temperatures. In a recent work, hydrothermal carbon spheres were graphitized at 1900 ∘ C, on which cobalt Fischer–Tropsch catalyst was subsequently dispersed by both chemical vapor deposition (CVD) and wet impregnation [40]. In both cases, the primary particles had a mean size of 5 nm, whereas the catalyst particles prepared by wet impregnation were aggregated up to 100 nm. The former produced 5 times more oxygenates than conventional Co on alumina catalyst. The use of hydrothermal carbon in catalysis is still in its infancy. The hydrothermal carbon material has the potential as a catalyst support due to its natural origin, sustainable production process, and the ability to tune the carbon material properties (porosity, wetting properties, amphiphilicity, doping) for different reactions.
4.3 Emerging Techniques for Carbon-Based Catalyst Synthesis Figure 4.5 displays some of the most relevant techniques for the preparation of metal catalyst on carbon supports. One of the core objectives is to produce metal catalysts of uniform and defined particle size, whether as single atoms, clusters, or nanoparticles. It has been recognized recently that SACs provide higher activity and selectivity in several catalytic reactions that involve small molecules such as oxygen reduction, hydrogen evolution, methane activation, CO2 reduction, CO oxidation, or organic synthesis among others (see Chapter 6). Likewise, there are reactions that require large ensembles of atoms or nanoparticles. For example, cobalt catalyst size 5–6 nm is required for Fischer–Tropsch synthesis (FTS) to achieve optimum activity and selectivity toward large-chain
69
70
4 Synthesis and Design of Carbon-Supported Highly Dispersed Metal Catalysts
Strong electrostatic adsorption (SEA) Dry impregnation
Homogeneous deposition–precipitation
Wet chemical methods Engineering catalyst
Colloidal nanoparticle deposition
Clusters
Single-atom catalyst (SACs)
Nanoparticles Chemical vapor deposition
Electrodeposition
Physical methods Photodeposition Mechanical: ball milling
Atomic layer deposition
Figure 4.5 Relevant techniques optimized to prepare engineered catalysts on carbon materials.
hydrocarbons [41]. The deposition of presynthesized colloidal nanoparticles of well-defined size and compositions on carbon supports with controlled interparticle distances is a major challenge that will be addressed below. The goal is to enable catalyst engineering with close to atomic-scale precision to enable efficient heterogeneous reactions. 4.3.1
Deposition of Colloidal Nanoparticles
Recent advances in colloidal synthesis have enabled preparing monometallic or bimetallic nanoparticles (e.g. alloys, core–shell structures) with well-defined structures and sizes. These nanocrystals have ligands or capping agents that need to be removed prior to catalysis to increase the activity, although in some cases the ligands can be reaction promoters when chosen judiciously. In general, the two main challenges to prepare an efficient and practical catalyst from colloidal nanoparticles are: (i) to uniformly disperse the nanoparticles on a support and (ii) the subsequent activation by the removal of the ligand using thermal treatment or mild oxidation. Carbon supports such as CNT or graphene are ideal supports to deposit preformed nanoparticles since the absence of microporosity and open porosity favors the infiltration of the nanoparticles to all the deep surfaces of the carbon material. To disperse the nanocrystals uniformly on a carbon support, the particle–support and particle–particle interactions should be balanced. If the interaction of the particles between themselves is stronger than with the support, the particles will tend to aggregate. The amount of ligand and the temperature can tune the strength of interactions of the nanoparticles with the support as demonstrated in the assembly of Fe nanocrystals on CNT [42]. To develop an active catalyst and make the metal accessible to reactants, the capping agent must be removed by thermal treatment or mild oxidation. The problem is that in this step, the nanoparticles may begin to coalesce. Therefore,
4.3 Emerging Techniques for Carbon-Based Catalyst Synthesis
the particles can change their size and shape if the oxidation step is severe. Therefore, the ligand should be removed under milder conditions as possible. In principle, stabilization should be achieved more easily on a carbon material with a higher amount of defects and oxygenated surface groups than on undefective graphitic carbon materials because the former has more anchoring sites for the nanoparticles after removing the capping agent. The removal of ligands is even more difficult for transition metals than for noble metal nanocrystals since the ligands usually bind stronger to the former. 4.3.2
Single-Metal Atom Deposition by Wet Chemistry
Carbons doped with heteroatoms (e.g. N, B, P, S) have the ability to coordinate transition-metal atoms and thereby stabilizing them as single-site catalyst. To prepare single-atom sites, loading of metal deposits on the heteroatom-doped carbon substrate can be carried out using traditional wet impregnation, followed by acid leaching to remove nanoparticles and noncoordinated metal atoms. This leaves behind the strongly bound single metal atoms, each coordinated by the surrounding heteroatom dopants. In particular, the strong interaction between N-dopant and metals has been verified theoretically and experimentally using techniques X-ray photoelectron spectroscopy (XPS), near edge X-ray absorption fine structure (NEXAFS) [43], and transmission electron microscopy (TEM)–electron energy loss spectroscopy (EELS) [44]. The interaction is enabled by the delocalized free electrons of N on carbon. Their origins in turn can be explained by the electron back-donation, where the 𝜎-type electron from the filled orbital of the N atom is donated to the empty orbital of the metal atom, leading to a π back-donation from filled metal atomic orbital to the antibonding orbital of the N atom. 4.3.3 Immobilization of Metal Clusters and SACs by Organometallic Approach Mixed-metal clusters modified with appropriate ligands can be immobilized onto CNTs and graphene using covalent or noncovalent π–π interactions [45] (see Chapter 5 on metal clusters synthesis). Subsequent removal of the stabilizing ligand by thermal treatment may lead to a different extent of coalescence, which depending on the exposed temperatures can result in carbon-supported nanoparticles. When the preparation of SACs on carbon is sought, an interesting approach is to use organometallic precursors containing both the dopant element such as nitrogen and the metal. Macrocyclic complexes bearing transition-metal atoms similar to porphyrins, which contain both the dopant element and the metal, are excellent precursors to CUS or SACs. These macrocycles are noncovalently bound on graphitic materials by π–π stacking. The covalent anchoring of these macrocycles on carbon lattice upon pyrolysis is favored by the affinity of nitrogen atom for carbon materials. This will pave the way for stabilizing the catalyst under different solvents and under severe reaction conditions. Several macrocyclic molecules containing both metal and nitrogen atoms such as iron porphyrin or
71
72
4 Synthesis and Design of Carbon-Supported Highly Dispersed Metal Catalysts
(a)
(b)
(c)
(d)
Figure 4.6 Different precursors to prepare nitrogen-coordinated SACs. (a) Iron porphyrin, (b) iron phthalocyanine, (c) phenanthroline + Fe salt, and (d) 2,2-bipyridine + Fe salt.
macrocycles formed by mixing phenanthroline or 2,2-bipyridine and Fe salt have been used to prepare atomically dispersed Fe–Nx species on different carbon materials. Figure 4.6 shows some macrocycles used as precursors of single-atom metal–Nx species on carbon materials. The special ligand endows the metal with the electronic density necessary to perform a reaction selectively suppressing undesired collateral reactions. The catalysts prepared have shown high activity in ORR comparable to Pt catalyst with high selectivity in the four-electron reduction. 4.3.4
Chemical Vapor Deposition Techniques on Carbon Supports
CVD of organometallic precursors under oxidative conditions is a conventional technique for the deposition of nanoparticles on supports. Recently, atomic-layer deposition (ALD) has emerged as a particular case of CVD. In contrast to conventional CVD, precursors are introduced as a series of sequential and nonoverlapping pulses. In each of these pulses, the precursor molecules react with the surface and the reaction ends once all the reactive sites on the surface are utilized. Therefore, the amount of material deposited depends on the precursor–surface interaction and the number of ALD cycles. The ALD on pristine carbon supports such as graphene can be hampered by the limited interactions associated with basal planes, and ALD is only able to deposit material on defects and edges. Moreover, noble metal ALD on graphene is reported to lead to the formation of a mixture of atoms, clusters, and nanoparticles of Pt. Accordingly, in order to deposit single-atom metal sites on graphene by
4.3 Emerging Techniques for Carbon-Based Catalyst Synthesis
ALD, a previous step of functionalization with oxygenated groups is necessary. Only recently, ALD has been used to prepare single-atom Pt site on graphene functionalized with oxygen reactive sites [46]. By repeating the number of ALD cycles (exposure to metal precursor – O2 ), it is possible to go from single atoms to clusters of controlled size. 4.3.5 Simultaneous Formation of Metallic Catalyst and Porous Carbon Support by Pyrolysis The loading of SACs by wet chemistry methods depends greatly on the density of anchoring sites on the support, and it is usually low. To circumvent the limitation, pyrolysis (i.e. treatment at high temperatures under inert gas) of molecularly defined porous structures containing both the source of metal and carbon can lead to higher loadings. In fact, it would be interesting if the porous structure is retained in the final carbon material. Otherwise, a sacrificial template can be used to preserve the porosity upon pyrolysis. Metal–organic frameworks (MOFs; see Chapter 8 for more details), composed of periodic and porous networks of organic linkers with metal atom centers, are an excellent platform to prepare single-site catalysts on porous carbon in one-step pyrolysis. This approach has been used to attain transition-metal single sites on nitrogen-doped carbon [47]. 4.3.6
Dry Mechanical Methods
There is direct formation of noble metal nanoparticles (99% PSM yield in fully functionalized MOFs since the size of the pores decreases dramatically when introducing functions. Techniques such as vapor-phase PSM can maximize the yield [32]. An alternative approach is to dilute the function within the MOF by using MixMOF chemistry. In the example above, if one can synthesize a MixMOF-5 with only 30% amino groups, again “diluting” its functions, one can expect the yield of PSM to be high for the formation of the O–O chelate under analogous conditions. Although it is not thoroughly described within this chapter, PSM can be performed also with reactive sites on the inorganic unit [16, 30].
155
156
8 Design of Molecular Heterogeneous Catalysts with Metal–Organic Frameworks
8.2.2.2
Post-synthetic Exchange (PSE)
PSE has emerged as a valuable alternative to PSM and isoreticular/MixMOF chemistry in the last ten years. The idea is that one can substitute either linkers of the same size and connectivity by reacting the preformed MOF with the linker of choice or metal composition on the inorganic unit by metal-exchanging the MOF. Linker PSE has been applied to MOF-5, UiO-66, and MIL-53 topologies, among others [33], whereas metal exchange PSE is employed with many topologies including MOF-5, UiO-66, and MIL-101, and a review has covered extensively the topic [14]. One can produce the active site 4 by linker PSE – again, for simplicity with MOF-5 topology – by treating MOF-5 with linker 6 (Figure 8.6f ). Analogously to PSM, the PSE yield has a major influence on what material is formed. Alternatively, one can perform metal PSE with Fe to include it in Al or Cr MOFs with MIL-53 [34] or MIL-101 [35] topologies (Figure 8.6g). Yield is normally less than quantitative in PSE, and similar considerations about the formation of MixMOFs discussed in the previous PSM paragraph can be drawn.
8.3 MOFs in Catalysis A general description of what types of catalysis are possible with such materials is needed to give an overview of what catalytic processes can be envisioned. This paragraph is only introductory, and the focus will be on significant selected examples rather than a comprehensive description. We advise the reader to explore published reviews on the topic [25, 36–38], but with 6000 publications on catalysis by MOFs, one can safely say that MOFs can be used as catalysts for any liquid- and gas-phase reaction that is compatible with their chemical and thermal stability. MOFs are constituted by organic linkers and inorganic units that, when assembled together, yield a porous structure, which is essential for catalytic activity and selectivity. The active sites can be located at the organic linker, at the inorganic unit, and within the pores as shown in Figure 8.7. Organic linkers can be active sites if functional moieties are present. For instance, one can introduce amino groups active for amine organocatalysis such
Figure 8.7 Summary of the possible catalytic sites found and build with and within MOFs. (See online version for color figure).
8.3 MOFs in Catalysis
as IRMOF-3 used in the Knøvenagel condensation [39] or sulfonic acid groups in MIL-101(Cr) for reactions catalyzed by Brønsted acids [40]. It is also feasible to use whole metal complexes as organic linkers to build MOFs such as in the case of a Mn–salen complex, which catalyzes the enantioselective olefin epoxidation [41]. Organic functional groups can be versatile precursors for PSM to create a wide variety of organocatalytic and metal catalysts, leading to a basically infinite number of possible active sites. A very interesting and didactical approach is the Lewis acidic Fe active site described above for the UMCM-1 topology, which catalyzes the aldol condensation of aldehydes [31]. Organic linkers can also be exchanged to produce catalytic active sites such as in the case of the pillared ZnZn-porphyrin MOF, whereby the pillared Zn porphyrins can be exchanged with isoreticular linkers featuring different M-porphyrin moieties and used for the ring opening of epoxides with trimethylsilyl azide [42]. Inorganic units are the foundations that keep the structure and topology of the MOFs together. However, they can also have terminal solvent ligands that can be removed by heating leading to CUS such as those in HKUST-1, MIL-101, and MOF-74 [25]. CUS can be employed in Lewis acid catalysis, but they can also undergo PSM with a ligand leading to different catalytic activity such as that observed in MIL-101, which reacted with a pyridine amino organic molecule to yield basic active sites for the Knøvenagel condensation [43]. Metal PSE is a viable tool to tune the catalytic properties of existing MOFs or introduce new ones. MOF-5 has inert inorganic units since the Zn2+ atoms are saturated. The metal exchange of one of the Zn2+ atoms in the MOF with metals such as Fe2+ and Cr2+ produces redox active sites [44]. Inorganic units can also help electron transfer and lead to semiconductors with application in catalysis. Metal exchange of UiO-66 with Ti4+ can tune its photocatalytic properties leading to a mixed metal MOF with good photocatalytic properties [45]. The pores can accommodate molecular active sites that are either physisorbed within the pores of the MOF, i.e. without covalent interaction, or can produce nanoparticles on the MOF, which acts as a standard support [46]. As an example MOF additive can adsorb homogeneous catalysts and give enhanced performance such as in the asymmetric hydrogenation of olefins with Rh diphosphine complexes [47]. An alternative way to adsorb molecular complexes is to use the ship-in-a-bottle approach whereby molecules react within the pores of an MOF and assemble to give a molecule that is larger than the pores of the MOF itself [48]. An example is a metal porphyrin that is assembled within the pores of an indium imidazolate MOF leading to a non-leaching active catalyst for the oxidation of alcohols to ketones. 8.3.1 The Difference Between MOFs and Standard Heterogeneous and Homogeneous Catalysts One of the salient features of studying catalysis by MOFs lies in the potential to develop novel catalyst materials based on MOFs with properties that cannot be obtained in homogeneous or heterogeneous catalysts. What is clear is that they can bridge the gap between homogeneous and heterogeneous catalysis and feature well-defined molecular active sites while being easy to separate
157
158
8 Design of Molecular Heterogeneous Catalysts with Metal–Organic Frameworks
and to recycle. However, confinement effects within the micropores of MOFs can do much more. Many examples in the literature show how MOFs can be unique because their active sites are surrounded in pores that resemble pockets of enzymes. With this very brief commentary, we would like the reader to be aware of the possibilities that designing active sites within MOFs can bring: (1) MOFs as models to understand standard heterogeneous catalysts: An interesting aspect that should not be overlooked about MOFs is that they can be characterized at the molecular level with a wide range of techniques because of their unique structural and chemical characteristics. This makes MOFs suitable as models to understand the catalytic behavior of standard heterogeneous catalysts [49]. (2) Selectivity of catalytic reactions: There are also a small number of examples that make use of the restricted space in the pocket of an MOF to tune the selectivity of catalytic reactions [50–52]. One can change regioselectivity, invert enantioselectivity, and even change completely the reactivity of catalytic intermediates within the pores of an MOF. (3) Monodispersion of active sites: One major difference between having a molecular active site within a porous matrix instead of dissolved in solution is that in an MOF the sites are physically isolated and cannot undergo oligomerization. This enhances the stability of a catalytic system in which the oligomerization of the active site is a known decomposition pathway [53].
8.4 Conclusion: Where to Go from Here This chapter has focused the attention on MOFs, how to build molecular active sites with the numerous synthetic tools available, and how such approaches can yield interesting catalytic performance. There are many aspects that could not be covered in detail – like defect engineering [54], MOFs as precursor for well-defined carbon materials [55], and advanced characterization [19] – and for which we give reference to reviews that cover such topics. No matter how one thinks of catalysis by MOFs, it is clear that it is only a matter of time before catalysis will find industrial application, but this will happen only if one can find catalytic processes that cannot be realized either with existing homogeneous or with heterogeneous catalysts. It is envisioned that such relatively complex materials can aid in the fine chemical industry aimed at building complex molecules with high selectivity.
References 1 Furukawa, H., Cordova, K.E., O’Keeffe, M., and Yaghi, O.M. (2013). Science
341: 1230444. 2 Batten, S.R., Champness, N.R., Chen, X.-M. et al. (2013). Pure Appl. Chem.
85: 1715–1724. 3 Hoskins, B.F. and Robson, R. (1989). J. Am. Chem. Soc. 111: 5962–5964.
References
4 Chui, S.S. (1999). Science 283: 1148–1150. 5 Li, H., Eddaoudi, M., O’Keeffe, M., and Yaghi, O.M. (1999). Nature 402:
276–279. 6 Farha, O.K., Eryazici, I., Jeong, N.C. et al. (2012). J. Am. Chem. Soc. 134:
15016–15021. 7 Ferey, G., Latroche, M., Serre, C. et al. (2003). Chem. Commun.: 2976–2977. 8 Ferey, G., Serre, C., Mellot-Draznieks, C. et al. (2004). Angew. Chem. Int. Ed.
43: 6296–6301. 9 Ferey, G., Mellot-Draznieks, C., Serre, C. et al. (2005). Science 309:
2040–2042. 10 Cavka, J.H., Jakobsen, S., Olsbye, U. et al. (2008). J. Am. Chem. Soc. 130:
13850–13851. 11 Rosi, N.L., Kim, J., Eddaoudi, M. et al. (2005). J. Am. Chem. Soc. 127: 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26
27 28 29 30
31 32
1504–1518. Bennett, T.D. and Cheetham, A.K. (2014). Acc. Chem. Res. 47: 1555–1562. Eddaoudi, M., Kim, J., Rosi, N. et al. (2002). Science 295: 469–472. Brozek, C.K. and Dinc˘a, M. (2014). Chem. Soc. Rev. 43: 5456–5467. Deng, H., Doonan, C.J., Furukawa, H. et al. (2010). Science 327: 846–850. Cohen, S.M. (2012). Chem. Rev. 112: 970–1000. Hoffmann, H., Debowski, M., Müller, P. et al. (2012). Materials 5: 2537–2572. Kim, M., Cahill, J.F., Prather, K.A., and Cohen, S.M. (2011). Chem. Commun. 47: 7629–7631. Bordiga, S., Bonino, F., Lillerud, K.P., and Lamberti, C. (2010). Chem. Soc. Rev. 39: 4885–4927. Stock, N. and Biswas, S. (2012). Chem. Rev. 112: 933–969. Paz, F.A.A., Klinowski, J., Silva, P. et al. (2011). Dalton Trans. 40: 321–330. Khan, N.A. and Jhung, S.H. (2015). Coord. Chem. Rev. 285: 11–23. Al-Kutubi, H., Gascon, J., Sudhölter, E.J.R., and Rassaei, L. (2015). ChemElectroChem 2: 462–474. T. Frišˇci´c, Metal–Organic Frameworks: Mechanochemical Synthesis Strategies, 2014. Liu, J., Chen, L., Cui, H. et al. (2014). Chem. Soc. Rev. 43: 6011–6061. Groom, C.R., Bruno, I.J., Lightfoot, M.P., and Ward, S.C. (2016). Acta Cryst. B72: 171–179. https://doi.org/10.1107/S2052520616003954. https://www.ccdc .cam.ac.uk/structures/. Devic, T., Horcajada, P., Serre, C. et al. (2010). J. Am. Chem. Soc. 132: 1127–1136. Horcajada, P., Chalati, T., Serre, C. et al. (2009). Nat. Mater. 9: 172–178. Bhattacharjee, S., Choi, J.-S., Yang, S.-T. et al. (2010). J. Nanosci. Nanotechnol. 10: 135–141. Morel, F.L., Xu, X., Ranocchiari, M., and van Bokhoven, J.A. (2015). Functional Metal–Organic Frameworks: Synthesis and Reactivity. Hoboken, NJ: Wiley-Blackwell. Tanabe, K.K., Cohen, S.M., and Cohen, S.M. (2009). Angew. Chem. Int. Ed. 48: 7424–7427. Servalli, M., Ranocchiari, M., and van Bokhoven, J.A. (2012). Chem. Commun. 48: 1904–1906.
159
160
8 Design of Molecular Heterogeneous Catalysts with Metal–Organic Frameworks
33 Karagiaridi, O., Bury, W., Mondloch, J.E. et al. (2014). Angew. Chem. Int. Ed.
53: 4530–4540. 34 Kim, M., Cahill, J.F., Fei, H. et al. (2012). J. Am. Chem. Soc. 134: 18082–18088. 35 Szilágyi, P.Á., Serra-Crespo, P., Dugulan, I. et al. (2013). CrystEngComm 15:
10175–10178. 36 Corma, A., Garcia, H., and Llabrés i Xamena, F.X. (2010). Chem. Rev. 110:
4606–4655. 37 Ranocchiari, M. and Bokhoven, J.A.V. (2011). Phys. Chem. Chem. Phys. 13:
6388–6396. 38 Kapteijn, F., Xamena, F.X.L.I., Gascon, J. et al. (2014). ACS Catal. 4: 361–378. 39 Gascon, J., Aktay, U., Hernandez-Alonso, M.D. et al. (2009). J. Catal. 261:
75–87. 40 Akiyama, G., Matsuda, R., Sato, H. et al. (2011). Adv. Mater. 23: 3294–3297. 41 Cho, S.-H., Ma, B., Nguyen, S.T. et al. (2006). Chem. Commun. (24):
2563–2565. 42 Takaishi, S., DeMarco, E.J., Pellin, M.J. et al. (2013). Chem. Sci. 4: 1509. 43 Hwang, Y.K., Hong, D.-Y., Chang, J.-S. et al. (2008). Angew. Chem. Int. Ed. 47: 44 45 46 47 48 49 50 51 52 53 54 55
4144–4148. Brozek, C.K. and Dinca, M. (2013). J. Am. Chem. Soc. 135: 12886–12891. Tu, J., Zeng, X., Xu, F. et al. (2017). Chem. Commun. 53: 3361–3364. Yang, Q., Xu, Q., and Jiang, H.-L. (2017). Chem. Soc. Rev. 46: 4774–4808. Vilhanová, B., Ranocchiari, M., and van Bokhoven, J.A. (2015). ChemCatChem 8: 308–312. Eddaoudi, M., Alkordi, M.H., Eubank, J.F. et al. (2008). J. Am. Chem. Soc. 130: 12639–12641. Beloqui Redondo, A., Morel, F.L., Ranocchiari, M., and van Bokhoven, J.A. (2015). ACS Catal. 5: 7099–7103. Seo, J., Whang, D., Lee, H. et al. (2000). Nature 404: 982–986. Zheng, M., Liu, Y., Lin, W. et al. (2012). Chem. Sci. 3: 2623. Bauer, G., Ongari, D., Xu, X. et al. (2017). J. Am. Chem. Soc. 139: 18166–18169. Zhang, T., Lin, W., Manna, K. et al. (2016). J. Am. Chem. Soc. 138: 3241–3249. Bueken, B., Fang, Z., Fischer, R.A. et al. (2015). Angew. Chem. Int. Ed. 54: 7234–7254. Kapteijn, F., Santos, V.P., Wezendonk, T.A. et al. (2015). Nat. Commun. 6: 6451.
161
9 Hierarchical and Anisotropic Nanostructured Catalysts Hamidreza Arandiyan 1 , Yuan Wang 2 , Christopher M.A. Parlett 3,4 , and Adam F. Lee 5 1 The University of Sydney, School of Chemistry, Laboratory of Advanced Catalysis for Sustainability, Eastern Avenue, Sydney 2006, Australia 2 The University of New South Wales, Sydney, School of Chemistry, High Street, Sydney 2052, Australia 3 University of Manchester, School of Chemical Engineering and Analytical Science, Sackville Street, Manchester M1 3AL, UK 4 The University of Manchester at Harwell, Diamond Light Source, Harwell Campus, Didcot OX11 0DE, UK 5 CAMIC, RMIT University, School of Science, La Trobe St, Melbourne, VIC 3001, Australia
9.1 Introduction Nanoporous materials have attracted significant attention due to their unique textural properties and, based on their pore dimensions, are divided according to International Union of Pure and Applied Chemistry (IUPAC) definitions into three categories: microporous (50 nm) [1]. Binary and ternary combinations of different pore networks within a single framework create hierarchical porous materials whose physical and chemical properties are particularly attractive in energy and environmental applications. Micro- and/or mesoporosity affords increased surface area, while macroporosity improves heat and mass transfer and reduces pressure drops [2]. Multiple pore networks also unlock the possibility to incorporate different chemical functions within a single structure, as achieved by a dual-templating route to fabricate a hierarchical macroporous–mesoporous silica containing Pd nanoparticles (NPs) spatially confined within the larger pores and Pt NPs confined within the smaller pores to catalyze a two-step cascade selective oxidation [3]. Janus NPs have also garnered interest since their highlighting by Gennes in his 1991 Nobel Prize address [4] and are an extreme example of anisotropic NPs whose properties such as shape, composition, and surface chemistry differ along different directions and hence are well suited to multiple catalytic reactions on different particle domains. Anisotropic NPs exhibit unique and sometimes superior physicochemical properties to isotropic particles [5]. This chapter illustrates the breadth of hierarchically porous and anisotropic NPs and approaches to their synthesis and catalytic applications (Scheme 9.1), Heterogeneous Catalysts: Advanced Design, Characterization and Applications, First Edition. Edited by Wey Yang Teoh, Atsushi Urakawa, Yun Hau Ng, and Patrick Sit. © 2021 WILEY-VCH GmbH. Published 2021 by WILEY-VCH GmbH.
162
9 Hierarchical and Anisotropic Nanostructured Catalysts
focusing on the impact of NP morphology, porosity, and chemical functionalization on activity, selectivity, and stability. In all cases, the ability to tune and spatially image material properties at the nanoscale is critical to understanding and improving catalytic performance.
9.2 Top-Down vs. Bottom-Up Approaches Numerous chemical and physical routes to NPs exist, with fabrication procedures classified as either top down or bottom up (Scheme 9.2) [8]. Top-down approaches create NPs by the removal of extraneous material from a bulk sample using reactive ion or chemical etching to produce highly uniform NPs. Hsu et al. [9] employed reactive ion etching to produce Si nanopillar/nanocones across an entire 4-inch wafer. In the bottom-up approach, molecular components arrange into more complex assemblies atom by atom, molecule by molecule, or cluster by cluster (akin to crystal growth). Colloid NP dispersions formed from mononuclear precursors are a common example of bottom-up synthesis. Bottom-up approaches are considered more straightforward and precise for creating small NPs (100 nm.
9.3 Shape Anisotropy and Nanostructured Assemblies Anisotropic NPs are versatile building blocks for the construction of functional materials due to their intrinsic asymmetry and resulting space-dependent interactions with magnetic, optical, mechanical, and chemical environments [10]. Control over the spatial arrangement and chemical composition of self-assembled anisotropic NPs affords access to unique structures, which may span multiple length scales and exhibit complex connectivity through so-called “bottom-up” fabrication. Such approaches encompass the synthesis and manipulation of particle–particle or particle–scaffold interactions to induce the assembly of ordered nanostructures, ideally in a scalable and predictable manner. Assembly (and disassembly) strategies can exploit anisotropic NPs such as nanocubes, nanotubes, and nanoplates into complex architectures such as three-dimensional ordered porous structures. This section discusses the catalytic applications of morphologically and chemically anisotropic NPs and their assemblies. Disassembly of mesostructured frameworks, such as La0.6 Sr0.4 MnO3 (LSMO) perovskites, can also yield anisotropic NP catalysts [11]. In a process akin to the retrosynthesis of organic compounds, hexapod-shaped building blocks can be harvested by fragmenting three-dimensional ordered macroporous (3DOM) structures in a controlled ultrasonic treatment for subsequent reassembly into a mesostructured architecture (Figure 9.1a–g). Fragmentation promotes the construction of high area periodic nanoarrays and, in this LSMO example, the genesis of new surface properties [11]; high-angle annular dark-field
O c vo ta. id
AgNPs AgNPs 10 nm
A
es
B
Co nt r
Highly dispersed NPs
, ur ct de es vi rti ru st pro pe s ds pro d es ou ffol cal ce cl or ca ani han rti to l p s s ch en s nopa istry e ca rou me nd ur na m hi po n , a ct nus che rc lly ee es f ru ra ca tw rti st fJa s o hi be pe ie H rarc gies t pro a no ce o ype icle ie r r e na urfa ent t part H ne po ar s s r e sy ans ce nu ue iffe am tr rfa Ja uniq wo d he s su is s t n t Th llow r o a ccu o
t en er rts s of iff a p st ly lass ctly d nter ta c n u es rful ng ca re a disti r co cl e ri a s ula rti le ow ee tic Ps) ertie lec pa re p ngin o ar (N p mo a e n op les pro nd na ticles rials c an artic with lk a ar ate N pi u op als b ro anopfor m an ri er ot n s N te thi s a m m ni ropic lock A fro b ot is ing An uild b
5 nm
nt me gn ali
ind siz ow e s
(110)
g lin ol
W
0.277 nm
Controlling layers
N wa eck ll
P siz ore e
C Catalyst
D
Anisotropic growth 500 nm
100 nm
Scheme 9.1 Overview of hierarchical and anisotropic nanostructured catalysts. (A) 3.63 wt% Ag/3DOM La0.6 Sr0.4 MnO3 . Source: Reproduced with permission from Arandiyan et al. [6]. Copyright 2014, American Chemical Society. (B) UCNP@SiO2 @mSiO2 (UCNP = upconversion nanoparticle). Source: Reproduced with permission from Li et al. [5]. Copyright 2014, American Chemical Society. (C) 3DOM-m La0.9 Ce0.1 CoO3 . Source: Reproduced with permission from Arandiyan et al. [7]. Copyright 2016, American Chemical Society. (D) Janus mesoporous silica nanocomposites of UCNP@SiO2 @mSiO2 &PMO (PMO = periodic mesoporous organosilica). Source: Reproduced with permission from Li et al. [5]. Copyright 2014, American Chemical Society.
Top-down approach
Bottom-up approach
Precursor containing molecules Bulk
Transformation by energy
Formation of ions, radicals and electrons
Fullerenes Transformation to smaller fragments
Exfoliation of multilayers to monloyers
Diamonds
Ionization by energy
Graphene nanosheets Condensation and formation of clusters
Powder
Clusters
Nanoparticles synthesized
Scheme 9.2 Bottom-up and top-down approaches to synthesizing carbon-based nanomaterials. Source: Ahmed and Ali [8]. http://www.onecentralpress .com/manufacturing-nanostructures/. Licensed under CCBY 4.0 International license.
9.4 Janus Nanostructures
(HAADF)-scanning transmission electron microscopy (STEM) evidenced that hexapod fragments exposed (001) facets rather than (110) facets present in the parent 3DOM LSMO material, which X-ray photoelectron spectroscopy (XPS) (Figure 9.1h) and density functional theory (DFT) calculations identified as offering a higher concentration of reactive oxygen adatoms and lower energy barrier for hydrogen abstraction from methane (CH4 * → CH3 * + H*), respectively, thereby promoting low-temperature methane combustion (Figure 9.1i). Catalytically active NPs can be fabricated in a rapidly expanding range of shapes and size [10], exhibiting a bewildering array of physicochemical properties. Pan and coworkers [12] investigated the influence of shape anisotropy of hexagonal CdS nanorods on their catalytic efficiency for H2 evolution and photoelectrochemical water splitting. CdS nanorods yielded 27 times more H2 and 19 times greater photocurrent density than CdS nanospheres. The superior catalytic performance of the nanorods was attributed to effective separation and transport of photogenerated electron–hole pairs. NP shape anisotropy can also tune selectivity: well-defined Pd nanocubes and nanopolyhedra embedded in mesoporous silica spheres exhibit excellent activity toward CO2 hydrogenation at 400–500 ∘ C, with polyhedra (presented the highest density of (111) facets) favoring CO formation over CH4 due to weak adsorption of the former (Figure 9.2) [13]. Encapsulation of anisotropic NPs within silica improved their stability to sintering with respect to palladium randomly impregnated over an amorphous silica support. Shape anisotropy may also arise from extreme gradients in surface chemistry and composition. Xia and coworkers synthesized Rh cubic nanoframes [14] and Pt cubic/octahedral [15] nanocages through selective chemical etching of palladium from Pd@Rh cubic and Pd@Pt cubic/octahedral NP precursors. The resulting Pt nanocages exhibited excellent electrocatalytic activity for oxygen reduction, with the (111) terminated nanooctahedra offering a high specific activity of 1.98 mA/cm2 at 0.9 V (versus reversible hydrogen electrode), twice that of (100) terminated nanocubes and eight times greater than that of a commercial Pt/C catalyst. Superior performance of the nanocages was attributed to enhanced OH desorption arising from contracted Pt–Pt interatomic distances.
9.4 Janus Nanostructures Janus nanomaterials (named after the Roman god depicted as having two faces, looking to the future and past simultaneously) exhibit spatially separated domains that differ in chemical composition and/or physical properties [16], e.g. bimetallic NPs in which each element is localized in separate hemispheres [17] or opposing ends of rods [18]. Catalytic Janus nanostructures include bimetallics [19], metals and oxides [20], and inorganic carbons and oxides [21] and find application in oxidation, reduction [22], photocatalysis [23], and electrocatalysis [24], and exemplars are outlined below. Wang et al. demonstrated the synthesis of AuNi Janus spindle nanostructures, comprising conjoined Au and Ni NPs of 10 and 15 nm, respectively [25]. Their formation mechanism was postulated to reflect the greater reducibility of Au3+
165
(a) PMMA template
(b) PMMA template
(c) 3DOM LSMO
(h)
O 1s
529.2 ads
210 nm
Intensity (a.u.)
1.2 μm
1.2 μm
0.2
0.2 0.4
0.4 0.6
0.6 0.8
0.8 1.0
1.0
latt
531.8 530.7
533.1 3D-hm LSMO
3DOM LSMO
55.5 nm 1.2 μm 1.2 μm –5.7 nm
nm
Macropore
Amorphous mesopores
leg
40 nm
80
534
530
532
528
526
Binding energy (eV)
(i)
–1.0
1DDN LSMO 3DOM LSMO 3D-hm LSMO
–1.5
nm
–2.0
dy
(g)
536
18 nm
Window
Bo
Hexapod building block
1DDN LSMO
(f) 3D-hm LSMO
20 nm
0 14
(e) 3D-hm LSMO
–2.5
Hexapod
In T
(d) 3DOM LSMO
–3.0 –3.5 –4.0 –4.5 –5.0 –5.5 1.4
1.5
1.6
1.7
1.8
1.9
2.0
2.1
1000/T (K–1)
Figure 9.1 Field-emission high-resolution scanning electron microscopy (FE-HRSEM) and 3D-environmental atomic force microscopy (3D-eAFM) analysis of LSMO catalysts: (a, b) PMMA template, (c, d) 3DOM LSMO, (e, f ) 3D-hm LSMO, (g) schematic illustration depicting synthesis of the 3D-hm LSMO catalysts, (h) O 1s XP spectra, and (i) Arrhenius plots of methane conversion by the as-prepared catalysts. Source: Wang et al. 2017 [11]. https://www.nature.com/articles/ ncomms15553#rightslink. http://creativecommons.org/licenses/by/4.0/. Licensed under CCBY 4.0.
9.4 Janus Nanostructures
(a)
(b) Impregnated Cubes Polyhedra
75 50 25 0 0
(d)
(c) 100 CO selectivity (%)
CH4 selectivity (%)
100
10 20 CO2 conversion (%)
30
75 50
Polyhedra Cubes Impregnated
25 0 0
(e)
10
20
30
CO2 conversion (%)
Figure 9.2 (a–c) High-resolution transmission electron microscopy (HR-TEM) images and (d, e) corresponding selectivity for CO2 hydrogenation over Pd/silica catalysts. Source: Martins et al. 2015 [13]. Adapted with permission of Elsevier.
ions by octadecylamine (to yield colloidal Au NPs) relative to Ni2+ ions whose reduction is promoted through adsorption over the in situ generated Au NP (termed noble-metal-induced reduction). Similar spindle AuNi Janus NPs were dispersed over magnetic halloysite nanocomposites (inorganic aluminosilicate clay) [26] and exhibited superior activity for nitrophenol and Congo red reduction by NaBH4 , relative to supported isotropic Au NPs. The enhanced activity of the Janus system was proposed to arise from an electronic junction effect at the bimetal interface, although no comparisons were made against monometallic Ni catalysts or AuNi alloys. Varying the size of both NPs to optimize the interfacial contact could provide further mechanistic insight and optimize reactivity of these magnetically separable Janus catalysts. Au–Ni Janus nanostructures prepared over catechol-modified silica spheres (the latter synthesized by the Stöber process and subsequent organofunctionalization of surface hydroxyls) [27] also exhibited good activity for nitrophenol reduction by NaBH4 , although no catalytic benchmarking was performed. Electrocatalysts for fuel cells typically employ high precious metal loadings (of notably Pt), and hence there is a significant driver to prepare lower-cost alternatives. Chen and coworkers [24] reported the Langmuir–Blodgett assembly of 1-hexanethiol capped Ag NP monolayers on a glass slide (Figure 9.3) whose surfaces were then subject to galvanic replacement by Au. The resulting Janus architecture exhibited a sixfold and threefold enhancement in electrocatalytic oxygen reduction compared with unmodified Ag NPs or AgAu bimetallic alloy NPs, respectively. This was attributed to the spatial partitioning of polar (mercaptopropanediol) and apolar (hexanethiolates) ligand stabilizers and the
167
168
9 Hierarchical and Anisotropic Nanostructured Catalysts
Air
Air
H2O
H2O
Aul–MPD
Ag Au
Figure 9.3 Schematic of bimetallic Janus NP synthesis employing interfacial galvanic exchange of hexanethiol-functionalized Ag NPs with an AuI –mercaptopropanediol complex. Source: Reproduced with permission from Song et al. [24]. Copyright 2012, American Chemical Society.
associated charge transfer from Ag to Au, promoting O2 dissociative adsorption over subnanometer Au regions. The weaker promotion observed over the alloy vs. Janus NPs, which possessed comparable Au loadings, may reflect differences in electronic transfer and/or interfacial spillover. Janus AuPt systems have also been developed for the electrocatalytic oxidation of Ru(bpy)3 2+ to Ru(bpy)3 3+ , a chemiluminescent transformation enabling electrogenerated chemiluminescence microscopy [28]. The Janus arrangement exhibited superior activity and resistance to passivation than monometallic systems; the differing reactivity of Au and Pt faces resulted in a concentration gradient and fluid slip around particles, suppressing undesired oxidation of the metal surfaces. Such examples of bimetallic Janus NPs as heterogeneous catalyst remain relatively rare, partly due to synthetic challenges and their unpredictable performance. Catalytic Janus NPs comprising a metal and an oxide, or two oxides, are more common. A general solvothermal route to the former yields Janus dumbbell-like NPs of MFe3 O4 (M = Ag, Au, Pd, Pt, or AgAu) with control over the dimensions of both components possible [29]. AuFe3 O4 and PtFe3 O4 Janus dumbbell NPs were evaluated for CO oxidation, with both exhibiting a significant decrease in light-off temperatures relative to conventional iron oxide-supported analogues. This enhancement was attributed to a stronger electronic interaction across the larger heterojunction interface in the nanodumbbells and concomitant superior thermal stability. Numerous noble metal-doped titania have been exploited as photocatalysts [30], including Au/TiO2 Janus NPs for the selective oxidation of methanol and aqueous-phase hydrogen production from isopropyl alcohol (Figure 9.4) [31]. Composites of 50–70 nm Au NPs in contact with 50–100 nm amorphous TiO2 were produced by slow hydrolysis of the titanium diisopropoxide bis(acetylacetonate) precursor in the presence of preformed colloidal Au NPs (stepwise rapid addition of the precursor resulted in an Au@TiO2 core–shell structure). The local surface plasmon resonance generated by large Au NPs in these Janus nanomaterials promoted efficient visible-light hydrogen production compared with either symmetric core–shell counterparts or 5 nm Au NP on a conventional P25 titania support. Strong coupling between the localized plasmon resonance at the Au/TiO2 interface and optical transitions involving localized electronic states in the TiO2 enhanced photogenerated electron–hole charge generation. A facile route to AuSiO2 Janus NPs was recently demonstrated, in
9.5 Hierarchical Porous Catalysts Janus
150
50 y (nm)
120
100 TiO2
–50
λ = 569 nm
(b)
0
Core–shell 50
y (nm)
0
60
Bare–gold
0 10 20 30 40 50 Time (min)
60
90
120
150
180
Time (min)
100
0
–50 30
0
50
0 0
λ = 595 nm
(c) y (nm)
2
30
50
Au
λ = 527 nm
–50
(d)
50
Au
0 x (nm)
|E(r)/Eo/2
VH (mL)
–50 1.5 1.0 0.5 0.0
100 TiO2
|E(r)/Eo/2
VH2 (ml)
90
(a)
50
Au
0
|E(r)/Eo/2
Janus Au–TiO2 Core–shell Au–TiO2 Amorphosus TiO2 Bare gold
0 50
Figure 9.4 (a) Hydrogen production over Janus and core–shell Au/TiO2 NPs, amorphous TiO2 , and colloidal Au, under visible-light irradiation, and (b–d) simulated plasmonic near-field maps showing intensity enhancement at their localized surface plasmon resonance. Source: Seh et al. 2012 [31]. Reproduced with permission of Wiley-VCH.
which partial encapsulation of 15 nm (citrate stabilized) gold colloids by low concentrations of poly(vinylpyrrolidone) promoted the subsequent formation of eccentric vs. concentric Au@SiO2 core–shell nanostructures [32]. The resulting AuSiO2 Janus NPs (in which Au NPs were only partially embedded in silica) exhibited high activity for nitrophenol reduction by NaBH4 relative to Au@SiO2 core–shell NPs. Nanomotors represent a niche catalytic application of Janus NPs. Ma and Sanchez reported selective enzyme (catalase) tethering, via a carboxylic acid function, to one hemisphere of silica nanospheres [33]. The spatial localization of catalase enzyme enabled subsequent NP propulsion through the biocatalytic decomposition of H2 O2 , with a 150% increase in the apparent diffusion coefficient relative to intrinsic Brownian (random) motion.
9.5 Hierarchical Porous Catalysts The coupling of two or more distinct pore networks, typically micro- and mesopores or meso- and macropores, yields material possessing hierarchical porosity. Such architectures have shown benefits in a variety of catalytic transformations, often attributed to enhanced internal mass transport and/or superior dispersion of active sites [34]. Bottom-up synthetic strategies to hierarchical porous architectures adopt templating strategies (using soft and/or
169
170
9 Hierarchical and Anisotropic Nanostructured Catalysts
hard organic or inorganic templates) or crystalline self-assemblies to generate intricate three-dimensional frameworks with morphologies that may mirror those observed in nature. Discrete nanoflowers of bismuth (Figure 9.5a) [35, 38, 39], titania [40], and mixed zinc oxide-ceria [41] arise from the self-assembly of nanoplatelets around a central cavity, thereby creating a hierarchical mesoporous–macroporous pore structure with 800 ∘ C) at which vaporization of the precursor or the initially formed nuclei occurs.
10.2 Flame Aerosol Synthesis and Reactors
10.2 Flame Aerosol Synthesis and Reactors The flame is a natural geometrically compact reactor that overcomes the temperature limitation for precursors vaporization while also providing the necessary steep temperature gradient for subsequent aerosol particles growth. A simple candle flame is a reactor for soot particles production, with the melted wax being the precursor that is fed to the flame. Within a few millimeters from the wax, at the brightest (and also the hottest) zone of the flame, the vaporized wax is fully converted to gaseous organic radicals that quickly form carbon soot nuclei that grow by coalescence, aggregation, and agglomeration as they move up the flame height, much like a plug flow reactor. The formed carbon soot aggregates leave the tip of the flame in the form of black smoke. It is possible to collect the soot particles by thermophoretic deposition simply by placing a cool water-filled metal spoon slightly above the flame tip. In fact, the modern synthesis of carbon blacks is produced at the global production rate of tons per year using a similar process by feeding natural gas as both the precursor feed and fuel and quenching with water sprays at some point along the flame when the carbon black particles of desired size have been formed [6, 7]. For the synthesis of non-carbonaceous or more specifically metal oxide particles, the respective gaseous metal precursors belonging to that of metal halides, carbonyls, and acetylacetonates can be fed into the flame. Where both the fed precursor and fuel are in gaseous phase, this process is specifically termed vapor-fed flame synthesis (VFS) (Figure 10.2). Combustion of metal precursor leads to the formation of metal oxide nuclei that further coalesces, aggregates, and agglomerates, in the same manner as explained above for carbon soot particles. Excess oxygen is provided to the flame to ensure complete combustion of carbon sources and the formation of pristine metal oxide particles. When inlet oxygen is below that of combustion stoichiometry, it gives rise to carbon-coated particles or in some isolated cases, carbides. When metal halide precursors such as TiCl4 or SiCl4 are used, the water vapor generated from the combustion of fuel source (e.g., natural gas, methane, hydrogen) hydrolyzes the metal precursors to form metal oxide and acid halide, e.g., SiCl4 + 2H2 O → SiO2 + 4HCl. The patent on industrial flame hydrolysis or Aerosil process was awarded to Degussa (now Evonik) in 1942 [8]. Other particles manufactured using the same process include Al2 O3 , TiO2 , ZrO2 , Fe2 O3 , NiO, WO3 , and AlBO3 . They are normally referred to as fumed particles. Incidentally, the fumed TiO2 , known by its commercial name of P25, has been the gold standard photocatalyst since the 1980s. An inherent limitation of VFS lies in the availability of volatile metal precursors, which may be too expensive, insufficiently volatile, or unavailable for many elements. Replacements with metal precursors that exist either as liquid in their pristine forms or as dissolvable salt or crystals can effectively overcome these limitations. In such case, retrofitting an atomizer is essential to disperse the precursors as spray droplets in the flame (Figure 10.2). If the liquid metal precursors are of low calorific values, e.g., dissolved metal nitrates in water or ethanol, where the contribution of enthalpy from the liquid precursor is less than that contributed by the oxy-fuel combustion, this is known as the flame-assisted spray pyrolysis (FASP). Because of the low calorific value of the precursor, it could be
185
186
10 Flame Synthesis of Simple and Multielemental Oxide Catalysts
Liquid-fed flame synthesis (FSP and FASP)
Vapor-fed flame synthesis (VFS)
Vacuum pump Hood
Hood Pump Filter assembly
Burner chamber with filter
NaOH solution
Spray flame Flamelets Gas exit Gap
Flame reactor
O2 CH4
Methane Air
O2 or air
Argon
HMDSO/fuel
(a)
Liquid precursor injection
M Motor
Metal precursor vaporization
(b)
Figure 10.2 Schematic of the configuration of flame aerosol reactors, including the particles collection, for the handling of (a) liquid- and (b) vapor-fed system. The former consists of an atomizer to pneumatically or ultrasonically disperse the liquid feed into the fine microdroplets for homogeneous combustion. HMDSO, hexamethyldisiloxane; PI: pressure indicator; FIC: flow indicator/control. Source: Adapted from Mädler et al. 2002 [4] and Pratsinis et al. 1996 [5].
possible that the hottest zone in the flame is insufficient to induce full vaporization/sublimation of the droplet or particle. If so, this leads to relatively large and inhomogeneous hollow and/or fragmented particles (Figure 10.1). If more than 50% the total enthalpy of combustion of the aerosol flame is contributed by the liquid precursor, the process can be categorized as flame spray pyrolysis (FSP) (Figure 10.2). The implication is that direct variation of the precursor feed flow rate in FSP directly influences the product particle size as well as other physical properties. The combination of exothermic metal precursors (e.g., metal alkoxides, naphthenates, 2-ethylhexanoates) and that of high calorific organic solvents (e.g., xylene, 2-ethylhexanoic acid, toluene) ensures complete vaporization/sublimation of the metal precursors and for the gas-to-particle route to take place (Figure 10.1) (see also Table 10.1 for different precursor formulations). It was revealed only in the recent years via single droplet studies that the transformation to vapor phase may also proceed through the microexplosion of the surface precipitated droplet (Figure 10.1, broken arrow) [34–36]. This is followed by the formation and growth of aerosol nanoparticles through the sequential processes of nucleation, coalescence, aggregation, and agglomeration. The combination of high-temperature flame and short residence time (on the order of milliseconds from droplet dispersion to agglomerated particles leaving flame) in FSP is the key to making ultrafine, highly crystalline nanoparticles, although, in some cases as will be shown below, amorphous nanoparticles can also be selectively obtained. Despite the complex fluid dynamics and the accompanying evolution of particle size growth, the theory on the post-nucleation stage within an FSP flame has been well understood and verified experimentally [37].
10.2 Flame Aerosol Synthesis and Reactors
Table 10.1 Liquid precursor formulations for the different FSP-derived metal oxide catalysts. Catalyst
Metal precursor
Solvent
References
Simple metal oxide-based catalyst TiO2
Titanium tetraisopropoxide
Xylene, acetonitrile
[9]
Fe-doped TiO2
Titanium tetraisopropoxide, iron naphthenate
Xylene, acetonitrile
[10]
Pt/TiO2
Platinum acetylacetonite, titanium tetraisopropoxide
Xylene, acetonitrile
[9]
Ag/TiO2
Silver benzoate, titanium tetraisopropoxide
Xylene
[11]
Pd/TiO2
Palladium acetylacetonate, titanium tetraisopropoxide
2-Ethylhexanoate, acetonitrile
[12]
ZnO
Zinc naphthenate
Toluene
[13]
Ag/ZnO
Silver nitrate, zinc naphthenate
Ethanol
[13]
Ru/ZnO
Ruthenium acetylacetonate, zinc 2-ethylhexanoate
Xylene, acetonitrile
[14]
CeO2
Cerium 2-ethylhexanoate
Xylene
[15]
Pt/CeO2
Platinum acetylacetonate, cerium 2-ethylhexanoate
Xylene
[16]
Ru/CeO2
Ruthenium acetylacetonate, cerium 2-ethylhexanoate
Xylene, acetonitrile
[14]
Al2 O3
Aluminum sec-butoxide
Xylene
[14]
Pt/Al2 O3
Platinum acetylacetonite, aluminum isopropoxide
Xylene, ethyl acetate
[17]
Pd/Al2 O3
Palladium acetylacetonate, aluminum sec-butoxide
Xylene, acetonitrile
[18]
Ru/Al2 O3
Ruthenium acetylacetonate, aluminum sec-butoxide
Xylene, acetonitrile
[14]
Multielemental oxide-based catalyst Cex Zr1−x O2
Cerium acetate hydrate, Lauric acid, acetic acid zirconium tetracetylacetonate
[19]
Cerium 2-ethylhexanoate, zirconium 2-ethylhexanoate
2-Ethylhexanoic acid, xylene
[20]
SiO2 -doped Ce0.5 Zr0.5 O2
Tetraethoxysilane, cerium acetate hydrate, zirconium 2-ethylhexanoate
2-Ethylhexanoic acid, toluene
[21]
Al2 O3 -doped Ce0.5 Zr0.5 O2
Aluminum 2-ethylhexanoate, cerium acetate hydrate, zirconium 2-ethylhexanoate
2-Ethylhexanoic acid, toluene
[21]
Pt/Ce0.5 Zr0.5 O2
Platinum acetylacetonate, cerium 2-ethylhexanoate, zirconium 2-ethylhexanoate
2-Ethylhexanoic acid, xylene
[20]
CuO/CeO2
Copper 2-ethylhexanoate, cerium 2-ethylhexanoate
Xylene
[15, 22] (continued)
187
188
10 Flame Synthesis of Simple and Multielemental Oxide Catalysts
Table 10.1 (Continued) Catalyst
Metal precursor
Solvent
References
CuO/Cex Zr1−x O2
Copper 2-ethylhexanoate, cerium 2-ethylhexanoate, zirconium propoxide
Xylene
[23]
Pt/FeOx /CeO2 (two flame synthesis)
Platinum acetylacetonate, iron Xylene 2-ethylhexanoate, cerium 2-ethylhexanoate
[16]
V2 O5 /TiO2
Vanadium oxo-triisopropoxide, titanium tetraisopropoxide
Xylene, acetonitrile
[24]
V2 O5 /SiO2
Ammonium metavanadate, hexamethyldisiloxane
Xylene, 2-ethylhexanoate, acetic anhydride
[25]
V2 O5 /Al2 O3
Vanadyl acetylacetonate, aluminum propoxide
C1–C8 linear alcohol, C1–C3 carboxylic acid
[26]
Co3 O4 /Al2 O3 (two-flame synthesis)
Cobalt 2-ethylhexanoate, aluminum sec-butoxide
Xylene
[27]
Co3 O4 /ZrO2
Cobalt 2-ethylhexanoate, zirconium propoxide
Xylene
[28, 29]
Ag/Co3 O4 /ZrO2
Silver benzoate, cobalt 2-ethylhexanoate, zirconium propoxide
Xylene
[29]
Ru/Co3 O4 /ZrO2
Ruthenium acetylacetonate, cobalt 2-ethylhexanoate, zirconium propoxide
Xylene
[29]
Pt/Co3 O4 /ZrO2
Platinum acetylacetonate, cobalt 2-ethylhexanoate, zirconium propoxide
Xylene
[29]
Rh/Co3 O4 /ZrO2
Rhodium acetylacetonate, cobalt 2-ethylhexanoate, zirconium propoxide
Xylene
[29]
Pd/Co3 O4 /ZrO2
Palladium acetylacetonate, cobalt 2-ethylhexanoate, zirconium propoxide
Xylene
[29]
SiO2 –Al2 O3
Tetraethyl orthosilicate, aluminum sec-butoxide
Xylene
[30]
SiO2 –AlPO4
Tetraethyl orthosilicate, aluminum sec-butoxide, tributyl phosphate
Xylene
[30]
BiVO4
Bismuth 2-ethylhexanoate, Xylene, 2-ethylhexanoate vanadium oxo-triisopropoxide
[31]
LaFeO3
Lanthanum acetate dihydrate, Propionic acid cobalt acetate hydrate
[32]
Pd/LaFeO3
Palladium acetate, lanthanum Propionic acid, acetate hydrate, cobalt acetate n-propanol, water hydrate
[33]
10.3 Simple Metal Oxide-Based Catalysts
Compared with VFS and FASP, the scientific and engineering development of FSP has been the most intense in the last 15 years that it represents the stateof-the-art advancement in the field. In fact, the FSP has emerged as a mainstream technique for the synthesis of heterogeneous catalysts that include those used for automobile exhaust treatment, fine chemicals synthesis, environmental abatement, and CO2 reduction [38]. As such, the subsequent subsections of the chapter will focus predominantly on FSP, although the same principles of catalyst design are applicable to the VFS and, to some extent, the FASP.
10.3 Simple Metal Oxide-Based Catalysts The early development of FSP was focused on the rapid and high-throughput synthesis of simple metal oxides, for example, SiO2 , 𝛾-Al2 O3 , SnO2 , TiO2 , ZnO, and CeO2 . These are the Generation I commodity flame-derived particles. Because the particles are produced via the gas-to-particle route, they are generally defined by their high surface area (typically 50–300 m2 /g), high purity, nonporosity, and relatively homogeneous particle size distribution. Metal oxides are most easily formed in FSP due to the oxygen-rich and oxidic combustion flame. There are only some isolated cases where non-oxide particles were made by FSP [7]. In fact, by restricting the oxygen content fed into the flame below stoichiometry, it would result in the formation of carbonaceous coatings around the particles [39]. Of the Generation I flame-derived particles, only a small handful of the simple metal oxides have readily usable catalytic functions, for example, TiO2 , ZnO, and WO3 as photocatalysts, 𝛾-Al2 O3 as an alcohol dehydration catalyst, and CeO2 as a three-way catalyst. The ability to directly tune the product particle size by controlling the rate of combustion or concentration of metal in the liquid precursor is a primary feature of the FSP. This allows, for example, the synthesis of very fine TiO2 nanoparticles (∼16 nm, 89 m2 /g) with significantly larger specific surface area (SSA) than the commercial VFS-made Evonik P25 TiO2 (48 m2 /g) [9]. The large SSA is accompanied by an inherently larger amount of surface charge-trapping defects that enhanced the photocatalytic degradation of organic compounds by direct charge transfer, although too much defects could have a detrimental effect of severe charge recombination [40, 41]. Defect sites are inevitable in many cases in FSP due to the rapid quenching of the flame that gives steep temperature gradient and short residence time. In a deliberate attempt to increase the defects, the Fe3+ substitutional doping can be carried out by mixing both Fe and Ti precursors during the flame synthesis. The approach introduces sub-bandgap states through the formation of oxygen vacancies and lattice defects, and as such allows sub-bandgap excitation of the photocatalyst under visible light [10]. Such approach requires a delicate balance between extending the visible light response by increasing the amount of dopant, but not excessively or the resultant high density defects would become charge recombination centers. One way to reduce the amount defects is to place an open-ended quartz tube around the FSP flame. The tube prevents the free entrainment of air into the flame (that gives rise to the steep temperature gradient), thereby prolonging the
189
10 Flame Synthesis of Simple and Multielemental Oxide Catalysts Quartz tube
(a)
Ambient air
O2/N2
(b)
O2/N2
190
(d)
50 nm
(c)
50 nm
(e) 3.25 Å (110)
3.52 Å (101)
2.49 Å (101)
Rutile Sheath gas (O2/N2)
Dispersant gas (O2)
Supporting flames (CH4/O2)
Liquid precursor
5 nm
Anatase
5 nm
Figure 10.3 (a) Schematic of the tube-enclosed FSP configuration with controllable O2 partial pressure surrounding the flame. TiO2 particles produced under (b, c) low O2 partial pressure led to the formation of rutile TiO2 single-crystal nanoparticles, while those prepared under (d, e) high O2 partial pressure yielded anatase TiO2 single-crystal nanoparticles. Source: Adapted from Kho et al. 2010 [42] and Kho et al. 2011 [43].
flame residence time (Figure 10.3). The now extended high-temperature zone in the tube-enclosed flame allows the healing of particle defects, but at the expense of more extensive coalescence [44]. By further reducing the amount of oxygen below stoichiometry into the quartz tube, it is possible to tune the amount of the two most common polymorphic phases of TiO2 , namely, anatase and rutile [43]. When synthesized under oxygen-rich flames (including in non-enclosed flame), the anatase phase is dominant, but with decreasing oxygen content below stoichiometry, the rutile phase increases. Carbon coatings are inevitable in the latter, but they can be easily removed by simple calcination. Because of the difference in conduction band edges of the two polymorphs, precise control over the anatase–rutile content is beneficial for the synergetic interfacial charge separation when used as a photocatalyst. This, coupled with the reduced charge-trapping defects from the tube-enclosed FSP, gives excellent performance in photocatalytic hydrogen evolution (the reductive half of the water splitting reaction) in the presence of methanol hole scavenger [42]. CeO2 is an essential component of the automobile three-way catalyst that effectively releases O2 (2Ce(IV) O2 ↔ Ce(III) 2 O3 + 1/2O2 ) during the oxygen-lean cycle and storing the oxygen during oxygen-rich cycle. Being synthesized at high temperature, the FSP-derived CeO2 , with its signature rhombohedral-shaped single nanocrystal (Figure 10.4), displays exceptional high thermal stability compared with other commercial CeO2 [45]. As such, it is highly suited to the application as three-way catalysts where typical operation temperatures range between 800 and 1200 ∘ C. More interesting design of CeO2 -based catalysts using the FSP will be discussed in the following section. Besides the specific catalytic applications as mentioned above, most of other Generation I metal oxides are used as catalyst supports as motivated by their
10.3 Simple Metal Oxide-Based Catalysts
(a)
(b)
(c)
20 nm
50 nm
20 nm (d)
(e)
5 nm
(f)
5 nm
20 nm
Figure 10.4 Transmission electron microscopy (TEM) images of (a) rhombohedral CeO2 , (b) ZnO nanorods, and (c) spherical TiO2 nanoparticles as-prepared using the FSP. Also shown are the analogous particles codeposited with noble metals (shown in red arrows) also prepared using the FSP, (d) Au/CeO2 , (e) Pt/ZnO, and (f ) Ag/TiO2 . Source: (a) and (f ) were adapted from Mädler et al. 2002 [45] and Teoh et al. 2010 [7], respectively. (See online version for color figure).
large SSA. Conventional preparation of supported catalysts such as wet impregnation and CVD require a two-step synthesis, first the preparation of the oxide support and then the deposition of the active metals [3]. On the contrary, the FSP allows an elegant one-step synthesis of supported metals on metal oxide. This has been made possible by the large number of catalytically active noble metals, e.g., Pt, Pd, Au, Ag, and alloys, that have much lower boiling points than the oxide supports. During the flame synthesis, the precursor travels from high (just after ignition point) to low (at the tip of the flame) temperature zone in the flame. Under such condition, the metal oxide would precipitate first as soon as the flame temperature zone falls below its boiling/sublimation point and grow, before the precipitation of the noble metal on the preformed metal oxide sets in at even lower temperature zone further downstream. Should the noble metal precipitates too early while the metal oxide is still in the liquid form, there would be a tendency to form miscible solid phase in the final product particles. In most cases, the limited solid–solid and solid–liquid miscibilities ensure that most of the noble metals are deposited on the surface, instead of being dispersed in the metal oxide or encapsulated. Successful examples of noble metal/metal oxide catalysts synthesized in the FSP include Pt/Al2 O3 and Pd/Al2 O3 for enantioselective hydrogenation [17, 18] and catalytic combustion; Pt/TiO2 , Ag/TiO2 , and Ag/ZnO for the photocatalytic degradation of organic pollutants [9, 11, 13]; Pd/TiO2 for photocatalytic deNOx [12]; Ag/CeO2 for selective oxidation of alcohol [46]; and Ru/MOx (MOx = Al2 O3 , ZnO, TiO2 , CeO2 ) for CO2 methanation [14]. Because of the oxidic nature of the FSP, the noble metal deposits are predominantly in their oxide forms. As such, a prereduction step can improve the catalytic activities of these catalysts, especially
191
192
10 Flame Synthesis of Simple and Multielemental Oxide Catalysts
if the catalysts are not utilized under highly reducing reactions [17]. Many of these catalysts fall under the category of Generation II flame particles that reproduce the standard compositions of what is already known as the best in each of the target applications; but improvising their physical properties to accentuate their catalytic performance.
10.4 Multielemental Oxide-Based Catalysts Multielemental oxide catalysts generally exist in the forms of solid solutions, composite metal oxides, or complex metal oxides. The FSP preparation of such catalysts has been extensively investigated that to date include Cex Zr1−x O2 solid solution as three-way catalysts or CuO/CeO2 ; FeOx /CeO2 composite metal oxides for the preferential oxidation of carbon monoxide (CO-PrOx); Co3 O4 /ZrO2 as CO hydrogenation catalysts; and the BiVO4 complex metal oxide as an oxygen evolution photocatalyst, the oxidative half of the water splitting reaction. 10.4.1
Solid Solution Metal Oxide Catalysts
An inherent advantage of the flame synthesis is the preserved stoichiometry between the metal elements (as precursor) fed to the flame and that in the product particles. This allows precise stoichiometric control in the overall composition of the FSP-derived Cex Zr1−x O2 (x = 0.1–0.9) solid solution. The precise control of the Ce and Zr content is imperative to improving the thermal stability and oxygen exchange capacity of the three-way catalyst [19]. Because of its solid solution nature, the FSP-derived Cex Zr1−x O2 single crystals replicate the rhombohedral shape of the CeO2 . Except for the linear shrinkage in lattice parameters (in accordance with the Vegard rule) [47], there exists minimal structural and lattice distortion of the original CeO2 structure when Ce4+ is gradually replaced by Zr4+ . In fact, the highly crystalline nature of the single crystals with little defects limited their oxygen exchange capacity. As such, the introduction of small amount of dopant such as SiO2 (3%) is necessary to distort the physical and crystal structure of the Cex Zr1−x O2 (accompanied by the transformation from perfect rhombohedral- to spherical-shaped particles), thereby improving the overall oxygen exchange capacity [21]. Further in situ flame deposition of noble metals, e.g., Pt and Rh, can be easily carried out in the same manner as that described above for simple metal oxides [20, 48]. 10.4.2
Composite Metal Oxide Catalysts
The CuO/CeO2 composite oxide catalyst is a promising candidate for the application in CO-PrOx to selectively remove the trace CO present in H2 -rich stream for the use in low-temperature polymer electrolyte membrane (PEM) hydrogen fuel cells. Depending on the loading of Cu, the Cu-enriched surface of the FSP-derived catalyst composes of monomer CuO (at up to 1 wt% Cu), dimer CuO (from 2 wt% Cu), or CuO clusters (from 5 wt% Cu) [15]. During the catalytic CO-PrOx reaction, the CeO2 through its switchable redox (Ce(IV) ↔ Ce(III) ) is capable of transporting lattice oxygen to the adsorbed CO on the CuOx site for the oxidation reaction. Although most of the Cu contents are
10.4 Multielemental Oxide-Based Catalysts
0.312 mm
(b)
(a)
(d)
(c)
Pristine CeO2
20 nm
20 nm
1 wt%
(e)
4 wt%
20 nm
12 wt%
(f) K1
100% K1
80% K2
D
CO conversion
EPR intensity (a.u.)
12 wt% 8 wt% 4 wt% 2 wt% 1 wt%
40% FSP Wet
20%
0.5 wt% C1
C2
C3
B
0%
DP 4 wt%
50 A
2.7
60%
2.5
2.3 2.1 g-Value
75
100
125
150
Temperature (°C)
1.9
1.7
Figure 10.5 TEM images of FSP-derived (a) pristine CeO2 , as well as the CuO/CeO2 composite metal oxides with different Cu loadings: (b) 1 wt%, (c) 4 wt%, and (d) 12 wt%. (e) Electron paramagnetic resonance (EPR) spectra of the CuO/CeO2 particles showing the strong hyperfine features (K1 and K2 doublets and D hyperfine features) belonging to the surface CuO dimer at above 4 wt% Cu, while the C1–C3 triplets at lower Cu loadings are characteristic of CuO monomer. The CuO/CeO2 prepared by wet deposition/precipitation yielded only CuO monomer. (f ) High CO-PrOx activity at low temperature was recorded for CuO/CeO2 with dimeric species, while the monomeric CuO species require higher temperature for the same conversion. DP, deposition-precipitation. Source: Kydd et al. 2009 [15]. Adapted with permission of John Wiley & Sons.
situated on the surface, trace amount of Cu is inevitably doped into the CeO2 lattice, resulting in the gradual transformation from the rhombohedral shape to that of nonuniform polyhedral and ultimately spherical-shaped crystals as a function of Cu loadings (Figure 10.5). The incidental trace doping is beneficial in enhancing the transport of lattice oxygen to the Cu active site - an effect similar to that described above on improving the oxygen exchange capacity of Cex Zr1−x O2 through doping. This, coupled with the exclusive CO adsorption and the high O iconicity (from the Cu—O—Cu bond) of the CuO dimer, gives rise to the highly active low-temperature CO-PrOx within its ideal range of 100–150 ∘ C [22]. Modification of the same catalyst through the addition of Zr4+ , i.e., Cu/Cex Zr1−x O2 , was found to promote the formation of Cu+ species that in turn promoted the adsorption of NO for the selective catalytic reduction by CO (to form N2 and CO2 ) [23]. The surface enrichment of active metal oxide on a different oxide support has also been observed for FSP-derived V2 O5 /TiO2 and V2 O5 /SiO2 catalysts. With
193
10 Flame Synthesis of Simple and Multielemental Oxide Catalysts
dist a nc
e, x
unique dispersion of amorphous VOx (mono- and oligomeric) species, these catalysts are effective for the partial oxidation of o-xylene to phthalic anhydride [24] and the oxidative dehydrogenation of propane [25], respectively. The retainment of surface amorphous layer is likely due to the strong surface interactions between the VOx species and the support especially that of TiO2 . For that reason, the retainment of surface amorphous layer of up to 15% is possible on TiO2 , while crystalline V2 O5 forms on SiO2 at above 10% loading. This is rare for most wet synthesis techniques, where crystalline V2 O5 with low dispersion is typical [49]. On the other hand, crystalline V2 O5 was formed when preparing V2 O5 /Al2 O3 by FSP even at low loadings (5%) due to the weaker surface interactions between the two oxide species [26]. The synthesis of Co-based catalysts for Fischer–Tropsch synthesis, i.e., the conversion of syngas to paraffins, xCO + (2x + 1)H2 → Cx H(2x+2) + xH2 O, is a challenging one using FSP. While the standard catalysts of Co3 O4 /Al2 O3 and Co3 O4 /SiO2 are typically prepared by wet impregnation and followed by the reduction to metallic Co, the direct flame spraying of comixed Co and Al or Si precursors would result in irreducible and hence inactive cobalt aluminate or cobalt silicate, respectively. To circumvent the problem, the two-nozzle FSP was devised to allow the separate formation of Co3 O4 and Al2 O3 in each independent flame (Figure 10.6) [27]. The two aerosol flames need to be carefully interfaced at the point of mixing where the temperature is sufficiently low to prevent the formation of the cobalt aluminate or silicate. At that point of mixing, the fine Co3 O4 nanoparticles will be anchored onto the Al2 O3 surface to minimize further coalescence. A small amount of Pt promoter can be further added on the Co3 O4 side to further enhance the reducibility of the Co3 O4 [50]. The two-nozzle strategy has been explored for the synthesis of other composite catalysts, including the Pt/BaO/Al2 O3 NOx storage-reduction catalyst [51], (Ba, K)-doped Rh/Al2 O3 CO2 hydrogenation catalyst [52], Pt/FeOx /CeO2 low-temperature CO-PrOx
rsec
tion
4 × 10−4
zleto
-inte
3 × 10−4
2 × 10−4 x = 26 cm
Noz
CTY (molCO) (gCO S)
194
x = 51 cm
1 × 10−4
20° 0 0
10
20 30 Time on stream (h)
40
50
Co naphthenate/xylene
(a)
Al sec-butoxide/xylene
(b)
Figure 10.6 (a) Cobalt-time-yield (CTY) of the two Co/Al2 O3 catalysts prepared at different nozzle-to-intersection distance, x, in (b) a double FSP configuration where Co (in the form of Co3 O4 ) and Al2 O3 were precipitated independently. Source: (a) Minnermann et al. 2013 [27]. Adapted with permission of Elsevier.
10.4 Multielemental Oxide-Based Catalysts
H2 consumption (a.u.)
Low temperature High temperature region region
0.4Pd–20Co–ZrO2 0.4Rh–20Co–ZrO2 0.4t–20Co–ZrO2 0.4Ru–20Co–ZrO2 0.4Ag–20Co–ZrO2 20Co–ZrO2 0
100
Rate of CH4 fromation (10–6 mol/(s gcat))
(a)
300 400 500 Temperature (°C)
600
700
800
4 Co–ZrO2 0.4Rh–Co–ZrO2 0.4Ru–Co–ZrO2 0.4Pt–Co–ZrO2 0.4Pd–Co–ZrO2 0.4Ag–Co–ZrO2
3
2
1
0 0
(b)
200
60
120 180 240 300 360 Time on stream (min)
420
480
Figure 10.7 (a) Temperature-programmed reduction (H2 -TPR) of FSP-prepared Co3 O4 /ZrO2 doped with 0.4% of different noble metal promoters, showing the two-step reduction Co3 O4 to CoO at low temperature and CoO to Co0 at high temperature. The introduction of noble metal promoters significantly increased the reducibility of the catalyst, leading to (b) higher rates of CO methanation activity. Source: Teoh et al. 2015 [29]. Adapted with permission of Elsevier.
195
10 Flame Synthesis of Simple and Multielemental Oxide Catalysts
catalyst [16], and SiO2 /Ce0.7 Zr0.3 O2 for the dry reforming of methane [53], where controllable extent of phase segregation of the metal oxides was desired. An alternative to the two-nozzle strategy for synthesizing effective Fischer–Tropsch synthesis catalyst would be to seek alternative support material altogether. In that respect, the flame spraying of mixed Co and Zr precursors gives rise to a unique Co3 O4 /ZrO2 structure, where the Co3 O4 clusters are dispersed within the ZrO2 matrix [28, 29]. Unlike SiO2 and Al2 O3 , Co does not form a complex metal oxide with Zr (i.e., cobalt zirconate) and hence can be easily reduced. By adding a small amount of Ru (0.4%) one can enhance the reducibility and activity of the FSP catalyst to match that of the standard catalyst of the same composition prepared by incipient wetness impregnation. The same catalyst can also be used for the CO methanation reaction, where the addition of noble metals significantly enhances the reducibility of the Co3 O4 in the order of undoped < Ag < Ru < Pt < Rh < Pd (Figure 10.7). The CO methanation is an alternative reaction to the CO-PrOx, where instead of selectively oxidizing the trace CO, it converts them to CH4 (CO + 3H2 → CH4 + H2 O) at the expense of a small amount of H2 . Besides crystalline materials, the FSP can also be used to synthesize amorphous catalysts, as exemplified by the preparation of SiO2 –Al2 O3 solid acids for the conversion of glucose to levulinic acid [30]. Although the high flame temperature (as high as 2800 K) warrants the crystallization of most catalytic materials, the extremely short residence time (in the range of milliseconds) in an FSP flame is insufficient to crystallize materials such as C and SiO2 . The amorphous nature of SiO2 –Al2 O3 catalyst ensures a homogeneous mixing of Al3+ and Si4+ that gives rise to the Brønsted and Lewis acid sites with acidity strength matching that of the zeolites [54]. Unlike the well-defined framework acid in zeolite, however, the acid content in FSP-derived SiO2 –Al2 O3 is at least an order of magnitude lower even at comparable SSA. Nevertheless, the tunable mixture of Brønsted and Lewis acids is beneficial for obtaining high levulinic 100 Glucose conversion Fructose 5-HMF Levulinic acid Formic acid
80 60 40 20 0 0.0
(a)
Conversion or carbon yield (%)
Conversion or carbon yield (%)
196
0.2
0.4 0.6 0.8 Al/(Al + Si)
1.0
100 80 60 40 20 0 0.0
(b)
0.2
0.4
0.6
0.8
1.0
Si/(P + Si)
Figure 10.8 Glucose conversion and products yields over FSP-prepared amorphous (a) SiO2 /Al2 O3 and (b) SiO2 /AlPO4 solid acids with tunable compositions. The high specific surface area SiO2 /Al2 O3 and ZSM-5 (Al/(Si + Al) = 0.08) are represented as open and open cross symbols, respectively. 5-HMF, 5-Hydroxymethyl furfural. Source: Beh et al. 2020 [30]. Adapted with permission of Royal Society of Chemistry.
10.5 Perspective and Outlook
acid yield, outperforming the commercial zeolite X and ZSM-5 catalysts that consist predominantly of Brønsted acids (Figure 10.8). For the same reason, the FSP-derived, composition-tunable amorphous SiO2 –AlPO4 solid acids also show high selectivity toward levulinic acid (Figure 10.8). 10.4.3
Complex Metal Oxide Catalysts
The synthesis of complex metal oxides is perhaps one of the most exciting classes of FSP-derived compounds for catalytic applications but to date is also one of the most underexplored. The challenge lies in the limited flame residence time required for the crystallization of the complex oxide catalysts. Such is the case of FSP-derived BiVO4 , where inactive amorphous phase was formed upon leaving the flame. Luckily however, the transition of amorphous to the active monoclinic BiVO4 could take place on the collection filter above the flame; a process that is sensitive to the exposure temperature [55]. As a visible-light photocatalyst for oxygen evolution reaction, the high density of surface defects on the as-collected BiVO4 (which becomes a source of charge recombination centers) was limiting its performance [31]. Simple aqueous-based posttreatment to heal the surface defects is possible, resulting in high oxygen evolution activity (Figure 10.9). For high-temperature applications such as the catalytic (flameless) combustion of methane, thermal stability is an important criterion. Increasing the flame temperature during the FSP synthesis is a way to achieve this, as described earlier for the CeO2 three-way catalyst and exemplified by the thermally resistant LaFeO3 perovskite catalyst prepared at controlled amount of O2 feed [32]. Upon deposition of Pd, the same catalyst can also be active for the selective catalytic reduction of NO by H2 under lean condition, although precalcination at 800 ∘ C followed by the H2 reduction at 300 ∘ C is necessary [33]. The precalcination step in the presence of 10% O2 allowed the oxidation to PdO and the doping of Pd into the perovskite lattice. The subsequent reduction step resulted in the PdCo alloy on the LaFeO3−𝛿 [56]. Such formation of surface alloy cluster was thought to prolong the lifetime of the active site compared with Pd alone.
10.5 Perspective and Outlook The development of active catalysts by flame synthesis has come a long way to establish itself as one of the mainstream synthesis techniques. Much of it was made possible through tireless demonstration of the as-prepared active catalysts for various reactions and benchmarking against the best-known catalysts. Industry leaders such as Evonik, Johnson Matthey, DuPont, and Haldor Topsoe have implemented the flame synthesis within their manufacturing facilities to produce catalysts with complex functions. This will increasingly break away from the Generation I commodity-type simple metal oxide particles (e.g., Al2 O3 , SiO2 , and TiO2 ) that was the basis of initial growth in the field of flame aerosols synthesis. With more complex formulations, the synthesis of Generation II flame-derived catalysts, which include a wide variety of multielemental oxide catalysts, tend
197
10 Flame Synthesis of Simple and Multielemental Oxide Catalysts
O2 evolved (mol)
XRD intensity (a.u.)
Tf = 430
80
Tf 430 360 378 345 310
Tf = 360
60
Tf = 378
40
Tf = 345
20
Tf = 310 Tf = 301
301
0 0
10
20
(a)
30
40
50
1
2
60
3 4 Time (h)
5
6
(b)
Bragg angle (°)
400
301b 430b
300
O2 evolved (mol)
XRD intensity (a.u.)
198
430b
200
430
301a
100
301b
430
301a
301 0
301 10
(c)
0 20
30
40
Bragg angle (°)
50
60
1
2
3 4 Time (h)
5
6
(d)
Figure 10.9 (a) X-ray diffraction (XRD) spectra of as-prepared BiVO4 at different exposure temperature at the filter (T f ) during the FSP synthesis showing the evolution from amorphous to crystalline monoclinic phase as T f increases. (b) The increased crystallinity of BiVO4 with increasing T f was beneficial to improving the overall oxygen evolution activity under visible-light irradiation. (c) Aqueous treatment without (labeled a) and with small addition of Bi and V (labeled b) on BiVO4 samples prepared at T f of 301 and 430 ∘ C resulted in improved crystallinity and reduced defect content. (d) Treatment b resulted in much improved activity in oxygen evolution reaction. Source: Kho et al. 2011 [31]. Adapted with permission of American Chemical Society.
to target more specific niche reactions. It is perhaps worth mentioning that whether from the lab-scale synthesis or industrial manufacturing perspective [57], much is to be gained from the elegance of the rapid flame synthesis that at the same time allows for flexible tuning of the catalytic properties over a large window. In fact, the extreme conditions of high temperature and short residence time during flame synthesis gave rise to properties otherwise not achievable by other techniques, for example, the formation of highly-dispersed CuO dimers on CeO2 or the extremely high surface density of VOx on TiO2 . Unfortunately, the discovery of some of these superior physicochemical properties is at times serendipitous. While the evolution of the size of these catalytic particles can already be accurately predicted modeled through the coupled computational fluid dynamics (CFD) and particle growth simulation, the modelling of the
References
chemical phase evolution of these particles remains elusive. The complexity lies in the fact that these particles, because of their short residence time in the flame, exist in metastable states, rendering the application of traditional phase diagrams less useful. A more comprehensive description of the flame design of advanced catalysts would hence involve the application of time-dependent molecular dynamics simulation (see Chapter 24 for an overview of the technique, although not specifically targeting the problem as described here) that takes into account of the flame environment and residence time (from CFD), physical size of the particles (from particle growth model), and their interactions. As computing power becomes increasingly affordable, such unified simulation shall in the near future enable a truly high-performance, flame-derived catalysts by design. Along the same lines, ensuing opportunities shall arise from the design of Generation III flame particles, where high-performance catalysts are to be synthesized based on formulations not previously explored by other syntheses. The discovery of new catalysts using the flame synthesis, especially the FSP, has a number of advantages, including the straightforward tuning of the particle compositions, creation of metastable phases, higher concentrations of achievable doping, and the formation of highly crystalline particles with tunable amount of surface defects (energetic sites). In fact, the flame technique is well-suited for combinatorial synthesis in the effort of uncovering new catalytic materials.
References 1 Kodas, T.T. and Hampden-Smith, M.J. (1999). Aerosol Processing of Materials.
New York: Wiley-VCH. 2 Pratsinis, S.E. (1996). Powder Technol. 88: 267–273. 3 Strobel, R. and Pratsinis, S.E. (2007). J. Mater. Chem. 17: 4743–4756. 4 Mädler, L., Kammler, H.K., Mueller, R., and Pratsinis, S.E. (2002). J. Aerosol
Sci. 33: 369–389. 5 Pratsinis, S.E., Zhu, W., and Vemury, S. (1996). Powder Technol. 86: 87–93. 6 Kühner, G. and Voll, M. (1993). Manufacture of carbon black. In: Carbon
7 8 9 10 11 12 13 14 15
Black (eds. J.-B. Donnet, R.C. Bansal and M.-J. Wang). New York: Marcel Dekker. Teoh, W.Y., Amal, R., Mädler, L., and Amal, R. (2010). Nanoscale 2: 1324–1347. Klopfer, F. (1942). German Patent DE 762,723. Teoh, W.Y., Mädler, L., Beydoun, D. et al. (2005). Chem. Eng. Sci. 60: 5852–5861. Teoh, W.Y., Amal, R., Mädler, L., and Pratsinis, S.E. (2007). Catal. Today 120: 203–213. Gunawan, C., Teoh, W.Y., Marquis, C.P. et al. (2009). Small 5: 341–344. Fujiwara, K., Müller, U., and Pratsinis, S.E. (2016). ACS Catal. 6: 1887–1893. Height, M.J., Pratsinis, S.E., Mekasuwandumrong, O., and Praserthdam, P. (2006). Appl. Catal., B 63: 305–312. Dreyer, J.A.H., Li, P., Zhang, L. et al. (2017). Appl. Catal., B 219: 715–726. Kydd, R., Teoh, W.Y., Wong, K. et al. (2009). Adv. Funct. Mater. 19: 369–377.
199
200
10 Flame Synthesis of Simple and Multielemental Oxide Catalysts
16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47
Dreyer, J.A.H., Grossmann, H.K., Chen, J. et al. (2015). J. Catal. 329: 248–261. Strobel, R., Stark, W.J., Mädler, L. et al. (2003). J. Catal. 213: 296–304. Strobel, R., Krumeich, F., Stark, W.J. et al. (2004). J. Catal. 222: 307–314. Stark, W.J., Maciejewski, M., Mädler, L. et al. (2003). J. Catal. 220: 35–43. Stark, W.J., Grunwaldt, J.-D., Maciejewski, M. et al. (2005). Chem. Mater. 17: 3352–3358. Schulz, H., Stark, W.J., Maciejewski, M. et al. (2003). J. Mater. Chem. 13: 2979–2984. Kydd, R., Ferri, D., Hug, P. et al. (2011). J. Catal. 277: 64–71. Zhang, R., Teoh, W.Y., Amal, R. et al. (2010). J. Catal. 272: 210–219. Schimmoeller, B., Schulz, H., Ritter, A. et al. (2008). J. Catal. 256: 74–83. Schimmoeller, B., Jiang, Y.J., Pratsinis, S.E., and Baiker, A. (2010). J. Catal. 274: 64–75. Rossetti, I., Fabbrini, L., Ballarini, N. et al. (2009). Catal. Today 141: 271–281. Minnermann, M., Grossmann, H.K., Pokhrel, S. et al. (2013). Catal. Today 214: 90–99. Teoh, W.Y., Setiawan, R., Mädler, L. et al. (2008). Chem. Mater. 20: 4069–4079. Teoh, W.Y., Doronkin, D.E., Beh, G.K. et al. (2015). J. Catal. 326: 182–193. Beh, G.K., Wang, C.T., Kim, K. et al. (2019). Green Chem. 22 (3): 688–698. https://doi.org/10.1039/C9GC02567G. Kho, Y.K., Teoh, W.Y., Iwase, A. et al. (2011). ACS Appl. Mater. Interfaces 3: 1997–2004. Chiarello, G.L., Rossetti, I., and Forni, L. (2005). J. Catal. 236: 251–261. Chiarello, G.L., Ferri, D., Grunwaldt, J.-D. et al. (2007). J. Catal. 252: 137–147. Rosebrock, C.D., Wriedt, T., Mädler, L., and Wegner, K. (2015). AIChE J. 62: 381–391. Li, H., Rosebrock, C.D., Riefler, N. et al. (2017). Proc. Combust. Inst. 36: 1011–1018. Rosebrock, C.D., Riefler, N., Wriedt, T. et al. (2013). AIChE J. 59: 4553–4566. Gröhn, A.J., Pratsinis, S.E., and Wegner, K. (2012). Chem. Eng. J. 191: 491–502. Koirala, R., Pratsinis, S.E., and Baiker, A. (2016). Chem. Soc. Rev. 45: 3053–3068. Grass, R.N., Athanassiou, E.K., and Stark, W.J. (2007). Angew. Chem. Int. Ed. 46: 4909–4912. Teoh, W.Y., Denny, F., Amal, R. et al. (2007). Top. Catal. 44: 489–497. Teoh, W.Y. (2013). Materials 6: 3194–3212. Kho, Y.K., Iwase, A., Teoh, W.Y. et al. (2010). J. Phys. Chem. C 114: 2821–2829. Kho, Y.K., Teoh, W.Y., Mädler, L., and Amal, R. (2011). Chem. Eng. Sci. 66: 2409–2416. Tsekuoras, G., Miyashita, M., Kho, Y.K. et al. (2010). IEEE J. Sel. Top. Quantum Electron. 16: 1641–1648. Mädler, L., Stark, W.J., and Pratsinis, S.E. (2002). J. Mater. Res. 17: 1356–1362. Beier, M.J., Hansen, T.W., and Grunwaldt, J.-D. (2009). J. Catal. 266: 320–330. Di Monte, R. and Kaspar, J. (2005). J. Mater. Chem. 15: 633–648.
References
48 Hotz, N., Stutz, M.J., Loher, S. et al. (2007). Appl. Catal., B 73: 336–344. 49 Schimmoeller, B., Pratsinis, S.E., and Baiker, A. (2011). ChemCatChem 3:
1234–1256. 50 Schubert, M., Pokhrel, S., Thomé, A. et al. (2016). Catal. Sci. Technol. 6:
7449–7460. 51 Strobel, R., Piacentini, M., Mädler, L. et al. (2006). Chem. Mater. 18:
2532–2537. 52 Büchel, R., Baiker, A., and Pratsinis, S.E. (2014). Appl. Catal., A 477: 93–101. 53 Lovell, E., Großmann, H., Horlyck, J. et al. (2019). ACS Appl. Mater. Interfaces
11 (29): 25766–25777. https://doi.org/10.1021/acsami.9b02572. 54 Huang, J., van Vegten, N., Jiang, Y. et al. (2010). Angew. Chem. Int. Ed. 49:
7776–7781. 55 Strobel, R., Metz, H.J., and Pratsinis, S.E. (2008). Chem. Mater. 20:
6346–6351. 56 Chiarello, G.L., Grunwaldt, J.-D., Ferri, D. et al. (2007). J. Catal. 252: 127–136. 57 Wegner, K., Schimmoeller, B., Thiebaut, B. et al. (2011). Kona Powder Part. J.
29: 251–265.
201
203
11 Band Engineering of Semiconductors Toward Visible-Light-Responsive Photocatalysts Akihide Iwase Meiji University, School of Science and Technology, Department of Applied Chemistry, 1-1-1 Higashi-Mita, Tamaku, Kawasakishi, Kanagawa 214-8571, Japan
11.1 Basis of Photocatalyst Materials A photocatalyst is defined as a material that induces chemical reactions by absorbing photons. A photocatalyst consists of a homogeneous (molecular) photocatalyst and a heterogeneous (solid) photocatalyst. This chapter focuses on solid photocatalyst materials. The solid (semiconductor) photocatalyst possesses conduction and valence bands consisting of empty and electron-filled orbitals, respectively (Figure 11.1). The gap between the conduction and valence bands is called the bandgap. When the photocatalyst absorbs photons with energy equal to or larger than the bandgap, electrons in the valence band are excited to the conduction band. The excited electrons (e− ) lead to reduction on the surface of the photocatalyst. Positive holes (h+ ) are generated in the valence band after excitation of electrons, and the holes lead to oxidation. The reduction and oxidation power of the excited electrons and holes depend on the levels of the bottom of the conduction band (conduction band minimum [CBM]) and top of the valence band (valence band maximum [VBM]). For example, thermodynamic requirement for a photocatalyst for water splitting into H2 and O2 is that the levels of its CBM and VBM are more negative than the reduction potential of water to H2 (0 V vs. normal hydrogen electrode [NHE] at pH 0) and more positive than the oxidation potential of water to O2 (+1.23 V vs. NHE at pH 0), respectively [1–3]. Thus, the photocatalyst material should be designed depending on the objective reactions. We need to bear in mind that photocatalysts with thermodynamically sufficient band structures are not always active for water splitting [2]. Regarding the processes from the generation of photoexcited electrons and holes to reaction on the surface (Figure 11.2), the photogenerated electrons and holes need to migrate to the photocatalyst surface. Most photocatalyst particles contain grain boundaries and defects inside and/or at the surface, which work as charge recombination centers of the photogenerated electrons and holes. Surviving electrons and holes are used for the reduction and oxidation of water to H2 and O2 , respectively. However, the reactions hardly proceed when their overpotentials are large. To reduce Heterogeneous Catalysts: Advanced Design, Characterization and Applications, First Edition. Edited by Wey Yang Teoh, Atsushi Urakawa, Yun Hau Ng, and Patrick Sit. © 2021 WILEY-VCH GmbH. Published 2021 by WILEY-VCH GmbH.
11 Band Engineering of Semiconductors Toward Visible-Light-Responsive Photocatalysts
(–)
Conduction band e–
H2 Reduction H+
Photon (hv > BG)
Potential
204
Band gap (BG)
H 2O h+
O2/H2O (+1.23 V vs. NHE)
Oxidation
Valence band
(+)
H+/H2 (0 V vs. NHE)
O2
Figure 11.1 A band structure of a semiconductor photocatalyst and thermodynamic requirements for water splitting. Cocatalyst
H2
Photon
Figure 11.2 Process of water splitting on a semiconductor photocatalyst.
H+ e–
h+
Photocatalyst Recombination O2 H2O
Boundaries and defects
the overpotential, metals, metal oxides, and metal sulfides are loaded on the photocatalyst surface [4]. The loaded materials are called cocatalysts and work as reaction sites for H2 and O2 evolution.
11.2 Photocatalyst Material Groups 11.2.1
Variety of Photocatalyst Materials
Various materials including metal oxides [1, 2, 5], metal (oxy)sulfides [2, 6, 7], and metal (oxy)nitrides [7, 8] are used as photocatalysts. Most photocatalysts developed for water splitting so far are metal oxides such as TiO2 [9] and SrTiO3 [10, 11]. The valence band of the metal oxides is generally formed by O2p orbitals, and the VBM is located at around +3.0 V vs. NHE at pH 0 (Figure 11.3) [12]. The level of the VBM is much deeper than the oxidation potential of water to O2 (+1.23 V vs. NHE at pH 0). Such a deep valence band is thermodynamically favorable for water oxidation to O2 . However, the bandgap inevitably becomes larger
11.2 Photocatalyst Material Groups
Figure 11.3 A band structure of a metal oxide photocatalyst. Potential (V) vs. NHE (pH 0)
(–) CBM H+/H2
0 +1
Band gap (BG) >3.0 eV
+2 +3
VBM (+)
Valence band formed by O2p
than 3.0 eV corresponding to the absorption only UV light when the conduction band possesses high enough potential for water reduction to H2 (0 V vs. NHE at pH 0). Thus, it is important to sensitize a wide bandgap material to visible light by band engineering. Valence bands of metal (oxy)sulfides and metal (oxy)nitrides are formed by S3p and N2p orbitals, respectively [2, 6–8]. The valence bands are shallower than that formed by O2p orbitals due to the smaller electronegativity of N and S than that of O. Thus, most of the metal (oxy)sulfides and metal (oxy)nitrides possess narrower bandgaps than metal oxides. On the other hand, the metal (oxy)sulfides and metal (oxy)nitrides often exhibit low chemical stability in water under irradiation of light [2]. In more detail, photogenerated holes oxidize the metal (oxy)sulfides and metal (oxy)nitrides themselves to produce S and N2 . For example, CdS is known as a photocorrosive material [13]. CdS + 2h+ −−−−→ Cd2+ + S
(11.1)
Although H2 is generated accompanied by the above elution of Cd2+ , the H2 generation cannot be called a photocatalytic reaction. The photocatalytic activity for H2 evolution in the presence of sacrificial reagents is required for photocorrosive materials. Although less common, some metal oxides such as ZnO also show photocorrosion [14]. ZnO + 2h+ −−−−→ Zn2+ + 1∕2O2 11.2.2
(11.2)
Main Constituent Metal Elements in Photocatalyst Materials
Most of the active photocatalysts contain metal cations with either d0 electron configuration (Ti4+ , Zr4+ , V5+ , Nb5+ , Ta5+ , Mo6+ , W6+ ) [1, 2] or d10 electron configuration (Zn2+ , Ga3+ , In3+ , Ge4+ , Sn4+ ) [5, 8]. Metal cations with dn (n ≠ 0 or 10) electron configurations take various oxidation states, and hence they work as charge recombination centers. Fe3+ and Mn2+ with a d5 electron configuration are also usable metal cations for photocatalysts, although the number of such photocatalysts is limited [15, 16]. The metal cations with d0 and d10 electron configurations contribute to the formation of conduction bands. The conduction band of the materials containing metal cations with a d0 electron configuration is
205
11 Band Engineering of Semiconductors Toward Visible-Light-Responsive Photocatalysts
(–) Ta5d Conduction band
Nb4d V3d
Potential
206
BG
CBM Driving force for water reduction to H2 H+/H2
BG
BG VBM
Valence band (+)
Vanadate
Niobate
Tantalate
Figure 11.4 Relative band structures of vanadate, niobate, and tantalite.
formed by empty d orbitals, while that of the materials containing metal cations with a d10 electron configuration is formed by hybridized s and p orbitals. In congeners, a metal cation with larger atomic number generally forms a higher conduction band. For example, CBMs formed by V3d in V5+ , Nb4d in Nb5+ , and Ta5d in Ta5+ usually become higher (more positive) in order of V < Nb < Ta (Figure 11.4). The conduction bands formed by Nb4d and Ta5d possess thermodynamically sufficient potential for water reduction to form H2 . Actually, niobates and tantalates are promising material groups for highly efficient water splitting into H2 and O2 [2]. In contrast, most of vanadates do not show activity for water reduction to H2 due to the low CBM level [17]. Although higher conduction band provides a larger driving force for water reduction to H2 , the bandgap also becomes wider.
11.3 Design of Band Structures of Photocatalyst Materials Strategies for the band engineering toward responding to the long-wavelength light can be divided into “formation of impurity levels by doping,” “control of the valence band,” and “formation of solid solution” [2]. 11.3.1
Doped Photocatalysts
Some elements form impurity levels in the forbidden band of a host material when small amounts (few atomic %) of the elements are doped into the material. For example, SrTiO3 is a white powder with a bandgap of 3.2 eV [10, 11], while SrTiO3 doped with Rh works as a visible-light-driven photocatalyst for H2 evolution in the presence of sacrificial reagents [18]. This is because the doped Rh ions formed impurity levels in the forbidden band of the SrTiO3 host (Figure 11.5). The Rh is originally doped as tetravalent ions at Ti sites and forms an electron acceptor level. During H2 evolution under visible-light irradiation in the presence of methanol as a sacrificial reagent, the doped Rh4+ ions are reduced to Rh3+ by
11.3 Design of Band Structures of Photocatalyst Materials
(–)
Potential
Ti3d
Ti3d
Ti3d
Impurity levels formed by Rh4+
3.2 eV
2.3 eV Impurity levels formed by Rh3+
1.7 eV O2p
O2p
O2p
SrTiO3
Rh-doped SrTiO3 in the dark
Rh-doped SrTiO3 under irradiation
(+)
Figure 11.5 Band structures of SrTiO3 and Rh-doped SrTiO3 in the dark and under irradiation.
photogenerated electrons at the beginning stage. The newly reduced Rh3+ forms an electron donor level in the forbidden band of the SrTiO3 host [19]. The H2 evolution proceeds by excitation from the electron donor level to the conduction band. In addition to Rh, Ir and Ru are also useful dopants to give visible-light response by forming electron donor levels [18, 20]. Although metal cation doping is widely used to develop photocatalysts with response to visible light for a long time, photocatalytic activity decreases in most cases. The dopants easily take various oxidation states, and hence they work as charge recombination centers between photogenerated electrons and holes. To address this issue, codoping to keep the charge balance is a useful strategy to stabilize a certain oxidation state of the dopants. For example, unstable Cr3+ is doped when only Cr is doped at Ti sites in SrTiO3 . Accordingly, only Cr-doped SrTiO3 does not show photocatalytic activity under visible-light irradiation although it possesses absorption band in the visible-light region. In contrast, stable Cr3+ can be doped upon keeping the charge balance when either Sb5+ or Ta5+ is codoped with Cr3+ at Ti sites, according to Eq. (11.3) [21, 22]. 2Ti4+ −−−−→ Cr3+ + Sb5+ (or Ta5+ )
(11.3)
The Cr- and Sb-codoped SrTiO3 shows photocatalytic activity under visible-light irradiation by excitation from impurity levels formed by Cr3+ to the conduction band of SrTiO3 (Figure 11.6a). The possibility of the charge recombination in such photocatalysts with keeping the charge balance is usually lower than that in the photocatalysts without keeping the charge balance. Anions such as N are also used as dopants [23]. The TiO2 doped with N at O sites possesses an absorption band in the visible-light region in addition to the bandgap absorption of the TiO2 host. Even for the anion doping, codoping is also useful to maintaining the charge balance. Codoping of F into TiO2 with N at O sites stabilizes the doped N as follows [24]. 2O2− −−−−→ N3− + F−
(11.4)
The photogenerated holes in the impurity levels formed by the stabilized N3− in N- and F-codoped TiO2 oxidize water to O2 , being different from only N-doped TiO2 .
207
11 Band Engineering of Semiconductors Toward Visible-Light-Responsive Photocatalysts
(–) Mo4d Ti3d Impurity levels formed by Cr6+
2.4 eV
Potential
208
3.2 eV
3.2 eV
2.3 eV
Impurity levels formed by Cr3+
Pb6s + O2p O2p (+)
(b)
(a)
Figure 11.6 Band structures of (a) Cr- and Sb-codoped SrTiO3 and (b) Cr-doped PbMoO4 .
Another doping strategy to sensitize wide bandgap materials to visible light is the formation of electron acceptor levels. Cr-doped PbMoO4 is the representative example [25]. Cr is doped as hexavalent at Mo sites in this case, being different from Cr- and Sb-codoped SrTiO3 . PbMoO4 is a valence-band-controlled photocatalyst as will be described in Section 11.3.2. The conduction and valence bands of PbMoO4 are formed by Pb6s and Mo4d, respectively. The doped Cr6+ forms an electron acceptor level in the forbidden band of PbMoO4 (Figure 11.6b). Cr-doped PbMoO4 shows photocatalytic activity for O2 evolution in the presence of sacrificial reagents under visible-light irradiation by excitation from the valence band to the electron acceptor level. Unfortunately, the electron acceptor level possesses insufficient potential for water reduction to H2 . 11.3.2
Valence-Band-Controlled Photocatalysts
As mentioned in Chapter 12, the valence bands are usually formed by O2p, S3p, and N2p orbitals in metal oxides, metal (oxy)sulfides, and metal (oxy)nitrides, respectively. In materials containing a certain metal ion, the metal ion contributes to the formation of the valence band above those formed by O2p, S3p, and N2p orbitals. For example, Cu(I), Ag(I), Pb(II), Sn(II), and Bi(III) are effective metal ions to reduce the bandgap of a metal oxide by forming a new valence band above the O2p orbitals [2, 26]. Among them, Cu(I) especially forms a shallow valence band. For example, Cu(Li1/3 Ti2/3 )O2 [27] and Cu(I)-substituted Ka2 La2 Ti3 O10 [28] photocatalysts produce H2 from water in the presence of a sacrificial reagent by responding to visible light up to 600 nm. In both Cu(I)-containing photocatalysts, the H2 evolution proceeds by excitation from the valence bands formed by Cu3d orbitals to the conduction band formed by Ti3d orbitals (Figure 11.7a). However, the valence band formed by Cu3d orbitals is usually too shallow to oxidize water to O2 . The valence bands formed by Ag(I), Pb(II), Sn(II), and Bi(III) possess enough potential to oxidize water to O2 . For example, BiVO4 with a bandgap of 2.4 eV is a representative photocatalyst for highly efficient water oxidation to O2 under visible-light irradiation [29]. Its valence band is formed by hybridized Bi6s and
11.3 Design of Band Structures of Photocatalyst Materials
(–) Ti3d Nb4d
Potential
2.1 eV
V3d
H+/H2 2.3 eV
2.4 eV O2/H2O
Cu3d Sn5s Bi6s (+) (a)
(b)
(c)
Figure 11.7 Band structures of (a) Cu(Li1/3 Ti2/3 )O2 , (b) BiVO4 , and (c) SnNb2 O6 .
O2p orbitals and locates shallower than the typical valence band formed by only O2p orbitals (Figure 11.7b) [30]. Unfortunately, the bottom of its conduction band formed by V3d orbitals is located at more positive than the potential for water reduction to H2 , and hence BiVO4 is only active for water oxidation to O2 . Although such photocatalysts are active in sacrificial H2 or O2 evolution, they are not suitable for water splitting by one-step photoexcitation. The use of Z-scheme and photoelectrode systems is required to split water into H2 and O2 . Actually, BiVO4 is widely used as freely suspended O2 -evolving photocatalyst in Z-scheme systems, as well as photoanode in photoelectrochemical systems [2, 31]. SnNb2 O6 with a bandgap of 2.3 eV is an active photocatalyst for both H2 and O2 evolutions under visible-light irradiation in the presence of sacrificial reagents [32]. The valence band is formed by Sn5s orbitals, while the conduction band is formed by Nb4d orbitals (Figure 11.7c). Note that overall water splitting into H2 and O2 using SnNb2 O6 has not been achieved yet, although SnNb2 O6 has thermodynamically sufficient band potential for the reaction. The chemical stability of Sn(II) sometimes becomes a problem not unlike other Sn(II)-containing materials. H2 evolution is observed under photoirradiation accompanied by oxidation of Sn2+ to Sn4+ by photogenerated holes, being regarded as photocorrosion. Characterization after photocatalytic reactions is quite important for these photocorrosive materials. Cu(I) in metal oxides may also have a problem with the chemical stability. In contrast, Cu(I) is stabilized in metal sulfides, while Ag(I) in metal sulfides is also effective in reducing the bandgaps. In other words, Cu3d and Ag4d orbitals form shallower valence bands than S3p orbitals. Most of the metal sulfides with narrow bandgaps contain Cu(I) or Ag(I) [6].
11.3.3
Solid Solution Photocatalysts
When a solid solution between materials with narrow and wide bandgaps is formed, the bandgap of the solid solution can generally be controlled depending
209
11 Band Engineering of Semiconductors Toward Visible-Light-Responsive Photocatalysts
(–) Zn4s4p
Zn4s4p + In5s5p In5s5p H+/H2
Potential
210
1.8 eV
3.5 eV
S3p
S3p + Ag4d
ZnS
(AgIn)xZn2(1–x)S2
S3p + Ag4d
(+) AgInS2
Figure 11.8 Band structures of ZnS, (AgIn)x Zn2(1−x) S2 , and AgInS2 .
on the mixing ratio. This strategy is mainly applied to nonoxide materials such as metal sulfides and selenides. Solid solutions between AgInS2 and ZnS are possibly formed, because both AgInS2 and ZnS can take the wurtzite structure [33]. The conduction and valence bands of AgInS2 are formed by In5s5p orbitals and hybridized S3p + Ag4d orbitals, respectively, and the bandgap is 1.8 eV (Figure 11.8). The conduction and valence bands of ZnS are formed by Zn4s4p and S3p orbitals, respectively, and the bandgap is 3.5 eV. The bandgap of the solid solution can be controlled in the range of 1.8–3.5 eV by changing the ZnS/AgInS2 ratio. The conduction bands of the solid solutions are formed by Zn4s4p and In5s5p orbitals, while the valence bands are formed by hybridized S3p + Ag4d orbitals. The contributions of elements to the conduction and valence bands depend on the ZnS/AgInS2 ratio, giving different bandgaps. The solid solutions possess much higher photocatalytic abilities for H2 evolution under visible-light irradiation than AgInS2 and ZnS. Narrowing of the bandgap increases the number of absorbable photons as a positive effect. In contrast, lowering of CBM decreases the driving force for water reduction to H2 as a negative effect. The balance between the positive and negative effects determines the photocatalytic activities of the solid solutions.
11.4 Preparation of Photocatalysts Most semiconductor photocatalysts are a combination of various metal cations and anions such as SrTiO3 . Such complex materials are usually prepared by reacting starting materials containing one metal cation. Even for photocatalysts of simple materials such as TiO2 and ZnS, the preparation process drastically affects their photocatalytic properties. In this section, representative synthesis methods for semiconductor photocatalysts are introduced. Each method has its own advantages and disadvantages, and hence suitable synthesis method depends on the photocatalyst material. Moreover, we need to bear in mind that
11.4 Preparation of Photocatalysts
a material prepared by a synthesis method can vary from a material prepared by other synthesis method even though the material is expressed by the same chemical formula. 11.4.1
Solid-State Reaction Method
A solid-state reaction method is widely and conventionally used for the synthesis of ceramic materials. The powdered starting materials are reacted by calcining at high temperatures for an extended duration after mixing and grinding well using a mortar and pestle. When the sample contains unstable species such as Cu(I) and materials such as sulfides and nitrides, the starting materials should be calcined in a vacuum or inert gases to avoid the oxidation by oxygen in air. Large aggregated particles with small surface area are usually obtained due to the calcination at high temperatures. The compositional homogeneity in each particle is not very high. 11.4.2
Flux Method
Flux method is similar to a solid-state reaction method. In flux method, fluxing agents are added in starting materials [34]. The calcination temperature should be higher than the melting point of the fluxing agent, allowing the crystals grow via dissolution–precipitation processes in the melted flux. The fluxing agents should be removed after the synthesis. A sample prepared by the flux method is usually large faceted particles with small surface area. Compositional homogeneity in one particle is high, and the particles do not contain many defects. Accordingly, crystallinity is usually high. 11.4.3 Hydrothermal Synthesis Method/Solvothermal Synthesis Method Hydrothermal reaction is a reaction in water under the condition with a temperature equal to or higher than 373 K and a pressure equal to or higher than 1 bar [35, 36]. The starting materials and water are heated in an autoclave. Crystals grow via dissolution–precipitation processes in the heated water. When an organic solvent such as alcohol is used instead of water, the process is called as solvothermal synthesis. Faceted particles reflecting the crystal structure are obtained. Moreover, particle size can be controlled upon changing the synthesis condition. Compositional homogeneity in each particle is high. 11.4.4
Polymerized (Polymerizable) Complex Method
Polymerized complex method is one of the sol–gel process-based synthesis methods [37]. Metal complexes are formed and subsequently polymerized in aqueous or alcohol media. The formed polymer is thermally decomposed to make an oxide precursor, and the precursor is further calcined to crystallize. The
211
212
11 Band Engineering of Semiconductors Toward Visible-Light-Responsive Photocatalysts
calcination temperature and time are usually lower and shorter, respectively, than those required for a solid-state reaction. Aggregated particles with featureless shape are usually obtained. Compositional homogeneity in each particle is high. 11.4.5
Precipitation Method
A precursor with low solubility is precipitated by adding suitable anions such as carbonate ions and sulfide ions into an aqueous metal salt solution to react with metal cations. Upon increasing the pH of an aqueous metal salt solution, metal ions are also precipitated as metal hydroxides. The precipitates are calcined to increase the crystallinity, if necessary. The precipitates are usually fine, crystalized particles with large surface area. When the crystallinity is not high enough, calcination is required. 11.4.6
Loading of Cocatalysts
Cocatalysts function as active sites to boost reduction and/or oxidation on the surface of photocatalysts, enhancing the photocatalytic activity [4]. The effect of the loaded cocatalyst on the photocatalytic activity depends on the loading method. Here, photodeposition and impregnation are introduced as basic deposition methods of cocatalysts. During the photodeposition, dissolved precursor ions of a cocatalyst in an aqueous solution are reduced to metal or metal oxide by photogenerated electrons and deposited on photocatalyst particles in the presence of hole scavengers such as methanol and Na2 S. When a suitable electron scavenger is used, the cocatalyst is oxidatively deposited by photogenerated holes. This method can be used to clarify whether the exposed surface works as reduction or oxidation sites [38]. In the more conventional impregnation method, photocatalyst powder is suspended in a precursor solution, and followed by heating of the solvent. The dried and impregnated powder is calcined in air to obtain a metal oxide cocatalyst, or in an atmosphere of H2 to obtain a metal cocatalyst. The calcination temperature should be given attention because calcination at high temperatures results in aggregated cocatalysts.
References 1 Osterloh, F.E. (2008). Chem. Mater. 20: 35–54. 2 Kudo, A. and Miseki, Y. (2009). Chem. Soc. Rev. 38: 253–278. 3 Walter, M.G., Warren, E.L., McKone, J.R. et al. (2010). Chem. Rev. 110: 4 5 6 7
6446–6473. Maeda, K. and Domen, K. (2016). Bull. Chem. Soc. Jpn. 89: 627–648. Inoue, Y. (2009). Energy Environ. Sci. 2: 364–386. Zhang, K. and Guo, L. (2013). Catal. Sci. Technol. 3: 1672–1690. Takata, T. and Domen, K. (2017). Dalton Trans. 46: 10529–10544.
References
8 Maeda, K. (2013). Phys. Chem. Chem. Phys. 15: 10537–10548. 9 Yamaguti, K. and Sato, S. (1985). J. Chem. Soc., Faraday Trans. 1 F 81:
1237–1246. 10 Lehn, J.M., Sauvage, J.P., and Ziessel, R. (1980). Nouv. J. Chim. 4: 623–627. 11 Domen, K., Naito, S., Soma, S. et al. (1980). J. Chem. Soc., Chem. Commun.:
543–544. 12 Scaife, D.E. (1980). Sol. Energy 25: 41–54. 13 Williams, R. (1960). J. Chem. Phys. 32: 1505–1514. 14 Rao, M.V., Rajeshwar, K., Verneker, V.R.P., and DuBow, J. (1980). J. Phys.
Chem. 84: 1987–1991. 15 Kakuta, S. and Abe, T. (2009). J. Mater. Sci. 44: 2890–2898. 16 Takayama, T., Tsuji, I., Aono, N. et al. (2017). Chem. Lett. 46: 616–619. 17 Lee, D.K., Lee, D., Lumley, M.A., and Choi, K.S. (2018). Chem. Soc. Rev.
https://doi.org/10.1039/c8cs00761f. 18 Konta, R., Ishii, T., Kato, H., and Kudo, A. (2004). J. Phys. Chem. B 108:
8992–8995. 19 Kawasaki, S., Akagi, K., Nakatsuji, K. et al. (2012). J. Phys. Chem. C 116:
24445–24448. 20 Iwase, A. and Kudo, A. (2017). Chem. Commun. 53: 6156–6159. 21 Kato, H. and Kudo, A. (2002). J. Phys. Chem. B 106: 5029–5034. 22 Ishii, T., Kato, H., and Kudo, A. (2004). J. Photochem. Photobiol., A 163:
181–186. 23 Asahi, R., Morikawa, T., Ohwaki, T. et al. (2001). Science 293: 269–271. 24 Miyoshi, A., Vequizo, J.J.M., Nishioka, S. et al. (2018). Sustainable Energy
Fuels 2: 2025–2035. 25 Shimodaira, Y., Kato, H., Kobayashi, H., and Kudo, A. (2007). Bull. Chem. Soc.
Jpn. 80: 885–893. 26 Sullivan, I., Zoellner, B., and Maggard, P.A. (2016). Chem. Mater. 28: 27 28 29 30 31 32 33 34 35 36 37 38
5999–6016. Iwashina, K., Iwase, A., Nozawa, S. et al. (2016). Chem. Mater. 28: 4677–4685. Iwashina, K., Iwase, A., and Kudo, A. (2015). Chem. Sci. 6: 687–692. Kudo, A., Ueda, K., Kato, H., and Mikami, I. (1998). Catal. Lett. 53: 229–230. Cooper, J.K., Gul, S., Toma, F.M. et al. (2014). Chem. Mater. 26: 5365–5373. Park, Y., McDonald, K.J., and Choi, K.S. (2013). Chem. Soc. Rev. 42: 2321–2337. Hosogi, Y., Shimodaira, Y., Kato, H. et al. (2008). Chem. Mater. 20: 1299–1307. Tsuji, I., Kato, H., Kobayashi, H., and Kudo, A. (2004). J. Am. Chem. Soc. 126: 13406–13413. Liu, X., Fechler, N., and Antonietti, M. (2013). Chem. Soc. Rev. 42: 8237–8265. O’Hare, D. (2001). Encyclopedia of Materials: Science and Technology, 2e (eds. K.H.J. Buschow, R.W. Cahn, M.C. Flemings, et al.), 3989–3992. Elsevier. Shi, W., Song, S., and Zhang, H. (2013). Chem. Soc. Rev. 42: 5714–5743. Kakihana, M. (2009). J. Ceram. Soc. Jpn. 117: 857–862. Ohno, T., Sarukawa, K., and Matsumura, M. (2002). New J. Chem. 26: 1167–1170.
213
215
Section II Surface Studies and Operando Spectroscopies in Heterogeneous Catalysis
217
12 Toward Precise Understanding of Catalytic Events and Materials Under Working Conditions Atsushi Urakawa Delft University of Technology, Department of Chemical Engineering, Catalysis Engineering, Van der Maasweg 9, Delft 2629 HZ, The Netherlands
Heterogeneous catalysis is indispensable for major chemical and energy conversion processes and for environmental protection to improve and sustain the quality of our lives and the environment. Despite the obvious importance, catalysts and catalytic processes are traditionally and even today developed using trial-and-error approaches. This is largely due to the intrinsic complexity of heterogeneously catalyzed processes involving various physical and chemical events such as electron transfer, atom motion, atomic and molecular sorption (adsorption and desorption), chemical reaction, molecular diffusion within the pore and outside of catalyst pellets, and fluid dynamics on the pellet to reactor scale (Figure 12.1). These events are taking place on the length scale of picometer to meter (10−12 –100 m) and on the time scale of attoseconds (10−15 seconds, e.g. for electron transfer) up to months and years (e.g. catalyst deactivation). In principle, precise comprehension of catalytic reactions expressed by conversion of reactants and product selectivity requires full understanding of all these events varying in time and length, which seems to be an impossible task. The most critical process among the different length scales is that taking place on the atomic scale because this defines the nature of catalysis; in other words, if this molecular-level catalysis does not work, nothing will happen on other length scales. For this reason, most advances in catalyst characterization are focused on precise understanding of catalytic activity through in-depth understanding of catalyst materials, active sites, and active species under controlled atmosphere (in situ) and, particularly, under catalysts’ working (operando) conditions [2–4]. The necessity of the latter has been stressed increasingly in the past few decades due to some gaps between the model catalyst and the real-world catalyst. Historically, heterogeneous catalysts were widely studied under ultrahighvacuum (UHV) conditions to precisely understand the surface structure and adsorbates over catalyst surfaces by X-ray photoelectron spectroscopy (XPS), low-energy electron diffraction (LEED), electron energy loss spectroscopy (EELS), low-energy electron microscopy (LEEM), and photoemission electron microscopy (PEEM), among other techniques [5, 6]. There were good rationales that these surface-sensitive techniques had to be carried out under UHV, which Heterogeneous Catalysts: Advanced Design, Characterization and Applications, First Edition. Edited by Wey Yang Teoh, Atsushi Urakawa, Yun Hau Ng, and Patrick Sit. © 2021 WILEY-VCH GmbH. Published 2021 by WILEY-VCH GmbH.
218
12 Toward Precise Understanding of Catalytic Events and Materials Under Working Conditions
Figure 12.1 Representative physicochemical events showing the complexity of heterogeneous catalytic processes [1].
evolved from Irving Langmuir’s early experiment on tungsten filament in a vacuum (light) bulb. Firstly, the surface could be cleaned of adsorbed impurities when heated to above Tamman temperature (∼half of melting point) under UHV to minimize the risk of airborne impurities readsorbing onto the surface. Secondly, the analyses of surfaces are best probed by the diffracted or emitted electrons since these low-energy electrons have extremely short mean free path of less than 2 nm in most solid materials. This means that only electrons from the near surface containing local information of the source environment are able to leave the surface and detected. Thirdly, unlike longer waves such as X-rays, electrons are easily diffracted and absorbed by any gas molecules en route to the sample (if using electron beam as probing source) or from sample to the detector, which means that the samples and detectors need to be contained in a UHV chamber. Although adsorbate molecules can be introduced into the UHV chamber after surface cleaning, non-adsorbed or loosely adsorbed molecules would need to be evacuated prior to analyses. For techniques such as LEED, model surfaces of single crystals are required. Over the last few decades, the precise surface information gained by the advancement of UHV and electron-probing techniques were formative to the foundation of surface chemistry and heterogeneous catalysis. However, more meticulous information is increasingly being demanded by the field in order to comprehend how active sites and species are formed over technical catalysts or even over model catalyst under efficiently reactive environment that is often under high pressure and temperature. This is because the catalyst structures, especially those on the surface, are dynamic, changing in response to the atmosphere, temperature, and the types and concentration of surface adsorbates. Three gaps, namely, pressure, temperature, and material gaps (Figure 12.2), arise due to such material structures and states uniquely formed under specific
12 Toward Precise Understanding of Catalytic Events and Materials
Industrial reaction P gap T
UHV
Material gap T gap In situ
P
Figure 12.2 Three major gaps (material gap, temperature gap, and pressure gap) often causing problems in practical relevance of spectroscopic and diffraction studies of catalyst materials. These gaps have to be filled to address real situation of catalyst material at work under technologically relevant conditions. Source: Adapted with permission from Urakawa [2]. Copyright 2016, Elsevier.
operando conditions, and the material gap arises largely due to the complex structure (defective nanostructures and with promoters) of technical catalysts that are often necessary for the catalysts to be highly active [7]. With the aim to fill these gaps, recently remarkable progresses are witnessed to perform in situ and especially operando spectroscopy. Spectroscopy and microscopy using incident and emitted electrons suffer from the short mean free path of electrons under non-vacuum conditions, but there are great advances, for example, in transmission electron microscopy (TEM) [8] and XPS [9, 10] using differential pressure techniques to enable detection of electrons under reactive atmosphere, albeit there is still an unavoidable pressure gap for most catalytic reactions. In this respect, new cell design, fully covering the sample under relevant pressure or even in the presence of liquid phase [11], for TEM studies should be highlighted, and the field is moving drastically forward to shed light on the structure of catalysts under operando conditions. A wider availability of synchrotron X-rays has impacted operando studies of catalyst materials not just because X-rays do not require confinement within UHV but because the extremely bright light source at synchrotron facilities allows unprecedented sensitivity of the measurements [12]. The wide energy band of synchrotron X-ray source was further advantageous for techniques based on X-ray absorption spectroscopy (XAS) that is quickly becoming the analytical norm in heterogeneous catalysis. Also, rather traditional analytical methods have evolved for this aim. For example, various sampling configurations have been developed for infrared (IR) spectroscopy to study gas/solid and solid/liquid interfaces under reactive environment [13], and electron paramagnetic resonance (EPR) can be nowadays used to sensitively uncover electronic states of catalysts under in situ and operando conditions [14]. It is also important to notice that there are more examples of the use of complementary multiple analytical (multimodal/multiprobe) methods simultaneously to elucidate the nature of catalytic reactions to add different views on catalytic processes for holistic understanding [2, 15–17]. Furthermore, the importance of heterogeneity of catalyst materials and various gradients, such as temperature, catalyst structure, oxidation state, surface
219
220
12 Toward Precise Understanding of Catalytic Events and Materials Under Working Conditions
species, and species in fluid phase, within catalytic reactors is increasingly recognized. In other words, catalyst material structure and physicochemical events not only dynamically change with time but also often vary spatially on the scale of atomic-level surface, pellet, and reactor [18, 19]. This is natural if a reaction takes place and there are intrinsic gradients of concentration and temperature due to the molecular conversion and exo- and endothermicity of chemical reactions. Development of in situ and operando methodologies along this direction is significant as represented by the use of X-ray tomography for catalyst design on the pellet scale [20] and space- and time-resolved spectroscopy on the reactor scale [18, 21]. On the other hand, there is need to more precisely probe the catalytic processes with better time and space resolution. For the former, ultrafast spectroscopy often enabled by pump–probe scheme is gaining popularity in taking snapshots of catalyst material structure, electronic state, and chemical transformation in real-time action [22–25]. For the latter, it has become possible to resolve the performance of catalyst in nanoscale using fluorescence spectromicroscopy [26, 27]. All this high-resolution information is pivotal in allowing efficient chemical engineering design and reactor scale-up solely from first principles. While performing in situ and operando studies to elucidate active sites and species under reaction conditions, one would realize that any analytical techniques are not selective toward the information we wish to gain, for example, seeing only active surface chemical species. This often limits our understanding of catalytic processes since spectator species, reactants, and products generally overwhelm the detected signals. To gain selectivity as well as sensitivity, there are advances in the experimental and post-processing mathematical methodologies discerning active ones from the others. The representative ones are modulation excitation spectroscopy [28], steady-state isotopic transient kinetic analysis (SSITKA) [29], and multivariate spectral analysis [30, 31]. In the following chapters (Chapters 13–21), recent developments and emerging analytical methodologies are comprehensively described. It should be noted that the coverage is not in full and some important and emerging operando methods such as scanning probe microscopy [32, 33], tip-enhanced Raman [34–37], advanced nuclear magnetic resonance (NMR) [38], magnetic resonance imaging (MRI) [39, 40], machine learning for spectral analysis [41, 42], and local temperature measurements [43], among many others, are not included, and interested readers are encouraged to look into the references.
References 1 Urakawa, A. (2006). Combined Spectroscopic and Theoretical Elucidation of
Catalytic Reaction Pathways. ETH Zurich. 2 Urakawa, A. (2016). Trends and advances in Operando methodology. Curr.
Opin. Chem. Eng. 12: 31–36. 3 Bañares, M.A. (2005). Operando methodology: combination of in situ spec-
troscopy and simultaneous activity measurements under catalytic reaction conditions. Catal. Today 100: 71–77.
References
4 Weckhuysen, B.M. (2003). Determining the active site in a catalytic process:
5 6
7 8
9
10
11
12
13
14 15
16
17
18 19
Operando spectroscopy is more than a buzzword. Phys. Chem. Chem. Phys. 5: 4351–4360. Prieto, M.J. and Schmidt, T. (2017). LEEM and PEEM as probing tools to address questions in catalysis. Catal. Lett. 147: 2487–2497. Freund, H.J., Baumer, M., Libuda, J. et al. (2003). Preparation and characterization of model catalysts: from ultrahigh vacuum to in situ conditions at the atomic dimension. J. Catal. 216: 223–235. Rupprechter, G. and Weilach, C. (2007). Mind the gap! Spectroscopy of catalytically active phases. Nano Today 2: 20–29. Jiang, Y., Zhang, Z.F., Yuan, W.T. et al. (2018). Recent advances in gas-involved in situ studies via transmission electron microscopy. Nano Res. 11: 42–67. Nguyen, L., Tao, F.F., Tang, Y. et al. (2019). Understanding catalyst surfaces during catalysis through near ambient pressure X-ray photoelectron spectroscopy. Chem. Rev. 119: 6822–6905. Takagi, Y., Uruga, T., Tada, M. et al. (2018). Ambient pressure hard X-ray photoelectron spectroscopy for functional material systems as fuel cells under working conditions. Acc. Chem. Res. 51: 719–727. Wu, J.B., Shan, H., Chen, W.L. et al. (2016). In situ environmental TEM in imaging gas and liquid phase chemical reactions for materials research. Adv. Mater. 28: 9686–9712. Newton, M.A. (2017). Time resolved operando X-ray techniques in catalysis, a case study: CO oxidation by O2 over Pt surfaces and alumina supported Pt catalysts. Catalysts 7 (2): 58. Zaera, F. (2014). New advances in the use of infrared absorption spectroscopy for the characterization of heterogeneous catalytic reactions. Chem. Soc. Rev. 43: 7624–7663. Bruckner, A. (2014). In situ EPR spectroscopy in heterogeneous catalysis: stepchild or ray of hope? Chem. Ing. Tech. 86: 1871–1882. Newton, M.A. and van Beek, W. (2010). Combining synchrotron-based X-ray techniques with vibrational spectroscopies for the in situ study of heterogeneous catalysts: a view from a bridge. Chem. Soc. Rev. 39: 4845–4863. Hinokuma, S., Wiker, G., Suganuma, T. et al. (2018). Versatile IR spectroscopy combined with synchrotron XAS–XRD: chemical, electronic, and structural insights during thermal treatment of MOF materials. Eur. J. Inorg. Chem. 2018 (17): 1847–1853. Bentrup, U. (2010). Combining in situ characterization methods in one set-up: looking with more eyes into the intricate chemistry of the synthesis and working of heterogeneous catalysts. Chem. Soc. Rev. 39: 4718–4730. Urakawa, A. and Baiker, A. (2009). Space-resolved profiling relevant in heterogeneous catalysis. Top. Catal. 52: 1312–1322. Weckhuysen, B.M. (2009). Chemical imaging of spatial heterogeneities in catalytic solids at different length and time scales. Angew. Chem. Int. Ed. 48: 4910–4943.
221
222
12 Toward Precise Understanding of Catalytic Events and Materials Under Working Conditions
20 Beale, A.M., Jacques, S.D.M., Gibson, E.K., and Di Michiel, M. (2014).
21
22 23
24 25
26
27
28
29
30 31
32
33
34
Progress towards five dimensional diffraction imaging of functional materials under process conditions. Coord. Chem. Rev. 277: 208–223. Urakawa, A., Maeda, N., and Baiker, A. (2008). Space- and time-resolved combined DRIFT and Raman spectroscopy: monitoring dynamic surface and bulk processes during NOx storage reduction. Angew. Chem. Int. Ed. 47: 9256–9259. Arnolds, H. and Bonn, M. (2010). Ultrafast surface vibrational dynamics. Surf. Sci. Rep. 65: 45–66. Alarcos, N., Cohen, B., Ziolek, M., and Douhal, A. (2017). Photochemistry and photophysics in silica-based materials: ultrafast and single molecule spectroscopy observation. Chem. Rev. 117: 13639–13720. Xu, J.Y., Tong, X., Yu, P. et al. (2018). Ultrafast dynamics of charge transfer and photochemical reactions in solar energy conversion. Adv. Sci. 5: 1800221. Rossi, G., Pasquini, L., Catone, D. et al. (2018). Charge carrier dynamics and visible light photocatalysis in vanadium-doped TiO2 nanoparticles. Appl. Catal., B 237: 603–612. Layek, A., Van Loon, J., Roeffaers, M.B.J., and Kubarev, A.V. (2019). Correlated super-resolution fluorescence and electron microscopy reveals the catalytically active nanorods within individual H-ZSM-22 zeolite particles. Catal. Sci. Technol. 9: 4645–4650. Roeffaers, M.B.J., Sels, B.F., Uji-i, H. et al. (2006). Spatially resolved observation of crystal-face-dependent catalysis by single turnover counting. Nature 439: 572–575. Urakawa, A., Bürgi, T., and Baiker, A. (2008). Sensitivity enhancement and dynamic behavior analysis by modulation excitation spectroscopy: principle and application in heterogeneous catalysis. Chem. Eng. Sci. 63: 4902–4909. Ledesma, C., Yang, J., Chen, D., and Holmen, A. (2014). Recent approaches in mechanistic and kinetic studies of catalytic reactions using SSITKA technique. ACS Catal. 4: 4527–4547. Jaumot, J., de Juan, A., and Tauler, R. (2015). MCR-ALS GUI 2.0: new features and applications. Chemom. Intell. Lab. Syst. 140: 1–12. Voronov, A., Urakawa, A., van Beek, W. et al. (2014). Multivariate curve resolution applied to in situ X-ray absorption spectroscopy data: an efficient tool for data processing and analysis. Anal. Chim. Acta 840: 20–27. Hendriksen, B.L.M., Bobaru, S.C., and Frenken, J.W.M. (2004). Oscillatory CO oxidation on Pd(100) studied with in situ scanning tunneling microscopy. Surf. Sci. 552: 229–242. Mom, R.V., Onderwaater, W.G., Rost, M.J. et al. (2017). Simultaneous scanning tunneling microscopy and synchrotron X-ray measurements in a gas environment. Ultramicroscopy 182: 233–242. Kumar, N., Wondergem, C.S., Wain, A.J., and Weckhuysen, B.M. (2019). In situ nanoscale investigation of catalytic reactions in the liquid phase using zirconia-protected tip-enhanced Raman spectroscopy probes. J. Phys. Chem. Lett. 10: 1669–1675.
References
35 Kumar, N., Kalirai, S., Wain, A.J., and Weckhuysen, B.M. (2019). Nanoscale
36 37 38
39
40 41
42
43
chemical imaging of a single catalyst particle with tip-enhanced fluorescence microscopy. ChemCatChem 11: 417–423. Wang, X., Huang, S.C., Huang, T.X. et al. (2017). Tip-enhanced Raman spectroscopy for surfaces and interfaces. Chem. Soc. Rev. 46: 4020–4041. Pettinger, B., Ren, B., Picardi, G. et al. (2004). Nanoscale probing of adsorbed species by tip-enhanced Raman spectroscopy. Phys. Rev. Lett. 92: 096101. Kovtunov, K.V., Pokochueva, E.V., Salnikov, O.G. et al. (2018). Hyperpolarized NMR spectroscopy: d-DNP, PHIP, and SABRE techniques. Chem. Asian J. 13: 1857–1871. Zhivonitko, V.V., Svyatova, A.I., Kovtunov, K.V., and Koptyug, I.V. (2018). Recent MRI studies on heterogeneous catalysis. In: Annual Reports on NMR Spectroscopy, vol. 95 (ed. G.A. Webb), 83–145. Lysova, A.A. and Koptyug, I.V. (2010). Magnetic resonance imaging methods for in situ studies in heterogeneous catalysis. Chem. Soc. Rev. 39: 4585–4601. Timoshenko, J. and Frenkel, A.I. (2019). “Inverting” X-ray absorption spectra of catalysts by machine learning in search for activity descriptors. ACS Catal. 9: 10192–10211. Guda, A.A., Guda, S.A., Lomachenko, K.A. et al. (2019). Quantitative structural determination of active sites from in situ and operando XANES spectra: from standard ab initio simulations to chemometric and machine learning approaches. Catal. Today 336: 3–21. Hartman, T., Geitenbeek, R.G., Whiting, G.T., and Weckhuysen, B.M. (2019). Operando monitoring of temperature and active species at the single catalyst particle level. Nat. Catal. 2: 986–996.
223
225
13 Pressure Gaps in Heterogeneous Catalysis Lars Österlund Uppsala University, The Ångström Laboratory, Department of Materials Sciences and Engineering, P.O. Box 35, SE-751 03 Uppsala, Sweden
13.1 Introduction Heterogeneous catalysis plays a key role in our society to make chemicals, fuels, and environmental remediation, with well over 80% of all catalytic processes using heterogeneous catalysts to achieve high conversion and/or selectivity through lowering activation barriers leading to desired products. Catalysts are expected to be instrumental to realize the world’s transition to renewable energy sources and carbon neutral industrial processes, realizing efficient fuel cells, synthetic biofuels, and selective oxidation catalysts. Heterogeneous catalysis is inherently a surface phenomenon whereby reactant molecules interact with the surface atoms of a catalyst to form product molecules. A heterogeneous catalyst works by lowering the activation barrier for a chemical reaction at its surface, thereby increasing the reaction rate to form a desired product. The catalyst can be activated by heating the catalyst (thermal catalysis), applying a voltage across the catalyst by connecting the catalyst with a counter electrode in an electrochemical reactor (electrocatalysis), irradiating the catalyst with energetic photons (photocatalysis), or combinations of them. For efficient use, mass transport of reactants and product molecules to and from the catalyst surface must also be accounted for. Here the so-called pressure gap in catalysis enters; at high pressures and temperatures where sufficient product yield is obtained, the concentration of reactants is necessarily high. High reactant pressure translates into high flux of reactants to the surface, high reactant coverage, and the possibility of a thermodynamic driving force of the reactants (and products) to change the physicochemical properties of the catalyst – beyond what occurs at low coverages and temperatures. In case of photocatalysis or electrocatalysis, electron transport, excited states, and potential gradients must additionally be considered, and the electron structure of bulk and subsurface of the catalyst must therefore also be understood, which in turn may be affected by high-pressure conditions. The holy grail of heterogeneous is to design and make durable catalysts that show both high turnover rate and selectivity toward to desired chemical reaction. Heterogeneous Catalysts: Advanced Design, Characterization and Applications, First Edition. Edited by Wey Yang Teoh, Atsushi Urakawa, Yun Hau Ng, and Patrick Sit. © 2021 WILEY-VCH GmbH. Published 2021 by WILEY-VCH GmbH.
226
13 Pressure Gaps in Heterogeneous Catalysis
To do so, not only is it necessary to unravel the elementary reaction steps that occur at gas–solid or liquid–solid interfaces but also to identify the active sites on the catalyst. These are extremely challenging tasks considering that reaction conditions to achieve appreciable and useful product yield usually imply high temperatures (typically several hundred degrees of Celsius) and high pressures (typically on the order 1–100 bar), employing nanostructured catalysts that have high surface area with ill-defined surface sites, or more precisely, surface sites that are difficult to characterize and that even may change in the course of the reaction. The purpose of this chapter is to present some basic background knowledge that set the stage of the high-pressure or operando heterogeneous catalysis, which typically is the term that is used to denote a catalytic process that occurs at conditions that are relevant for practical applications. As such, operando does not necessarily mean high pressures and temperatures; some catalysts, e.g. supported Au catalysts or TiO2 photocatalysts, work at fairly low temperatures – even at room temperatures and atmospheric pressures. In this chapter we will review attempts to bridge the pressure gap in heterogeneous catalysis by departing from a few case studies that illustrate different possibilities highlighted as “Boxes” embedded in the main text. We will show examples where a pressure gap does not exist per se and kinetics dictates the outcome, a situation that may be studied accurately in idealized vacuum conditions at low catalyst temperatures. In other examples we will demonstrate that the catalyst can undergo dramatic restructuring as a result of reactions that are thermodynamically driven by adsorbed molecules and the catalyst, a result that necessitates sufficient concentration of reactive adsorbate species, which at cannot be realized low pressures, e.g. due to an activation barrier for dissociation of the reactant, and artificial means are required to simulate operando conditions. Oxidation of metal catalysts under operando conditions is one such example; another is dissociation of hydrogen over copper. Further, we will have a deeper look into the physical and chemical properties of the catalysts upon exposure to high pressures, as compared with low pressures, and in particular ultrahigh vacuum (UHV), which traditionally has been the surface science approach. Finally, catalysis on nanoparticles, nanocatalysis, will be discussed and inherent differences of such systems compared with extended (infinite) surfaces [1]. There are a number of good reviews that treat the pressure gap in heterogeneous catalysis, where the reader can find more details of physicochemical properties of catalysts at elevated pressures and methodology developments to study catalysts under near-ambient or operando conditions [2–7] (Figure 13.1).
13.2 High-Pressure Studies of Catalysts At high gas pressures the number of gas molecules colliding with the catalyst surface per unit time is vast. Kinetic gas theory [8, 9] tells us that the number of molecules impinging on surface per unit time and unit area, √or the impinge2 ment rate, is 𝜈 = 1∕4 ⋅ (N∕V ) ⋅ c molecules/(s m ), where c = 8RT∕𝜋M is the arithmetic mean velocity defined by the Maxwell velocity distribution and R is
13.2 High-Pressure Studies of Catalysts
- Particles of active material (nanoclusters) on porous support material
High pressure (p > 1 bar) Sample cleanliness difficult Electron- and ion-based probes obstructed High adsorbate coverages at high T
The pressure and structure gap
Technical catalysts
Planar model catalysts ity ex l p m co g in Single crystals as re
c
In
Nanofabricated particles
Deposited nanoparticles
UHV (p < 10–9 bar) Clean and well-defined samples All surface science techniques available High adsorbate coverages only at low T
Figure 13.1 Illustration of a dispersed technical catalyst consisting of a heterogeneous distribution of nanoparticles on a supporting material on the left-hand side and model catalysts with increasing degree of complexity on the right-hand side. The so-called pressure gap connecting model studies under idealized conditions (typically vacuum or ultrahigh vacuum) spans more than 9 orders of magnitude. (See online version for color figure).
Avogadro’s constant, T is the temperature, and M is the molar mass. For nitrogen c = 471 m/s at 293 K, which is the expected order of magnitude since we know that the speed of sound in air at this temperature is 343 m/s. Using the ideal gas law, N/V = P/kT, where P is the gas pressure and k Boltzmann constant, we obtain the well-known equation for the impingement rate: 2.63 × 1026 P P[mbar] (molecules∕(m2 s)) ≈ √ 𝜈=√ MT 2𝜋mkT
(13.1)
For the lightest element, hydrogen, Eq. (13.1) yields 𝜈 = 1.1 × 1028 m−2 s−1 , and for nitrogen 𝜈 = 2.9 × 1027 m−2 s−1 . Even though the velocity of the molecules is large, they do not travel far before they collide with other gas molecules. Simple geometrical considerations tell us that the mean free path of a molecule, i.e. the distance it travels between each collision, is l= √
kT
(13.2)
2𝜋d2 P
Equation (13.2) yields l = 64 nm for nitrogen at 1 atm and 293 K. In reality we will have a distribution of mean free paths since all molecules do not have the same likelihood of collision. In fact, the number of molecules that do not collide over a distance x follows a Poisson distribution and decreases exponentially, N = N 0 e−x/l . Thus the number of molecules that travel further than those with mean value l predicted by Eq. (13.2) is N 0 /e ≈ 0.37N 0 , and hence the number of molecules having shorter mean free path is 0.63N 0 . A consequence of Eq. (13.1) is that even at very low UHV base pressure, say, 10−10 mbar, a surface with an area of A = 1 cm2 is bombarded by about 3 × 1010 molecules/s (assuming that
227
228
13 Pressure Gaps in Heterogeneous Catalysis
the vacuum contains mainly the air molecules N2 and O2 ). Even when the gas molecule does not engage in chemical interaction with the surface (i.e. covalent bonding or chemisorption), van der Waals forces (polarization forces) still attract the molecules to the surface, and the molecules become physisorbed on the surface. In general molecules minimize their energy by adsorbing on a surface. At atmospheric pressures the surface will be instantly covered by molecules. Adopting a typical value of 1015 adsorption sites per cm2 , it will take about N s /𝜈 ∼ 1 ns to completely covered the surface with a monolayer (ML) of molecules if the sticking coefficient is unity and all molecules that hit the surface become adsorbed. In contrast, under UHV conditions, it will take many hours to cover the surface. This is one of the reasons for conducting model studies under UHV conditions. If one is able to prepare a well-defined model catalyst in UHV, a time window of several hours exists before residual gas molecules present in the UHV reactor interfere with the experiment. The other fundamental reason to conduct model studies of catalyst surfaces in UHV is due to Eq. (13.2). At high pressures experimental probe techniques that utilize particles to interrogate physical and chemical properties of the catalyst are prohibited by scattering with gas molecules, and the situation becomes worse with denser media (high pressures and liquids). Thus there is no simple way to transfer the information carried by a particle that has interacted with the catalyst surface to a detector without loss of information by collisions with molecules in the surrounding dense media. For electrons l ∼ 1 μm in air and at least low vacuum is needed for electrons to escape to a detector without collisions with surrounding media molecules, unless specialized tricks are used. Such tricks can involve methods to differentially pump the detector and bring it sufficiently close to the sample while maintaining a local high pressure over the sample or injecting gas through microcapillary tubes over the sample to minimize the total number of gas molecules entering the reaction chamber while still maintaining a spatially confined high-pressure region. The fact that electron-based techniques are limited by collisions in the reaction media (gas or liquid) is particularly cumbersome since they are well suited to provide detailed physical and chemical information, such as photoelectron spectroscopy and electron microscopy. The past decade has however seen a tremendous improvement of so-called ambient pressure, or more correctly near-ambient pressure, photoelectron spectroscopy [10–13] and electron microscopy [14–17] techniques utilizing variations of the “tricks” above to create local high pressures. Today, it is possible to obtain atomic-scale structure and high-resolution photoelectron emission spectra data of catalysts at pressures on the order ∼1 mbar, and even up to 100 mbar has recently been reported. Another electron-based technique to unravel atomic-scale information of catalysts is scanning tunneling microscopy (STM). In STM the tunneling current between the electrically conducting sample and a conducting tip with atomic dimension is measured. A schematic illustration of an STM instrument consisting of a conducting tip mounted on piezoelectric scanner tubes that moves a tip laterally across a samples is shown in (Figure 13.2) [18]. With STM it is possible to operate at high pressures, readily up to 1 atm and beyond [18, 20]. The success of STM in high-pressure studies lies in the fact that the electron tunneling from the catalyst surface to the scanning tip (or vice versa) occurs
13.3 Adsorption on Solid Surfaces at Low and High Pressures
Figure 13.2 Schematic view of a scanning tunneling microscope. Source: Adapted from Laegsgaard et al. 2001 [18] and Besenbacher et al. 2005 [19]. (See online version for color figure).
± 300 V
at length scales of ångström, which is short enough to be unaffected by ambient media molecular scattering. In this chapter several examples of STM studies performed on single-crystal model catalyst surfaces will be presented. Although STM is a proven technique to achieve atomically resolved images across UHV to 1 atm, it is technically very challenging to conduct such experiments and in particular for many technically relevant catalysts such as nonconducting oxides. These limitations explain why there are relatively few studies employing STM in high-pressure heterogeneous catalysis. Photon-based techniques, on the other hand, are relatively unaffected by high pressures, and there are numerous techniques that have proven their usability for high-pressure and operando studies, and in this chapter we will give examples from Fourier transform infrared (FTIR) spectroscopy studies across different pressure regions from UHV to 1 atm [21–23]. A potential difficulty with infrared (IR) spectroscopy and other spectroscopy techniques is possible interference with ambient media absorption bands that can complicate quantitative analysis. Other technical issues are related to infrared beam and detector stability; mercury–cadmium–telluride (MCT) detectors need to be cooled by liquid nitrogen. Nevertheless, FTIR is fairly easy to integrate with reaction cells to preform operando studies and is suitable to combine with other spectroscopic techniques [24, 25], as exemplified in Figure 13.3.
13.3 Adsorption on Solid Surfaces at Low and High Pressures 13.3.1
Kinetically Restricted Adsorbate Structures
The heat of adsorption of a molecule is much higher than the activation barrier for diffusion on the surface and also much higher than diffusion into the bulk of the solid. As a consequence, adsorbed molecules readily exchange position on the surface, quickly equilibrate, and find the most thermodynamically stable configuration on the surface. The adsorbed layer may therefore be viewed as a 2D phase that is confined from its surrounding gas (or liquid) environment and solid
229
230
13 Pressure Gaps in Heterogeneous Catalysis
Light guide IR Heated tubes MS
Gas Diffusion tube
Sample holder Capillary colon with PID regulator with heater
Liq.
H2O
PID contolled bath
Figure 13.3 A diffuse reflectance Fourier transform infrared spectroscopy (DRIFTS) setup suitable for combined operando DRIFTS, UV–vis, and gas chromatography–mass spectrometry (GC-MS). Source: Adapted from Österlund 2009 [23] and van der Meulen et al. 2007 [25]. (See online version for color figure).
substrate by a high activation energy for desorption and a high activation energy for diffusion into the solid bulk. At high pressures high adsorbate coverages build up and are maintained on the surface by the high impingement rate (Eq. (13.1)). Under UHV conditions it can be necessary to cool down the sample to achieve high surface coverages. The residence time, 𝜏, of an adsorbed molecule on the surface reflects this situation: 𝜏 = 𝜏0 eΔEad ∕kT
(13.3)
where ΔEad is the heat of adsorption and 𝜏 0 is roughly the inverse of the escape frequency (attempts to desorb), typically on the order ∼10−12 seconds. To give some example, for a weakly adsorbing Xe atom on the Ni(100) surface with Ead ∼ 5 kcal/mol [26], we obtain 𝜏 ∼ 5 ns from Eq. (13.3) at T = 300 K, but when the surface is cooled to liquid nitrogen temperature (77 K), the residence time becomes 𝜏 ∼ 244 seconds. Another example, the heat of adsorption for CO on Pt(111) is around 35 kcal/mol [27], which yields 𝜏 ∼ 11 million years at T = 300 K! Thus, the adsorption process can be considered to be irreversible at room temperature. A simple estimate of the distance a CO molecule can diffuse on the √ Pt(111) surface in time 𝜏 can be obtained from the mean square displacement, ⟨x2 ⟩, and the measured diffusion coefficient D = D0 e−Ediff ∕kT , which is connected through the mean jump frequency h = h0 e−Ediff ∕kT , by Eq. (13.4): D = a2 h∕4 where a is the mean jump distance (e.g. from adsorption site to site), or √ √ ⟨x2 ⟩ = 2 D𝜏
(13.4)
(13.5)
Using Ediff = 12.6 kcal/mol and D0 ∼ 1 cm2 /s measured between 320 and 360 K [28], and 𝜏 from Eq. (13.3), as rough estimates of the diffusion parameters across a wider temperature range, it is seen that the mean displacement is
13.3 Adsorption on Solid Surfaces at Low and High Pressures
231
>100 μm for T < 500 K and increases roughly exponentially with decreasing temperature. If some of the inter-site activation energies for hopping is much larger (∼20 kcal/mol), e.g. at corner and edge site of a nanocatalyst particle, the mean displacement would still be on the order ∼10 μm. Thus in this simple model, CO molecules are predicted to access all possible adsorption sites on a Pt nanoparticle by surface diffusion. As shown in Box 13.1, the activation energy for CO diffusion on the Pt(111) surface is sufficiently low to allow long-range-ordered CO structures to form at 170 K, while at 80 K the CO structure becomes kinetically restricted, and only short-range order is observed [29]. In this latter case, as well as in many cases, the so-called pressure gap is not really a pressure gap, but merely a shift of the pressure and temperature window where a specific adsorbate structure is formed depending on kinetic limitations. Studies where one mimics high-pressure exposure situations by exploring low temperature phases, or enables activated adsorption processes by artificial means, are the surface science “static” approach to study high-pressure catalysis. In cases where thermodynamics do not drive the system irreversibly to a new stable phase upon high-pressure exposure, it is usually possible to lower the surface temperature to accommodate the same surface coverage that is achievable at high pressures. It is however worth to point out that even though the CO/Pt(111) belong to one of the most studied systems in surface science since the 1970s, even the basic question on which sites CO preferentially adsorbs has been the subject of intense research for many decades [30, 31]. In light of recent environmental high-resolution transmission electron microscopy (TEM) studies of CO adsorption on Au/CeO2 , where a slight outward Au lattice relaxation was observed (Box 13.1), one may anticipate that further work in this field will further scrutinize possible influence of lattice relaxation on CO adsorption geometries. Box 13.1 CO Adsorption Metal Surfaces CO Adsorption on Pt(111) The CO/Pt system is of most studied system in surface science. Not only is it technically relevant in a range of different applications from in energy technology, chemical synthesis, and environmental technologies, but also it is a prototype system to model and probe active sites of catalysts. The Blyholder model [32] is the generally accepted model for CO bonding to metal surfaces, which, despite its simplicity where the interaction is described by the LUMO 2π* and HOMO 5σ CO molecular orbitals, has been verified and refined by numerous experimental [27, 33–37] and theoretical [30, 31] studies. High-pressure STM has been used to study the CO/Pt(111) system in the whole pressure range from UHV to 1 bar [29, 38]. Although first claimed to exhibit unique high-pressure adsorption structures [38], later studies [29] have established that at a pressure of 1 bar of CO, two rotational domains of a hexagonal CO Moiré pattern with a periodicity of 11.8 ± 0.4 Å rotated Ψ = 24∘ ± 2∘ with respect to the underlying substrate (Figure 13.4). High-resolution STM data showed that (Continued)
13 Pressure Gaps in Heterogeneous Catalysis
Box 13.1 (Continued) 1 bar 300 K Θsat = 0.68
(a)
ωvis
ωIR (tunable)
(b)
ωsf = ωIR + ωvis
2090 300 K
(c)
2097 300 K
ML
(a)
0.5 (4 L) 0.5 (3 L) 0.47 0.45 2086 2081
(b) (A)
(c)
0.38 0.26
2040 2080 2120 2160 Energy (cm–1)
SFG - signal (a.u.)
232
500 mbar 200 mbar 100 mbar 10 mbar 1 mbar 2040 2080 2120 2160 Energy (cm–1)
(B)
Figure 13.4 (A) STM images of compressed CO adlayers on Pt(111) obtained at 1 bar CO pressure at different magnifications (a: 240 × 125 Å2 ; b: 55 × 51 Å2 ) and, (c) corresponding proposed model based on DFT calculations. Source: Vestergaard et al. 2002 [29]. Reproduced with permission of American Physical Society. (B) (a) Schematic drawing of the SFG measurement principle. Source: Shen 1989 [39]. Reproduced with permission from Springer Nature. Sum frequency generation (SFG) spectra of CO/Pt(111) at 300 K: (b) At various CO exposures at UHV yielding the well-known c(4 × 2) saturation structure accommodating a CO coverage 𝜃 = 0.5, and (c) at pressures between 1 and 500 mbar, showing an approximately 40% increase of the 𝜈(C–O) stretching frequency corresponding to on-top CO species. Source: Rupprechter et al. 2001 [40]. Reproduced with permission of American Chemical Society. (See online version for color figure).
√ √ the Moiré pattern arises from a 19 × 19R23.4-13CO unit cell, corresponding to a surface coverage of 0.68 ML (Figure 13.4). Thus a compressed CO structured is formed, yielding 36% higher CO coverage. The same CO Moiré structure is however also formed at low-pressure conditions when the Pt(111) surface was saturated with CO at 170 K. At still lower temperatures (90 K), a kinetically hindered domain structure was observed. Importantly, this STM study, covering 13 decades of pressure at varying substrate temperatures, shows that the very same CO adsorption structure is formed after saturation coverage at low temperatures as at atmospheric pressures at room temperature once the substrate temperature is adjusted to allow for surface diffusion while still kept sufficiently low not to cause CO desorption. These conclusions were supported in a sum frequency generation (SFG) study by Rupprechter et al. [40]. They found only on-top CO species up to 500 mbar but at much higher concentrations; about ∼40% increase of on-top CO and concomitant increase of coverage ∼𝜃 ≈ 0.7 were deduced – in quantitative agreement with the STM results of Vestergaard et al. [29]. Still there exist open questions regarding possible high-coverage induced structural modification of the substrate and whether the same conclusion can be drawn for supported Pt
13.3 Adsorption on Solid Surfaces at Low and High Pressures
nanoparticles as we show in the following example. Metal atom relaxation and small changes in bond distances are intrinsically difficult to measure, and our understanding today mostly relies in large part on analysis of low-energy electron diffraction (LEED) data [41]. CO Adsorption on Au Nanoparticles on CeO2 In a study of CO oxidation on supported Au nanoparticles on CeO2 , high-resolution TEM images under vacuum and reaction conditions revealed a reversible distortions in the two topmost layers of the {100} Au facets in a reaction gas mixture of 1% CO diluted in air at a pressure of 45 Pa. As shown in Figure 13.5, the interatomic Au spacing along the was extended by 25% [17]. The outward relaxation was shown to correlate with the catalytic activity. It was further shown that the distortion of the Au atoms was due to CO molecules adsorbed to the surface (Figure 13.5). The CO-induced Au{100} facets accommodated a significantly higher CO density to compensate for the larger Au—Au bond distanced. These findings may suggest that adsorbate structures formed at high pressures or at low temperatures may be guided to new adsorption sites by an interplay between adsorbate interactions and relaxation of surface metal atoms, and it remains to be proven if this phenomenon is more general and if it in fact also is a factor for the CO/Pt(111) system discussed above that stabilizes the compressed CO structures seen at high coverages. 1 vol% CO/air
Vacuum II I
2 nm 0.25 nm
0.29 nm I
0.20 nm 0.5 nm
0.25 nm 0.25 nm
0.29 nm II 0.20 nm
0.25 nm 0.5 nm
(a)
(b)
Figure 13.5 Au{100}-hex reconstructed surface under catalytic conditions. A gold nanoparticle supported on CeO2 in (a) vacuum and (b) a reaction environment (1 vol% CO in air at 45 Pa) at room temperature. The two {100} facets in the rectangular regions indicated by I and II are shown enlarged in vacuum and in CO in air. The images were recorded using a Cs-corrected environmental TEM. Source: Yoshida et al. 2012 [17]. Reproduced with permission of American Association for the Advancement of Science.
233
234
13 Pressure Gaps in Heterogeneous Catalysis
13.3.2
Thermodynamically Driven Reactions on Solid Surfaces
At high pressures new phases may form that do not exist at lower pressure for thermodynamic reasons, which makes it impossible to extrapolate results obtained at low temperature and vacuum with those conducted at elevated pressures at high temperatures. Similarly, minority reaction pathways that are insignificant at low pressure due to kinetic limitations may become dominant at high pressure. It is, however, still possible to mimic high-pressure reaction conditions in those cases where it is possible to drive the system toward a thermodynamically favored high-pressure phase and maintain the system in this state (for a sufficiently long time) even when the pressure is lowered. Examples of this are activated adsorption processes where typically the adsorption is limited by an energy barrier for dissociative adsorption. These are called dynamically limited adsorption processes. An example is hydrogen adsorption on Cu(110) [42], where dissociative hydrogen adsorption is an activated process and prevents the formation of a reconstructed (1 × 2) H phase to be formed under vacuum conditions. However, at high H2 pressures, this dynamic limitation is overcome simply by a large impingement rate of H2 [42]. Similarly, if hydrogen is predissociated, the same (1 × 2) H structure is formed also at UHV conditions. Other examples include oxidation of metals. Oxygen adsorption on Al(111) is an activated dissociative process, which prevents oxidation of alumina in vacuum conditions but can be accelerated by using elevated kinetic energy of the O2 molecules by molecular beam techniques [43]. In contrast, at elevated pressures, it is well known that a native oxide layer forms on aluminum. Similarly, platinum is a noble metal resistant toward oxidation but which at sufficiently high pressures and temperatures nevertheless becomes oxidized. Upon exposure of Pt(110) toward O2 in UHV, the surface reconstructs to for a (1 × 2) reconstructed phase that eventually limits the maximum O coverage that is accessible upon exposure in vacuum. In contrast, under operando conditions, at high pressures and temperatures, it is well known that Pt oxide forms [44]. Pt oxide formed in such experiments is however stable upon evacuation to vacuum and accessible to scrutiny of surface science techniques. Similarly, O2 may be predissociated in an atomic doser and used to simulate the high-pressure and high-temperature conditions prevalent under operando conditions in a real Pt oxidation catalyst. Doing so, experiments can be conducted in UHV to study the early phases of Pt oxidation [45] (Box 13.2). Box 13.2 CO Oxidation on Pt(110) The CO oxidation reaction is arguably the most well-studied reaction in the surface science of heterogeneous catalysis. On Pt surfaces O2 dissociates and reacts with CO and yields CO2 , which is desorbed into the gas phase. The reaction is a classic example of the principle of moderate bonding of the reactants on the catalyst to activate them for further reaction (O2 dissociation, CO + O reaction, and CO2 desorption); only if all elementary steps proceed the catalytic cycle can be repeated. At low temperature the reaction is adsorption limited on the low index Pt surfaces due to so-called CO poisoning (the adsorption sites are fully covered
13.3 Adsorption on Solid Surfaces at Low and High Pressures
235
by CO), whereby the CO + O reaction to yield CO2 is rate limiting. The chemical state of Pt(110) under operando CO oxidation conditions has been studied at the atomic level using STM [20, 45], where the catalytic activity of the Pt surface atoms under oxidation conditions has been reported to change along with the appearance of new surface structures. Using STM and temperature-programed desorption, Li et al. [45] study the Pt(110) surface under strongly oxidizing conditions involving either high-pressure O2 or atomic oxygen exposure (Figure 13.6). At low temperatures, only disordered Pt oxide structures are observed. After annealing ordered surface oxide islands are observed to coexist with a highly stable reconstructed (12 × 2)-O chemisorption structure. From density functional theory (DFT) calculations, a consistent model for the surface oxide phase is proposed. The phase is found to be metastable, and its presence is explained in terms of stabilizing defects in the chemisorption layer and reduced Pt mobility.
Mass spectrometer m/Z = 32 signal (A)
2 ML O – Pt(110)-(1 × 2) (11 × 2) superstructure
2.0 × 10–9
60 min atomic O @ 500 K Oxide II: PtO2
Oxide I: PtO1
1.5 1.0 0.5 0.0 300
400
500
600 700 Temperature (K)
800
900
1000
Figure 13.6 Temperature-programmed desorption spectra of O2 from a Pt(110) single crystal exposed to atomic oxygen at 500 K or at high pressures O2 . The red and blue curves show spectra after 60 minutes atomic O exposure at a substrate temperature of 500 K. The green curves show the result from Gauss–Lorentzian deconvolution of the blue spectrum. Source: Li et al. 2004 [45]. Reproduced with permission of American Physical Society. (See online version for color figure).
More recently, combined ambient-pressure X-ray photoelectron spectroscopy (XPS), high-pressure scanning tunneling microscopy, and DFT calculations have shown that at low oxygen pressure, only chemisorbed oxygen is observed on the Pt(110) surface [46]. At higher pressure (PO2 = 0.5 Torr), nanometer-sized islands of multilayer α-PtO2 -like surface oxide form along with chemisorbed oxygen, in good agreement with the results on Li et al. [45]. Both chemisorbed oxygen and the surface oxide are removed in the presence of CO, similar to what was shown in the STM study by Hendriksen and Frenken [20] (Figure 13.7). The rate of disappearance of the surface oxide was found to be close to that of the chemisorbed oxygen at 270 K. Interestingly, the spectroscopic features of the (Continued)
13 Pressure Gaps in Heterogeneous Catalysis
Box 13.2 (Continued) 1 Partial pressure (bar)
236
O2
10–1 10–2 Rth
G
10–3
H
D C
B
10–4
CO
Bhigh
Rlow
A
CO2
F E
1000
0 A
2000
3000 4000 Time (s)
5000
6000
B
C
D
F
G
H
7000
0] [11 E
Figure 13.7 A series of high-pressure STM images acquired during CO oxidation in alternatively CO-rich and O2-rich flow conditions at T = 425 K, where the Pt(110) surface is reversibly oxidized and reduced [20]. The partial pressures of O2 and CO are such that CO is in excess until region D is reached, after which the surface becomes accessible for oxidation by high pressures of O2 , and the CO2 formation rapidly increases. This is a result of the bistable nature of the CO oxidation reaction over Pt that agrees quantitatively with previous reports [47]. The surface gradually roughens as dissociated O atoms interact with the open Pt surface similar to what is shown in Figure 13.6. A gradual decrease of the reactivity is seen in region D as the surface oxide develops. Source: Hendriksen and Frenken 2002 [20]. Reproduced with permission of American Institute of Physics.
surface oxide were reported to be similar to the oxide observed on Pt nanoparticles of a similar size, highlighting that the Pt structures formed under oxidizing conditions and their reactivity toward CO can provide insights into catalytic processes not only on open Pt surfaces but also for Pt nanoparticle catalysts. A conclusion from these studies is that the surface oxide formed on Pt crystal surfaces and on Pt nanoparticles is catalytically active for CO oxidation. Note, however, that this conclusion does not imply that the fully oxidized Pt surface is catalytically active. Similar conclusions were arrived at in the recent review by Park [6]. Finally, we comment that platinum oxide is a well-known catalyst in hydrogenation catalysis for organic synthesis (primarily ketones to alcohols or ethers) [48] that usually is activated in its reduced form to form platinum black, which is a porous Pt metal that typically is made from Adams’ catalyst (PtO2 ) by reaction with hydrogen.
13.3 Adsorption on Solid Surfaces at Low and High Pressures
237
When an adsorbate molecule reacts with the catalyst, the products can either bind to the surface or desorb into the gas phase. In most cases it is desirable that the reaction products should bind weaker than the reactant molecules and desorb into the gas phase (or liquid) to free a surface sites for next reactant molecules to adsorb and react. At high reactant pressure the reactant coverage is nevertheless maintained high to ensure that turnover number of the catalytic process is high. This is a true nonequilibrium process whereby the dynamical surface processes at high pressure and high adsorbate coverage govern the reactivity. The bistability of the CO oxidation over Pt catalysts is an example where self-organization will result in a spatiotemporal organization of catalytic processes [49]. More generally, high-pressure conditions can favor formation of surface structures that are stabilized by nonequilibrium conditions, such as strain or structural defects, which under operando conditions will be the active surface structures and which are different from ideal structures stable at static UHV conditions. In Box 13.2 high-pressure STM and mass spectrometry studies of CO oxidation over Pt(110) model catalysts show how Pt oxide is formed under oxygen-rich conditions and that under CO excess the surface is reduced and blocked (CO poisoned). The situation is different from UHV model studies where high coverage can only be achieved by lowering the surface temperature, unless the adsorbate binds strongly to the surface. As a consequence, at high pressure, adsorbate phases may exist that find no analogue under UHV conditions. The different adsorbate coverage that can be maintained at high reactant pressure may have profound influence of the catalytic activity. It is well known that adsorbates can change the surface structure and induce structural reconstructions, and new surface phases such as oxides, carbides, and sulfides form at operando conditions. Although our understanding of adsorption and reaction on metal surfaces ranging from vacuum to high pressures has advanced remarkably the past 30 years, allowing us to reach atomic-scale insights and devise predictive theoretical models, corresponding studies on oxide surfaces that make up the majority of the Earth’s crust are still in its infancy [13, 50, 51]. The irreversible phase transformations, compound formation, and dissolution processes that take place on mineral surface are disguised by the diversity of active sites and complex reaction milieu. Nevertheless, the sedulous surface science work that has been conducted on metal oxides in vacuum [52, 53] represents the necessary building blocks that pave the way for the fundamental insights. In Box 13.3 CO and HCOOH adsorptions on TiO2 surfaces are presented, which serve as model systems for corresponding reactions on other metal oxides. Box 13.3 Adsorption on Metal Oxides CO Adsorption on TiO2 (110) Titanium dioxide (TiO2 ) is wide band gap semiconductor that stands out as the prototype catalyst in photocatalysis [54]. Mixed binary oxides have been used extensively as selective NO reduction catalysts (Ce/TiO2 , TiO2 /V2 O5 ), and the (Continued)
238
13 Pressure Gaps in Heterogeneous Catalysis
Box 13.3 (Continued) sulfated TiO2 is reported to be a superacid-like catalyst that also exhibits good NO reduction properties [55]. Furthermore, metal nanoparticles supported on TiO2 are important oxidation catalysts, notably the low-temperature Au/TiO2 and the Pt/TiO2 catalysts. The CO interaction with supported Au nanoparticles was discussed in Box 13.1 and highlights the CO metal bonding. In the CO oxidation reaction on Au/TiO2 , the TiO2 support plays the dual role of providing CO through surface diffusion and to participate in O2 binding at the Au perimeter [56]. The CO (photo)oxidation reaction on the rutile TiO2 (110) surfaces has attracted much attention and is analogous to CO oxidation on metals considered to be a benchmark model reaction in heterogeneous oxidation catalysis [57–59]. Numerous experimental and theoretical studies have been devoted to CO adsorption on rutile TiO2 (110) [60–65]. Here, the combination of infrared spectroscopy and DFT calculations stands out as particularly powerful tools to unravel adsorbate structures. Compared with the corresponding data available for CO adsorption on metal surfaces, much less data is however still available on TiO2 surfaces and still less on other oxides. Nevertheless, we have learned from adsorption studies on metal single crystals that for reversible adsorption and reaction, low-temperature studies can be used to mimic high-coverage phases formed at high pressures and high temperatures in case where activation barriers for adsorption can be overcome (also by artificial means, e.g. by predissociation of diatomic molecules such as H2 and O2 as exemplified in Box 13.2). In this example, we review the CO adsorption on the rutile TiO2 (110) surfaces from low to high coverage with the aim to give a foundation for possible structures that may be observed at high pressures [65]. The CO molecule adsorbs weakly via the C atom on fivefold coordinated Ti ions (Ti5c ; see Figure 13.8) on the stoichiometric rutile TiO2 (110) surface when the coverage is 1 ML or lower [61]. At 1 ML coverage, a (2 × 1) adlayer structure with rows of CO molecules running along the [001] direction has been proposed [63]. The crystal directions are also shown in Figure 13.8. As for CO adsorption on metals surfaces (see Box 13.1), it is crucial that the high-coverage CO adlayers can accommodate the lateral CO–CO repulsion energies. The C∞v symmetry axes of the CO molecules were reported to be tilted by about 20∘ away from the surface normal in a zigzag pattern [62, 63]. In a recent infrared reflection-absorption spectroscopy (IRRAS) study by Petrik and Kimmel [64], it was however proposed that the CO molecules along the [001] rows were not tilted. A new 𝜈(CO) IR peak close to the gas-phase CO value was observed at lower temperatures for CO coverages exceeding 1 ML. The peak was attributed to loosely bound CO molecules residing above the ridge of two-coordinated bridging oxygen atoms (O2c ) and attributed to additional CO molecules above 1 ML coverage oriented parallel to the surface plane and along the [110] direction, i.e. perpendicular to the O2c ridges. Their proposed structures of the ML and the additional half-layer of the 1.5 ML coverage system and their corresponding measured spectra are shown in Figure 13.8b. New insight was recently reported by Hu et al. [65]. In this work it was found that a tilted CO adsorption geometry for CO molecules bonded to Ti5c sites at 1 ML coverage is indeed consistent with the reported IRRAS data by Petrik and Kimmel [64] Furthermore, an adsorption structure for 1.5 ML
13.3 Adsorption on Solid Surfaces at Low and High Pressures
239
CO is predicted to consist of weakly bound CO molecules on top of the ML. We expect that future high-pressure studies both can use and elaborate on the results presented in this section regarding the high-coverage CO structures.
Figure 13.8 Proposed high-coverage 1.5 ML CO/TiO2 structure generates PBE-simulated sand p-polarized IRRAS spectra [65] (a) that agree very well with the published experimental spectra of Petrik and Kimmel [64] (b). We note that the agreement between the simulated and the experimental IRRAS spectra is remarkably good although the absolute values of the CO vibrational frequencies – and thus the IRRAS peaks/dips – are slightly shifted with respect to experiment. Source (a) and (b): Petrik and Kimmel 2012 [64]. Reproduced with permission of American Chemical Society. (See online version for color figure).
HCOOH Adsorption on TiO2 As a final example of adsorption on metal oxide surfaces, the formic acid adsorption system is presented. Formic acid adsorption is not only a model system to understand carboxylate coordination on metal oxide surfaces, but formic acid is also a key molecule in a vast number of applications ranging from fine chemical synthesis, intermediate in hydrogenation reactions to make synthetic fuels (e.g. (Continued)
240
13 Pressure Gaps in Heterogeneous Catalysis
Box 13.3 (Continued) from CO2 ), oxidation product of hydrocarbons, and a linker group in inorganic chemistry to bind organic molecules to metal cations. Formic acid is one of the most abundant acids in the atmosphere, with important implications on precipitation chemistry and acidity, including oxidant for other important atmospheric gases such as SO2 . In urban regions ppb levels of HCOOH are observed, i.e. at partial pressures on the order 10−6 mbar, which is intimately connected to human activity [66]. Formic acid is known to dissociate on the rutile TiO2 (110) [67–69] and TiO2 (101) [70] surfaces to yield bidentate coordinated formate species and one H atom, where each O atom in the formate molecule is bonded to one fivefold coordinated surface Ti5c atom each, and the H atom protonates low-coordinated surface O atoms, notably the twofold coordinated bridging O2c atoms as depicted in Figure 13.9 [68, 69, 71, 72]. (110) surface B A
Light direction [001] Light direction
D
[110]
Formate O Bridging O Ti
C
O
H
Figure 13.9 Schematic picture of possible formate (HCOO− ) structures on TiO2 (110) surface denoted species A, B, C, and D. The B and C species bind to an oxygen vacancy, but A and D do not. The crystal orientation and the directions of electromagnetic vectors as defined in the text are shown. The orientation of the incident IR beam with respect to the x-axis and projected onto the xy-plane is denoted by azimuthal angle 𝜙, where 𝜙 = 0∘ is defined to be in the [110] direction and 𝜙 = 90∘ in the [001] direction. (See online version for color figure).
Type A is the thermodynamically preferred structure. If an O vacancy exists, then type B is preferred. However, as is usually the case for metal oxide surfaces, even trace amount of water modifies the surface chemistry. Even under UHV conditions ( E0 + 100 eV
Can be c. ±50 eV across specific non-monochromatic emission line
c. 0.1–5 eV (depends on element of interest)
Density of occupied states, valence state, spin state, nature of neighbors
vtc-XES
As previous
c. ±50 eV across valence-to-core transition appearing when Eins > E0
As previous
Density of occupied states, nature of neighbors, local geometry
As previous
3. Resonant methods using high-resolution fluorescence detection RXES or RIXS
c. ±50 eV across E0
c. 0.2–2 eV (depends on Eins and monochromator)
c. ±50 eV across specific emission line appearing when Eins > E0
c. 0.1–2 eV (depends on element of interest)
Density of occupied and unoccupied states under resonant conditions
HERFD XANES or PFY XANES
c. ±50 eV across E0
As previous
Fixed at maximum of specific emission line appearing when Eins > E0
As previous
As XANES but with better resolution
HEROS
Fixed energy c. 10–20 eV below E0
c. 0.2–6 eV (depends on Eins and monochromator)
c. ±50 eV across specific emission line appearing when Eins > E0
As previous
As HERFD XANES
19 Operando X-Ray Spectroscopies on Catalysts in Action
Kapton, 25 µm 1.0
Transmission (a.u.)
342
Be, 200 µm
0.8 0.6 Air, 10 cm 0.4 Quartz, 10 µm 0.2 Water, 1 mm 0.0 2000
4000
6000 Energy (eV)
8000
10 000
Figure 19.2 Transmission of X-rays through different materials as a function of X-ray energy.
XAS and XES are generally not surface sensitive. The penetration depth of hard X-rays in a material ranges from microns to centimeters depending on the material density, X-ray energy, and concentration of the element of interest. The lack of surface sensitivity of XAS and XES is less critical for catalytic materials consisting of finely dispersed atoms, clusters, or nanoparticles of the active phase on high surface area supports, because the high fraction of active atoms on the surface of these materials facilitate their chemical speciation.
19.2 X-Ray Absorption Spectroscopy Methods XAS spectra in the hard X-ray region are typically measured in transmission or fluorescence modes. The transmission geometry is ideal for concentrated samples; however the sample thickness needs to be adjusted to have an optimal ratio between the incident (I 0 ) and the transmitted X-ray intensity (I 1 ). The transmission XAS signal is expressed as 𝜇(Eins ) = ln(I0 ∕I1 ) where 𝜇(Eins ) is the absorption coefficient. Practically, 𝜇 should be c. 0.1–2 to measure a good quality XAS signal in transmission mode. I 0 and I 1 signals are commonly measured by ionization chambers filled with inert gases placed between two electrodes, with one at high voltage. When X-rays pass through such a chamber, they ionize gases, and an electron current proportional to the X-ray intensity is registered. When the concentration of the element of interest in the sample is low (typically below 0.5 wt%) or samples contain a large concentration of other heavy elements additionally absorbing X-rays, only fluorescence detection is possible. Fluorescence (or emission) X-rays are produced by an absorbing atom after an X-ray absorption event while refilling
19.2 X-Ray Absorption Spectroscopy Methods
the electron hole (Figure 19.1a). The intensity of fluorescence photons is lower in comparison to the intensity of incident X-rays, and they are emitted in all directions. When an electron is excited above its K-edge, the most intense XES lines are Kα1 and Kα2 . For acquiring fluorescence-detected XAS, the integral intensity of these photons can be measured by a solid-state detector (e.g. a silicon drift detector [SDD]), which is capable to discriminate emitted X-rays with >150 eV resolution or with an X-ray photodiode without energy resolution. Good energy resolution of the detector helps in reducing the background signals originating from elastic scattering of incident X-rays (having higher energies than the emitted X-rays) and the emission lines of other elements present in the sample (having absorption edges below the incident X-ray energy). Placing the detector close to the sample helps to increase the signal-to-noise ratio; however, a solid-state detector should not be saturated. The intensity of unwanted elastic X-rays (produced by scattering of the incident beam by air or the sample) can be reduced by placing a special filter in front of the detector. For K-edges, this filter typically contains an element with the atomic number just before the element of interest, a so-called Z-1 filter. Due to the specific edge energy position of the element in the Z-1 filter, the transmission for the fluorescence signal of the element of interest (with the atomic number Z) is much stronger than for the incident X-rays. The fluorescence XAS signal (I fluo ) is divided by the incident beam intensity (I 0 ) and defined as 𝜇(E) = Ifluo ∕I0 For very diluted samples, the shape of fluorescence and transmission XAS is identical. For more concentrated samples however, the amplitude of fluorescence XAS features is damped and needs to be corrected to be used for precise chemical speciation. This dampening is due to the reabsorption of fluorescence photons emitted by the element of interest by the neighboring atoms of the same element, so-called self-absorption or over-absorption [5, 6]. Table 19.1 summarizes the main XAS methods. Figure 19.3a shows an example of the Cr K-edge XAS of Na2 CrO4 . The XAS spectrum around the absorption edge (E0 ± 50 eV) is called the X-ray absorption near edge structure (XANES) or the near edge X-ray absorption fine structure (NEXAFS), which is the term used by the soft X-ray community (e.g. referring to O K-edge NEXAFS). The K-edge XANES reflects the density of unoccupied p states (p-DOS [density of states]), while the L3 -edge XANES probes the unoccupied d- and s-states. XANES spectra contain chemical information about the electronic as well as the geometric structure of the absorbing atom. Moreover, L2,3 XANES is strongly affected by multi-electron interactions. Figure 19.3b compares the Cr K-edge XANES spectra of Cr2 O3 and Na2 CrO4 . Cr3+ ions (which are essential trace nutrients) in Cr2 O3 have sixfold oxygen coordination in octahedral symmetry, while Cr6+ ions (which are toxic) in Na2 CrO4 are surrounded by four oxygen atoms in tetrahedral symmetry (Figure 19.3c). The position of the X-ray absorption edge (E0 ) is typically ascribed to the maximum of the first derivative of the rising edge. The E0 typically (but not always) shifts to higher energies when the valence state of an element increases. The cause of this shift is the removal of electrons from outer shells, leading to the core shell electrons held slightly
343
19 Operando X-Ray Spectroscopies on Catalysts in Action
Pre-edge
Positive interference
1.2 1.0 0.8
Neighboring atom Absorbing atom
0.6
Na2CrO4 Edge
5960
0.4
XANES
6000
6000
(b)
Negative interference
0.2 0.0
Cr2O3
Normalized absorption
1.4
6040 6080 Energy (eV)
EXAFS
6200
(a) Na2CrO4
6400 6600 Energy (eV) Cr2O3
6800
20
|χ (R)|
Normalized absorption (a.u.)
344
Na2CrO4 10 Cr2O3 0 0
(c)
(d)
1
2 R (Å)
3
4
Figure 19.3 (a) Normalized Cr K-edge XAS of Na2 CrO4 indicating XANES and EXAFS regions. (b) Cr K-edge XANES spectra of Cr2 O3 and Na2 CrO4 . (c) Geometric structure of Cr6+ in Na2 CrO4 and Cr3+ in Cr2 O3 . (d) Fourier transform magnitude Cr K-edge EXAFS spectra of Na2 CrO4 and Cr2 O3 (solid lines) together with the corresponding fit (dotted lines).
tighter. Figure 19.3b illustrates this effect for the Cr K-edge. The edge position also depends on the nature of ligands since they also affect the outer electrons; therefore, the valence state and the edge position are not always proportional. The peaks appearing before the absorption edge are called pre-edges. These distinct absorption features reflect electron transitions from the core levels to the lowest unoccupied states. Commonly the most intense pre-edges appearing below K-edges originate from the orbital mixing between p and d electron shells. These pre-edges are more intense for non-centrosymmetric structures, for example, having tetrahedral coordination (like Cr6+ in Na2 CrO4 in Figure 19.3d), and are weaker for centrosymmetric structures having, for example, octahedral coordination (like Cr3+ in Cr2 O3 in Figure 19.3d). To compare XANES spectra, spectra are typically background corrected and normalized by setting the edge jump to 1. After such normalization, at c. −100 eV below E0 , the XANES signal should be close to 0, and above the edge (>100 eV), it should oscillate around 1. This normalization is typically done by a subtraction of a linear background (fitted below the edge) and dividing the resulting spectrum by a normalization function (similar to a step function), generated by fitting a polynomial function through the post-edge spectrum. As the shape of XANES directly reflects the average local coordination of the absorbing atom in the sample, the simplest way of XANES analysis is visual comparison with the spectra of the reference samples (also called fingerprinting) or their linear combination fitting. Figure 19.4 shows an example of linear combination fitting of time-resolved Cu K-edge XANES collected from a Cu-SSZ-13 zeolite-based catalyst under conditions of
19.2 X-Ray Absorption Spectroscopy Methods
Normalized absorption (a.u.)
1.2
(a)
Time = 0 s
(b)
Time = 75 s (c)
1.0
Time = 940 s
Data Fit
0.8
CuI(NH3)2
0.6 CuII-Z
0.4 CuII(NH3)2
0.2
CuII(NOx)y
0.0 8980
9000
9020
8980
9000
9020
8980
9000
9020
Energy (eV) Figure 19.4 Cu K-edge XANES of Cu-SSZ-13 catalyst at different delay time after switching off NH3 flow (0 s (a), 75 s (b), and 940 s (c)), the resulting linear combination fits, the weighted spectra of the reference samples, and the corresponding residuals. Source: Marberger et al. 2018 [7]. Adapted with permission of Springer Nature. (See online version for color figure).
selective catalytic reduction (SCR) of NOx by NH3 [7] taking place in a diesel car exhaust. The reference samples mimicking the state of copper sites are aqueous solutions of Cu(I) and Cu(II) amino complexes, CuII (NOx )y structure standing for CuII (NO2 )x (H2 O)6 or CuII (NO3 )2 (showing very similar spectra), and ionic Cu(I)/Cu(II) in a SSZ-13 network produced by pretreatment of Cu-SSZ-13 at 400 ∘ C in an O2 /N2 mixture. In this example the reference samples allow precise chemical speciation of copper, which is the ideal case, because references satisfactory mimicking the structure of active sites in catalysts under operando conditions are often not available. In that case, analytical procedures including principal component or multivariate analysis can be helpful to guess the number of components and their XANES spectra from a large dataset [6, 8, 9]. Further analysis may also involve theoretical simulations of XANES spectra of nonexisting references proposed by structural modeling. Extended X-ray absorption fine structure (EXAFS) spectroscopy refers to the part of the XAS spectrum starting from c. 100 eV above the absorption edge and extended up to c. 1000 eV above the absorption edge (Figure 19.3a). EXAFS originates from the scattering of the photoelectron wave generated when the excited photoelectron is ejected from the atom. This photoelectron wave scatters on the nearest neighbors returning to the original atom. This produces constructive and destructive interference, visible as oscillations of the absorption coefficient. To extract chemical information from an EXAFS spectrum, one needs to fit the oscillation part using the EXAFS theory [2, 5, 6, 10]. Before the analysis, a background approximating an EXAFS spectrum of the same atom without any neighbors is subtracted from the data, leaving only the oscillating part of absorption spectrum 𝜒(Eins ). Then 𝜒(Eins ) is converted into k-space (𝜒(k)), where k is the wavenumber related to the photon energy by the following equation: 1√ 2me (Eins − E0 ) k= ℏ
345
346
19 Operando X-Ray Spectroscopies on Catalysts in Action
where ℏ is Plank’s constant divided by 2𝜋, me is the mass of an electron, Eins is the incident energy, and E0 is the edge energy. The EXAFS equation is the following: ∑ fi (k) − 2Ri 2 2 Ni 2 e 𝜆(k) e−2k 𝜎i sin(2kRi + 𝛿i (k)) 𝜒(k) = S02 kR i i The sum corresponds to all (i) electron scattering paths involved in the fitting model. The scattering path is single or direct when the photoelectronic wave scatters elastically from the neighboring atom and returns. It is multiple when the photoelectronic wave scatters consequently from several neighbors before returning. S02 is the amplitude reduction factor, which is an element-specific constant typically varying between 0.6 and 1; N i is the number of equivalent scattering paths (corresponding to the coordination number for single scattering paths); Ri is the interatomic distance for the ith scattering path; 𝜎i2 is the mean square disorder; f i (k) is the photoelectron (back)scattering amplitude; 𝛿 i (k) is the (back)scattering phase. EXAFS signal in k-space 𝜒(k) can be also Fourier transformed (FT) and presented in R-space as 𝜒(R). Before FT the 𝜒(k) data can be multiplied (weighted) by k, k 2 , or k 3 to make the amplitude of oscillations similar for the whole k region. Due to the phase shift, the FT EXAFS spectra are not equivalent to the radial distribution function. The peaks in EXAFS typically appear at shorter distances in comparison to the corresponding chemical bonds, and a single chemical bond can also generate multiple peaks in EXAFS. More details on the EXAFS equation and different fitting strategies can be found in dedicated books [2, 5, 6]. Figure 19.3d shows the magnitude of the FT Cr K-edge EXAFS spectra of Cr2 O3 and Na2 CrO4 and the corresponding fits. Due to the limited mean free path of photoelectrons, EXAFS spectroscopy provides information only about the nature of the closest neighbors. Table 19.2 presents the results of the Cr K-edge EXAFS fits for Na2 CrO4 and Cr2 O3 . In these fits using only the single scattering paths, the coordination numbers (N i ) were fixed to the values known from X-ray diffraction (XRD). Thus, the varied structural parameters were only the coordination distances (Ri ) changing from the initial values originating from the XRD model (Reff ) Table 19.2 Results of Cr K-edge EXAFS fits for Na2 CrO4 and Cr2 O3 .
Compound
Starting Number of Coordination coordination Debye–Waller Energy distance (Reff ) factor shift Coordination neighbors distance (N) (R) (Å) (𝚫E) (eV) path (Å) (𝝈 2 ) (Å2 )
Na2 CrO4
Cr–O
4a)
1.65(1)
1.645 3
0.003 2(7)
Cr2 O3
Cr–O
6a)
1.989(5)
1.959 4
0.003 8(4)
2.5 ± 0.9
Cr–Cr
1a)
2.65(1)
2.670 6
0.003 9(4)
2.5 ± 0.9
Cr–Cr
3a)
2.907(6)
2.890 4
0.003 9(4)
2.5 ± 0.9
Cr–Cr
3a)
3.430(8)
3.413 5
0.004 30(4)
2.5 ± 0.9
Cr–Cr
6a)
3.668(6)
3.650 1
0.004 3(4)
2.5 ± 0.9
The error bars are given in brackets. a) These values were not varied in the fit.
1.5 ± 3.5
19.3 High-Energy-Resolution (Resonant) X-Ray Emission Spectroscopy
Pt L3 edge
0.0 (a)
PtRu alloy Ru-core/Pt-shell
0.5
0
2
4 r (Å)
6
FT magnitude (Å–3)
FT magnitude (Å–3)
1.0 2
Ru K edge
1
PtRu alloy Ru-core/Pt-shell
0
8
0 (b)
2
4 r (Å)
6
8
Figure 19.5 Fourier transform magnitude of the k2 -weighted EXAFS spectra for the Pt L3 - (a) and Ru K- (b) edges in the alloy and core–shell bimetallic nanocatalysts supported on 𝛾-Al2 O3 . Source: Frenkel 2012 [13]. Adapted with permission of Royal Society of Chemistry. (See online version for color figure).
and the Debye–Waller factors (𝜎i2 ) corresponding to the local disorder. S02 was fixed at 0.72 after fitting the EXAFS spectrum of Cr foil, which is a common procedure because the structure of an element in the metal state is much better defined in comparison with any other state. ΔEi corresponds to the E0 shift. When the local coordination of the element of interest is not known, the coordination numbers (N i ) can also be fitted. Fitting the coordination numbers in metal nanoparticles gives an indication about their average size and shape, especially if their diameter is below 3 nm [11, 12]. Atoms at the surface of metal nanoparticles have a lower number of nearest neighbors in comparison with those in the nanoparticle core. Since EXAFS measures the average number of neighbors in all atoms forming the nanoparticles, this number is smaller for smaller and nonspherical nanoparticles having a larger fraction of atoms on the surface. By comparing the number of specific neighbors in bimetallic nanoparticles, EXAFS can help to uncover whether metals form a simple alloy or a more complex structure [12, 13]. Figure 19.5 illustrates the sensitivity of EXAFS to the local coordination for supported Pt–Ru bimetallic nanoparticles. It shows that the alloyed structure can be distinguished from a core shell. Determining by EXAFS the coordination numbers in supported metal oxides can also provide information about their local structure and crystallinity. However, one should keep in mind that EXAFS fits typically contain large error bars, especially for the number of neighbors. This is due to structural disorder described by the pseudo-Debye–Waller factors strongly correlating in the fit with the number of neighbors. More details about each fitting parameter and fitting strategies can be found elsewhere [2, 5, 6].
19.3 High-Energy-Resolution (Resonant) X-Ray Emission Spectroscopy Table 19.1 summarizes the details of the main XES methods. All these methods are based on the detection of emitted (fluorescence) photons. While for fluorescence XAS we detect the integral intensity of emitted X-rays (Eem ) with >150 eV
347
19 Operando X-Ray Spectroscopies on Catalysts in Action
n sio er
Cylindrically bent crystal
sp
Di
Dispersive plane
ΔE
E2
⊝2
Position-sensitive detector
Focusing
348
E1
Cylindrically bent crystal
ΔE Detector
⊝1 R
X-rays in (a)
Sample
Sample
(b)
Figure 19.6 X-ray emission spectrometer in von Hamos geometry (a) and application of the Bragg’s law for energy discrimination (b). Source: Szlachetko et al. 2012 [15]. Adapted with permission of American Institute of Physics.
energy resolution, for XES and resonant spectroscopy, we detect the emitted X-rays with a resolution comparable to the lifetime broadening of the initial and final core holes of the absorbing element [14]. To measure Kα lines of 3d elements and L𝛼 lines of 5d elements, the required energy resolution of the emission spectrometer should be around 0.1–5 eV depending on the studied transition. XES and resonant methods are chemically sensitive only if they are measured with enhanced energy resolution. As solid-state detectors cannot provide sufficient energy resolution, one needs to use an emission spectrometer that performs energy discrimination based on Bragg’s law. Such spectrometers are typically equipped with single-crystal analyzers made of oriented Si, Ge, or SiO2 wafers. Figure 19.6a shows an example of such a spectrometer in the von Hamos geometry [15]. In this spectrometer, the fluorescence photons emitted from the sample are discriminated by energy in the horizontal direction (in the same way as light is dispersed by a quartz prism) and focused in the vertical direction on a position-sensitive detector. Figure 19.6b explains the application of Bragg’s law for the emission energy (Eem ) discrimination. Bragg’s law describes the condition at which constructive interference between reflected waves is obtained at angle Θ: 2d sin Θ = n𝜆
(19.1)
where d is the distance between crystal planes (d-spacing) of the analyzer crystal. The X-ray energy is inversely proportional to the wavelength (𝜆); therefore fluorescence X-ray photons with lower energy (E1 ) diffract from the analyzer crystal at higher angle (Θ1 ). As a side note, Bragg’s law is also used to select the incident X-ray energy in a monochromator, for the hard X-ray regime, typically Si(111) or Si(311) single crystals are used. Figure 19.7 shows examples of non-resonant Mn, Kα, and Kβ XES spectra. When a Mn-containing sample is excited with the incident X-ray beam >100 eV above the Mn K-edge (E0 = 6539 eV), the majority of emitted X-rays correspond to Kα and Kβ XES lines (Figure 19.7). These lines are also called core-to-core X-ray emission spectroscopy (ctc-XES) because of electron transfer from core to core shells. The chemical sensitivity of Kβ emission lines is stronger than for Kα lines, because 2p electrons interact weaker with valence electrons than 3p
19.3 High-Energy-Resolution (Resonant) X-Ray Emission Spectroscopy
Kβ main lines
Kβ satellite lines
Intensity (a.u.)
Kβ1,3
Kβ2,5 Kβ″
Kα1
×500
Kα2
KLβ Kβ′ Kβ
5880
5900
×8
5920 6480 6520 Fluorescence energy (eV)
6560
Figure 19.7 ctc-XES (Kα and Kβ main lines) and vtc-XES (Kβ satellite lines) of Mn in MnO resulting from filling of a 1s core hole. Source: Glatzel and Bergmann 2005 [14]. Adapted with permission of Elsevier.
electrons. For example, the intensity of the so-called Kβ′ line, i.e. the shoulder of the Kβ line of 3d metals, is proportional to the number of unpaired electrons (spin state) of the 3d element under study (Figure 19.7). This is due to the exchange interaction between the 3d and 2p orbitals resulting in the appearance of a Kβ′ feature. Weak valence-to-core X-ray emission spectroscopy (vtc-XES) features (Figure 19.7), originating from valence to core transitions, are the most chemically sensitive non-resonant XES lines [14, 16, 17]. Figure 19.8 demonstrates the chemical sensitivity of the vtc-XES spectra of Cr. Unfortunately, vtc-XES transitions have very low probability (a factor of 103 lower than Kα XES lines), and therefore it is challenging to measure them with good statistics on low concentrated samples. High-energy-resolution fluorescence detection using crystal-based X-ray emission spectrometers also allows to measure RXES, which is alternatively called resonant inelastic X-ray scattering (RIXS). RXES spectra can be acquired when an element of interest is excited with the monochromatic incident energy tuned close to the absorption edge c. ±50 eV. In this case, the combined instrumental resolution of the incident beam (monochromator) and the emission spectrometer should be better than the lifetime broadening of the initial core hole formed during the absorption process and the core hole on the upper orbital formed during the X-ray emission process [18]. The shape and intensity of RXES spectra strongly depend on the incident energy due to many electron processes. RXES spectra at variable incident energy are commonly presented as 2D maps, where one scale corresponds to the incident energy (Eins ) and the other to the emission energy (Eem ) or the energy transfer (Eins − Eem ). The 2D map in Figure 19.9 shows the Ce 2p3/2 3d5/2 RXES of a Pt/CeO2 catalyst, where 2p3/2 indicates the position of the initial core hole and 3d5/2 refers to the position of the electron hole formed during X-ray emission. The top panel in Figure 19.9
349
19 Operando X-Ray Spectroscopies on Catalysts in Action
Figure 19.8 The vtc-XES (Kβ satellite lines) of Cr metal (a), CrB (b), Cr3 C2 (c), CrN (d), Cr2 O3 (e), CrSi (f ), CrP (g), FeCr2 O4 (h), and K2 Cr2 O7 (i). Source: Eeckhout et al. 2009 [17]. Adapted with permission of Royal Society of Chemistry.
(i) (h) (g) Intensity (a.u.)
350
(f)
(e) (d) (c) (b)
(a)
0.0 5960
5970
5980
5990
Emission energy (eV)
illustrates strong changes in the Ce 2p3/2 3d5/2 RXES as a function of incident energy, obtained by cutting through the 2D RXES map at a single incident energy. The right panel in Figure 19.9 demonstrates the high-energy-resolution fluorescence detected (HERFD) XANES or partial fluorescence yield (PFY) XANES obtained by cutting through the 2D RXES map at a single emitted X-ray energy. Since the combined energy resolution of the XES spectrometer and incident monochromator (c. 1 eV) is better than the lifetime broadening of the Ce 2p3/2 hole and comparable to the Ce 5d5/2 hole [18], the sharpness of the Ce L3 HERFD XANES spectrum is enhanced producing sharper features in comparison with standard Ce L3 XANES [19]. Another RXES method is so-called high-energy-resolution off-resonant spectroscopy (HEROS) [20]. In contrast to XES methods probing the density of occupied states of the absorbing element, HEROS probes the density of unoccupied
19.4 In Situ and Operando Cells
5.740 5.732 5.726 5.715
Emission
4.860
4.847
4.835
4.822
4.810
Incident energy (keV)
5.738 5.734 5.730
5.722 5.718
Absorption
5.726
Emission energy (keV)
Figure 19.9 RXES plane of 1.5 wt% Pt/CeO2 measured in 1% carbon monoxide at 150 ∘ C. The incident energy was scanned around the Ce L3 edge from 5.715 to 5.740 keV, and the X-ray emission was recorded from 4.788 to 4.909 keV. The emission energy scale is inverted. Top panel corresponds to RXES spectra at fixed incident energies. Right panel shows HERFD XANES extracted at the maximum of Lα1 line. Source: Kopelent et al. 2016 [37]. Adapted with permission of Royal Society of Chemistry. (See online version for color figure).
states providing spectra similar to HERFD XANES. To measure HEROS the incident X-ray energy should be tuned to a single energy lying 10–20 eV below the absorption edge. In this case, the electron can be excited into an unoccupied state above the Fermi level by taking energy from the emitted photon. In spite of the fact that this process has a low probability, HEROS spectra can be easily measured at synchrotron facilities especially on concentrated samples and without being influenced by self-absorption effects.
19.4 In Situ and Operando Cells To obtain relevant information about the catalyst structure under reaction conditions, it is important to design a cell reproducing reaction conditions and allowing XAS and XES data acquisition [2]. The reaction conditions comprise many parameters including temperature, pressure, reactant space velocity, catalytic conversion, and absence of diffusion limitations and temperature gradients. The catalytic conversion of the sample in the operando cell can be monitored online with a mass spectrometer, chromatograph, etc. XAS and XES spectra are typically measured through X-ray transparent windows, which need to be integrated in the cell design and be thin enough to transmit enough X-rays, be inert toward the reactants, be tight for reactants and products, sustain pressure, not increase the dead volume, and not add temperature gradients. Ideally,
351
352
19 Operando X-Ray Spectroscopies on Catalysts in Action
Insulation shield
Gaseous reactants Laser
Capillary
Gas out
Gaseous products (GC-MS, RGA)
Window 1
Gas in
Heat cartridges
Cell body Window 2 Gas out
SS-bracket Slide Slide
Surface species (micro-IR)
NP charge, size, shape, structure (XAFS, TEM)
X-rays
(b)
Micro X-rays
Gasket
(a) Gas in
e-Beam Micro infrared
(c)
Support (Raman)
Particle/support interface (EELS)
(d)
Figure 19.10 Examples of flow reactors suitable for operando X-ray spectroscopy. (a) and (b) show different versions of a so-called capillary reactor, (c) demonstrates a metal body reactor, and (d) exhibits a micro reactor also suitable for transmission electron microscopy. Source: (a) Chupas et al. 2008 [21]. Adapted with permission of International Union of Crystallography; (b) van Beek et al. 2011 [22]. Adapted with permission of Taylor & Francis; (c) Chiarello et al. 2014 [25]. Adapted with permission of American Institute of Physics; (d) Zhao et al. 2015 [24]. Adapted with permission of John Wiley & Sons.
such windows should also not be crystalline as this adds unwanted diffraction peaks to the XAS spectra decreasing their quality. Typical window materials are kapton (polyimide), graphite foil, polycarbonate, Teflon, quartz glass, beryllium, silicon nitride, glassy carbon, etc. Transmission through window materials (plotted in Figure 19.2) depends on the X-ray energy and window thickness. Figure 19.10 shows examples of flow reactors suitable for operando spectroscopy. Figures 19.10a,b demonstrate two versions of capillary plug flow reactors [21, 22] originally proposed in [23], where a catalyst powder is fixed between two quartz wool plugs inside a quartz or kapton capillary. Figure 19.10c shows a different version of a plug flow reactor made of stainless steel, where catalyst powder is fixed by quartz wool plugs and X-ray spectroscopy can be measured through graphite or kapton windows. Figure 19.10d depicts an example of in situ electron microscopy microreactor that can also be used for operando X-ray spectroscopic studies [24]. The dead volume can limit the possibilities for time-resolved studies in flow reactors. One can use a tracer gas and a mass spectrometer to evaluate how fast the gases in the reactor can be exchanged upon a gas switching experiment. Even working with very low dead volumes and high flow rates, it is difficult to have complete gas exchange in the reactor in less than one second at atmospheric pressure. Therefore fast processes and short-lived intermediates are challenging to detect using gas switching strategy [25, 26]. Nevertheless, practically the turnover frequency of many industrially relevant catalytic processes is slower than one catalytic cycle per second. Therefore, in many cases, the kinetics of structural changes involved in the rate-limiting step can be successfully studied by operando X-ray spectroscopy providing ideas how to accelerate this step
19.5 Application of Time-Resolved Methods
and thus to design more active catalyst. Much faster processes can be studied when the catalytic reaction can be triggered by light, for example, in the case of photocatalysts [2, 27, 28]. In this case, different electronic processes taking place in femtosecond–millisecond regimes can be analyzed using laser pump–X-ray probe techniques at synchrotrons and free electron laser facilities. When analyzing catalyst structure in flow reactors both under steady-state and transient conditions, one should be aware that the catalyst structure and reaction kinetics may differ along the catalyst bed. This is typical for chemical processes limited by the propagation of reactants in the catalyst bed (diffusion-limited regime) and not by the reaction itself (kinetics-limited regime) [29]. When switching gases in the diffusion-limited regime, chemical transformations at the beginning of the catalyst bed will happen faster than in the middle and at the end of the catalyst bed [26]. In the steady-state regime and at high conversion, the structure of the catalyst can also vary along the catalyst bed [30, 31]. In this case, the concentration of reactants is higher at the beginning of the catalyst bed, while the end of the bed mainly interacts with reaction products, complicating evaluation of structure–activity relationships. Therefore, for correct chemical speciation of active sites (especially under steady-state conditions), one needs to work at low conversions (deferential conditions), assuring that the structure of the catalyst is uniform and has similar activity all over the reactor.
19.5 Application of Time-Resolved Methods Synchrotron facilities offer various possibilities for time-resolved XAS studies [2, 32]. The desired time resolution depends on the dynamics of the process that researches need to follow and the capabilities of the experimental facility. A typical step-by-step energy scan of a XAS spectrum, collected by rotating the Si(111) or Si(311) monochromator crystal, covering an energy range of 1000 eV, can take at least 20–30 minutes and may take longer to increase the signal-to-noise ratio. In the same mode, a 100 eV XANES scan will take at least two to three minutes. This acquisition mode is acceptable to study catalysts under steady-state conditions and to monitor slow processes of catalysts deactivation. Nowadays more and more synchrotrons offer faster measuring modes combining rapidly oscillating monochromators with continuous data acquisition allowing 1000 eV energy scans in minutes, seconds, and even tens of milliseconds. These modes are generally called quick XAS or on-the-fly XAS [2, 32, 33]. Instead of using a continuously scanning monochromator, dispersive polychromators (with bent Si single crystals) are also an option for time-resolved studies offering even faster data acquisition rate. However, the quality of EXAFS spectra in this case is typically lower due to the different types of detectors and larger technical challenges, which will only get worse at new generation synchrotrons [34]. Time-resolved XES and 1D RXES at fixed incident energy is also possible using, for example, the energy-dispersive von Hamos-type spectrometer (as introduced in Section 19.3), which allows to acquire XES spectra using a 2D pixel detector in “one shot” without the necessity of moving any optical element [35, 36]. Time-resolved XAS and
353
19 Operando X-Ray Spectroscopies on Catalysts in Action
CuI(NH3)2
CuII(NH3)4
LC fit weights (a.u.)
NH3 on
0.8
CuII(OH)2
CuII-Z
225 °C
190 °C
Concentration (ppm)
354
(a)
NH3 on
CuII(NOx)z 400 °C
275 °C (b)
NH3 on
(d)
(c)
0.6
NH3 on
0.4 0.2 0.0
1000
(e)
(g)
(f)
(h)
NO
800
NO
600
NH3
400
NO
NO
NH3
NH3
200
NH3
0
0
400 800 1200 Time (s)
0
400 800 1200 Time (s)
0
400 800 1200 Time (s)
0
400 Time (s)
Figure 19.11 Time-resolved chemical speciation of copper in the Cu-SSZ-13 catalyst and the corresponding gas-phase analysis at the reactor outlet during addition of 1000 ppm NH3 flow to the flow of 1000 ppm NO, 6 vol% O2 , 6 vol% H2 O in N2 at 190 ∘ C (a, e), 225 ∘ C (b, f ), 275 ∘ C (c, g), and 400 ∘ C (d, h). Source: Marberger et al. 2018 [7]. Adapted with permission of Springer Nature. (See online version for color figure).
XES can be very useful to follow fast deactivation and activation processes in operando as well as changes in the structure of catalysts operating under transient conditions (e.g. automotive catalysts, chemical looping catalysts, etc.). Automotive catalysts typically operate under fast switching between oxidizing and reducing atmospheres to remove CO, NOx , and hydrocarbons from the car exhaust. Therefore, their structure needs to be followed under transient conditions to decipher structure–activity relationships. For example, a time-resolved Cu K-edge XANES study uncovered recently the reason for low-temperature ammonia slip in copper-based zeolite catalyst (Cu-SSZ-13) used for SCR of NOx by ammonia [7]. Figure 19.11 shows the time-resolved chemical speciation of Cu K-edge XANES spectra during cutoff of a 1000 ppm NH3 flow (mimicking pulsing of ammonia to a real exhaust gas system) to the catalyst preconditioned in SCR conditions (a flow of 1000 ppm NO, 1000 ppm NH3 , 6 vol% O2 , 6 vol% H2 O in N2 ). Linear combination fitting of time-resolved XANES spectra (Figure 19.4) demonstrated that high concentrations of Cu(I) amino complex at lower temperatures (e.g. 190 ∘ C) inhibit the NOx conversion. This mechanistic finding helped to improve the efficiency of a catalytic converter at low temperatures by reducing ammonia concentration in the feed. Specially designed transient experiments can also help in detecting short-lived intermediates invisible under steady-state conditions and correlate the rate of their formation and decay to the rate-limiting step. Recently this approach was applied to study Pt-loaded ceria-based catalysts used for low-temperature CO oxidation [36, 37]. A quantitative correlation of the Ce3+ formation rate under
19.5 Application of Time-Resolved Methods
transient conditions with the steady-state CO2 formation rate suggested that formation of short-lived Ce3+ intermediates at the Pt–CeO2 interface is kinetically coupled to the rate-determining step and accordingly that this rate needs to be accelerated to make a more active catalyst. At the same time, it was shown that there is a fraction of long-lived Ce3+ spectator species that do not participate in the catalytic cycle. Figure 19.12 illustrates how the time-resolved Ce 2p3/2 3d5/2 Pixel array 3s 2s detector 1s
Time-resolved RXES 11% Ce3+ 0% Ce3+ Fit
Normalized intensity (a.u.)
CO
y ra n X- ssio i em
Ge (400)
Eemission
Energy dispersive optics
O2 4800
(b)
Plug flow reactor in-situ cell
Monochromatic X-ray beam
(a) CO + O2 CO
24 °C
Initial rate
2
Ce3+ (%)
4
Rate (molco )(gcatalysts)
5
1E–5
3 Active Ce3+
2
Spectator Ce3+
1
(c)
60
120
40
180
Time (s)
21
1E–6 Transient EA = 36 (±7) kJ/mol
(d)
Slow
T (°C) 30
4860
Steady state EA = 44 (±1) kJ/mol
1E–7 3.1
0 0
50
4840 4820 Emission energy (eV)
3.2
3.3 1000 (T/K–1)
3.4
Oxygen atom Carbon monoxide Dioxygen Carbon dioxide
Fast
Ce3+
(e)
Ce4+ Active ceria
Ce3+ Spectator
Figure 19.12 Experimental setup (a), time-resolved Ce 2p3/2 3d5/2 RXES of 1.5 wt% Pt/CeO2 catalyst with the reference spectra (b), time-resolved Ce3+ profiles under transient conditions (c), quantitative correlation between the initial Ce3+ formation rate in CO and the steady-state CO oxidation rate under steady-state conditions in CO + O2 (d), and the mechanism of CO oxidation at the Pt–CeO2 interface involving short-lived Ce3+ intermediate (e). Source: Kopelent et al. 2015 [36]. Adapted with permission of John Wiley & Sons. (See online version for color figure).
355
FT(k3* (χ(k))) (Å–4)
19 Operando X-Ray Spectroscopies on Catalysts in Action
(k3* (χ(k))) (Å–3)
356
1.2 V RHE, anodic 1.55 V RHE, anodic 1.2 V RHE, cathodic
4
2
0 3 0
–3
0
1
2
3
4
5
6
7
Radial distance (Å)
Figure 19.13 Fourier transform Co K-edge EXAFS spectra for the BSCF electrode (magnitude and imaginary part) measured under operando conditions. The shoulder at ∼3.5 Å and the peak at ∼4 Å correspond to Co–Ba/Sr and Co–Co/Fe paths in the perovskite structure. The major peak located at ∼2.9 Å suggests the presence of Co–Co/Fe edge-sharing complex in the Cox Fe1−x O(OH) structure increasing in intensity under water splitting conditions. Source: Fabbri et al. 2017 [38]. Adapted with permission of Springer Nature. (See online version for color figure).
RXES spectra were acquired under operando conditions (panel a) and how the Ce3+ concentration was quantified using reference samples with known Ce3+ concentration (panel b). It also demonstrates how the initial Ce3+ formation rate in CO was fitted (panel c) and correlated to the steady-state CO2 formation rate (panel d) to prove that the formation of short-lived Ce3+ intermediates is involved in the rate-determining step (panel e). Electrocatalytic processes can also be followed by time-resolved XAS under real operando conditions. Recently it helped to get insight in the active structure of a perovskite-based Ba0.5 Sr0.5 Co0.8 Fe0.2 O3−𝛿 (BSCF) oxygen evolution catalyst under the conditions at which this catalyst evolves oxygen (overpotentials of 1.45 V), demonstrating that the formation of a new Cox Fe1−x O(OH) phase is responsible for progressive activation of this catalyst under operation conditions [38]. Figure 19.13 illustrates the identification of the formation of the Cox Fe1−x O(OH) phase based on Co K-edge EXAFS.
19.6 Limitations and Challenges A lack of chemical sensitivity sometimes limits the applicability of XAS and XES methods for operando studies in catalysis. X-ray spectroscopy is particularly sensitive to the oxidation state and significant changes in the local coordination of the element of interest. At the same time X-ray spectroscopy is not surface sensitive and probes the average structure of all absorbing atoms in the sample (both
19.7 Concluding Remarks
at the surface and in the bulk). Insufficient chemical sensitivity can be due to several reasons. First, sometimes only a very small fraction of atoms of the element of interest forms the active sites, while a larger fraction remains buried in the bulk or forms inactive species on the surface. Second, the active sites may seem to be inactive because most of the time during the catalytic cycle they do not see a reactant. Third, the lifetime of surface intermediates formed during the catalytic cycle can be so short that it is not detected due to a lack of time resolution. In all cases chemical sensitivity can be improved: In the first case, the fraction of active species in the catalyst can be increased by depositing the element of interest on a support material or by using a dedicated synthesis method, increasing the fraction of active sites. The second problem is difficult to resolve until experimental conditions can be tuned to allow stronger interaction of reactive molecules with the active sites. In the third case, experiments under transient conditions can help to detect short-lived intermediates, as shown above for the detection of the Ce3+ intermediate in the low-temperature CO oxidation over a Pt/CeO2 catalyst [36]. Detection of weak spectral features related to the reactivity of a minor fraction of active sites can be facilitated if the catalyst state can be reversibly cycled between different states (a modulated excitation experiment) [39–41]. In this case the resulting time-resolved XAS spectra can be processed using the phase-sensitive detection approach that allows to filter out the reversible spectral changes related to the active sites. The sensitivity of XANES and XES can also be enhanced by enhancing the energy resolution and by improving the signal-to-noise ratio. The energy resolution of XANES can be enhanced using fluorescence detection with an energy resolution better than the core hole lifetime broadening [14, 19, 42]. RXES methods can also offer higher chemical sensitivity in comparison with traditional XANES due to the stronger difference between different states of the metal [37]. This way of detection, however, requires much longer acquisition time due to much smaller number of photons arriving to the detector. Another known challenge for X-ray spectroscopy that is important to be aware of is X-ray damage. Typically damage appears as photoreduction of the element of interest by the X-ray beam and more often happens for materials containing metalorganic compounds than for purely inorganic structures. Each material should be carefully checked for possible X-ray damage by monitoring the evolution of spectral features when exposing it to the beam. To slow the X-ray damage, it is possible to reduce the X-ray beam density on the sample by defocusing the beam. Once the rate of X-ray damage is quantified, one can also consider moving the beam to fresh sample spots after a certain time. Cooling of the sample is another known strategy to slow down X-ray damage; however, in this case an operando approach is not possible.
19.7 Concluding Remarks To summarize, X-ray spectroscopy presents great opportunities for the operando studies of catalysts in the next 20 years. First, as we demonstrated in the Sections
357
358
19 Operando X-Ray Spectroscopies on Catalysts in Action
19.4 and 19.5, operando X-ray spectroscopy will continue to help in solving challenging problems in heterogeneous catalysis by establishing structure–activity relationships and allowing the rational design of better catalysts. For this task, targeted design of well-defined catalytic materials combining high catalytic activity and optimized for X-ray spectroscopic studies will help to get deeper insight in specific reaction mechanisms. Second, development of more efficient detectors and more precise X-ray optics will allow increasing the chemical sensitivity of XAS and XES methods. Third, time-resolved studies and experiments under transient conditions will continue to emerge helping to uncover reaction intermediates and guide the rational design of better catalysts. Fourth, the combination of X-ray spectroscopies with complementary methods such as infrared and Raman spectroscopies, XRD and total X-ray scattering, electron paramagnetic spectroscopy (EPR), nuclear magnetic resonance (NMR), X-ray photoelectron spectroscopy (XPS), isotope labeling, etc. will help to tackle complex reaction mechanisms where X-ray spectroscopy does not provide sufficient information. Fifth, joint efforts of synchrotron facilities, research laboratories, and industry will help to build dedicated experimental setups and new expertise in the field of fundamental and applied catalysis. And finally, the continued development of theoretical methods for XANES and RXES interpretation will facilitate the analysis of experimental results, moving from a fingerprint approach to detailed knowledge.
References 1 Willmott, P. (2019). An introduction to synchrotron radiation: techniques and
applications, 2e. Hoboken, NJ: Wiley. 2 van Bokhoven, J.A. and Lamberti, C. (2016). X-Ray Absorption and X-Ray
3
4
5 6 7
8
Emission Spectroscopy: Theory and Applications. Wiley. http://www.wiley.com/ WileyCDA/WileyTitle/productCd-1118844238.html. Seidler, G.T., Mortensen, D.R., Remesnik, A.J. et al. (2014). A laboratory-based hard x-ray monochromator for high-resolution x-ray emission spectroscopy and x-ray absorption near edge structure measurements. Rev. Sci. Instrum. 85: 113906. https://doi.org/10.1063/1.4901599. Padamati, S.K., Angelone, D., Draksharapu, A. et al. (2017). Transient formation and reactivity of a high-valent nickel(IV) oxido complex. J. Am. Chem. Soc. 139: 8718–8724. https://doi.org/10.1021/jacs.7b04158. Bunker, G. (2010). Introduction to XAFS by Grant Bunker. Core: Camb. https://doi.org/10.1017/CBO9780511809194. Calvin, S. (2013). XAFS for Everyone. CRC Press. https://www.crcpress.com/ XAFS-for-Everyone/Calvin/p/book/9781439878637. Marberger, A., Petrov, A.W., Steiger, P. et al. (2018). Time-resolved copper speciation during selective catalytic reduction of NO on Cu-SSZ-13. Nat. Catal. 1: 221. https://doi.org/10.1038/s41929-018-0032-6. Voronov, A., Urakawa, A., van Beek, W. et al. (2014). Multivariate curve resolution applied to in situ X-ray absorption spectroscopy data: an efficient tool
References
9
10
11 12
13
14
15
16
17
18 19
20
21
22
for data processing and analysis. Anal. Chim. Acta 840: 20–27. https://doi.org/ 10.1016/j.aca.2014.06.050. Rochet, A., Baubet, B., Moizan, V. et al. (2016). Co-K and Mo-K edges quick-XAS study of the sulphidation properties of Mo/Al2 O3 and CoMo/Al2 O3 catalysts. C.R. Chim. 19: 1337–1351. https://doi.org/10.1016/ j.crci.2016.01.009. Sayers, D.E., Stern, E.A., and Lytle, F.W. (1971). New technique for investigating noncrystalline structures: Fourier analysis of the extended X-ray – absorption fine structure. Phys. Rev. Lett. 27: 1204. https://doi.org/ 10.1103/PhysRevLett.27.1204. Frenkel, A. (2009). Solving the 3D structure of metal nanoparticles. Z. Kristallogr. Cryst. Mater. 222: 605–611. https://doi.org/10.1524/zkri.2007.222.11.605. Beale, A.M. and Weckhuysen, B.M. (2010). EXAFS as a tool to interrogate the size and shape of mono and bimetallic catalyst nanoparticles. Phys. Chem. Chem. Phys. 12: 5562–5574. https://doi.org/10.1039/B925206A. Frenkel, A.I. (2012). Applications of extended X-ray absorption fine-structure spectroscopy to studies of bimetallic nanoparticle catalysts. Chem. Soc. Rev. 41: 8163–8178. https://doi.org/10.1039/C2CS35174A. Glatzel, P. and Bergmann, U. (2005). High resolution 1s core hole X-ray spectroscopy in 3d transition metal complexes – electronic and structural information. Coord. Chem. Rev. 249: 65–95. https://doi.org/10.1016/j.ccr.2004 .04.011. Szlachetko, J., Nachtegaal, M., de Boni, E. et al. (2012). A von Hamos x-ray spectrometer based on a segmented-type diffraction crystal for single-shot x-ray emission spectroscopy and time-resolved resonant inelastic x-ray scattering studies. Rev. Sci. Instrum. 83: 103105. https://doi.org/10.1063/1 .4756691. Hall, E.R., Pollock, C.J., Bendix, J. et al. (2014). Valence-to-core-detected X-ray absorption spectroscopy: targeting ligand selectivity. J. Am. Chem. Soc. 136: 10076–10084. https://doi.org/10.1021/ja504206y. Eeckhout, S.G., Safonova, O.V., Smolentsev, G. et al. (2009). Cr local environment by valence-to-core X-ray emission spectroscopy. J. Anal. At. Spectrom. 24: 215–223. https://doi.org/10.1039/b808345m. Campbell, J.L. and Papp, T. (2001). Widths of the atomic K–N7 levels. At. Data Nucl. Data Tables 77: 1–56. https://doi.org/10.1006/adnd.2000.0848. Paun, C., Safonova, O.V., Szlachetko, J. et al. (2012). Polyhedral CeO2 nanoparticles: size-dependent geometrical and electronic structure. J. Phys. Chem. C 116: 7312–7317. https://doi.org/10.1021/jp300342b. Szlachetko, J., Nachtegaal, M., Sa, J. et al. (2012). High energy resolution off-resonant spectroscopy at sub-second time resolution: (Pt(acac)2 ) decomposition. Chem. Commun. 48: 10898–10900. https://doi.org/10.1039/c2cc35086f. Chupas, P.J., Chapman, K.W., Kurtz, C. et al. (2008). A versatile sample-environment cell for non-ambient X-ray scattering experiments. J. Appl. Crystallogr. 41: 822–824. https://doi.org/10.1107/S0021889808020165. van Beek, W., Safonova, O.V., Wiker, G., and Emerich, H. (2011). SNBL, a dedicated beamline for combined in situ X-ray diffraction, X-ray absorption
359
360
19 Operando X-Ray Spectroscopies on Catalysts in Action
23
24
25
26
27
28
29
30
31
32
33
34
and Raman scattering experiments. Phase Transitions 84: 726–732. https://doi .org/10.1080/01411594.2010.549944. Clausen, B.S., Gråbæk, L., Steffensen, G. et al. (1993). A combined QEXAFS/XRD method for on-line, in situ studies of catalysts: examples of dynamic measurements of Cu-based methanol catalysts. Catal. Lett. 20: 23–36. https://doi.org/10.1007/BF00772594. Zhao, S., Li, Y., Stavitski, E. et al. (2015). Operando characterization of catalysts through use of a portable microreactor. ChemCatChem 7: 3683–3691. https://doi.org/10.1002/cctc.201500688. Chiarello, G.L., Nachtegaal, M., Marchionni, V. et al. (2014). Adding diffuse reflectance infrared Fourier transform spectroscopy capability to extended x-ray-absorption fine structure in a new cell to study solid catalysts in combination with a modulation approach. Rev. Sci. Instrum. 85: 074102. https://doi .org/10.1063/1.4890668. Guda, A.A., Bugaev, A.L., Kopelent, R. et al. (2018). Fluorescence-detected XAS with sub-second time resolution reveals new details about the redox activity of Pt/CeO2 catalyst. J. Synchrotron Radiat. 25 https://doi.org/10.1107/ S1600577518005325. Smolentsev, G., Soldatov, M.A., Probst, B. et al. (2018). Structure of the CoI intermediate of a cobalt pentapyridyl catalyst for hydrogen evolution revealed by time-resolved X-ray spectroscopy. ChemSusChem 11: 3087–3091. https:// doi.org/10.1002/cssc.201801140. Penfold, T.J., Szlachetko, J., Santomauro, F.G. et al. (2018). Revealing hole trapping in zinc oxide nanoparticles by time-resolved X-ray spectroscopy. Nat. Commun. 9: 478. https://doi.org/10.1038/s41467-018-02870-4. Chorkendorff, I. and Niemantsverdriet, J.W. (2003). Concepts of Modern Catalysis and Kinetics. Wiley Online Library. http://onlinelibrary.wiley.com/ book/10.1002/3527602658. Grunwaldt, J.-D., Hannemann, S., Schroer, C.G., and Baiker, A. (2006). 2D-mapping of the catalyst structure inside a catalytic microreactor at work: partial oxidation of methane over Rh/Al2 O3 . J. Phys. Chem. B 110: 8674–8680. https://doi.org/10.1021/jp060371n. Hartfelder, U., Singh, J., Haase, J. et al. (2016). Detecting and utilizing minority phases in heterogeneous catalysis. Sci. Rep. 6: srep37597. https://doi.org/10 .1038/srep37597. Frenkel, A.I., Khalid, S., Hanson, J.C., and Nachtegaal, M. (2013). QEXAFS in catalysis research: principles, data analysis, and applications. In: In Situ Characterization of Heterogeneous Catalysts (eds. J.A. Rodriguez, J.C. Hanson and P.J. Chupas), 23–47. Wiley. http://onlinelibrary.wiley.com/doi/10.1002/ 9781118355923.ch1/summary. Müller, O., Nachtegaal, M., Just, J. et al. (2016). Quick-EXAFS setup at the SuperXAS beamline for in situ X-ray absorption spectroscopy with 10 ms time resolution. J. Synchrotron Radiat. 23: 260–266. https://doi.org/10.1107/ S1600577515018007. Abe, H., Aquilanti, G., Boada, R. et al. (2018). Improving the quality of XAFS data. J. Synchrotron Radiat. 25: 972–980. https://doi.org/10.1107/ S1600577518006021.
References
35 Marchionni, V., Szlachetko, J., Nachtegaal, M. et al. (2016). An operando
36
37
38
39
40
41
42
emission spectroscopy study of Pt/Al2 O3 and Pt/CeO2 /Al2 O3 . Phys. Chem. Chem. Phys. 18: 29268–29277. https://doi.org/10.1039/C6CP05992A. Kopelent, R., van Bokhoven, J.A., Szlachetko, J. et al. (2015). Catalytically active and spectator Ce3+ in ceria-supported metal catalysts. Angew. Chem. Int. Ed. 54: 8728–8731. https://doi.org/10.1002/anie.201503022. Kopelent, R., van Bokhoven, J.A., Nachtegaal, M. et al. (2016). X-ray emission spectroscopy: highly sensitive techniques for time-resolved probing of cerium reactivity under catalytic conditions. Phys. Chem. Chem. Phys. 18: 32486–32493. https://doi.org/10.1039/C6CP05830B. Fabbri, E., Nachtegaal, M., Binninger, T. et al. (2017). Dynamic surface self-reconstruction is the key of highly active perovskite nano-electrocatalysts for water splitting. Nat. Mater. 16: 925–931. https://doi.org/10.1038/ nmat4938. Ferri, D., Newton, M.A., and Nachtegaal, M. (2011). Modulation excitation X-ray absorption spectroscopy to probe surface species on heterogeneous catalysts. Top. Catal. 54: 1070. https://doi.org/10.1007/s11244-011-9727-5. Urakawa, A., Buergi, T., and Baiker, A. (2008). Sensitivity enhancement and dynamic behavior analysis by modulation excitation spectroscopy: principle and application in heterogeneous catalysis. Chem. Eng. Sci. 63: 4902–4909. https://doi.org/10.1016/j.ces.2007.06.009. König, C.F.J., Schildhauer, T.J., and Nachtegaal, M. (2013). Methane synthesis and sulfur removal over a Ru catalyst probed in situ with high sensitivity X-ray absorption spectroscopy. J. Catal. 305: 92–100. https://doi.org/10.1016/j .jcat.2013.05.002. Safonova, O.V., Tromp, M., van Bokhoven, J.A. et al. (2006). Identification of CO adsorption sites in supported Pt catalysts using high-energy-resolution fluorescence detection X-ray spectroscopy. J. Phys. Chem. B 110: 16162–16164. https://doi.org/10.1021/jp063416t.
361
363
20 Methodologies to Hunt Active Sites and Active Species Atsushi Urakawa Delft University of Technology, Department of Chemical Engineering, Catalysis Engineering, Van der Maasweg 9, 2629 HZ Delft, The Netherlands
20.1 Introduction Heterogeneously catalyzed reactions typically take place at so-called active sites defined by unique geometrical and electronic structures of the catalyst surface. Such unique structures are the core of heterogeneous catalysis and accelerate formation of active chemical species undergoing desired selective chemical transformation at these sites. Understanding such active sites and active species is considered as one of the most critical elements toward knowledge-based rational design of catalyst materials and maximizing selectivity to a desired chemical product. In situ spectroscopy and particularly operando methodologies, which signify in situ studies under catalytically relevant conditions (i.e. reactivity is measured simultaneously), are widely used nowadays, and numerous analytical tools are available, elucidating geometrical and electronic structures and chemical compositions of bulk material, fluid, and surface species under reaction conditions, even with spatial resolutions on different length scales (nanometer to meter scales) [1–5]. The library of the analytical tools is expanding its size and capabilities. These analytical methods are even combined, in so-called multiprobe or multimodal approaches, providing multifaceted and more holistic information about the nature of catalyst materials and active species [6, 7]. The advances are significant; however, there are still general difficulties we encounter during in situ and operando studies in the detection of active sites and species. They are selectivity and sensitivity to detect what we wish to see. This dynamic nature of active sites and species in addition to the challenging detection of active sites and species at gas/solid and solid/liquid interfaces, where heterogeneously catalyzed reactions commonly take place, has kept catalysis research and development highly empirical. The general situation of catalytic reaction and issues in spectroscopic studies are illustrated in Figure 20.1 where a man is playing golf in front of many spectators (Figure 20.1a). Obviously, what we wish to observe is the action of the golf player (active species/sites) and not those of spectators. The spectators may play key roles in affecting the performance of the player through cheerful actions (like Heterogeneous Catalysts: Advanced Design, Characterization and Applications, First Edition. Edited by Wey Yang Teoh, Atsushi Urakawa, Yun Hau Ng, and Patrick Sit. © 2021 WILEY-VCH GmbH. Published 2021 by WILEY-VCH GmbH.
364
20 Methodologies to Hunt Active Sites and Active Species
(a)
(b)
(c)
(d)
(e)
Figure 20.1 Illustration of sensitivity and selectivity issues in the detection active one (the gold player) and distinguishing him from the spectators. From d, through c, to b, the detection sensitivity is improved with a black-and-white camera. After the color addition to b, a is obtained. e shows sensitive and selective detection of only the active one. (See online version for color figure).
the effect of solvent and support material on active sites/species), but if we imagine that we are cameramen, we just wish to observe the precise action of the player. When an outdated low-quality black-and-white camera is used, we may obtain an unmeaningful picture (Figure 20.1d), and it is impossible to understand the action of the active one (i.e. the player). This is because of (i) poor signal-to-noise (S/N) ratio (low sensitivity), (ii) dynamic nature and time resolution (only an averaged picture is obtained), and (iii) low image resolution. When all these resolutions and sensitivity are improved, the picture would improve from Figure 20.1d to Figure 20.1c and b where a golf player is more easily identified. When another dimension, here color, is added, the picture (Figure 20.1a) becomes much clearer to identify the active one at quick glance. These issues on the detection sensitivity are common in in situ and operando studies of catalytic reactions where better detectors improve the spectral quality (i.e. from Figure 20.1d to Figure 20.1b). The color addition is similar to the case of multimodal spectroscopy or spectromicroscopy (doing spectroscopy at each pixel during imaging) where detailed and/or complementary information is added to identify the chemical nature more precisely. What is even better for research would be to use the method to see only the player, the active one, without seeing spectators by adding selectivity into the
20.2 Modulation Excitation Technique
detection. The above detection sensitivity and selectivity are the major challenges in operando studies where various kinds of molecules are abundantly present and measurements are performed under compromised conditions for sensitivity due to high temperature, high pressure, and the use of practically relevant reactor and complex materials, hindering the detection of minute fraction of signals from active ones. There are significant advances toward the detection of active ones, mainly using transient approach where a catalytic system is perturbed by an external parameter such as concentration of reactant(s). In this chapter, the general concepts and representative examples of the three methodologies for this aim are briefly explained.
20.2 Modulation Excitation Technique Modulation excitation (ME) technique (when used with spectroscopy, called modulation [or modulated] excitation spectroscopy [MES]) is the method particularly designed for adding selectivity to detect active ones and, simultaneously, improving the detection sensitivity drastically [8–10]. An ME experiment uses a perturbation of an external experimental parameter, so-called stimulation (or stimulus), which can be a temperature, concentration, light flux, pH, and electric current, modulated at a constant frequency. Stimulation is chosen so that it influences the kinetics of the system (e.g. catalytic reaction), especially the concentration of what we wish to see. If a system is perturbed by a periodic stimulation, all the parameters in the system, which are affected by this perturbation, will also change periodically and at the same frequency as the stimulation (𝜔) or harmonics thereof (2𝜔, 3𝜔, etc.). The response of the affected parameters typically shows a frequency-dependent amplitude and phase delay with respect to the stimulation, and one can learn about kinetics of physicochemical processes from the amplitude, phase delay, and their frequency dependency. During the initial periodic cycles, the affected parameters relax to new quasi-steady-state values around which they oscillate (Figure 20.2, showing an example of the response at one energy point). When the response of the periodically oscillating signal is denoted as S(t), the phase-domain response Sk at the fundamental (k = 1, k: demodulation index) and harmonic (k = 2,3, …) frequencies are obtained by a mathematical treatment of Phase delay Active species response φ Signal intensity
Figure 20.2 Typical profile of active species responding to the stimulation. Note that the delay between stimulation and response and the response amplitude contains kinetic information. Source: Reprinted with permission from Urakawa et al. [10]. Copyright 2008, Elsevier.
Stimulation
Time
365
366
20 Methodologies to Hunt Active Sites and Active Species
an experimental signal S(t), so-called phase-sensitive detection (PSD) or simply demodulation (Eq. (20.1)): T
2 S(t) sin(k𝜔t + 𝜙PSD )dt (20.1) k T ∫0 As essence, PSD allows selective extraction of the k𝜔 component of the response. This is extremely useful in removing fully the noise at frequencies different from k𝜔. Figure 20.3 illustrates the typical outputs of an ME experiment and the PSD principle. In an MES experiment, spectra are acquired in a time-resolved manner during a number of modulated cycles of stimulation A(t), and the obtained spectra are averaged √ into one cycle. This simple averaging procedure increases the S/N by about N according to Poisson statistics (N is the number of cycles). An actual spectral response (at one energy point in Figure 20.3) can be shown as the sum of three components B(t) + C(t) + D(t), where B(t) is a signal of species actively responding to the stimulation (the one we are interested in), C(t) is spectator species response, and D(t) is noise (Figure 20.3). Typically, the active species respond at the same frequency 𝜔 with a phase delay (or phase lag) 𝜑 with respect to the stimulation. On the other hand, spectator species, e.g. solvent and catalyst support, do not respond to the stimulation, and the response C(t) is simply a constant. On the other hand, the noise D(t) typically contains high frequency elements as noticed by the Fourier decomposition. After the PSD at k = 1 using Eq. (20.1) applied to the time-domain response signal that is expressed by B(t) + C(t) + D(t), the corresponding phase-domain total response signal B(𝜙PSD ) + C(𝜙PSD ) + D(𝜙PSD ) is obtained. What is most important is that the signals of spectator C and most of the noise D vanish in the phase domain thanks to the frequency filtering. For the spectator signal (a constant value), the signal cancellation can be easily understood by multiplying the constant value with a sinusoidal wave of one period and then by integrating over the period. Due to the same positive and negative areas, the integration yields 0. On the other hand, noise typically contains frequency elements higher than k = 1, and due to the nature of the frequency filtering at k = 1, most of the noise is filtered off by PSD. The active species term in the phase domain B(𝜙PSD ) is almost identical to that in the time domain, B(t), without losing the kinetic information (amplitude and phase delay) inside. Generally, the C and D components are much larger than the signal of our interest B; hence PSD drastically enhances the S/N ratio and allows selective detection of the interesting active species by carefully choosing the type of stimulation. One can do much more using PSD with k > 1, by which the amplitude and phase delay of k𝜔 component of a response can be accurately extracted [11]. The higher frequency terms (k𝜔 component) are useful for kinetic analysis and separation of signals with fast kinetics because the transformation and kinetics of the species we wish to capture need to be in the same order of modulation frequency. The most common shape of stimulation created by simple on/off switch, namely, square wave, contains odd frequency terms (1𝜔, 3𝜔, 5𝜔, 7𝜔, etc.) with decreasing amplitudes at higher k𝜔. This allows convenient extraction of high frequency )= Sk (𝜙PSD k
20.2 Modulation Excitation Technique
(i.e. fast kinetics) responses by performing PSD at higher k𝜔, although the S/N of the responses becomes lower due to the lower amplitudes and thus corresponding responses at higher k𝜔. Furthermore, the phase delay determined by PSD is related to the delay observed in the time domain. This allows studying sequence of elementary reactions and their kinetics [9, 10]. The ME technique can be combined with any analytical techniques, and its power can be most noticeable when combined with a broadband analytical method such as infrared (IR) spectroscopy [9, 10]. The combination allows the separation of overlapping bands of different kinetic behavior utilizing the different phase delays of responses originating from respective species [10]. The advantages and power have led to a broad range of application of the ME technique from catalysis, gas sensors [12], and biological systems [13, 14]. A number of examples in catalysis are summarized in the recent review by Müller and Hermans [15]. The ME technique has been applied also for diffraction studies where the actively responding signals (e.g. sublattices) appear at 2𝜔 (𝜔 is the fundamental frequency of the stimulation) [16–18]. When X-ray energy is modulated, it is also possible for X-ray diffraction (XRD) to be even chemical element selective [19]. IR spectroscopy is the most common application of the ME technique. Here an example in combination with X-ray absorption spectroscopy (XAS) is described (Figure 20.4). Ferri et al. reported the power of the ME technique for Time domain
Signal intensity
Stimulation A(t) = A sin(ωt)
Phase domain
Active species response B(t) = B sin(ωt + φ) (φ < 0)
φ Phase delay
Active species response B(ϕPSD) = B cos(φ–ϕPSD)
PSD Spectator response C(t) = C (≠ 0)
0
0.5 Time (T)
1.0
Spectator response C(ϕPSD) = 0
Noise
Noise
D(t) = ΣDi (iωt + φi)
D(ϕPSD) = 0 (when i ≠ 1)
Figure 20.3 Schematic illustration of sensitivity enhancement by PSD. A(t): stimulation function, B(t): response of active species perturbed by the stimulation where A(t) is the phase delay with respect to the stimulation, C(t): response of spectator species that is not affected by the perturbation, D(t): Fourier-decomposed noise. The sum of the response components, B(t) + C(t) + D(t), is the actual experimental response. Source: Reprinted with permission from Urakawa et al. [10]. Copyright 2008, Elsevier.
367
368
20 Methodologies to Hunt Active Sites and Active Species
a combined XAS-Diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS)-mass spectrometry (MS) study of surface reactions between CO and NO over Al2 O3 -supported Rh and Pd [20]. Figure 20.4a shows how the Rh K-edge spectra change when atmosphere changes between 5% CO and 5% NO. Obviously the changes are too small and not possible to study by eyes. The difference can be studied by looking into the difference spectra (Figure 20.4b). However, the high noise level does not permit to study the changes with high confidence. When the spectra shown in Figure 20.4a are processed by PSD, one obtains phase-domain spectra shown in Figure 20.4c with drastic S/N improvement. Typical improvement in S/N is 1–3 orders of magnitude by this approach. In this study, what is most striking is that the changes observed can be firmly attributed to the surface oxidation of Rh by comparing with the difference spectrum between Rh2 O3 and Rh foil (Figure 20.4d). This is important since from the bulk point of view (Figure 20.4a), Rh is in the metallic state (Rh0 ) but there is a minute fraction of Rh that is oxidized, which is due to the surface Rh oxidation (leading to active sites, Figure 20.1). By the transient and ME approach, the bulk-sensitive analytical technique like XAS can become surface sensitive. In this study, the authors also correlate this surface redox process of the active catalyst metal with the evolutions of surface chemical species and gaseous species observed by IR spectroscopy and MS, respectively, thus clarifying the whole picture of the catalytic reaction [20].
20.3 Steady-State Isotopic Transient Kinetic Analysis (SSITKA) The previously described ME technique is very powerful to study active sites and active species particularly when catalytic activity is simultaneously monitored in a time-resolved fashion. However, it is intrinsically an unsteady-state operation, breaking the balance of the chemical processes taking place at/in the gas, surface, and catalyst bulk by an external stimulation, and it is not best condition to precisely study the kinetics of steady-state chemical reactions. Steady-state isotopic transient kinetic analysis (SSITKA) is the method of choice in such cases. It is also a transient method like the ME technique. However, SSITKA only perturbs the intermediates and product concentrations by means of isotopes, thereby uninfluencing the steady state of the reaction, rendering SSITKA a combination of steady-state and transient technique. A typical SSITKA setup is described in Figure 20.5a, where it consists of a flow of gas, either reactant R or its isotope *R, entering a reactor and the effluent gas concentrations are monitored by MS whose values are calibrated by gas chromatography (GC) for precise quantitative analysis. Using the same concentration of R or *R in the flow, the reaction is operated under steady state. From the normalized response of the product P and/or the corresponding isotope *P, one can extract rich chemical information such as reversibility of chemical reaction, reaction kinetics, and residence time of surface species (𝜏, Figure 20.5a) [23]. The gas flow contains typically an inert (shown as I, e.g. noble gas) to determine and exclude the effect of the gas holdup and mixing in the reactor.
0.2
0.010
0.1
0.005
Absorption
Absorption
20.3 Steady-State Isotopic Transient Kinetic Analysis (SSITKA)
0.0 –0.1
–0.005
(c)
–0.2
(a)
23.2
23.4
23.6
23.8
24.0
24.2
23.35
23.40
0.012 0.009
Absorption
0.01
Absorption
0.000
0.00
–0.01
0.006 0.003 0.000 –0.003 –0.006
–0.02 23.2
(b)
23.4
23.6
23.8
Energy (keV)
24.0
24.2
23.15
(d)
23.20
23.25
23.30
Energy (keV)
Figure 20.4 (a) Time-resolved X-ray absorption spectra at the Rh K-edge recorded at 573 K of one period averaged over seven modulation (5 vol% NO in He vs. 5 vol% CO in He) periods over reduced 2 wt% Rh/Al2 O3 . (b) Selected difference spectra within a single period of the ME experiment (calculated from (a)). (c) Phase-resolved spectra (𝜙PSD = 0–120∘ ) obtained by PSD from the set of time-resolved spectra (a). (d) Enlargement of the same set of spectra together with the scaled spectra of bulk Rh2 O3 (black) and the Rh foil (blue) for comparison. The spectrum in red corresponds to 𝜙PSD = 0∘ ; the spectrum in green is the scaled difference between the reference spectra (Rh–Rh2 O3 ). Source: Adapted with permission from Ferri et al. [20]. Copyright 2010, Royal Society of Chemistry. (See online version for color figure).
The application examples of SSITKA range widely from CO hydrogenation/oxidation [24–26], water–gas shift (WGS) [27, 28], and methane reforming [29, 30], to name a few, and its unique usefulness in understanding reaction kinetics and mechanisms has been demonstrated over a few decades. It is important to highlight that the power of SSITKA has been brought to another level by combining it with operando spectroscopy. Figure 20.5b presents an example where SSITKA is combined with simultaneous monitoring of surface chemical species (intermediates/spectators) by IR (DRIFTS) and products by MS [31]. With the conventional SSITKA, one cannot know the nature of the implied intermediate species, but in this approach it is possible to elucidate what kind of surface species is present under reaction conditions, which ones are involved in the catalytic cycle (i.e. reaction mechanisms), and their kinetics. One of the notable examples of combined IR-MS-SSITKA was reported by Meunier et al. (Figure 20.6) where the reaction mechanism and kinetics of WGS reaction (CO + H2 O ⇌ CO2 + H2 ) over Pt/CeO2 were investigated [32]. Although the reaction has been studied by numerous researchers, there have been great controversies about the nature of active species, especially about the role of surface formate species. The study clearly shows that surface formates elucidated by IR are not the intermediate species but rather spectators, based on the different kinetic response at 433 K (Figure 20.6a) where the decay rate of surface formates does not match with that of the product (CO2 ) formation. However, they become
369
20 Methodologies to Hunt Active Sites and Active Species
the main intermediate at a slightly higher temperature of 493 K (Figure 20.6b) where the decay and formation rates match. Such unique insights can be only gained by combining with spectroscopic inspections. The theory of SSITKA and its application examples reported prior to 1995 and including recent examples are elegantly summarized by Goodwin and Holmen, respectively [21, 23], and interested readers are encouraged to learn what can be studied with SSITKA.
20.4 Multivariate Analysis The ME technique and SSITKA show their full power when combined with broadband spectroscopic methods. However, there is a troublesome issue when transient spectra are analyzed when overlaps of spectral signals/bands arise. Unfortunately this situation is very common. Both MES and SSITKA suffer from such overlaps and make the spectral analysis complicated or impossible for precise kinetic studies. Also, spectral interpretations and decomposition are often performed using reference spectra measured ex situ, despite the fact that catalytically active species and sites/phases are often present stably or transiently only under operando conditions. One powerful method to facilitate disentangling complex spectra into chemically pure and meaningful ones is multivariate spectral analysis. Some of the mathematical engines are called blind source separation methods because there is no need of references for disentangling complex spectra. In multivariate spectral analysis, spectra are typically recorded in a time-resolved fashion, and the set of spectra is processed by a mathematical engine, yielding kinetically pure component spectra and respective concentration profiles (Figure 20.7). Among others, multivariate curve resolution (MCR) is a very popular algorithm for blind source separation [33, 34]. 12
CO + H2O
R GC
Vent
Reactor
MS
Vent
IR reaction cell
*R
Time
l
CO + H2O Abs
t=0
(a)
0 Time
P
t=0
I* S* S
τ
P
0
Relate products to surface intermediates
13
1 F(t)
P* F(t)
370
I Time
(b)
Figure 20.5 (a) Flow scheme of a typical SSITKA setup and normalized transient curve for decaying response of the unlabeled product (P) and labeled product (P*) including the inert response (I). Source: Adapted with permission from Ledesma et al. [21]. Copyright 2014, American Chemical Society. (b) Flow scheme of the IR-MS-SSITKA system for the operando investigation of catalytic reactions. True reaction and minor/spectator intermediates are denoted as I and S, respectively. Source: Adapted with permission from Meunier [22]. Copyright 2010, Elsevier.
20.4 Multivariate Analysis
Figure 20.6 Relative intensity of 13 CO2 (g) (•) and of 12 C-containing surface formate species (◽) following a 12 CO–13 CO isotopic switch under steady-state WGS conditions over a 2% Pt/CeO2 at (a) 433 K and (b) 493 K. Feed: 2% 13 CO + 7% H2 O in 2% Kr/Ar. Source: Reprinted with permission from Meunier et al. [32]. Copyright 2007, Elsevier.
Relative IR and MS signal
There are many recent examples for the use of multivariate spectral analysis in catalysis and related fields [35–41]. Figure 20.8 shows an application of MCR onto the time-resolved DRIFT spectra obtained during on/off cycle of UV light during photocatalytic CO2 reduction over Pt/TiO2 [36]. The raw spectra are extremely complex due to coexisting multiple chemical species on the surface. Thanks to the disentangling power of the multivariate spectral analysis, it was possible to separate signals arising from CO on metallic and oxidized Pt and on defective Ti and from surface formates and carbonates. Based on the temporal evolution and product profiles, a detailed reaction mechanism can be extracted (Figure 20.8).
1 0.8
13CO
2(g) C-formates
12
0.6 0.4 0.2 0 0
5
Relative IR and MS signal
(a)
10 Time (min)
15
20
13CO
1
2(g)
0.8 0.6 0.4 0.2
12C-formates
0 0
5
Time
10 Time (min)
15
Concentration
Multivariate analysis
Intensity
Intensity
(b)
Energy Energy
Time
Figure 20.7 Multivariate analysis to identify spectroscopically “pure” components and their concentration profiles often without the need of reference sample measurements. Source: Adapted with permission from Urakawa [5]. Copyright 2016, Elsevier.
20
371
20 Methodologies to Hunt Active Sites and Active Species
Wavenumber (cm–1) 2160 0
2120
2080
Concentration (a.u.)
2040
0.00 0.02 0.04 0.06 0.04 0.06 0.08
1 Dark1
2
Time (min)
60
3 4
UV1
5
Dark2
6
UV2
120
180
240
Pt0–CO Ti3+–CO Ptn+–CO
(a) 1
Dark1
(b)
C
O
O
O
C Pt0 O O Ti4+ Ti4+ Ti3+
Pt0
O Ti4+
Ti3+
Ti3+
Ti3+
3
O Ti3+
H2 CH4
O C H Pt0 H O O O O 4+ 4+ 4+ 4+ Ti Ti Ti Ti
Ti4+
Ti4+
CHO 2 Pt0
O C Ti4+ Ti3+
O Ti4+
O Ti4+ Ti3+
Ti3+
4 O
C
O O Pt0 O O O 4+ 4+ 4+ Ti Ti Ti Ti4+
O 4+
Ti
C Ptn+ O O Ti4+ Ti4+ Ti4+
6 O
O
O
C Ptn+ C O O O Ti4+ Ti4+ Ti4+ Ti4+
UV2
Dark2
Pt0 O
H
5
(c)
Formates Carbonates
Ti3+–CO Pt0–CO
2 O
UV1
372
O
O O C C Ptn+ O O O O Ti4+ Ti4+ Ti4+ Ti4+
Figure 20.8 (a) Evolution of surface carbonyl species on Pt/TiO2 during on/off cycles of UV light. (b) Component concentration profiles of Ti3+ –CO (pink) and Pt0 –CO (blue) obtained by MCR analysis of (a) and of those of formate (red) and carbonate (cyan) species obtained by MCR analysis of the spectra in the region between 1300 and 1700 cm−1 (not shown). (c) Proposed CO2 reduction mechanism over Pt/TiO2 . Source: Reprinted with permission from Borges Ordoño and Urakawa [36]. Copyright 2019, American Chemical Society. (See online version for color figure).
The practical advantage of multivariate analysis is its versatility. It has been used with many analytical techniques and for many reactions. It is useful whenever concentrations (signal response) vary. For example, it was used to study catalytic reaction kinetics through liquid concentration profiles determined by dip-in ATR (attenuated total reflection)-IR probe [42]. It is also useful to separate noise as separate component, thus improving S/N, and to detect small signal changes. One example is to add surface sensitivity to XAS by filtering out the signal of surface Cu oxidized by CO2 by multivariate spectral analysis [38] in a similar way as demonstrated for the case of Rh surface oxidation by MES (Figure 20.4).
References
20.5 Outlook There are increasing efforts to uncover the nature of active sites and active species with high selectivity and sensitivity. Toward the general goal of researchers in catalysis, that is, rational catalyst design, there is no doubt that more sensitive and selective analytical methods need to be developed for the purpose, and three approaches described in this chapter will gain more popularity in the coming decades. There are still further developments to be made. For example, kinetic analysis is possible for all methods described, but it is not often reported, likely due to the mathematical barrier to do so. More efforts need to be made together with further refinement of the experimental methodologies such as combining with more analytical probes, as referred to as a multiprobe or multimodal approach to gain complementary insights into the nature of active species/sites. For SSITKA, it is almost exclusively used to investigate gas-phase chemistry. Further efforts to study liquid-phase chemistry and kinetics, differentiating the detection of product isotopes, e.g. by IR detector, are awaited. Proper analysis of kinetics also implies precise understanding of chemical and concentration gradients in catalytic reactors. Spatiotemporal spectroscopic methods are expected to be combined with the methodologies described in this chapter.
References 1 Weckhuysen, B.M. (2003). Determining the active site in a catalytic process:
2
3
4 5 6
7
8
Operando spectroscopy is more than a buzzword. Phys. Chem. Chem. Phys. 5: 4351–4360. Bañares, M.A. (2005). Operando methodology: combination of in situ spectroscopy and simultaneous activity measurements under catalytic reaction conditions. Catal. Today 100: 71–77. Weckhuysen, B.M. (2009). Chemical imaging of spatial heterogeneities in catalytic solids at different length and time scales. Angew. Chem. Int. Ed. 48: 4910–4943. Urakawa, A. and Baiker, A. (2009). Space-resolved profiling relevant in heterogeneous catalysis. Top. Catal. 52: 1312–1322. Urakawa, A. (2016). Trends and advances in Operando methodology. Curr. Opin. Chem. Eng. 12: 31–36. Newton, M.A. and van Beek, W. (2010). Combining synchrotron-based X-ray techniques with vibrational spectroscopies for the in situ study of heterogeneous catalysts: a view from a bridge. Chem. Soc. Rev. 39: 4845–4863. Bentrup, U. (2010). Combining in situ characterization methods in one set-up: looking with more eyes into the intricate chemistry of the synthesis and working of heterogeneous catalysts. Chem. Soc. Rev. 39: 4718–4730. Baurecht, D. and Fringeli, U.P. (2001). Quantitative modulated excitation Fourier transform infrared spectroscopy. Rev. Sci. Instrum. 72: 3782–3792.
373
374
20 Methodologies to Hunt Active Sites and Active Species
9 Bürgi, T. and Baiker, A. (2002). In situ infrared spectroscopy of catalytic
10
11
12
13 14
15 16
17
18 19
20
21
22
23
24
solid-liquid interfaces using phase-sensitive detection: enantioselective hydrogenation of a pyrone over Pd/TiO2 . J. Phys. Chem. B 106: 10649–10658. Urakawa, A., Bürgi, T., and Baiker, A. (2008). Sensitivity enhancement and dynamic behavior analysis by modulation excitation spectroscopy: principle and application in heterogeneous catalysis. Chem. Eng. Sci. 63: 4902–4909. Urakawa, A., Bürgi, T., and Baiker, A. (2006). Kinetic analysis using square-wave stimulation in modulation excitation spectroscopy: mixing property of a flow-through PM-IRRAS cell. Chem. Phys. 324: 653. Pavelko, R.G., Choi, J.-K., Urakawa, A. et al. (2014). H2 O/D2 O exchange on SnO2 materials in the presence of CO: Operando spectroscopic and electric resistance measurements. J. Phys. Chem. C 118: 2554–2563. Fringeli, U.P. and Günthard, H.H. (1971). Modulated excitation infrared spectrophotometer. Appl. Opt. 10: 819–824. Müller, M., Buchet, R., and Fringeli, U.P. (1996). 2D-FTIR ATR spectroscopy of thermo-induced periodic secondary structural changes of poly-(l)-lysine: a cross-correlation analysis of phase-resolved temperature modulation spectra. J. Phys. Chem. 100: 10810–10825. Müller, P. and Hermans, L. (2017). Applications of modulation excitation spectroscopy in heterogeneous catalysis. Ind. Eng. Chem. Res. 56: 1123–1136. Chernyshov, D., van Beek, W., Emerich, H. et al. (2011). Kinematic diffraction on a structure with periodically varying scattering function. Acta Crystallogr., Sect. A: Found. Adv. 67: 327–335. Chernyshov, D., Dyadkin, V., van Beek, W., and Urakawa, A. (2016). Frequency analysis for modulation-enhanced powder diffraction. Acta Crystallogr., Sect. A: Found. Adv. 72: 500–506. Caliandro, R., Chernyshov, D., Emerich, H. et al. (2012). Patterson selectivity by modulation-enhanced diffraction. J. Appl. Crystallogr. 45: 458–470. van Beek, W., Emerich, H., Urakawa, A. et al. (2012). Untangling diffraction intensity: modulation enhanced diffraction on ZrO2 powder. J. Appl. Crystallogr. 45: 738–747. Ferri, D., Kumar, M.S., Wirz, R. et al. (2010). First steps in combining modulation excitation spectroscopy with synchronous dispersive EXAFS/DRIFTS/mass spectrometry for in situ time resolved study of heterogeneous catalysts. Phys. Chem. Chem. Phys. 12: 5634–5646. Ledesma, C., Yang, J., Chen, D., and Holmen, A. (2014). Recent approaches in mechanistic and kinetic studies of catalytic reactions using SSITKA technique. ACS Catal. 4: 4527–4547. Meunier, F.C. (2010). The power of quantitative kinetic studies of adsorbate reactivity by operando FTIR spectroscopy carried out at chemical potential steady-state. Catal. Today 155: 164–171. Shannon, S.L. and Goodwin, J.G. (1995). Characterization of catalytic surfaces by isotopic-transient kinetics during steady-state reaction. Chem. Rev. 95: 677–695. Stockwell, D.M., Chung, J.S., and Bennett, C.O. (1988). A transient infrared and isotopic study of methanation over Ni/Al2 O3 . J. Catal. 112: 135–144.
References
25 Eckle, S., Anfang, H.G., and Behm, R.J. (2011). Reaction intermediates and
26 27
28
29
30 31
32
33
34 35
36
37 38
39
side products in the methanation of CO and CO2 over supported Ru catalysts in H2 -rich reformate gases. J. Phys. Chem. C 115: 1361–1367. van Neer, F. and Bliek, A. (1999). The feedback mechanism in self-oscillations for CO oxidation over EUROPT-3. Chem. Eng. Sci. 54: 4483–4499. Petallidou, K.C., Kalamaras, C.M., and Efstathiou, A.M. (2014). The effect of La3+ , Ti4+ and Zr4+ dopants on the mechanism of WGS on ceria-doped supported Pt catalysts. Catal. Today 228: 183–193. Kalamaras, C.M., Americanou, S., and Efstathiou, A.M. (2011). “Redox” vs “associative formate with –OH group regeneration” WGS reaction mechanism on Pt/CeO2 : effect of platinum particle size. J. Catal. 279: 287–300. Bobin, A.S., Sadykov, V.A., Rogov, V.A. et al. (2013). Mechanism of CH4 dry reforming on nanocrystalline doped ceria–zirconia with supported Pt, Ru, Ni, and Ni–Ru. Top. Catal. 56: 958–968. Verykios, X.E. (2003). Mechanistic aspects of the reaction of CO2 reforming of methane over Rh/Al2 O3 catalyst. Appl. Catal., A 255: 101–111. Goguet, A., Meunier, F.C., Tibiletti, D. et al. (2004). Spectrokinetic investigation of reverse water-gas-shift reaction intermediates over a Pt/CeO2 catalyst. J. Phys. Chem. B 108: 20240–20246. Meunier, F.C., Tibiletti, D., Goguet, A. et al. (2007). On the complexity of the water-gas shift reaction mechanism over a Pt/CeO2 catalyst: effect of the temperature on the reactivity of formate surface species studied by operando DRIFT during isotopic transient at chemical steady-state. Catal. Today 126: 143–147. de Juan, A. and Tauler, R. (2006). Multivariate curve resolution (MCR) from 2000: progress in concepts and applications. Crit. Rev. Anal. Chem. 36: 163–176. Jaumot, J., de Juan, A., and Tauler, R. (2015). MCR-ALS GUI 2.0: new features and applications. Chemom. Intell. Lab. Syst. 140: 1–12. Corral-Pérez, J.J., Billings, A., Stoian, D., and Urakawa, A. (2019). Continuous hydrogenation of carbon dioxide to formic acid and methyl formate by a molecular iridium complex stably heterogenized on a covalent triazine framework. ChemCatChem 11: 4725–4730. Borges Ordoño, M. and Urakawa, A. (2019). Active surface species ruling product selectivity in photocatalytic CO2 reduction over Pt- or Co-promoted TiO2 . J. Phys. Chem. C 123: 4140–4147. Roso, S., Degler, D., Llobet, E. et al. (2017). Temperature-dependent NO2 sensing mechanisms over indium oxide. ACS Sens. 2: 1272–1277. Hyakutake, T., van Beek, W., and Urakawa, A. (2016). Unravelling the nature, evolution and spatial gradients of active species and active sites in the catalyst bed of unpromoted and K/Ba-promoted Cu/Al2 O3 during CO2 capture-reduction. J. Mater. Chem. A 4: 6878–6885. Voronov, A., Urakawa, A., van Beek, W. et al. (2014). Multivariate curve resolution applied to in situ X-ray absorption spectroscopy data: an efficient tool for data processing and analysis. Anal. Chim. Acta 840: 20–27.
375
376
20 Methodologies to Hunt Active Sites and Active Species
40 Lesage, C., Devers, E., Legens, C. et al. (2019). High pressure cell for edge
jumping X-ray absorption spectroscopy: applications to industrial liquid sulfidation of hydrotreatment catalysts. Catal. Today 336: 63–73. 41 Vantelon, D., Davranche, M., Marsac, R. et al. (2019). Iron speciation in iron-organic matter nanoaggregates: a kinetic approach coupling Quick-EXAFS and MCR-ALS chemometrics. Environ. Sci. Nano 6: 2641–2651. 42 Fakhrnasova, D., Chimentão, R.J., Medina, F., and Urakawa, A. (2015). Rational and statistical approaches in enhancing the yield of ethylene carbonate in urea transesterification with ethylene glycol over metal oxides. ACS Catal. 5: 6284–6295.
377
21 Ultrafast Spectroscopic Techniques in Photocatalysis Chun Hong Mak 1,2 , Rugeng Liu 1,2 , and Hsien-Yi Hsu 1,2 1 City University of Hong Kong, School of Energy and Environment & Department of Materials Science and Engineering, G5703, 5/F, Yeung Kin Man Academic Building (YEUNG), Tat Chee Avenue, Kowloon Tong, Hong Kong 999077, PR China 2 Shenzhen Research Institute of City University of Hong Kong, 8 Yuexing 1st road, Shenzhen Hi-Tech Industrial Park, Nanshan District, Shenzhen 518057, PR China
E–k diagram (Scheme 21.1a, b) illustrates the characteristics of a semiconductor material and the relationship between the energy and momentum of any accessible quantum mechanical states for electrons inside the semiconductor. When the electron in the semiconductor absorbs the photon with sufficient energy (>Eg ), the electrons can overcome the potential barrier to further promote the electron from the valence band to the conduction band. After that, the electrons may release the excess energy through radiative and non-radiative pathways, which are similar with the excited-state relaxation processes, as listed in Jablonski energy diagram. Ultrafast spectroscopic method targets the investigation of dynamic processes inside the semiconductor and provides crucial scientific value, such as intermolecular charge transfer and exciton diffusion length. Ultrafast spectroscopic techniques have been utilizing for the exploration of fundamental mechanisms in photocatalytic reaction. In this chapter, we will illustrate some of the common ultrafast spectroscopic techniques, including transient absorption spectroscopy (TAS), time-resolved photoluminescence (TRPL) spectroscopy, and time-resolved microwave conductivity (TRMC) spectroscopy for the dynamic study of photocatalysts.
21.1 Transient Absorption Spectroscopy 21.1.1
Introduction
Transient spectroscopy was invented by Norrish and Porter in 1950. They developed an instrument consisting of flash photolysis tube, high-pressure gas-filled flash discharge tube, and a high-resolution absorption spectrometer to study the radical species and their dynamics on the sub-microsecond time range. Their pioneering work opened a door for the investigation in the time domain with microsecond accuracy. The most successful part of this technique allows Heterogeneous Catalysts: Advanced Design, Characterization and Applications, First Edition. Edited by Wey Yang Teoh, Atsushi Urakawa, Yun Hau Ng, and Patrick Sit. © 2021 WILEY-VCH GmbH. Published 2021 by WILEY-VCH GmbH.
378
21 Ultrafast Spectroscopic Techniques in Photocatalysis Excited singlet states
Excitation (absorption) 10–15 s
5 4 3 2 S2 1 0
Vibrational energy states (10–13–10–11 s) Internal conversion (10–9–10–7 s)
Conduction band
E
5 4 3
S1 21
Intersystem crossing
0
Quenching
Fluorescence (10–9–10–7 s)
Non-radiative relaxation
5 4 3 S0 21 0
Ground state
(a)
Ec
Delayed fluorescence 5 4 3 2 1 0
p Electrons Eg
Excited triplet state (T1)
Holes
EV
Non-radiative relaxation (triplet) Phosphorescence (10–3–102 s)
Valence band (b)
Scheme 21.1 (a) Jablonski energy diagram, illustrating the excitation and the relaxation pathways with the correlated time scale. (b) Basic E–k diagram, x-axis is momentum, p, or wavenumber, k (since p = ℏk), y-axis is energy, E g is band gap energy, E c is energy level of conduction band, and E v is energy level of valence band. Source: Adapted from Norrish et al. 1967 [1]. (See online version for color figure).
direct observation for short-lived species, leading them to receive the Nobel Prize in Chemistry in 1967 [1–4]. Until the accomplishment of the first ruby laser developed by Inoue in 1960 [5], this achievement brought an intense pulsed light source to reality. After the great efforts on the laser technologies, the pulse width was tremendously reduced to nanoseconds, picoseconds, and sub-picosecond pulse laser within six years. Until 1991, the successful breakthrough on the duration of pulses minimized to femtosecond was introduced by Shank and coworkers [6] During the development of advanced laser technologies, Ahmed Zewail used the laser as an excitation source to study the dynamics in atomic scale (i.e. chemical bonding), and his finding was awarded the Nobel Prize in Chemistry in 1999. His works provided a powerful tool to understand the transition states inside the chemical reactions by utilizing femtosecond transient spectroscopy [7]. Transient absorption spectroscopy (TAS), also known as pump–probe spectroscopy, measures time-correlated changes in the absorbance or transmittance in the sample. To acquire the absorption spectrum, pulsed laser passes through the beam splitter and creates the pump pulse and the probe pulse for exciting the sample and obtaining the absorption spectrum of the excited species, respectively. The delay time of the “probe” pulse is defined by the additional length passing through the retroreflector that is constructed on a motorized translation stage. Then, the sample is excited by the pump pulse, and the probe pulse will arrive to the sample by a defined time from nanosecond to femtosecond. Afterward, part of the laser pulse is absorbed by the sample. Then, the remaining irradiation, representing the absorption spectrum of the sample at defined delay time after excitation, passes through an iris and is collected by the detector. To investigate the absorption spectrum, Beer–Lambert law is applied: A = log10
I0 I
(21.1)
21.1 Transient Absorption Spectroscopy
We can illustrate the intensity of the excitation pulse labeled as I e . After I e excited the sample, I e can be expressed as Ie = I0 ⋅ 10−Ae
(21.2)
where Ae is the absorption of the excited sample and I 0 is the intensity of the incident pulse. Also, we need to obtain the intensity of the non-excitation pulse labeled as I ne , and I ne can be expressed as Ine = I0 ⋅ 10−Ane
(21.3)
where Ane is the absorption of the non-excited sample and I 0 is the intensity of the incident pulse. Thus, the equation regarding to change in absorption (ΔA) can be obtained by dividing (21.2) by (21.1) and expressed in logarithm equation: ΔA ≡ Ae − Ane = log10
Ine Ie
(21.4)
The change in absorption, ΔA, is the target signal in the transient absorption spectrometer. The data of I ne and I e can be obtained from the periodic blocking of the pump pulses by an optical chopper, and the snapshots of non-excited and excited periods can be taken by the photodiodes or array detectors. The typical schematic layout of the pump–probe experiment is shown in Figure 21.1. At the beginning, the pump pulse is passed through the sample to generate the excited state. The relatively weak probe pulse tracks will arrive to the same area of the sample after a defined delayed time to track the absorption caused by pump pulse. In general, a beam splitter is employed to “split” short laser pulse ( NiW > NiFe > NiCr, which were prepared on steel substrate by electrodeposition [80]. NiMo possesses not only the most HER active sites but also excellent durability, and hence its binary or ternary alloys are among the best HER catalysts. Such a merit originates from the hypo-hyper-d-electronic coupling effects induced by two unpaired and half-filled d-electron orbitals of Ni and Mo, respectively. Accordingly, electronic adjustment by a third element in NiMo ternary alloys has been followed recently. Huang and coworkers [81] identified the essential role of Zn in boosting NiMo intrinsic electrocatalytic behavior. They identified that 2 atom% as the optimum Zn content facilitates the formation of high-valent Ni and Mo, thus featuring a noble-metal-like HER activity for NiMoZn alloy. Ternary NiMoTi alloy synthesized by mechanical alloying and heat treatment also exhibited a positive influence of Ti doping on the alkaline HER efficiency of NiMo alloy [82]. Synergistic effect between NiMo intermetallic phase and Ni(Ti) solid solution was mostly promoted in Ni50 Mo40 Ti10 system. Moreover,
30.3 Electrocatalyst Materials in Liquid Electrolyte Water Splitting
hybridizing with conductive carbon materials as in the case of recently reported alloys, e.g. CoNi@NC [83] and CoFe@N-rGO [84], is another perspective to get advanced HER catalysts. 30.3.2.2
Non-precious Metal Composites
In addition to intermetallic compounds, nonmetal species such as Se, S, P, N, C, and B are potential additives to noble metal–free-based composites. Like OER electrocatalysts, transition metal chalcogenides, pnictides, carbides, and borides have been extensively evaluated for their electrocatalytic HER behavior. In this section, recent improvements on nonmetal composites are discussed, and the corresponding HER performances are listed in Table 30.2. Nanostructured MoS2 has been one of the well-studied transition metal chalcogenides and is considered a promising alternative to noble Pt for electrochemical reduction of water. Bulk MoS2 possesses a graphite-like hexagonally packed layered structure (hexagonal close-packed [HCP] crystal system), and crystallinity brings a variety of surface sites. Distinct polytypes exist in bulk MoS2 crystallites where electron–hole mobility happens with much faster kinetics along the basal planes (1T-MoS2 polytype) compared with the perpendicular transfer between sheets (2H-MoS2 polytype). Much of the surface of bulk MoS2 is composed of these Table 30.2 HER performances of earth-abundant nonmetal electrocatalysts.
Catalyst
Substrate
Overpotential @10 mA/cm2 (mV vs. RHE)
MoS2 /reduced graphene oxide (RGO)
GCa)
∼150
MoS2
Fluorinedoped tin oxide (FTO)
MoS2
Tafel slope (mV/dec)
Electrolyte
References
41
0.5 M H2 SO4
[85]
260
50
0.5 M H2 SO4
[86]
Au
170
60
pH = 0.2 H2 SO4
[87]
CoS2
GC
145
51
0.5 M H2 SO4
[88]
Mo2 C
Carbon
240b)
54
1M KOH
[89]
Mo2 C
GC
151
59
1M KOH
[90]
Ni2 P
Ti foil
130
46
0.5 M H2 SO4
[91]
Co-P
Ti foil
85b)
50
0.5 M H2 SO4
[92]
FeP
Ti plate
55
38
0.5 M H2 SO4
[93]
a) GC denotes as glassy carbon. b) Overpotential obtained at 20 mA/cm2 .
549
550
30 Electrochemical Water Splitting
Geometric surface area Electrochemically active surface area
HER active edge site
Insert basal plane site
Figure 30.10 Two-dimensional representation of crystalline 2H polytype MoS2 catalyst electrochemically active surface area and projected geometric surface area. The edge sites, indicated by green dots, are active for HER. Basal plane sites, indicated by blue squares, are not active for HER. Source: Reprinted with permission from Benck et al. [94]. Copyright 2014, American Chemical Society. (See online version for color figure).
basal plate sites that are favored thermodynamically but inert catalytically [94]. In contrast, edge sites in each layer of MoS2 octahedral coordination show high activity for HER by having optimal hydrogen binding energies close to zero that could be observed in noble metals such as Pt. Considerably large number of active edge sites could be brought into effect through nanostructuring [95]. Hence, unlike bulk structure, nanostructured MoS2 is ideal for HER, especially since it offers enhanced surface roughness. Generally, active edge sites are more available in a roughened surface, and by taking ECSA as an intrinsic activity measurement into account, higher electrocatalytic activity could be projected for nanocrystalline MoS2 per unit of geometric surface area. Distinctive active edge and inert basal sites in MoS2 are illustrated in Figure 30.10. A considerable amount of research has been conducted on exposing more active sites in MoS2 , and the results proposed several strategies for excellent Pt-like HER activity, namely, (i) design of active site-rich nanostructures with considerable porosity such as reticular nanosheets [86]; (ii) doping of active heteroatoms such as Co, Ni, V, and Li [96]; (iii) functionalization with reactive anions containing O, Se, and N [97]; (iv) addition of adjunct conductive structures such as graphene [98] and CNTs [85]; as well as (v) amorphization [99]. Mo2 C is among the best metal carbides with preferable electronic structure and a Pt-like electrocatalytic performance. The cumulative effects of Mo activity, C conductivity, combinatorial durability, and favorable structural phase were reported by Vrubel and Hu for the substantial HER performance of 𝛽-Mo2 C nanoparticles in both acidic and alkaline media [89]. Shortly after that, multiple phases of MoC (𝛼-MoC1−x , 𝜂-MoC, 𝛾-MoC, and 𝛽-Mo2 C) were thoroughly examined, and a decreasing trend of 𝛽-Mo2 C > 𝛾-MoC > 𝜂-MoC > 𝛼-MoC1−x was obtained for their HER ability [100], evincing the significance of structural optimization for nanoscale Mo2 C. Morphology also has a great influence, which was demonstrated by Mo2 C nanowires prepared via simple heat treatment under inert conditions by Liao et al. [101]. Having a porous, aggregation-free structure
30.4 Conclusions and Outlook
along with large specific surface area, the as-prepared Mo2 C nanowires yielded small overpotentials at high current densities about 8 and 50 times higher than those of bulk and commercial Mo2 C catalysts, respectively. MOF-template strategy was examined by Lou and coworkers [90] for Mo2 C catalyst, and the results verified its electrocatalytic HER capability with small overpotentials in both acidic and basic solutions. In pnictide group, one of the most promising HER catalysts is Ni2 P. Initial theoretical studies by Rodriguez’s group in 2005 suggested an ensemble effect where P sites participate in HER and promote a moderate bonding between intermediate species and products with the catalyst surface, placing Ni2 P among the best practical HER catalysts. Inspired by their work, Popczun et al. [91] confirmed Ni2 P exceptional HER activity experimentally in acid electrolyte. However, the behavior was not satisfactory in basic conditions because of the formed Ni phase from Ni2 P. Herein, structural changes such as nanostructuring, heteroatom doping, anion regulation, and affixing conductive nanocarbons are the plausible options to boosting Ni2 P alkaline HER efficiency and stability, and numerous recent research reports signify their efficacy. 30.3.2.3
Metal-Free Electrocatalysts
Nonmetallic compounds have attracted interests recently toward economical H2 production. Certain features seem necessary, namely, large surface area, excellent charge transfer, facile mass transport, and chemical stability in corrosive electrolytes that potentially were supplied by nanoscale carbonaceous compounds such as CNTs and graphene as support materials. However, further improvement of the electronic structure of these nanocarbon materials can be achieved by heteroatom doping, leading to a dramatic increase in their intrinsic HER activity. Accordingly, metal-free carbon-based electrocatalysts emerged as alternative candidates such as O-, S-, N-, P-, and B- (dual) doped graphene [102], N-doped activated carbon [103], B-doped multiwalled carbon nanotubes (MWCNTs) [104], and N-doped hexagonal carbon [105]. Using theoretical computations, Qiao and coworkers reported that the valence orbital energy level of heteroatom-doped graphene could be originated from the N-/P-co-activated adjacent carbon atoms in graphene matrix providing higher activity compared with single-atom doped graphene. Carbon nitride (C3 N4 ) also possesses a 2D structure like graphene and is capable of showing a low HER onset overpotential in pure phase but with low current density due to poor electroconductivity. Hence, combining C3 N4 with conductive heteroatom-doped graphene could be an effective resolution. C3 N4 -N graphene introduced by Zheng et al. [106] shows an inherent chemical–electronic synergy suitable for optimum proton adsorption and reduction kinetics, which makes it comparable to transition metal catalysts.
30.4 Conclusions and Outlook In this chapter, the main technologies and promising electrocatalysts used in electrochemical water splitting to date were summarized. Practical considerations
551
552
30 Electrochemical Water Splitting
necessitate ongoing development in cell design and different cell components to reach maximum efficiency. ALKWE is a mature and the simplest unit currently utilized in the industrial sector. PEMWE benefits from the compact design and improved kinetics, but considering the capital costs due to membrane and electrode materials, it is at near-term commercialization. SOE water electrolyzer depicts the highest efficiency thanks to high-temperature electrolysis; however, different issues in cell components have kept it at laboratory scale. Although considerable achievements have been made based on non-precious metal and nonmetal compounds for both OER and HER procedures, developed systems still face several issues for the use in large-scale applications: (i) Prolonged mechanical durability and electrochemical stability often fall behind the state-of-the-art electrode advances though some have demonstrated comparable activity with commercial electrodes. While the lack of or poor connecting interface between the current collector and catalyst is a continuing problem for the electrodes prepared via binder-assisted electrocatalysts coating, 3D conductive scaffolds are not completely robust for long-term use. Hence, engineering advancement in electrode configuration is highly desired. (ii) Insufficient catalytic activity is usually accompanied with poor stability in corrosive acidic solutions. Current electrochemical water splitting technology utilizes alkaline and acidic electrolytes for OER and HER, respectively. But this entails complexity and increased cost in electrode preparation and cell design. Therefore, both efficient catalytic and electrolytic evaluations in various electrolytes should be expanded. (iii) Industrial electrolyzers often require large current densities of 1 or 2 A/cm2 for both anodic and cathodic reactions, while the majority of reported electrocatalysts either were studied below 100 mA/cm2 or unable to deliver such amounts. Besides, in common three-electrode systems, evolved O2 and H2 may diffuse to the opposite area and react with each other to form water, which occurs more sharply at high currents. In any case, mass transport effect is crucial for the electrodes with high current densities, and such systems should be evaluated in novel electrolytic cells with functional separators and more advanced techniques beyond cyclic voltammetry and Tafel analysis. Accordingly, the urge for more active and stable materials, bifunctional electrocatalysts, along with pragmatic approaches in catalyst and cell design, is envisaged for future electrochemical water splitting.
References Chen, J.G., Crooks, R.M., Seefeldt, L.C. et al. (2018). Science 360: eaar6611. Liu, Y., Liang, X., Gu, L. et al. (2018). Nat. Commun. 9: 2609. Schlapbach, L. and Zuttel, A. (2001). Nature 414: 353. Cook, T.R., Dogutan, D.K., Reece, S.Y. et al. (2010). Chem. Rev. 110: 6474. Roger, I., Shipman, M.A., and Symes, M.D. (2017). Nat. Rev. Chem. 1: 0003. Vanýsek, P. (2012). Electrochemical series. In: CRC Handbook of Chemistry and Physics, 93e (ed. M.W. Haynes), 5–80. Chemical Rubber Company. ISBN: 9781439880494. 7 Li, X., Hao, X., Abudula, A., and Guan, G. (2016). J. Mater. Chem. A 4: 11973. 1 2 3 4 5 6
References
8 Barbir, F. (2005). Sol. Energy 78: 661. 9 Sapountzi, F.M., Gracia, J.M., Weststrate, C.J. et al. (2017). Prog. Energy
Combust. Sci. 58: 1. 10 Zeng, K. and Zhang, D. (2010). Prog. Energy Combust. Sci. 36: 307. 11 Carmo, M., Fritz, D.L., Mergel, J., and Stolten, D. (2013). Int. J. Hydrogen
Energy 38: 4901. 12 Hauch, A., Ebbesen, S.D., Jensen, S.H., and Mogensen, M. (2008). J. Mater.
Chem. 18: 2331. 13 Acar, C. and Dincer, I. (2014). Int. J. Hydrogen Energy 39: 1. 14 Kim, S.W., Kim, H., Yoon, K.J. et al. (2015). J. Power Sources 280: 630. 15 Mamoon Rashid, M.D., Mohammed, K., Mesfer, A. et al. (2015). Int. J. Eng.
Adv. Technol. 4: 80. 16 Manabe, A., Kashiwase, M., Hashimoto, T. et al. (2013). Electrochim. Acta
100: 249. 17 Pool, D.H., Stewart, M.P., O’Hagan, M. et al. (2012). Proc. Natl. Acad. Sci. 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33
34 35 36 37 38
U.S.A. 109: 15634. Todd, D., Schwager, M., and Merida, W. (2014). J. Power Sources 269: 424. Ganley, J.C. (2009). Int. J. Hydrogen Energy 34: 3604. Parrondo, J., Arges, C.G., Niedzwiecki, N. et al. (2014). RSC Adv. 4: 9875. Phillips, R. and Dunnill, C.W. (2016). RSC Adv. 6: 100643. Peng, Y., Jiang, K., Hill, W. et al. (2019). ACS Appl. Mater. Interfaces 11: 3971–3977. https://doi.org/10.1021/acsami.8b19251. You, B. and Sun, Y. (2018). Acc. Chem. Res. 51: 1571. Han, G., Jin, Y.H., Burgess, R.A. et al. (2017). J. Am. Chem. Soc. 139: 15584. Pan, J., Chen, C., Li, Y. et al. (2014). Energy Environ. Sci. 4: 354. Varcoe, J.R., Atanassov, P., Dekel, D.R. et al. (2014). Energy Environ. Sci. 7: 3135. Siracusano, S., Baglio, V., Di Blasi, A. et al. (2010). Int. J. Hydrogen Energy 35: 5558. Ebbesen, S.D., Jensen, S.H., Hauch, A., and Mogensen, M.B. (2014). Chem. Rev. 114: 10697. Fabbri, E., Bi, L., Pergolesi, D., and Traversa, E. (2011). Energy Environ. Sci. 4: 4984. Li, S.S. and Xie, K. (2013). J. Electrochem. Soc. 160: F224. Bi, L., Boulfrad, S., and Traversa, E. (2014). Chem. Soc. Rev. 43: 8255. Udagawa, J., Aquiar, P., and Brandon, N.P. (2007). J. Power Sources 166: 127. van de Loosdrecht, J. and Niemantsverdriet, J.W. (2013). Synthesis gas to hydrogen, methanol and synthetic fuels. In: Chemical Energy Storage (ed. R. Schoegl), 443–457. Berlin: De Gruyter. Rostrup-Nielsen, J. and Christiansen, L.J. (2011). Concepts in Syngas Preparation, Catalytic Science Series. London: Imperial College Press. Trotochaud, L., Ranney, J.K., Williams, K.N., and Boettcher, S.W. (2012). J. Am. Chem. Soc. 134: 17253. Lu, X. and Zhao, C. (2015). Nat. Commun. 6: 6616. Zhou, H., Yu, F., Zhu, Q. et al. (2018). Energy Environ. Sci. 11: 2858. Long, X., Li, J., Xiao, S. et al. (2014). Angew. Chem. Int. Ed. 53: 7584.
553
554
30 Electrochemical Water Splitting
39 Friebel, D., Louie, M.W., Bajdich, M. et al. (2015). J. Am. Chem. Soc. 137:
1305. 40 Gerken, J.B., Shaner, S.E., Masse, R.C. et al. (2014). Energy Environ. Sci. 7: 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78
2376. Liu, Y., Li, J., Li, F. et al. (2016). J. Mater. Chem. A 4: 4472. Zhang, B., Zheng, X., Voznyy, O. et al. (2016). Science 352: 333. Liu, L., Zhang, H., Mu, Y. et al. (2016). J. Power Sources 327: 599. Xiao, C., Lu, X., and Zhao, C. (2014). Chem. Commun. 50: 10122. Song, F. and Hu, X. (2014). J. Am. Chem. Soc. 136: 16481. Zhang, P., Li, L., Nordlund, D. et al. (2018). Nat. Commun. 9: 381. Stern, L.A., Feng, L., Song, F., and Hu, X. (2015). Energy Environ. Sci. 8: 2347. Zhou, W., Wu, X.J., Cao, X. et al. (2013). Energy Environ. Sci. 6: 2921. Xu, K., Chen, P., Li, X. et al. (2015). J. Am. Chem. Soc. 137: 4119. Zhao, S., Wang, Y., Dong, J. et al. (2016). Nat. Energy 1: 16184. Ma, T.Y., Dai, S., Jaroniec, M., and Qiao, S.Z. (2014). J. Am. Chem. Soc. 136: 13925. Haber, J.A., Cai, Y., Jung, S. et al. (2014). Energy Environ. Sci. 7: 682. Liu, P.F., Yang, S., Zheng, L.R. et al. (2017). Chem. Sci. 8: 3484. Suntivich, J., May, K.J., Gasteiger, H.A. et al. (2011). Science 334: 1383. Chen, G.F., Ma, T.Y., Liu, Z.Q. et al. (2016). Adv. Funct. Mater. 26: 3314. Liu, Y., Cheng, H., Lyu, M. et al. (2014). J. Am. Chem. Soc. 136: 15670. Zou, X., Wu, Y., Liu, Y. et al. (2018). Chem 4: 1139. Yang, J., Zhu, G., Liu, Y. et al. (2016). Adv. Funct. Mater. 26: 4712. Zou, X., Liu, Y., Li, G.D. et al. (2017). Adv. Mater. 29: 1700404. Dong, B., Zhao, X., Han, G.Q. et al. (2016). J. Mater. Chem. A 4: 13499. Zhu, H., Jiang, R., Chen, X. et al. (2017). Sci. Bull. 62: 1373. Xu, X., Song, F., and Hu, X. (2016). Nat. Commun. 7: 12324. Xiong, D., Wang, X., Li, W., and Liu, L. (2016). Chem. Commun. 52: 8711. Han, A., Chen, H., Sun, Z. et al. (2015). Chem. Commun. 51: 11626. Duan, J., Chen, S., Vasileff, A., and Qiao, S.Z. (2016). ACS Nano 10: 8738. (a) Li, B.Q., Zhang, S.Y., Tang, C. et al. (2017). Small 13: 1700610. (b) Hou, Y., Qiu, M., Zhang, T. et al. (2017). Adv. Mater. 29: 1701589. Fan, H., Yu, H., Zhang, Y. et al. (2017). Angew. Chem. Int. Ed. 56: 12566. Masa, J., Weide, P., Peeters, D. et al. (2016). Adv. Energy Mater. 6: 1502313. Masa, J., Sinev, I., Mistry, H. et al. (2017). Adv. Energy Mater. 7: 1700381. Liu, G., He, D., Yao, R. et al. (2018). Nano Res. 11: 1664. Miner, E.M. and Dinc˘a, M. (2016). Nat. Energy 1: 16186. Duan, J., Chen, S., and Zhao, C. (2017). Nat. Commun. 8: 15341. Senthil Raja, D., Chuah, X.-F., and Lu, S.-Y. (2018). Adv. Energy Mater. 8: 1801065. You, B., Jiang, N., Sheng, M. et al. (2015). Chem. Mater. 27: 7636. Zou, X. and Zhang, Y. (2015). Chem. Soc. Rev. 44: 5148. Miles, M.H. and Thomason, M.A. (1976). J. Electrochem. Soc. 123: 1459. Zou, X., Huang, X., Goswami, A. et al. (2014). Angew. Chem. Int. Ed. 53: 4372. Qiu, H.J., Ito, Y., Cong, W. et al. (2015). Angew. Chem. Int. Ed. 54: 14031.
References
79 Soares, D.M., Teschke, O., and Torriani, I. (1992). J. Electrochem. Soc. 139:
98. 80 Raj, I.A. (1993). J. Mater. Sci. 28: 4375. 81 Wang, X., Su, R., Aslan, H. et al. (2015). Nano Energy 12: 9. 82 Panek, J., Kubisztal, J., and Bierska-Piech, B. (2014). Surf. Interface Anal. 46:
716. 83 Deng, J., Ren, P., Deng, D., and Bao, X. (2015). Angew. Chem. Int. Ed. 54:
2100. 84 Wu, Z., Li, P., Qin, Q. et al. (2018). Carbon 139: 35. 85 Li, Y., Wang, H., Xie, L. et al. (2011). J. Am. Chem. Soc. 133: 7296. 86 Kibsgaard, J., Chen, Z., Reinecke, B.N., and Jaramillo, T.F. (2012). Nat. Mater.
11: 963. 87 Li, H., Tsai, C., Koh, A.L. et al. (2015). Nat. Mater. 15: 48. 88 Faber, M.S., Dziedzic, R., Lukowski, M.A. et al. (2014). J. Am. Chem. Soc.
136: 10053. 89 Vrubel, H. and Hu, X. (2012). Angew. Chem. Int. Ed. 124: 12875. 90 Wu, H.B., Xia, B.Y., Yu, L. et al. (2015). Nat. Commun. 6: 6512. 91 Popczun, E.J., McKone, J.R., Read, C.G. et al. (2013). J. Am. Chem. Soc. 135:
9267. 92 Popczun, E.J., Read, C.G., Roske, C.W. et al. (2014). Angew. Chem. Int. Ed.
53: 5427. 93 Jiang, P., Liu, Q., Liang, Y. et al. (2014). Angew. Chem. Int. Ed. 53: 12855. 94 Benck, J.D., Hellstern, T.R., Kibsgaard, J. et al. (2014). ACS Catal. 4: 3957. 95 Hinnemann, B., Moses, P.G., Bonde, J. et al. (2005). J. Am. Chem. Soc. 127:
5308. 96 Wang, H., Lu, Z., Xu, S. et al. (2013). Proc. Natl. Acad. Sci. U.S.A. 110:
19701. 97 Zhou, W., Hou, D., Sang, Y. et al. (2014). J. Mater. Chem. A 2: 11358. 98 Firmiano, E.G.S., Cordeiro, M.A.L., Rabelo, A.C. et al. (2012). Chem. Com-
mun. 48: 7687. 99 Ge, X., Chen, L., Zhang, L. et al. (2014). Adv. Mater. 26: 3100. 100 Wan, C., Regmi, Y.N., and Leonard, B.M. (2014). Angew. Chem. Int. Ed. 126: 101 102 103 104 105 106
6525. Liao, L., Wang, S., Xiao, J. et al. (2014). Energy Environ. Sci. 7: 387. Jiao, Y., Zheng, Y., Davey, K., and Qiao, S.Z. (2016). Nat. Energy 1: 16130. Zhang, B., Wen, Z., Ci, S. et al. (2014). RSC Adv. 4: 49161. Cheng, Y., Tian, Y., Fan, X. et al. (2014). Electrochim. Acta 143: 291. Liu, Y., Yu, H., Quan, X. et al. (2014). Sci. Rep. 4: 6843. Zheng, Y., Jiao, Y., Zhu, Y. et al. (2014). Nat. Commun. 5: 3783.
555
557
31 New Visible-Light-Responsive Photocatalysts for Water Splitting Based on Mixed Anions Kazuhiko Maeda Tokyo Institute of Technology, Department of Chemistry, School of Science, 2-12-1-NE-2 Ookayama, Meguro-ku, Tokyo 152-8550, Japan
31.1 Introduction As discussed in Chapter 11, band engineering of a semiconductor is very important for the development of a heterogeneous photocatalyst. Scaife pointed out that the flat band potential (EFB ) of various metal oxides with d0 or d10 electronic configuration follows an empirical relation: EFB (NHE) ≈ 2.94 − Eg where Eg is the band gap of an oxide semiconductor [1]. This means that if a metal oxide has a band gap smaller than 2.94 eV, the EFB that is nearly equal to the conduction band minimum (CBM) lies at a potential more positive than the water reduction potential (i.e. 0 V vs. normal hydrogen electrode [NHE] at pH 0). A typical example is WO3 , which has a band gap of 2.7 eV and does not have the ability to reduce proton to give H2 . Mixed anion compounds that contain more than two anionic species in the same phase are good candidates of band-engineered photocatalysts, in particular, to produce a visible-light-responsive narrow-gap photocatalyst for overall water splitting [2]. The key concept is to utilize p orbitals of less electronegative anion than oxygen, thereby forming a new valence band at more negative positions while keeping the conduction band potential, as shown in Figure 31.1. Oxynitrides (oxide nitrides) are in general stable, nontoxic materials and can be readily obtained by nitriding a corresponding metal oxide powder. Domen et al. have developed oxynitrides and nitrides as photocatalysts for water splitting under visible light [3–5]. Figure 31.2a shows ultraviolet (UV)–visible diffuse reflectance spectra of Ta2 O5 , TaON, and Ta3 N5 . The absorption edge is red shifted from Ta2 O5 to Ta3 N5 via TaON. The band structures of Ta2 O5 , TaON, and Ta3 N5 determined by UV photoelectron spectroscopy and electrochemical analysis are schematically drawn in Figure 31.2b [6], which clearly shows that the valence band maximum (VBM) is shifted negatively in the order Ta2 O5 < TaON < Ta3 N5 without significant change in the CBM.
Heterogeneous Catalysts: Advanced Design, Characterization and Applications, First Edition. Edited by Wey Yang Teoh, Atsushi Urakawa, Yun Hau Ng, and Patrick Sit. © 2021 WILEY-VCH GmbH. Published 2021 by WILEY-VCH GmbH.
31 New Visible-Light-Responsive Photocatalysts for Water Splitting Based on Mixed Anions
Potential vs. NHE at pH 0 0
Conduction band (metal d orbitals)
Conduction band (metal d orbitals)
Wide band gap (>3 eV)
Narrow band gap ( 400 nm), both TaON and Ta3 N5 served as stable photocatalyst for water reduction/oxidation in the presence of suitable electron donor/acceptor, meaning that both have a potential to split pure water under visible light [7, 8]. Since the initial reports on TaON and Ta3 N5 , a variety of (oxy)nitrides have been reported as candidates for photocatalytic water splitting under visible light. GaN–ZnO and ZnGeN2 –ZnO solid solutions with d10 electronic configurations are active photocatalysts to split pure water into H2 and O2 under visible
31.2 New Doped Rutile TiO2 Photocatalysts for Efficient Water Oxidation
light without any sacrificial reagents [9, 10]. In particular, GaN–ZnO solid solutions have been demonstrated to be the first reproducible example of overall water splitting among semiconductors having a band gap smaller than 3 eV. Significant improvement of water splitting activity was accomplished upon surface modification of GaN–ZnO with RuO2 or Rh–Cr mixed oxide nanoparticles that served as H2 evolution cocatalysts [9, 11], with stable operation even as long as three months [12]. Followed by these d10 -based oxynitrides, visible-light overall water splitting using d0 oxynitrides has recently been achieved. The key to accomplish overall water splitting is proper surface modification. For example, ZrO2 -grafted TaON became an active photocatalyst for overall water splitting under visible light when modified with core/shell-structured RuOx /Cr2 O3 nanoparticles and colloidal IrO2 , which work as water reduction and oxidation sites, respectively [13]. A similar surface modification strategy was applied for LaTaO2 N–LaMg2/3 Ta1/3 O3 solid solutions [14] and CaTaO2 N [15] to achieve overall water splitting. One of the drawbacks of oxynitride photocatalysts is their instability against oxidation reactions by photogenerated holes. Even with benchmarking GaN–ZnO photocatalyst, degradation of photocatalytic activity was observed, because the photogenerated holes oxidize the N3− anion in the material [12]. The stability issue is a common challenge in not only oxynitrides but also other non-oxide-type photocatalysts such as sulfides and nitrides. While visible-light-driven overall water splitting using a single semiconductor photocatalyst still remains a difficult subject, a two-step photoexcitation scheme, so-called Z-scheme shown in Figure 31.3, using two different semiconductors allows one to apply a variety of mixed anion materials, which are not limited to oxynitrides, to one (or both) side of the system in the presence of a suitable shuttle redox mediator (e.g. IO3 − /I− and Fe3+ /Fe2+ ) or with the aid of intimate interfacial contact between the two semiconductors [16–23].In the Z-scheme system, a wider range of visible light can be utilized because the energy to drive each reaction is reduced. It is also possible to apply a semiconductor material that satisfies either water oxidation or reduction potential to one side of the system. Recent technological advancement to assemble both H2 and O2 evolution photocatalysts based on metal oxides in a conductive sheet enables efficient Z-scheme water splitting, with a solar energy conversion efficiency of 1.1% [24]. This photocatalyst sheet strategy could be applicable to oxynitrides, although the efficiency still remains low [25]. Thus, significant progress has been made in photocatalytic water splitting community in the past 20 years. Meanwhile, some of mixed anion photocatalysts, most of which have been developed for overall water splitting, have been applied to CO2 fixation scheme, with the aid of a functional metal complex [26–29]. In this chapter, the latest development of new mixed anion photocatalysts is described not only for water splitting but also CO2 reduction.
31.2 New Doped Rutile TiO2 Photocatalysts for Efficient Water Oxidation Rutile TiO2 having a band gap of 3.0 eV is a good photocatalyst for water oxidation under UV irradiation ( Eg
hv > Eg
e–
CB (Ox/red) +0.82 (O2/H2O)
H2O O2
Eg e–
e–
Red Ox
e–
VB O2 evolution photocatalyst
e– H2
Eg
h+
(a)
e–
CB
H+
VB h+
H2 evolution photocatalyst
(b)
Figure 31.3 Basic principle of overall water splitting using two different semiconductor photocatalysts through two-step photoexcitation. Source: Reproduced with permission from Maeda and Domen [4]. Copyright 2010, American Chemical Society.
donor that is more susceptible to oxidation than water, rutile TiO2 is capable of selectively oxidizing water into O2 . While this property is very useful as a water oxidation photocatalyst for the construction of a Z-scheme water splitting system (see Figure 31.3b), the band gap of rutile TiO2 is too wide to effectively absorb visible light. Nitrogen-doped TiO2 is a visible-light-responsive photocatalyst that has been mostly used for the degradation of harmful organic compounds [31]. However, the use of anion-doped TiO2 for light energy conversion scheme such as water splitting and CO2 fixation is very rare. The key to produce an active doped TiO2 photocatalyst is to compensate the charge imbalance that is usually caused by aliovalent ion doping into TiO2 (e.g. O2− /N3− exchange). Such charge imbalance in the crystal lattice leads to the generation of vacancies to compensate the charge balance, which potentially works as recombination centers of electrons and holes. For example, Cr-doped rutile TiO2 is able to absorb an entire range of visible light, but does not work as a photocatalyst for individual H2 /O2 evolution in the presence of sacrificial regents under visible light due to the formation of Cr6+ [32]. By contrast, rutile TiO2 codoped with Cr and Sb shows photocatalytic activities, because the generation of Cr6+ is suppressed by the Sb5+ codopant. The codoping strategy can be applied to anion-doped TiO2 , as exemplified by N- and F-codoped anatase TiO2 [33, 34]. Recently, the codoping strategy was applied to cation–anion combinations to synthesize a new doped rutile TiO2 photocatalyst [35]. Ta- and N-codoped rutile TiO2 (TiO2 :Ta,N) powder was synthesized by nitriding Ta-doped rutile TiO2 that had been in prior synthesized by a microwave-assisted solvothermal method. X-ray diffraction and elemental analyses indicated that TiO2 :Ta,N (1.0 mol% Ta and 420 nm) and under AM 1.5G simulated sunlight. The solar-to-hydrogen energy conversion efficiency achieved in this system was 0.02%. Although this value is low, it is almost double compared with that using TaON (Figure 31.6), which is one of the most active oxynitride photocatalysts for visible-light water oxidation [3], even though the visible-light absorption of TiO2 :Ta,N is obviously inferior to that of TaON (see Figure 31.4a).
31.2 New Doped Rutile TiO2 Photocatalysts for Efficient Water Oxidation
(a)
[100] (b)
2 nm
2 nm c
Ti N/O/F
b
a
(c)
[101]
2 nm Ti N/O/F
[100]
(d)
[101]
2 nm ~ 50 pmb c
a
Figure 31.7 Results of STEM observations for TiO2 :N,F. (a, b) High-angle annular dark-field (HAADF) and annular bright-field (ABF) images taken along the [100] and (c, d) [101] direction. As the signal intensity in HAADF imaging is approximately proportional to Z 2 (where Z represents the atomic number), Ti atomic columns in the HAADF images can be seen as bright dots. Source: Reproduced with permission from Miyoshi et al. [38]. Copyright 2018, Royal Society of Chemistry. (See online version for color figure).
Another interesting example of doped rutile TiO2 is N/F-codoped rutile TiO2 (TiO2 :N,F) [38]. This material could be synthesized by heating a mixture of rutile TiO2 and (NH4 )2 TiF6 at 773 K under a flow of ammonia gas. Scanning transmission electron microscopy (STEM) observations indicated the production of the rutile-type structure of TiO2 :N,F with good crystallinity and no disorder of anion arrangement (Figure 31.7). Visible-light absorption capability of TiO2 :N,F was dependent on the concentration of nitrogen in the material, as determined by the relative concentration of (NH4 )2 TiF6 in the starting mixture. As shown in Figure 31.8, the visible-light absorption band was more pronounced with increasing the relative concentration of (NH4 )2 TiF6 . Photocatalytic activity of TiO2 :N,F for visible-light water oxidation was also influenced by the (NH4 )2 TiF6 concentration. It is noted that rutile TiO2 nitrided at the same temperature without (NH4 )2 TiF6 showed negligible activity, due
563
31 New Visible-Light-Responsive Photocatalysts for Water Splitting Based on Mixed Anions
3.0 Kubelka–Munk (au)
564
C = 50 C = 30 C = 20 C = 15 C = 10 C = 0 (873 K)a R-TiO2
2.5 2.0 1.5
C = 50
C = 15
1.0 0.5 0.0
C=0
300
400 500 600 Wavelength (nm)
700
Figure 31.8 UV–visible diffuse reflectance spectra of TiO2 :N,F obtained by nitriding the mixture of TiO2 and (NH4 )2 TiF6 with various concentrations of (NH4 )2 TiF6 at 773 K for one hour. a Nitrided at 873 K. Here “C” indicates the molar concentration of (NH4 )2 TiF6 in the mixture. Source: Reproduced with permission from Miyoshi et al. [38]. Copyright 2018, Royal Society of Chemistry. (See online version for color figure).
primarily to its poor visible-light absorption. The highest activity for water oxidation into O2 from aqueous AgNO3 solution was obtained with the sample prepared at the (NH4 )2 TiF6 /TiO2 ratio of 15/85. Similar to the TiO2 :Ta,N case, transient absorption spectroscopy revealed that appropriate N/F codoping reduces the density of mid-gap states working as deep traps of photogenerated electrons and increases the number of free electrons compared with only N-doped rutile TiO2 . Thus, photocatalytic activity of the TiO2 :N,F for water oxidation could be enhanced by improving visible-light absorption capability through the N/F codoping while suppressing the density of deep trap sites. The optimized TiO2 :N,F, further modified with RuO2 cocatalyst, was applied as the water oxidation component in Z-scheme water splitting in combination with a H2 evolution photocatalyst Ru/SrTiO3 :Rh in the presence of [Co(bpy)3 ]3+/2+ (bpy = 2,2′ -bipyridine) as a shuttle redox mediator under visible light. It is known that most of mixed anion photocatalysts, more or less, suffer from oxidative degradation by photogenerated holes, which results in low stability [3]. Interestingly, the optimized RuO2 /TiO2 :N,F provided stable H2 and O2 evolution during the 35 hours of reaction, as shown in Figure 31.9.
31.3 Unprecedented Narrow-Gap Oxyfluoride As discussed above, oxynitrides are promising materials as visible-light-driven photocatalysts. This is because N-2p orbitals can form a valence band that has a more negative potential, as compared with O-2p orbitals. In this regard, oxyfluorides (oxide fluorides) seem unsuitable because of the higher electronegativity of fluorine. Very recently, the oxyfluoride Pb2 Ti2 O5.4 F1.2 that has an anion-ordered A2 B2 X6 X′ 0.5 -type pyrochlore structure (Figure 31.10a) [40] was found to possess an unprecedented small band gap (c. 2.4 eV), absorbing visible light of up to
Amount of evolved gases (μmol)
31.3 Unprecedented Narrow-Gap Oxyfluoride Evacuation
Evacuation
40 H2 30 H2
O2
20
2+
H+
[Co(bpy)3] 3+
10
[Co(bpy)3]
0 0
20 30 10 Reaction time (h)
H 2O O2 RuO2/rutile-TiO2:N,F
Ru/SrTiO3:Rh
Figure 31.9 Time course of H2 and O2 evolution from mixtures of RuO2 /TiO2 :N,F (50 mg) and Ru/SrTiO3 :Rh (25 mg) dispersed in an aqueous solution (120 ml) containing tris(2,2′ -bipyridyl)cobalt(II) sulfate (0.5 mM) under visible-light irradiation (𝜆 > 420 nm). Source: Reproduced with permission from Miyoshi et al. [38]. Copyright 2018, Royal Society of Chemistry.
pb O O
O/F
(a)
2.0
(b)
Conduction band (CB)
–2
1.5
–1
1.0 0.5 0.0 300
Ti
Potential (V vs. Ag/AgNO3)
Vacancy
Kubelka–Munk (au)
2.5
H+ /H2 (–0.98) Band gap (c. 2.4 eV)
0
CO2/HCOOH (–1.17) CO2/CO (–1.09)
O2/H2O(+0.25)
+1
400 500 600 Wavelength (nm)
Valence band (VB)
700
(c)
Figure 31.10 (a) Crystal structure of Pb2 Ti2 O5.4 F1.2 . The annotations indicate elements, sites, and Wyckoff positions. In this space group, the X and X′ sites are divided into two sites, which are denoted as X1/X2 and X′ 1/X′ 2, respectively. (b) UV−visible diffuse reflectance spectrum of Pb2 Ti2 O5.4 F1.2 . The inset shows a picture of Pb2 Ti2 O5.4 F1.2 . (c) Schematic band structure diagram of Pb2 Ti2 O5.4 F1.2 , along with some redox potentials. Source: Reproduced with permission from Kuriki et al. [39]. Copyright 2018, American Chemical Society. (See online version for color figure).
510 nm (see Figure 31.10b) [39]. Electrochemical impedance spectroscopy and UV–visible diffuse reflectance spectroscopy indicated that Pb2 Ti2 O5.4 F1.2 has band-edge potentials suitable for water reduction/oxidation and CO2 reduction, as illustrated in Figure 31.10c. Actually, Pb2 Ti2 O5.4 F1.2 worked as a stable photocatalyst for visible-light-driven H2 evolution and CO2 reduction when modified with a suitable promoter unit such as Pt nanoparticles (for H2 evolution) or a Ru(II) binuclear complex (for CO2 reduction), respectively, in the presence of triethanolamine (TEOA) as an electron donor. In addition, water oxidation to form O2 was possible using AgNO3 as an electron acceptor with the aid of RuO2 cocatalyst. Density functional theory (DFT) calculations were performed to investigate the origin of the visible light response of Pb2 Ti2 O5.4 F1.2 . Figure 31.11a shows the total and partial density of state (DOS) of the material; the CBM of Pb2 Ti2 O5.4 F1.2 consists mainly of Ti-3d orbitals with some hybridization of Pb-6p (in particular,
565
31 New Visible-Light-Responsive Photocatalysts for Water Splitting Based on Mixed Anions
–20
(a)
–8
Antibonding
Total O/F (X1) – p orbital O (X2) – p orbital O (X′1) – p orbital Ti – d orbital Pb – d orbital Pb – s orbital Pb – p orbital –7.5
–7 –6.5 Energy (eV)
–6
Pb-6p Energy
Density of states (au)
Density of states (au)
566
Antibonding
Bonding
O-2p
–5.5
Pb-6s
–15
–10
–5
0
Energy (eV)
5
10
15
Bonding
20
(b)
Figure 31.11 (a) Total and partial DOS of Pb2 Ti2 O5.4 F1.2 . In Pb2 Ti2 O5.4 F1.2 , the X′ 1 (4b) and X′ 2 (4d) sites are occupied only by oxide anion, with the occupation factors of 0.9 and 0.16, respectively. DFT calculations were performed with a simple model of X′ 1/X′ 2 = 1.00/0.00, assuming that the X′ 1 site is fully occupied by oxide anion, while the X′ 2 site is vacant. The inset shows an enlarged view of the lower valence band region. (b) Orbital interaction between Pb-6s and O-2p in Pb2 Ti2 O5.4 F1.2 along with an energy level diagram based on the RLP model. Source: Reproduced with permission from Kuriki et al. [39]. Copyright 2018, American Chemical Society. (See online version for color figure).
higher-energy side), whereas the VBM is formed by O-2p orbitals with certain contributions of Pb-6s orbitals. In lower-energy side (from −8 to −5 eV) of the valence band, it is clear that Pb-6s orbitals are hybridized with O-2p orbitals (Figure 31.11a inset). It was also found that the O-2p contribution to Pb-6s band (per atom) is largest in the X′ 1 site, which is fivefold those in X1 and X2 sites, on the basis of integral calculation for each partial DOS. The valence band character can be interpreted in terms of the revised lone pair (RLP) model, proposed by Walsh et al. [41]. In this model (schematically shown in Figure 31.11b), the antibonding orbitals formed by Pb-6s and O-2p orbitals are stabilized through the interaction with the empty Pb-6p orbitals, accounting for the elevated O-2p orbitals in Pb2 Ti2 O5.4 F1.2 . It is noted that some Pb(II)-based oxides such as PbO and PbTiO3 are known to exert the RLP effect [41]. However, the band gaps of such known Pb(II)-containing oxides are wider than that of Pb2 Ti2 O5.4 F1.2 , indicating that the Pb-6s/O-2p hybridization in Pb2 Ti2 O5.4 F1.2 is much stronger. This is at least in part due to the unique coordination environment around the A-site in the pyrochlore structure, where in the ideal A2 B2 O7 case there are two short A—O bonds and six long A—O bonds. The length of the shortest Pb—O bond in Pb2 Ti2 O5.4 F1.2 is 2.248 Å [40], which is much shorter than that in perovskite PbTiO3 (2.510 Å). This difference in the local coordination environment could qualitatively explain the narrower band gap of Pb2 Ti2 O5.4 F1.2 (c. 2.4 eV) than PbTiO3 (c. 2.8 eV). A follow-up study on another oxyfluoride of Pb2 Ti4 O9 F2 supported the idea [42]. Thus, the unprecedented visible light response of Pb2 Ti2 O5.4 F1.2 is concluded to originate from strong interaction between Pb-6s and O-2p orbitals, which is caused by a short Pb—O bond in the pyrochlore lattice due to the fluorine substitution. The low coordination number of the A-site of pyrochlore (or defect-pyrochlore) structure is preferable for strengthening the electronic interaction between M-6s (M = Pb and Bi) and O-2p, thereby producing more prominent visible-light absorption.
31.4 Conclusion and Future Perspective
31.4 Conclusion and Future Perspective Mixed anion materials such as oxynitrides are promising as photocatalysts for water splitting. In this chapter, several examples of recently reported mixed anion materials were discussed. Rutile TiO2 codoped with Ta/N or N/F served as an effective O2 evolution photocatalyst in visible-light Z-scheme water splitting systems, in combination with SrTiO3 :Rh as a H2 evolution photocatalyst in the presence of a suitable shuttle redox mediator. In such doped TiO2 materials, the suppression of oxygen vacancies, which are generated due to aliovalent cation and/or anion doping (e.g. Ti4+ /Rh3+ or O2− /N3− exchange) and work as deep traps of photogenerated charge carriers, is essential to obtain measurable photocatalytic activity for water oxidation, as clearly demonstrated by transient absorption spectroscopy. Ta/N-codoped rutile TiO2 exhibited higher performance as a photocatalyst for O2 evolution in solar-driven Z-scheme water splitting with SrTiO3 :Rh than TaON, which is one of the most active non-oxide photocatalysts for visible-light water oxidation. Very recently, an electrode material consisting of the rutile TiO2 :Ta,N has been shown to work as a stable water oxidation photoanode [43]. With post-modification of the rutile TiO2 :Ta,N electrode by conductive TiOx layer and a RuOx cocatalyst, the optimized photoanode produced O2 upon simulated sunlight over five hours of operation with a faradaic efficiency of 94% and no sign of deactivation. This is distinct from the ordinary oxynitride-based photoanodes that usually suffer from self-oxidative degradation, even with the aid of a water oxidation cocatalyst. The TiO2 :Ta,N photoanode could be coupled to not only H2 evolution but also CO2 reduction using a molecular photocathode. Another important finding in heterogeneous photocatalysis using mixed anion materials is that the pyrochlore oxyfluoride Pb2 Ti2 O5.4 F1.2 functions as a stable visible-light-responsive photocatalyst. Even though it is an oxyfluoride that had been believed to be unsuitable for visible-light photocatalysis, crystal engineering through the use of mixed anion would enable the development of oxyfluorides as new class of visible-light photocatalysts for solar-to-fuel energy conversion. Approximately 20 years have passed since the research on mixed anion materials as water splitting photocatalysts started. While a satisfactory photocatalyst has not been developed to date, significant progress has been made in the recent 10 years, in collaboration with synthetic materials chemistry and spectroscopy. Recent research activities on mixed anion photocatalysts have also extended to construct a CO2 reduction system with functional metal complexes, which is another important reaction in artificial photosynthesis [26–29]. This was a result of interdisciplinary interaction between different research fields (i.e. semiconductor photocatalysis and coordination chemistry). A tentative goal of photocatalytic overall water splitting to produce H2 is to develop a stable system that works with a solar-to-hydrogen energy conversion efficiency of ∼10% [4]. The state of the art in water splitting photocatalyst is at most a few percent [24]. The major obstacle to achieve the goal is still the lack of an efficient photocatalytic material that is able to absorb longer wavelength photons and to show high quantum efficiency. Nevertheless, the material’s constraint may be relaxed using a photovoltaic device and a photoelectrochemical
567
568
31 New Visible-Light-Responsive Photocatalysts for Water Splitting Based on Mixed Anions
cell [44], although such device combination appears unsuitable for large-scale application. Anyway, continuous efforts are still necessary to realizing artificial photosynthesis in the future.
References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34
Scaife, D.E. (1980). Sol. Energy 25: 41. Kageyama, H., Hayashi, K., Maeda, K. et al. (2018). Nat. Commun. 9: 772. Maeda, K. and Domen, K. (2007). J. Phys. Chem. C 111: 7851. Maeda, K. and Domen, K. (2010). J. Phys. Chem. Lett. 1: 2655. Maeda, K. and Domen, K. (2016). Bull. Chem. Soc. Jpn. 89: 627. Chun, W., Ishikawa, A., Fujisawa, H. et al. (2003). J. Phys. Chem. B 107: 1798. Hitoki, G., Takata, T., Kondo, J.N. et al. (2002). Chem. Commun.: 1698. Hitoki, G., Ishikawa, A., Takata, T. et al. (2002). Chem. Lett. 7: 736. Maeda, K., Takata, T., Hara, M. et al. (2005). J. Am. Chem. Soc. 127: 8286. Lee, Y., Terashima, H., Shimodaira, Y. et al. (2007). J. Phys. Chem. C 111: 1042. Maeda, K., Teramura, K., Lu, D. et al. (2006). Nature 440: 295. Ohno, T., Bai, L., Hisatomi, T. et al. (2012). J. Am. Chem. Soc. 134: 8254. Maeda, K., Lu, D., and Domen, K. (2013). Chem. Eur. J. 19: 4986. Pan, C., Takata, T., Nakabayashi, M. et al. (2015). Angew. Chem. Int. Ed. 54: 2955. Xu, J., Pan, C., Takata, T., and Domen, K. (2015). Chem. Commun. 51: 7191. Abe, R., Takata, T., Sugihara, H., and Domen, K. (2005). Chem. Commun.: 3829. Higashi, M., Abe, R., Teramura, K. et al. (2008). Chem. Phys. Lett. 452: 120. Higashi, M., Abe, R., Ishikawa, A. et al. (2008). Chem. Lett. 37: 138. Maeda, K., Higashi, M., Lu, D. et al. (2010). J. Am. Chem. Soc. 132: 5858. Maeda, K., Abe, R., and Domen, K. (2011). J. Phys. Chem. C 115: 3057. Maeda, K., Lu, D., and Domen, K. (2013). ACS Catal. 3: 1026. Ma, G., Chen, S., Kuang, Y. et al. (2016). J. Phys. Chem. Lett. 7: 3892. Fujito, H., Kunioku, H., Kato, D. et al. (2016). J. Am. Chem. Soc. 138: 2082. Wang, Q., Hisatomi, T., Jia, Q. et al. (2016). Nat. Mater. 15: 611. Pan, Z., Hisatomi, T., Wang, Q. et al. (2016). ACS Catal. 6: 7188. Sekizawa, K., Maeda, K., Domen, K. et al. (2013). J. Am. Chem. Soc. 135: 4596. Yoshitomi, F., Sekizawa, K., Maeda, K., and Ishitani, O. (2015). ACS Appl. Mater. Interfaces 7: 13092. Muraoka, K., Kumagai, H., Eguchi, M. et al. (2016). Chem. Commun. 52: 7886. Nakada, A., Kuriki, R., Sekizawa, K. et al. (2018). ACS Catal. 8: 9744. Miyoshi, A., Nishioka, S., and Maeda, K. (2018). Chem. Eur. J. 24: 18204. Asahi, R., Morikawa, T., Ohwaki, T. et al. (2001). Science 293: 269. Kato, H. and Kudo, A. (2002). J. Phys. Chem. B 106: 5029. Nukumizu, K., Nunoshige, J., Takata, T. et al. (2003). Chem. Lett. 32: 196. Maeda, K., Shimodaira, Y., Lee, B. et al. (2007). J. Phys. Chem. C 111: 18264.
References
35 Nakada, A., Nishioka, S., Vequizo, J.J.M. et al. (2017). J. Mater. Chem. A 5:
11710. 36 Yamakata, A., Ham, Y., Kawaguchi, M. et al. (2015). J. Photochem. Photobiol.
A 313: 168. 37 Yamakata, A., Vequizo, J.J.M., and Matsunaga, H. (2015). J. Phys. Chem. C
119: 24538. 38 Miyoshi, A., Vequizo, J.J.M., Nishioka, S. et al. (2018). Sustainable Energy
Fuels 2: 2025. 39 Kuriki, R., Ichibha, T., Hongo, K. et al. (2018). J. Am. Chem. Soc. 140: 6648. 40 Oka, K., Hojo, H., Azuma, M., and Oh-ishi, K. (2016). Chem. Mater. 28: 5554. 41 Walsh, A., Payne, D.J., Egdell, R.G., and Watson, G.W. (2011). Chem. Soc. Rev.
40: 4455. 42 Wakayama, H., Utimula, K., Ichibha, T. et al. (2018). J. Phys. Chem. C 122:
26506. 43 Nakada, A., Uchiyama, T., Kawakami, N. et al. (2019). ChemPhotoChem 3: 37. 44 Grätzel, M. (2001). Nature 414: 338.
569
571
32 Electrocatalysts in Polymer Electrolyte Membrane Fuel Cells Stephen M. Lyth 1,2 and Albert Mufundirwa 3 1 Kyushu University, Platform of Inter/Transdisciplinary Energy Research (Q-PIT), 744 Motooka, Nishi-ku, Fukuoka 819-0395, Japan 2 Kyushu University, International Institute for Carbon-Neutral Energy Research, 744 Motooka, Nishi-ku, Fukuoka 819-0395, Japan 3 Kyushu University, Department of Hydrogen Energy Systems, Faculty of Engineering, 744 Motooka, Nishi-ku, Fukuoka 819-0395, Japan
32.1 Introduction Hydrogen reacts exothermically with oxygen resulting in explosive generation of water, heat, and light (Figure 32.1a). This looks impressive but is dangerous. In 1936 a hydrogen-filled zeppelin airship known as the Hindenburg ignited and exploded, resulting in 36 deaths (Figure 32.1b). In 2011 a tsunami in Japan triggered a failure of the cooling systems at Fukushima Daiichi nuclear power plant. Water reacted at hot zirconium surfaces in the reactor generating hydrogen, which exploded. This destroyed the containment buildings and helped spread radioactive materials into the environment. Despite these isolated disasters, hydrogen can be relatively safe compared with liquid fuels like gasoline. Hydrogen fires generally burn quickly in a localized area and are over in a few seconds (Figure 32.1c). Meanwhile, gasoline fires burn for a long time and tend to be more destructive (Figure 32.1d) [1]. Another advantage of using hydrogen fuel is that it can be generated from electrolysis of water using renewable energy. This avoids CO2 emissions and will help mitigate climate change. As such, hydrogen is being seriously considered as an alternative fuel for heating and powering homes, vehicles, and factories. But how can we control the potentially destructive force of hydrogen, and how can we utilize it safely? One of the simplest solutions is to physically separate the hydrogen and oxygen using a membrane and use electrochemistry to regulate the reaction. Fuel cells are devices that generate electricity from electrochemical reactions between a fuel and an oxidant at the anode and cathode, respectively. These are physically separated by an electrolyte membrane. The principle is similar to a battery, but where batteries are self-contained, in fuel cells the reactants can be continuously supplied. Here we focus specifically on polymer electrolyte fuel cells (PEFCs) utilizing hydrogen as a fuel. The operation principle of a PEFC is
Heterogeneous Catalysts: Advanced Design, Characterization and Applications, First Edition. Edited by Wey Yang Teoh, Atsushi Urakawa, Yun Hau Ng, and Patrick Sit. © 2021 WILEY-VCH GmbH. Published 2021 by WILEY-VCH GmbH.
572
32 Electrocatalysts in Polymer Electrolyte Membrane Fuel Cells
(a)
(b)
(c)
(d)
Figure 32.1 (a) An exploding hydrogen-filled balloon. Source: Maxim Bilovitskiy (cropped) Licensed under CC BY SA-4.0. (b) The Hindenburg disaster. Sam Shere (cropped)/public domain. Fires in (c) a hydrogen fuel cell vehicle (FCV) and (d) a conventional gasoline vehicle. Source: US Department of Energy 2011 [1].
(a)
ox hy
Oxygen molecule
Hydrogen Electron molecule
Water molecule
ox
hy
ox
hy
ox
hy
ox
hy
Hydrogen ions (b)
(a)
(c)
Anode Catalyst Polymer electrolyte Cathode membrane
(b)
Figure 32.2 (a) Schematic diagram of a modern PEFC. (b) The first fuel cell, designed by William Grove in 1842. Source: (a) Reprinted with permission from Schlapbach and Züttel [2]. Copyright 2001, Springer Nature; (b) Grove 1842 [3]. Reprinted with permission of Taylor & Francis Ltd.
represented in Figure 32.2a [2]. At the anode, the hydrogen oxidation reaction (HOR) occurs on the surface of a platinum electrocatalyst, liberating protons and electrons (Eq. (32.1)). At the cathode, the oxygen reduction reaction (ORR) occurs: oxygen from air combines with protons and electrons and is reduced to form water (Eq. (32.2)). These two “half-reactions” generate a potential difference across the cell, driving protons through the electrolyte and electrons around the external circuit. Completing the circuit sustains the reaction indefinitely until the fuel or oxidant supply stops. The overall reaction is obtained by combining both half-reactions and canceling out the protons and electrons (Eq. (32.3)).
32.1 Introduction
This balanced reaction is identical to the explosive combination of H2 and O2 . However, in a PEFC most of the energy released can be captured as useful electricity and heat: 2H2 → 4H+ + 4e−
(32.1)
O2 + 4H + 4e → 2H2 O
(32.2)
2H2 + O2 → 2H2 O
(32.3)
+
−
Hydrogen fuel cells may seem like a futuristic concept, but they have been around for almost 180 years. In 1842, Grove discovered the fuel cell when he reversed a water electrolysis experiment (Figure 32.2b) [3]. In his pioneering experiment the electrodes were platinum foil, doubling as the electrocatalyst. Sulfuric acid was used as the electrolyte. Hydrogen and oxygen were trapped in tubes containing the anode and cathode, respectively. Grove linked several cells together, generating sufficient voltage to electrolyze water. He had built the world’s first fuel cell, naming it the “gaseous voltaic battery.” However, Grove’s fuel cell was inefficient and generated only a tiny current. One reason was the large distance between the electrodes, resulting in high ionic resistance (i.e. ohmic losses). Minimizing ohmic losses is key to improving performance and can be achieved by decreasing the distance between electrodes or by increasing the conductivity of the electrolyte. Another reason for the poor performance was unoptimized electrode design. For the HOR or ORR reactions to take place, three conditions must be met at the catalyst surface: (i) H2 or O2 molecules must be present, (ii) electron transport must be possible, and (iii) proton transport must be possible. The place where these three things can happen is the triple phase boundary [4, 5]. In Grove’s cell, this is just the contact point between the gas, the acid, and the Pt foil, shown schematically by a dotted line in Figure 32.3a. In modern PEFC systems, a complex electrode design is employed to maximize the triple phase boundary. Pt nanoparticles are used in place of bulk platinum because of their larger surface area. These are supported on carbon black, which provides electronic transport and porosity for the gas supply. A very thin layer of polymer ionomer is coated on the surface to provide ionic transport (Figure 32.3b). The triple phase boundary of a modern electrocatalyst is shown H
Pt
+
H2
O2
O2 e– Carbon black
Triple phase boundary
(a)
Triple phase boundary
Platinum H+ e–
Inactive Pt H2SO4
Carbon black
Active Pt
(b)
Ionomer
(c)
Figure 32.3 (a) Schematic of the triple phase boundary in Grove’s fuel cell. (b) Structure of the electrocatalyst layer in a modern PEFC. Source: Image courtesy of David Rivera, Copyright 2018. (c) The triple phase boundary in a modern electrocatalyst layer.
573
32 Electrocatalysts in Polymer Electrolyte Membrane Fuel Cells
in Figure 32.3c. In practice, the triple phase boundary is approximated using the electrochemical surface area (ECSA), normalized to the mass of catalyst. In modern systems, this is generally ∼100 m2 /g, whereas in Grove’s cell it would have been orders of magnitude lower. One more modern advance is the solid polymer electrolyte, replacing the aqueous electrolyte. This is usually Nafion, a sulfonated fluoropolymer with high proton conductivity (∼0.1 S/cm). This is made into thin membranes (e.g. 20 μm thick) to decrease the ohmic losses. In addition, Nafion is mixed into the electrocatalyst layer to provide an interpenetrating electrolyte, increasing the triple phase boundary.
32.2 Platinum Electrocatalysts Buying platinum jewelry is expensive, and using it in a fuel cell vehicle (FCV) is no different. In a typical PEFC it is estimated that the Pt electrocatalyst and its application make up 21–45% of the stack cost, depending on the production scale (Figure 32.4a) [6]. For a FCV such as the Toyota Mirai, this translates to ∼US$3700 (estimated from Figure 32.4b) [7]. Intensive research over the decades has sought to improve the activity and thus reduce the cost through advances in nanoparticle synthesis, such as alloying with cheaper metals [8], or creation of complex nanoparticle geometries [9]. Modern commercial PEFCs use Pt loadings of 99%). Considering the CO2 advantages, Yan et al. synthesized Pd particles supported on SiO2 and Al2 O3 by using liquid CO2 , showing high activity and durability for LA hydrogenation [116, 117]. On the other hand, Du et al. studied several metals (e.g. Pd, Pt, Ru, Au) supported on TiO2 , SiO2 , ZrO2 , and C, showing the Au/ZrO2 catalyst to have the best performance in terms of high conversion with high yields of ∼97% [118, 119]. Likewise, Son et al. confirmed that the supported Au nanoparticle catalyst was efficient for the synthesis of GVL (yGVL = 90%) after screening a series of Ru/C, Ru/SBA, Au/ZrC, and Au/ZrO2 catalysts [120]. Due to the high cost of noble metal catalysts, other inexpensive metal-based catalysts have been proposed in the literature. Upare et al. attained high GVL yields (yGVL = 96%) produced by LA hydrogenation on Al2 O3 -supported Ni–Cu bimetallic catalysts at 265 ∘ C and 25 bar H2 [121]. Haan et al. patented a Ni catalyst that provided a 71% yield of GVL [122]. On the other hand, bimetallic nanoparticle catalysts tend to display better performances for liquid-phase hydrogenation of LA to GVL. Shimizu et al. evaluated a series of base metal (Ni, Co, Cu, and Fe) and metal oxides (Mo, V, and W oxides) co-loaded carbon (C) and Ni-loaded metal oxides for the hydrogenation of LA to GVL. The best catalytic behavior was found in Ni–MoOx /C (99% yield) [123]. Obregón et al. attained high GVL yields for the hydrogenolysis of LA by using the Al2 O3 -supported Ni and Ni–Cu catalysts after two hours at 50 ∘ C and 6.5 MPa [124]. Recently, Hengne and Rode reported the use of nanocomposites of
33.3 Biofuel Production from Lignocellulosic Biomass
Cu–ZrO2 and Cu–Al2 O3 as catalysts for the LA hydrogenation to GVL, obtaining GVL selectivity of 90% [125]. Another alternative route for GVL production is the catalytic transfer of hydrogen (CTH) of LA using alcohol as hydrogen donor through the Meerwein–Ponndorf–Verley (MPV) reaction. Chia and Dumesic reported the use of different metal oxides such as ZrO2 , γ-Al2 O3 , MgO/Al2 O3 , MgO/ZrO2 , and CeZrOx to produce GVL by CTH [126]. The highest GVL yield (yGVL = 92%) was found in the presence of ZrO2 as catalyst by using 2-butanol at 150 ∘ C. A large number of studies about GVL production from LA have been reported in the literature. However, the transformation from carbohydrates or even biomass would be more interesting economically, which makes the use of bifunctional catalysts able to hydrogenate and containing acid or basic properties necessary. The GVL production from fructose is considerably easier than from other carbohydrates. Zhou et al. reported the one-pot conversion of fructose into GVL, with EL being the intermediate [127]. A polymer of divinylbenzene highly cross-linked with an acidic ionic liquid was employed as heterogeneous acid catalyst for the alcoholysis of fructose into EL in a mixed solvent of ethanol and DMSO, attaining full conversion and EL yield of 65% after 12 hours at 150 ∘ C. Then, EL was extracted by ethyl acetate, and direct hydrogenation was carried out at 130 ∘ C for six hours under 3 MPa H2 over TiO2 -supported Co nanoparticles, achieving a GVL yield of 55%. However, the GVL yield decreased from third cycle due to Co leaching. Son et al. also carried out the one-pot conversion of fructose into GVL, but they employed FA as the hydrogen source [120]. It has been previously mentioned that first they found Au-supported catalysts were the most active for LA hydrogenation to GVL. Then, they realized the one-pot reaction from fructose in the presence of FA in which FA acted as both acid catalyst for fructose dehydration to LA and hydrogen source for LA hydrogenation to GVL, attaining a final GVL yield of 48%. Yang et al. carried out GVL production from fructose [128]. Firstly, they realized alcoholysis of fructose in ethanol by using Amberlyst-15 obtaining EL after 20 hours at 120 ∘ C, which was used for GVL production (yGVL = 50%) using 2-propanol as hydrogen source and Raney Ni as catalyst. The use of glucose as feedstock is preferred to fructose because it is more abundant and inexpensive. One of the main disadvantages for this process is the deactivation of the heterogeneous catalysts by poisoning due to the presence of acids and humins. To solve this problem, alternatives such as the use of an organic cosolvent have been proposed in the literature. Alonso et al. carried out the GVL synthesis from corn stover using 0.5 M H2 SO4 aqueous solution and alkylphenol to extract LA [129]. Then, the LA hydrogenation to GVL took place over Ru–Sn catalyst. Molinari et al. used 2-methyltetrahydrofuran (MTHF) to extract LA from aqueous feeds obtained by the H2 SO4 -catalyzed hydrolysis of lignocellulosic materials [130]. The LA extracted by MTHF was successfully converted into GVL in the presence of Raney Ni, with an LA conversion >99% and a GVL yield of 96%. The use of MTHF is preferred because of its low cost, sustainability, and availability. Wettstein et al. designed a biphasic reaction system for GVL production from cellulose using an aqueous-phase solution containing HCl saturated with NaCl and GVL as solvent to extract LA (Figure 33.7) [131]. This biphasic system provided high yields of LA and FA (e.g. 70%). The LA recovered in
609
610
33 Conversion of Lignocellulosic Biomass to Biofuels OH H
HO
H
OH
O
HO O
O O
OH OH
n
Cellulose
O
H3C
O O
O H3C
GVL
H
H3C O
H2 H3C OH Metal catalyst
LA
O
O
Fuels GVL
O
O HO
O OH
O HO
Green solvent
H
H3C
OH O
HCI
NaCI
H2O
Chemicals
Figure 33.7 Schematic representation of biphasic reaction system for GVL production from cellulose and main GVL applications. Source: Wettstein et al. 1991 [131]. Reproduced with permission of Royal Society of Chemistry. (See online version for color figure).
GVL was subsequently converted into GVL over a carbon-supported Ru–Sn catalyst. These research groups improved this catalytic system using Amberlyst-70 as solid acid catalyst instead of HCl for conversion of cellulose into LA [132]. They affirmed that the solvent had a positive effect on LA production since unconverted cellulose and degradation products were detected after reaction using pure water. These products were not observed for GVL/water systems. High LA yields were obtained from cellulose and real biomass (69% and 54%, respectively) after 16 hours and 160 ∘ C in the presence of Amberlyst-70. Finally, LA was converted into GVL by using RuSn4 /C as catalyst, which was stable for 60 hours of time onstream. Although homogeneous solutions of mineral acids were very active for the hydrolysis of lignocelluloses or carbohydrates into LA and FA, they are difficult to recycle and lead to environmental problems, which makes the use of heterogeneous catalysts advisable. Ding et al. reported a one-pot two-step method for the synthesis of GVL from cellulose [133]: They used mesoporous Al–NbOPO4 and commercial 5% Ru/C as solid catalysts for LA and GVL production, respectively. Without LA separation, it attained a GVL yield of 57%. Al–NbOPO4 accelerated LA formation due to its high acid concentration and the coexistence of Lewis and Brønsted acid sites. Putro et al. reported a similar procedure for the production of GVL from sugarcane bagasse over Pt/TiO2 and acid-activated bentonite as catalysts, obtaining an interesting performance with respect to LA conversion (100%) and GVL selectivity (95%) [134].
33.4 Outlook and Conclusions Lignocellulosic biomass conversion to biofuels is a promising alternative to traditional fossil fuels since lignocellulose is a low-cost and sustainable feedstock and that is widely available. It has been demonstrated that carbohydrate-containing
References
biomass can be converted to platform molecules such as furfural and HMF, in which further transformation to other chemicals and fuels is feasible. In this chapter, recent advances in procedures to obtain different furanic-derived fuels including EMF, EL, DMF, MF, and GVL have been discussed. These biofuels can be directly blended with diesel or gasoline with or without modifications, obtaining high combustion efficiency and reducing the negative environmental impact of traditional fuels. Currently, the conversion of lignocellulosic biomass into biofuels involves multistep processes, but one-pot reaction directly from biomass would be more economical and sustainable since lignocellulose is an abundant and inexpensive feedstock. The design of multifunctional catalysts plays a key role in the development of these processes as they are required to carry out the production of these biofuels directly from biomass in one-pot synthesis. This fact saves cost, time, and energy for an economic industrial-scale production. These multifunctional catalysts must possess a variety of catalytic functions, including acid sites, basic sites, and metal hydrogenation sites. Moreover, these catalysts must be recoverable and must maintain their catalytic performance for multiple catalytic runs. The main bottleneck of the development of these catalysts is the catalytic deactivation found in most of these processes. The side reactions involved in these processes due to the high reactivity of HMF and furfural molecules are also well known. This deactivation is still more pronounced if biomass is directly employed as feedstock since more number of secondary reactions can take place. The formation of soluble and insoluble by-products such as humins accounts for 10–50% carbon loss of the feed, causing unfavorable process economics. These species are frequently deposited on the catalytic surface, resulting in deactivation of catalyst. Therefore, a deeper analysis on the deactivation processes is required in order to understand and to assist in the design of effective catalysts specifically tailored to produce biofuels from lignocellulosic biomass. By identifying, optimizing, and tuning important catalytic parameters to prevent deactivation, the design of multifunctional catalysts for cascade reactions will be more effective in decreasing the coke formation, improving the selectivity to desirable chemicals, and increasing the rate of reaction. Therefore, future attention may also focus more on the deactivation and regeneration of catalysts to enhance their stability and subsequently their economic viability. Finally, the development of efficient method of separation and purification of these biofuels is also required. With these technological advances, the synthesis of these biofuels will make a more promising progress and play an important role in the transportation sector.
References 1 Hu, L., Lin, L., Wu, Z. et al. (2017). Renewable Sustainable Energy Rev. 74:
230–257. 2 De Bhowmick, G., Sarmah, A., and Sen, R. (2018). Bioresour. Technol. 247:
1144–1154.
611
612
33 Conversion of Lignocellulosic Biomass to Biofuels
3 Qian, Y., Zhu, L., Wang, Y., and Lu, X. (2015). Renewable Sustainable Energy
Rev. 41: 633–646. 4 Wang, T., Nolte, M., and Shanks, B. (2014). Green Chem. 16: 548–572. 5 Demirbas, A. (2001). Energy Convers. Manage. 42: 1357–1378. 6 FitzPatrick, M., Champagne, P., Cunningham, M.F., and Whitney, R.A.
(2010). Bioresour. Technol. 101: 8915–8922. 7 Lynd, L., Sow, M., Chimphango, A. et al. (2015). Biotechnol. Biofuels 8 (1): 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29
30
18. Saha, B., Bohre, A., Dutta, S., and Abu-Omar, M. (2015). ACS Sustainable Chem. Eng. 3: 1263–1277. Agarwal, A. (2007). Prog. Energy Combust. Sci. 33: 233–271. Balat, M. (2011). Energy Convers. Manage. 52: 858–875. Demirbas, A. (2007). Prog. Energy Combust. Sci. 33: 1–18. De, S., Dutta, S., and Saha, B. (2012). ChemSusChem 5: 1826–1833. Dutta, S., De, S., Alam, M. et al. (2012). J. Catal. 288: 8–15. Taherzadeh, M. and Karimi, K. (2008). Int. J. Mol. Sci. 9: 1621–1651. Brandt, A., Grasvik, J., Hallett, J., and Welton, T. (2013). Green Chem. 15: 550–583. Alonso, D., Bond, J., and Dumesic, J. (2010). Green Chem. 12: 1493–1513. Avanthi, A., Kumar, S., Sherpa, K., and Banerjee, R. (2017). Biofuels-UK 8: 431–444. Kubicka, D., Kubickova, I., and Cejka, J. (2013). Catal. Rev. 55: 1–78. Chandra, R., Takeuchi, H., and Hasegawa, T. (2012). Appl. Energy 94: 129–140. Demirbas, A. (2011). Appl. Energy 88: 17–28. Kumari, D. and Singh, R. (2018). Renewable Sustainable Energy Rev. 90: 877–891. van Putten, R., van der Waal, J., de Jong, E. et al. (2013). Chem. Rev. 113: 1499–1597. Zeitsch, K.J. (2000). The Chemistry and Technology of Furfural and Its Many By-Products. Elsevier Science B.V. Lange, J.P., Van Der Heide, E., Van Buijtenen, J., and Price, R. (2012). ChemSusChem 5: 150–166. Mascal, M. and Nikitin, E. (2008). Angew. Chem. Int. Ed. 47: 7924–7926. Tang, X., Wei, J., Ding, N. et al. (2017). Renewable Sustainable Energy Rev. 77: 287–296. Horvath, I., Mehdi, H., Fabos, V. et al. (2008). Green Chem. 10: 238–242. Yan, K., Wu, G., Lafleur, T., and Jarvis, C. (2014). Renewable Sustainable Energy Rev. 38: 663–676. Balan, V. (2014). Current challenges in commercially producing biofuels from lignocellulosic biomass. ISRN Biotechnol. 2014: 463074 (31). Hindawi Publishing Corporation ISRN Biotechnology. https://doi.org/10.1155/2014/ 463074. Alipour, S., Omidvarborna, H., and Kim, D. (2017). Renewable Sustainable Energy Rev. 71: 908–926.
References
31 Gruter, G.J.M. and Dautzenberg, F. Method for the synthesis of
32 33
34
35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57
5-hydroxymethylfurfural ethers and their use. US Patent 2011/0082304 A1, filed 15 December 2010 and issued 7 April 2011. Ras, E., Maisuls, S., Haesakkers, P. et al. (2009). Adv. Synth. Catal. 351: 3175–3185. Imhof, P., Dias, A.S., de Jong, E., and Gruter, G.J. (2009). OA02 Furanics: Versatile Molecules for Biofuels and Bulk Chemicals Applications. Avantium Technologies BV. Sanborn, A.J. (2008). Processes for the preparation and purification of hydroxymethylfuraldehyde and derivatives. US Patent 2006/0142599, filed 09 December 2005 and issued 29 June 2006. Balakrishnan, M., Sacia, E., and Bell, A. (2012). Green Chem. 14: 1626–1634. Lanzafame, P., Temi, D., Perathoner, S. et al. (2011). Catal. Today 175: 435–441. Neves, P., Antunes, M., Russo, P. et al. (2013). Green Chem. 15: 3367–3376. Che, P., Lu, F., Zhang, J. et al. (2012). Bioresour. Technol. 119: 433–436. Liu, A., Zhang, Z., Fang, Z. et al. (2014). J. Ind. Eng. Chem. 20: 1977–1984. Liu, A., Liu, B., Wang, Y. et al. (2014). Fuel 117: 68–73. Ren, Y., Liu, B., Zhang, Z., and Lin, J. (2015). J. Ind. Eng. Chem. 21: 1127–1131. Raveendra, G., Rajasekhar, A., Srinivas, M. et al. (2016). Appl. Catal., A 520: 105–113. Li, H., Zhang, Q., and Yang, S. (2014). Int. J. Chem. Eng.: 481627. https://doi .org/10.1155/2014/481627. Liu, B. and Zhang, Z. (2013). RSC Adv. 3: 12313–12319. Wang, S., Zhang, Z., Liu, B., and Li, J. (2013). Catal. Sci. Technol. 3: 2104–2112. Zhang, Z., Wang, Y., Fang, Z., and Liu, B. (2014). ChemPlusChem 79: 233–240. Yin, S., Sun, J., Liu, B., and Zhang, Z. (2015). J. Mater. Chem. A 3: 4992–4999. Wang, H., Deng, T., Wang, Y. et al. (2013). Green Chem. 15: 2379–2383. Morales, G., Paniagua, M., Melero, J., and Iglesias, J. (2017). Catal. Today 279: 305–316. Kazi, F., Patel, A., Serrano-Ruiz, J. et al. (2011). Chem. Eng. J. 169: 329–338. Zhao, H., Holladay, J., Brown, H., and Zhang, Z. (2007). Science 316: 1597–1600. Kunkes, E., Simonetti, D., West, R. et al. (2008). Science 322: 417–421. Lew, C., Rajabbeigi, N., and Tsapatsis, M. (2012). Ind. Eng. Chem. Res. 51: 5364–5366. Li, H., Saravanamurugan, S., Yang, S., and Riisager, A. (2016). Green Chem. 18: 726–734. Bai, Y., Wei, L., Yang, M. et al. (2018). J. Mater. Chem. A 6: 7693–7705. Mariscal, R., Maireles-Torres, P., Ojeda, M. et al. (2016). Energy Environ. Sci. 9: 1144–1189. Joshi, H., Moser, B., Toler, J. et al. (2011). Biomass Bioenergy 35: 3262–3266.
613
614
33 Conversion of Lignocellulosic Biomass to Biofuels
58 Silva, J., Grekin, R., Mariano, A., and Maciel, R. (2018). Energy Technol. 6:
613–639. 59 Unlu, D., Ilgen, O., and Hilmioglu, N. (2016). Chem. Eng. J. 302: 260–268. 60 Nandiwale, K., Sonar, S., Niphadkar, P. et al. (2013). Appl. Catal., A 460:
90–98. 61 Nandiwale, K., Niphadkar, P., Deshpande, S., and Bokade, V. (2014). J. Chem.
Technol. Biotechnol. 89: 1507–1515. 62 Yan, K., Wu, G., Wen, J., and Chen, A. (2013). Catal. Commun. 34: 58–63. 63 Pasquale, G., Vazquez, P., Romanelli, G., and Baronetti, G. (2012). Catal.
Commun. 18: 115–120. 64 Patil, C., Niphadkar, P., Bokade, V., and Joshi, P. (2014). Catal. Commun. 43: 65 66 67 68 69 70 71 72 73 74 75
76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93
188–191. Melero, J., Morales, G., Iglesias, J. et al. (2013). Appl. Catal., A 466: 116–122. Oliveira, B. and da Silva, V. (2014). Catal. Today 234: 257–263. Fernandes, D., Rocha, A., Mai, E. et al. (2012). Appl. Catal., A 425: 199–204. Peng, L., Lin, L., Zhang, J. et al. (2011). Appl. Catal., A 397: 259–265. Peng, L., Lin, L., Li, H., and Yang, Q. (2011). Appl. Energy 88: 4590–4596. Liu, Y., Liu, C., Wu, H., and Dong, W. (2013). Catal. Lett. 143: 1346–1353. Kuo, C., Poyraz, A., Jin, L. et al. (2014). Green Chem. 16: 785–791. Deng, W., Liu, M., Zhang, Q., and Wang, Y. (2011). Catal. Today 164: 461–466. Demolis, A., Essayem, N., and Rataboul, F. (2014). ACS Sustainable Chem. Eng. 2: 1338–1352. Lange, J., van de Graaf, W., and Haan, R. (2009). ChemSusChem 2: 437–441. van de Graaf, W.D. and Lange, J.P. (2007). A catalytic process for the convers ion of furfuryl alcohol into levulinic acid or alkyl levulinate. US Patent WO2007023173, filed 24 August 2006 and issued 01 March 2007. Neves, P., Lima, S., Pillinger, M. et al. (2013). Catal. Today 218: 76–84. Chen, B., Li, F., Huang, Z. et al. (2014). ChemSusChem 7: 202–209. Hu, L., Lin, L., and Liu, S. (2014). Ind. Eng. Chem. Res. 53: 9969–9978. Hu, E., Hu, X., Wang, X. et al. (2012). Tribol. Int. 55: 119–125. Zhong, S., Daniel, R., Xu, H. et al. (2010). Energy Fuels 24: 2891–2899. Román-Leshkov, Y., Barrett, C., Liu, Z., and Dumesic, J. (2007). Nature 447: 982–986. Binder, J. and Raines, R. (2009). J. Am. Chem. Soc. 131: 1979–1985. Zu, Y., Yang, P., Wang, J. et al. (2014). Appl. Catal., B 146: 244–248. Upare, P., Hwang, D., Hwang, Y. et al. (2015). Green Chem. 17: 3310–3313. Saha, B., Bohn, C., and Abu-Omar, M. (2014). ChemSusChem 7: 3095–3101. Chidambaram, M. and Bell, A. (2010). Green Chem. 12: 1253–1262. Nishimura, S., Ikeda, N., and Ebitani, K. (2014). Catal. Today 232: 89–98. Chatterjee, M., Ishizaka, T., and Kawanami, H. (2014). Green Chem. 16: 1543–1551. Huang, Y., Chen, M., Yan, L. et al. (2014). ChemSusChem 7: 1068–1072. Kong, X., Zhu, Y., Zheng, H. et al. (2014). RSC Adv. 4: 60467–60472. Braun, M. and Antonietti, M. (2017). Green Chem. 19: 3813–3819. Wei, Z., Lou, J., Li, Z., and Liu, Y. (2016). Catal. Sci. Technol. 6: 6217–6225. Wang, C., Xu, H., Daniel, R. et al. (2013). Fuel 103: 200–211.
References
94 De, S., Saha, B., and Luque, R. (2015). Bioresour. Technol. 178: 108–118. 95 Bremner, J.G.M. and Keeys, R.K.F. (1947). J. Chem. Soc.: 1068–1080. 96 Burnett, L.W., Johns, I.B., Holdren, R.F., and Hixon, R.M. (1948). Ind. Eng.
Chem. 40: 502–505. 97 Lessard, J., Morin, J., Wehrung, J. et al. (2010). Top. Catal. 53: 1231–1234. 98 Zhu, Y., Xiang, H., Li, Y. et al. (2003). New J. Chem. 27: 208–210. 99 Jiménez-Gómez, C., Cecilia, J., Moreno-Tost, R., and Maireles-Torres, P.
(2017). ChemSusChem 10: 1448–1459. 100 Sitthisa, S., An, W., and Resasco, D. (2011). J. Catal. 284: 90–101. 101 Lee, W., Wang, Z., Zheng, W. et al. (2014). Catal. Sci. Technol. 4: 2340–2352. 102 Xiong, K., Lee, W., Bhan, A., and Chen, J. (2014). ChemSusChem 7:
2146–2149. 103 Peters, F.N. (1933). Method for the reduction of furfural and furan derivates 104 105 106 107 108 109 110 111
112 113 114 115 116 117 118 119 120 121 122
123 124 125
US Patent 1906873, filed 26 April 1928 and issued 02 May 1933. Linares, G. and Nudelman, N. (2003). J. Phys. Org. Chem. 16: 569–576. Yan, K. and Chen, A. (2014). Fuel 115: 101–108. Alonso, D., Wettstein, S., and Dumesic, J. (2013). Green Chem. 15: 584–595. Bereczky, A., Lukács, K., Farkas, M., and Dóbé, S. (2014). Nat. Resour. 5: 177–191. Fábos, V., Koczó, G., Mehdi, H. et al. (2009). Energy Environ. Sci. 2: 767–769. Serrano-Ruiz, J.C., West, R.M., and Dumesic, J.A. (2010). Annu. Rev. Chem. Biomol. Eng. 1: 79–100. Gurbuz, E., Alonso, D., Bond, J., and Dumesic, J. (2011). ChemSusChem 4: 357–361. Dunlop, A.P. and Madden, J.W. (1957). Process of preparing gamma-valerolactone. US Patent 2786852, filed 19 August 1953 and issued 26 March 1957. Hengne, A., Biradar, N., and Rode, C. (2012). Catal. Lett. 142: 779–787. Upare, P., Lee, J., Hwang, D. et al. (2011). J. Ind. Eng. Chem. 17: 287–292. Manzer, L. (2004). Appl. Catal., A 272: 249–256. Bourne, R., Stevens, J., Ke, J., and Poliakoff, M. (2007). Chem. Commun.: 4632–4634. Yan, K., Lafleur, T., Wu, G. et al. (2013). Appl. Catal., A 468: 52–58. Yan, K., Jarvis, C., Lafleur, T. et al. (2013). RSC Adv. 3: 25865–25871. Du, X., Bi, Q., Liu, Y. et al. (2011). ChemSusChem 4: 1838–1843. Du, X., He, L., Zhao, S. et al. (2011). Angew. Chem. Int. Ed. 50: 7815–7819. Son, P., Nishimura, S., and Ebitani, K. (2014). RSC Adv. 4: 10525–10530. Upare, P., Lee, J., Hwang, Y. et al. (2011). ChemSusChem 4: 1749–1752. Haan, R., Lange, J.P., Petrus, L., and Petrus-Hoogenbosch, C. (2007). Hydrogenation process for the conversion of a carboxylic acid or an ester having a carbonyl group. US Patent 20070208183 A1, filed 28 February 2007 and issued 06 September 2007. Shimizu, K., Kanno, S., and Kon, K. (2014). Green Chem. 16: 3899–3903. Obregón, I., Corro, E., Izquierdo, U. et al. (2014). Chin. J. Catal. 35: 656–662. Hengne, A. and Rode, C. (2012). Green Chem. 14: 1064–1072.
615
616
33 Conversion of Lignocellulosic Biomass to Biofuels
126 Chia, M. and Dumesic, J. (2011). Chem. Commun. 47: 12233–12235. 127 Zhou, H., Song, J., Kang, X. et al. (2015). RSC Adv. 5: 15267–15273. 128 Yang, Z., Huang, Y., Guo, Q., and Fu, Y. (2013). Chem. Commun. 49:
5328–5330. 129 Alonso, D., Wettstein, S., Bond, J. et al. (2011). ChemSusChem 4: 1078–1081. 130 Molinari, V., Antonietti, M., and Esposito, D. (2014). Catal. Sci. Technol. 4:
3626–3630. 131 Wettstein, S., Alonso, D., Chong, Y., and Dumesic, J. (2012). Energy Environ.
Sci. 5: 8199–8203. 132 Alonso, D., Gallo, J., Mellmer, M. et al. (2013). Catal. Sci. Technol. 3:
927–931. 133 Ding, D., Wang, J., Xi, J. et al. (2014). Green Chem. 16: 3846–3853. 134 Putro, J., Kurniawan, A., Soetaredjo, F. et al. (2015). RSC Adv. 5:
41285–41299.
617
34 Conversion of Carbohydrates to High Value Products Isao Ogino Hokkaido University, Faculty of Engineering, , N13W8 Kita-ku, Sapporo, 060-8628, Japan
34.1 Introduction Biomass is an organic matter typically formed from CO2 and H2 O through photosynthesis: h𝜈
xCO2 + yH2 O −−→ Cx (H2 O)y + xO2 Lignocellulose consists of cellulose (40–50%), hemicellulose (15–30%), and lignin (Figure 34.1) [1] and has been considered as the most renewable source of carbon to produce value-added chemicals. Because the chemical formula of cellulose and hemicellulose is represented by Cx (H2 O)y and most carbon atoms in them bear a molecule of water attached in the form of an H and an OH, they are also referred to as in the group of carbohydrates (“hydrates of carbon”) [2]. Carbohydrates are usually defined as polyhydroxy aldehydes and ketones or substances that hydrolyze to yield polyhydroxy aldehydes and ketones. Carbohydrates that cannot be hydrolyzed into simpler carbohydrates are called monosaccharides. Glucose, fructose, and xylose are examples of monosaccharides. Carbohydrates that undergo hydrolysis to form two or three molecules are called disaccharides and trisaccharides, respectively. Cellulose is an example of polysaccharides that give a large number of molecules of monosaccharides (>10) upon hydrolysis. Cellulose is a crystalline polymer of glucose in which β-d-glucopyranose units are linked through β-glycosidic bonds (Figure 34.1) [3]. It exhibits high resistance to chemical attack because of its crystalline structure with extensive hydrogen bonding through OH groups. Hemicelluloses are other examples of polysaccharides that have β-(1→4)-linked backbones with an equatorial configuration (Figure 34.1) [4]. They are amorphous with low degrees of polymerization (∼200) and branched polymers. The structure of hemicelluloses depends on the source and may consist of C5 carbohydrates (pentoses) such as xylose and arabinose and C6 sugars (hexoses) such as glucose, fructose, galactose and uronic acids. A general strategy to convert carbohydrates to value-added chemicals starts with deconstruction of polysaccharides into smaller molecules (Figure 34.2) [6]. Heterogeneous Catalysts: Advanced Design, Characterization and Applications, First Edition. Edited by Wey Yang Teoh, Atsushi Urakawa, Yun Hau Ng, and Patrick Sit. © 2021 WILEY-VCH GmbH. Published 2021 by WILEY-VCH GmbH.
34 Conversion of Carbohydrates to High Value Products
Cellulose H OH
HO HO
H OH
H
H H O
H HO
H
H
H
H
H
H
OH
n
HO
HO
OH H O
OH
H
OH
OH
HO O
HO
H O
OH
H H
O
O
O H
H H
OH
Lignin
HO
O
O
O
OH
OH
OH
p-Coumaryl alcohol Coniferyl alcohol Sinapyl alcohol H COOH OH
H O H3CO HO
H
HO
H
H
O
H OH HO
H H H
C OH
H H
H
CH2OH
H H
O
OH
HO
O
O HO
H O
HO
O
H
H O
OH
H
OH H
HO
C O
O
HO O
HO
H
H H
O
C O
OH
HH
H
O
Hemicellulose
H
C
H
OH
H
H
H
OH
H
Figure 34.1 Structure of lignocellulosic biomass with cellulose, hemicellulose, and lignin represented. The building blocks of lignin, p-coumaryl, coniferyl, and sinapyl alcohol are also shown. Source: Adapted from Alonso et al. [1]. Copyright 2012, Royal Society of Chemistry. (See online version for color figure). Carbohydrate fractions
“Initial” Platform Chemicals Chemocatalytic conversion
Lignocellulosic biomass
O
O
HO HO
O
O
OH
OH
Pretreatment steps
HO
OH
n
CHO
Furfural (FAL)
Xylose
C5-units
O
HO
OH
C6-units
HO HO n
CHO or
5-HMF
OH
O
O
O
HOCH2 HOCH2 O
Biocatalytic conversion
OH OH Glucose
O OH Levulinic acid (LA)
O
OH
O
OH
Ethanol
Propionic acid
O HO
HO
OH
Carbon-based chemicals and value-added end products
618
3-Hydroxypropionic acid
OH O
Succinic acid
Isoprene
Figure 34.2 Synthesis of platform chemicals from biomass. Source: Adapted from Mika et al. [5]. Copyright 2018, American Chemical Society. (See online version for color figure).
The molecules derived from polysaccharides are often rich with functionality such as –OH, –C=O, and –COOH, which are reactive toward by-products. Thus, it is difficult to control selectivity to produce target chemicals in a single processing step. Hence, these molecules are converted into other molecules that possess lower oxygen content and reduced reactivity of functional groups [5]. The derived molecules serve as building blocks for further transformations [7]. The US Department of Energy has selected such building molecules and reported them as platform chemicals (Figure 34.2) [8]. Some of the platform chemicals are derived by biochemical methods, whereas others by chemical methods. In this chapter, some examples of the recent and emerging technologies related to heterogeneous catalysts are described for the conversion of carbohydrates to value-added chemicals.
34.2 Overview of Strategy for Catalyst Development and Routes for Conversion
34.2 Overview of Strategy for Catalyst Development and Routes for Conversion of Carbohydrates The current trends include the syntheses of nontoxic green solvents and polymer precursors that potentially substitute petroleum-derived compounds. Heterogeneous catalysts offer technological advantages over homogeneous catalysts because they allow easier separation of reaction products and enable continuous flow reaction processes [9]. Continuous flow processes generally facilitate mass production of chemicals, which is critical to providing biomass-derived chemicals at reasonable prices. General strategy for development of catalysts starts by identifying the type of reactions to convert certain functional groups or configurations of a molecule to those of a target molecule. Types of reactions often used in the conversion of carbohydrates and molecules derived from carbohydrates are dehydration, hydrolysis, isomerization, hydrogenation, hydrogenolysis, and C–C coupling. Some of these reactions (e.g. dehydration, hydrolysis, and isomerization) are often catalyzed by acid or base sites, whereas others (e.g. hydrogenation) are catalyzed by metal sites or by a combination of acid sites and metal sites. Then, a catalyst is synthesized by aiming to incorporate active sites (acid or basic sites or metal sites) possibly in a desirable environment (nonpolar or polar) in nanometer-sized pores of a catalyst. When conversion of carbohydrate-derived molecules requires sequential reactions at acid and metal sites, those sites are coupled as a bifunctional catalyst or as a packed-bed reactor consisting of a dual bed of acid catalysts and supported metal catalysts. Some of these catalysts may be similar to those used in petroleum industry. However, the conversion of carbohydrate-derived molecules is often conducted in the presence of water, sometimes under hydrothermal conditions at elevated temperatures, and possibly in an acidic or alkaline solution. Thus, new catalysts that are robust and stable under these conditions are required. Carbon materials possess hydrophobic surfaces and exhibit high resistance against acidic or alkaline media. Some metal oxides such as titania, niobia, and zirconia exhibit higher stability than other metal oxides and thus may be desirable for the conversion of carbohydrates. Alumina and aluminosilicates, which have been used widely in petroleum industry, undergo structural collapse in liquid-phase reactions for biomass conversion [10]. Coating of surface of alumina particles with graphitic carbon is known to improve their hydrothermal stability [11]. An ultimate goal in catalyst development is to synthesize a robust solid catalyst that enables high selectivity under mild reaction conditions like enzyme catalysts. Various efforts have been paid to synthesize solid catalysts that mimic enzyme-like performance. Representative examples are crystalline silicate-based catalysts that possess Lewis acid sites in the molecular-sized channels. These materials possess the same topology as some zeolites. Because they are crystalline and possess molecular-sized pores and may allow entrance and conversion of specific molecules, they are usually referred to as molecular sieves. A representative example is a molecular sieve with β topology that contains Sn (Sn-β) as the Lewis acidic center (Figure 34.3) [12]. Zeolites are crystalline aluminosilicates consisting of interconnected SiO4 and AlO4 − tetrahedra. The negative charge
619
620
34 Conversion of Carbohydrates to High Value Products
OSi OSi HO OH
Sn SiO
Si
(b) (a)
Figure 34.3 Schematic drawings of zeolite β framework (a) and an open site (b). (See online version for color figure).
on the AlO4 − tetrahedra is balanced by extra-framework cations such as H+ and Na+ , rendering ion-exchange properties. When protons are present as extra-framework cations, they act as strong Brønsted acids. The framework Si in pure silica molecular sieves may be substituted with other heteroatoms such as Fe and Sn. Because Sn has the same number of valence electron as Si, it requires no extra-framework cations like H+ , and hence Sn sites act as Lewis acid sites. Many metal cations bind water strongly and thus are poisoned by it and not desirable for the conversion of carbohydrates. However, Sn sites in Sn-β exhibit weaker interaction with water and still activate functional group such as carbonyl groups. In addition, when Sn-containing molecular sieves are highly crystalline, they offer highly hydrophobic environments that mimic those in active sites of enzymes [13]. Furthermore, the pore apertures of Sn-β molecular sieves (6–7 Å) are large enough to accommodate glucose. The environment inside the pores is highly hydrophobic and seems to be unfavorable for the adsorption of hydrophilic glucose. However, a favorable entropy change upon adsorption of glucose from an aqueous solution into the pore is known to drive its facile transfer [14]. Various works indicate that the location of Sn sites may not be random [15] and that a so-called open site consisting of (–Si–O)3 –Sn–OH with a neighboring SiOH (Figure 34.3) enables high selectivity in isomerization of glucose to fructose [13]. Representative routes to derive platform chemicals and their derivatives with examples of catalysts used in the reactions are shown in Figure 34.4, and details are described in Section 34.3. A sequence of selective conversions of cellulose yields dihydrolevoglucosenone, a nontoxic green solvent, dihydrolevoglucosenone, via levoglucosenone (LGO) as will be described in Section 34.3.1 [16]. Dehydration of C6 carbohydrates (glucose and fructose) forms a platform chemical, 5-hydroxymethylfurfural (HMF) [17], which can be converted further into various chemicals as will be described in Sections 34.3.2–34.3.4 [18]. Glucose is more abundant than fructose as a feedstock. However, dehydration of fructose to HMF is more facile than that of glucose to HMF. Thus, isomerization of glucose to fructose is desirable [19]. A Sn-β can isomerize glucose to fructose efficiently [12, 13, 20] and allows a one-pot synthesis of HMF from glucose when it is coupled with dehydration reaction catalyzed by a Brønsted acid catalyst [21].
OH
OH O HO
OH
O
O
O
HO O
HO
O
OH
OH n
OH
Cellulose
Green solvent
Hydrolysis
Hydrolysis
–H2O
O
HO HO
Brønsted acid
O
O
O
O
Brønsted acid
H2
O
Ni/SiO2
Levoglucosenone (LGO)
Levoglucosan
O
O
OH
Dihydrolevoglucosenone
–H2O H OH
O OH
Acrylic acid
Surfactant and solvent
–H2O
H
OH OH
H
H
–H
OH
2O
O
OH
HO O OH
OH
H2
OH
H 2 O –H 2 5-Hydroxymethylfurfural Pt/C (HMF)
OH
H2
O O
OH
O
O
n
Fructose
O
HO HO
H
OH H
HO
OH
Resin
O O
Furfural
n
Hemicellulose
Polylactic acid (PLA)
OH
O
O
O
O
n
Polyethylene terephthalate (PET)
O
Levulinic acidS H 2 up Brønsted acid H2 po rte HO O d O O Furfuryl alcohol Au or Ru-S n Formic acid /ZrO 2 γ-Valerolactone
O
HO
OH
O
Terephthalic acid
OH O
O
O OH
Xylose
Lactide
O
Biopolymer
H O
H O
Polyethylene furanoate (PEF)
Brønsted acid
H H
OH
1,6-Hexanediol (1,6-HDO)
O
O
OH
HO
O
2,5-Furandicarboxilic acid (FDCA)
–H2O
O
HO
HO
HO
Lactic acid
O
O
O HO
OH
OH
H2
Pt-WOx/TiO2 1,2,6-Hexanetriol (1,2,6-HTO) Tetrahydrofurandimethanol (THFDM)
Ni/SiO 2
O
Brønsted acid
Isomerization Lewis acid (Sn-β)
OH
OH O
Glucose
O
Monomer
Isomerization
H O
HO HO
Monomer
Plasticizers Pharmaceuticals Green solvent
Biodegradable polymer
Figure 34.4 Conversions of carbohydrates to value-added chemicals and their intermediates. Examples of type of reactions, catalysts, and applications are presented. (See online version for color figure).
622
34 Conversion of Carbohydrates to High Value Products
Hydrogenation of HMF forms tetrahydrofurandimethanol (THFDM). THFDM may also be obtained from LGO and can be converted further into 1,2,6-hexanetriol (1,2,6-HTO) that serves as surfactant and solvent or 1,6-hexanediol (1,6-HDO) that serves as polymer precursor [22]. Oxidation of HMF yields furandicarboxylic acid (FDCA), which has been considered as a substitute of petroleum-based terephthalic acid for the production of polyethylene terephthalate (PET) as will be described in Section 34.3.3 [23]. Diels–Alder cycloaddition of reduced or oxidized HMF with ethylene forms terephthalic acid as shown in Section 34.3.4 [24], which can be used to synthesize PET. Acid-catalyzed dehydration of xylose or hemicellulose produces furfural, a five-carbon building block [25]. Solid acid catalysts combined with a biphasic reaction system allow the continuous flow synthesis of furfural from xylan hemicellulose and xylose [25]. Partial or total hydrogenation of furfural or HMF produces furfuryl alcohol or tetrahydrofurfuryl alcohol, respectively [26]. Furfuryl alcohol has been used in the synthesis of foundry resin. Tetrahydrofurfuryl alcohol produces 1,5-pentanediol, which can be used as a polymer precursor. Acid-catalyzed conversion of hydration of furfuryl alcohol yields another platform chemical, levulinic acid [27], which serves as a building block for pharmaceuticals, plasticizers, and cosmetics [28]. Alkyl esters formed from levulinic acid via esterification reaction with alcohol using acid catalysts [29] may be used for flavors, fragrances, and plasticizers. Hydrogenation of levulinic acid forms γ-valerolactone (GLV) [30], which has been demonstrated as an effective solvent for various biomass-related conversions [31]. For example, HMF readily undergoes condensation reactions at moderate temperatures like 373 K in the presence of an acid or a base. Thus, dehydration of fructose to HMF in water results in low yields of HMF (70%) from fructose. Fermentation of pentoses or hexoses forms lactic acid (2-hydroxypropanoic acid), which is another platform molecule that can be converted into various useful molecules [32]. Because the fermentation pathway to obtain lactic acid is costly, heterogeneous catalytic systems have been developed as will be described in Section 34.3.5 [33]. Lactic acid also serves as the building block for polylactic acid (PLA), which is a biodegradable polymer. A heterogeneous catalytic system based on molecular sieving catalysts has been developed to enable an economical route to produce an intermediate molecule for PLA as will be described in Section 34.3.6 [34]. Some of the platform chemicals have been produced on a large scale in industry. For example, AVA Biochem has produced HMF on a commercial scale by hydrothermal processing. GFBiochemicals has produced levulinic acid from C6 carbohydrates, which are derived from biomass through an acid-catalyzed hydrolysis treatment under high temperature and pressure. In addition to the platform chemicals, other useful chemicals have been produced on a large scale from biomass. For example, Circa has developed a Furacell technology, which converts biomass to LGO through a continuous reaction system. TM
34.3 Synthesis of Value-Added Chemicals from Carbohydrates
34.3 Synthesis of Value-Added Chemicals from Carbohydrates 34.3.1
Dihydrolevoglucosenone
When dehydration of cellulose is conducted in aprotic solvents, it forms LGO. Partial hydrogenation of LGO over a supported Pd catalyst forms dihydrolevoglucosenone (Cyrene ), which has been produced on a large scale by Circa. Like this reaction, hydrogenation reactions are generally carried out using supported metals such as Pd, Pt, Ni, or Ru to saturate C—C bonds (C=C and C≡C bonds) and C=O bonds [35]. Hydrogen atoms formed on supported Pd metals from H2 add C=C bond in LGO. Cyrene has been certified as a nontoxic solvent by the EU recently. However, Cyrene exhibits solvent properties similar to conventional aprotic solvents such as dimethylformamide (DMF) and N-methylpyrrolidone (NMP), which are toxic. Beneficial effects of Cyrene as a solvent have been demonstrated in several chemical reactions such as fluorination of an aromatic ring, which is of significant interest in pharmaceutical industry. LGO and Cyrene can be converted to THFDM via hydrogenation of C=O bond to form threo- and erythro-levoglucosanol (Lgol) and a subsequent hydrogenolysis reaction [36]. TM
34.3.2
1,6-Hexanediol
1,6-HDO serves as a monomer for polyesters and polyurethanes [37]. It has been produced on a large scale via a petrochemical route using catalytic hydrogenation of adipic acid or its esters. Rennovia has developed a pilot-scale plant for an alternative biomass route and has reported that it enables a better profit margin than the petrochemical route. 1,6-HDO may be obtained by hydrogenation of HMF or LGO to THFDM, followed by hydrogenolysis [22]. Hydrogenolysis is a reaction that cleaves C—C and C—O bonds by hydrogen and is often referred to as hydrodeoxygenation (HDO). HDO is typically carried out on a bifunctional solid catalyst that performs both hydrogenation at metal sites and dehydration reactions at acid sites. The conversion of HMF to 1,6-HDO has been accomplished at high yields by using Pd/SiO2 and Ir-ReOx /SiO2 catalysts in a H2 O/tetrahydrofuran (THF) mixture [38]. Other researchers have demonstrated a high-yield synthesis of 1,6-HDO by hydrogenation of LGO by a Ni/SiO2 or Pt/SiO2 –Al2 O3 catalyst and hydrogenolysis of THFDM over a Pt–WOx /TiO2 or Pt–WOx /ZrO2 catalyst (Figure 34.4) [22]. 34.3.3
Furandicarboxylic Acid (FDCA)
FDCA has been considered as a substitute of petroleum-based terephthalic acid for the production of polyesters with ethylene glycol (EG). Polymerization of FDCA with EG yields polyethylene 2,5-furandicarboxylate (PEF) (Scheme 34.1). PEF is considered to reduce energy usage and greenhouse gas emissions by about 50% and an improved oxygen permeability relative to PET. PET is generally synthesized by polymerization of EG and dimethyl terephthalate or terephthalic acid.
623
624
34 Conversion of Carbohydrates to High Value Products O
O n
O
O HO
OH
+ n HO
O O
OH
O
O
O n
(a) O
O
O + n HO
n HO
O O
OH
OH
O
O
(b)
n
Scheme 34.1 Synthesis of polyethylene 2,5-furandicarboxylate (a) and polyethylene terephthalate (b).
Terephthalic acid has been produced by selective oxidation of p-xylene. The manufacture of PET is the largest and about one-fifth in the total production of polymer in the world. It is used as bottles for water and soft drinks as well as food packaging. Avantium has produced a homogeneous catalyst system that converts methoxymethylfurfural to FDCA. FDCA can also be obtained by the oxidation of HMF using solid catalysts. Selective oxidation of HMF to FDCA may be achieved by the chemistry used in the synthesis of terephthalic acid from p-xylene with some modifications. However, FDCA exhibits low solubility in many solvents, and two to four equivalent bases are often required to prevent deactivation of solid catalysts by precipitation of FDCA on their surfaces. Consequently, a subsequent neutralization step, which produces a large amount of wastes, is required to obtain FDCA. To overcome this challenge, solid base materials such as MgO and hydrotalcites [39] were used as catalyst supports for metals that enable the oxidation reaction. Hydrotalcites are layered double hydroxides consisting of octahedrally coordinated Mg2+ and Al3+ (Figure 34.5). A partial substitution of Mg sites in brucite Mg(OH)2 with Al forms positive charges in the layer, which are compensated by interlayer anions. Hydrotalcites show anion-exchange ability [40] and exhibit strong Brønsted basicity when it incorporates OH− anions [41]. Calcination of hydrotalcites forms Mg–Al mixed oxides that exhibit strong Lewis basicity, depending on the Mg–Al ratio. The approach described in the preceding paragraph often needs a highly diluted HMF feed to prevent precipitation of products. A more recent work demonstrated that fructose can be converted to HMF at 70% yield in a Mg-Al double hydroxides
Anion, An– Water
Figure 34.5 Schematic drawing of Mg–Al layered double hydroxides. (See online version for color figure).
34.3 Synthesis of Value-Added Chemicals from Carbohydrates
γ-valerolactone (GVL)/H2 O mixed solvent and subsequently oxidized to FDCA at 93% yield over a carbon-supported platinum nanoparticle catalyst (Pt/C catalyst) [23]. The high solubility of FDCA in GVL/H2 O solution enables the oxidation at high concentrations. Thus, the Pt/C catalyst can be used to yield FDCA at a high yield without a homogeneous base catalyst. In addition, FDCA can be separated from the solvent by crystallization to obtain >99% pure FDCA. Additionally, FDCA acts as an effective catalyst to dehydrate fructose to HMF. Therefore, corrosive mineral acid catalysts such as HCl and H2 SO4 are not required. By-products formed by the dehydration of carbohydrates can be separated and converted into activated carbon. Another work has demonstrated the synthesis of FDCA by oxidation of an acetal derivative of HMF using a CeO2 -supported gold nanoparticle catalyst and Na2 CO3 in water [23]. The acetal derivative of HMF is formed by the reaction of HMF with 1,3-propanediol (Scheme 34.2) and exhibits high stability against humin formation. The work has suggested two crucial steps in the reaction mechanism through first principles calculations (Figure 34.6). Partial hydrolysis of the acetal into 5-formyl-2-furan carboxylic acid is catalyzed by OH− and Lewis acid sites on CeO2 , and subsequent oxidative dehydrogenation of the in situ generated hemiacetal involves Au nanoparticles. O
O
HO
H
O
or HO
O
O HO
O
O
O
OH
Scheme 34.2 Synthesis of FDCA from HMF or an acetal derivative of HMF. HO
O
HO
O
O HO
O
HO H
O
100 Au
Energy (kJ/mol)
0
–100
O
OH–
CeO2
OO
CeO2
H O O
Au
OO O
O
O O Ce
O O O H O O
O HO
TS1 (75)
Ce
Au
O
O
CeO2
O
O O
IM1
HO Au CeO2
O
O
O
HO O
O
TS3 (–20)
H O
H O Ce O O
Au
HO
O
–
O
O HO
CeO2
TS2 (–76)
IM2 (–135)
O Ce O
IM4 (–154)
O O
–200
HO
Au
–300
Ce
CeO2
O
O H O O Ce O O
IM3 (–258) O O
HO
O OH O
–400 Au CeO2
O OH Ce O O
Figure 34.6 Computed reaction energy diagram for the rate-determining step during the oxidation of acetal-derivative of HMF by 1,3-propanediol (PD-HMF) into FDCA over the CeO2 -supported Au catalyst. Source: Adapted from Kim et al. [23]. Copyright 2018, Wiley-VCH. (See online version for color figure).
625
626
34 Conversion of Carbohydrates to High Value Products
CH3 O
H3C
CH3 +C2H4
O H3C
[1]
[2]
–H2O –H2O
H+
+H2O CH3
O
[4]
O
H 3C
[3]
Figure 34.7 Diels–Alder cycloaddition of dimethylfuran [1] and ethylene produces an oxanorbornene cycloadduct [2], which dehydrates to p-xylene [3]. Water hydrolyzes dimethylfuran [1] to 2,5-hexadione [4] in equilibrium. Source: Adapted from Chang et al. [24]. Copyright 2014, Royal Society of Chemistry.
34.3.4
Terephthalic Acid
Recently, a new catalytic route from HMF to terephthalic acid has been developed using Diels–Alder reaction (Figure 34.7) [24]. A Diels–Alder reaction proceeds between a conjugated diene and a compound containing a double bond. The reaction forms two σ bonds at the expense of two π bonds. HMF may be converted to dimethylfuran by hydrogenolysis using a carbon-supported CuRu bimetallic catalyst. Then, the dimethylfuran can be reacted with ethylene in heptane solvent at 523 K using various solid acid catalysts, which results in the symmetry-allowed [4 + 2] Diels–Alder cycloaddition of ethylene to dimethylfuran to form an oxanorbornene and subsequent Brønsted-acid-catalyzed dehydration to yield p-xylene. A β zeolite that possesses Brønsted acid sites exhibited the best mass-specific activity and a high selectivity to p-xylene of ∼90% at conversion of 99%. When the conversion of dimethylfuran is low, the reaction forms a side product of 2,5-hexanedione by hydration. However, as the conversion was increased, it reverted back to dimethylfuran and was converted to p-xylene. Because hydrogenation of HMF requires a supported precious metal catalyst, a more economical route, which uses oxidized derivatives of HMF, has been reported (Figure 34.8) [24, 42]. In the work, a partially oxidized HMF, 5-(hydroxymethyl)furoic acid (HMFA), and each of the ether and ester derivatives of HMFA were reacted with ethylene to yield aromatic products. The fully oxidized HMF, which is FDCA, can be reacted with ethylene to form p-xylene. However, the strong electron-withdrawing effects of two –COOH groups in FDCA deactivate the diene and result in a prohibitory slow rate of reaction.
34.3 Synthesis of Value-Added Chemicals from Carbohydrates Solid Brönsted acids, homogeneous Lewis acids
O
O
n tio
uc
d Re
O
O
HO OH
HO
O
Ox id
ati on
OH
O
n atio
id
Ox
Diels–Alder/dehydration reactions
O O
Biomass
HO
O
Oxidation
O
HMF
O OH
HO
HMFA O O
O
O
O O
O O
O
HMBA
O
HO
OH
HO
O
O
HO
O
Oxidation Lewis acid molecular sieves
HO
O
O
OH
OH
PTA
O
O
and/or O
O
DMT
O O
O O
OH
Figure 34.8 Synthesis of p-xylene through Diels–Alder cycloaddition of reduced or oxidized HMF. Source: Adapted from Pacheco and Davis [24]. Copyright 2014, American Chemical Society.
The partially oxidized derivatives of HMF allows the reaction of ethylene at reasonable rates over molecular sieves with zeolite β topology containing Lewis acid sites (Sn4+ and Zr4+ ). The molecular-sized pores present in the catalysts facilitate the cycloaddition reaction by confinement effects.
34.3.5
Lactic Acid
PLA has been currently produced on a large scale through an anaerobic fermentation route. The sequential synthesis route consists of fermentation of carbohydrates (pentoses or hexoses) to lactic acid, conversion of lactic acid to a cyclic dimer called as lactide, and ring-opening polymerization of lactide to form PLA [43]. The first fermentation step dominates the total cost of the production of PLA. Thus, to achieve a more economical production of PLA, various catalytic routes that replace the fermentation route have been investigated over a decade. Initially, a homogeneous Lewis acid catalyst based on Sn halide was used to convert trioses to alkyl lactates [44]. Later, heterogeneous acid catalysts based on molecular sieves have been investigated as potential catalysts to convert trioses to lactic acid or esters [45]. Conversion of hexoses to lactic acid is more challenging because it involves scissions of C—C bonds, which requires more harsh conditions and consequently facilitates the formation of side products. Direct conversion of fructose or sucrose to methyl lactate was demonstrated by using solid Lewis acid catalysts such as Sn-β molecular sieves [33]. The reaction was performed in methanol solvent at 433 K, forming methyl lactate in carbon-based yields of up to 44% for fructose and 68% for sucrose. The reaction proceeds via a retro-aldol reaction that forms two trioses from fructose, which undergo sequential dehydration and addition of methanol (Figure 34.9).
627
628
34 Conversion of Carbohydrates to High Value Products OH
OH O
HO
OH
r1
O OH
HO OH OH Aldohexose
r5
r6
O O
OH OH Ketohexose
r2
Humins
OH
5-Hydroxymethylfurfural
r11
r3
OH
OH
O
r12
O
O HO
OH
Glycolaldehyde
OH
OH OH OH 2-C-(hydroxymethyl)-aldopentose
Dihydroxyacetone
r7
OH O HO
OH
O O
Pyruvaldehyde
OH Aldotetrose
r8
O
OH
Glyceraldehyde
r9
O OH OR Pyruvaldehyde hemiacetal
OH O OR Alkyl lactate
r4 C2 and C4 products
r10 C2 by- products
Figure 34.9 Schematic representation of reaction network in which ketohexoses can isomerize to aldohexoses via 1,2-hydride shift (r1 ) and to 2-C-(hydroxymethyl)-aldopentoses via 1,2-carbon shift (r11 ) reactions. Retro-aldol reactions of hexose species (r2 , r3 , and r12 ) lead to the formation of C2 , C3 , and C4 carbohydrate fragments. Lewis acids can then catalyze the formation of α-hydroxy carboxylic acids from these smaller fragments (e.g. r7 , r8 , and r9 in the formation of alkyl lactate from trioses). Side reactions, involving dehydration reactions of fructose to 5-HMF (r5 ), redox and fragmentation reactions of unstable intermediates, and various humin-forming condensation reactions, lead to loss yield of desired products. Source: Adapted from Orazov and Davis [33]. Copyright 2015, American Chemical Society.
The high-temperature conditions used in the preceding work results in the formation of HMF and subsequent degradation of HMF to insoluble humins when sugars are converted. Moreover, the Sn-β facilitates aldose–ketose isomerization via 1,2-hydride shift, which results in the formation of C2 and C4 products from aldoses in addition to desirable C3 products. To minimize the undesirable reactions, a tandem catalyst system has been developed recently [33]. The catalyst system combined heterogeneous catalysts such as molybdenum oxide, alkali-exchanged stannosilicate molecular sieves, or amorphous TiO2 –SiO2 coprecipitates, which are coupled with solid Lewis acid catalysts such as Sn-containing molecular sieves with MFI topology (Sn-MFI). The former catalysts catalyze retro-aldol reactions but do not readily catalyze the aldose–ketose isomerization reaction. Because Sn-MFI possesses smaller pore apertures (5–6 Å) than Sn-β, it prevents access of hexoses to Lewis acid sites and hence does not cause undesirable side reactions of aldose–ketose isomerization. 34.3.6
Lactide
The current process converts lactic acid to lactide via polycondensation of lactic acid with removal of water to a prepolymer and a subsequent reaction to form
34.3 Synthesis of Value-Added Chemicals from Carbohydrates
PLA production cost breakdown
Renewable New catalytic processes carbohydrates or classic fermentation
Feedstock
O L-Lactic
acid (LA) = HO (aqueous)
Current process Step 1. Polycondensation LA LA H2O LA
L2A L3A
Zeolite H2O Proposed process
O Highly disperse prepolymer low Mw
HO
O
n
Step 2. Backbiting O O O O
n Dissolved metal salt vacuum, 453–523 K
HO
O
O
O O (L,L)-Lactide
O
O
ROP
Direct cyclization O HO OH O O
L4A
Vacuum, 423–493 K
O
Lactide synthesis
OH
LA synthesis
O O
O ROP
O n Controlled high Mw PLLA
-(D,L)-Lactide: 5–10% -Dead end side products
Figure 34.10 Current and new chemical process for making lactide and polylactic acid from lactic acid. Red: current industrial two-step process; blue: proposed direct lactide synthesis by selectivity control during condensation. Ln A, linear oligomer of n lactyl units; ROP, ring-opening polymerization. Pie chart shows a cost estimate for current PLA production from sugar to pellet. Source: Adapted from Dusselier et al. [34]. Copyright 2015, AAAS. (See online version for color figure).
lactide from the prepolymer (Figure 34.10), both of which are energy-intensive conversions. In particular, the second synthesis step requires continuous removal of lactide by an energy-intensive separation process like distillation to shift the equilibrium transesterification reaction. A more energy-efficient route has been demonstrated using zeolites as shape-selective catalysts [46]. In general, Brønsted-acid-catalyzed conversion of lactic acid forms lactide as well as undesired lactyl oligomers. However, molecular-sized pores present in zeolites restrict the formation of oligomers and facilitate the cyclization to form lactide. A β zeolite forms lactide with nearly 79% yield at full conversion of lactic acid. In addition, an integrated process has been proposed. In this process, a stirred tank reactor was used to perform the shape-selective conversion of lactic acid under reflux, water formed in the reaction is continuously removed from the reaction mixture, and a dry solvent is recycled to the reactor. A more work
629
630
34 Conversion of Carbohydrates to High Value Products
from the same group demonstrated the synthesis of lactide from alkyl lactates in the gas-phase reaction over a supported TiO2 /SiO2 catalyst [34].
34.4 Perspective As shown by examples shown in this chapter, promising solid catalysts and catalytic processes have been developed to convert carbohydrates into value-added products. The ultimate goal in catalyst development in twenty-first century is to design and synthesize a solid catalyst that achieves 100% selectivity and high activity under mild reaction conditions like enzyme catalysts. To achieve this goal, the research work in twenty-first century should aim to deepen the fundamental understanding of structure–performance relationships to design catalyst architectures to enable such performance. Systematic investigations of solid catalysts with a series of well-defined structures using state-of-the-art analytical techniques that track their dynamic structural changes at the atomic scale will facilitate such efforts as described in other sections. In addition, theoretical calculations that guide interpretation of experimental results and predict desirable structures will facilitate the rational design of new catalysts further. In addition, a synthetic method that creates active sites with atomic precision [47] is required to enable the synthesis of solid catalysts with structures predicted through rational design. Achieving such goal may be still far away. However, as demonstrated in the examples of Sn-β catalysts and other various catalysts, efforts to achieve the ultimate success have continued.
References 1 (a) Gibson, L.J. (2012). J. R. Soc. Interface 9: 2749–2766. (b) Alonso, D.M.,
Wettstein, S.G., and Dumesic, J.A. (2012). Chem. Soc. Rev. 41: 8075–8098. 2 Solomons, T.W.G. and Fryhle, C.B. (2011). Organic Chemistry, 10e. Hoboken,
NJ: Wiley. 3 O’Sullivan, A.C. (1997). Cellulose 4: 173–207. 4 Scheller, H.V. and Ulvskov, P. (2010). Annu. Rev. Plant Biol. 61: 263–289. 5 (a) Corma, A., Iborra, S., and Velty, A. (2007). Chem. Rev. 107: 2411–2502. (b)
Mika, L.T., Csefalvay, E., and Nemeth, A. (2018). Chem. Rev. 118: 505–613. 6 Onda, A., Ochi, T., and Yanagisawa, K. (2008). Green Chem. 10: 1033–1037. 7 (a) Sheldon, R.A. (2014). Green Chem. 16: 950–963. (b) Zhang, X., Wilson, K.,
and Lee, A.F. (2016). Chem. Rev. 116: 12328–12368. 8 Bozell, J.J. and Petersen, G.R. (2010). Green Chem. 12: 539–554. 9 (a) Ertl, G., Knözinger, H., Schüth, F., and Weitkamp, J. (2008). Handbook
of Heterogeneous Catalysis, vol. 8. Wiley. (b) Gates, B.C. (1992). Catalytic Chemistry. New York, NY: Wiley. 10 Ravenelle, R.M., Copeland, J.R., Kim, W.-G. et al. (2011). ACS Catal. 1: 552–561. 11 Pham, H.N., Anderson, A.E., Johnson, R.L. et al. (2012). Angew. Chem. Int. Ed. 51: 13163–13167.
References
12 (a) Corma, A., Nemeth, L.T., Renz, M., and Valencia, S. (2001). Nature 412:
13
14 15
16 17
18
19
20 21 22 23
24
25 26 27
28
29 30
423–425. (b) Moliner, M., Román-Leshkov, Y., and Davis, M.E. (2010). Proc. Natl. Acad. Sci. U.S.A. 107: 6164–6168. (c) Davis, M.E. (2015). Top. Catal. 58: 405–409. (a) Bermejo-Deval, R., Assary, R.S., Nikolla, E. et al. (2012). Proc. Natl. Acad. Sci. U.S.A. 109: 9727–9732. (b) Gounder, R. and Davis, M.E. (2013). AlChE J. 59: 3349–3358. Bai, P., Siepmann, J.I., and Deem Michael, W. (2013). AlChE J. 59: 3523–3529. (a) Bare, S.R., Kelly, S.D., Sinkler, W. et al. (2005). J. Am. Chem. Soc. 127: 12924–12932. (b) Yang, G., Pidko, E.A., and Hensen, E.J.M. (2013). J. Phys. Chem. C 117: 3976–3986. (c) Wolf, P., Valla, M., Núñez-Zarur, F. et al. (2016). ACS Catal. 6: 4047–4063. Cao, F., Schwartz, T.J., McClelland, D.J. et al. (2015). Energy Environ. Sci. 8: 1808–1815. (a) Román-Leshkov, Y., Chheda, J.N., and Dumesic, J.A. (2006). Science 312: 1933. (b) Chen, S.S., Maneerung, T., Tsang, D.C.W. et al. (2017). Chem. Eng. J. 328: 246–273. (c) Simeonov, S.P., Coelho, J.A., and Afonso, C.A. (2012). ChemSusChem 5: 1388–1391. (a) van Putten, R.J., van der Waal, J.C., de Jong, E. et al. (2013). Chem. Rev. 113: 1499–1597. (b) Rosatella, A.A., Simeonov, S.P., Frade, R.F.M., and Afonso, C.A.M. (2011). Green Chem. 13: 754–793. (a) Takagaki, A., Ohara, M., Nishimura, S., and Ebitani, K. (2009). Chem. Commun. (Cambridge, UK): 6276–6278. (b) Osatiashtiani, A., Lee, A.F., Granollers, M. et al. (2015). ACS Catal. 5: 4345–4352. Ren, L., Guo, Q., Kumar, P. et al. (2015). Angew. Chem. Int. Ed. 54: 10848–10851. Nikolla, E., Román-Leshkov, Y., Moliner, M., and Davis, M.E. (2011). ACS Catal. 1: 408–410. He, J., Burt, S.P., Ball, M. et al. (2018). ACS Catal. 8: 1427–1439. (a) Motagamwala, A.H., Won, W.Y., Sener, C. et al. (2018). Sci. Adv.: 4. (b) Kim, M., Su, Y., Fukuoka, A. et al. (2018). Angew. Chem. Int. Ed. 57: 8235–8239. (a) Chang, C.-C., Green, S.K., Williams, C.L. et al. (2014). Green Chem. 16: 585–588. (b) Pacheco, J.J. and Davis, M.E. (2014). Proc. Natl. Acad. Sci. U.S.A. 111: 8363–8367. Aellig, C., Scholz, D., Dapsens, P.Y. et al. (2015). Catal. Sci. Technol. 5: 142–149. Nakagawa, Y., Tamura, M., and Tomishige, K. (2013). ACS Catal. 3: 2655–2668. (a) González Maldonado, G.M., Assary, R.S., Dumesic, J., and Curtiss, L.A. (2012). Energy Environ. Sci. 5: 6981–6989. (b) Weingarten, R., Cho, J., Xing, R. et al. (2012). ChemSusChem 5: 1280–1290. (a) Démolis, A., Essayem, N., and Rataboul, F. (2014). ACS Sustainable Chem. Eng. 2: 1338–1352. (b) Pileidis, F.D. and Titirici, M.M. (2016). ChemSusChem 9: 562–582. Ogino, I., Suzuki, Y., and Mukai, S.R. (2018). Catal. Today 314: 62–69. Du, X.L., He, L., Zhao, S. et al. (2011). Angew. Chem. Int. Ed. 50: 7815–7819.
631
632
34 Conversion of Carbohydrates to High Value Products
31 (a) Delidovich, I., Leonhard, K., and Palkovits, R. (2014). Energy Environ.
32 33 34
35
36 37 38 39 40 41 42 43 44 45 46 47
Sci. 7: 2803–2830. (b) Qi, L., Mui, Y.F., Lo, S.W. et al. (2014). ACS Catal. 4: 1470–1477. Dusselier, M., Van Wouwe, P., Dewaele, A. et al. (2013). Energy Environ. Sci. 6: 1415. Orazov, M. and Davis, M.E. (2015). Proc. Natl. Acad. Sci. U.S.A. 112: 11777–11782. (a) Dusselier, M., Van Wouwe, P., Dewaele, A. et al. (2015). Science 349: 78. (b) De Clercq, R., Dusselier, M., Makshina, E., and Sels, B.F. (2018). Angew. Chem. Int. Ed. 57: 3074–3078. Ogino, I., Serna, P., and Gates Bruce, C. (2017). Supported catalysts. In: Handbook of Solid State Chemistry, vol. 6 (eds. R. Dronskowski, S. Kikkawa and A. Stein), 313–337. Wiley-VCH. Krishna, S.H., Assary, R.S., Rashke, Q.A. et al. (2018). ACS Catal. 8: 3743–3753. Krishna, S.H., Huang, K., Barnett, K.J. et al. (2018). AlChE J. 64: 1910–1922. Xiao, B., Zheng, M., Li, X. et al. (2016). Green Chem. 18: 2175–2184. Ardemani, L., Cibin, G., Dent, A.J. et al. (2015). Chem. Sci. 6: 4940–4945. Ogino, I., Tanaka, R., Kudo, S., and Mukai, S.R. (2018). Microporous Mesoporous Mater. 263: 181–189. Ebitani, K., Motokura, K., Mori, K. et al. (2006). J. Organomet. Chem. 71: 5440–5447. Pacheco, J.J., Labinger, J.A., Sessions, A.L., and Davis, M.E. (2015). ACS Catal. 5: 5904–5913. De Clercq, R., Dusselier, M., and Sels, B.F. (2017). Green Chem. 19: 5012–5040. Hayashi, Y. and Sasaki, Y. (2005). Chem. Commun. (Cambridge, UK): 2716–2718. Dapsens, P.Y., Menart, M.J., Mondelli, C., and Pérez-Ramírez, J. (2014). Green Chem. 16: 589–593. Holm, M.S., Saravanamurugan, S., and Taarning, E. (2010). Science 328: 602. Hermans, S. and Visat de Bocarmé, T. (2015). Atomically-Precise Methods for Synthesis of Solid Catalysts. Cambridge: Royal Society of Chemistry.
633
35 Enhancing Sustainability Through Heterogeneous Catalytic Conversions at High Pressure Nat Phongprueksathat and Atsushi Urakawa Delft University of Technology, Department of Chemical Engineering, Catalysis Engineering, Van der Maasweg 9, 2629 HZ Delft, The Netherlands
35.1 Importance of High-Pressure Reaction Condition The most successful example of the high-pressure chemistry in heterogeneous catalysis and chemical processing is the commercialization of ammonia synthesis process (300–700 bar) developed by Haber and Bosch in 1913 [1]. This breakthrough had a significant impact on large-scale production of fertilizers supporting almost half of the world’s population through enhanced food production [2]. Following this innovation, high-pressure methanol synthesis process (300 bar) was commercialized by BASF in 1923 [3]. This marked the emergence of the mass production of organic chemicals since methanol is considered one of the most promising building blocks to obtain more complex chemical structures in petrochemical industries [4]. Another success is the development of a still up-to-date polymerization process of ethylene at an ultrahigh pressure (3000 bar) in 1935 [5]. Since then, polyethylene has evolved into a material critical to our life. The major characteristics of these high-pressure processes are still similar to those of the state-of-the-art processes existing today. Nowadays, the global trend of those high-pressure catalytic processes tends to shift toward a milder condition at lower pressure and temperature or even at ambient conditions [6–8]. This movement was partially triggered by the famous “12 principles of green chemistry.” At a glance, the high-pressure reaction condition does not comply with the “design for energy efficiency” and “safety” principles. Still, the high-pressure condition offers distinctive advantages for many catalytic reactions by significantly increasing the product yield and improving the incorporation of materials used in the process into the final product, thus “maximizing atom economy.” The smaller but more productive process enabled by the high-pressure condition can render such processes safer and more energy efficient. Moreover, high pressure allows the use of supercritical states of various media such as carbon dioxide and water as a “safer solvent or auxiliary” instead of often hazardous organic solvents. The unique advantages of the high-pressure condition can be so prominent that the overall catalytic processes can be designed Heterogeneous Catalysts: Advanced Design, Characterization and Applications, First Edition. Edited by Wey Yang Teoh, Atsushi Urakawa, Yun Hau Ng, and Patrick Sit. © 2021 WILEY-VCH GmbH. Published 2021 by WILEY-VCH GmbH.
634
35 Enhancing Sustainability Through Heterogeneous Catalytic Conversions at High Pressure
Reaction kinetics
Chemical equilibrium P
Mass transport Low P
High P
Phase behaviors
Low P
High P
Stagnant film
Process efficiency
T > Tcrit P > Pcrit
Figure 35.1 The main motivations to use high-pressure conditions for heterogeneously catalyzed reactions.
in accordance with the green chemistry principles despite the apparent imaginal discrepancy. In the industrial chemical and catalytic processes, the motivation to use high-pressure reaction condition varies. Figure 35.1 summarizes common advantageous characteristics offered by such reaction conditions, and each point is described below. 35.1.1
Chemical Equilibrium (One Phase)
The most common motivation to employ high-pressure condition is to induce thermodynamic advantages by shifting chemical equilibrium when the number of molecules (thus volume for gas-phase reaction) reduces toward product formation, according to Le Châtelier’s principle. The shift in chemical equilibrium ultimately affects the maximum value of attainable conversion and product selectivity. The syntheses of ammonia, methanol, and low-density polyethylene are obvious examples of exploiting high-pressure reaction conditions for such advantages. The role of catalyst in this respect is to accelerate the reaction so that thermodynamics play more dominant roles in defining reaction performance. Generally, high-pressure conditions offer fewer advantages for liquid-phase reaction due to negligible or the less compressive nature of liquids. 35.1.2 35.1.2.1
Phase Behavior (Multiphase) Phase Separation
High-pressure conditions are prevalently used for phase separation in the industrial processes, mainly to liquefy some condensable components out of the gas
35.1 Importance of High-Pressure Reaction Condition
mixture based on an increased vapor density with pressure. Some liquid-phase reactors take advantage of high-pressure conditions to prevent vaporization of the light component such as light hydrocarbons, thus retaining reactive components with a catalyst in the liquid phase. Relatively novel utilization of high-pressure condition for phase separation is an in situ condensation of product from the gas phase during the reaction. In this case, the reaction performance is ruled by multiphase equilibrium (chemical and vapor–liquid equilibrium) and can override the limit defined by the one-phase thermodynamic equilibrium of the reaction. This concept resembles multiphase equilibrium exploitation in reactive distillation that integrates chemical reaction and product separation in a single unit. An example is the methyl tert-butyl ether (MTBE) production where MTBE is continuously removed from a catalytic distillation column to enhance equilibrium-limited isobutene conversion [4]. 35.1.2.2
Supercritical State
Increasing pressure of a medium can force two separate phases to collapse into one. When pressure and temperature are above the critical point, the phase boundary is indistinguishable, and the fluid density is somewhere between that of gas and liquid at standard conditions. The diffusion rate of molecules in their supercritical state is comparable with that of the gas phase. Dense medium and enhanced molecular diffusion are favorable for chemical reactions since molecular collisions and consequently the reaction rate can be enhanced. Also, high fluid density promotes heat transfer, which is helpful for heat removal. These characteristics enable the use of supercritical fluids as a unique solvent or even as reactant (e.g. CO2 conversion under a supercritical state of CO2 ) as frequently applied in hydrogenation and water oxidation under supercritical conditions. 35.1.3
Mass Transfer and Kinetics
In a heterogeneous reaction involving a solid catalyst, the chemical reaction is merely one of several steps. The required interaction of reactants with catalyst surface highlights the importance of mass transport, which can affect the progress of reaction apart from reaction equilibrium and kinetics. For instance, convective and diffusive transport of reactants from the bulk fluid into catalyst pore and adsorption, desorption, and surface diffusion of reactants on catalyst surface can play decisive roles in defining the reaction rate. Also, high-pressure conditions are known to induce higher coverage of reactants and consequently the concentration of surface intermediates at elevated pressures. This generally leads to enhanced surface reaction rate, favorable for overall reaction kinetics [9, 10]. 35.1.3.1
Molecular Diffusion
The aforementioned enhanced molecular diffusion under pressurized conditions, e.g. supercritical states, can be advantageous for reaction kinetics. In reality, the effect of pressure on diffusion rate in heterogeneous catalysis is complex since
635
636
35 Enhancing Sustainability Through Heterogeneous Catalytic Conversions at High Pressure
reaction pressure affects more than one parameter at the same time (e.g. viscosity of fluids). Another important nature of diffusion is that the driving force for molecular diffusion is not only intrinsic but also extrinsic. For example, the concentration gradient, the main driving force for molecular diffusion as illustrated by Fick’s law, is related to partial pressure in the gas phase. Thus, the partial pressure driving force can be conveniently generated and tuned in case of gas-phase reaction by creating pressure difference in a reaction system. A convenient approach that applies such pressure gradient is membrane reactor, where a product could be continuously separated by a membrane, and thus reaction equilibrium is not limited by one-phase thermodynamics. 35.1.3.2
Multiphase Reaction
In case of a multiphase reaction taking place in a trickle-bed or slurry reactor, dissolution and mass transport of reactant from the gas phase into the liquid phase are often enhanced by increasing reaction pressure (Henry’s law). This was applied for high-pressure hydrotreating processes where hydrogen transfer into liquid heavy oil and the reaction occurs on the solid surface of the catalyst after the diffusion of molecular hydrogen through the liquid medium. 35.1.4
Process Efficiency and Economy
High-pressure operation effectively decreases reactor volume and thus the amount of catalyst required, which could offer economic advantages despite obvious higher capital cost for thicker-wall reactor and compression. Another note is that an industrial catalytic process hardly consists of only one reaction unit. Usually, a product from the unit will be treated and/or processed by the preceding or the subsequent units for further reaction and separation operated at different temperature and pressure. For high-pressure reactions, prior compression of reactants is required, and it is one of the most expensive operations in a catalytic process and consumes a large amount of energy. Therefore, it is imperative to evaluate the process efficiency and economy not only on the basis of the catalytic reactor but also as the whole process. For example, in the steam reforming of methane to produce syngas, the steam reformer is operated at a relatively high pressure (30 bar), although the reaction is not thermodynamically favorable at higher pressures and it requires higher reaction temperature and more excess steam to counterbalance the negative effect on the reaction equilibrium [4]. Still, most applications of syngas such as methanol and ammonia synthesis require syngas at high pressure (50−150 bar) or even higher [1, 3], and operation of the steam reforming at high pressure makes sense as the whole process. Furthermore, by taking advantage of the thermodynamic stability of the reaction products, typical compression could be avoided. An example is the generation of hydrogen from formic acid. Using a nano-Pd heterogeneous catalyst, the system can generate high-pressure gases containing only hydrogen and carbon dioxide at over 360 bar under mild temperature (80 ∘ C) [11]. This system is suitable for the practical application of fuel cell vehicles (FCVs), which requires hydrogen at high pressure over 350 bar [12].
35.2 State-of-the-Art Application of High Pressure in Heterogeneous Catalysis
35.2 State-of-the-Art Application of High Pressure in Heterogeneous Catalysis 35.2.1 Boosting CO2 Conversion and Surpassing One-Phase Chemical Equilibrium by In situ Phase Separation at High-Pressure Reaction Condition Hydrogenation of CO2 allows the production of hydrocarbons and oxygenates such as methane, methanol, dimethyl ether, and formic acid. Production of these chemicals will play pivotal roles in energy storage technology and utilization of CO2 known as the most emitted greenhouse gas. Particularly, methanol has gained increasing attention due to the ease in its transport and safety compared with hydrogen as a chemical energy carrier [13]. Methanol can be conveniently handled by the existent liquid fuels infrastructure and serve as a starting material (C1 feedstock) for producing various chemicals (olefin and gasoline, to name a few) [14]. Methanol production utilizing waste CO2 (e.g. contained in flue gases) and H2 produced by water electrolysis sourced from natural and renewable energy, such as wind and solar power, is a sustainable path for carbon recycling urged to be implemented and spread in near future. Methanol synthesis by hydrogenation of CO2 (i) is more complex compared with that from synthesis gas (ii) due to the fact that reverse water–gas shift (RWGS) reaction (35.3) takes place in parallel with methanol synthesis reaction (35.1): CO2 + 3H2 ⇌ CH3 OH + H2 O
(35.1)
CO + 2H2 ⇌ CH3 OH
(35.2)
CO2 + H2 ⇌ CO + H2 O
(35.3)
According to the stoichiometry of these reactions, the number of molecules reduces for the methanol synthesis reactions (35.1) and (35.2), whereas the number remains the same for the RWGS reaction. This means that increasing reaction pressure can shift the chemical equilibrium toward methanol formation since the pressure dependency of the RWGS reaction is negligible [15]. In addition, the methanol synthesis reactions (35.1) and (35.2) are exothermic, while RWGS reaction is endothermic; thus lower temperature should be favored for high-yield methanol synthesis. This thermodynamic nature of the methanol synthesis reaction from CO2 and H2 is illustrated by the equilibrium CO2 conversion and methanol selectivity at the stoichiometric ratio (CO2 : H2 = 1 : 3) in the temperature range where generally reported catalysts show activity [16]. Obviously, the reaction pressure has a drastic influence on the yield of methanol (molar basis – not time yield), showing a clear advantage of operating the reaction at higher pressures. At 100 and 200 bar, there is a steep increase in CO2 conversion toward lower reaction temperature (Figure 35.2, gray line, at c. 170 and 240 ∘ C, respectively), and this is due to the condensation of the liquid products (methanol and water) below the transition temperatures. This condensation removes the products from the gas-phase reaction and shifts
637
35 Enhancing Sustainability Through Heterogeneous Catalytic Conversions at High Pressure
(a)
400 bar 200 bar 100 bar 30 bar
150
200
250
300
Temperature (°C)
100 90 80 70 60 50 40 30 20 10 0
Methanol selectivity (%)
100 90 80 70 60 50 40 30 20 10 0
CO2 conversion (%)
638
350
(b)
400 bar 200 bar 100 bar 30 bar
150
200
250
300
350
Temperature (°C)
Figure 35.2 Equilibrium CO2 conversion (a) and methanol selectivity (b) at different temperatures in stoichiometric CO2 hydrogenation at 30, 100, 200, and 400 bar. Only CO was considered as another carbon-containing chemical produced by the reaction. The thick gray line in CO2 conversion at 100 and 200 bar shows the phase transition due to liquid-phase condensation at lower temperatures. Source: Álvarez et al. 2017 [16]. Adapted with permission of American Chemistry Society.
forward the equilibrium of Reaction (35.1); thus >95% methanol yield is within reach in theory. The phase separation of product stream was demonstrated at 150 bar and 206 ∘ C using a high-pressure view cell by Kommoß et al. [17]. In this cell, a mixture of H2 , CO2 , CH3 OH, and H2 O was prepared to simulate a product stream outlet from a reactor at 50% CO2 conversion and 67% CH3 OH selectivity. This suggested a possible in situ phase separation of the products within the reactor. At the same time, the results imply the necessity to perform the reaction at a kinetically unfavorable temperature lower than 206 ∘ C to allow the condensation of methanol and water from the gas phase and eventually shift the equilibrium toward higher methanol yield. It should be noted that the phase separation from methanol synthesis from synthesis gas feed has also been demonstrated for the first time under reaction conditions by van Bennekom et al. [18]. The liquid condensation was observed at 200 ∘ C and 200 bar from a view cell, as shown in Figure 35.3a. When the phase condensation takes place in situ in CO2 hydrogenation, one can expect gradual phase separation along the axial direction of the catalytic reactor (Figure 35.3b). The optimal temperature for this reaction is reported to be at 260–280 ∘ C to maximize CO2 conversion, CH3 OH selectivity, and thus methanol yield. However, as evident from Figure 35.2, it is not possible to take advantage of both phase separation and optimal reaction rate at 100 bar, since the phase condensation temperature (170 ∘ C) is lower than optimal reaction temperature (260–280 ∘ C). Increasing liquid condensation temperature can be achieved by increasing reaction pressure or rather such a pressure increase induces the creation of one dense phase as indicated by the smoothened CO2 conversion curve at 400 bar (Figure 35.2a). Under such very high-pressure conditions, it is possible to achieve c. 90% methanol yield under continuous operation with unprecedentedly high weight time yield (gram of methanol produced per gram of catalyst per hour) when mass transfer limitations are minimized [19].
35.2 State-of-the-Art Application of High Pressure in Heterogeneous Catalysis
Figure 35.3 (a) Liquid product formation during methanol synthesis from syngas over a commercial Cu/ZnO/Al2 O3 catalyst. P = 200 bar, T = 200 ∘ C, syngas: H2 /CO/CO2 = 0.70/ 0.28/0.02. (b) The proposed in situ phase separation within the catalyst bed inside the reactor. Source: van Bennekom et al. 2013 [18]. Reproduced with permission of Elsevier.
Catalyst
Stirrer
Liquid level (a) CO2 H2 Gas reactant CO 2 H2
(b)
CH OH 3 H O 2
Condensed product
CH3OH H2O CO CO2 H2
Catalyst bed
35.2.2 Exploitation of Supercritical Fluid Properties for Catalytic Reactions In industrial processes, supercritical fluids have been used as solvent, reactant, and catalyst itself in catalytic reactions to play vital roles. The unique properties of supercritical fluids can offer advantages for heterogeneously catalyzed reactions. As discussed above, supercritical fluids exhibit a high diffusion rate similar to the gas phase while providing a high collision rate due to its high density. The mixed and distinct characteristics are due to the molecular heterogeneity of supercritical fluids resulting from the formation of local clusters, which can be observed and simulated for supercritical CO2 and H2 O [20, 21]. The polarity of fluids often changes under supercritical state compared with that under subcritical state due to a considerable decrease in dielectric constant. A highly polar water exhibits almost nonpolarity at supercritical condition (above 374 ∘ C and 221 bar), which makes it suitable for dissolving organic molecules. For this reason, many supercritical reactions dealing with organic matters are considered a green process since water can be used instead of organic solvents. Complete miscibility between water and organic compounds also facilitates a single-phase reaction without mass transport limitation at phase boundaries. This advantage is now commercially utilized for the treatment of water containing hazardous waste [22]. Moreover, the acidity (and basicity) of water is greatly increased due to an increase in the dissociation constant, which can be exploited for acid- and base-catalyzed reactions. The combination of the increased acidity and organic solubility of water can be applied for acid-catalyzed reactions of organic compounds such as hydrolysis of esters for biodiesel production, Friedel–Crafts alkylation for aromatic substitution, and alcohol dehydration to olefin [23]. Catalysts are also employed to circumvent
639
640
35 Enhancing Sustainability Through Heterogeneous Catalytic Conversions at High Pressure
the requirement of harsh reaction conditions and improve the economics of the supercritical process, although most of them are catalyzed homogenously. The state-of-the-art process in this category is HydrofactionTM developed by Steeper Energy that combines supercritical water chemistry (390–420 ∘ C and 300–350 bar) and homogeneous catalysis to produce renewable crude oil from woody biomass [24]. On the other hand, supercritical water oxidation and metal-catalyzed heterogeneous reaction are considered the last option because of deactivation and metal leaching problem [23, 25]. More efforts are needed to improve the stability of catalysts by preventing the active site leaching and by improving sulfur tolerance as one of the major applications of supercritical water, biomass conversion, typically deals with sulfur-containing organic substances that often poison catalysts. A relatively new heterogeneous catalytic process employs supercritical properties of water for biomass conversion to produce renewable fuels. A notable example is supercritical water gasification (SCWG) that produces high-quality biofuel such as hydrogen and methane from any organic compound. This technology provided a high thermal efficiency (70–77%) with a short residence time (99.9%) under the practical condition (500 ∘ C and 20 bar) and reasonable gas hourly space velocity (15 000 h−1 ). This CO conversion obtained from the membrane reactor exceeded the thermodynamic limit (92.5%). This is obviously impossible for a conventional fixed-bed reactor in the same condition. Although elevated pressure cannot alter equilibrium conversion due to the equal number of moles before and after the reaction, reaction pressure affects the contact time between gas and catalyst surface and resulted in enhanced CO conversion. Moreover, increasing reaction pressure within the membrane
References
reactor will greatly enhance the driving force for H2 diffusion across the membrane and result in much higher CO conversion. This membrane reactor was able to produce a high-pressure pure CO2 retentate suitable for a carbon capture process, although there was a trade-off in H2 purity due to the moderate H2 permselectivity of zeolite membrane. However, that compromise is overcome by long-term stability up to 600 hours.
35.3 Concluding Remark Despite the global trends to use milder pressure reaction conditions according to green chemistry principles, high-pressure technologies are still considered vital because of various unique advantages. The positive influences of high pressure on “thermodynamics” especially for “chemical equilibrium” and “phase behavior,” as well as “reaction kinetics,” in the bulk and on the catalyst surface are representative. High pressure also allows creating a driving force for molecular diffusion that can be used to enhance product separation and shift chemical equilibrium. Furthermore, a high-pressure condition facilitates the process intensification, reducing the size and increasing the efficiency of overall catalytic processes. Such prominent advantages that cannot be achieved at ambient pressure are expected to result in the innovative design of future greener catalytic processes.
References 1 Liu, H. (2014). Ammonia synthesis catalyst 100 years: practice, enlightenment
2
3
4 5
6
7
and challenge. Chin. J. Catal. 35: 1619–1640. https://doi.org/10.1016/S18722067(14)60118-2. Erisman, J.W., Sutton, M.A., Galloway, J. et al. (2008). How a century of ammonia synthesis changed the world. Nat. Geosci. 1: 636. https://doi.org/10 .1038/ngeo325. Sheldon, D. (2017). Methanol production – A technical history. Johnson Matthey Technol. Rev. 61: 172–182. https://doi.org/10.1595/ 205651317x695622. Moulijn, J.A., Makkee, M., and van Diepen, A.E. (2013). Chemical Process Technology. Wiley https://books.google.es/books?id=7C0eQ1AiQRcC. Demirors, M. (2011). The history of polyethylene. In: 100+ Years Plastics. Leo Baekeland and Beyond (eds. E.T. Strom and S.C. Rasmussen), 115–145. https://doi.org/10.1021/bk-2011-1080.ch009. Zhang, X., Kong, R.-M., Du, H. et al. (2018). Highly efficient electrochemical ammonia synthesis via nitrogen reduction reactions on a VN nanowire array under ambient conditions. Chem. Commun. 54: 5323–5325. https://doi.org/10 .1039/C8CC00459E. Wildfire, C., Abdelsayed, V., Shekhawat, D., and Spencer, M.J. (2018). Ambient pressure synthesis of ammonia using a microwave reactor. Catal. Commun. 115: 64–67. https://doi.org/10.1016/j.catcom.2018.07.010.
645
646
35 Enhancing Sustainability Through Heterogeneous Catalytic Conversions at High Pressure
8 Van Tran, T., Le-Phuc, N., Nguyen, T.H. et al. (2018). Application of NaA
9
10
11
12
13
14
15
16
17
18
19
20
membrane reactor for methanol synthesis in CO2 hydrogenation at low pressure. Int. J. Chem. Reactor Eng. 16: 1–7. https://doi.org/10.1515/ijcre-20170046. Corral-Pérez, J.J., Bansode, A., Praveen, C.S. et al. (2018). Decisive role of perimeter sites in silica-supported Ag nanoparticles in selective hydrogenation of CO2 to methyl formate in the presence of methanol. J. Am. Chem. Soc. 140: 13884–13891. https://doi.org/10.1021/jacs.8b08505. Bansode, A., Tidona, B., von Rohr, P.R., and Urakawa, A. (2013). Impact of K and Ba promoters on CO2 hydrogenation over Cu/Al2 O3 catalysts at high pressure. Catal. Sci. Technol. 3: 767–778. https://doi.org/10.1039/ C2CY20604H. Zhong, H., Iguchi, M., Song, F.-Z. et al. (2017). Automatic high-pressure hydrogen generation from formic acid in the presence of nano-Pd heterogeneous catalysts at mild temperatures. Sustainable Energy Fuels 1: 1049–1055. https://doi.org/10.1039/C7SE00131B. Alazemi, J. and Andrews, J. (2015). Automotive hydrogen fuelling stations: an international review. Renewable Sustainable Energy Rev. 48: 483–499. https:// doi.org/10.1016/j.rser.2015.03.085. Olah, G.A., Goeppert, A., and Prakash, G.K.S. (2009). Beyond Oil and Gas: The Methanol Economy, 2e. Weinheim: Wiley-VCH Verlag GmbH & Co. KGaA https://doi.org/10.1002/9783527627806. Bertau, M., Offermanns, H., Plass, L. et al. (2014). Methanol: The Basic Chemical and Energy Feedstock of the Future: Asinger’s Vision Today. Heidelberg, New York, Dordrecht, London: Springer https://doi.org/10.1007/978-3-64239709-7. Tidona, B., Urakawa, A., and Rudolf von Rohr, P. (2013). High pressure plant for heterogeneous catalytic CO2 hydrogenation reactions in a continuous flow microreactor. Chem. Eng. Process. Process Intensif. 65: 53–57. https://doi.org/ 10.1016/j.cep.2013.01.001. Álvarez, A., Bansode, A., Urakawa, A. et al. (2017). Challenges in the greener production of formates/formic acid, methanol, and DME by heterogeneously catalyzed CO2 hydrogenation processes. Chem. Rev. 117: 9804–9838. https:// doi.org/10.1021/acs.chemrev.6b00816. Kommoß, B., Klemenz, S., Schmitt, F. et al. (2017). Heterogeneously catalyzed hydrogenation of supercritical CO2 to methanol. Chem. Eng. Technol. 40: 1907–1915. https://doi.org/10.1002/ceat.201600400. van Bennekom, J.G., Venderbosch, R.H., Winkelman, J.G.M. et al. (2013). Methanol synthesis beyond chemical equilibrium. Chem. Eng. Sci. 87: 204–208. https://doi.org/10.1016/j.ces.2012.10.013. Gaikwad, R., Bansode, A., and Urakawa, A. (2016). High-pressure advantages in stoichiometric hydrogenation of carbon dioxide to methanol. J. Catal. 343: 127–132. https://doi.org/10.1016/j.jcat.2016.02.005. Sedunov, B. (2012). The analysis of the equilibrium cluster structure in supercritical carbon dioxide. Am. J. Anal. Chem. 03: 899–904. https://doi.org/10 .4236/ajac.2012.312A119.
References
21 Kalinichev, A.G. and Churakov, S.V. (1999). Size and topology of molecular
22
23 24
clusters in supercritical water: a molecular dynamics simulation. Chem. Phys. Lett. 302: 411–417. https://doi.org/10.1016/S0009-2614(99)00174-8. Marrone, P.A. (2013). Supercritical water oxidation – current status of full-scale commercial activity for waste destruction. J. Supercrit. Fluids 79: 283–288. https://doi.org/10.1016/j.supflu.2012.12.020. Savage, P.E. (1999). Organic chemical reactions in supercritical water. Chem. Rev. 99: 603–622. https://doi.org/10.1021/cr9700989. Jensen, C.U., Rodriguez Guerrero, J.K., Karatzos, S. et al. (2017). Fundamentals of Hydrofaction : renewable crude oil from woody biomass. Biomass Convers. Biorefin. 7: 495–509. https://doi.org/10.1007/s13399-017-0248-8. Qian, L., Wang, S., Xu, D. et al. (2016). Treatment of municipal sewage sludge in supercritical water: a review. Water Res. 89: 118–131. https://doi.org/10 .1016/j.watres.2015.11.047. Gassner, M., Vogel, F., Heyen, G., and Maréchal, F. (2011). Optimal process design for the polygeneration of SNG, power and heat by hydrothermal gasification of waste biomass: thermo-economic process modelling and integration. Energy Environ. Sci. 4: 1726–1741. https://doi.org/10.1039/c0ee00629g. Schubert, M., Aubert, J., Müller, J.B., and Vogel, F. (2012). Continuous salt precipitation and separation from supercritical water. Part 3: Interesting effects in processing type 2 salt mixtures. J. Supercrit. Fluids 61: 44–54. https://doi.org/10.1016/j.supflu.2011.08.011. Kruse, A. (2008). Supercritical water gasification. Biofuels, Bioprod. Biorefin. 2: 415–437. https://doi.org/10.1002/bbb.93. Guo, Y., Wang, S.Z., Xu, D.H. et al. (2010). Review of catalytic supercritical water gasification for hydrogen production from biomass. Renewable Sustainable Energy Rev. 14: 334–343. https://doi.org/10.1016/j.rser.2009.08 .012. Elliott, D.C. (2008). Catalytic hydrothermal gasification of biomass. Biofuels, Bioprod. Biorefin. 2: 254–265. https://doi.org/10.1002/bbb.74. Nielsen, R.P., Olofsson, G., and Søgaard, E.G. (2012). CatLiq – high pressure and temperature catalytic conversion of biomass: the CatLiq technology in relation to other thermochemical conversion technologies. Biomass Bioenergy 39: 399–402. https://doi.org/10.1016/j.biombioe.2012.01.035. Singh, J. and Gu, S. (2010). Commercialization potential of microalgae for biofuels production. Renewable Sustainable Energy Rev. 14: 2596–2610. https://doi.org/10.1016/j.rser.2010.06.014. Peng, G., Vogel, F., Refardt, D., and Ludwig, C. (2017). Catalytic supercritical water gasification: continuous methanization of Chlorella vulgaris. Ind. Eng. Chem. Res. 56: 6256–6265. https://doi.org/10.1021/acs.iecr.7b00042. Reimer, J., Peng, G., Viereck, S. et al. (2016). A novel salt separator for the supercritical water gasification of biomass. J. Supercrit. Fluids 117: 113–121. https://doi.org/10.1016/j.supflu.2016.06.009. Duan, P.G., Li, S.C., Jiao, J.L. et al. (2018). Supercritical water gasification of microalgae over a two-component catalyst mixture. Sci. Total Environ. 630: 243–253. https://doi.org/10.1016/j.scitotenv.2018.02.226. TM
25
26
27
28 29
30 31
32
33
34
35
647
648
35 Enhancing Sustainability Through Heterogeneous Catalytic Conversions at High Pressure
36 Lerou, J.J., Tonkovich, A.L., Silva, L. et al. (2010). Microchannel reactor archi-
37
38
39
40
41
42
43
44
45
46
tecture enables greener processes. Chem. Eng. Sci. 65: 380–385. https://doi .org/10.1016/j.ces.2009.07.020. Tonkovich, A.Y., Perry, S., Wang, Y. et al. (2004). Microchannel process technology for compact methane steam reforming. Chem. Eng. Sci. 59: 4819–4824. https://doi.org/10.1016/j.ces.2004.07.098. Zhang, N., Chen, X., Chu, B. et al. (2017). Catalytic performance of Ni catalyst for steam methane reforming in a micro-channel reactor at high pressure. Chem. Eng. Process. Process Intensif. 118: 19–25. https://doi.org/10.1016/j.cep .2017.04.015. Murzin, D.Y. (2015). Chapter 5. Hydrogen and syngas generation. In: Chemical Reaction Technology. Berlin, München, Boston: De Gruyter https://doi.org/ 10.1515/9783110336443-007. Marlin, D.S., Sarron, E., and Sigurbjörnsson, Ó. (2018). Process advantages of direct CO2 to methanol synthesis. Front. Chem. 6: 1–8. https://doi.org/10 .3389/fchem.2018.00446. Voldsund, M., Jordal, K., and Anantharaman, R. (2016). Hydrogen production with CO2 capture. Int. J. Hydrogen Energy 41: 4969–4992. https://doi.org/10 .1016/j.ijhydene.2016.01.009. Kim, S.J., Yang, S., Reddy, G.K. et al. (2013). Zeolite membrane reactor for high-temperature water-gas shift reaction: effects of membrane properties and operating conditions. Energy Fuels 27: 4471–4480. https://doi.org/10.1021/ ef302014n. Rahimpour, M.R., Samimi, F., Babapoor, A. et al. (2017). Palladium membranes applications in reaction systems for hydrogen separation and purification: a review. Chem. Eng. Process. Process Intensif. 121: 24–49. https://doi .org/10.1016/j.cep.2017.07.021. Wang, H. and Lin, Y.S. (2012). Effects of water vapor on gas permeation and separation properties of MFI zeolite membranes at high temperatures. AlChE J. 58: 153–162. https://doi.org/10.1002/aic.12622. Masuda, T., Fukumoto, N., Kitamura, M. et al. (2001). Modification of pore size of MFI-type zeolite by catalytic cracking of silane and application to preparation of H2 -separating zeolite membrane. Microporous Mesoporous Mater. 48: 239–245. https://doi.org/10.1016/S1387-1811(01)00358-4. Arvanitis, A., Sun, X., Yang, S. et al. (2018). Approaching complete CO conversion and total H2 recovery for water gas shift reaction in a high-temperature and high-pressure zeolite membrane reactor. J. Membr. Sci. 549: 575–580. https://doi.org/10.1016/j.memsci.2017.12.051.
649
36 Electro-, Photo-, and Photoelectro-chemical Reduction of CO2 Jonathan Albo, Manuel Alvarez-Guerra, and Angel Irabien University of Cantabria, School of Industrial Engineering and Telecommunications, Department of Chemical and Biomolecular Engineering, Avda. Los Castros s/n, 39005 Santander, Spain
36.1 Introduction The large-scale consumption of fossil fuels for the production of energy has raised CO2 levels from a preindustrial level of about 270 ppm to surpass the 400 ppm. According to International Energy Agency (IEA) statistics, oil, coal, and natural gas still represent 80.6% of the total energy supply in the world in 2015, while renewable energies only accounted for 10.2% [1]. The same report also indicated that CO2 emissions have substantially grow accordingly, going from approximately 5500 MTCO2 in 1973 to 14 500 MTCO2 in 2015. The natural carbon cycle has been exceeded, and it is presently unable to keep up with the input of large amounts of anthropogenic CO2 emitted, causing severe environmental problems such as the undesirable effects of global warming. Renewable energies, and, in particular, wind and solar energy, are becoming more technically and economically feasible to replace fossil fuels, but their implementation is being slower than expected, and the world energy supply will still rely on fossil fuels in the next decades. In this scenario, the development of methodologies to reduce CO2 emissions needs to be a priority in the political agenda worldwide in order to advance toward a carbon-neutral energy cycle. A variety of strategies may alleviate the CO2 issue via improving the combustion efficiency of fossil fuels or exploring clean and renewable energy sources, but these strategies should be really hard pressed to replace fossil fuels, at least in the short term. A better efficiency in energy utilization would also help to save fossil fuels and reduce the emissions of CO2 . At the same time, great efforts are being undertaken to develop carbon capture and storage (CCS), as one of the technologies to handle large quantities of CO2 emissions, where CO2 capture (i.e. absorption, adsorption, or membrane separation) seems to be the bottleneck step where the efforts have to be applied [2]. Other methods include direct CO2 capture from air, which could be important to address the dispersed emissions in the atmosphere from cars, domestic heating, or trains [3]. Limitations to CCS technologies, however, include high costs and energy requirements, the uncertainty on the permanence of stored CO2 in storage sites or the impact of Heterogeneous Catalysts: Advanced Design, Characterization and Applications, First Edition. Edited by Wey Yang Teoh, Atsushi Urakawa, Yun Hau Ng, and Patrick Sit. © 2021 WILEY-VCH GmbH. Published 2021 by WILEY-VCH GmbH.
650
36 Electro-, Photo-, and Photoelectro-chemical Reduction of CO2
large volumes of CO2 on natural systems. Additionally, processes for converting captured CO2 into useful and valuable products, in the so-called carbon capture and utilization (CCU), are being developed in parallel. In view of the vastness of CO2 supplied in the atmosphere, it makes sense to consider CO2 a resource rather than a waste for its direct use or its transformation into new products. It is also being recognized that exploiting part of the captured CO2 to generate value from it can complement its storage via CCS, thus contributing to meet the carbon reduction targets. The utilization of CO2 can be grouped into three main categories: (i) technological utilization, which may not require the conversion of CO2 ; (ii) enhanced biological utilization; and (iii) production of chemicals and fuels [4]. Firstly, CO2 can be directly used in numerous technological applications. For several decades, CO2 has been used for enhanced oil recovery (EOR). In this process, a percentage of the CO2 stays in the reservoir, and EOR is therefore considered as a form of long-term storage. CO2 can be also used as a solvent for supercritical extraction of components such as pigments, oils, or caffeine. CO2 finds use in the food industry, including carbonated beverages and food packaging, or in water treatment to reduce pH levels in alkaline water. Secondly, CO2 is a key component of photosynthesis, and by adding additional CO2 to greenhouses, the growth of plants can be enhanced. Lastly, CO2 can be chemically converted into a variety of value-added chemicals, including salicylic acid (an important active metabolite of aspirin), urea, and different polymers. In fact, CO2 can be activated and chemically converted through different approaches such as thermochemical and biochemical methods and more innovative technologies such as the electrochemical, photochemical, or photoelectrochemical reduction [5]. These latter innovative technologies are appealing since they could enable an economically competitive industrial production of CO2 -based chemicals by using renewable energy, contributing to rebalance the carbon cycle. The final oxidation state of the carbon atom in the value-added chemicals synthesized with these technological solutions is determined by the specific reaction pathway and the number and rates of electrons exchanged. Table 36.1 shows the sequence of reactions and the corresponding redox potentials for CO2 to commonly reported products (e.g. CO, HCOOH, CH3 OH, and CH4 ) [5]. Since CO2 is a thermodynamically stable molecule, its multistep reduction via electro-, photo-, and photoelectro-chemistry is challenging and confronts Table 36.1 Redox potentials for CO2 reduction. Reaction
E 0 (V vs. NHE) at pH = 7
CO2 + e− → CO⋅− 2
−1.9
CO2 + 2H+ + 2e− → HCOOH
−0.61
CO2 + 2H+ + 2e− → CO + H2 O
−0.52
CO2 + 4H+ + 4e− → HCHO + H2 O
−0.51
CO2 + 6H+ + 6e− → CH3 OH + H2 O
−0.38
CO2 + 8H+ + 8e− → CH4 + 2H2 O
−0.24
2H+ + 2e− → H2
−0.42
36.2 Fundamentals
many fundamental technical hurdles. The kinetics for CO2 reduction are also, in general, more sluggish than the thermodynamically favorable two-electron H2 evolution reaction (HER), which competes with CO2 reduction. In any case, the achievements so far are exciting, and the authors believe that with continued research efforts, these innovative technologies for CO2 conversion might become technically and economically feasible in the near future. This chapter provides the basic principles of the electro-, photo-, and photoelectro-chemical processes for CO2 conversion, taking mainly into consideration the applied catalytic materials and cell geometries. The key challenges to be faced are envisaged, proposing future steps to be undertaken to advance as fast as possible toward realistic solutions for CO2 utilization.
36.2 Fundamentals 36.2.1
Redox Processes
There are many reactions in which it is convenient to regard electrons as participants in the transformation of reactants into products. Reactions of this kind are known generally as electron transfer reactions, or more commonly as oxidation–reduction reactions [6]. In order to convert CO2 into useful products, CO2 must be “reduced,” i.e. it must gain electrodes from another species (which is “oxidized”), in a transactional process of “reduction–oxidation,” commonly known as “redox.” The equipment for carrying out a redox reaction is normally called electrochemical reactor or electrochemical cell. These systems are most generally defined as two electrodes separated by at least one electrolyte phase [7]. When a net reaction proceeds in an electrochemical cell, oxidation occurs at one electrode (the anode), and reduction takes place at the other electrode (the cathode). We can think of the cell as consisting of two half-cells joined together by an external circuit through which electrons flow and an internal pathway that allows ions to migrate between them so as to preserve electroneutrality [6]. The redox process occurs at the same active surface in photocatalysis as explained later in the chapter, but for the sake of clarity, we focus at this point in electrochemical cells. The thermodynamics of electrochemical cells is treated in all textbooks of physical chemistry as are the conventions. The equilibrium (or reversible) cell potential can be calculated by subtracting the equilibrium potential of the anode (EeA ) from that of the cathode (EeC ), and this is related to the free energy of the overall cell reaction (ΔG) by the well-known equation [8]: e = −nF(EeC − EeA ) ΔG = −nFECell
(36.1)
where n is the number of moles of electrons per mole reacted and F is the Faraday constant (i.e. the charge on a mole of electrons, 96 485 C⋅mol−1 ). Therefore, obtaining the equilibrium cell potential for the redox reaction, we can then calculate the free energy change associated with the redox reaction. If ΔG is negative, this tells us that the reaction is favorable, while if it is positive, it is thermodynamically unfavorable and can only occur when (electrical) energy
651
652
36 Electro-, Photo-, and Photoelectro-chemical Reduction of CO2
is supplied. The equilibrium potentials for some CO2 reduction reactions are listed in Table 36.1, showing us that several different products can be obtained as reduced forms of CO2 . In any case, this is a thermodynamic discussion, but it does not consider the rate of chemical change (i.e. the current that will flow). The reactivity of CO2 is low because it is the most stable carbon-based structure in nature, and an input of energy must be supplied to reduce CO2 and convert it into reduced, useful products. Much larger voltages than the equilibrium cell potentials have to be applied in practice across the two electrodes to drive the conversion. The rate of chemical conversion will depend on the kinetics of the two electrode reactions, and an overpotential (𝜂) is necessary to carry out the reaction at a required rate (measured by the current density, j, the current per unit area of electrode surface, as explained later). Moreover, an input of energy is also necessary to drive the migration of ions between the electrodes through the cell, which leads to a potential drop iRCell (where RCell represents the internal resistance of the cell, which depends on the electrolyte properties, the configuration of the electrodes, and the design of the cell). Therefore, the cell voltage required to observe a current i in a real cell can be expressed by ∑ ECell = EeC − EeA − |𝜂| − iRCell (36.2) Both the 𝜂 and iRCell terms increase with current density and may be regarded as inefficiencies whereby electrical energy is converted into heat [8]. 36.2.2
Assessment of Reaction Performance: Figures of Merit
Several different figures of merit are commonly used to assess the performance of electrocatalytic CO2 reduction processes: (a) Rate of product formation: The amount of the desired product from CO2 reduction obtained per unit of cathode area and unit of time (e.g. expressed in units of mol⋅m−2 ⋅min−1 ) is sometimes used as a useful figure for quantifying the rate at which the product can be obtained in the reduction process. (b) Current density (j, usually expressed in mA⋅cm−2 ): The current density is defined as the electric current flow divided by the area of the electrode (usually the geometric area). It measures the rate of the conversion, and a high j as possible is desirable to minimize the area of electrode (and hence the size of the reactor) needed for a desired production rate. (c) Faradaic efficiency ( FE, %) for a certain product: It is the yield based on the electrical charge passed, i.e. the % of the total charge supplied that is used in forming the desired CO2 reduction product, so ideally it should be as close as possible to 100%. Also known as “current efficiency,” FE measures the selectivity of the process of reducing CO2 to a certain product, and it can be calculated using the equation: FE =
zFm × 100 Q
(36.3)
36.2 Fundamentals
where z is the number of electrons exchanged (e.g. z = 2 for the reduction of CO2 to HCOOH or CO and z = 6 for the reduction to CH3 OH; see Table 36.1), F is the Faraday constant, m is the number of moles of the certain product, and Q is the total charge passed. (d) Energetic efficiency ( EE, %): It is a measure of the overall energy utilization toward the desired CO2 reduction products, which can be defined by Eq. (36.4): EE =
E∘ × 100 E
(36.4)
∘
where E is the theoretical equilibrium cell potential and E represents the real potential applied in the process (which includes all the cell overpotentials). The EE describes the ratio between energy stored in the desired CO2 reduction product and the input energy needed to produce it, so a high EE is desirable since this means a small energy penalty of the process. 36.2.3
Role of Heterogeneous Catalysts
The electrochemical reaction occurs at the interface between the electrode (an electronic conductor) and the electrolyte (an ionic conductor). It is composed of a series of steps, including (i) the approach of the reactant species to the electrode surface, (ii) the reaction via heterogeneous electron transfer across the interface (actual electrochemical step), and (iii) the transfer of the reduced CO2 product species away from the reaction area into bulk solution [9]. Most of the proposed mechanisms for direct CO2 reduction on an electrode involve the initial formation of intermediate species. That species is usually called ⋅− CO⋅− 2 , although the exact structure of the species is unknown [10]. Hence, CO2 ⋅− does not necessarily represent a bare CO2 anion, but it denotes whatever intermediate species is formed when an electron is transferred to a CO2 molecule. The detailed mechanistic pathways for each product are not clear at present, and in many cases, several different schemes have been proposed [9]. We already mentioned that the actual potentials for CO2 reduction are much more negative than the values estimated from the thermodynamic data. Now we may understand the reason: it costs energy to create the CO⋅− 2 intermediate species (which requires highly negative potential for the formation), and extra energy has to be provided to get the reaction to occur. Figure 36.1 represents the qualitative reaction scheme for CO2 conversion. As shown in Figure 36.1, the overpotential required for the CO2 reduction reaction to proceed is much lower when catalysts are present (“𝜂 cat ”) than when no catalyst is involved in the reaction (“𝜂 no cat ”). Accordingly, catalysts play a key role by lowering the energy of the intermediates and therefore improving the EE of the CO2 conversion. Catalytic strategies thus aim to avoid or at least reduce the penalty of the high-energy intermediate CO⋅− 2 for obtaining the various products from reduction of CO2 at lesser energetic costs. Energetically more favorable strategies
653
36 Electro-, Photo-, and Photoelectro-chemical Reduction of CO2
Figure 36.1 Qualitative reaction scheme for CO2 conversion. Source: Whipple and Kenis 2010 [11]. Reproduced with permission of American Chemical Society.
CO2˙−
CO2˙−+ cat
Energy
654
ηno cat
ηcat CO + 1/2O2
E0 CO2 Reaction coordinate
consist of catalyzing the target reaction by means of a homogeneous or a heterogeneous catalyst. Homogeneous catalysis has essentially, but not exclusively, involved reduced states of transition metal complexes [12]. The Section 36.3 will briefly summarize the importance of heterogeneous catalysts on the electro-, photo-, and photoelectro-catalytic reduction of CO2 .
36.3 Innovative Technologies for CO2 Reduction 36.3.1
Electrochemical Reduction
36.3.1.1
How Does the Technology Work?
Figure 36.2 schematically shows the process that takes place in an electrochemical reactor when electrical energy is supplied to establish a potential between a cathode and an anode to allow CO2 to be electrochemically transformed into reduced forms. One of the advantages of this technology for CO2 valorization is that the conversion of CO2 to useful materials can be carried out under mild e–
Figure 36.2 Scheme of the CO2 electrochemical reduction process and different possible products generated in an electrochemical cell. Source: Zhang et al. 2018 [13]. https://onlinelibrary.wiley.com/ doi/10.1002/advs.201700275. Licensed under CCBY 4.0.
e– e–
e–
H+ H+
e–
e– H+
Anode CO2 CH4
Separator H2O C2H4
O2
CO
CH3OH
Cathode HCOOH
HCHO
CH3CH2OH
C2H6
36.3 Innovative Technologies for CO2 Reduction
conditions of temperature and pressure (normally ambient conditions). As the main goal is the reduction of CO2 , research efforts have obviously been focused on the cathodic compartment where CO2 can be converted to different products, simply leaving that in the anodic compartment a reaction like oxygen evolution (i.e. oxidation of water to give molecular oxygen) can take place without adversely affecting the reduction of CO2 . Different types of electrochemical reactors can be used. Undivided cells (where both the anode and the cathode are in the same compartment) can be useful for studies that aim at gaining fundamental understanding of the CO2 electroreduction process. Nevertheless, the most typical configuration is shown in Figure 36.2: an electrochemical cell in which cathode and anode are placed in different compartments separated by an ion conducting membrane, usually a cation-exchange membrane like Nafion. The separated two-compartment configuration prevents the oxidation in the anode of the desired products obtained from CO2 reduction in the cathode. The “H-type cells” (name derived from their typical “H” shape) have been popular two-compartment cells for studying CO2 electroreduction, but flow or filter-press-type reactors (in which the electrolytes flow through their own compartment) allow continuous operation and appear as a more suitable cell configuration for an industrial real application. Recent studies to evaluate the techno-economic feasibility of producing different chemicals through CO2 electroreduction indicated that CO and formic acid appear as the most economically viable products [14]. Important research has also been carried out to other products with more potential such as methanol or gaseous hydrocarbons, although their performance is still further from practical implementation [15]. 36.3.1.2
Key Factors Influencing Reaction Performance
The electrochemical reduction of CO2 is a promising and challenging process in which many different variables can have an influence on its performance. Undoubtedly, the catalyst used has a crucial influence on the process. As explained in the Section 36.2, catalysts are needed to reduce the high overpotentials, and they can induce the selective formation of desired products. In the last few decades, research efforts on the direct heterogeneous reduction of CO2 have mostly focused on different metal catalysts, showing that different products can be obtained depending on the metal used. For example, in aqueous media at ambient conditions, CO is the main product with metal electrodes like Au, Ag, or Zn [16]; Cu mainly yields mixtures of hydrocarbons (e.g. methane and ethylene [17]) and alcohols (e.g. methanol [18]); and other metals such as In, Sn, Hg, or Pb are selective for the production of formic acid/formate [19]. Over the last few years, researchers have also started to study other materials beyond bulk monometals, including metal oxides, bimetallic materials, metal–organic frameworks (MOFs), and organometallic or nonmetallic (e.g. nitrogen- or boron-doped diamond) catalysts, whose promising results represent new trends and opportunities in CO2 electroreduction [20, 21]. In catalysis, the Sabatier principle suggests that the interactions between a catalyst and the reactant should be neither too strong nor too weak. According to this principle, if the reaction rate is represented vs. a catalyst property, such as the energy of adsorption (which can be calculated by density functional theory
655
36 Electro-, Photo-, and Photoelectro-chemical Reduction of CO2
(a)
0.0 Pt3Co
–0.2
–1.2
O (g) →
*C H
→* CH *C O
Ex
pe
rim
en
Cu
ta l
Th
Pt3Sc Pt
Activity
–1.0
Pd
Pt
CO
–0.8
Ni Rh
Pt3Ni Pt3Y
O
H
–0.6
→* CO
–0.4
*C O
Limiting/onset potential (V vs RHE)
656
Au
er or et
ica
Sta
l Ag
–1.4 (111) surfaces 0.0 –1.5 –1.0 –0.5 *CO binding energy (eV)
bilit
(b)
y
ion rpt so ngth d A tre s
Figure 36.3 Volcano plots: (a) volcano plot for carbon dioxide reduction on metals. (b) A three-dimensional volcano scheme of metal alloy catalysts. Source: (a) Seh et al. 2017 [22]. Reproduced with permission of Royal Society of Chemistry; (b) Mayrhofer and Arenz 2009 [23]. Reproduced with permission of Springer Nature.
[DFT]), plots with the shape of a volcano should be obtained, such as those shown in Figure 36.3. Understanding how to control binding energies of reactive intermediates on a surface is the key to designing catalytic materials with improved performance. Therefore, the further development of volcano plots, such as those pioneered by Norskov’s group [22, 24], will help in the understanding of trends in electrocatalytic activity for CO2 reduction over different catalysts, and hence, they will be useful in the design of advanced, more active catalysts. Moreover, apart from the nature of the material used as catalyst, the configuration of the electrode has also a great influence on the performance of the CO2 electroreduction. For example, a certain metal can be used in electrodes with different forms, such as plates, granules, powders, or nanoparticles deposited on porous supports. Particularly, the configuration of gas diffusion electrodes (GDEs) has resulted to be especially successful. GDEs are usually consist of a carbonaceous support, a microporous layer, and a catalytic layer formed by metal nanoparticles, as represented in Figure 36.4. The improved performance of GDEs can be attributed to an enhancement of the three-phase boundary area between the solid catalyst, the gaseous reactants, and the electrolyte, which avoids mass transfer limitations, and therefore, allows working at higher current densities with high FEs [26]. In addition to the catalyst, the electrolyte used has also a great influence on CO2 electroreduction. The vast majority of studies on electrochemical reduction of CO2 have been carried out in aqueous media. However, due to the low solubility of CO2 in water and the presence of the competing side reaction of hydrogen evolution, attempts in other nonaqueous media have also been reported. Propylene carbonate, acetonitrile, dimethylformamide (DMF), or dimethyl sulfoxide (DMSO) are aprotic solvents, and therefore hydrogen evolution is greatly suppressed in them [10]. Using the same catalyst, the major products from CO2 reduction are different in a nonaqueous electrolyte than in an aqueous electrolyte (for more details, see, e.g. [27]). Using aprotic solvents, typical products obtained
36.3 Innovative Technologies for CO2 Reduction
KHCO3 + KCI + HCOO–
Nafion 117
KOH
CO2 KHCO3 + KCI
GDE
Reference electrode
DSA-O2 KOH
Carbon paper Microporous layer Catalytic layer
Figure 36.4 Filter-press electrochemical cell with a GDE configuration. Source: Del Castillo et al. 2017 [25]. Reproduced with permission of Elsevier.
do not need protons to take part in CO2 reduction (e.g. CO or oxalates), but when protons are needed for the reaction (e.g. for formic acid, alcohols, etc.), small amounts of water may be added. Interestingly, in recent years the use of ionic liquids (a family of compounds – organic salts that are liquid below 100 ∘ C – with unique properties) has allowed remarkable improvements in CO2 electroreduction [28, 29]. The electrochemical reduction of CO2 in gas phase can be carried out using a solid polymer electrolyte instead of a liquid electrolyte. Ion-exchange membranes (e.g. Nafion cation-exchange membranes or Selemion anion-exchange membranes) coated with metal catalyst have been mainly used for the reduction of CO2 to gaseous products like CO, CH4 , or C2 H6 [30]. 36.3.1.3
Main Challenges
Significant progress has been achieved in recent years, as denoted by the rapidly increasing number of research publications, but the maturity of CO2 electrochemical technology has yet to reach the requirements for commercialization. Key challenges that will have to be tackled in this exciting area of research include: (a) Improved and/or novel electrocatalytic materials: The great research efforts on research on the catalysis of the electroreduction of CO2 during the last decade have allowed significant advances, although improvements in electrocatalysts are definitely necessary. An ideal electrocatalyst should combine: (i) high FE to the desired CO2 reduction product (which means that a high % of the electrical charge supplied is used in the desired product, and therefore it implies that the selectivity of the process is high); (ii) high formation rates of the product (measured by the current density, the electric current flow divided by the area of the electrode, at which the process can be operated); and (iii) a low overpotential for CO2 reduction to allow an energy efficient process (a lower overpotential means that the real potential applied in the process will be closer to the equilibrium potential, so the energy penalty of the process will be lower).
657
658
36 Electro-, Photo-, and Photoelectro-chemical Reduction of CO2
Future research will definitely continue tackling the challenge of optimizing all these figures of merit (i.e. FE, j, EE, etc.) at the same time, so improving one is not at the expense of making other worse. The still insufficient stability and durability of the materials used as electrodes can also be highlighted as another important challenge to be addressed for economically viable industrial processes. Moreover, further understanding of reaction mechanisms will indeed help in the design of better electrocatalytic materials. Due to the complexity of the reaction environment and multiple bond-forming and bond-breaking processes, there is a lack of understanding of the CO2 electroreduction mechanism, pathways, and intermediates. The determination of the transient catalytic intermediates is a real challenge due to their very short lifetimes and low concentrations. Coupled to advanced methods of experimental analysis (e.g. scanning tunneling microscope [STM], electron paramagnetic resonance [EPR], Fourier transform infrared spectroscopy [FTIR], etc.), the use of first principles computational techniques, such as DFT, will be helpful for improving the understanding of how the CO2 electroreduction reaction proceeds on catalytic materials. Such theoretical modeling techniques could also guide the development of new catalysts with improved performance. As explained before, further development of Norskov’s volcano plot will be useful as a rational platform to understand trends in activity of different catalysts, helping to design better and more active electrocatalytic materials for CO2 electroreduction. (b) Improved/novel electrolytes: An interesting way of improvement is related to the nature of the solvent used as liquid electrolyte, which, together with the electrocatalyst, plays a crucial role in the CO2 electroreduction. The use of aqueous media presents limitations, especially due to the concomitant side reaction of hydrogen evolution that takes place at similar thermodynamic potential. Therefore, relevant research efforts point to the use of new solvents, like ionic liquids, as a promising way of enhancing the performance. Extending the application of solid polymer electrolytes to avoid the use of liquid electrolytes that dilute liquid products of CO2 electroreduction may also open opportunities for improvement. (c) More efficient electrocatalytic reactors: As in any electrochemical process, the design of the electrochemical reactor for CO2 conversion is crucial for a successful performance at a large scale. Undoubtedly, future work will have to bridge the gap between laboratory-scale reactors and industrial reality. Engineering research should focus on the design and scale-up of electrochemical reactors that are able to operate in a continuous mode for the industrial implementation of economically viable CO2 electroreduction processes. 36.3.2 36.3.2.1
Photochemical Reduction How Does the Technology Work?
Since the first report on photocatalysis by Fujishima and Honda in 1972 [31], the research community has been attempting to mimic nature and pursue the spontaneous transformation of atmospheric CO2 and H2 O to chemicals using sunlight as the sole energy input and over the same active surface, in contrast
36.3 Innovative Technologies for CO2 Reduction
CO2 Diffusion adsorption CO2, H+ Reduction CH4,CO, Desorption CH3,CO, Diffusion etc.
e– e– CB e–
–
e
e– VB h+
h+
Volume recombination
Surface recombination
+
h
h+
H +, O 2
h+
Oxidation H2O
Figure 36.5 Photoinduced generation of electron–hole pairs in CO2 reduction with H2 O. Source: Wu et al. 2017 [5]. https://onlinelibrary.wiley.com/doi/10.1002/advs.201700194. Licensed under CCBY 4.0.
with the electroreduction process. Although the efficiencies reported are still low [32], the artificial photosynthesis is believed to have a great potential to enable a shift to a sustainable energy economy and chemical industry. The process starts with the adsorption of CO2 in the photocatalyst surface, where the molecule is susceptible to reduction by photogenerated electrons. Under the irradiation of light source, the electrons are excited from the valence band (VB) to the conduction band (CB), and, simultaneously, an equal number of holes (which serve as a positive charge carrier) are formed in the CB due to the absorption of photons from the incident light. These electron–hole pairs separate from each other, producing the reduction of CO2 molecule and the oxidation of H2 O (releasing H+ ) at the surface of a semiconductor material, as represented in Figure 36.5. In order for these photogenerated electrons and holes to be energetically favorable to reduce CO2 and oxidize H2 O, photocatalysts should have a suitable band structure. Their CB edge must be more negative than the redox potential of CO2 reduction, and the VB edge should be more positive than the redox potential of water oxidation (+0.82 V vs. normal hydrogen electrode [NHE] in pH = 7). After reduction, the products need to desorb and diffuse to accomplish the whole conversion process. The redox potential levels of the adsorbate species and the band gap energy determine the likelihood and rate of the charge transfer processes for electrons and holes [33], resulting in different products (Table 36.1). The general criterion to evaluate the performance for the photocatalytic conversion of CO2 is to measure the concentration of a certain product, n, produced within a period, t, under light irradiation per gram of catalyst, m, according to Eq. (36.5). A common unit for this formation rate, r, is μmol⋅gcat. −1 ⋅h−1 : r=
n t⋅m
(36.5)
659
660
36 Electro-, Photo-, and Photoelectro-chemical Reduction of CO2
Another relevant parameter to evaluate the photocatalytic performance is the effectivity of the catalyst to convert light into chemical energy. The efficiency of light absorption, charge separation, and surface redox reaction can be expressed by the apparent quantum yield, AQY (%), defined as the rate of electrons transferred toward a certain product per rate of incident photons on the photoelectrode surface. The AQY can be comparable with the FEs used in electrocatalysis and is defined as: n −R AQY = e × 100 (36.6) I where ne− , R, and I denote the number of electrons involved in the photocatalytic reaction, the molecular production rate, and the rate of incident photons, respectively. 36.3.2.2
Key Factors
The performance of a photocatalysis system for CO2 conversion is mainly related to the photoactive material and photoreactor design applied. An ideal semiconductor photocatalyst may include an adjusted band gap and proper position of CB and VB edges to utilize visible-light range (which correspond to about 50% of the energy received at the Earth’s surface), together with an enhanced light harvesting, suitable electron–hole separation, large surface area, favorable surface chemistry/morphology, and sufficient stability. For reasons that include affordability and stability, titanium dioxide (TiO2 ), mainly in its anatase and rutile forms, has been the most widely used semiconductor in photocatalysis. Unfortunately, after an intensive research in this area, it appears that TiO2 possesses important drawbacks as photoactive material for CO2 conversion since it can mainly absorb in the UV part of the spectrum (𝜆 < 390 nm) and it presents a poor electron–hole pair separation [34]. Various strategies have been followed by the research community to manipulate TiO2 band gap and enhance light utilization [35], including doping with metal/nonmetal catalysts, surface decoration, dye sensitization, heterojunction formation, or structuralization, among others, as described in detail in other chapters in this book. These techniques may promote electron–hole pair separation and CO2 adsorption and activation, enhancing CO2 photoconversion efficiency to desired products. Particular interest has been lately directed toward the use of low-dimensional structured materials, composed of ultrafine particles, as photocatalysts primarily for their tunable optical and electronic properties due to quantum size effects as well as their enhanced photoelectric performance compared with bulk materials [36]. Nonetheless, the charge separation and selectivity of TiO2 -based materials are still low, even after applying in combination several of the aforementioned strategies. Therefore, it is of crucial importance to explore the performance of alternative materials to titanium-based semiconductors. For example, silicon carbide (SiC), copper oxides (Cu2 O), gallium (Ga), or cadmium sulfide (CdS)-based compounds may show better light harvesting properties than TiO2 and may be good candidates to substitute it [37]. Figure 36.6 shows common semiconductor materials that may be applicable for the photoreduction of CO2 , attending to their CB and VB edge positions. Besides, efficient photoreactor designs for real-world applications need to consider various factors, including (i) light source and geometrical configuration, to
36.3 Innovative Technologies for CO2 Reduction
–3
Potential vs.NHE (V)
–2 CO2/HCOOH (–0.61 V)
–1 Si
0 CdSe
+1
SrTiO3
+2
TiO2(A)
CO2/CH4 (–0.24 V) H2O/O2 (0.82 V)
BiVO4 TiO2(R)
GaP Cu2O TaON ZnS SiC Ta3N5 C3N4 CdS
CO2/HCHO (–0.48 V) 2H+/H2 (–0.41 V) CO2/CH4HO (–0.38 V)
Zno
+3 (pH = 7)
Figure 36.6 Band gap energies for common semiconductors materials relative to the redox potentials of CO2 reduction products at pH = 7. Source: Li et al. 2014 [38]. Reproduced with permission of Springer Nature.
have uniform light distribution throughout the entire system; (ii) construction material, limited by requirements of light transmission; (iii) heat exchange, to remove the heat generated by the lamp; (iv) mixing and flow characteristics, to ensure an appropriate contact between reactants, electrons/holes, and catalysts; and (v) phases involved and mode of operation [39]. A number of different laboratory photoreactor designs have been assessed for the photocatalytic reduction of CO2 . Normally, batch-type reactors (with the catalyst introduced as powder on the bottom of the cell) and slurry reactors (in which light scattering occurs) have been applied. These simple configurations have been demonstrated to be inefficient to induce the CO2 conversion to more reduced species, due to a low surface area-to-volume ratio produced by particle agglomeration, and the required separation of the photocatalyst material from the products obtained. Fixed reactors are also applied, although they present low specific area and light scattering, leading to poor light utilization [40]. The agglomeration phenomena occurring in the aforementioned configurations may be partially solved by the use of reactors equipped with coated catalysts (e.g. optical fiber technology and internally illuminated reactor) [41], although their optical properties vary for different coating strategies that lead to different particles sizes and porosities. In any case, the application of these reactor designs is still scarce, and further developments are needed. Figure 36.7 shows a representation of slurry, optical fiber, and internally illuminated reactors. 36.3.2.3
Main Challenges
The most relevant issues that need to be tackled, together with a further exploration of CO2 photoreduction mechanisms, are: (a) Innovative photoactive materials: The research efforts will require a multidisciplinary approach to develop novel photocatalytic materials with higher stability, selectivity, and efficiency toward products under visible-light irradiation. It should take the fullest advantage of new nanoscale structures
661
662
36 Electro-, Photo-, and Photoelectro-chemical Reduction of CO2 Incident light rays Flow inlet
Flow outlet Flow outlet
Incident light rays
Catalyst coated optical fibers
Suspended catalyst
(b)
(a)
Flow inlet
Incident light rays Light propagating optical fibers Flow inlet
(c)
Catalyst coated monolithic channels
Flow outlet
Figure 36.7 Scheme for a (a) slurry reactor with top illumination, (b) optical fiber reactor with side illumination, and (c) internally illuminated reactor with top illumination. Source: Oluwafunmilola and Maroto-Valer 2015 [39]. https://www.sciencedirect.com/science/article/ pii/S1389556715000271. Licensed under CCBY 4.0.
with large surface area, mixed crystal phase, defect disorders, material modifications, improved charge separation, and directional electron transfers. The breakthrough seems to be in the synthesis of non-TiO2 materials with improved physicochemical properties. Metal-free cocatalysts are also under development to help structuring robust artificial photocatalysts. In particular, carbonaceous materials including carbon nitride, graphitic carbon nitride (g-C3 N4 ), graphene, and carbon nanotubes (CNTs) show great potential to replace some of the existing metal catalysts, providing an enhanced electrical and thermal conductivity, light harvesting, surface CO2 adsorption capacity, and low cost. MOFs may also represent an opportunity in the photocatalytic reduction of CO2 . The integration of active semiconductors with bio-based materials presents also promise for an enhanced transformation of CO2 . (b) Efficient photocatalytic reactors: The research efforts must consider important parameters such as reactor geometry, flow rate, irradiation source, or catalyst mass. Modeling of the effect of operation parameters on CO2 reduction is also required in order to extrapolate results and design a pilot-scale system. Besides, considering the limited rates of CO2 photocatalytic reduction to products, the recirculation of unconverted products is also advised. These aspects may have an important implication on both economy and efficiency of the overall process. 36.3.3 36.3.3.1
Photoelectrochemical Reduction How Does the Technology Work?
The photoelectrocatalytic reduction of CO2 was initially reported by Halmann in 1978 using a GaP semiconductor as photoelectrode in aqueous solution [42]. Since then, photoelectrochemistry and the development of efficient photoelectrocatalysts for an efficient reduction of CO2 have received much attention [43].
36.3 Innovative Technologies for CO2 Reduction
Basically, a photoelectrochemical system is similar to a standard electrochemical cell, except that light is explicitly involved in providing energy to cause the redox reactions to proceed. In principle, such integration reduces the system capital cost and enables higher efficiency by reducing loses in transporting electricity to the electrolysis cell, eliminating current collectors, and interconnections between devices. Compared with photocatalysis, the applied bias can cause band bending and help the oriented transfer of the photogenerated electrons, decreasing the recombination of the photogenerated electron–hole pairs. Besides, photocatalytic materials with unfavorable band positions for CO2 reduction and H2 O oxidation can be still used in photoelectrocatalytic systems when applying an external bias. In addition, the obtained redox products can be readily separable in two-compartment photoelectrochemical cells. Overall, the options of photoelectrocatalysis may be broader than photocatalysis for CO2 transformation. At the photoelectrodes, light is absorbed by a semiconductor material, exciting electrons to a higher energy level. These electrons and the corresponding holes generated are capable of carrying out redox reactions at the semiconductor/liquid interface. The photoelectrochemistry is therefore a multidisciplinary field involving surface science, electrochemistry, solid-state physics, and optics. The efficiency of a CO2 photoelectrochemical reduction system is commonly expressed as the product formation rate, r (Eq. (36.5)), and the percentage of light energy input converted to chemical energy output, AQY (Eq. (36.6)) or FE (Eq. (36.3)), since generally an external bias potential is required. 36.3.3.2
Key Factors
The development of high-efficiency photoelectrodes that satisfy the requirements of real applications entails the processing of materials with enhanced solar energy conversion efficiency, durability, and low cost. Photoelectrocatalysis normally exploits semiconductor materials for CO2 conversion including inorganic binary compounds, such as TiO2 , ZnO, GaP CdS, and SiC [44]. As it is for photocatalysis, TiO2 -based materials with various structures (e.g. nanocrystals, nanocomposites, nanotube arrays, etc.) are the most investigated photoelectrocatalysts. The same strategies followed in photocatalysis to overcome the important limitations of TiO2 (see Section 36.3.2.2) are of application in the development of materials for photoelectrocatalysis. Research efforts have also been focused on multielement metal oxides (e.g. Bi2 WO6 , ZnGaNO, Ca2 Nb2 TaO10 , or MnCo2 O4 ), as their band gaps and band edge positions can be rationally tuned and also metal complexes, which may exhibit excellent CO2 adsorption ability. Moreover, metal-free compounds (e.g. graphitic g-C3 N4 or MOFs) give good hopes to improve the photoelectrochemical conversion of CO2 . For a practical use of photoelectrochemical cells, anode and cathode should be separated by an ion-exchange membrane in a dual-compartment cell in order to allow an efficient separation/transport of charges, dealing with a reduction–oxidation reaction in each compartment and enhancing the separation of products. Gas-phase operation on the cathode compartment is also preferable in order to avoid the limitations of CO2 solubility [45]. Research
663
664
36 Electro-, Photo-, and Photoelectro-chemical Reduction of CO2
on CO2 reduction to liquid products has mostly focused on batch photoelectrochemical reactors, but production of valuable chemicals on an industrial scale will likely require continuous flow reactors to minimize capital costs and maximize product consistency. Moreover, there are different electrode configurations for photoelectrochemical systems depending on which electrode (i.e. anode, cathode, or both) acts as photoelectrode [44] as shown in Figure 36.8. (a) Photocathode–dark anode: The most common configuration usually employs a photocathode made of a p-type semiconductor such as those based on Cu (e.g. Cu2 O, CuO, Cu2 ZnSnS4 , etc.), Co3 O4 , InP, or CdSeTe, among others, due e– CO2 O2
Fuel
O2 H+ h+
V
RE
CO2+H+
e–
+
H
CH4,CO, CH3OH, HCOOH
H2O
Anode
h+
Photocathode
(a) e–
V
CO2 O2
e–
Fuel
RE
CO2+H+ H+
O2 H+
e–
CH4,CO, CH3OH, HCOOH
h+ H2O
Photoanode
Cathode
(b)
e– CO2 O2
Fuel
RE
CO2+H+
e– O2 +
H h+
H2O
Photoanode
V
e–
H+ CH4,CO, CH3OH, HCOOH
h+
Photocathode
(c)
Figure 36.8 Representation of a (a) photocathode–dark anode, (b) photoanode–dark cathode, and (c) photocathode–photoanode electrode configurations. Source: Xie et al. 2016 [44]. Reproduced with permission of Royal Society of Chemistry.
36.3 Innovative Technologies for CO2 Reduction
to their high CB energy suitable for CO2 reduction and a metallic anode for H2 O oxidation. However, CO2 reduction on p-type semiconductors requires a high bias potential since their VB are not sufficiently positive to oxidize H2 O. These materials are also expensive, toxic, or unstable during the reaction, and two-electron compounds such as CO and formic acid are usually the main products of CO2 reduction. Moreover, combination of a photocathode with a cocatalyst able to activate CO2 molecules is needed since p-type semiconductors do not act as a true catalyst for the activation of CO2 molecules, but just as light harvester to generate electrons and holes. (b) Photoanode–dark cathode: The use of n-type semiconductors such as TiO2 , ZnO, BiVO4 , or WO3 , which are earth abundant, cheap, and stable, as photoanode for H2 O oxidation and a metallic electrocatalyst that is active for CO2 reduction as the cathode is also an attractive alternative. This configuration allows reducing the external electric bias over an electrochemical cell configuration, since the voltage generated by the light in the anode supplies an extra negative potential to the cathode for CO2 reduction. It also provides active protons through a proton-exchange membrane toward the cathode side, which plays an important role for CO2 reduction. (c) Photocathode–photoanode: The combination of a photocathode made of a p-type semiconductor for CO2 reduction with a photoanode made of a n-type semiconductor for H2 O oxidation can also be seen. Ideally, this configuration allows an efficient CO2 reduction without external voltage by using two appropriate n-type and p-type semiconductors to form multi-semiconductor systems, in which the materials have matched band gap positions to transfer electrons and holes. In some cases, the system may need an extra electrical bias to facilitate charge separation and overcome parasitic loses and reaction overpotentials. Some recent reports also consider photovoltaic (PV) cell tandem devices, where PV panels are coupled to photoelectrochemical cells to supply the voltage required for CO2 reduction/H2 O splitting [46]. 36.3.3.3
Main Challenges
Photoelectrocatalysis combines the advantages of both electrocatalysis and photocatalysis. In spite of that, the efficiencies of the systems presented to date are modest. The challenges are basically those mentioned above for electro- and photo-catalysis for CO2 transformation, with some particularities. (a) Photoactive materials: Most of the n-type semiconductors applied suffer photocorrosion during the process, and modification of photoelectrodes using surface coatings (e.g. TiO2 , Al2 O3 , CNT, etc.) could be an effective way to improve the stability for CO2 photoelectroreduction. In addition, Z-scheme junction systems with semiconductors with matched CB and VB edge positions (as both photocathodes and photoanodes) can lead to systems with facilitated charge transfer to promote its reduction efficiency and selectivity and conduct the reduction of CO2 and oxidation of H2 O simultaneously without (or with reduced) external applied bias. Besides, the development of more efficient cocatalysts with strong adsorption and
665
666
36 Electro-, Photo-, and Photoelectro-chemical Reduction of CO2
activation capacity of CO2 will lead to the selective production of desirable chemicals in the photocathode. The development of materials with high conductivity for proton-exchange membrane is also key for the process. (b) Photocatalytic reactors: The systems reported have evolved from photocathode to photoanode-driven cells, but more promising strategies such as two-compartment bias-free photocathode–photoanode systems and photoelectrochemical cells coupled with PV panels in stand-alone tandem cells should receive more attention to demonstrate their feasibility in the practical application of CO2 reduction processes under sunlight illumination.
36.4 Concluding Remarks The conversion of CO2 into useful products is challenging because CO2 is the most stable carbon-based structure in nature. Electro- and photoelectro-chemical approaches appear as attractive options for CO2 reduction, especially when they are coupled to intermittent renewable energy, where there is a very interesting future possibility to use the excess electric energy from renewable sources, like solar or wind energy, to reduce CO2 to value-added products and/or fuels. The products could then be used in fuel cells or in industrial processes, when and where needed. More potential for a shift toward a low-carbon economy presents the photochemical approach, where CO2 reduction and H2 O oxidation occurs at the same catalytic surface using sunlight as the sole energy input. The number of interesting studies, especially in the field of photo- and photoelectro-chemical reduction of CO2 , has grown rapidly. As highlighted in this chapter, promising results on electrocatalytic CO2 conversion to different chemicals and fuels have already been achieved, but further research efforts are still needed before these technologies can reach practical commercialization. The challenges discussed in this chapter, including the development of catalytic materials for improved performance of CO2 reduction, can be seen as exciting opportunities of research in this vibrant field of CO2 conversion.
Acknowledgments The authors gratefully acknowledge the financial support from the Spanish Ministry of Economy and Competitiveness (MINECO), under the projects CTQ2013-48280-C3-1-R and CTQ2016- 76231-C2-1-R, and PID2019-104050RAI00, as well as Ramón y Cajal programme (RYC-2015-17080).
References 1 International Energy Agency (IEA). (2017). Key World Energy Statistics 2017. 2 Boot-Handford, M.E., Abanades, J.C., Anthony, E.J. et al. (2014). Carbon cap-
ture and storage update. Energy Environ. Sci. 7: 130–189.
References
3 Sanz-Perez, E.S., Murdock, C.R., Didas, S.A., and Jones, C.W. (2016). Direct
capture of CO2 from ambient air. Chem. Rev. 116: 11840–11876. 4 Aresta, M., Dibenedetto, A., and Angelini, A. (2014). Catalysis for the val-
5 6
7 8 9 10
11 12 13
14
15
16
17
18
19
orization of exhaust carbon: from CO2 to chemicals, materials, and fuels. Technological use of CO2 . Chem. Rev. 114 (3): 1709–1742. Wu, J., Huang, Y., Ye, W., and Li, Y. (2017). CO2 reduction: from the electrochemical to photochemical approach. Adv. Sci. 4: 1700194. Lower, S. (2018). General Chemistry Virtual Textbook. Chem1 Introduction to Electrochemistry. http://www.chem1.com/acad/webtext//elchem/ec-1.html (accessed 29 October 2018). Bard, A.J. and Faulkner, L.R. (2001). Electrochemical methods: Fundamentals and Applications, 2e. New York, NY: John Wiley & Sons. Pletcher, D. and Walsh, F.C. (1993). Industrial Electrochemistry, 2e. London, UK: Chapman and Hall. Qiao, J., Liu, Y., and Zhang, J. (eds.) (2016). Electrochemical Reduction of Carbon Dioxide: Fundamentals and Technologies. Boca Raton, FL: CRC Press. Masel, R., Liu, Z., Zhao, D. et al. (2016). CO2 conversion to chemicals with emphasis on using renewable energy/resources to drive the conversion. In: Commercializing Biobased Products: Opportunities, Challenges, Benefits, and Risks, RSC Green Chemistry No. 43 (ed. S.W. Snyder), 215–257. Royal Society of Chemistry (RSC). Whipple, D.T. and Kenis, P.J.A. (2010). Prospects of CO2 utilization via direct heterogeneous electrochemical reduction. J. Phys. Chem. Lett. 1: 3451–3458. Costentin, C., Robert, M., and Savéant, J.M. (2013). Catalysis of the electrochemical reduction of carbon dioxide. Chem. Soc. Rev. 42: 2423–2436. Zhang, W., Hu, Y., Ma, L. et al. (2018). Progress and perspective of electrocatalytic CO2 reduction for renewable carbonaceous fuels and chemicals. Adv. Sci. 5: 1700275. Verma, S., Kim, B., Jhong, H.R.M. et al. (2016). A gross-margin model for defining technoeconomic benchmarks in the electroreduction of CO2 . ChemSusChem 9: 1972–1979. Merino-Garcia, I., Alvarez-Guerra, E., Albo, J., and Irabien, A. (2016). Electrochemical membrane reactors for the utilisation of carbon dioxide. Chem. Eng. J. 305: 104–120. Zhu, D.D., Liu, J.L., and Qiao, S.Z. (2016). Recent advances in inorganic heterogeneous electrocatalysts for reduction of carbon dioxide. Adv. Mater. 28: 3423–3452. Merino-Garcia, I., Albo, J., and Irabien, A. (2018). Tailoring gas-phase CO2 electroreduction selectivity to hydrocarbons at Cu nanoparticles. Nanotechnology 29: 014001. Albo, J., Alvarez-Guerra, M., Castaño, P., and Irabien, A. (2015). Towards the electrochemical conversion of carbon dioxide into methanol. Green Chem. 17: 2304–2324. Du, D., Lan, R., Humphreys, J., and Tao, S. (2017). Progress in inorganic cathode catalysts for electrochemical conversion of carbon dioxide into formate or formic acid. J. Appl. Electrochem. 47: 661–678.
667
668
36 Electro-, Photo-, and Photoelectro-chemical Reduction of CO2
20 Jhong, H., Ma, S., and Kenis, P.J. (2013). Electrochemical conversion of CO2
21 22
23 24 25
26
27
28
29
30
31 32
33 34 35
36
to useful chemicals: current status, remaining challenges, and future opportunities. Curr. Opin. Chem. Eng. 2: 191–199. Lu, Q. and Jiao, F. (2016). Electrochemical CO2 reduction: electrocatalyst, reaction mechanism, and process engineering. Nano Energy 29: 439–456. Seh, Z.W., Kibsgaard, J., Dickens, C.F. et al. (2017). Combining theory and experiment in electrocatalysis: insights into materials design. Science 355: eaad4998. Mayrhofer, K.J.J. and Arenz, M. (2009). Log on for new catalysts. Nat. Chem. 1: 518–519. Liu, X., Xiao, J., Peng, H. et al. (2017). Understanding trends in electrochemical carbon dioxide reduction rates. Nat. Commun. 8: 15438. Del Castillo, A., Alvarez-Guerra, M., Solla-Gullón, J. et al. (2017). Sn nanoparticles on gas diffusion electrodes: synthesis, characterization and use for continuous CO2 electroreduction to formate. J. CO2 Util. 18: 222–228. Irabien, A., Alvarez-Guerra, M., Albo, J., and Domínguez-Ramos, A. (2018). Electrochemical conversion of CO2 to value-added products. In: Electrochemical Water and Wastewater (eds. C.A. Martínez Huitle, M.A. Rodrigo and O. Scialdone). Elsevier. ISBN: 9780128131602. Sánchez-Sánchez, C.M., Montiel, V., Tryk, D.A. et al. (2001). Electrochemical approaches to alleviation of the problem of carbon dioxide accumulation. Pure Appl. Chem. 73: 1917–1927. Rosen, B.A., Salehi-Khojin, A., Thorson, M.R. et al. (2011). Ionic liquid-mediated selective conversion of CO2 to CO at low overpotentials. Science 334: 643–644. Alvarez-Guerra, M., Albo, J., Alvarez-Guerra, E., and Irabien, A. (2015). Ionic liquids in the electrochemical valorisation of CO2 . Energy Environ. Sci. 8: 2574–2599. Hori, Y. (2016). CO2 reduction using electrochemical approach. In: Solar to Chemical Energy Conversion, Lecture Notes in Energy 32 (eds. M. Sugiyama, K. Fujii and S. Nakamura). Cham, Switzerland: Springer International Publishing. ISBN: 978-3-319-25400-5. Fujishima, A. and Honda, K. (1972). Photolysis of water at a semiconductor electrode. Nature 238: 37–38. Sohn, Y., Huang, W., and Taghipour, F. (2017). Recent progress and perspectives in the photocatalytic CO2 reduction of Ti-oxide-based nanomaterials. Appl. Surf. Sci. 396: 1696–1711. Yuan, L. and Xu, Y.-J. (2015). Photocatalytic conversion of CO2 into value-added and renewable fuels. Appl. Surf. Sci. 342: 154–167. Chen, X. and Mao, S.S. (2017). Titanium dioxide nanomaterials: synthesis, properties, modifications, and applications. Chem. Rev. 107 (7): 2891–2959. Tu, W., Zhou, Y., and Zou, Z. (2014). Photocatalytic conversion of CO2 into renewable hydrocarbon fuels: state-of-the-art accomplishment, challenges, and prospects. Adv. Mater. 26 (27): 4607–4626. Voiry, D., Suk Shin, H., Ping Loh, K., and Chhowalla, M. (2018). Low-dimensional catalysts for hydrogen evolution and CO2 reduction. Nat. Rev. Chem. 2: 0105.
References
37 Navalón, S., Dhakshinamoorthy, A., Alvaro, M., and Garcia, H. (2013). Pho-
38
39
40
41
42
43
44
45
46
tocatalytic CO2 reduction using non-titanium metal oxides and sulfides. ChemSusChem 6: 562–577. Li, X., Wen, J., Low, J. et al. (2014). Design and fabrication of semiconductor photocatalyst for photocatalytic reduction of CO2 to solar fuel. Sci. China Mater. 57: 70–100. Oluwafunmilola, O. and Maroto-Valer, M.M. (2015). Review of material design and reactor engineering on TiO2 photocatalysis for CO2 reduction. J. Photochem. Photobiol. C: Photochem. Rev. 24: 16–42. Roupp, G.B., Nico, J.A., Annangi, S. et al. (1997). Two-flux radiation-field model for an annular packed-bed photocatalytic oxidation reactor. AlChE J. 43: 792–801. Tahir, M. and Amin, N.A.S. (2013). Photocatalytic CO2 reduction with H2 O vapors using montmorillonite/TiO2 supported microchannel monolith photoreactor. Chem. Eng. J. 230: 314–327. Halmann, M. (1978). Photoelectrochemical reduction of aqueous carbon dioxide on p-type gallium phosphide in liquid junction solar cells. Nature 275: 115–116. Wang, P., Wang, S., Wang, H. et al. (2018). Recent progress on photo-electrocatalytic reduction of carbon dioxide. Part. Part. Syst. Char. 35: 1700371. Xie, S., Zhang, Q., Liu, G., and Wang, Y. (2016). Photocatalytic and photoelectrocatalytic reduction of CO2 using heterogeneous catalysts with controlled nanostructures. Chem. Commun. 52: 35–59. Ampelli, C., Centi, G., Passalacqua, R., and Perathoner, S. (2010). Synthesis of solar fuels by novel photoelectrocatalytic approach. Energy Environ. Sci. 3: 292–301. Jang, Y., Jeong, I., Lee, J. et al. (2016). Unbiased sunlight-driven artificial photosynthesis of carbon monoxide from CO2 using a ZnTe-based photocathode and a perovskite solar cell in tandem. ACS Nano 10: 6980–6987.
669
671
37 Photocatalytic Abatement of Emerging Micropollutants in Water and Wastewater Lan Yuan 1,2 , Zi-Rong Tang 2 , and Yi-Jun Xu 1,2 1 Fuzhou University, State Key Laboratory of Photocatalysis on Energy and Environment, College of Chemistry, Xueyuan Road No. 2, University New Area, Fuzhou, 350116, PR China 2 Fuzhou University, College of Chemistry, New Campus, Xueyuan Road No. 2, University New Area, Fuzhou, 350116, PR China
37.1 Introduction The rapid development of manufacturing technology after the industrial revolution has improved the standards of living significantly but also threatened human health and the environment. Micropollutants discharged by human activities into the aquatic environment, including wastewater, surface water, groundwater, and drinking water, has become a serious global issue [1, 2]. It has been estimated that around 4 billion people have no or little access to clean and sanitized water supply worldwide, and millions of people die due to severe waterborne diseases every year. These statistical figures are expected to continue to grow in the near future [3, 4]. Micropollutants are commonly present in waters at trace concentrations. The “low concentration” and diversity of micropollutants not only complicate the associated detection and analysis procedures but also create challenges for water and wastewater treatment processes [5, 6]. It is an imperative task to develop newer eco-friendly methods to treat micropollutants in water and wastewater at a lower cost and with less energy [7, 8]. For treating micropollutants, traditional physical techniques (adsorption on activated carbon, ultrafiltration, reverse osmosis, coagulation by chemical agents, ion exchange on synthetic adsorbent resins, etc.) can generally be used efficiently [1, 2, 6]. Nevertheless, they are nondestructive, since they just transfer organic compounds from water to another phase, thus potentially causing secondary pollution. Consequently, regeneration of the adsorbent materials and posttreatment of solid wastes are needed, which are expensive operations [8]. In contrast, recently, heterogeneous photocatalysis using semiconductor photocatalysts is emerging as a destructive and green technology, which has received intense attention and been widely applied to the abatement of water micropollutants [4, 7]. The unique advantages of heterogeneous photocatalysis for the aquatic environment treatment include the following: (i) it is driven by sunlight, a completely renewable source of energy; (ii) it can occur under Heterogeneous Catalysts: Advanced Design, Characterization and Applications, First Edition. Edited by Wey Yang Teoh, Atsushi Urakawa, Yun Hau Ng, and Patrick Sit. © 2021 WILEY-VCH GmbH. Published 2021 by WILEY-VCH GmbH.
672
37 Photocatalytic Abatement of Emerging Micropollutants in Water and Wastewater
mild conditions (room temperature and atmospheric pressure); (iii) it does not involve mass transfer and thus not causing secondary pollution; and (iv) semiconductor photocatalysts such as TiO2 are abundant, cheap, and nontoxic and show relatively high chemical stability [4, 7]. The utilization of combined photocatalysis and other solar technologies may be developed to reduce water pollution. This chapter first provides an overview of the basic principles of heterogeneous photocatalysis for micropollutant abatement in water and wastewater, followed by the introduction of different kinds of micropollutants and examples of those degradable by photocatalysis. For the low efficiency of traditional photocatalysts in micropollutant abatement, available strategies for the efficiency enhancement of the reactions are outlined. Finally, the future research challenges and prospects are concisely discussed.
37.2 Main Processes for Photocatalytic Abatement of Micropollutants in Water and Wastewater In general, micropollutants consist of a vast and expanding array of anthropogenic as well as natural substances, which include pharmaceuticals, personal care products, surfactants, industrial chemicals, pesticides, heavy metals, and many other emerging compounds [1, 2, 6]. The categories and typical examples that are degradable by photocatalysis are listed in Table 37.1. As for classical heterogeneous catalysis, the overall process can be divided into five independent steps: (i) transfer of the reactants in the fluid phase to the surface of photocatalyst, (ii) adsorption of the reactants, (iii) reaction in the adsorbed phase, (iv) desorption of the products, and (v) removal of the products from Table 37.1 Nonexhaustive examples of micropollutants degradable by photocatalysis. Category
Important subclasses
Examples degradable by photocatalysis
Pharmaceuticals
Lipid regulator, anti-inflammatory drugs, anticonvulsants, antibiotics
Tetracycline, lincomycin, salicylic, ciprofloxacin, naproxen, paracetamol, caffeine
Personal care products
Fragrances, preservatives, Benzene, naphthalene, antimicrobials, insect repellents chloroxylenol
Surfactants
Nonionic surfactants
Industrial chemicals
Plasticizers and plastic additives Phthalic, bisphenol A
Pesticides
Insecticides, herbicides, and fungicides
Heavy metals
Al, Cd, Cr, Cu, Fe, Pb, Hg, Ni, Zn Al, Cu, Zn, Cr, Fe
Sodium dodecyl sulfate, sodium dodecyl benzene sulfonate, trimethyl phosphate, tetrabutylammonium phosphate
Parathion, lindane, tetrachlorvinphos, fenitrothion
A rather complete list of the micropollutants in water and wastewater and more examples of photocatalytic degradation can be found in [2, 7], respectively.
37.2 Main Processes for Photocatalytic Abatement of Micropollutants
Figure 37.1 Schematic illustration of semiconductor-based photocatalytic processes. P stands for micropollutants.
Adsorption Reduction (iii) (ii)
O2 O2• –
e– (i)
Recombination Adsorption
(ii)
hv
h+
P+
(iii) Oxidation
Further degradation
P
H+ + OH– H2O
the interface region [9]. Photocatalytic reactions differ from conventional catalysis only by the mode of activation of the catalyst in which thermal activation is replaced by photonic activation, which occurs in the adsorbed phase in step (iii). In general, this photonic activation process based on semiconductors involves three main steps (Figure 37.1). Specifically, (i) upon light irradiation, the photocatalyst absorbs supra-bandgap photons (≥Eg , Eq. (37.1)), and photoexcited electron (e− ) and hole (h+ ) pairs are produced in the conduction band (CB) and the valence band (VB), respectively (Eq. (37.2)). (ii) The photogenerated electrons and holes are either separated and migrated to catalytically active sites at the semiconductor surface or recombined. (iii) An efficient charge utilization of the electron–hole pairs allows the respective oxidation and reduction reactions with the adsorbates on the particle surface (Eqs. (37.3) and (37.4)). Bandgap (eV) =
1240 𝜆(nm)
(37.1)
PC + h𝜈(≥ Eg ) −−−→ h+ + e−
(37.2)
A(ads) + e− −−−−→ A− (ads)
(37.3)
D(ads) + h+ −−−−→ D+ (ads)
(37.4)
e− + O2 → O⋅− 2
(37.5)
h+ + OH− (or HO) −−−−→ ⋅ OH
(37.6)
O2 + 2H+ + 2e− −−−−→ H2 O2
(37.7)
The first reaction type for water pollutant abatement is oxidative degradation. It is well known that O2 and water are essential for photooxidation. Molecular oxygen usually functions as an electron acceptor by interacting with the photogenerated electrons on the CB of the photocatalyst, thus forming reactive oxygen species (ROS), superoxide radicals (O⋅− 2 ), an oxidizing agent (Eq. (37.5)). The photogenerated holes can oxidize the micropollutants (P) to form P+ or react with OH− or oxidize H2 O into OH⋅ radicals, a different type of ROS acting as a strong oxidizing agent (Eq. (37.6)). Together with other highly oxidant species, peroxide radicals (Eq. (37.7)) are reported to be responsible for heterogeneous
673
674
37 Photocatalytic Abatement of Emerging Micropollutants in Water and Wastewater
photocatalytic oxidation [4, 5]. Another reaction type is the direct reduction of micropollutants such as heavy metal ions in solution by photogenerated electrons [9, 10]. Heavy metals are generally toxic and can be removed from industrial waste effluents as small crystallites deposited on the photocatalyst according to the redox process, provided the redox potential of the metal cation is more positive than the flat band potential of the semiconductor.
37.3 Advancements in Photocatalysts for Photocatalytic Abatement of Micropollutants in Water and Wastewater Various semiconductors including TiO2 [11], ZnO [12], CeO2 [13], WO3 [14], C3 N4 [15], BiVO4 [16], Bi2 WO6 [17] can be used as photocatalysts for micropollutant abatement in the aqueous environment [5, 18]. Among them, TiO2 has been the most commonly used due to its well-known advantages such as cost-effectiveness, nontoxicity, regeneration ability, photocatalytic efficiency, and high chemical stability [8, 19]. However, a major drawback of pure TiO2 is its relatively large bandgap, as it can only absorb a small portion of the ultraviolet (UV) radiation spectrum [20]. In addition, most of the photocatalysts encounter a common problem of fast charge carrier recombination, which results in low photocatalytic conversion efficiency. Therefore, it is highly desired to carry out proper modifications to enhance the photolysis potential of semiconductors. 37.3.1
Photocatalysts Components Optimization
An efficient approach is the modification of catalyst structure and composition, which is able to largely influence each reaction process, including light harvesting, charge carrier separation, and surface reaction kinetics, and ultimately results in photocatalytic performance regulation. The strategies for optimization of semiconductor-based photocatalyst components include semiconductor doping, metal deposition, quantum dots sensitization, plasmon-based photocatalysts, semiconductor combinations, and so on. 37.3.1.1
Semiconductor Doping
In most cases, proper dopants not only enhance the absorption of visible light due to the formation of localized or delocalized electronic states but also enhance the separation efficiency of photogenerated electron–hole pairs [5]. Metal ion doping with a lower dopant content can introduce localized electronic states, such a donor level above the VB (Figure 37.2a) or an acceptor level below the CB (Figure 37.2b) in the forbidden band of wide bandgap photocatalysts, such as TiO2 and ZnO, which can narrow their bandgaps and thus enhance their activity under visible-light irradiation [20, 21]. Metal ion doping with a high dopant content often leads to the formation of the band of delocalized states in the middle of the bandgap (Figure 37.2c). On the other hand, some ions, especially nonmetal ions, such as N, I, B, Cl, F, S, P, and C, can contribute to new valence
37.3 Advancements in Photocatalysts for Photocatalytic Abatement of Micropollutants
e – e– e–
e– e – e – CB
CB
e– e– e–
e– e– e– CB
Acceptor level
CB
Mid-gap states
Donor level VB
VB
•O
(b)
(c)
O2
(d)
Vis light
Dye* 2
New valence band
h+ h+ h+
h+ h+ h+
h+ h+ h+
h+ h+ h+
(a)
VB
VB
e– e– e–
•O
Metals 2
CB O2
e– e– e–
e–
UV light
CB Fermi level
Dye
h+ h+ h+
(e)
(h) Empty state
•O 2
O2
–OH
VB
VB
e– e– e–
e–
Carbon materials
Vis light •O 2
CB O2
e–
e–
•OH
e– e– e–
UV light
CB Fermi level
Filled state VB
h+ h+ h+
(f) •O
(i) e– e–
2
O2
e – e– e –
CB
CB
VB
•OH
Vis light
VB
(g)
–OH
VB
LSPR effect
h+ h+ Narrow band gap semiconductor
•O 2
O2
e– e– e– CB
e– e– e– Vis light
UV light
–OH
h+ h+ h+
VB
h+ h+ h+
•OH
(j)
Figure 37.2 (a) Donor level, (b) acceptor level, (c) mid-gap states formed by metal ion doping; (d) new valence band formation by nonmetal ions doping; schematic basic principle of (e) dye-sensitized, (f ) Localized surface plasmon resonance (LSPR)-sensitized, and (g) narrow bandgap semiconductor-sensitized activity under visible-light irradiation; schematic illustration of charge transfer for (h) metal deposition, (i) carbon materials combination, and (j) type II semiconductors combination. The * in Figure 37.2e stands for the excited state of the dye.
band formations and shift valence band edge upward to narrow the bandgap (Figure 37.2d) [20, 22, 23]. For example, N-doped TiO2 has been reported to show excellent visible-driven photocatalytic activity for the degradation of pharmaceutical pollutants, including ciprofloxacin, naproxen, and paracetamol [24]. C-, N-, and S-doped mesoporous anatase–brookite nanoheterojunction titania have been synthesized through a simple sol–gel method in the presence of triblock copolymer pluronic P123. The photocatalytic efficiency of the catalyst under visible-light illumination dramatically increased with the addition of the C, N, and S nonmetals as dopants, achieving complete degradation (close to 100%) of cyanotoxin microcystin [25]. The results demonstrate the great potential of
675
676
37 Photocatalytic Abatement of Emerging Micropollutants in Water and Wastewater
the visible-light-activated C-, N-, and S-doped titania photocatalysts for the treatment of organic micropollutants in contaminated waters under visible light. 37.3.1.2
Surface Sensitization
The surface modifications of wide bandgap photocatalysts using a combination of dye sensitization, metals with surface plasmon resonance (SPR) effects or other visible-light-responsive photocatalysts provide another way for visible-light activity enhancement. In dye sensitization, the dye is excited by absorbing visible light, causing charge injection into the CB of the semiconductor at sub-bandgap excitation, which is followed by catalytic processes through interfacial electron transfer (Figure 37.2e) [26]. The metallic plasmonic nanoparticles (such as Au, Ag, and Cu), anchored to a semiconductor, acting similarly to a dye sensitizer, which absorbs resonant photons and transfers the energetic electrons formed in the process of SPR excitation to the nearby semiconductor to initiate the photocatalytic reactions (Figure 37.2f ) [20, 27]. In addition, combining wide bandgap semiconductors with narrow bandgap semiconductors has also been reported to result in visible-light activity (Figure 37.2g) [21, 28]. For example, nanocrystalline TiO2 has been combined with D35 organic dye to fabricate a visible-light photocatalyst, which exhibits excellent visible-light degradation of bisphenol A (BPA). Under the experimental conditions with initial BPA concentration of 5 mg/L, initial pH of 7, external bias of 0.25 V, and sensitizer concentration of 0.1 mM, BPA was almost completely degraded in 300 minutes, and the four intermediates were gradually mineralized [29]. Interestingly, using a photosensitizing material tris(4,7-diphenyl-1,10-phenanthroline)ruthenium(II) chloride immobilized in a porous poly(dimethylsiloxane) inert support, Diez-Mato et al. have investigated the elimination of three common micropollutants such as ibuprofen, paracetamol, and BPA. The process effectively removed ibuprofen and BPA in ultrapure water with conversion rates of 100% and 80%, respectively [30]. Graphitic carbon nitride (g-C3 N4 ), possessing a suitable bandgap (2.7 eV), has been employed in the modification of TiO2 to form g-C3 N4 /TiO2 nanocomposite, which showed superior photocatalytic performance for the degradation of ciprofloxacin (CIP) under visible-light irradiation, compared with pure TiO2 and commercial P25 [31]. 37.3.1.3
Metal Deposition
Considering the practical applications, a higher reaction rate is still required since the quantum efficiency under visible light is still lower than that under UV light [18, 32]. One way to improve the photocatalytic activity is to deposit metallic nanoparticles of noble metals (such as Pt, Pd, Au, Ag, Ru, and Fe) onto semiconductors to enhance the photocatalytic activity by suppressing the electron–hole pair recombination [28]. The photoinduced electrons migrate to the metal due to the relatively low Fermi level of metals (Figure 37.2h), which makes the photoinduced holes more stable by increasing the lifetime of the charge carrier. Therefore, more OH⋅ radicals and superoxide radicals are generated and enhanced the redox reaction. For example, Xu’s group has reported an efficient visible-light (𝜆 > 420 nm)driven photoreduction of Cr(VI) to Cr(III) over noble metal nanoparticles
37.3 Advancements in Photocatalysts for Photocatalytic Abatement of Micropollutants
loaded on TiO2 (P25) with oxygen vacancies (OV). The results show that the noble metal deposition not only decreases the concentration of OV, the charge recombination sites on the defective P25-OV, but also serve as electron sinks to promote the separation of electron–hole pairs, thereby enhancing the photoactivity [33]. Modifying TiO2 surface with Pt and SiOx has accelerated the degradation of all the tested pharmaceuticals (i.e. caffeine, cimetidine, propranolol, and sulfamethoxazole) to a certain degree [14]. Moreover, Ag and Au cocatalysts decorated on BiVO4 nanofibers have been developed to show greater photocatalytic performance for phenol degradation. Ag–BiVO4 showed enhancement due to increased carrier traps, while Au–BiVO4 showed enhancement due to both carrier traps and SPR. It is considered that BiVO4 nanofibers have the potential to become efficient photocatalysts alternative to TiO2 for the removal of emerging organic contaminants [34]. 37.3.1.4
Carbon Materials Combination
Recently, combining semiconductors with various carbon materials, such as activated carbon nanofibers, carbon nanotubes (CNTs), and graphene, has been investigated as promising materials toward enhanced photocatalytic activities through suppression of e− /h+ recombination (the electron-sink effect of carbon materials benefited from their low Fermi level and high conductivity) (Figure 37.2i). In addition, the carbon materials provide high-surface-area scaffolds for semiconductor particles distribution and immobilization, resulting in greater photocatalytic efficiency [35, 36]. For example, enhanced photocatalytic activity of graphene-modified TiO2 in comparison to bare TiO2 has been demonstrated in the degradation of different types of micropollutants, including diuron, alachlor, isoproturon, microcystin-LA, BPA, and diphenhydramine. The enhanced activity was attributed to the good assembly and interfacial coupling between TiO2 and graphene sheets as well as to the respective quenching of photoluminescence [37, 38]. 37.3.1.5
Combining Semiconductors
Another promising approach is to combine different types of semiconductors to attain simultaneously a sufficient redox potential of photoexcited electron–hole pairs and efficient absorption of a large proportion of sunlight. Figure 37.2j illustrates the type II band structure (the most effective type for charge carrier transfer among three types of semiconductor heterojunctions) [39] of a composite photocatalyst prepared by a mixture of wide and narrow bandgap photocatalysts. Upon light irradiation, the photogenerated electrons can flow from the semiconductor with a higher CB minimum to the other with a lower CB minimum. Moreover, the formation of heterojunction leads to a more efficient inter-electron transfer between the two catalysts, thus enhancing photocatalytic activity. For example, using sunlight-driven g-C3 N4 /P25 photocatalyst, the degradation of clofibric acid (CA) was investigated. A very low g-C3 N4 content of 8.0 wt% resulted in a 3.36 and a 2.29 times faster reaction rate for CA photodegradation compared to pristine g-C3 N4 and P25, respectively [40]. A broad-spectrum N-doped carbon quantum dot/TiO2 nanosheet composites with significantly improved broad-spectrum utilization exhibited higher photocatalytic activity toward
677
678
37 Photocatalytic Abatement of Emerging Micropollutants in Water and Wastewater
the degradation of diclofenac. This excellent photocatalytic performance was largely attributed to the efficient charge separation induced by the fabricated heterostructures [41]. Semiconductors combination is a primary strategy to develop broad-spectrum photocatalysts for photocatalytic pollutant elimination. 37.3.2 37.3.2.1
Photocatalysts Configuration Optimization Freestanding Particulate
In general, photocatalysts can be used either as freestanding particulates or as thin film upon coating on a substrate. As for particulate photocatalysts, the structure, size, and shape can be optimized to maximize the photocatalytic activities during the water treatment process, since the morphology or scale modification improves the adsorption capacity of the contaminants and the active surface area with nanoscales [5]. Figure 37.3a shows the various dimensions of the structure of photocatalysts: 0D spheres, 1D fibers, rods, and tubes, 2D nanosheets, and 3D interconnected architectures, as well as their property advantages. Typically, it is suggested that with the decreasing size of the nanoparticles, the e− /h+ recombination behaviors decrease due to the enhanced interfacial charge carrier transfer on the semiconductor surfaces [5]. 1D structure was reported to be efficient in Interconnected architectures
(a)
(b)
Sheets
3D
Tubes
2D
Fibers Spheres
1D
Light Smooth scattering surface High specific surface area Nonwoven mat High adhesion
(c)
0D
Unique structures, porosity, High carrier mobility
FCRIC 20.0kV 13.2mm ×50.0k SE(M)
1.00μm
Graphene-EY aerogel Take out, wash, and reuse Take out and wash
(d) Nitro compounds and Cr(VI) solution
Hydrothermal Regenerated graphene-EY aerogel
Fresh EY solution and used graphene-EY aerogel
Figure 37.3 (a) Semiconductor structures according to the structural dimensionality and expected property. (b) SEM image of the nanocomposite Au NPs/TiO2 films. NP, nanoparticles. (c) Images of borosilicate tube before and after abrasive blasting treatment. (d) Schematic illustration of the regeneration of 3D graphene–EY aerogel. Source: (a) Reprinted with permission from Lee and Park [5]. Copyright 2013, Elsevier; (b) Reprinted with permission from Nil et al. [46]. Copyright 2018, Springer-Verlag; (c) Reprinted with permission from Espino-Estévez et al. [47]. Copyright 2015, Elsevier; (d) Reprinted with permission from Yang et al. [48]. Copyright 2017, Elsevier.
37.3 Advancements in Photocatalysts for Photocatalytic Abatement of Micropollutants
the adsorption and decomposition of nonbiodegradable organic compounds by exhibiting a shorter carrier diffusion pathway and faster mass transfer of organic compounds [34]. Because of the unique mesoporous channel for facilitating the intraparticle molecular transfers of ions, the well-defined 3D architectures with a large specific surface area could further help increase their photocatalytic efficiency [12]. 37.3.2.2
Film, Immobilized, and Aerogel-Based Catalysts
A great challenge for powder photocatalysts to be used in the treatment of wastewater lies in the difficulty of phase separation and recovery of the catalysts. In this context, coated catalyst configurations eliminate the need for catalyst filtration or centrifugation, thus possessing greater applicability in the repeated catalytic operations [42]. Therefore, more coated photocatalysts and immobilization techniques have been investigated, such as preparation of film materials and reactive photocatalysts immobilized on inert supports [43–45]. Recently, Au NPs/TiO2 has been coated onto borosilicate glass disk to obtain thin-film nanocomposite materials and used in the photocatalytic degradation of emerging micropollutants [46]. The scanning electron microscopy (SEM) image (Figure 37.3b) of Au NPs/TiO2 thin film showed excellent distribution of fine grains of TiO2 on the surface of borosilicate glass, while Au nanoparticles were evenly distributed on the TiO2 structure. The film exhibited efficient photocatalytic removal of sulfamethoxazole and triclosan from aqueous solutions using UV-A light (𝜆max = 330 nm). The stability of the film was reassessed with the repeated use of the catalyst through simple wash and dry, which showed no significant decrease in photocatalytic degradation efficiency. The preparation of stable and highly photoactive titania coatings on inert supports is necessary for realistic applications of TiO2 photocatalysis in water purification treatments. The coatings of TiO2 on the outer wall of the inner tube of a glass tubular reactor by dip-coating method (Figure 37.3c) have been reported [47]. The results of the adhesion tape test showed that either milling of aggregate material with a planetary mill or chemical stabilization of the particles was necessary to obtain TiO2 coatings on the glass tube with acceptable quality to be used in water treatment applications. The photoactivity results showed that the coatings prepared with 30 minutes wet milling of the catalyst followed by chemical disaggregation were more efficient in degradation and mineralization of phenol, diclofenac, and isoproturon. The reusability of the TiO2 coatings was evaluated, and a promising photocatalytic performance was observed with a very low variation in the decay rate after five consecutive usages. 3D graphene aerogel materials are another viable option of using the photocatalysts that make the phase separation easy. The 3D graphene aerogel architectures possess not only the inherent properties of graphene (GR) of high electrical conductivity but also unique hierarchically porous structures for reactant adsorption, both of which are beneficial for the degradation of pollutants. Typically, organic dyes Eosin Y (EY) as photosensitizers have been spatially confined and distributed in the graphene framework to form the 3D macroscopic graphene–organic aerogel photocatalysts, which display efficient visible-light-driven activity toward photoreduction of Cr(VI) to Cr(III) [48].
679
680
37 Photocatalytic Abatement of Emerging Micropollutants in Water and Wastewater
In addition, such a bulk aerogel manifests excellent regenerability via simple replenishment of fresh dyes, which guarantees the long-term photoactivity of the 3D aerogel (Figure 37.3d).
37.4 Reaction System Optimization 37.4.1
Reaction Conditions
The rate and efficiency of a photocatalytic reaction depend on a number of factors that govern the kinetics of photocatalysis. An extensive review of the influential variables showed that dissolved oxygen, light intensity distribution, loading of the photocatalyst, air flow rate, temperature, and hydrogen peroxide concentration had constructive impacts on the process performance, while initial concentration of the reactant, light wavelength, feed flow rate, irradiation time, and pH showed detrimental effects [3, 9]. For more details on the effect of variables on the process performance, readers are referred to Ref. [18]. 37.4.2
Solar Reactors
Depending on the reaction considered, various photoreactors can be chosen, including fixed-bed photoreactors and slurry batch photoreactors, either mechanically or magnetically stirred [9, 49]. In laboratory experiments, near-UV light is provided by a lamp placed in front of an optical window of the photoreactor, with wavelength adjustable using optical filters. For different types of photocatalytic reactors tested, refer to Ref. [18]. For pilot experiments using sunlight, extensive effort has been devoted to the design of efficient solar light collectors. The original solar photoreactor designs for photochemical were based on line-focusing parabolic trough concentrator (PTC) (Figure 37.4a), which is based on conventional solar thermal collector designs. Subsequently, it has been found that PTCs were unsuitable for photocatalytic applications since water was heated and radiation flux was too high. Most of the photons were not used efficiently, and their cost was high. Attempts were made to use nonconcentrating solar collectors (NCCs) (Figure 37.4b) as an alternative to PTCs. Despite important advantages, the design of NCC is not trivial, due to the need for weather-resistant and chemically inert UV-transmitting reactors. Besides, NCC designs possess other disadvantages of low mass transfer and reactant vaporization and contamination. In this regard, compound parabolic concentrators (CPCs) (Figure 37.4c), a type of low-concentration collector used in thermal applications, becomes an option of interest. Having the advantages of both nonconcentrating and concentrating and testing of small nontracking systems without their original disadvantages, CPCs seem to be the best option for solar photocatalytic processes. Figure 37.4d shows the photographs of a commercial nonconcentrating solar detoxification system using the CPC technology, consisting of a TiO2 separation system (left) and a CPC (right). For more details of the development of the photoreactors, see the reviews from the group of S. Malato [3, 50].
37.5 Future Challenges and Prospects
(a)
(b)
(c)
(d)
Figure 37.4 Design concepts for solar water photocatalytic reactors: (a) concentrating (parabolic trough), (b) nonconcentrating (one-sun reactor), and (c) compound parabolic collector. (d) Photographs of solar detoxification demonstration plant constructed in “SOLARDETOX” project at HIDROCEN (Madrid, Spain). Left: TiO2 separation system. Right: compound parabolic collector. Source: Reprinted with permission from Malato et al. [3]. Copyright 2009, Elsevier.
In summary, the design procedure for a solar photocatalytic system requires the selection of a reactor, catalyst operating mode (slurry or fixed matrix), reactor-field configuration (series or parallel), treatment system mode (once-through or batch), flow rate, pressure drop, pretreatment, catalyst and oxidant loading method, pH control, and so on. The optimization of these reaction conditions from a system level is of great importance to achieve a desirable photocatalytic efficiency.
37.5 Future Challenges and Prospects Heterogeneous photocatalysis appears to be one of the most promising potential applications for micropollutant abatement in water and wastewater, since many toxic micropollutants, either organic or inorganic, can be oxidized or reduced into harmless final compounds under ambient conditions. Despite the attractive advantages and great progresses made in the design of efficient materials and photocatalytic systems, it still remains a big challenge to achieve efficiency and practicality potentially applied in industrial operations. Thus, it is more realistic to address the package of challenges as a whole in this research field. First, more efficient photocatalytic engineering should be developed in terms of photocatalysts and reaction systems optimization. In particular, for practical applications, it is necessary for the water to be transparent before the photocatalytic treatment. In this regard, conventional wastewater treatment plants (WWTPs), including physical and chemical adsorption, coagulation, precipitation, and biotransformation, as well as advanced oxidation processes (AOPs) such
681
682
37 Photocatalytic Abatement of Emerging Micropollutants in Water and Wastewater
as UV radiation, UV/Cl2 , UV/O3 , and UV/H2 O2 treatment, should be combined to pave the way for novel integrated water treatment technologies, thereby overcoming high treatment costs. Second, monitoring the disappearance rate of the target micropollutants is not the most appropriate parameter to classify the efficiency. More efforts should be devoted to understanding the thorough mechanisms with detailed reaction steps of the different pathways, the quantification of various intermediates, and by-product evaluation, which are key factors to optimize each treatment and to maximize the overall process. Third, toxicity tests of the treated water will also gather useful information about the practical application of the photocatalytic process, particularly when incomplete degradation is planned. A better understanding of the photocatalytic process and the operative conditions could give great opportunities for its application in the abatement of emerging micropollutants in water and wastewater.
Acknowledgments The support from the National Natural Science Foundation of China (NSFC) (U1463204, 21872029, and 21173045), the Award Program for Minjiang Scholar Professorship, the Natural Science Foundation (NSF) of Fujian Province for Distinguished Young Investigator Rolling Grant (2017J07002), the Independent Research Project of State Key Laboratory of Photocatalysis on Energy and Environment (No. 2014A05), the 1st Program of Fujian Province for Top Creative Young Talents, and the Program for Returned High-Level Overseas Chinese Scholars of Fujian Province is gratefully acknowledged.
References 1 Kim, M.-K. and Zoh, K.-D. (2016). Environ. Eng. Res. 21: 319–332. 2 Margot, J., Rossi, L., Barry, D.A., and Holliger, C. (2015). WIREs Water 2:
457–487. 3 Malato, S., Fernández-Ibáñez, P., Maldonado, M.I. et al. (2009). Catal. Today
147: 1–59. 4 Chong, M.N., Jin, B., Chow, C.W.K., and Saint, C. (2010). Water Res. 44:
2997–3027. 5 Lee, S.-Y. and Park, S.-J. (2013). J. Ind. Eng. Chem. 19: 1761–1769. 6 Luo, Y., Guo, W., Ngo, H.H. et al. (2014). Sci. Total Environ. 473–474:
619–641. 7 Gaya, U.I. and Abdullah, A.H. (2008). J. Photochem. Photobiol. C: Photochem.
Rev. 9: 1–12. 8 Akpan, U.G. and Hameed, B.H. (2009). J. Hazard. Mater. 170: 520–529. 9 Herrmann, J.-M. (1999). Catal. Today 53: 115–129. 10 Amin, M.T., Alazba, A.A., and Manzoor, U. (2014). Adv. Mater. Sci. Eng. 2014:
24. 11 Mahmoud, W.M.M., Rastogi, T., and Kummerer, K. (2017). Curr. Opin. Green
Sustainable Chem. 6: 1–10.
References
12 Chang, J.S., Tan, J.K., Shah, S.N. et al. (2017). J. Taiwan Inst. Chem. Eng. 81:
206–217. 13 Sponza, D.T. and Guney, G. (2017). Water Sci. Technol. 76: 2603–2622. 14 Choi, J., Lee, H., Choi, Y. et al. (2014). Appl. Catal., B 147: 8–16. 15 Zheng, Q., Durkin, D.P., Elenewski, J.E. et al. (2016). Environ. Sci. Technol. 50:
12938–12948. 16 Chen, S.J., Zhai, X.N., Li, Y.J. et al. (2018). J. Nanosci. Nanotechnol. 18:
2472–2480. 17 Chen, S.J., Li, Y.J., Lu, R.J., and Wang, P. (2013). J. Nanosci. Nanotechnol. 13:
5624–5630. 18 Zangeneh, H., Zinatizadeh, A.A.L., Habibi, M. et al. (2015). J. Ind. Eng. Chem. 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43
26: 1–36. Konstantinou, I.K. and Albanis, T.A. (2004). Appl. Catal., B 49: 1–14. Dong, H., Zeng, G., Tang, L. et al. (2015). Water Res. 79: 128–146. Li, X., Yu, J., Low, J. et al. (2015). J. Mater. Chem. A 3: 2485–2534. Serpone, N. and Emeline, A.V. (2012). J. Phys. Chem. Lett. 3: 673–677. Lee, K.M., Lai, C.W., Ngai, K.S., and Juan, J.C. (2016). Water Res. 88: 428–448. Shetty, R., Chavan, V.B., Kulkarni, P.S. et al. (2017). Indian Chem. Eng. 59: 177–199. El-Sheikh, S.M., Zhang, G.S., El-Hosainy, H.M. et al. (2014). J. Hazard. Mater. 280: 723–733. Chatterjee, D. and Mahata, A. (2001). Appl. Catal., B 33: 119–125. Zhang, N., Han, C., Fu, X., and Xu, Y.-J. (2018). Chem 4: 1832–1861. Pelaez, M., Nolan, N.T., Pillai, S.C. et al. (2012). Appl. Catal., B 125: 331–349. Bai, X., Yang, L., Hagfeldt, A. et al. (2019). Chem. Eng. J. 355: 999–1010. Diez-Mato, E., Cortezon-Tamarit, F.C., Bogialli, S. et al. (2014). Appl. Catal., B 160: 445–455. Yang, Z., Yan, J., Lian, J.B. et al. (2016). ChemistrySelect 1: 5679–5685. Hu, Y., Zhang, X., and Wei, C. (2010). Res. Chem. Intermed. 36: 95–101. Pan, X. and Xu, Y.-J. (2013). J. Phys. Chem. C 117: 17996–18005. Nalbandian, M.J., Zhang, M.L., Sanchez, J. et al. (2015). J. Mol. Catal. A: Chem. 404: 18–26. Zhang, N., Yang, M.-Q., Liu, S. et al. (2015). Chem. Rev. 115: 10307–10377. Yang, M.-Q., Zhang, N., Pagliaro, M., and Xu, Y.-J. (2014). Chem. Soc. Rev. 43: 8240–8254. Moreira, N.F.F., Narciso-da-Rocha, C., Polo-Lopez, M.I. et al. (2018). Water Res. 135: 195–206. Martins, P.M., Ferreira, C.G., Silva, A.R. et al. (2018). Composites Part B 145: 39–46. Wang, Y., Wang, Q., Zhan, X. et al. (2013). Nanoscale 5: 8326–8339. Chen, P., Wang, F.L., Zhang, Q.X. et al. (2017). Chemosphere 172: 193–200. Wang, F.L., Wu, Y.L., Wang, Y.F. et al. (2019). Chem. Eng. J. 356: 857–868. Luna, R., Solis, C., Ortiz, N. et al. (2018). Int. J. Chem. Reactor Eng. 16. Bellobono, I.R. and Groppi, F. (2017). Curr. Opin. Green Sustainable Chem. 6: 69–77.
683
684
37 Photocatalytic Abatement of Emerging Micropollutants in Water and Wastewater
44 Sanches, S., Nunes, C., Passarinho, P.C. et al. (2017). J. Chem. Technol.
Biotechnol. 92: 1727–1737. 45 Arlos, M.J., Hatat-Fraile, M.M., Liang, R. et al. (2016). Water Res. 101:
351–361. 46 Nil, L., Tiwari, A., Shukla, A. et al. (2018). Environ. Sci. Pollut. Res. 25:
20125–20140. 47 Espino-Estévez, M.R., Fernández-Rodríguez, C., González-Díaz, O.M. et al.
(2015). Chem. Eng. J. 279: 488–497. 48 Yang, M.-Q., Zhang, N., Wang, Y., and Xu, Y.-J. (2017). J. Catal. 346: 21–29. 49 Pronina, N., Klauson, D., Rudenko, T. et al. (2016). Photochem. Photobiol. Sci.
15: 1492–1502. 50 Malato, S., Blanco, J., Alarcón, D.C. et al. (2007). Catal. Today 122: 137–149.
685
38 Catalytic Abatement of NOx Emissions over the Zeolite Catalysts Runduo Zhang, Peixin Li, and Hao Wang College of Chemical Engineering, Beijing University of Chemical Technology, State Key Laboratory of Chemical Resource Engineering, No.15 North 3rd Ring Road, Chaoyang District, Beijing, 100029, PR China
For the past 50 years, there has been a growing focus on NOx emission control from both stationary and mobile sources, leading to worldwide research on deNOx technologies. Over 40% of NOx emissions are from stationary sources, such as power plants and industrial boilers using fossil fuels, and over 50% come from automotive sources, such as gasoline and diesel engine cars. Currently, three commercial catalytic systems are available: noble-metal-based three-way catalysts (TWCs) for the purification of gasoline engine exhausts, transition-metal–zeolite catalysts for the purification of diesel engine exhausts, and vanadium–titania catalysts for the elimination of power plant effluent gases. However, due to stricter environmental legislation and the demand to achieve energy savings, there is an increasing desire to develop more efficient deNOx catalysts. For the stationary deNOx technology, especially for that used in power plants, it is required that the industrial catalyst possesses excellent low-temperature activity as well as high N2 selectivity. In addition to these criteria, the automobile deNOx demands a wide temperature window for ideal performance due to the large temperature swings in the exhaust gases. Selective catalytic reduction (SCR) was first discovered by Engelhard and patented in 1957, and it was mainly summarized into the following four parts: (1) Noble-metal-supported alumina (Pt, Ph, Pd, Ag): These catalysts were developed as catalytic materials for emission control in the early 1970s and became the standard for SCR reaction, particularly for diesel engine emissions. The reaction uses CO and H2 or hydrocarbons as the reducing agents while requiring temperature above 300 ∘ C to achieve decent conversions [1, 2]. The noble metal catalyst is not only expensive but is also prone to sulfur poisoning [3–5] and restricted to a narrow operation temperature range [6, 7]. These factors limit their wider implementations. (2) Metal oxide catalysts: Japan developed the SO2 -resistant V2 O5 /TiO2 catalyst in the 1970s [8, 9], to satisfy the request according to country’s environmental protection policy. The catalyst exhibited high activity for NOx reduction in the temperature range of 300–400 ∘ C [10, 11] and for oxidative Heterogeneous Catalysts: Advanced Design, Characterization and Applications, First Edition. Edited by Wey Yang Teoh, Atsushi Urakawa, Yun Hau Ng, and Patrick Sit. © 2021 WILEY-VCH GmbH. Published 2021 by WILEY-VCH GmbH.
686
38 Catalytic Abatement of NOx Emissions over the Zeolite Catalysts
desulfurization of flue gas [12, 13]. The reaction used ammonia or urea as the reducing agent. Despite being commercialized since 1978, the vanadium catalyst remains widely used today. However, the temperature of the plant obviously declined after desulfurization and dedusting, which reduces the corresponding catalytic activity. Simultaneously, due to the highly biological toxicity and relatively high temperature necessary for the reaction, the use of the vanadium catalyst has been largely limited in the United States and Japan. To overcome this drawback, an effort on designing an efficient and environmentally friendly low-temperature catalyst for NH3 -SCR has been made. (3) Carbon-based catalysts: These catalysts were used as a catalyst carrier for SCR reaction with the large surface area and high adsorption capacity [14, 15]. The currently studied carbon-based catalysts mainly include activated carbon (AC) [16], activated carbon fibers (ACFs) [17], and AC moldings [18]. Due to complicated preparation process, high cost, and instability under an oxidative atmosphere, they lack practical value and are difficult to popularize. At present, the carbon-based catalysts are still in the state of laboratory development. (4) Zeolite catalysts: Zeolites are water-containing aluminosilicates of natural or synthetic origin with a highly ordered crystal structure, generally formulated as MI MII 0.5 [(AlO2 )x ⋅(SiO2 )y ⋅(H2 O)z ], where MI and MII are preferentially alkali and alkaline-earth metals. They consist of SiO4 − and AlO4 − tetrahedra, which are interlinked by common oxygen atoms to form a three-dimensional network. The tetrahedra are the smallest structural units by which zeolites can be classified. Linking these primary building units together leads to 23 possible secondary building blocks (polygons). Zeolites are predominantly distinguished on the basis of the geometries of the cages (α, β, γ, and super) and channels (straight and sinusoidal) formed by the rigid tetrahedral frameworks. To date, 229 types of zeolite framework structures have been recognized by the Structure Commission of the International Zeolite Association [19]. With the high catalytic activity for NH3 -SCR and a wide range of active temperature, zeolites have been attracting attention in the research. The types of zeolites used mainly include Y (FAU), ZSM-5 (MFI), β (BEA), SSZ-13 (CHA), mordenite (MOR), etc. The zeolite catalysts for the SCR process are mainly prepared by ion-exchange method, and the metal elements that can be used for ion exchange usually include Mn, Cu, Co, Pd, V, Ir, Fe, Ce, etc. More details are given in the following sections. Summarized above, zeolite catalysts are the most promising deNOx catalysts. The research direction has thus been pursued: designing an efficient low-temperature zeolite catalyst for the NH3 -SCR system.
38.1 Zeolite Catalysts with Different Topologies Metal-exchanged zeolites represent one kind of SCR catalysts, which exhibit considerable performances across a wide temperature range. Fe- and Cu-modified
38.1 Zeolite Catalysts with Different Topologies
zeolite catalysts were widely investigated for NH3 -SCR, due to their higher activities and thermal stabilities with respect to the commercialized vanadium catalysts [20]. Cu-based zeolite catalysts (Cu–ZSM-5, Cu-β, Cu–mordenite) commonly exhibited higher activities at lower temperature ranges ( 400 ∘ C). Activity evaluation suggested that the simultaneous presence of Cu and Fe species in β-type zeolites significantly improved the low-temperature NO conversion. Although a large amount of research have been done over the conventional zeolite catalysts with medium micropores, such as β (BEA), ZSM-5 (MFI), mordenite (MOR), and ferrierite (FER) [23, 24], several practical challenges arise when these metal-exchanged zeolites are used in lean-burn or diesel vehicles. Two of the most serious problems are hydrothermal deactivation and chemisorption of impurities (hydrocarbon, phosphorus, and potassium) on the active sites of catalysts [25, 26]. Recently, many researchers concentrate on the newly discovered small-pore Cu-exchanged zeolite catalysts with a CHA topology, such as Cu–SAPO-34 and Cu–SSZ-13, which were frequently reported to be extremely active for NH3 -SCR with high hydrothermal stability [27, 28]. NH3 -SCR performance was gradually improved along with the zeolite framework structures changing from straight channel (ZSM-5, β, Y) to cage type (SSZ-13, SSZ-16, SSZ-17), bridged by hybrid structures (OFF-ERI series: offretite, ZSM-34, UZM-12) [29]. The catalytic activities for NH3 -SCR of NOx over Cu-exchanged zeolites with diverse topologies are illustrated in Figure 38.1. Cage-type zeolites exhibit a superior activity (NOx conversion >95% at a relatively wide temperature range of 150–400 ∘ C) compared to Cu–ZSM-5 and Cu-β at either low temperatures (400 ∘ C). For the structural stability, cage-type zeolites exhibit the best resistance to dealumination and have long life,
NO conversion (%)
100 Cu–β Cu–ZSM-5 Cu–Y Cu–Offretite Cu–ZSM-34 Cu–UZM-12 Cu–SSZ-13 Cu–SSZ-17 Cu–SSZ-16
80 60 40 20
SCu(1)
Cage units
LCu(1) High active Cu ions located in d6r unit of cage-type zeolites
0 100
200 300 400 Temperature (°C)
500
600
Cage-type zeolites Cu-β
Straight channels
Cu-ZSM-5
Cage units
Cu-Y
Hybird structured zeolites
Cu-Offretite Cu-ZSM-34
Cu2+ Cuo
Cu-UZM-12
NOx + NH3 + O2 Straight channel zeolites
N2 + H2O
Cu+
Cu-SSZ-17 Cu-SSZ-13 Cu-SSZ-16 0
100 200 300 400 500 Amount (μmol/g)
Figure 38.1 Catalytic activity profiles for NH3 -SCR of NOx over Cu zeolites. Source: Reprinted with permission from Wang et al. [29]. Copyright 2018, Elsevier B.V.
687
688
38 Catalytic Abatement of NOx Emissions over the Zeolite Catalysts
due to their small pore openings. For the exchanged Cu sites, straight channels favor the formation of dimeric [Cu–O–Cu]2+ , while eight-membered ring enclosed cages favor the formation of isolated monomers Cu2+ and Cu(OH)+ . It was realized that cage-type zeolites favored the formation of copper cations, especially generating much more Cu+ ions than the others, rather than CuO [30].
38.2 Essential Nature of Novel Cu–CHA catalyst 38.2.1
Shape Selectivity
CHA-type SSZ-13 has a three-dimensional tetrahedral framework composed of double six-membered rings (d6r) in an AABBCC sequence, forming a “cage” per unit cell as depicted in Figure 38.2. It has been pointed out that SSZ-13 exhibits outstanding SCR performance (∼100% NO conversion) and excellent N2 yields (>95%) across a wide temperature range of 150–400 ∘ C after copper ion exchange [30, 32]. This remarkable performance is certainly attributable to the special topology of the SSZ-13 zeolite. a = b = 13.719 Å A A B B C C A
c = 14.953 Å
Figure 38.2 Hexagonal unit cell of an SSZ-13 zeolite (dashed lines) illustrating the AABBCCAA stacking sequence and equivalent rhombohedral unit cell (solid lines). The zeolite cage (delimited by 12 4-MR, 6 8-MR, and 2 highlighted 6-MR) is depicted as well. Light gray spheres (yellow line) are Si atoms and dark gray spheres (red line) are O atoms. Source: Reprinted with permission from McEwen et al. [31]. Copyright 2012, Elsevier. (See online version for color figure).
A
An attempt has been made to correlate the SCR behavior with the unique topology of SSZ-13 on the basis of the “shape selectivity” concept [33], which means that only the reactant, intermediate, and product of a certain size and shape can penetrate into the interior of the zeolite pores and undergo reaction at catalytically active sites. The supersmall pores (∼3.80 Å, 8-membered rings [MR]) of SSZ-13, which serve as the exits of the reaction channels, are smaller than the dimensions of the undesired by-products (NO2 [>3.83 Å] and N2 O [3.83 Å]); however, they are passable by the target product (N2 [3.64 Å]). Accordingly, the pore size effect may provide an explanation for the lesser amounts of harmful NO2 and N2 O by-products and the high N2 yields achieved over Cu–SSZ-13. In addition, it has been widely accepted that both C3 H6 and SO2 often cause deactivation during the catalytic elimination of NO by Cu zeolites. Although C3 H6 can participate in the SCR reaction as a reductant, it has the disadvantage of giving rise to carbonaceous deposition on the reactive centers, especially at low temperatures [28, 34]. Moreover, SO2 poisoning is frequently observed, primarily because of its reaction with Cu active sites to form CuSO3 or CuSO4 species at T < 200 ∘ C. Fortunately, the pore openings (3.8 Å) of Cu–SSZ-13 are
38.2 Essential Nature of Novel Cu–CHA catalyst
smaller than the kinetic diameters of both C3 H6 (4.68 Å) and SO2 (4.11 Å). As a result, the pore size restriction can greatly inhibit the diffusion of C3 H6 and SO2 molecules into the main channels of SSZ-13, thereby preserving a majority of the active sites from poisoning. In other words, the shape selectivity exhibited by SSZ-13 due to its supersmall pores is confirmed to be responsible for its high resistance to C3 H6 and SO2 poisoning. 38.2.2
Cation Location
Due to its unique structural topologies with supersmall pores, the chemical status of active sites constituted one of the hottest topics during NH3 -SCR investigation on the CHA-type zeolite catalysts. The location of Cu2+ cations inside the zeolite pores was investigated by using Rietveld refinement of synchrotron-based powder X-ray diffraction (XRD) [35]. The results suggested that Cu2+ ions were mostly isolated and exclusively occupied in the plane of 6-MR of SSZ-13. Moreover, XRD patterns further suggested that the thermal stability of SSZ-13 was improved significantly after copper introduction compared with the parent zeolite with an acidic form. The characterization methods of electron paramagnetic resonance (EPR) and temperature-programmed reduction with hydrogen (H2 –TPR) were used to study the distribution of Cu species for Cu–SSZ-13 prepared by the wet ion-exchange (WIE) method [27]. Five kinds of locations for Cu2+ cations (A–E) were proposed, as depicted in Figure 38.3, showing one unit cell of CHA topology. The following conclusions were accordingly reached: (i) Only under dehydrated conditions and low Cu loading amount (ion-exchange level 80%). The second Cu2+ could be located Figure 38.3 Schematic of the SSZ-13 hexagonal unit cell structure and possible Cu2+ locations. Source: Reprinted with permission from Gao et al. [27]. Copyright 2013, Elsevier.
C d ≈ 3.8 Å
A′ A D
E B
a = b = 13.416 Å
c = 15.044 Å
689
38 Catalytic Abatement of NOx Emissions over the Zeolite Catalysts
in position B or C if it stays close to a 6-MR; or instead, it might be located inside the large cage if it is close to an 8-MR (position D or E). The density functional theory (DFT) method was used to check the location and energy of Cu ions of Cu–SSZ-13 zeolite [36]. The isolated Cu2+ cations were confirmed to be more favorably located at 6-MR, which is consistent with those reported by experimental approaches [37]. However, in the presence of various adsorbates with –OH ligand, for example, [CuII (OH)]+ , the extra-framework site in the 8-MR was found to be more energetically stable than that of 6-MR for Cu2+ ions. Furthermore, the effect of Si/Al and Cu/Al ratios was also taken into account during DFT calculations. It was revealed that the Cu ions could be three- and fourfold coordinated to the lattice O atoms of Cu–SSZ-13 with different Si/Al ratios, as shown in Figure 38.4 for ZCu and Z2 Cu models. Examination of the partial density of state (PDOS) further verified that the Cu ions were, respectively, in +1 and +2 oxidation states in ZCu and Z2 Cu. Additionally, the infrared (IR) vibration frequencies of NO adsorption on Cu+ , Cu2+ , and [CuII (OH)]+ sites were DFT calculated, being located in the range of 1770–1808, 1850–1950, and 1870–1915 cm−1 , respectively, and also consistent with those derived from the IR experimental results. This finding verified the correctness of DFT simulations [31]. Based on the DFT, the oxidation state and coordination environment of Cu active sites in Cu–SSZ-13 were also investigated during NH3 -SCR. This indicated that the fourfold coordinated Cu(II) was the dominant Cu species under the “fast SCR” and “slow SCR” conditions, wherein the NO2 /NOx ratios were 0.5 and 1, respectively. Under the standard SCR conditions, containing no NO2 in the feed, the mixed Cu(I) and Cu(II) oxidation states could be both observed. As
Density of state (state/eV)
12
Cu+
10 8
1.948 O3
6 4
O3
2
2.004
0
–7
2.363
Cu O3
Al
–6
–5
9.7 e –4
–3
–2
–1
0
1
(a) Density of state (state/eV)
690
Cu2+
4 2 0
1.997 O3 2.010 O3 Al Al Cu 2.061 O2 O3
–2
4.3 e 5.0 e
2.102 –7
(b)
Spin Spin
–6
–5
–4 –3 –2 Energy (eV)
–1
0
1
Figure 38.4 Partial density of state (PDOS) of Cu 3d states in (a) ZCu and (b) Z2 Cu. Insets are the local structures of Cu in the 6-MR sites. Cu—O bond lengths are indicated in units of Å. The numbers of integrated electrons from PDOS plots are shown. In (b), the dashed and solid lines show spin-up and spin-down states for Cu 3d, respectively, and the dotted vertical line highlights the location of the Fermi level. Source: Reprinted with permission from Zhang et al. [36]. Copyright 2014, American Chemical Society.
38.2 Essential Nature of Novel Cu–CHA catalyst
also reported, twofold Cu(I) and fourfold Cu(II) bound with H2 O or OH were found to be the most stable species over a wide range of deNOx conditions [36]. 38.2.3
Copper Status
As stated earlier, the Si/Al ratio could influence the dispersion as well as the detailed chemical status of the introduced Cu active species. Combination of the experimental and theoretical (DFT) approaches could investigate the chemical status of Cu species doped on SSZ-13 [38]. This indicates that at least two kinds of Cu species existed over this zeolite catalyst, which was closely dependent on the ratio of copper species to total Al atoms in zeolite (Cu/Altotal ). As Cu/Altotal < 0.2, Cu ions were dominant in the status of isolated Cu2+ cations located near the 6-MR, while Cux Oy species could be formed as Cu/Altotal > 0.2 being located in the 8-MR. A series of Cu-exchanged SSZ-13 catalysts (Cu/Altotal = 0–0.41) for the standard NH3 -SCR were investigated to verify the effect of Cu/Altotal ratio on copper status [39]. The isolated Cu2+ ions, acting as the active centers, were confirmed to be located at the 6-MR of SSZ-13 during NH3 -SCR as Cu/Altotal < 0.2, and the standard SCR reaction rate increased linearly up to Cu/Altotal = 0.2, with a maximum of 3.8 × 10−6 mol NO/g cat/s. Because NH3 acting as the actual reducing agent plays a key role in SCR reaction, investigations to identify the species formed by NH3 adsorption upon Cu–SSZ-13 and their involvement were put forward based on in situ EPR, solid-state nuclear magnetic resonance (NMR), and DFT calculations [40]. Five kinds of NH3 adsorption modes were observed under different conditions: [Cu(NH3 )5 ]2+ , [Cu(Of )2 (NH3 )2 ]2+ , [Cu(Of )3 NH3 ]2+ , [Cu(NH3 )2 ]+ , and [CuOf (NH3 )]+ (Of representing the framework oxygen). The adsorbed NH3 was demonstrated to be able to reduce Cu2+ ions into Cu+ ions. 38.2.4
CHA-Type Silicoaluminophosphate
Silicoaluminophosphate-34 (SAPO-34), being generated from the incorporation of an Si atom into neutral AlPO4 , has the same spatial topology as that of SSZ-13. According to the literature, Cu–SAPO-34 catalysts exhibit outstanding activities and durability for the NH3 -SCR process, and their active site status has attracted an especially large amount of attention. In an early study, a series of Cu–SAPO-34 samples were prepared by a WIE and precipitation method for NO reduction [41]. Various techniques (XRD, H2 –TPR, scanning transmission electron microscopy [STEM], and diffused reflectance infrared Fourier transform spectroscopy [DRIFTS]) were used to identify the location and status of Cu species in these samples. The results consistently indicated that the Cu species existed predominantly as isolated ions at the exchange sites in the ion-exchanged samples; however, as for the precipitated sample, CuO on the external surface was the dominant species. Superior NH3 -SCR activity was observed for the ion-exchanged samples, suggesting that isolated Cu cations at the exchange sites constituted the active centers. A series of Cu–SAPO-34 samples with various Cu loadings (0.7–3.0 wt%) through the solid-state ion-exchange (SSIE) method were also prepared [41]. The chemical statuses of the loaded Cu species were characterized by in situ DRIFTS, XRD,
691
692
38 Catalytic Abatement of NOx Emissions over the Zeolite Catalysts
H2 –TPR, and ultraviolet and visible spectroscopy (UV/Vis), which suggest that two different Cu species existed on the prepared Cu–SAPO-34 samples: Cu2+ ions and Cux Oy clusters (dimeric or oligomeric Cu species). Cu2+ ions were verified to be the active centers for NO abatement, while Cux Oy could promote the NH3 oxidation, leading to the observed decrease in the standard SCR process at high temperatures. The acidity–activity correlation for Cu–SAPO-34 during NH3 -SCR was subsequently studied by the same group, wherein a series of Cu–SAPO-34 samples with varying numbers of Brönsted acid sites were prepared by WIE with potassium to investigate the role of the Brönsted acidity. Along with an increase in the potassium loading, the SCR activity of the Cu/K–SAPO-34 catalysts diminished in accordance with the decreasing Brönsted acidity. The reaction rate was further found to be dependent on the NH3 coverage on Brönsted acid sites at 180–280 ∘ C. At elevated temperatures, the acidic sites could act as a source of NH3 for the SCR reaction because the NH4 + was initially adsorbed on the Brönsted acid sites and could then gradually migrate to the copper sites to finally react with the NOx species.
38.3 SCR Reaction Mechanism NH3 -SCR process generally occurs via three types of reaction paths according to the literature [42], which depends on the fraction of NO2 attended in reaction as listed in Eqs. (38.1)–(38.4): 4NO + 4NH3 + O2 → 4N2 + 6H2 O (standard SCR)
(38.1)
NO + NO2 + 2NH3 → 2N2 + 3H2 O (fast SCR)
(38.2)
6NO2 + 8NH3 → 7N2 + 12H2 O (slow SCR to N2 )
(38.3)
2NO2 + 2NH3 → N2 + N2 O + 3H2 O (slow SCR to N2 O)
(38.4)
As for the standard SCR, many authors suggest that the reaction starts with an oxidation of NO to NO2 on the active sites. Subsequently, NO2 can react with the adsorbed NH3 , yielding NH4 NO2 or NH4 NO3 , which can be further decomposed into the final products of N2 and H2 O as well as the undesired pollutants of N2 O and NOx [43]. The NO oxidation into NO2 is reported to be the rate-determining step. Most researchers believed that the standard SCR could take place over the zeolite catalysts, fulfilling the Langmuir–Hinshelwood mechanism. However, strange SCR behavior was observed in the case of Cu–SSZ-13 prepared through one-pot synthesis [44]. Although the “fast SCR mechanism,” the first step of which being a rapid oxidation of partial NO resulting in a NO + NO2 mixture, is typical in effect for the catalytic reduction of NO by NH3 at low temperatures, it has been proposed that the “standard SCR mechanism” is more relevant for Cu–SSZ-13 because of the lesser NO2 generation caused by a transition state constraint imposed by the small window of the CHA cage. However, the standard SCR mechanism is still a subject of debate, which is mainly because the generated intermediates are so active that they cannot be readily identified. An intermediate of nitrite/HONO during a
38.3 SCR Reaction Mechanism
standard SCR mechanism study of Fe–ZSM-5 was found to exist by a novel method [45]. During the research, BaO/Al2 O3 , known as an LNT (lean NOx trap) catalyst, was mixed with the SCR catalyst of Fe–ZSM-5. A trap experiment was thereafter conducted by exposing this mixed catalyst under an atmosphere of NO and O2 at a low temperature (120 ∘ C) [46]. As reported, the inclusion of BaO/Al2 O3 in a physical mixture with Fe–ZSM-5 resulted in stabilizing nitrite species adsorbed on Fe sites after exposure to NO + O2 and captured upon BaO via gas-phase equilibrium with HONO. Finally, an alternative mechanism for the standard SCR process was also proposed: (i) the nitrite species (NO2 − ) in equilibrium with gaseous nitrous acid (HONO) could be initially generated in the presence of NO and O2 ; (ii) the formed nitrite/HONO species were subsequently decomposed into gaseous NO2 to a certain degree. However, in the presence of NH3 , the nitrite/HONO could quickly react with NH3 to form the final product of N2 . It was also reported that NO oxidation to NO2 hardly was the rate-determining step of the standard SCR reaction, whereas the nitrite/HONO served as the most important intermediate. Many researchers suggest that most of typical microporous zeolites (such as ZSM-5, Y, and β) favor “fast SCR mechanism,” due to which the pore size of zeolites are larger than the dimensions of the undesired by-products. Different copper species have different effects on the SCR mechanism of Cu zeolite catalysts [28]. The CuO generated at high Cu loadings can catalyze the oxidation of NO into NO2 , which is favorable for low-temperature deNOx activity because the NH3 + NO + NO2 (fast SCR) and NH3 + NO2 (slow SCR) reactions are known to be faster than the NH3 + NO reaction (standard SCR) at low temperatures. An overall SCR scheme as a function of the NO2 /NOx ratio as well as reaction temperature was thereafter profiled, as shown in Figure 38.5. It is seen that a NO2 SCR
Standard SCR 2NO2(ads) Common
N2 + H2O
reaction
+H2O
>250 °C
O2
[HONO](ads)+[HNO3](ads)
NO Temperature (°C)
NH3 N2+[NH4 +...NO3–](ads) NO >180 °C
NO2 NO2
>200 °C
N2O + H2O
Low-temperature route 250 °C
NH3
50 NO2/NOx ratio (%)
NH4NO3(s) 100
Figure 38.5 Proposed overall SCR scheme as a function of the NO2 /NOx ratio and temperature. Source: Reprinted with permission from Iwasaki and Shinjoh [46]. Copyright 2010, Elsevier.
693
694
38 Catalytic Abatement of NOx Emissions over the Zeolite Catalysts
common step in each kind of SCR mechanism is the formation of surface species of adsorbed ammonium nitrate. These mechanism studies are beneficial for clarifying the structure–performance relationship and give us ideas on the design of SCR catalyst with considerable activity. The main differences among the three pathways are reaction sequence and reaction rate. NO is easy to be oxidized to NO2 , and for “fast SCR” and “slow SCR,” NO2 is used as a reactant to combine with NH3 to form NH4 NO3 and further react with NO to form NH4 NO2 (rate-determining step) to decompose into N2 and H2 O. In the standard SCR, NO as a reactant combines with NH3 to directly form NH4 NO2 and decompose into N2 and H2 O. The reaction sequences of the three pathways are distinguishable, and since the reaction of NH4 NO3 to NH4 NO2 is the rate-determining step, the reaction rates are accordingly different.
38.4 Conclusions and Perspectives In brief, zeolites are so far the superior catalysts for the reduction of NOx . Zeolites are highly porous materials having special pore structures with diverse topologies, associated with relatively high surface areas [47]. Their ion-exchange ability is beneficial for the better dispersion of active components such as cations [48, 49]. Simultaneously, static electrical field of the zeolite framework facilitates the activation of reactants [50]. The aforementioned special characteristics of zeolites make them promising materials served as SCR catalysts. Furthermore, acidity will help in stabilizing NH3 at high temperatures [51, 52], which is the reducing agent for NO elimination. The low cost and the convenience of industrialization have promoted the rapid development of zeolite catalysts. Perspectives in deNOx reaction system are listed as follows: (a) NH3 -SCR has become one of the most promising deNOx methods under lean conditions. However, due to some drawbacks of commercialized V–W–Ti catalysts, more attention is now paid to the development of new superiorly active materials for NOx removal. Fortunately, Cu–SSZ-13 zeolite has recently been proposed to exhibit extremely high low-temperature activity and N2 selectivity. The related structure–activity relationship is also illustrated on the basis of both experimental and theoretical approaches. However, using the very expensive and toxic template of N,N,N-trimethyl-1-adamantylammonium hydroxide (TMA-daOH) for Cu–SSZ-13 synthesis leads to a significant increase in the catalyst cost, which seriously prevents industrial realization. Therefore, developing an economical way to synthesize SSZ-13 by using a cheap template has become an important direction for the NH3 -SCR technique. (b) On the basis of the concept of “quasi shape selectivity” illustrated for SSZ-13, investigations of other microporous zeolite catalysts with supersmall apertures, such as OFF (3.6 × 4.9 Å2 ), ERI (3.6 × 5.1 Å2 ), LEV (3.6 × 4.8 Å2 ), and AFX (3.6 × 3.4 Å2 ), should also be taken into account for the NH3 -SCR study to obtain a common principle for the design of highly efficient zeolite catalysts.
References
(c) Developing deNOx technologies other than the traditional NH3 -SCR, such as H2 -SCR and NO direct decomposition. Although some research has already been done on these topics, they still need a lot of efforts. (d) The hot topic of special structures in zeolites, such as “hierarchical,” “core–shell,” “yolk–shell,” may play a role in the deNOx reaction. (e) Theoretical simulation based on DFT is believed to be promising to describe the structure and local environment of active sites, as well as to determine the depollution mechanism at the molecular or atomic level. However, due to the complications of some depollution systems, such as NO-SCR involving many molecules (NO, O2 , and reductant), causing a huge amount of calculation, and being more uncertain, related DFT studies on the deNOx mechanism were scarcely conducted. Hence, these mechanism simulations might become prevalent in the future with the development of computational technology.
References 1 Vaccaro, A.R., Mul, G., Pérez-RamíRez, J. et al. (2003). On the activation of
2
3
4
5
6
7
8
9
Pt/Al2 O3 catalysts in HC–SCR by sintering: determination of redox-active sites using multitrack. Appl. Catal., B 46: 687–702. Bion, N., Saussey, J., Haneda, M. et al. (2003). Study by in situ FTIR spectroscopy of the SCR of NOx by ethanol on Ag/Al2 O3 – evidence of the role of isocyanate species. J. Catal. 217: 47–58. Guenin, M., Breysse, M., Frety, R. et al. (1987). Resistance to sulfur poisoning of metal catalysts: Dehydrogenation of cyclohexane on Pt/Al2 O3 catalysts. J. Catal. 105: 144–154. Corro, G., Cano, C., and Fierro, J.L.G. (2010). A study of Pt–Pd/γ-Al2 O3 catalysts for methane oxidation resistant to deactivation by sulfur poisoning. J. Mol. Catal. A: Chem 315: 35–42. Angelidis, T.N., Christoforou, S., Bongiovanni, A. et al. (2002). On the promotion by SO2 of the SCR process over Ag/Al2 O3 : influence of SO2 concentration with C3 H6 versus C3 H8 as reductant. Appl. Catal., B 39: 197–204. Jagtap, N., Umbarkar, S.B., Miquel, P. et al. (2009). Support modification to improve the sulphur tolerance of Ag/Al2 O3 for SCR of NOx with propene under lean-burn conditions. Appl. Catal., B 90: 416–425. Wang, J., He, H., Feng, Q. et al. (2004). Selective catalytic reduction of NOx with C3 H6 over an Ag/Al2 O3 catalyst with a small quantity of noble metal. Catal. Today 93: 783–789. Kasaoka, S., Sasaoka, E., and Senda, K. (1977). Catalysts of V2 O5 -TiO2 & system for reduction of nitrogen oxide with ammonia. J. Fuel Soc. Jpn. 56: 818–823. Phil, H.H., Reddy, M.P., Kumar, P.A. et al. (2008). SO2 resistant antimony promoted V2 O5 /TiO2 catalyst for NH3 -SCR of NOx at low temperatures. Appl. Catal., B 78: 301–308.
695
696
38 Catalytic Abatement of NOx Emissions over the Zeolite Catalysts
10 Wang, X., Wang, J., and Zhu, T.Y. (2013). Coupled control of chlorobenzene
11
12
13
14
15
16
17
18
19 20
21
22
23
24 25
and NO over V2 O5 /TiO2 catalyst in NH3 -SCR reaction. Adv. Mater. Res. 811: 83–86. Jung, S.M. and Grange, P. (2001). Characterization and reactivity of V2 O5 –WO3 supported on TiO2 –SO4 2− catalyst for the SCR reaction. Appl. Catal., B 32: 123–131. Morikawa, S., Takahashi, K., Yoshida, H. et al. (1982). Life of V2 O5 –TiO2 catalyst for the NOx reduction with NH3 in flue gas from oil fired boiler. J. Fuel Soc. Jpn. 61: 1024–1030. Wan, Q., Duan, L., He, K.B. et al. (2011). Removal of gaseous elemental mercury over cerium doped low vanadium loading V2 O5 –WO3 /TiO2 in simulated coal-fired flue gas. Environ. Sci. 32: 2800–2804. Shen, B., Chen, J., Yue, S., and Li, G. (2015). A comparative study of modified cotton biochar and activated carbon based catalysts in low temperature SCR. Fuel 156: 47–53. Wei, L., Shan, T., Yun, S. et al. (2015). Utilization of Sargassum based activated carbon as a potential waste derived catalyst for low temperature selective catalytic reduction of nitric oxides. Fuel 160: 35–42. Boyano, A., Gálvez, M.E., Moliner, R. et al. (2008). Carbon-based catalytic briquettes for the reduction of NO: effect of H2 SO4 and HNO3 carbon support treatment. Fuel 87: 2058–2068. Shen, B.X. and Liu, T. (2010). Deactivation of MnOx –CeOx /ACF catalysts for low-temperature NH3 -SCR in the presence of SO2 . Acta Phys. Chim. Sin. 26: 3009–3016. Li, Y., Guo, Y., Xiong, J. et al. (2016). The roles of sulfur-containing species in selective catalytic reduction of NO with NH3 over activated carbon. Ind. Eng. Chem. Res. 55: 12341–12349. Baerlocher, Ch. and McCusker, L.B. Database of Zeolite Structures: http:// www.iza-structure.org/databases/. Delahay, R., Kieger, S., Tanchoux, N. et al. (2004). Kinetics of the selective catalytic reduction of NO by NH3 on a Cu-faujasite catalyst. Appl. Catal., B 52: 251–257. Komatsu, T., Nunokawa, M., Moon, I.S. et al. (1994). Kinetic studies of reduction of nitric oxide with ammonia on Cu2+ -exchanged zeolites. J. Catal. 148: 427–437. Rahkamaa-Tolonen, K., Maunula, T., Lomma, M. et al. (2005). The effect of NO2 on the activity of fresh and aged zeolite catalysts in the NH3 -SCR reaction. Catal. Today 100: 217–222. Kröcher, O., Devadas, M., Elsener, M. et al. (2006). Investigation of the selective catalytic reduction of NO by NH3 on Fe–ZSM5 monolith catalysts. Appl. Catal., B 66: 208–216. Long, R.Q. and Yang, R.T. (2001). Selective catalytic oxidation (SCO) of ammonia to nitrogen over Fe-exchanged zeolites. J. Catal. 201: 145–152. Brandenberger, S., Krocher, O., Tissler, A., and Althoff, R. (2008). The state of the art in selective catalytic reduction of NOx by ammonia using metal-exchanged zeolite catalysts. Catal. Rev. 50: 492–531.
References
26 Bartholomew, C.H. (2001). Mechanisms of catalyst deactivation. Appl. Catal.,
A 212: 17–60. 27 Gao, F., Walter, E.D., Karp, E.M. et al. (2013). Structure–activity relationships
28
29
30
31
32
33
34
35
36
37
38 39
40
in NH3 -SCR over Cu–SSZ-13 as probed by reaction kinetics and EPR studies. J. Catal. 300: 20–29. Pereda-Ayo, B., De La Torre, U., Illán-Gómez, M.J. et al. (2014). Role of the different copper species on the activity of Cu/zeolite catalysts for SCR of NOx with NH3 . Appl. Catal., B 147 (5): 420–428. Wang, H., Xu, R., Yi, J., and Zhang, R.D. (2019). Zeolite structure effects on Cu active center, SCR performance and stability of Cu-zeolite catalysts. Catal. Today 327: 295–307. Chen, B., Xu, R., Zhang, R., and Liu, N. (2014). Economical way to synthesize SSZ-13 with abundant ion-exchanged Cu+ for an extraordinary performance in selective catalytic reduction (SCR) of NOx by ammonia. Environ. Sci. Technol. 48: 13909–13916. McEwen, J.S., Anggara, T., Schneider, W.F. et al. (2012). Integrated operando X-ray absorption and DFT characterization of Cu–SSZ-13 exchange sites during the selective catalytic reduction of NOx with NH3 . Catal. Today 184: 129–144. Kwak, J.H., Tonkyn, R.G., Kim, D.H. et al. (2010). Excellent activity and selectivity of Cu–SSZ-13 in the selective catalytic reduction of NOx with NH3 . J. Catal. 275 (2): 187–190. Hussain, M., Fino, D., and Russo, N. (2014). Development of modified KIT-6 and SBA-15-spherical supported Rh catalysts for N2 O abatement: from powder to monolith supported catalysts. Chem. Eng. J. 238: 198–205. Ye, Q., Wang, L., and Yang, R.T. (2012). Activity, propene poisoning resistance and hydrothermal stability of copper exchanged chabazite-like zeolite catalysts for SCR of NO with ammonia in comparison to Cu/ZSM-5. Appl. Catal., A 427–428: 24–34. Fickel, D.W. and Lobo, R.F. (2010). Copper coordination in Cu–SSZ-13 and Cu–SSZ-16 investigated by variable-temperature XRD. J. Phys. Chem. C 114: 1633–1640. Zhang, R.Q., McEwen, J.S., Kollár, M. et al. (2014). NO chemisorption on Cu/SSZ-13: a comparative study from infrared spectroscopy and DFT calculations. ACS Catal. 4: 4093–4105. Deka, U., Juhin, A., Eilertsen, E.A. et al. (2012). Confirmation of isolated Cu2+ ions in SSZ-13 zeolite as active sites in NH3 -selective catalytic reduction. J. Phys. Chem. C 116: 4809–4818. Verma, A.A., Bates, S.A., Anggara, T. et al. (2014). NO oxidation: a probe reaction on Cu–SSZ-13. J. Catal. 312: 179–190. Bates, S.A., Verma, A., Paolucci, C. et al. (2014). Identification of the active Cu site in standard selective catalytic reduction with ammonia on Cu–SSZ-13. J. Catal. 312: 87–97. Moreno-González, M., Hueso, B., Boronat, M. et al. (2015). Ammonia-containing species formed in Cu-chabazite as per in situ EPR, solid-state NMR, and DFT calculations. J. Phys. Chem. Lett. 6: 1011–1017.
697
698
38 Catalytic Abatement of NOx Emissions over the Zeolite Catalysts
41 Wang, L., Li, W., Schmieg, S.J., and Weng, D. (2015). Role of Brönsted acid-
42
43
44
45
46 47
48
49 50
51
52
ity in NH3 selective catalytic reduction reaction on Cu/SAPO-34 catalysts. J. Catal. 324: 98–106. Zhang, R.D., Liu, N., and Lei, Z. (2016). Selective transformation of various nitrogen-containing exhaust gases toward N2 over zeolite catalysts. Chem. Rev. 116 (6): 3658. Wang, D., Zhang, L., Kamasamudram, K., and Epling, W.S. (2013). In situ-DRIFTS study of selective catalytic reduction of NOx by NH3 over Cu-exchanged SAPO-34. ACS Catal. 3: 871–881. Xie, L., Liu, F., Ren, L. et al. (2014). Excellent performance of one-pot synthesized Cu–SSZ-13 catalyst for the selective catalytic reduction of NOx with NH3 . Environ. Sci. Technol. 48: 566–572. Ruggeri, M.P., Selleri, T., Colombo, M. et al. (2014). Identification of nitrites/HONO as primary products of NO oxidation over Fe–ZSM-5 and their role in the standard SCR mechanism: a chemical trapping study. J. Catal. 311: 266–270. Iwasaki, M. and Shinjoh, H. (2010). A comparative study of “standard”, “fast” and “NO2 ” SCR reactions over Fe/zeolite catalyst. Appl. Catal., A 390: 71–77. Puertolas, B., Garcia-Andujar, L., Garcia, T. et al. (2014). Bifunctional Cu/H–ZSM-5 zeolite with hierarchical porosity for hydrocarbon abatement under cold-start conditions. Appl. Catal., B 154–155: 161–170. Hunger, B., Heuchel, M., Clark, L.A., and Snurr, R.Q. (2002). Characterization of acidic OH groups in zeolites of different types: an interpretation of NH3 –TPD results in the light of confinement effects. J. Phys. Chem. B 106: 3882–3889. Weitkamp, J. (2000). Zeolites and catalysis. Solid State Ionics 131: 175–188. Liu, N., Zhang, R., Li, Y., and Chen, B. (2014). Local electric field effect of TMI (Fe, Co, Cu)-BEA on N2 O direct dissociation. J. Phys. Chem. C 118: 10944–109560. Miyamoto, Y., Katada, N., and Niwa, M. (2000). Acidity of β zeolite with different Si/Al ratio as measured by temperature programmed desorption of ammonia. Microporous Mesoporous Mater. 40: 271–281. Zhang, W., Burckle, E.C., and Smirniotis, P.G. (1999). Characterization of the acidity of ultrastable Y, mordenite, and ZSM-12 via NH3 -stepwise temperature programmed desorption and Fourier transform infrared spectroscopy. Microporous Mesoporous Mater. 33: 173–185.
699
Index a Ax By Cz Oq mixed oxide structures 542 ab initio molecular dynamics 419–437, 454 absorption energies, of reactants 170 absorption-CT imaging 265–266 accelerated degradation test (ADT) 271 acetylene cyclotrimerization 104, 108 acetylenic chains 63 acid leaching 71, 136–137 acid liquid electrolyte water electrolyser (ACIWE) 535 activated carbon (AC) 57, 58, 60, 68, 74, 82, 89, 109, 120, 551, 605, 625, 671, 677, 686 activated carbon fibers (ACF) 686 activation losses 583, 584, 588 active carbon-supported nickel–tungsten carbide 605 active HER electrocatalysts 547 active sites 105–106, 152–156, 363–373 active species 92, 217, 295, 302, 357, 363–373, 691 profile of 365 activity volcano curve 507 adsorbate–adsorbate interactions for CO methanation 516 adsorbate interactions 233, 507, 516–518 adsorbed CO species, classification of 322, 325
adsorption on metal oxides CO adsorption, on TiO2 (110) 237–239 HCOOH adsorption, TiO2 239–243 adsorption process 230, 231, 234, 463, 507 adsorption, on solid surfaces kinetically restricted adsorbate structures 229–231 supported nanoparticle catalysts reactions 244–245 thermodynamically driven reactions 234, 237 advanced NMR 123, 136, 153, 220, 289, 295, 307, 358, 691 advanced oxidation processes (AOPs) 681 aerosol particles formation 183, 184 aerosol syntheses 183 AgAu bimetallic alloy NPs 167 agglomeration 50, 67, 128, 172, 184–186, 273, 575, 585, 661 Ag1 H2 PW 600 air-borne particles (or aerosols) 183 alcohol dehydration catalyst 189 alcohols synthesis 509–511 algae feedstocks 595 algebraic reconstruction technique (ART) 268 aligned nanotube 45 alkali metal hydroxide solution 138
Heterogeneous Catalysts: Advanced Design, Characterization and Applications, First Edition. Edited by Wey Yang Teoh, Atsushi Urakawa, Yun Hau Ng, and Patrick Sit. © 2021 WILEY-VCH GmbH. Published 2021 by WILEY-VCH GmbH.
700
Index
alkaline liquid electrolyte water electrolyser (ALKWE) 535–537, 539–541, 552 alkoxysilanols 9 alkynes 13, 85 Al2 O3 600 Al2 O3 supported Ni and Ni-Cu catalysts 608 aluminate (AlO4 5− ) 9, 172, 194, 273 aluminium chloride 8, 131 aluminophosphate 9, 131, 134 aluminosilicates 8, 119, 600, 619, 686 Amberlyst-15 599, 600, 603–605, 609 Amberlyst-131 601 amine organocatalysis 156 amino functionalized material 151–152, 155 aminoterephthalates 151 ammonia synthesis 5, 6, 533, 633, 636, 641 amorphous or crystalline structure 46 amphiphilic organosilanes 130 anion exchange membrane water electrolyzer (AEMWE) 539 anion-doped TiO2 560 anion-ordered A2 B2 X6 X′ 0.5 -type pyrochlore structure 564 anionic silicate oligomers 130 anisotropic adsorption 27–28 anisotropic nanostructures 40, 42–44 anisotropic NPs, encapsulation of 165 anisotropic photocatalysts 11 anisotropic properties, of crystal facets 22, 27 anisotropic surface energy 22 anisotropic surface properties 32 anisotropy of crystal lattice 27, 28 of hexagonal CdS nanorods 165 anodic electrophoretic deposition 48, 49 anodization 11, 15, 40–46, 50–52 catalytic applications 44 formation of, anodic oxide layer 42 growth mechanism, anodized metal foil 43 pulse/step anodization 45–46
antibiotics 530 anti-bonding orbital of H2 71, 80, 81, 321, 566 of N atom 71 apparent diffusion coefficient 169 aqueous polyalcohols 13 arabinose 617 arene (C6 R6 ) ligands 85 arithmetic mean velocity 226 aromatic cyclopentadienyl (C5 R5 – ) 85 aromatic polymer lignin 595 array nanostructures, of nanotubes 42 ascorbic acid 62 “as-deposited” cluster-based catalysts 90, 94 atom motion 217 atomic and molecular sorption 217 atomic force microscopy 166, 253, 291 atomic layer deposition (ALD) 72–73, 111 atomic orbital (AO) 29, 71, 80, 296 atomic probe tomography (APT) 272–273 atomically-thick CoSe2 nanosheets 545 ATR-IR spectroscopy 315–317, 327–328, 331 alcohol oxidation 331 Au10 -cluster-based catalyst 79–96 AuNi alloys 167 AuNi Janus spindle nanostructures 165 AuSiO2 Janus NPs 161, 167–169 automobile three-way catalyst 190 automotive catalysts 354 Avantium 599, 624 Avantium employed EMF 599 average electron transfer number 580 Avogadro’s constant 227
b π back-donation 71, 321 background spectrum 312, 318, 319 “ball-and-stick” representations 86, 87 bamboo-type nanotubes 45 band engineering, of semiconductors cocatalysts, loading of 212
Index
impregnation method 212 photodeposition method 212 photocatalyst materials 203–204 band structures design 206–210 main constituent metal elements 205–206 variety of 204–205 synthesis methods, of semiconductor photocatalysts 210–211 flux method 211 hydrothermal reaction 211 polymerized (polymerizable) complex method 211 precipitation method 212 solid-state reaction method 211 band-engineered photocatalysts 557 bandgap 11, 29, 32, 33, 67, 189, 203–210, 472, 673–677 barrier oxide layer 42 BASF 4, 6, 148, 633 Ba0.5 Sr0.5 Co0.8 Fe0.2 O3–𝛿 543 B-doped MWCNTs 551 Beer–Lambert law 264, 378 benzene oxidation 107 benzenedicarboxylate 148–150, 155 1,4-benzenedicarboxylic acid (BDC) linker 11 benzenetricarboxylate organic linkers 148 benzyl alcohol oxidation 92 BEP relation and the activation barrier 503–507 bidentate formate 241, 243, 331 bimetal interface 167 bimetallic Janus NP synthesis 168 bimetallic nanoparticles 70, 347, 608 bio-inspired hydrogen-producing catalyst 413, 415 biobutanol 594–596 biochemical methods 618, 650 biodiesel 595, 602, 639 bioethanol 594–596 biohydrogen 595, 596 biomass conversion 9, 11, 57, 528, 593, 610, 619, 640 biomass conversion/biomass upgrading 528
biorefinery concept 593 biphenyldicarboxylate linkers 151 1,4-biphenyldicarboxylic acid (BPDC) 11 bis(dicarbonylcyclopentadienyliron, Fe2 O4 C14 H10 67 bismuth subcarbonate 171 bismuth tungstate (Bi2 WO6 ) thin film 52 BiVO4 single crystal 31, 198, 209 Blyholder model 231 B3LYP functional 410 Bohr excitonic radius 13 Bohr magneton 296 bond-counting contribution factor (BCCF) 516, 517 Born–Oppenheimer (BO) approximation 406, 409, 420, 483 bottom-up fabrication 162, 174 Brønsted acid 136 Brønsted acid catalysts 600 Brønsted–Evans–Polanyi (BEP) relation 497, 498, 502–507, 511, 513, 516, 517, 519, 520 Brillouin zone 32 brown solids 130 Brownian collision 184 Bruker EMXPlus cw-EPR spectrometer 299 BSCF electrode 356 bulk MOFs 546 bulk MoS2 549 bulky molecules 121, 123, 140 1,3-butadiene 105 butadiyne linkages 63 n-butane 33, 128 1-butene 105 t-butyl hydroperoxid 170
c calcine 125 capillary plug-flow reactors 352 capped Ag NPs monolayers 167 capping agents 25, 26, 35, 61, 70, 86 carbocatalyst 57, 59
701
702
Index
carbohydrates 9, 57, 594–596, 602–605, 609, 610, 617–630 carbohydrate-containing biomass 596 carbonaceous materials 125, 662 carbon allotrope, sp3 hybridization 58, 63 carbon-based catalysts 57, 75, 686 carbon-based fuel 593 carbon-based semiconductor material 74 carbon capture and storage (CCS) technology 529, 649, 650 carbon capture and utilization (CCU) 650 carbon corrosion 266, 575, 585 carbon dioxide (CO2 ) emission 529 carbon emissions 527 carbon feedstock 528 carbon nano-onions 63 carbon nanotubes (CNTs) 58, 127, 497, 548, 551, 575, 587, 603, 662, 677 carbon nitride 11, 14, 64, 67–68, 74, 387, 551, 662, 676 carbon nitride (C3 N4 ) catalyst 387–389, 551 carbon steel reactor 6 carbon-supported copper-ruthenium catalyst (CuRu/C) 604 carbon-supported CuRu bimetallic catalyst 626 carbon-supported platinum nanoparticle catalyst 625 carboxylate ligand 148 carboxylic acids 13, 602, 628 carboxylic acid function 169 catalyst deactivation 217, 268 catalyst development 619–622, 630 catalyst fabrication methods 79 catalyst on graphene graphene, as starting material 60, 61 graphene derivatives 62–64, 75 graphene oxide, as precursor 61–62 catalyst precursor, deposition of 58
catalysts preparation, new carbon materials catalyst on GO 59, 60 catalyst on graphene graphene derivatives 62–64 graphene oxide, as precursor 61–62 graphene, as starting material 60–61 catalyst on nanodiamonds 63, 66 catalyst on onion like carbon 63 CVD, techniques 72–73 dry mechanical methods 73 electrodeposition 73 from hydrothermal carbonisation 68–69 photodeposition 74 SACs, on carbon nitrides 67, 68 simultaneous metallic catalyst 73 techniques 69–74 wet-chemistry synthesis, optimization of 70, 71 catalyst structure in flow reactors 353 catalytic benchmarking 167 catalytic CO-PrOx reaction 192 catalytic cracking 8–11, 171, 273, 291 catalytic deNOx 9 catalytic Janus nanostructures 165 catalytic Janus NPs 168 catalytic lignite-to-gasoline process 8 catalytic materials 5, 10, 39, 48, 52, 53, 81, 199, 253, 265–267, 271, 275, 286, 287, 317, 333, 342, 404, 497–520, 651, 656–658, 661, 663, 666, 685 catalytic oxidation 14, 63, 105, 168, 170, 290, 674 catalytic oxide films 42 catalytic pyrolysis, of biomass 124 catalytic reaction 5–8, 46, 363, 639–641 catalytic supercritical water gasification (CSCWG) 640 catalytic thin films 39–53 catalytic transformations 169, 290
Index
catalytically active sites (CASs) 103, 105–106, 283, 673, 688 identification 103, 105–106, 283, 673, 688 CatApp 502, 509 cathodic electrodeposition 41, 46–48, 52 pulse electrodeposition 47–48 cathodic electrolytic deposition 40 cathodic electrophoretic deposition 48, 49 cathodic pulse 47, 48 cationic metals 148 CATMAP 506, 511 C5 carbohydrates 617 CCD (charged-coupled device) cameras 257, 259, 284–285 CDFT 412 cell holder 582, 583 cellulose 595, 600, 602, 603, 609, 610, 617, 618, 620, 623 cerium-rich quaternary systems 543 cetyl trimethyl ammonium bromide (CTAB) 10, 13, 131, 132, 139 chalcogen-based ligands 85 charcoal-supported catalyst 68 charge, of an electron 28, 456 CHA-type silicoaluminophosphate 691–692 CHA-type SSZ-13 Cu–SAPO-34 catalysts 691 CHA-type zeolite catalysts 689 chemical compounds, purification of 86 chemical dissolution, of oxide layer 43 chemical equilibrium 503, 634, 637–639, 643–645 chemical etching 162, 165 chemical imaging techniques 265, 275 chemical methods, SAC bottom-up synthetic methods adsorption method 109–111 co-precipitation method 109, 110
galvanic-replacement method 112 top-down synthetic methods 112–113 chemical reaction 103, 155, 203, 217, 220, 225, 256–257, 259, 284, 286, 288, 331, 369, 378, 380, 385–386, 402–403, 412, 419, 423–424, 439, 455, 463, 472, 484, 518, 520, 571, 623, 635, 653 chemical transformation 220, 353, 363, 596 chemical vapor deposition (CVD) 11 12, 22, 69, 72, 172, 184, 190 chemically synthesised metal clusters 85–95 catalysis using homogeneous catalysts 89, 90, 92 chemiluminescent transformation 168 18 C-hexagonal pores 63 chloroplatinic acid method 576 chrome aluminophosphate 134 classical molecular dynamics (CMD) simulations 402 clusters 3–16, 71–72, 79–96 CO adsorption 231, 233, 237, 238, 321–324, 468 CO adsorption metal surfaces 231 Au nanoparticles, CeO2 233 CO adsorption, on Pt(111) 231–232 CO adsorption, on TiO2 (110) 237–238 coalescence 12, 71, 184–186, 190, 194 coarse-grain approach 402 cobalt aluminophosphate 134 cobalt catalysts 7 cobalt-time-yield (CTY) 194 cocatalyst 31, 204, 212, 564, 565, 567, 665 cocatalysts function 212 co-electrolytic SOEWE 541 CO2 electroreduction process 655, 658 co-embedded N-doped carbon nanotubes 548 CoFe@N-rGO 549
703
704
Index
coherent diffraction imaging (CDI) 273 CO hydrogenation catalysts 192 CO2 hydrogenation 84, 165, 194, 331, 638 coke formation 124, 305, 611, 640 collapsed bifunctional sites 95 colloid nanoparticle dispersions 162 colloidal silica particles 128 colloidal stability 49, 50 colloidal TiO2 particles, interfacial electron/hole-transfer reaction of 380 combinatory methods 40–42, 50–52 Combined electrophoretic deposition–anodization (CEPDA) approach 51–52 combined XAS-DRIFTS 367–368 combustion, of metal precursors 185 CO methanation reaction 196 commercial SiN membranes 258 CO Moiré structure 232 complementary metal–oxide–semiconductor (CMOS) 284 complex metal oxide catalysts 197 composite catalysts 46, 59 composite metal oxide catalysts 192–197 compositional homogeneity 211 comprise mesoporosity 122 computed tomography (CT) 264, 267–268, 270, 274 computer simulations 401–403 conduction band (CB) 29, 32, 190, 203, 205–210, 377, 378, 382, 384, 388, 389, 391, 392, 472, 557, 659, 673 conduction band minimum (CBM) 32, 203, 557 confocal laser scanning microscopy (CLSM) 284, 285, 287, 289, 291 conformal coating 48 Congo Red reduction 167 CoNi@NC 549
constrained density-functional theory (CDFT) 412, 416 contact process 3, 4 conventional absorption-CT 265–267 conventional direct current method 48 conventional heating 62 conventional microporous zeolites 123, 126 coordination number approximation 244–245 coordination polymer 147, 148 coordinative unsaturated metal sites (CUS) 71, 73, 153, 157 CO oxidation 14, 81, 83–85, 92, 105–107, 168, 234–238, 245, 259, 325, 354, 355, 357, 516–517 on multi-component alloy surfaces 516 on Pt(110) 234–236 CO (photo-)oxidation reaction 41, 193, 234, 236, 238, 392, 538, 559, 561, 624, 664 copper based zeolite catalyst (Cu–SSZ-13) 354 copper–zinc catalyst 7 coprecipitation technique 13 CO2 reduction 27, 45 electrochemical reduction electrochemical cell 655 GDEs 656 high Faradaic efficiency 657 improved/novel electrolytes 657, 658 in gas phase 657 low overpotential for 657 Sabatier principle 655 figures of merit 652–653 heterogeneous catalysts 653–654 photochemical reduction efficient photocatalytic reactors 662 electron-hole pairs 659 factors 660–661 innovative photoactive materials 661–662
Index
photocatalyst surface 659 photoelectro-chemical reduction efficiency of 663 electrode configurations for 664 GaP semiconductor 662–663 inorganic binary compounds 663 photoactive materials 665–666 photoanode–dark cathode 665 photocatalytic reactors 666 photocathode–dark anode 664–665 photocathode–photoanode 665 redox potentials for 650 redox processes 651–652 reservoir and EOR 650 thermodynamically stable molecule 650–651 utilization of 650 core-to-core XES 348 CoSe2 nanosheets/GCE 545 counter electrode 11, 41, 42, 51, 225, 576, 577, 584 covalent organic frameworks (COFs) 11 covalent triazine frameworks (CTF) 64, 67–68 Cr-doped rutile TiO2 560 critical coupling 300 critical process 217, 640 crystal facet engineering 21 crystal morphology 26 crystalline materials 153, 171, 269 crystalline microporous 8, 133 crystalline microporous aluminophosphates (AlPO4 ) 8, 133–134 crystalline nanoparticles 63, 186 crystalline porous catalysts 171 crystalline silicate-based catalysts 619 crystalline V2 O5 194 crystalline wall structure 128 crystalline zeolites 9 crystallization 7, 9, 22, 46, 52, 86, 124, 126, 128, 131, 132, 136, 153, 196, 197, 546, 625
crystallization process 9, 46, 52, 124, 126, 128, 131, 135, 153 Cs-exchanged silicotungstic acid 600 C6 sugars 617 cubic topology 148, 151 Cu2 O crystals 26 cuprous oxide 48 current efficiency 46–47, 652 Cu–SZZ-13 catalyst 345, 354 Cu-zeolite catalysts, in situ EPR study of 301 Cu-zeolite systems 303–305 cyclic voltammetry 440, 462, 552, 577–578, 581 cyclohexanedimethanol (CHDM) 94 cylindrical internal reflection (CIR) 327 CyreneTM 623
d data mining and machine learning in catalyst design 509 Davy lamp 3 3D continuous macroporous structure 265 dealumination 124, 136–137, 139, 140, 289, 290, 687 acid leaching 136–137 calcination process 136 mesopores, formation 137 steaming process 136 zeolite modification 136 debris, of nanotubes 46 delay time of probe pulse 378 3D electron density maps 92 deNOx reaction system 530, 609, 610, 622, 636, 680–681, 694 deNOx technology 685 density functional theory (DFT) 106, 165, 235, 307, 403, 405, 497, 658 basis sets 409 Born–Oppenheimer approximation 406 calculations 565
705
706
Index
density functional theory (DFT) (contd.) Hohenberg–Kohn theorem 406–409 Kohn–Sham equation 407, 408 density, of layer 46 deposited metal (Met) 46 deposited space 46–47 design for energy efficiency 633 desilication 124, 138–140 ammonium surfactants 139–140 high-silica zeolites 138 modified procedure 139 pore-growth moderator 139 3D graphene aerogel materials 679 dhbdc linkers 150 didactical approach 157 dielectric constant 32, 50, 639 diesel engine emissions 685 differential pumping effect 256, 258 differential pumping–type system 254, 256–259 diffraction anomalous fine structure-CT (DAFS-CT) 274 diffraction/scattering techniques 264 diffractive techniques 150 diffuse reflectance Fourier transform infrared spectroscopy (DRIFTS) 230, 316–317, 325, 329–331, 368, 370, 433, 691 cells 325, 327 of CO adsorbed on various Pd/Al2 O3 samples 323 isotopic transient kinetic analysis 331 on Pd/Al2 O3 331 of V2 O5 –WO3 –TiO2 319 diffusion coefficient 45, 169, 230, 287, 579 diffusion length 45, 124, 244, 377 dihydrolevoglucosenone 620, 623 2,5-dihydroxymethylfuran 605 diluted surfactant solution 130 3-dimensional ordered macroporous (3DOM) structures 162, 164, 165, 170, 172, 174
dimer (M2 ) yields, formation of 80 dimer methods 424 2,5-dimethylfuran (DMF) 594, 604–606 dimethyl hexahydroterephtalate (DMHT) 94 dimethyl terephthalate (DMT) 94, 95, 623 dimethylformamide (DMF) 605, 623, 656 diquaternary ammonium surfactants 131 direct current (DC) 47, 48, 82 direct inclusion methods 172 discrete nanoflowers, of bismuth 170 dispersed niobia particles 69 dispersed technical catalyst 227 dispersing and stabilizing nanodiamonds 63 dispersive forces 59 dispersive polychromators 353 dissociative adsorption 28, 168, 234 of water 28 dissociative hydrogen adsorption 234 3D macrostructures 62 2D nanolace sheets, of alumina 45 2D NiFe-based bimetallic organic framework 546 3D NiFe hydroxide thin nanosheets 542 doped photocatalysts 206–208 doped rutile TiO2 photocatalyst 559–564, 567 doubled water electrolysis (DWE) 538 Dowex–type resins 603 downstream mass spectrometer 81, 108 drive cycle testing 266 dual-target magnetron sputtering system 84 dual-templating approach 131, 161, 170 dye-sensitized heterogenous photocatalyst 384–386
Index
dynamically limited adsorption processes 234
e electrocatalysis 11, 21, 33, 39, 57, 68, 105, 110, 165, 225, 292, 440, 660, 665 electrocatalyst inks 582 electrocatalysts 167, 439–441, 541–551, 571–588 electrocatalytic oxygen reduction 167 electrocatalytic reactions 11, 39, 57 electrochemical anodization 11, 15 electrochemical configuration 46 electrochemical method, fabricating thin film cathodic electrodeposition 46–48 pulse electrodeposition 47–48 combinatory methods, electrochemical process 50–52 electrophoretic deposition 48–50 anodization 42–45 catalytic applications 44 formation of, anodic oxide layer 42 growth mechanism, anodized metal foil 43 pulse/step anodization 45–46 electrochemical processes 40, 441 electrochemical reactor (electrocatalysis) 225, 439, 651, 654, 655, 658 electrochemical reduction definition 651 electrochemical surface area (ECSA) 545, 550, 574, 577–581 electrochemical water splitting 39, 533, 534, 551, 552 fundamentals of 533–535 hydrogen evolution reaction electrocatalysts 547–549 liquid electrolyte water electrolysis 535–537 oxygen evolution reaction electrocatalysts 541–545
polymer electrolyte membrane water electrolysis 539–540 solid oxide electrolyte water electrolysis 540 electrodeposition 41, 46–49, 52, 73, 548 electro-oxidative polymerization 40 electron cloud 32 electron configuration 205, 206, 303 electron energy loss spectroscopy (EELS) 71, 217, 255, 256, 259, 260, 607 electronic interactions 13, 333 electronic transfer 168 electronic wavefunctions 403, 406, 407, 409, 410 electron magnetic resonance (EMR) 295 electron nuclear double resonance (ENDOR) 299, 307 electron paramagnetic resonance (EPR) 193, 219, 295–307, 658, 689 -active species 302 magnetic field intensity 298 principles of 296–298 spectra of Cu-CHA samples 305 electron scattering 255, 256, 346 electron spin echo envelop modulation (ESEEM) 299, 307 electron spin resonance (ESR) 295 electron transfer 60, 157, 217, 303, 306, 348, 380, 382, 383, 385, 386, 388, 389, 391, 392, 402, 412, 416, 478, 529, 541, 546, 579–581, 651, 653, 676, 677 electron transfer number 579–581 electrophoretic deposition 40, 41, 48–52 electroplating 39, 41 electroreduction 33, 46, 68, 504, 655–659 electroreduction, of metal cations 46 electrostatic repulsion 50 electroxidation 33, 68
707
708
Index
elementary reactions and materials selection 498–502 Eley–Rideal (ER) mechanism 504 enantioselective hydrogenation 191 end plates 583 energetic efficiency 653 energetic photons (photocatalysis) 225 energy consumption 535, 536, 593 energy functionals, development of 410–412 energy levels Zeeman splitting 297 energy-dispersive von Hamos type spectrometer 353 enhanced oil recovery (EOR) 650 epifluorescence microscope 281, 282 equilibrium adsorption 57 ethanol/dimethyl sulfoxide mixture 601 ethanol/tetrahydrofuran mixture 601 ethene photodecomposition 170 5-ethoxymethylfurfural (EMF) 594–602, 609, 611, 624 ethyl levulinate (EL) 594, 602–604 European Synchrotron Radiation Facility (ESRF) 274 exchange-correlation density functionals 410 exchange-correlation functional 408, 410, 411, 414 excited electrons (e– ) 33, 203, 384, 472 exo-and endo-thermicity 220 exothermic metal precursors 186 explicit kinetic simulations based on DFT calculations 507 exsolution method 172 extended X-ray absorption fine structure (EXAFS) spectroscopy 109, 274, 345–347 extended X-ray absorption fine structure-CT (EXAFS-CT) 109, 274, 345 Extremely Brilliant Source (EBS) 274
f fabrication, of model catalysts fabrication procedures 162
83
face-centered cubic (fcc) 22, 88, 413 crystal structure 27 facet engineering, on catalysts anisotropic adsorption 27–28 anisotropic properties, of crystal facets 27–31 effects of 32–34 optical properties 32–33 activity & selectivity 33–34 mechanisms 22–27 solid-phase methods 22 solution-phase methods 22 vapour-phase methods 22 surface electric field 29–31 surface electronic structure 28–29 faceted nanocrystals 22 faceted semiconductor 29, 33 faradaic efficiency 567, 652–653 Faraday constant 47, 579, 651, 653 Faraday law for electrolysis 46 fcc-structured metal crystal 22–23 Fe-family-based OER catalysts 543 Fe3 O4 clusters 62 α-Fe2 O3 nanotubes 11, 44 SiO2 601 SiO2 –HPW 601 SiO2 –SH–Im–SO4 H 601 SiO2 –SO3 H 601 FeOx surfaces 104 Fermi level 28, 81, 351, 456, 457, 465, 546, 676, 677, 690 Fick’s law 636 film thickness 46–47 filter-press elecotrochemical cell 657 first generation biofuels 595, 596 Fischer–Tropsch catalyst 69, 271 Fischer–Tropsch synthesis (FTS) 7, 8, 13, 69, 194, 196, 641 flame aerosol reactors 186 flame aerosol synthesis and reactors 185–189 flame-assisted spray pyrolysis (FASP) 185, 186, 189 flame spray pyrolysis 186 flame synthesis
Index
flame aerosol synthesis and reactors 185–189 multielemental oxide-based catalysts complex metal oxide catalysts 197 composite metal oxide catalysts 192–197 solid solution metal oxide catalysts 192 from natural aerosols formation 183–184 simple metal oxide-based catalysts 189–192 flash photolysis time-resolved microwave conductivity experimental layout 390 flat thin films 39, 40 flow-field plates 583 flow reactors 352, 353, 664 fluidized catalytic cracking (FCC) process 8, 27, 171, 321, 322, 413 fluorescence basics of 280–282 CLSM 284, 285 contrast mechanism 283 strategies, resolve catalytic processes 283 super-resolution fluorescence microscopy 285–286 WFM 284, 285 fluorescence-detected XAS 343 fluorescence microscopy 279–288, 290–292 fluorescence spectromicroscopy 220 fluorescent oligomers 289 fluorogenic aminophenyl fluorescein 290 fluorogenic molecules 283, 288 focused ion beam (FIB) milling 45, 253, 260 forbidden gaps 81 formation of, carboxylate dicopper 148 formic acid 33, 111, 239–242, 246, 499, 605, 636, 637, 655, 657, 665 fossil fuels 528, 533, 593, 596, 611, 649, 685
foundry resin 622 Fourier transform infrared (FTIR) spectroscopy 229, 230, 327, 658 fourth generation biofuels 595 fragmentation 83, 86, 90, 162, 628 free electron laser (FEL) 274, 353 free energy calculations metadynamics 430–432 reaction coordinate, defining 425 sampling points, selection of 428 thermodynamic integration constrained MD simulations 425 ergodicity assumption 425 Lagrange formalism 426 Lagrange multipliers 427, 428 schematic illustration 425 umbrella sampling 428–430 free energy profile, for an elemental reaction 424 free-standing particulate 678–679 Friedel–Crafts alkylation 639 fructose 596, 599–601, 603–605, 609, 617, 620, 622, 624–625, 627–628 fuel cell test station 583 fuel cell vehicles (FCVs) 572, 574–575, 581, 636 fuel crossover 584 furandicarboxylic acid (FDCA) 539, 622–626 furfural 5, 196, 596, 603–604, 606–607, 611, 622 furfuryl alcohol (FOL) 289, 603–607, 622 oligomerization 290
g galactose 617 galvanic replacement, Au 167 galvanic-replacement method 109, 112 galvanostatic condition 48 galvanostatic method 46 γ-valerolactone (GLV) 594, 607, 622 GaN-ZnO 558–559 gas diffusion electrode (GDE) 582–583, 656–657
709
710
Index
gas diffusion layer (GDL) 266, 539, 574 gas physisorption 150 gas-to-particle formation 183–184, 186, 189 gaseous nitrous acid (HONO) 693 gaskets 539, 583 Gauss–Lorentzian deconvolution 235 gelled FeCoW oxyhydroxides 543 Gemini-type surfactant 132–133 generalized-gradient approximation (GGA) 411–412 Gibbs energy 558 Gibbs–Wulff theorem 22 glucose 60, 68, 170, 196, 596, 600–601, 603, 605, 609, 617, 620 gold-thiolate clusters 85–86 gradual transformation of nanodiamond 66 grafting metallocene complexes 104 graphdiyne 62–64, 75 graphene 62, 73 derivatives 62–64, 75 liquid cell 258 nanoplatelets 584 graphene-like carbon foam 584–585 graphene oxide (GO) catalyst on 59–60 pre-impregnated metal precursor 62 structure of 59 graphite intercalation compounds (GICs) 61 graphite-like hexagonally-packed layered structure 549 graphitic carbon materials 71 graphitic carbon nitride (g-C3 N4 ) 11, 64, 67, 662, 676 graphitic/amorphous carbon 63 graphyne 62–64, 75 green chemistry principles 633, 634, 645 greenhouse gases 529, 593, 623, 637 Grove’s fuel cell 573 g-value 296, 298
h Haber–Bosch process 4, 6 half-cell measurements 576, 583–584 half-reactions 41, 533–534, 572 half-wave potential 579, 586–587 hard and soft dual-templating 170 hard templates carbon nanotubes and nanofibres 127–128 ordered mesoporous carbons 128–130 space-confined method 125–127 hard-templating methods 15, 124–125, 140 hard X-rays 272, 339, 348 penetration depth of 342 heat of adsorption 229–230 Hellmann–Feynman theorem 409–410, 483 hemicellulose 595, 606, 617–618, 622 n-heptane 33 heptazine 64, 67 heteroatoms 14, 59, 67, 71, 74–75, 545, 547, 550–551, 600, 620 heterogeneity, of catalyst materials 219–220 heterogeneous catalysis 21, 217, 401, 405, 597 biomass conversion/biomass upgrading 528 carbon capture and storage technology 529 carbon dioxide (CO2 ) emission 529 carbon emissions 527 chemical and energy industries 527 density functional theory in 405–416 DFT applications 412–414 forces on the ions, calculation of 409–410 hydrogen 527 MD technique 419 nitrogen oxides (NOx ) 529 quantum mechanical techniques in 403–404
Index
SCR technology 529 wastewater treatment processes 529 heterogeneous catalysts 3–5, 83, 103–114, 147–158, 168, 170, 311–334, 619 CO2 conversion 653–654 heterogeneous catalytic conversions in high-pressure reaction condition 633–636 heterogeneous photocatalysis 484, 567, 681 advantages of 671–672 heterogeneous photocatalysts 380–384, 391–392, 530, 557–558 heterogeneous reaction 15, 27, 70, 635, 640 heterogeneous (solid) photocatalyst 203 heterogenized Re-complex catalyst with TiO2 nanoparticles 384 heterojunction interface 168 hexaethynylbenzene 63 1,6-hexanediol (1,6-HDO) 622, 623 1-hexanethiol 167 1,2,6-hexanetriol (1,2,6-HTO) 622 hexapod-shaped building blocks 162 hexoses 596, 617, 622, 627–628 hierarchical and anisotropic nanostructured catalysts Janus nanostructures 165, 167–169 hierarchical porous catalysts 169–171 nanostructured assemblies 162 porous/anisotropic substrates, functionalisation of 170–174 shape anisotropy 162, 165 top-down vs. bottom-up approaches 162, 164 hierarchical macroporous–mesoporous SBA-15 network 171, 173 hierarchical macroporous–mesoporous silica 161 hierarchical mesoporous–macroporous pore structure 170–171 hierarchical porous catalysts 169–170
hierarchical structure zeolites 122 hierarchical titania 170 hierarchical zeolites, synthesis strategies 122 advantages 123 changed product selectivity 124 decreased coke formation 124 increased intracrystalline diffusion 123 modern strategies for 124 dealumination 136–137 desilication 138–140 hard templates 124–130 soft templates 130–136 reduced steric limitation 123 high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) 14, 92, 104, 110, 112, 114, 162–163, 175, 563 high-energy bottom-up ball-milling synthesis 109 high-energy resolution fluorescence detection 349–350 high-energy-resolution off-resonant spectroscopy (HEROS) 350–351 high energy resolution (resonant) X-ray emission spectroscopy 347–351 high-index facets 27 high-performance computing (HPC) clusters 401, 493 high-pressure catalytic reactions 5–8 high-pressure hydrotreating processes 636 high-pressure methanol synthesis process 633 high-pressure reaction condition advantages 633 catalytic processes 634 chemical equilibrium 634 CO2 conversion methanol synthesis 637 phase separation 638
711
712
Index
high-pressure reaction condition (contd.) green chemistry principles 634 industrial chemical 634 mass transfer and kinetics molecular diffusion 635–636 multi-phase reaction 636 microchannel reactor 641–643 phase separation 634–635 process efficiency and economy 636 supercritical fluid properties for 639–641 supercritical state 635 water-gas shift reaction 643–645 high-pressure scanning tunneling microscopy 235 high-pressure steel reactors 7 high-pressure studies, of catalysts 226–229 high-throughput (HT) computational catalyst design activity volcano curve 507 adsorbate-adsorbate interactions for CO methanation 516–518 alcohols synthesis 509–511 BEP relation and the activation barrier 503–507 CO oxidation on multi-component alloy surfaces 516 data mining and machine learning 509 DFT calculations 498 elementary reactions and materials selection 498–502 explicit kinetic simulations based on DFT calculations 507–508 hydrogen evolution reactions (HER) 511–516 RhAu alloy nanoparticles for NO decomposition by machine learning 518–520
scaling relation and reaction energy 502–503 highest occupied molecular orbital (HOMO) 80, 231, 280, 388, 485–486, 490–491 highly active NiFe oxides/hydroxides 542 hinder energy-dispersive X-ray spectroscopy 255 2H-MoS2 polytype 549 Hohenberg–Kohn theorem 406–409, 472 hollandite-type manganese oxide (HMO) 106–107, 113–114 HOMO 5σ CO molecular orbitals 231 homogeneous catalysts 39, 68, 75, 89, 157–158, 302, 619, 624 homogeneous or heterogeneous catalysts 157 homogeneous Lewis–Brønsted acid mixture 605 homogeneous (molecular) photocatalyst 203 Houdry process 8 HPW/k-10 clay 600 HPW/MCM-41 600 H-type cells 655 Hubbard U corrections 412 hybrid water electrolysis (HWE) 538 hydrochloric acid 27, 127, 136 hydrodeoxygenation (HDO) 596, 604, 606–607, 623 HydrofactionTM 640 hydrogen 527 bonds 59, 81, 595, 617 -filled zeppelin airship 571 mass production 533 hydrogenation 3, 5, 13, 57, 60, 63, 67–68, 84, 89, 94–95, 105, 121, 157, 165, 167, 191–192, 194, 272, 329, 331–333, 369, 505, 596, 604–609, 611, 619, 622–623, 625–626, 635, 637–638 of n-olefins 121
Index
hydrogen evolution reaction (HER) 39, 63, 511–516, 534–535, 538–542, 545, 547–552, 651 electrocatalysts 548 metal-free electrocatalysts 551 noble metal–free alloys 548 non-noble metals 547–549 non-precious metal composites 549–551 hydrogen oxidation reaction (HOR) 572–573, 577 hydrophilic mesopores 171 hydrophobic macropores 171 hydrophobic metal precursors 68 hydro/solvothermal synthesis 11 hydrothermal carbon material 69 hydrothermal conditions 57, 68, 128, 619 hydrothermal method 22, 26, 62 hydrothermal modifications 124 hydrothermal procedure 61 hydrothermal treatment 52, 60, 68, 125, 130, 135 5-hydroxymethyl furfural (HMF) 196, 539, 596, 599–600, 603–606, 611, 620, 622–624, 626–628 5-(hydroxymethyl)furoic acid (HMFA) 626 2-hydroxypropionic acid 622 γ-hydroxyvaleric acid 608 hyperfine sublevel correlation (HYSCORE) 299
i Imperial Chemical Industries (ICI) 7 impregnation method 111, 212 increased intracrystalline diffusion 123 industrial catalytic process 636 industrial catalytic reactions 5 industrial Ruhrchemie catalyst 266 infrared (IR) spectroscopy 311 attenuated total reflection mode 314–318
challenges 333 frequency ranges 312 of organic molecules, vibration types 312 theory on 311–313 transmission mode 314–315 infrared microspectroscopy 317–318, 334 inherent microporosity 122 initial oxide layer 43 inorganic cluster 152 inorganic monolithic aerogels 265 inorganic radicals and radical ions 303 in situ and operando methodologies 220, 363 in situ EPR 302–303 components 299 of Cu-zeolite catalysts 301–302 Cu-zeolite systems 303–305 experimental methods and setup for 298 microwave frequencies 299, 300 radicals and radical ions 305–306 sample cavity 300 setup issues 301 signal channel 300 in situ flame deposition 192 in-situ measurement 301, 304 in situ transmission electron microscopy observation applications 259–261 differential pumping–type system 256–257 in gas and liquid phases window-type system 254–256 other systems 257–258 π–π interactions 59, 71 intercrystalline mesopores 122 interfacial spillover 168 International Energy Agency (IEA) statistics 649 intracrystalline mesopores 122–123, 135–136 intracrystalline transport 123
713
714
Index
intrinsic Brownian (random) motion 169 intrinsic reaction mechanisms 106–107 invert enantioselectivity 158 ion-exchanged zeolites 14, 302 IR microspectroscopy 317–318, 334 IRMOF-3 152, 155, 157 iron carbonyl reaction 6 iron oxide nanocrystallites (Pt1 /FeOx ) 109 iron porphyrin 67, 72 irreducible thorium oxide 7 irregular nanoporous morphology 45 IR spectroscopy 219, 327, 367 background spectrum 318–320 NH3 320 under operando conditions 325–328 using probe molecules 320–325 isomerization, of hydrocarbons 8–9 isoreticular linkers 157 isoreticular MOFs 151–152
j Jablonski energy diagram 281, 377, 378 Jacob’s ladder of functional development 410, 411, 414 Janus arrangement 168 Janus AuPt systems 168 Janus nanomaterials 165, 168 Janus nanostructures 165–169
k K-edge XANES 343–345, 354 kinetically controlled region 579 kinetically restricted adsorbate structures 229–231 kinetic control, of zeolite seeds 131 kinetic gas theory 226 kinetic Monte Carlo simulations 245, 402 Knøvenagel condensation 157 Knudsen diffusion 123 Kohn–Sham approach 406–409
equation 407, 408, 471 wavefunctions 407–408, 413 Koutecky–Levich (KL) theory 579 Kubelka–Munk (K–M) function 317, 318 Kuhlmann reaction 4
l lactic acid 622, 627–629 lactide 627–630 Lagrange formalism 426 Lagrange multipliers 426–428 Lambert–Beer law 317 laminated graphene layers 258 Langmuir–Hinshelwood (LH) mechanism 504–505, 692 lanthanum strontium manganite (LSM) 540 La0.6 Sr0.4 MnO3 (LSMO) 162, 165, 166, 172, 174 lateral resolution 282, 285 lattice constant 23 layered (oxy)hydroxides 542 Le Chatelier’s principle 90, 634 Lenz’s law 32 Lerf–Klinowski model 59 Levich equation 579 levulinic acid 196–197, 602, 622 Lewis acid catalysis 148, 153, 157 light absorption, of semiconductors 32–33 lignin 120, 595, 617, 618 lignocellulose 9, 594, 610–611, 617 conversion 57 lignocellulosic biomass biological and chemical pathways 594 composition and resources 595–596 cost of 593 DMF 604–606 ethoxymethylfurfural (EMF) 597–602 ethyl levulinate (EL) 602–604 γ-valerolactone (GVL) 607–610 2-methylfuran (MF) 606–607
Index
linear shrinkage 192 linear sweep voltammetry (LSV) 578–581, 587 liquid electrolyte water electrolysis 537 DWE 538 HWE 538–539 OWE 537–538 TWE 539 liquid metal precursors 185 liquid phase transformations 170 liquid precursor formulations 187 liquid water quantification method 266 lithium-ion batteries 21 load-cycling protocol 581 local density approximation (LDA) 410, 411, 474 localization-based super-resolution microscopy 285, 289 localized Gaussian-type basis 409 localized SPR (LSPR) 15, 29, 169, 675 low energy electron microscopy (LEEM) 217 low-energy electron diffraction (LEED) 217–218, 233 low-index facets 22, 24, 27 lowest unoccupied molecular orbital (LUMO) 80, 280–281, 388, 485–486, 490
m mackinawite-type nanosheet arrays of FeS 545 macrocycles 72 macrocyclic molecules 71 macroporosity 122, 125, 140, 161 macroporous 124, 161, 162, 170–171, 173, 174, 265 magnetic resonance imaging (MRI) 220, 295, 307 mass activity 439, 579–580, 584–586 mass transport limited current density 579 mass transport losses 584
mass transport-limited region 579 mass-select nanoclusters 81 material gap 218–219, 324 Materials Project 509 maximizing atom economy 633 Maxwell velocity distribution 226–227 mean jump distance 230 Meerwein–Ponndorf–Verley (MPV) reaction 609 membrane-electrode assembly (MEA) 271, 539, 582–583 Mercury–Cadmium–Telluride (MCT) detectors 229 meso/microporous MFI-Sn/Al zeolite 601 mesoporous 161 aluminosilicates 600 carbon 130 silica supported Keggin heteropolyacid catalysts 602 mesoporous silica ([Si]MCM-41) materials 110 mesostructured architecture 162 mesostructured frameworks 162 meta-GGA functional 411 metadynamics 420, 424, 430–432 metal aerosol particles 183 metal-based catalysts 79, 608 metal-carbon hybrid materials 547 metal carbonyl clusters 85, 88, 90 metal catalysts, miniaturization 12–14 metal cations 10–12, 46, 61, 109, 120, 205–207, 210, 212, 295, 620, 674 metal cation-doping 207 metal chalcogenides 545, 549 metal clusters 3–16, 21, 58, 61, 71–72, 79–96, 105, 108, 244 metal clusters-based catalysts catalysts, deposition of clusters 81–85 chemically synthesised metal clusters 85–88 metal clusters/single metal atom deposition
715
716
Index
metal clusters/single metal atom deposition (contd.) dopant elements, on carbon 71 metal clusters, immobilization of 71–72 metal crystal catalysts 22 metal deposition 674–677 metal-exchanged zeolite 686–687 metal-free carbon-based electrocatalysts 551 metal-free catalyst 57, 59, 63 metal-free electrocatalysts 551 metal-ion doping 674–675 metal-organic chemical vapour deposition 22 metal–organic frameworks (MOFs) 10, 73, 546–547, 655 -derived carbon (nano)materials 547 -derived functional electrocatalysts 547 -derived hybrid Co3 O4 -carbon porous nanowire array 547 metal halide precursors 185 metallic cations 46, 48 metallic foil 42–43, 51 oxidation of 42 metallosilicates 9 metal oxides 61, 204, 542, 543, 545 catalysts 685 on graphene 61 interface 43 nanostructures 46 supports 584 metal (oxy)nitrides 204–205, 208 metal (oxy)sulfides 204–205, 208 metal precursor 12, 57–62, 67–68, 72, 74–75, 108–109, 185–186 metal precursor–O2 73 metal-to-metal charge transfer (MMCT) 14 methanation of CO2 329–331 methanol 633, 637 synthesis 5–7, 15, 259, 633, 637–639, 641
to-hydrocarbon conversions 8 methanol-to-olefins (MTO) reaction 128, 268 5-methylfurfural 605 2-methylfuran (MF) 594, 596, 597, 605–607, 611 methyl tert-butyl ether (MTBE) production 635 methylene blue decomposition 170 meticulous information 218 MFI zeolite membrane 644 micelle templating 170 micro-and/or mesoporosity affords 161 micro-channel plate (MCP) 387 micro-reactors 352, 642–643 microcanonical (NVE) ensemble 422 microelectromechanical system (MEMS) 254–256, 258–260 microkinetic modelling 497, 499, 502, 503, 506, 507, 511, 516, 520 micropollutants 14, 16, 530, 671–682 microporous 161 aluminophosphates zeolite analogs 131 channels 123 crystalline AlPO structure 134 microwave-assisted heating 61 mixed-metal clusters 71, 86, 94 modified hierarchical zeolites 126 modular infrared spectrometers 334 modulation excitation (ME) technique 220, 325, 365–369, 371 molar enthalpy, of metal sublimation 22 molar weight, of materials 46 molecular active sites, tools post-synthetic methodologies 155–156 PSE 156 PSM 155 pre-synthetic methodologies 153–155 molecular adsorption, of water 28 molecular diffusion 217, 635–636, 645
Index
molecular dynamics (MD) simulations 420 challenge 419 conservation of total energy 422–424 description 419 examples of 433–437 free energy calculations 425–432 in canonical ensemble 422 Neural Network 432–433 in NVT ensemble 422 TiO2 /H2 O model system 433, 435 velocity Verlet algorithm 421–422 Verlet algorithm 421 water-gas shift reaction 436, 437 molecular heterogeneous catalysis 58 molecular heterogeneous catalysts isoreticular MOFs 151–152 MOF, catalysis catalytic reactions, selectivity of 158 monodispersion, of active sites 158 standard heterogeneous catalysts 158 SBU 151–152 tools, molecular active sites post-synthetic methodologies 155–156 pre-synthetic methodologies 153–155 molecular orbitals (MOs) 29, 80, 280, 388, 485 molecular sieves 120, 130, 134, 619–620, 627–628 molybdenum disulfide 11 monodispersion, of active sites 158 monolayer (ML) 83, 111, 167, 228, 461–468 monometallic Ni catalysts 167 monometallic systems 168 monosaccharides 601, 604, 617 MoO3 truncated rhombohedral 44 multidentate carboxylate linkers 148 multielemental oxide-based catalysts
complex metal oxide catalysts 197 composite metal oxide catalysts 192–197 solid solution metal oxide catalysts 192 multifunctional catalysts 611 multi-phase reaction 636 multivariate spectral analysis 220 in catalysis 371–372 multiwall carbon nanotubes (MWNTs) 127, 551
n Nafion 271, 306, 539, 574, 577, 582, 655, 657 bound carbon layer 48 nano-carbon 547, 554 nanocrystalline metals 46 nanodiamond core/graphitic shells (ND@G) 63 nanodiamonds 58, 63, 66, 74 nanodumbbells 168 Nanometer Accuracy by Stochastic Chemical reActions (NASCA) 286 -type reactivity mapping 290 nanomotors 169 nano-onions 58, 63 nanoparticles (NPs) 81, 103 propulsion 169 shape anisotropy 165 nanoporous materials 161 nanorods 10–11, 27, 40, 42, 44–45, 48, 62, 112, 165, 191, 545, 546 nanoscale Mo2 C 550 nano-sized zeolite crystals 122–123, 128 nanostructured metal oxide 42 nanostructured MoS2 549–550 nanotomography 266 nanotubular substrate 48 National Institute of Standard and Technology (NIST) 275 natural emissions, of aerosols 183 Nb2 O5 nanorods 11, 44
717
718
Index
N-doped activated carbon 551 N-doped hexagonal carbon 551 N-doped TiO2 207, 560, 675 near edge X-ray absorption fine structure (NEXAFS) 71, 343 NEDO cell holder 582–583 Nernst equation 577 Neural Network (NN) 420, 432 N/F-codoped rutile TiO2 (TiO2 :N,F) 563 NH3 adsorption on a metal oxide 320 on V2 O5 –WO3 –TiO2 318 NH3 assisted SCR reaction 303 NH3 -SCR process 689, 691, 692 Ni-and Mo-based HER catalysts 548 Ni-based composites 540 niche catalytic application 169 NiFe hydroxides 542, 545 based oxygen electrode 542 NiFe mixed oxides 543 NiMo ternary alloys 548 Ni3 N nanosheets 546 Ni2 P-based catalysts 607 Ni3 S2 nanorods/Ni foam 545 4-nitrobenzaldehyde to 4-nitrobenzyl alcohol 14 nitrobenzene 13, 147, 272 3-nitrobenzene 13 nitrogen (N2 ) coordinated SACs 72 fixation 5 physisorption 148 nitrogen-doped carbon decorated with iron atoms (Fe–N–C) 586 nitrogen-doped TiO2 560 nitrogen oxides (NOx ) 302–303, 529 catalysts 272 emissions NH3 -SCR of 687 stationary and mobile sources 685 zeolites 694 inhalation 529 nitrophenol 167, 169 reduction, by NaBH4 167, 169
N-methylpyrrolidone (NMP) 623 NO, selective catalytic reduction of 197, 318, 328–329 noble metal nanoparticles 60–62, 73, 676 noble metal-doped titania 168 noble metal–free alloying 548 NOMAD 509 non-conventional liquid water electrolysis 538 non-empirical functionals 410 non-equilibrium process 237 non-noble metals and noble metal-free alloys 547–549 non-noble transition metals 547 non-polar solvent 61 non-precious metal composites 549–551 non-solubilizing solvent 94 non-toxic green solvents 619–620 Nosé–Hoover thermostat 422–423, 435–436 nuclear magnetic resonance (NMR) 123, 136, 153, 220, 289, 295, 307, 328, 358, 691 nudged elastic band (NEB) method 410, 413, 424, 504 numerical aperture (NA) 281–282, 285
o octadecylamine 167 octahedral geometry 148 octahedrons 24–25, 29, 151 O2 dissociative adsorption 168 OH desorption 165 ohmic losses 536–537, 540–541, 573–575, 584 one-dimensional (1D) photocatalysts 11 one-pot synthesis 85, 603, 611, 620, 692 one-step photoexcitation 209 onion-line carbon (OLC) 63 onset potential 539, 546, 579
Index
on-the-fly XAS 353 open circuit voltage 493, 584 operando IR spectroscopies, case study of 328–333 operando spectroscopy 219, 328, 352, 370 ordered mesoporous carbons 128–130 organic electrolyte 44, 51 organic/inorganic capping agents 25 organic linkers 10, 73, 148, 153, 155–157 organogold precursor 110 organometallic approach 61, 71–72 organometallic ligands 85 organometallic precursors 71 organosilane surfactants 131, 132 organosilanes 130 orifice 256–258 orthogonal catalysts 171 orthosilicic acid 9 O-, S-, N-, P-, and B-(dual) doped graphene 551 Ostwald ripening 113, 575, 581 overall water electrolysis (OWE) 537–538 overarching synthesis methodology 86 overpotentials 203–204, 356, 534–535, 538, 542–547, 551, 584, 652–653, 655, 657, 665 over-reactive anodization 46 oxidation of pyrite (100) surface 413 of cyclohexane to cyclohexanone 13–14 oxidation rate 43, 355 oxidation–reduction (redox) reactions 40 oxidative coupling of methane reaction (OCM) 268, 303 oxide etching agent 42 oxide semiconductor 45, 259, 557 oxy-fuel combustion 185 oxygen adsorption 172, 234, 545
oxygen evolution reaction (OER) electrocatalysts 39, 67, 533, 544 borides 546 carbides 546 metal chalcogenides 545 metal–organic frameworks (MOFs) 546–547 metal oxides 542–545 metal pnictides 546 oxygen reduction reaction (ORR) 14, 67, 72, 84, 105, 107, 507, 572, 576 oxynitrides (oxide-nitrides) 557, 559, 562, 564, 567
p paddle-wheel geometry 148 pair distribution function (PDF) technique 269 pair distribution function (PDF)-CT 269–270 palladium–titanium oxide catalyst (Pd1 /TiO2 ) 109 parallel macropore channels 170 partial oxidation, of o-xylene 194 particle detachment 575 particle–particle or particle–scaffold interactions 162 Pauli exclusion principles 29 Pb2 Ti2 O5.4 F1.2 565, 566 Pdx Auy /C bimetallic catalysts 605 Pd-alloy membrane 644 Pd/Al2 O3 catalysts 322 Pd nanoparticles 74, 161, 259, 333 penetration depth of hard X-rays 342 1,5-pentanediol 622 pentoses 596, 617, 622, 627 Perdew–Burke-Ernzerhof (PBE) functional 410 perfluorosulfonic acid (PFSA) polymers 539 periodic macropore arrays 170 periodic nanostructures 45 periodic stimulation 365 permselectivity 643–645
719
720
Index
perovskite based Ba0.5 Sr0.5 Co0.8 Fe0.2 O3-𝛿 (BSCF) oxygen evolution catalyst 356 perovskite lattice 172–173, 197 perovskite oxides 542 persistent organic pollutants (POPs) 529–530 petroleum hydrotreating 641, 643 phase sensitive detection (PSD) 334, 357, 365–368 sensitivity enhancement by 366 phenanthroline or 2,2-bipyridine 72, 676 phosphines (PR3 ) 85 photo-Fenton process 74 photo/electrochemical CO2 reduction 39 photoanode 45, 209, 567, 664–666 photoanode–dark cathode 664–665 photocatalysis 11, 21, 39, 45, 67, 74, 165, 225, 237, 292, 306, 377–393, 471–472, 484, 530, 567, 658, 660, 663, 665, 671–672, 679–681 photocatalyst powder 212 photocatalyst surface 203–204, 659 photocatalytic abatement of micropollutants, water and wastewater heterogeneous catalysis 672 non-exhaustive 672 oxidative degradation 673 photocatalysts components optimization carbon materials combination 677 metal deposition 676–677 semiconductor doping 674–676 semiconductors combination 677–678 surface sensitization 676 photocatalysts configuration optimization 3D graphene aerogel materials 679–680
free-standing particulate 678–679 immobilization techniques 679 thin film nanocomposite materials 679 reaction system optimization reaction conditions 680 solar reactors 680–681 photocatalytic H2 reduction 392 photocatalytic mechanism mesoporous anatase 382, 384 mesoporous rutile with intelligent ink 383, 384 of Re-based photocatalyst immobilized on TiO2 386 photocatalytic reaction 11, 27, 51, 205, 209, 377, 381–383, 386, 561, 660, 673, 676, 680 of mesoporous anatase 381 photocatalytic water splitting 306, 484, 558, 559 photocathode–dark anode 664–665 photocathode–photoanode 664–666 photocharge transport 45 photocharge, for surface reaction 45 photocorrosion 205, 209, 665 photocorrosive material 205, 209 photodegradation, of organic pollutants 27 photodeposition method 212 photoelectrocatalytic reactions 11 photoelectrochemical catalysis 33 photoelectrode systems 209 photoelectrons, diffusion of 45 photoelectrons, fraction of 45 photoemission electron microscopy (PEEM) 217 photoexcitation 45, 209, 385, 387, 390, 412, 559, 560 photogenerated electron-hole charge generation 168 photogenerated electron-hole pairs 165, 663, 674
Index
photogenerated electrons 29, 203, 207, 212, 383, 564, 659, 663, 673–674, 677 photoholes 45, 94 photon based techniques 229, 246 phthalic anhydride 194 pillared Zn porphyrins 157 plane-wave basis set 409, 489 plant-derived biomass 596 plasmon frequency 32 platinum catalyst 4, 588 platinum dissolution 575, 581 platinum electrocatalyst 572, 574–576, 585 plug flow reactor 185, 352 Pluronic P123 triblock copolymer 10 Poisson distribution 227 polarization modulation-infrared reflection-absorption spectroscopy (PM-IRRAS) 324 poly(triazine imide) 67 poly(vinylpyrrolidone) (PVP) 13, 25, 26, 169 polybenzimidazole (PBI) 585 polyether sulfone (PES) 424, 432, 536 polyethylene 633–634 polyethylene oxide templated 170 polyethylene terephthalate (PET) 622–624 polyethyleneimine 135 polylactic acid (PLA) 174, 622, 627, 629 polymer electrolyte fuel cells (PEFCs) cyclic voltammetry 577–578 durability measurement 581 electrochemical surface area (ECSA) 574 electron transfer number 580–581 future perspectives 587–588 Grove’s fuel cell 573 half-reactions 572 hydrogen oxidation reaction 572 linear sweep voltammetry (LSV) 578–580 MEA measurements 583–584
membrane-electrode assembly (MEA) 582–583 ohmic losses 573 operation principle 571 oxygen reduction reaction (ORR) 572 platinum electrocatalyst 574–576 research 584–586 solid polymer electrolyte 574 sulfuric acid 573 triple phase boundary 573 unoptimized electrode design 573 voltammetry 576–577 polymer electrolyte membrane (PEM) fuel cells 12, 39, 192, 571–588 polymer electrolyte membrane water electrolyser (PEMWE) 536, 539–540 polymer precursors 619, 622 polymeric carbohydrates 595 polymerized (polymerizable) complex method 211 polyphenylene sulfide (PPS) 536 popping 62 pore dimensions 9, 122, 161 porosity dispersion 126, 140 porous/anisotropic substrates, functionalisation of 17–0174 porous carbon support formation 73 porous catalysts 8–11, 169–171, 283, 287 porous materials 10, 83, 95, 148, 152, 161, 171–172, 287, 684 post-synthetic exchange (PSE) 154–157 post-synthetic modification (PSM) 154–157 potential energy, of electrons 28 potentiostat 40, 42, 46, 48, 467 potentiostatic condition 48 potentiostatic method 46 pre-edges 344 pre-impregnated metal precursors 60, 62 pre-synthesized nanoparticles 61
721
722
Index
precipitation method 212 pressure gap 219, 225, 231, 246, 324, 339 pressure gaps, in heterogeneous catalysis adsorption, on solid surfaces kinetically restricted adsorbate structures 229–231 supported nanoparticle catalysts reactions 244–245 thermodynamically driven reactions 234, 237 high-pressure studies, of catalysts 226–229 pristine carbon supports 72 propane dehydrogenation 13, 63 proton-exchange membrane fuel cells (PEMFCs) 48, 265, 665, 666 protonic SOEWE 541 pseudo-ternary Ni0.2 Co0.3 Ce0.5 Ox 543 Pt catalyst 72, 172, 237, 271, 272 Pt loaded ceria-based catalysts 354 Pt(acac) method 575–576 Pt/CeO2 244, 349, 351, 355, 357, 370 ptychographic X-ray CT 273, 274 pulse current (PC) 47–48, 257 pulse EPR 307 pulse reverse current (PRC) 47 pulse/step anodization 45–46 pump–probe spectroscopy 378
q quantitative correlation 106, 354–355 quantum mechanics 402–403, 406, 483, 488, 497 quasi-in situ measurement 301 quenching process of triethanolamine (TEOA) 385 quick XAS 353
r radiofrequency (RF) potentials 295 random walk motion 45 Raney catalyst 5
82,
rate-determining steps (RDS) 123, 243, 355–356, 383, 502, 516, 625, 692–694 rational catalyst design 263, 373, 509–520 Rayleigh criterion 282 reactive ion 42, 162 reactive oxygen adatoms 165 reactive oxygen species (ROS) 67, 539, 673 red-hot metals 3 redox processes 368, 472, 577, 651–652, 674 redox reactions 11, 40–41, 52, 291, 306, 472, 576, 651, 660, 663, 676 reduced graphene oxide (RGO) composite 51, 60–62, 549 reduced Ni/Al2 O3 catalyst absorbance spectrum of 312–313 reduced steric limitation 123 reduction-oxidation 651, 663 reference electrode 40–41, 457, 576–577 reference topology 152–154 refractive index 282, 291, 315 regioselectivity 158 residence time 186, 189, 196, 197, 199, 230, 244, 325, 369, 640, 643 resonant inelastic X-ray scattering (RIXS) 349 resonant oscillation, of conduction electrons 29 resonant XES (RXES) 340, 349–351, 353, 355, 357–358 retrofitting 185 retrosynthesis, of organic compounds 162 reverse water-gas shift (RWGS) 331, 637 revised lone pair (RLP) model 566 Rh wet impregnation 172 RhAu alloy nanoparticles for NO decomposition by machine learning 518–520
Index
rhodium–platinum gauze 4 rhombohedral-shaped single nanocrystal 190 ring electrode 580 ring-opening, of epoxides 157 rotating disk electrode (RDE) 578–579 rotating ring-disk electrode (RRDE) 578, 580 Ru nanoparticles 62 Ru5 PtSn catalyst 94–95 RuO2 /TiO2 :Ta,N photocatalyst 561 ruthenium 5, 7, 535, 604, 676 rutile TiO2 photocatalysts 559–564
s Sabatier principle 507, 655 sacrificial reagents 205–206, 208–209, 559 sacrificial stabilizing agent 61 “safer solvent or auxiliary,” 633 salicylic acid 650 scaling relation and reaction energy 502–503 scanning electron microscopy (SEM) 31, 44, 134, 166, 244, 268, 290–291, 547, 586, 678, 679 scanning probe microscopy 220 scanning tunneling microscopy (STM) 228–229, 231–232, 235–237, 242, 253, 440, 658 scanning-type atomic-scale imaging techniques 253 Schrödinger equation 403, 405–407, 410, 419, 424, 432, 471–473, 477 second generation biofuels 595 secondary building units (SBUs) 151–152 selective catalytic reduction (SCR) process 344, 529 carbon-based catalysts 686 CHA-type SSZ-13 cation location 689–691 copper status 691 shape selevctivity 688–689
deNOx methods 694 HONO 693 metal oxide catalysts 685–686 microporous zeolites 693 NH3 692–694 of NO 328 noble metal-supported alumina 685 reaction sequence and reaction rate 694 zeolite catalysts 686–688 selective enzyme (catalase) tethering 169 self-consistent field (SCF) calculations 408, 409, 479 self-interaction error (SIE) 410, 412 self-supporting wafer 314, 327 semiconductor doping 674–676 semiconductor photoanode 45 semiconductor photocatalyst 13, 34, 203, 204, 210, 559–560, 660, 671–672 synthesis methods flux method 211 solid-state reaction method 211 semiconductor-based photocatalytic processes 673 semi-empirical functionals 410 sequential and non-overlapping pulses 72 shape anisotropy 162–165 silanized zeolitic seeds 130 silanol-rich domains 136 silica-supported Wells–Dawson heteropolyacid 602 silicate (SiO4 4- ) 9 silicate polyanions 130 silicate polymerization 130 silicoaluminophosphate 134, 691–692 silylane cationic polymers 130 simple metal oxide-based catalysts 187, 189–192, 197 simultaneous metallic catalyst 73 Si nanopillar/nanocones 162
723
724
Index
single atoms (SAs) 3–16, 34, 58, , 62, 68, 69, 71–74, 80, 103–114, 548, 551 single crystal X-ray diffraction (XRD) 86, 150 single-site catalysts 71 single walled CNTs 60 single-atom catalysts (SACs) 14, 58, 71, 104 advantages CAS, identification 105–106 intrinsic reaction mechanisms 106–107 maximum atom efficiency 105 unique catalytic properties 105, 106 chemical methods 108 bottom-up synthetic methods 109–112 top-down synthetic methods 112–113 concepts 104 physical methods 108 synthesis 107 SiO2 nanoparticles 94 SiO2 -supported Ni and Ni–Fe bimetallic catalysts 607 “site-isolated” catalysts 89 small angle X-ray scattering-CT (SAXS-CT) 274–275 small Fe nanoparticles 61 smooth overlap atomic position (SOAP) 519 Sn-BEA zeolite 601 Sn-containing molecular sieves 620, 628 Sn-zeolites 600 “soft-landing” conditions 83 SO3 H-functionalized polymer catalysts 600 SO3 H-silica catalysts 601 soft templates silanization templating methods 135–136
templating, with surfactants 130–135 soft-templating methods 124 sol–gel method 22, 675 sol–gel silica particles 10 solar reactors 680–681 solar water photocatalytic reactors 681 solid alkaline water electrolyzer (SAWE) 539–540 solid materials 124, 148, 218 solid molecules 153, 155 solid oxide electrolyte water electrolysis (SOEWE) 536, 540–541 solid polymer electrolyte 574, 657–658 solid solution metal oxide catalysts 192 solid-solution photocatalysts 209–210 solid-state heterogeneous catalysis 58 solid (semiconductor) photocatalyst 203 solution-phase synthesis 25 solvent assisted linker exchange (SALE) 155 solvent-free aerobic oxidation, of cyclohexene 92 solvothermal 153 solvothermal method 22, 560 solvothermal reduction, in liquid phase 61 solvothermal synthesis method 62, 211 Sonoco–Vacuum Oil Company 8 space-confined method 58, 125–127 carbon nanotubes and nanofibres 127–128 space-filled cluster representation 86, 87, 89 spatial localisation 169 spatially resolved surface photovoltage spectroscopy (SRSPS) 31 special matrix isolation technique 303 specific activity 165, 579–580, 585, 626 specific surface area (SSA) 60, 68, 120, 189, 196, 439, 551, 679 spectroscopic studies, issues in 363–364
Index
spectroscopic technique 229, 264, 271, 274, 279, 377–393 sp2 -hybridized carbon atoms 59 spin-trapped OH radicals 306 spin trapping process 303 spin traps 303, 306 spin/dip/doctor-blade coating 39 spinel oxides 542 spray pyrolysis 11, 15, 185–186 sputtering 39, 84, 241 stabilization of suspension 50 standard Gibbs free energy change 505 start–stop protocol 581 steady-state isotope transient kinetic analysis (SSITKA) 220, 325, 334, 369–371, 373 steam reformer 7, 636, 641–643 steam reforming of methane 636, 641–642 steric hinderance 123, 385 strong metal–support interaction (SMSI) 63, 90, 104, 113, 172 structure-directing agents (SDAs) 9, 11, 131–135, 138–139, 607 sub-bandgap excitation 189, 676 subnanometer Au regions 168 sulfuric acid 3–4, 67, 573, 576 super-resolution fluorescence microscopy 285–290 case studies 286, 287, 289, 290 superacid-like catalyst 238 superatoms 81, 88 supercritical fluids 327, 635, 639–641 supercritical water gasification (SCWG) 640 superimposed alternating current (AC) 82 surface energy 12, 13, 22–24, 35, 259 surface energy order 24 surface grafting 14, 15 surface metal-oxygen bond 42 surface modification of metallic nanoparticles (NPs) 391 surface plasmon resonance (SPR) effects 15, 29, 81, 168–169, 675–676
surface science approach 226 surface science studies 27, 322 surface sensitization 676 surfactant-silicate mesostructures 130 sustainable chemistry 527 Suzuki coupling 13 synthesis process 7, 22, 79, 85, 119, 633 synthetic zeolites 8
t Tamman temperature 218 tandem water electrolysis (TWE) 538–539 Ta2 O5 nanotubes 11, 557 temperature-programmed desorption (TPD) 140, 235 terephthalic acid 153, 622–624, 626–627 terminated nanocubes 165 ternary NiMoTi alloy 548 ternary oxide films 52 tetradentate nitrile ligand 147, 148 tetraethyl orthosilicate (TEOS) 9 tetrahedral geometry, of ligand 147 tetrahydrofurandimethanol (THFDM) 622, 623 tetrahydrofurfuryl alcohol 622 tetrakis(acetonitrile)copper(I) 147–148 theoretical cell voltage 583 theory of relativity 403 thermal catalysis 21, 39, 225 thermal decomposition method 22 thermal stability 150, 156, 168, 192, 197, 689 thermal treatment, in H2 gas 58 thermal/chemical/physical/vapour deposition 39 thermochemical methods 650 thermodynamic stability 113, 636 thermogravimetric analysis (TGA) 150, 327 thermophoretic deposition 185 thiol ligands 81
725
726
Index
thiolate-protected metal nanoclusters 86 third generation biofuels 595, 596 three-dimensional ordered macroporous La0.6 Sr0.4 MnO3 (3DOM LSMO) 172 three-dimensional ordered porous structures 162 three-electrode cell 576, 578, 580–582 three-electrode configuration 41 three-way catalysts (TWCs) 189–190, 192, 685 time-correlated single-photon counting (TCSPC) unit 386 time-resolved Cu K-edge XANES study 344, 354 time-resolved microwave conductivity (TRMC) 389 charge-carrier dynamics of Pd-Ni bimetal nanoparticles modified TiO2 391–393 experimental setup 390 time-resolved photoluminescence (TRPL) 386 components 386 experimental setup for 386, 387 heterogenous dye sensitized-C3 N4 catalysts 388 streak tube operating principle 387 TiO2 nano-crystalline 676 TiO2 nanorod 27 TiO2 nanotube (TNT) 11, 42, 44, 51, 290 tip-enhanced Raman 220 titanium diisopropoxide bis(acetylacetonate) precursor 168 titanium dioxide (TiO2 ) 27, 44, 237, 660 nanotube arrays 42 1T-MoS2 polytype 549 tomographic data collection strategy 264 tomography, in catalysts design
conventional absorption-CT 265–267 imaging, with X-rays 264–265 PDF-CT 269–270 XRD-CT 267–268 top-down approach, of fabricating clusters 82, 108, 162, 164 transesterification, of bulk triacyl glycerides 170 transient absorption spectroscopy (TAS) Beer–Lambert Law 378 history 378 schematic layout 379 uses 380 transition metal nanoparticles 61 transition state theory (TST) 420, 423–424, 497, 503, 505 transmission electron microscopy (TEM) 71, 88, 92, 94, 107, 113, 127, 129, 134, 139, 191, 193, 219, 231, 233, 253–255, 258–260, 291, 352, 547, 584–586 transmission IR spectroscopy 314, 315, 332 of CO adsorbed on Pd(111), Pd(100) and silica supported Pd catalyst 324 trickle-bed or slurry reactor 636 triethylamine (TEA) 306 trimethoxysilane (3-glycidoxypropyl) 135 triple phase boundary 573–575 tube-enclosed FSP configuration 190 tungsten disulfide 11 turbostratic structure 58 turnover frequencies (TOFs) 13, 92, 172, 245, 352, 519 σ-type donation 71 two half-reactions 41
u ubiquitous formation 183 ultrafast spectroscopic techniques time-resolved microwave conductivity 389–393
377
Index
time-resolved photoluminescence 386–389 transient absorption spectroscopy 377–386 ultrafine nanoparticles 103 ultrahigh-vacuum (UHV) 82, 217, 226–227, 324, 333 ultrathin NiCo bimetal-organic framework nanosheets 546 UMCM-1 topology 155, 157 under ultra-high vacuum (UHV) 81–86, 94, 95, 217–219, 226–232, 234, 237, 240–241, 257–258 unmodified Ag NPs 167 unoptimized electrode design 573 unprecedented narrow-gap oxyfluoride 564–566 uronic acids 617
v vacuum-annealed rutile TiO2 108 valence band (VB) 28, 29, 32, 203–206, 208–210, 377–378, 472, 490, 557, 564, 659–660, 665, 673–675 valence band controlled photocatalysts 208–209 valence band maximum (VBM) 32, 33, 203, 204, 557, 561, 566 value-added chemicals synthesis from carbohydrates dihydrolevoglucosenone 623 furandicarboxylic acid (FDCA) 623–626 1,6-hexanediol (1,6-HDO) 623 lactic acid 627–628 lactide 628–630 terephthalic acid 626–627 van der Waals forces (polarization forces) 50, 59, 228 vapor-fed flame synthesis (VFS) 185, 189 vaporization/sublimation, of droplet 186 vapour deposition methods 39 vectorial charge transport 11, 44, 45
velocity Verlet algorithm 421–422 Velocys methane steam reformer 642 Verlet algorithm 421–422, 427 vibrational spectroscopy 246, 301, 307, 311, 312 visible-light-driven overall water splitting 559 visible-light-responsive narrow-gap photocatalyst 557 visible-light-responsive photocatalysts GaN-ZnO 558 oxynitrides (oxide-nitrides) 557 rutile TiO2 photocatalysts 559–564 unprecedented narrow-gap oxyfluoride 564–566 visible-light-driven overall water splitting 559 water splitting 558 ZnGeN2 -ZnO 558 volatile organic compounds 183, 529 volatile organic molecules 184 voltammetry 440, 463, 468, 552, 576–581 voltammogram 576–578, 585–587 V-W-TiO2 -sepiolite catalyst for selective catalytic reduction (SCR) 328
w wastewater treatment plants (WWTPs) 681 wastewater treatment processes 529, 671 water-gas shift (WGS) reaction 13–14, 105–106, 110, 325, 331, 369–370, 436, 637, 641, 643–644 water oxidation 204, 208–209, 533, 542, 558–565, 567, 635, 640, 659 water photoelectrolysis 45 water splitting 11, 14, 26, 27, 39, 49, 57, 84, 94, 165, 190, 192, 203–204, 206, 209, 306, 356, 484, 528, 533–552, 557–568 water splitting reaction 49, 190, 192, 528, 536
727
728
Index
water-stable NH2 -MIL-88B(Fe2 Ni) MOF materials 547 wave-particle duality 403 well-defined metal clusters 15 wet-chemistry synthesis, optimization of colloidal nanoparticles, deposition 71 metal clusters/single metal atom deposition 71 dopant elements, on carbon 71 metal clusters, immobilization of 71–72 wet impregnation method 12, 69, 71, 106, 109, 172, 190, 194 wet synthesis techniques 194 wide-field fluorescence microscopy (WFM) 284–286, 291 window-type system 254–256, 258, 259 WO3 52, 94, 185, 189, 380, 557–558, 665, 674 nanoflowers 44 nanosheets 11 Wulff construction 22, 24, 260
x X-ray absorption near edge structure (XANES) 343, 345 energy resolution 357 sensitivity 357 X-ray absorption near edge structure-CT (XANES-CT) 270–272 X-ray absorption spectroscopy (XAS) 219, 367, 368 fluorescence-detected 343 limitations 356–357 time-resolved 353–356 transmission signal 342 X-ray damage 357 X-ray diffraction (XRD) 113, 114, 133–135, 150, 198, 269, 275, 307, 325, 327, 346, 358, 367, 560, 689 X-ray diffraction computed tomography (XRD-CT) 267–272
X-ray emission spectroscopy (XES) 347–351 high energy resolution (resonant) 348 limitations 356–357 sensitivity 357 time-resolved 354 X-ray fluorescence-CT (XRF-CT) 270–272 X-ray photoelectron spectroscopy (XPS) 71, 94, 165, 217, 219, 235, 246, 358 X-ray spectroscopy 339 applications 341 fundamentals of 339–342 in situ and operando cells 351–353 xylose 596, 606, 617, 622
y yttria stabilized zirconia (YSZ) 540
z Zeeman effect 296 Zeeman splitting 297–298 zeolite 110, 119 catalysts 686, 688 deactivation 124 microporous framework of 121 rings 120 shape selectivity 121, 122 structure 120 zeolite-templated carbon (ZTC) 10 zeolitic channels 644 zeolitic materials 126, 130, 131, 139 zero-gap electrode cell 537 zeta potential 50 ZIF-67 547 zinc chromite (Cr2 O3 -ZnO) catalyst 6 zinc oxide 48, 170 zirconyl chloride 11 ZnGeN2 -ZnO 558 ZrO2 600 Z-scheme system 209, 559–561, 564, 567, 665