High Temperature Electrolysis 3031225074, 9783031225079

Citation preview

Lecture Notes in Energy 95

Miguel Angel Laguna-Bercero   Editor

High Temperature Electrolysis

Lecture Notes in Energy Volume 95

Lecture Notes in Energy (LNE) is a series that reports on new developments in the study of energy: from science and engineering to the analysis of energy policy. The series’ scope includes but is not limited to, renewable and green energy, nuclear, fossil fuels and carbon capture, energy systems, energy storage and harvesting, batteries and fuel cells, power systems, energy efficiency, energy in buildings, energy policy, as well as energy-related topics in economics, management and transportation. Books published in LNE are original and timely and bridge between advanced textbooks and the forefront of research. Readers of LNE include postgraduate students and nonspecialist researchers wishing to gain an accessible introduction to a field of research as well as professionals and researchers with a need for an up-to-date reference book on a well-defined topic. The series publishes single- and multi-authored volumes as well as advanced textbooks. **Indexed in Scopus and EI Compendex** The Springer Energy board welcomes your book proposal. Please get in touch with the series via Anthony Doyle, Executive Editor, Springer ([email protected])

Miguel Angel Laguna-Bercero Editor

High Temperature Electrolysis

Editor Miguel Angel Laguna-Bercero Instituto de Nanociencia y Materiales de Aragón (INMA) CSIC—Universidad de Zaragoza Zaragoza, Spain

ISSN 2195-1284 ISSN 2195-1292 (electronic) Lecture Notes in Energy ISBN 978-3-031-22507-9 ISBN 978-3-031-22508-6 (eBook) https://doi.org/10.1007/978-3-031-22508-6 © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors, and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer imprint is published by the registered company Springer Nature Switzerland AG The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

Contents

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Miguel A. Laguna-Bercero

1

Fundamentals of Solid Oxide Electrolysis Cells (SOEC) . . . . . . . . . . . . . . . Miguel A. Laguna-Bercero, Yudong Wang, Xiao-Dong Zhou, and Liangzhu Zhu

5

Solid-State Electrolytes for Solid Oxide Electrolysis Cells . . . . . . . . . . . . . . Sivaprakash Sengodan

35

Oxygen Electrode Materials for Solid Oxide Electrolysis Cells (SOECs) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Vaibhav Vibhu, Amir Reza Hanifi, Thomas H. Etsell, and Jean-Marc Bassat Fuel Electrode Materials for Solid Oxide Electrolysis Cells (SOECs) . . . Muhammad Shirjeel Khan and Ruth Knibbe

59

91

Ceramic Coatings for Metallic Interconnects and Metal Alloys Support for Solid Oxide Electrolysis Applications . . . . . . . . . . . . . . . . . . . . 117 Elisa Zanchi, Antonio Gianfranco Sabato, Hassan Javed, Agnieszka Drewniak, Damian Koszelow, Sebastian Molin, and Federico Smeacetto Glass Ceramic Sealants for Solid Oxide Cells . . . . . . . . . . . . . . . . . . . . . . . . . 153 Jochen Schilm, Mihails Kusnezoff, and Axel Rost Modeling of Solid Oxide Electrolysis Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . 207 Yang Wang, Chengru Wu, Kui Jiao, Qing Du, and Meng Ni Protonic Ceramic Electrolysis Cells (PCECs) . . . . . . . . . . . . . . . . . . . . . . . . . 245 Laura Almar, Sonia Escolástico, Laura Navarrete, David Catalán-Martínez, Jesús Ara, Sonia Remiro-Buenamañana, Imanol Quina, and José M. Serra

v

vi

Contents

Durability and Degradation Issues in Solid Oxide Electrolysis Cells . . . . 277 Pattaraporn Kim-Lohsoontorn, Patthiya Prasopchokkul, Aritat Wongmaek, Parintorn Temluxame, Ramin Visvanichkul, Saharat Bairak, and Natthamon Nuengjumnong Emerging Trends in Solid Oxide Electrolysis Cells . . . . . . . . . . . . . . . . . . . . 313 Albert Tarancón, Marc Torrell, Federico Baiutti, Lucile Bernadet, Simone Anelli, Natalia Kostretsova, and Maritta Lira Stack/System Development for High-Temperature Electrolysis . . . . . . . . . 383 Hamza Moussaoui, Vanja Suboti´c, Jan Van herle, Ligang Wang, Xinyi Wei, and Hangyu Yu Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 409 Miguel A. Laguna-Bercero

Abbreviations

AAO AC ADL AFL AISI ALD ALS AM ANN ASR AST BCY BCZY BCZYYb7111 BoP BSCF BSCFNb BSCFTa BSCFV BSCFW B-V BZCY BZCY72 BZCYYb BZCYYb4411 BZY BZY20 CAD CAPEX CC CCS

Anodized aluminum oxide Alternate current Anode diffusion layer Anode functional layer American Iron and Steel Institute Atomic layer deposition Adler-Lane-Steele Additive manufacturing Artificial neural network Area specific resistance Accelerated stress test Y-doped barium cerate BaCe1-x-y Zry Yx O3-δ BaZr0.7 Ce0.1 Y0.1 Yb0.1 O3-δ Balance-of-plant Ba1-x Srx Co1-x Fex O3 Ba0.5 Sr0.5 Co0.8-x Fe0.2 Nbx O3-δ Ba0.5 Sr0.5 (Co0.8 Fe0.2 )1-x Tax O3-δ Ba0.5 Sr0.5 (Co0.7 Fe0.3 )1-x Vx O3-δ Ba0.5 Sr0.5 (Co0.7 Fe0.3 )1-x Wx O3-δ Butler–Volmer Yttria and ceria-doped barium zirconate BaZr0.7 Ce0.2 Y0.1 O3-δ Ytterbia, yttria and ceria-doped barium zirconate BaZr0.4 Ce0.4 Y0.1 Yb0.1 O3-δ Y-doped barium zirconate BaZr0.8 Y0.2 O3-δ Computer-aided design Capital expenditure Current collector Carbon capture and storage vii

viii

CDL CFD CFL CFY CGSZ CMR CPE CVD CYSZ DC DEM DFT DGM DIP DIW DLP DMC DME DOD DR DRT DTU EC EC EDS EERA EIS ELD EPD ESC FESC FIB-SEM FLW GAN GC GDC G-LSTM HEM HEMAA HEO HER HHV HRR HTE IC

Abbreviations

Cathode diffusion layer Computational fluid dynamic Cathode functional layer Chromium-based interconnects Ceria and gadolinia-doped zirconia Catalytic membrane reactor Constant phase element Chemical vapor deposition Zirconia doped with yttria and ceria Direct current Discrete element method Density functional theory Dust gas model Direct inkjet printing Direct ink writing Digital light processing Dynamic Monte Carlo Direct dimethyl ether Drop-on-demand Dry reforming reaction Distribution of relaxation times Technical university of Denmark Electronic conductor Equivalent circuit Energy-dispersive X-ray spectroscopy European energy research alliance Electrochemical impedance spectroscopy Electrolytic deposition Electrophoretic deposition Electrolyte-supported cell Fuel electrode-supported cell Focused ion beam-scanning electron microscopy Finite-length Warburg Generative adversarial network Gas chromatography Gadolinia-doped ceria Grid-long short-term memory High entropy materials High-energy micro-arc alloying High entropy oxide Hydrogen evolution reaction High heating value Hydrogen reduction reaction High temperature electrolysis Ionic conductor

Abbreviations

IEDP IIP IJP KMC kNN KPI LBM LCeNT LCM LDC LED LHHW LHV LNCO LNO LPG LPNCO LPNO LSBT LSC LSCeCrFN LSCF LSCMF LSCN LSCoM LSCrM LSCuF LSF LSFT LSFTC LSGM LSM LSNMF LST LTE LV MDA MEMS MIEC ML MOO MPEC MRT MS MSR

ix

Isotopic exchange depth profiling Indirect inkjet printing Inkjet printing Kinetic Monte Carlo k-nearest neighbor Key performance indicator Lattice Boltzmann method La0.8 Ce0.1 Ni0.4 Ti0.6 O3-δ Lanthanum calcium manganite Ceria doped with lanthanum Light emitting diode Langmuir–Hinshelwood–Hougen–Watson Low heating value La2 Ni0.8 Co0.2 O4+δ La2 NiO4+δ Liquefied petroleum gas La1.5 Pr0.5 Ni0.8 Co0.2 O4+δ La1.5 Pr0.5 NiO4+δ Lanthanum barium-doped strontium titanate Lanthanum strontium cobaltite (La0.65 Sr0.3 Ce0.05 )0.9 (Cr0.5 Fe0.5 )0.85 Ni0.15 O3Lanthanum strontium cobalt ferrite La0.6 Sr0.4 Co0.2 Mn0.2 Fe0.6 O3 Lanthanum strontium cobalt nickelate La1-x Srx Co1-x Mnx O3 La0.75 Sr0.25 Cr0.5 Mn0.5 O3-δ Lanthanum strontium copper ferrite Lanthanum strontium ferrite La0.3 Sr0.7 Fe0.7 Ti0.3 O3 Cobalt and titanium substituted lanthanum strontium ferrite Strontium and magnesium-doped lanthanum gallate Lanthanum strontium manganite La0.6 Sr0.4 Ni0.2 Mn0.2 Fe0.6 O3 La-doped SrTiO3 Low temperature electrolysis Low voltage Methane dehydroaromatization Microelectromechanical processes Mixed ionic and electronic conductor Machine learning Multi-objective optimization Mixed protonic and electronic conductivity Multiple-relaxation-time Mass spectrometry Methane steam reforming reaction

x

MS-SOEC NA NNCO NNO NS OCV OER OPEX ORR OSOEC PBCO PCA PCEC PCER PEM PEN PET PFM PMMA PMR PNCO PNO PrDC PSCF PSOEC PtG PtM PtX PVA PVD RC REO RePCEC RePEM ReSOC RP RWGS SACN ScSZ SDC SEM SFM SIMS SLA SMM

Abbreviations

Metal-supported solid oxide electrolysis cell Avogadro number Nd2 Ni0.8 Co0.2 O4+δ Nd2 NiO4+δ Navier–Stokes Open-circuit voltage Oxygen evolution reaction Operational expenditure Oxygen reduction reaction Oxygen ion-conducting solid oxide electrolysis cell PrBaCo2 O5+δ Principal component analysis Protonic ceramic electrolysis cells Protonic ceramic electrochemical reactor Proton exchange membrane fuel cell Positive-electrolyte-negative Porous electrode theory Phase-field model Poly methyl methacrylate Protonic membrane reformer Pr2 Ni0.8 Co0.2 O4+δ Pr2 NiO4+δ Praseodymia-doped ceria Pr0.58 Sr0.4 Fe0.8 Co0.2 O3-δ Proton-conducting solid oxide electrolysis cell Power-to-gas Power-to-methane Power-to-X Poly(vinyl alcohol) Physical vapor deposition Robocasting Reactive element oxides Reversible protonic ceramic electrochemical cell Reversible polymer electrolyte membrane cell Reversible solid oxide cell Ruddlesden–Popper Reverse water gas shift SiO2 -Al2 O3 -CaO-Na2 O Scandia-stabilized zirconia Samaria-doped ceria Scanning electron microscopy Sr2 Fe2-x Mox O6-δ Secondary ion mass spectroscopy Stereolithography Stefan–Maxwell model

Abbreviations

SMR SoA SOC SOE SOEC SOFC SoH SSC SSCF STFC STN SVD SVM TEC TGA THD TPB TPBl TPCF TXM UV VOED WGS WGSR XPS YbSZ YSZ

xi

Steam methane reforming State of the art Solid oxide cell Solid oxide electrolyzer Solid oxide electrolysis cell Solid oxide fuel cell State-of-the-health Samarium strontium cobaltite Sm1-x Srx Co1-x Fex O3 SrTi0.3 Fe0.6 Co0.1 O3-δ Sr0.95 Ti0.9 Nb0.1 O3 Singular value decomposition Support vector machine Thermal expansion coefficient Thermogravimetric analysis Total harmonic distortion Triple-phase boundary Three-phase boundary length Two-point correlation function Transmission X-ray microscopy Ultraviolet Virkar’s oxygen electrode delamination Water gas shift Water gas shift reaction X-ray photoelectron spectroscopy Ytterbia-stabilized zirconia Yttria-stabilized zirconia

Introduction Miguel A. Laguna-Bercero

The growing emissions of CO2 and other greenhouse gases has led to a global environmental warning. In order to restraint climate change, the use of sustainable and renewable alternatives to fossil derived products is needed nowadays more than ever. Renewable energies are partially covering those needs in recent decades. However, the demand is growing constantly to reach carbon neutrality, and one of the issues of renewable energies is their intermittent nature that avoids the continuous energy supply. Thus, the surplus of renewable energy should be stored, e.g., by its conversion to chemical carriers such as hydrogen, methane (Ebbesen et al. 2014). In fact, hydrogen is considered as one of the most attractive chemical carriers which can be used as a clean fuel for different applications (Bi et al. 2014). The urgent need of an energy transition worldwide has meant that practically, all countries have developed (or are urgently developing) national hydrogen roadmaps, including real plans for the implementation of this new technology. The hydrogen strategic vision for a climate-neutral EU will give a boost to clean hydrogen production in Europe (European Commission 2020). It this then clear that the condition of hydrogen as an energy vector and its high versatility gives the ability to position itself as a key tool for the integration of the different sectors energy, which will favor greater energy flexibility, availability, and security, as well as greater efficiency and profitability in the energy transition, contributing to the decarbonization of the economy. All these plans aim to identify the challenges and opportunities for the full development of renewable hydrogen, providing a series of measures aimed at promoting investment action. It is expected, at least in Europe, a minimum contribution of 25% of renewable hydrogen with respect to the total consumed in the industry

M. A. Laguna-Bercero (B) Instituto de Nanociencia y Materiales de Aragón (INMA), CSIC—Universidad de Zaragoza, C/María de Luna 3, E-50018 Zaragoza, Spain e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M. A. Laguna-Bercero (ed.), High Temperature Electrolysis, Lecture Notes in Energy 95, https://doi.org/10.1007/978-3-031-22508-6_1

1

2

M. A. Laguna-Bercero

for the year 2030. On the other hand, short and long-term energy storage can materialize through the use of renewable hydrogen as energy vector, facilitating the use of existing infrastructures. In this sense, the role of renewable hydrogen applications in fuel cells, or also as an intermediate element in Power-to-X technologies, is one of the key parameters for the energy storage strategy worldwide. It is also important to remark that this is not the first time that hydrogen has been impulse as the solution to replace fossil fuels to mitigate climate change. But this time the strongest driver is the decarbonization of the energy sector in order to get zero neutrality. This is the reason why recently emerged key enablers such as the availability of low-cost renewable electricity and the matureness of hydrogen technologies. The 2015 Paris Agreement has provoked an impulse on the part of relevant actors including government, industry, and investors (United Nations 2015). The aforementioned proclamations of net-zero targets by major economies, including the European Union by 2050 and China by 2060, have accelerated this preference (Fetting 2020). Water electrolysis is then a clear technology for the hydrogen economy, as falling costs for hydrogen produced with renewable energy, combined with the urgency of cutting greenhouse gas emissions, have given clean hydrogen unprecedented political and business momentum, being hydrogen the missing link in the transformation of the global energy system (Taibi et al. 2020). Water electrolysis for green hydrogen production or, more generally, Power-to-X (PtX) technologies are considered one of the most promising solutions for massive energy storage of the electricity generated intermittently by renewable power sources. In addition, it is also the solution for decarbonizing energy-intensive industries such as for example iron and steel manufactures, refineries or chemical plants. During the electrolysis process, it is very important to remark that the efficiency in the conversion from electricity to chemical energy is critical since electricity costs represent more than half of the price. Consequently, there is an important need to develop highly performing and cost-efficient electrolyzers. Among the different types of electrolysis systems, those based on high-temperature solid oxide electrolysis cells (SOECs) are the most efficient (>80%LHV) and also present higher production yields and lower specific electric energy (5 kWh/Nm3 ) and proton exchange membrane (PEM, > 6 kWh/Nm3 ) electrolyzers. SOEC significantly lowers the electrical energy consumption of the process compared to commercial low-temperature electrolysis. However, operational lifetime is one of the main actual concerns for the launch of co-electrolysis systems into the market. In despite of this, several efforts are still needed to understand the degradation processes occurring upon long-term operation, as the degradation rates are still far from those of SOFC mode. All of these aspects will be covered in detail in subsequent sections of this book. In this context, an efficient and environmentally-friendly production of H2 can be obtained by using high-temperature electrolysis combining renewable electricity and available sources of heat and steam from non-fossil energy sources (Duan et al. 2020). In addition, clean synthetic fuels or high added-value chemicals can be produced by power to gas and power to liquids technologies, if CO2 electrolysis or H2 O/CO2 co-electrolysis is performed (Remiro-Buenamañana

Introduction

3

et al. 2021). Direct production of highly reactive syngas by co-electrolysis represents an additional advantage of SOECs, especially when looking for higher value-added products. Despite their high potential, SOECs still remain the less mature electrolysis technology, in comparison with alkaline and PEM technologies. SOEC comprises the use of oxide-ion or proton conducting ceramic electrolytes. Particularly, protonic ceramic electrolysis cells (PCEC) pump protons and form dry H2 , leaving O2 on the steam side, requiring fewer separation steps and obtaining directly pressurized dry H2 . Besides PCECs can operate at lower temperatures than SOEC due to the lower activation energy of the proton transport as compared with oxide ions. Several demonstration projects are under way in which electrolyzers are connected to electricity and gas grids, and a huge number of new projects are continuously being announced in 2022. However, there are still several gaps remaining in the knowledge of what electrolyzers can ultimately achieve, at what cost, and where they may be most effective in meeting policy and market needs. Some of these aspects are very difficult to predict, as for example electricity prices are continuously fluctuating due to the instability of the market. A good overview about the SOFC history is given by Smolinka et al. (2022). The first SOEC applications were those performed by Westinghouse and General Electric in the 1960s. Solid electrolyte stacks were already investigated for electrolysis of the atmosphere in spacecrafts for recovery of oxygen from and H2 O (Elikan et al. 1971) and in a NASA project performing CO2 electrolysis with solid oxide electrolyte cells for oxygen recovery in life support systems (Isenberg and Cusick 1988; Spacil and Tedmon 1969). The first significant results were reported by Donitz et al. in the 1980s with the HotElly project (high operating temperature electrolysis) system led by Dornier GmbH (Dönitz et al. 1980), which consisted on SOEC research including a pilot plant test. High-temperature electrolysis started to have more attention in the first decade of this millennium, especially with the advances reported at Ceramatec, the Japanese Central Research Institute of Electric Power Industry, and the efforts at DTU in Denmark. An additional boost was when Topsoe Fuel Cells decided to stop fuel cell activities and focused on electrolysis development, which could be later connected with chemical engineering and ammonia production plants from the company. By this period, important companies such as Haldor Topsoe and Sunfire have recognized that the change to zero-emission economy can gradually change the view on this technology making it suitable for hydrogen and syngas generation (e-fuels and e-chemicals) by solid oxide electrolysis. The gas industry in France (ENGIE, GDF) is also following this trend. In USA, a new company was founded recently by Ceramatec (OxEon Energy), and one of their goals is the development of a CO generator for Mars mission using SOECs (Hartvigsen et al. 2019). In addition, a huge number of new projects are continuously being announced in Europe in the last five years, including companies such as Sunfire, SolidPower, NexCeris, CeresPower, Bosch, Elcogen, and Convion. The scale of those projects is ranging from a few kW up to about 2.5 MW. We are therefore faced with a historic opportunity to give a definitive boost to hydrogen technologies and in particular to high-temperature electrolysis. This book serves to describe the state of the art of technology, collecting aspects that range from

4

M. A. Laguna-Bercero

fundamental thermodynamics, through the description of the materials that compose the different components, up to the development of SOEC cells and stacks. The current state of modeling and important aspects such as new strategies and designs within this still incipient technology is also covered. It is to be hoped that in the coming years, this technology will already be a reality in our daily lives.

References Bi L, Boulfrad S, Traversa E (2014) Steam electrolysis by solid oxide electrolysis cells (SOECs) with proton-conducting oxides. Chem Soc Rev 43(24):8255–8270 Dönitz W, Schmideberger R, Steinheil E, Streicher R (1980) Hydrogen production by high temperature electrolysis of water vapour. Int J Hydrogen Energy 5:55–63 Duan C, Huang J, Sullivan N, O’Hayre R (2020) Proton-conducting oxides for energy conversion and storage. Appl Phys Rev 7(1):011314 European Comission (2020) ‘A hydrogen strategy for a climate-neutral Europe. European Commission’, Brussels Ebbesen SD, Jensen SH, Hauch A, Mogensen MB (2014) High temperature electrolysis in alkaline cells, solid proton conducting cells, and solid oxide cells. Chem Rev 114(21):10697–10734 Elikan L, Morris JP, Saunders CG, Wu CK (1971) 180-day life test of solid electrolyte system for oxygen regeneration. ASME paper 71-Av-32:1–12 Fetting C (2020) The European green deal. ESDNReport, Vienna Hartvigsen J, Elangovan S, Frost L (2019) OxEon energy demonstration of manned-mission scale ISRU process systems. ICES-2019–257, Boston, MA Isenberg AO, Cusick RJ (1988)Carbon dioxide electrolysis with solid oxide electrolyte cells for oxygen recovery in life support systems. J Aerospace, SAE Technical Paper 881040 Remiro-Buenamañana S et al. (2021) Membranes technologies for efficient CO2 capture–conversion. Engineering Solutions for CO2 Conversion:55–83 Smolinka T, Bergmann H, Garche J, Kusnezoff M (2022) The history of water electrolysis from its beginnings to the present’ in electrochemical power sources: fundamentals, systems, and applications: hydrogen production by water electrolysis, Elsevier, pp 83–164 Spacil HS, Tedmon CS Jr (1969) Electrochemical dissociation of water vapor in solid oxide electrolyte cells: I. thermodynamics and cell characteristics. J Electrochem Soc 116:1618–1633 Taibi E, Blanco H, Miranda R, Carmo M (2020) Green hydrogen cost reduction. IRENA Report United Nations (2015) Paris agreement

Fundamentals of Solid Oxide Electrolysis Cells (SOEC) Miguel A. Laguna-Bercero, Yudong Wang, Xiao-Dong Zhou, and Liangzhu Zhu

1 Thermodynamics of SOEC at Equilibrium An SOEC consists of two porous electrodes that are separated by a layer of dense ion-conducting ceramic electrolyte. For steam electrolysis, water is reduced in the porous fuel electrode under an applied voltage, forming hydrogen and oxide ions. The oxygen ions transfer through the electrolyte and are oxidized into oxygen in the oxygen electrode, as shown in Eqs. (1) and (2), and also in Fig. 1a. Fuel electrode: 2H2 O(g) + 4e−  2O2− + 2H2 (g)

(1)

2O2−  4e− + O2 (g)

(2)

Oxygen electrode:

These reactions occur at the active sites, where the gas phase, the ionic conductive phase, and the electronic conductive phase meet with each other at a so-called M. A. Laguna-Bercero (B) Instituto de Nanociencia y Materiales de Aragón, Universidad de Zaragoza-CSIC, 50018 C/María de Luna 3, Zaragoza, Spain e-mail: [email protected] Y. Wang · X.-D. Zhou Department of Chemical Engineering and Institute for Materials Research and Innovation, University of Louisiana at Lafayette, Lafayette, LA 70504, USA L. Zhu Key Laboratory of Advanced Fuel Cells and Electrolyzers Technology of Zhejiang Province, Ningbo Institute of Materials Technology and Engineering, Chinese Academy of Sciences, Ningbo 315201, China © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M. A. Laguna-Bercero (ed.), High Temperature Electrolysis, Lecture Notes in Energy 95, https://doi.org/10.1007/978-3-031-22508-6_2

5

6

M. A. Laguna-Bercero et al.

Fig. 1 a Schematic representation of an SOEC for water splitting. b Energy demand for water electrolysis as a function of the electrolysis temperature (Graves et al. 2011a, b)

triple-phase boundary (TPB). To facilitate the conduction of oxygen ions in the ceramic electrolyte, SOECs are operated at elevated temperatures (500–850 °C). Although a high operation temperature may lead to issues of material degradation, the high-temperature operation does offer thermodynamic benefits for the electrochemical hydrogen production. The total reaction occurring in a SOEC is highly energy demanding. 2H2 O(g)  2H2 (g) + O2 (g)

(3)

The total energy required equals the molar enthalpy change of the reaction (3), which is equivalent to the low heating value (LHV) of hydrogen. The electrical energy demand for the electrolysis is the Gibbs free energy change, which is a function of operating temperature given by: G = H − T S

(4)

The subtracted term, T S, is coming from the heat conversion. Figure 1b shows the change of different energy demand as a function of electrolysis temperature at standard state conditions (partial pressures of all species equal to 1 atm). The dash line shows the boiling point of water and categorizes the electrolysis processes into high-temperature electrolysis (HTE) and low-temperature electrolysis (LTE). The LTE occurs at T < 100 °C to split liquid water, during which the total energy required corresponding to the high heating value (HHV) of hydrogen. The extra energy demand for the LTE is originated from water evaporation energy. The electricity demand decreases with increasing temperature. Therefore, HTE is more cost-effective because of the less use of electricity power. Note that the electrical energy demand also depends on the partial pressures of individual species, electrolysis current, and electrode properties.

Fundamentals of Solid Oxide Electrolysis Cells (SOEC)

7

The electrical voltage of an electrolyzer is determined by the difference in oxygen chemical potential, μ O2 , between the oxygen electrode and the fuel electrode. The μ O2 is defined as partial molar Gibbs free energy of oxygen in the system:  μ O2 =

∂G ∂ N O2

 (5) T, p

The oxygen electrode is typically fed by air, which can be considered as an ideal E gas mixture. The chemical potential of oxygen at the oxygen electrode, μ O O2 , follows:  E μO O2

=

μ0O2

+ RT ln

p OO2E

 (6)

p0

where μ0O2 is the chemical potential of pure oxygen gas at 1 atm, p OO2E is the oxygen partial pressure of the sweeping in the oxygen electrode, and p0 is the reference pressure, which is 1 atm. At the fuel electrode, hydrogen and steam are fed, which leads to a very small amount of oxygen in the fuel electrode, following the equilibrium of reaction (3) and the following relation among the partial pressures: Kp =

p 2H2 p O2

(7)

p 2H2 O p0

where K p is the equilibrium constant, which is a function of standard molar Gibbs free energy change of reaction (3). r G 0 = −RT lnK p

(8)

FE The oxygen chemical potential in the fuel electrode, μ O , is given by the controlled 2 FE steam partial pressure ( p H2 O ) and hydrogen partial pressure ( p HF 2E ):

 FE μO = μ0O2 − r G 0 + 2RT ln 2

p HF 2EO



p HF 2E

(9)

The Nernst potential or reversible cell voltage is usually measured by open circuit voltage (OCV) of the electrolyzer and is given by: Vrev =

E FE μO O2 − μ O2

4F

⎡  2 ⎤ r G 0 RT ⎣ p OO2E p HF 2E ⎦ = + ln 4F 4F p0 p HF 2EO

(10)

8

M. A. Laguna-Bercero et al.

Besides the standard Gibbs free energy term considered in Fig. 1b, the gas composition and the system pressure affect the reversible cell voltage and the thermodynamics of the electrolyzer reaction.

2 Energy Loss Mechanism in SOEC The current density characterizes the hydrogen production rate in the electrolyzer. Increasing the area specific hydrogen production rate of a solid oxide cell can reduce the capital cost of the SOEC. The Faraday’s law of electrolysis gives the area specific hydrogen production rate from the electrolysis current: n˙ H2 = η F,H2

I 2F

(11)

where I is the average current density and η F,H2 is the Faradaic efficiency of hydrogen production. For the steam electrolysis in an SOEC, the η F,H2 can reach nearly 1. Therefore, a higher electrolysis current can enable one to achieve less material and capital cost for an SOEC stack. From the efficiency perspective, the electricity cost is proportional to the electrolysis voltage. The driving force of the electrolysis current is the higher voltage applied on the electrolyzer than the reversible voltage. Figure 2 gives the current density (I)-voltage (V ) curve of an SOEC. Vcell = Vrev + ηact + ηohm + ηcon

Fig. 2 I-V relation of an SOEC and the contribution of different overpotentials

(12)

Fundamentals of Solid Oxide Electrolysis Cells (SOEC)

9

The additional voltage required is the overpotential of an electrolyzer, including the activation overpotential (ηact ), the ohmic overpotential (ηohm ), and the concentration overpotential (ηcon ). The activation polarization results from the electrochemical reaction kinetics of both electrodes. The Butler–Volmer equation offers the relation between ηact and I. For example, at the oxygen electrode side, there exists:

 OE  O E  OE  αa z Fηact αcO E z Fηact I = I0 exp − exp − RT RT

(13)

I0 is the exchange current density. A higher I0 value indicates more active electrode, indicating a smaller overpotential is needed to achieve the same electrolysis current density. The resistance from the electrode polarization can be written by taking derivatives of Eq. (13). dη O E r O E = act dI  OE  OE OE  O E  −1 αa z Fηact αa z F αcO E z Fηact αcO E z F exp exp − = I0 (14) + RT RT RT RT where ηact is dominating at a low operation current. As the current increases, the electrode is highly polarized and activated with a decreased r O E . A highly active electrode is needed to reduce ηact . The ηohm originates from the electrical resistances of cell components. The electrodes are made of either electronically conductive ceramics or cermets, and their conductivities are usually higher than 100 S cm−1 . On the other hand, oxygen ions, O2− , are majority charge carriers in an yttrium-stabilized zirconia (YSZ) electrolyte. The mobility of oxygen ions is much lower than that of electrons, resulting in a lower conductivity. The ohmic overpotential, ηohm , exhibits a linear relationship with the electrolysis current density. ηohm = I

l σ Oel2−

(15)

where l is the thickness of the solid electrolyte, and σ Oel2− is the oxygen ion conductivity of the electrolyte. The ionic conductivity is a key material property that contributes to the ohmic loss. The conductivity of an electrolyte is determined by the ionic mobility and the amount of oxygen ion vacancies (VO·· ). For a YSZ electrolyte, yttrium ion occupies the zirconium site. The concentrations of yttrium defects  (Y Zr )and oxygen vacancies(VO·· ) are governed by the electroneutrality conditions:      2 VO·· = Y Zr

(16)

10

M. A. Laguna-Bercero et al.

The oxygen ionic conductivity (σ O 2− ) can be expressed as:    z 2 2− F 2   z2 2 − F 2 (17) σ O 2− = σVO·· = VO·· O D[VO·· ] = Y Zr O D ·· RT 2 RT [VO ]    where Y Zr can be calculated from the doping level in the yttria stabilized zirconia (YSZ), and D[VO·· ] is the diffusion coefficient of oxygen vacancies. Arrhenius equation can be used to describe the temperature effect on D[VO·· ] . 

 1 Ea 1 − D[VO·· ] (T ) = D[VO·· ],T0 exp R T0 T

(18)

where D[VO·· ],T0 is the diffusion coefficient of oxygen vacancies at T 0 , and E a is the activation energy for oxygen ionic conduction. Due to the diffusion limit of reactants and products, the concentrations of the species at the triple-phase boundary are different from the inlet gas. The difference in concentration is more profound when the reaction rate is substantially high and requires a higher voltage for the electrolysis. The concentration overpotential can be estimated by the following equation (Larminie et al. 2018): ηcon

  I RT ln 1 − =− 2F IL

(19)

where I L is the limiting current density of the cell, which is governed by the inlet gas flow rates and the microstructure of the electrodes. The higher electrolysis voltage requires an extra electrical energy. Therefore, the minimization of the overpotentials becomes the key to reduce the energy loss and enhance the energy efficiency of hydrogen production.

3 Energy Efficiency in SOEC The high-temperature water electrolysis entails the conversion of the electrical energy into the chemical energy stored in hydrogen. Therefore, the electrical efficiency can be calculated by using the ratio between the LHV of the produced hydrogen and the consumed electrical energy (Harrison et al. 2010): Eletrical efficiency =

L H V · n˙ H2 VT N ,L H V = I Vcell Vcell

(20)

where VT N,L H V is the thermal neutralvoltage of the electrolysis based on LHV of HV = 1.29V . The electrolyzer, however, can be operated hydrogen VT N ,L H V = L2F below VT N ,L H V , which results in an electrical efficiency higher than 100%. The reason is because the total energy for high-temperature water electrolysis comes not

Fundamentals of Solid Oxide Electrolysis Cells (SOEC)

11

only from the electrical energy but also from the thermal energy. The consumed heat is included in the electrical efficiency calculation. T S is the heat needed from the thermodynamic analysis, which can be partially or fully compensated by Joule heat generated from the electrical conduction. The external heat power required can be calculated by the following equation:   L H V · n˙ H2 − I Vcell = I VT N ,L H V − Vcell

(21)

The endothermic/exothermic regions can be differentiated by comparing VT N ,L H V with the operation voltage Vcell , Q > 0 or VT N ,L H V > Vcell refers to an endothermic process while Q < 0 or VT N ,L H V < Vcell indicates to an exothermic process. The energy efficiency of the electrolyzer is given by the expression: Energy efficiency =

L H V · n˙ H2 I Vcell + Heat supply

(22)

When the electrolyzer is operating under a voltage lower than VT N ,L H V , external heat supply is needed, and the system efficiency can reach a very high value (even close to 100%). The loss of energy comes from the heat leakage. Figure 3 compares the current–voltage curve range of SOEC with other room temperature electrolysis technologies. The reversible cell voltage is lower because the electrolysis starts from high energy steam instead of liquid water. The advantage of SOEC is further boosted because high-temperature cells often exhibit a lower internal resistance than low-temperature cells. The overpotential of SOEC is lower than the room temperature electrolyzer, due largely to faster electrochemical and transport kinetics at high temperatures. A lower operating voltage enables a higher energy efficiency, which makes SOEC a promising system for highly effective energy conversion. Furthermore, no precious metal is needed in SOEC, which further reduces the capital cost for hydrogen production. In addition to water splitting, CO2 reduction is an important application for SOEC. A much lower electrolysis voltage for CO2 electrolysis is required in SOEC than lowtemperature electrolyzer, as shown in Fig. 3b. The highly activated catalyst at high temperatures reduces the activation overpotential and the energy loss. When CO2 and CO are fed to a fuel electrode, the Faradaic efficiency for CO production is ~ 100%. Moreover, the reversible cell voltage decreases from 1.33 V at 25 °C to 0.97 V at 800 °C due to the positive S of the CO2 reduction reaction (Küngas et al. 2020). Though the total energy change (H ) remains almost constant, less electrical energy is needed, and the additional heat requirement can be met by the Joule heat in the electrolyzer. The high-temperature electrolysis is the energy effective approach to achieve the electrochemical conversion from CO2 to CO.

12

M. A. Laguna-Bercero et al.

Fig. 3 Typical performance range of various water electrolysis technologies. The thermal neutral voltage divides the operation map into endothermic/exothermic regions (Graves et al. 2011a, b; Hauch et al. 2020)

4 Protonic Ceramic Electrolysis Cell (PCEC) Proton conducting ceramics are gaining more attention to reduce the operation temperature of reversible solid oxide cells. In fact, a dedicated chapter will be later discussed. The protons are major carriers in the electrolyte instead of large size oxygen ions in YSZ, which enables a lower ohmic loss in the electrolyzer in a temperature range from 400 to 600 °C. Steam is fed and split into protons and oxygen in the oxygen electrode. The protons then form hydrogen in the fuel electrode (Fig. 4). Fuel electrode: 4H+ + 4e−  2H2 (g)

(23)

2H2 O  4H+ (g) + O2 (g) + 4e−

(24)

Oxygen electrode:

The hydrogen can be produced with a low humidity. Moreover, these reactions enable the separation of hydrogen generation and steam supply. In addition, the electrochemically active nickel particles stay under dry atmosphere and low temperature, so that its coarsening can be suppressed, and the microstructure of fuel electrode can be preserved. As an emerging technology, the primary challenge for PCEC is the electronic current leakage, which results in a lower Faradaic efficiency than the traditional

Fundamentals of Solid Oxide Electrolysis Cells (SOEC)

13

Fig. 4 Schematic of a proton conducting electrolyzer cell

SOEC. Protonic ceramics generally exhibit mixed proton, oxygen ion, and electron hole conduction, while the protonic flux is dominating. The transference number of each charge carrier depends on the material composition, operating temperature, atmosphere, and polarization current density. At a higher electrolysis current, more electronic defects are generated locally, resulting in a higher electronic conductivity and lower Faradaic efficiency (Vøllestad et al. 2019).

5 Performance and Characterization of SOEC The typical characterization of an electrolyzer is very similar to that one used in fuel cell mode, which is the voltage (V ) versus current density (I) curves, as the one previously illustrated in Fig. 3. A comparison between alkaline, PEM electrlyzers, and solid oxide electrolyzers in terms of voltage vs. current density can be found in the work from Barelli et al. (2017). On the contrary as in fuel cell mode, when electricity is supplied to the cell (current increase), the voltage increases due to the internal resistance, overpotentials, and concentration effects. Considering that excess of steam is supplied, the voltage linearly increases with current up to the thermoneutral voltage. In general, the linear relationship remains up to a fuel utilization of about 80%. Then, the cell voltage applied to an electrolyzer at a current different than zero is always larger than the

14

M. A. Laguna-Bercero et al.

thermodynamic cell voltage at the same operating conditions to provide the driving force to assure charge carriers exchange. When operating in fuel cell mode, there is an agreement to give current densities (in mAcm−2 or Acm−2 ) at a specific operating voltage, typically about 0.7 V. Maximum power densities (in mWcm−2 or Wcm−2 ) at voltages close to 0.5 V are also used as reference. On the contrary to fuel cell mode, power density values are not given in electrolysis mode, being current density values the most common reference parameter. In electrolysis mode, as previously explained, thermoneutral voltage for high-temperature electrolysis occurs at 1.29 V, and operating around this voltage is generally recommended. For this reason, when comparing different electrolysis cells or stacks, current densities at a voltage of about 1.3 V are typically given, for each temperature, pressure, and fuel conditions. In terms of reversibility, SOEC presents a much more linear voltage–current density behavior in both operational modes (SOFC/SOEC) than low-temperature alkaline and protonic electrolyzers. This is indicating a high degree of reversibility or in other words, that electrode potentials are close to equilibrium with no influence of the current density direction. In theory, this could be an advantage, as SOFC is a mature technology. For example, lifetime of SOFC stacks has been demonstrated for a period of more than 100,000 h (Fang et al. 2019). In fact, this reversibility good behavior opens novel applications, such as solid oxide regenerative (or reversible) fuel cells (Wang et al. 2017). If that is the case, reliability, power to weight, and power to volume ratios of a real system will be significantly improved if both power production (fuel cell), and power storage (electrolysis) could be integrated in the same single unit, also simplifying system requirements considerably. This design will operate in a similar manner than a rechargeable battery. In any case, there are several factors to be considered in SOEC mode which are not identical as the ones on SOFC mode. These aspects will be further discussed in detail in the following sections, including steam management at the fuel electrode, higher oxygen partial pressures at the electrolyte/oxygen electrode interfaces, and higher current densities in general along the cells.

5.1 Electrochemical Impedance Spectroscopy Electrochemical impedance spectroscopy (EIS) is probably the most used technique to characterize energy devices. EIS is measured using a small excitation signal (AC potential) to an electrochemical cell and then measuring the current through the cell. By this excitation, the response of the cell is pseudo-linear, and the current response to a sinusoidal potential will be a sinusoid at the same frequency but shifted in phase. A detailed description of the technique can be found in the book from Barsoukov and McDonald (2005). There is an expression, analogous to Ohm’s law, which is used to calculate the impedance (Z) of the system:

Fundamentals of Solid Oxide Electrolysis Cells (SOEC)

Z=

sin(ωt) Vt V0 sin(ωt) = Z0 = It I0 sin(ωt + ϕ) sin(ωt + ϕ)

15

(25)

where V t and I t are the potential and the current at a certain time t, V 0 and I 0 are the amplitude and the current of the signal, ω is the radial frequency, t is the time, and ϕ is the phase shift. As reported by Lvovich (2008), impedance is described as the ratio between voltage and current, demonstrating the ability of a circuit to resist the flow of electrical current (real impedance term), but it also reflecting the ability to store electrical energy like a capacitance (imaginary impedance term). Impedance is then defined as a complex resistance composed of different resistors, capacitors and inductors. The impedance can be also represented as a complex number using Euler’s relationship, being a combination of real (Z R ) and imaginary (Z IM ) parts according with the following equation: Z (ω) = Z 0 exp( j ϕ) = Z 0 cos ϕ + j sin ϕ

(26)

It is worth mention that typical electrochemical systems are usually not stable with time; as a consequence, it is very important to assure that no evolution of the system occurs during the measurement, and also that the amplitude is low enough to consider a linear range in the current domain. According with Eq. (26), when the real part is plotted as a function of the imaginary part, a typical Nyquist plot is then obtained (Fig. 5 left). In this figure, the Y-axis is represented by negative values, and each point corresponds to the impedance of the system at different frequencies, where points on the left side correspond to higher frequencies and lower frequencies data are represented on the right side of the plot. Typical EIS measurements for SOFC/SOEC systems are recorded using a wide frequency range, typically ranging from about 100 kHz to 1 MHz. In addition, another usual way of representing EIS data is using a Bode plot (Fig. 5 right), where log frequency is plotted as a function of the absolute value of the impedance (|Z|=Z 0 ) and the phase shift (°). This technique is considered as a fundamental analysis when characterizing any type of SOEC cells or short stacks, as it allows direct characterization under real cell operation, as a consequence is considered an in situ characterization technique. It is also very versatile, as many parameters could be varied during measurements, including DC applied vias (both positive and negative), fuel rates and composition, pressure, temperature, etc. Each or these parameters could be affecting differently at each of the system component, and as the measurements are performed as a function of the frequency, it is possible to extract information of the different electrochemical processes taking place in the SOEC cell. In any case, each sample is unique, as several factors such as grain boundaries or impurities on the single phase materials, reactivity on composite cathodes as well as the interphase with the electrolyte, surface effects, microstructural differences including electrode thickness and porosity will have a direct impact on the electrochemical processes. In order to understand the different processes, performing measurements varying the aforementioned parameters is essential. But even performing these experiments, direct data analysis of the

16

M. A. Laguna-Bercero et al.

Fig. 5 Example of Nyquist (left) and Bode (right) plots of EIS results for a cobalt and titanium substituted SrFeO3 -based perovskite (LSFTC) symmetric cell tested under different fuel and temperature conditions (Song et al. 2020)

measurements is always a challenge, and additional fitting is generally performed after data acquisition using mathematical models. The most common analyses used for a better understanding are the use of equivalent circuit fitting (EC) (Mcdonald 1987) or the analysis by the distribution of relaxation times (DRT) (Yager 1936). Equivalent circuit fitting (EC) An EC is typically formed by ideal electrical elements such as resistors, accounting for the ohmic resistance or inductors, typically related with the electrochemical cell system. These elements appear in series and/or parallel with non-ideal electrical elements such as constant phase elements (CPE), which are accounting for the different electrochemical processes at the electrodes. CPEs are widely used to fit impedance diagrams in electrochemistry, and their shape is easily identified as a depressed semicircle describing a non-ideal capacitive behavior. The CPE impedance could be expressed as: ZC P E =

1 C( j ω)n

(27)

where C is the capacitance and n in an exponent which is equals to 1 for a real capacitor. Other elements generally applied for equivalent circuit fitting are the finite-length Warburg (FLW, associated with diffusion followed by a reaction process), Gerischer component (for coupled diffusion and reaction processes) or PET de Levie elements

Fundamentals of Solid Oxide Electrolysis Cells (SOEC)

17

(porous electrode theory describing the coupling between ionic conduction in an electrolyte and a reaction), which are used for MIECs, mixed ionic, and electronic conductors. Coming back to the description of the CPE, it can describe several behaviors depending on the value of the n exponent. As defined, when n = 1, the CPE corresponds to a pure capacitor. Another typical situation is in case of n = 0.5, where the CPE corresponds to a Warburg element accounting for a diffusion resistance caused by the current flow. Lvovich defined that the CPE can be associated to many different causes depending on the value of n, including electrode porosity and thickness, surface roughness, conductivity of the surface slow, inhomogeneous potential and current distribution, grain boundaries, impurities, etc. For a detailed analysis of CPE interpretation, the study of this work is highly recommended (Lvovich 2008). As a summary, by performing this deconvolution into an equivalent circuit, it is possible to distinguish the different contributions governing each electrochemical system. However, the use of this approach must be carried out with great attention since as the number of model parameters is increased, the quantity of mathematical solutions of the fit also increases considerably. Therefore, the most important thing to consider is to find an electrochemical sense for each system, generally making use of the existing literature. In this respect, a detailed review about the use of impedance spectroscopy applied for SOEC components can be found here (Nechache et al. 2014). DRT analysis on electrode reactions Apart from equivalent circuit fitting, there are other interesting methods for analyzing impedance data described in the literature, as for example the analysis of the difference in impedance spectra (ADIS) proposed by the group at the Technical University of Denmark (DTU) (Ebbesen et al. 2010; Jensen et al. 2007), or the distribution of relaxation time (DRT) analysis, which was firstly proposed by the group from Prof. Ivers-Tiffée in Karlsruhe (Schichlein et al. 2002). DRT analysis can then be used to obtain the parameters of polarization resistances in each of the processes of the Voigt elements circuit of an electrochemical cell. The theoretical aspects of the method of optimal regularization in the DRT are detailed in (Saccoccio et al. 2014) and (Ciucci and Chen 2015). Ciucci and coworkers also developed the DRTTools software for performing the DRT analysis from an impedance data plot (Wan et al. 2015), where test functions are used in discrepancy mode and cross validation of real and imaginary data to reproduce the approximate values in DRT as explained in the literature. The modeling of the Voigt circuit represents the behavior of the impedance through the equation: Z model (x, f ) = R∞ +

N

 n=1

+∞

1 φ (lnτ )dln τ 2 n −∞ 1 + (2π f τ )

xn ∫



18

M. A. Laguna-Bercero et al.

−i

N

 n=1

 2π f τ φ τ τ + ez ( f ) (ln )dln 2 n −∞ 1 + (2π f τ )

+∞

xn ∫

(28)

where R∞ is the ohmic resistance, φn = 1 is the phase angle in the ideal model, f is the frequency, ez ( f ) is the mathematical error based on the frequency, and τ is the relaxation time. The estimates of the variable xn are normally carried out by adjusting the expression of the model with the experimental data by means of the expression: N

2 2   1   1     Z exp ( f n ) − Z model (x, f n ) +  Z exp ( f n ) − Z model (x, f n ) S(x) = wn wn n=1

(29) 



where wn and wn are pondered diagonal matrices of ones to simplify calculations. The super indexes “and” indicate real and imaginary data, respectively. The x of the expansion can be obtained via regularized regression. In this case, the data obtained from the Nyquist and Bode representations with noise are adjusted by adding an additional term to the least square regression. It requires a minimization of the equation with the general form: N  [xn − h(x1 , x2 , . . . x N , f n )]2 + λP(x1 , x2 , . . . x N , f n )

(30)

n=1

for N data points in f n frequency samples and a regularization function h(x1 , x2 , . . . x N , f n ). The penalization parameter P(x1 , x2 , . . . x N ) is weighted employing a regularization parameter λ ≥ 0. The DRT analysis is typically combined with equivalent circuit fitting, as the one showed as an example in Fig. 6, where three R-CPE parallel elements are connected in series with a resistance and an inductance. For each circuit representation of Voigt elements as the one shown in the figure, a distribution of time constraints can be obtained. This can be interpreted as the response of the impedance spectra with an infinite set of small resistor circuits in parallel with capacitors connected in series. However, the DRT problem may not have a single solution, or it may be unstable in some results depending on experimental errors or numerical stalling errors. This method is complementary to the fitting of equivalent circuits and provides the polarization data present in each process. The DRT method is then used in combination with equivalent circuit fitting in order to get detailed information about the dynamic constants of an impedance plot, by determining the relaxation times and amplitudes of the different processes taking place in the impedance data. As relaxation times cannot be directly measured, a deconvolution method such as DRT must be performed after the measurement, assuring much detailed resolution, as observed in the example on Fig. 7. It is there clearly shown that DRT analysis facilitates the interpretation, revealing kinetic

Fundamentals of Solid Oxide Electrolysis Cells (SOEC)

19

Fig. 6 Example of an equivalent circuit diagram for the analysis of distribution of relaxation times on the EIS measurement of a SOEC cell

Fig. 7 Example of DRT versus EC comparison (Wang et al. 2021)

parameters of different underlying processes. This analysis is now widely used for SOFC/SOEC analysis, where many of the different processes are overlapped in frequencies, and usually, it is possible to differentiate both contributions from both fuel and oxygen electrode reactions.

5.2 Hydrogen Production Determination Hydrogen production rates are in general directly calculated from the I-V curves, taking into account the electric current density passing through the cell. It can be also calculated independently from the inlet and outlet dew point measurements, by analyzing the change mole fraction of steam, since one mole of consumed steam corresponds to one mole of produced hydrogen. A detailed explanation for this calculation can be found in the following reference (Independent Review Team 2017). In addition, a useful and common tool is the use of a gas chromatographer (GC) to analyze the outlet gas composition. In these experiments, it is very important to

20

M. A. Laguna-Bercero et al.

include a cooling water reflux system on the outlet gas prior the GC, in order to reduce the amount of steam of the reactant gases to a minimum, as steam will damage the chromatographer column. These analyses are typically performed at a fixed current density, assuring that the cell reached a stationary state, which are then confirmed by subsequent GC measurements. From these measurements, it is possible to estimate the faradaic efficiency of the process in order to conduct the energetic balance and then to identify the endothermic, thermoneutral, and exothermic operation regimes. GC analysis of the produced hydrogen in the output gas should reveal a good agreement with the necessary steam content required for each current density, also in concordance with the Faraday efficiency.

5.3 Chronoamperometry/Accelerated Degradation Tests As it will be described in detail in the subsequent sections, there are still several challenges for large-scale deployment of high-temperature electrolysis regarding durability, where the actual lifetime is about 20,000 h, and there is target for a commercial lifetime of at least 80,000 h for the year 2050, according with the key performance indicator (KPI) values from EERA (European Energy Research Alliance) (EERA 2020). In order to achieve these goals, a decrease in the degradation rates to less than 0.5%kh−1 (in voltage) will be required. It is worth mentioning that the current state of the art for SOEC operation is the rage of 1 to 10%kh−1 , but also under certain SOEC operation conditions, the target of 0.5%kh−1 can be achieved (Lang et al. 2020; Schefold et al. 2017; Irvine et al. 2016; Sohal et al. 2012). Since large durability studies are very time consuming, one goal to further the development of SOEC technology is to develop accelerated stress test (AST) protocols (Königshofer et al. 2021). The idea is to shorten different test programs aiming to accelerate certain degradation mechanisms while producing results comparable to those of ordinary long-term tests, with the goal of making qualitative predictions for the overall lifetime of SOECs. This is achieved mainly applying higher current densities and also by the introduction of impurities. However, there are very limited studies in this sense (Nechache et al. 2019; Suboti´c et al. 2021). The work of Königshofer et al. confirmed that operation at lower temperatures and higher voltages and operation at high steam conversion rates result in higher initial voltages and ASR degradation rates. On the contrary, operation at high steam partial pressures presented limited influence during the first 100 h of operation. They suggested that a few hundred operating hours may be sufficient to predict performance over longer periods using the proposed operating strategies. In any case, many additional experiments under different test conditions are needed in order to validate the accuracy of the prediction of long-term performance of SOECs.

Fundamentals of Solid Oxide Electrolysis Cells (SOEC)

21

6 Degradation Phenomena in Solid Oxide Electrolyzer Cells The stability and degradation rate of a SOEC play a vital role on its commercial application. Numerous factors such as cell fabrication, materials selection including both cell components and metallic interconnects, operation conditions (e.g., applied voltage or current density), working temperatures, gas compositions, stacking techniques, can deleteriously affect the stability (Chen et al. 2016; Laguna-Bercero 2012; Hauch et al. 2020; Moçoteguy and Brisse 2013; Wang et al. 2020). In this section, we will mainly cover one of the very intrinsic parameters, i.e., electronic/ionic conductivity and interface transfer parameters, without much discussion on phenomenological observations, process conditions, and cell or stack components, which will be discussed in more detail in subsequent chapters. The formation and validation of the Virkar’s Oxygen Electrode Delamination Theories (VOED theories) actually expand across about three decades and may be divided into three stages: the conceptualization stage (Virkar 1991, 2001), the finalization stage (Virkar 2005, 2012, 2007, 2010), and the validation and application stage (Virkar 2015; Zhu 2017). Although some of these earlier studies are not specifically dealing with SOEC or electrode delamination, important theoretical foundation was laid therein. We are not making new derivations and conclusions in this section beyond those reported by the VOED theories, but rather we aim to provide a more general and perhaps an easier interpretation. Some classical thermodynamics and electrochemical background were skipped in those references. As we consider readers of different backgrounds in this reference book, all these aspects will be covered in detail. Another highlight is that, although Virkar has made the notations clear in each of the aforementioned publications, the alternative use in the notations and current flow directions from one to another over the past years may be inconvenient for readers who are not familiar with these series of publications. For example, the oxygen electrode is assigned subscripts “c” (cathode) or the current for SOFC mode is defined negative (Virkar 2005, 2012), while the oxygen electrode may be assigned subscripts “a” (anode) or the current for SOFC mode is defined as positive elsewhere (Virkar 2010) due to the reversibility of SOEC. While not affecting the derivations and interpretation of the theories, we shall make some arbitrary and slightly redefining of the symbols which will not change the model but may help the readers to follow the mechanisms more easily. Important statements on notations i. Arbitrarily choosing oxygen electrode side as electrode/electrolyte interface II, and hydrogen electrode side as electrode/electrolyte interface I. To be distinguished with the interfaces, the properties far from Interface II and inside the oxygen electrode are denoted with symbol “P” which stands for positive terminal, and properties far from Interface I and inside the hydrogen electrode are denoted with symbol “N” which stands for negative terminal.

22

M. A. Laguna-Bercero et al.

ii. Set ionic current flow is negative for SOEC mode and is positive for SOFC mode. Note the partial electronic current is always negative for both SOEC and SOFC modes.

6.1 The Classical and VOED-Theories From classical thermodynamics, we can write dG = −SdT + V dp

(31)

where G, S, V, T, and p are the abbreviations for the Gibbs free energy, entropy, volume, temperature, and pressure, respectively. For an infinitely small change in a time dt, by considering the system irreversibility, Eq. (31) is further written as dG = −SdT + V dp +

n 

μi d Ni

(32)

1

where  μi =

∂G ∂ Ni

 (33) p, T,N j=i

is known as “chemical potential”. It is worth mentioning that μi reflects the extensive property of a system and is a state function; its absolute value cannot be determined unless a reference state is predefined. As Eqs. (31) and (32) are dealing with electrically neutral species and system, they are insufficient if electric field is added on top of that, therefore, we have the following expression dG = −SdT + V dp +

n 

μi d Ni + F

1

n 

z i d Ni

(34)

1

The —containing term in Eq. (34) essentially modifies the change of the system energy upon application of electric field, and then it is necessary to define a new term called “electrochemical potential” ∼

μi =μi + z i F

(35)

where z i is the valence of ion i, and F is the Faraday constant (the product of the ∼ electronic charge e and the Avogadro number NA). Note in some references, μi may be represented as ηi . If charge is counted on the basis of single electron, then (35) is given as

Fundamentals of Solid Oxide Electrolysis Cells (SOEC)

23

∼

μi =μi + z i e

(36)

Then Eq. (34) simplifies as dG = −SdT + V dp +

n 

∼

μi d Ni

(37)

1

The above equations are the basis for understanding an open system with considering both irreversibility and electric field. In most energy devices involving transport of ionic and electronic species, it is more convenient to represent system properties and states in term of current density or voltage. Under applied electric field, the general transport equation for a charged species is given as (Virkar 2005; Heyne 1968): Ii = −

σi ∼ ∇ μi zi F

(38)

where σi is the ionic conductivity. If we consider one-dimensional transport only, which is the case in the following discussion, then ∼

Ii = −

σi d μi zi F d x

(39)

It is important to recall the previous statements on defining of interfaces and symbols “P” and “N”. At the thin Interface I as shown in Fig. 8, IiI =

N μIO2 − μ O 2

4FriI

−

ϕI − ϕN riI

(40)

N where μIO2 and μ O are the oxygen chemical potential at the Interface I and the 2 hydrogen electrode, respectively; ϕ I and ϕ N are the electric potential at the Interface I and the hydrogen electrode, respectively; and riI is the area specific ionic charge transfer (polarization) resistance at the Interface I. μ I −μN

Note in the above equation, the O24F O2 part is actually the Nernst potential due to pO2 difference across interface I, which is opposing the applied voltage under electrolysis mode. As a consequence, the above equations may be rewritten as IiI =

E AI − E NI E NI E AI − = − riI riI riI

(41)

where E NI and E AI are the Nernst potential and applied voltage across the Interface I, respectively. Note that usually they cannot be separated from the whole cell during

24

M. A. Laguna-Bercero et al.

Fig. 8 Schematic illustration on the interface and electrode definitions

normal testing, the separation of these voltages here enables easier understanding of potential distribution one-dimensionally. Since E AI > E NI under electrolysis mode, IiI in the above equation is negative. At the same interface IeI = −

ϕI − ϕN reI

(42)

where IeI and reI are the electronic current density and the area specific electronic charge transfer (direct electron transfer) resistance at the Interface I, respectively. Since ϕ I > ϕ N under electrolysis mode, IeI is negative, at the thin Interface II we get the following expression, IiI I =

μ OP 2 − μ OI I2 4FriI I

−

ϕP − ϕI I riI I

(43)

where μIIO2 and μ OP 2 are the oxygen chemical potential at the Interface II and the oxygen electrode, respectively; ϕ I I and ϕ P are the electric potential at the Interface II and the oxygen electrode, respectively; and riI I is the area specific ionic charge transfer (polarization) resistance at the Interface II.

Fundamentals of Solid Oxide Electrolysis Cells (SOEC)

25

μ P −μ I I

O2 2 Similarly, the O4Fr part is the Nernst potential due to pO2 difference across II i interface II, the above equations may be rewritten as

IiI I =

E NI I E AI I − E NI I E AI I − = − riI I riI I riI I

(44)

where E NI I and E AI I are the Nernst potential and applied voltage across the Interface II, respectively. Since E AI I > E NI I under electrolysis mode, IiI I is negative as is expected. At the same interface IeI I = −

ϕP − ϕI I reI I

(45)

where IeI I and reI I are the electronic current density and the area specific electronic charge transfer (direct electron transfer) resistance at the Interface II, respectively. Since ϕ P > ϕ I I under electrolysis mode, IeI I is negative. Across the whole electrolyte layer excluding the two thin interfaces, similar equations can be given as Iiel =

μ OI I2 − μ OI 2 4Flρi

−

ϕI I − ϕI lρi

(46)

for ionic current density, where Iiel , ρi and l are the ionic current density, ionic resistivity, and thickness of the electrolyte layer, respectively. And Ieel = −

ϕI I − ϕI lρe

(47)

where Ieel and ρe are the electronic current density and thickness of the electrolyte layer, respectively. In steady state, IiI = IiI I = Iiel = Ii

(48)

IeI = IeI I = Ieel = Ie

(49)

and

where Ii and Ie are the ionic and electronic current densities through the whole membrane, respectively. Based on the total ionic and electronic area specific, resistance, Ii and Ie may be directly given as Ii =

N μ OP 2 − μ O 2

4F(riI + lρi + riI I )

−

ϕP − ϕN riI + lρi + riI I

(50)

26

M. A. Laguna-Bercero et al.

and Ie = −

reI

ϕP − ϕN + lρe + reI I

(51)

The above two equations may be further given as Ii =

EN EA EA − EN − =− Ri Ri Ri

(52)

and Ie = −

EA Re

(53)

From Eqs. (40) and (52), N μIO2 − μ O 2

4FriI

−

ϕI − ϕN EA − EN =− Ri riI

(54)

Combining Eqs. (42) and (49) gives the following expression: N μIO2 − μ O 2

4FriI

+

Ie reI EA − EN =− I Ri ri

(55)

−

E A reI EA − EN =− Ri Re riI

(56)

Substitute Ie by Eq. (53) gives N μIO2 − μ O 2

4FriI

After substituting a portion of the local interface parameters such as ϕ I and IiI , which are difficult to be measured by globally measurable parameters such as E A , E N , Ii and Ie through Eqs. (54) to (56), finally μ OI 2 at interface I just inside the electrode and near the hydrogen electrode can be written as

μ OI 2

=

N μO 2

(E A − E N )riI E A reI − + 4F Re Ri

(57)

Alternatively, μ OI 2

=

N μO 2

(E A − E N )riI I N + 4F (ϕ − ϕ ) − Ri

(58)

And similarly, for μ OI I2 at interface II just inside the electrode and near the oxygen electrode can be written as

Fundamentals of Solid Oxide Electrolysis Cells (SOEC)



μ OI I2

=

μ OP 2

(E A − E N )riI I E A reI I − 4F − Re Ri

27

(59)

Alternatively,

(E A − E N )riI I μ OI I2 = μ OP 2 − 4F (ϕ P − ϕ I I ) − Ri

(60)

The above four equations replicate the same results reached by Eqs. (59) and (60) (Virkar 2010), which are the fundamental basis for SOEC degradation being the root cause of oxygen electrode delamination in the Virkar’s theory. Now, we shall expand slightly further on the above equations and discuss more on what conclusions may be reached depending on the variation of the interface parameters and operating conditions. From classical chemical potential theory, we can write μO2 = μ0O2 + RT ln pO2

(61)

Combining Eqs. (57) to (61), the corresponding oxygen partial pressure inside de electrolyte at the Interface I and just near the hydrogen electrode, can be given by

p OI 2 = p ON2 exp

4F RT



(E A − E N )riI E A reI − Re Ri

 (62)

and the corresponding oxygen partial pressure inside the electrolyte at the Interface II and just near the oxygen electrode, can be given by

p OI I2 = p OP 2 exp

−4F RT



(E A − E N )riI I E A reI I − Re Ri

 (63)

The very last equation is the most important part of the VOED theories, and depending on the values of the parameters and the sign inside the curly brackets, p OI I2 can be magnitudes higher than p OP 2 . The resulting consequences are discussed in detail in the VOED theories, and are schematically shown here in Fig. 9 as a “hybrid” format. The term “hybrid” here refers to the different length scale for the interface region and the normal electrolyte and electrode layers, so that the transition of the oxygen pressure change may be seen more intuitively. The readers are encouraged to refer to the work of Virkar for more information and calculated examples (Virkar 2010).

28

M. A. Laguna-Bercero et al.

Fig. 9 Schematic variation of oxygen chemical potential through a SOEC under certain conditions where pO2 at Interface II may well exceed the pO2 in the oxygen electrode leading to delamination. The model here assumes “ideal” electrode for simplification, i.e., no electrode polarization resistance; therefore, oxygen chemical potential is flat in the two electrode regions. Note the schematic is in a “hybrid” format due to different length scale for the interface region and the normal electrolyte and electrode layers

6.2 Experimental Observations and Evidences on the VOED Theories Oxygen electrode delamination phenomena have been evidenced in many studies. Graves et al. have observed delamination and morphological changes in the YSZ electrolyte under high current densities for CO2 electrolysis (Graves et al. 2011b), as shown in Fig. 10. The authors suggested that high current density operation of the cells resulted in a build-up of oxygen pressure at (or in the electrolyte close to) the interface between the electrolyte and the oxygen electrode. This observation is exactly the same as the conclusions reached by the VOED theories. Note that in this study, the micro-cracks are all perpendicular to the electrolyte/electrode interface and inside the electrolyte layer. Kim et al. have investigated SOEC degradation mechanism via YSZ supported symmetrical cells where LSM and LSM/YSZ were used as functional layer and current collection layer, respectively (Kim et al. 2013). As shown in Fig. 11a, the cell voltage gradually increased with time under an anodic current density of −1.5 A cm−2 at 750 °C, followed by a sharp rise of the voltage and cell failure after 120 h operation. The EIS data indicate both the ohmic and polarization resistances increased substantially with increasing operating time (Fig. 11b). Postmortem analysis revealed intergranular fracture along YSZ grains and just close to the electrode/air electrode

Fundamentals of Solid Oxide Electrolysis Cells (SOEC)

29

Fig. 10 Scanning electron micrographs of a cell that was long-term tested under electrolysis at high current density: a The lanthanum strontium manganite (LSM)/YSZ electrode active layer and the entire electrolyte. b Closer view of the electrolyte damage near the interface with the LSM/YSZ electrode. Reproduced with permission from (Graves et al. 2011b). Copyright Infor

interface, along with LSM precipitates within the fractured grain surfaces. Complete delamination of the oxygen electrode from the electrolyte was observed (Fig. 11d) and was considered to occur at the moment of cell failure with the drastic increase of the cell voltage. The authors compared their observation with Knibbe et al.’s work where evidences of intergranular fracture in the YSZ layer propagating along the grain boundaries parallel to the LSM/YSZ interface were also observed (Knibbe et al. 2010). In Knibbe et al.’s study, they proposed that the high current density allowed for oxygen to form in the grain boundaries, and the formation rate depended on the nucleation and growth process in the YSZ grain boundaries. Mawdsley et al. examined oxygen electrode delamination mechanisms via various techniques in stacks operated for 1000 and 2000 h, where clear oxygen electrode delamination was observed near the oxygen electrode/electrode interface (Fig. 12a) (Mawdsley et al. 2009). The authors proposed two possible reasons for the observed delamination. The first is related to oversintering of the electrode or bond layer which could cause enough pressure build-up to cleave the oxygen electrode grains and cause delamination and the second considers “high oxygen flux at the electrode–electrolyte interface which may lead to a mismatch between the greater ability of zirconia to

30

M. A. Laguna-Bercero et al.

Fig. 11 Studies on degradation mechanism of electrode and air electrode in SOEC: a Cell voltage (LSM-YSZ anode potential vs. Pt reference electrode) as a function of time with the anodic current passage of -1.5 A cm−2 at 750 °C in air, b a series of EIS data obtained at time interval of 5–20 h during the operation, c SEM image of the interface area between LSM-YSZ electrode and YSZ electrolyte, showing extensive interdiffusion between two phases and intergranular fracture along YSZ grain boundaries after anodic current passage (−1.5 A cm−2 ) for 120 h at 750 °C in air, d SEM image of the LSM-YSZ electrode which delaminated from the YSZ electrolyte after anodic current passage (−1.5 A cm−2 ) for 120 h at 750 °C in air, e intergranular fracture running along grain boundary under 2.0 A cm−2 electrolysis current density. (a to d) Reproduced with permission from (Kim et al. 2013). e Reproduced with permission from (Knibbe et al. 2010).

Fundamentals of Solid Oxide Electrolysis Cells (SOEC)

31

release oxygen and the lesser ability of the electrode material to conduct oxide ions away. In this case, pressure would build up and oxygen could be released to the defect at the interface after electron is transferred to the electrode, and the defect would grow into a crack that would eventually cleave the interface.” While both mechanisms refer to oxygen pressure accumulation, the second one is more related to oxygen pressure build-up due to oxygen ionic and electron flow and can be well explained by the VOED theories. Laguna-Bercero et al. have also observed similar oxygen electrode delamination evidences (Laguna-Bercero et al. 2011). Cracks perpendicular to current flow direction formed inside the YSZ electrolyte and close to the oxygen electrode/electrolyte interface (Fig. 12b). The morphology of the cracks is very similar to that reported by Graves et al. (2011b) and Knibbe et al. (2010). Unlike the previous studies, this study also measured oxygen elemental distribution after cell degradation, as shown in Fig. 12c. The average oxygen content is about 3% higher near the oxygen electrode/electrolyte interface in the degraded sample than in the blank sample, and

Fig. 12 a SEM micrographs of a polished cross section of a cell from the 2000-h stack (near the oxygen exit, steam inlet corner), the area where the oxygen electrode delaminated, b SEM micrographs showing cracking of the YSZ electrolyte, c oxygen atomic content along the YSZ electrolyte measured by EDS. As standard reference, oxygen content of a reference cell is also plotted in the graph, and d Raman spectra along the YSZ electrolyte. Raman spectrum of a blank sample is also shown for comparison. The inset shows a SEM micrograph for the analyzed region. a Reproduced with permission from Mawdsley et al. (2009). (b to d) Reproduced with permission from Laguna-Bercero et al. (2011).

32

M. A. Laguna-Bercero et al.

meanwhile, the oxygen content is about 1% lower near the fuel electrode/electrolyte interface in the degraded sample than in the blank sample. Both observations are well explainable by the VOED theories. The author also found that the band widening occurred at 620 cm−1 by Raman measurements may be related to the increase of oxygen observed by energy dispersive X-ray spectroscopy(EDS) analysis (Fig. 12c). Zhu et al. have also observed increased oxygen content at the positive (oxygen) electrode/electrolyte and decreased oxygen content in the negative (fuel) electrode/electrolyte interface in thick electrolyte supported cells after applying constant current for about 200 h (Zhu et al. 2017). All those experimental findings validated the aforementioned VOED theories and explained one of the degradation issues on SOEC.

References Barelli B, Bidini G, Cinti G (2017) Airflow management in solid oxide electrolyzer (SOE) operation: performance analysis. ChemEngineering 1(2):13 Barsoukov E, MacDonald JR (2005) In: Impedance spectroscopy, Wiley, Hoboken, New Jersey Chen K, Jiang SP (2016) Review—materials degradation of solid oxide electrolysis cells. J Electrochem Soc 163(11):F3070–F3083 Ciucci F, Chen C (2015) Analysis of electrochemical impedance spectroscopy data using the distribution of relaxation times: a Bayesian and hierarchical approach. Electrochim Acta 167:439–454 Ebbesen SD, Graves C, Hauch A, Jensen SH, Mogensen M (2010) Poisoning of solid oxide electrolysis cells by impurities. J Electrochem Soc 157:B1419–B1429 EERA (2020) Key performance indicators (KPIs) for FCH research and innovation, 2020–2030 Fang Q, Blum L, Stolten D (2019) Electrochemical performance and degradation analysis of an SOFC short stack following operation of more than 100,000 hours. 166(16):F1320–F1325 Graves C, Ebbesen SD, Mogensen M, Lackner KS (2011a) Sustainable hydrocarbon fuels by recycling CO2 and H2 O with renewable or nuclear energy. Renew Sustain Energy Rev 15(1):1–23 Graves C, Ebbesen SD, Mogensen M (2011b) Co-electrolysis of CO2 and H2 O in solid oxide cells: performance and durability. Solid State Ionics 192(1):398–403 Harrison KW, Remick R, Hoskin A, Martin G (2010) Hydrogen production: fundamentals and case study summaries. (National Renewable Energy Lab (NREL), Golden, CO (United States)) Hauch A et al (2020) Recent advances in solid oxide cell technology for electrolysis. Science 370(6513):6118 Heyne L (1968) Ionic conductivity in oxides. In: Mass transport in oxides. NBS Special Publications 296:149–164 Independent Review Team (2017) Measurement of hydrogen production rate based on dew point temperatures. U. S. Department of Energy Hydrogen Program, NREL/MP-150–42237 Irvine JTS, Neagu D, Verbraeken MC, Chatzichristodoulou C, Graves C, Mogensen MB (2016) Evolution of the electrochemical interface in high-temperature fuel cells and electrolysers. Nat Energy 1:15014 Jensen SH, Hauch A, Hendriksen PV, Mogensen M, Bonanos N, Jacobsen T (2007) A method to separate process contributions in impedance spectra by variation of test conditions. J Electrochem Soc 154:B1325–B1330 Kim J et al (2013) Degradation mechanism of electrolyte and air electrode in solid oxide electrolysis cells operating at high polarization. Int J Hydrogen Energy 38(3):1225–1235

Fundamentals of Solid Oxide Electrolysis Cells (SOEC)

33

Knibbe R et al (2010) Solid oxide electrolysis cells: degradation at high current densities. J Electrochem Soc 157(8):B1209 Königshofer B, Pongratz G, Nusev G, Boškoski P, Höber M, Juriˇci´c D, Kusnezoff M, Trofimenko N, Schröttner H, Hochenauer C, Suboti´c V (2021) Development of test protocols for solid oxide electrolysis cells operated under accelerated degradation conditions. J Power Sour 497:230982 Küngas R (2020) Review—electrochemical CO2 reduction for CO production: comparison of lowand high-temperature electrolysis technologies. J Electrochem Soc 167(4):044508 Laguna-Bercero MA et al. (2011) Electrolyte degradation in anode supported microtubular yttria stabilized zirconia-based solid oxide steam electrolysis cells at high voltages of operation. J Power Sour 196(21):8942—7 Laguna-Bercero MA (2012) Recent advances in high temperature electrolysis using solid oxide fuel cells: a review. J Power Sour 203:4–16 Lang M, Raab S, Lemcke MS, Bohn C, Pysik M (2020) Long-term behavior of a solid oxide electrolyzer (SOEC) stack. Fuel Cells 20(6):690–700 Larminie J, Dicks A, McDonald MS (2018) In: Fuel cell systems explained, J. Wiley Chichester, UK Leonide A, Sonn V, Weber A, Ivers-Tiffée E (2008) Evaluation and modeling of the cell resistance in anode-supported solid oxide fuel cells. J Electrochem Soc 155:B36–B41 Lvovich VF (2008) Impedance spectroscopy. J. Wiley & Sons, Hoboken, New Jersey Mawdsley JR et al (2009) Post-test evaluation of oxygen electrodes from solid oxide electrolysis stacks. Int J Hydrogen Energy 34(9):4198–4207 Mcdonald JR (1987) Impedance spectroscopy and its use in analyzing the steady-state AC response of solid and liquid electrolytes. J Electroanal Chem 223(1–2):25–50 Moçoteguy P, Brisse A (2013) A review and comprehensive analysis of degradation mechanisms of solid oxide electrolysis cells. Int J Hydrogen Energy 38(36):15887–15902 Nechache A, Cassir M, Ringuedé A (2014) Solid oxide electrolysis cell analysis by means of electrochemical impedance spectroscopy: a review. J Power Sour 258:164–181 Nechache A, Boukamp BA, Cassir M, Ringuedé A (2019) Accelerated degradation of yttria stabilized zirconia electrolyte during high-temperature water electrolysis. J Solid State Electrochem 23(3):871–881 Saccoccio M et al (2014) Optimal regularization in distribution of relaxation times applied to electrochemical impedance spectroscopy: ridge and lasso regression methods—a theoretical and experimental study. Electrochim Acta 147:470–482 Schefold J, Brisse A, Poepke H (2017) 23,000 h steam electrolysis with an electrolyte supported solid oxide cell. Int J Hydrogen Energy 42(19):13415–13426 Schichlein H, Müller AC, Voigts M, Krügel A, Ivers-Tiffée E (2002) Deconvolution of electrochemical impedance spectra for the identification of electrode reaction mechanisms in solid oxide fuel cells. J Appl Electrochem 32(8):875–882 Sohal MS, O’Brien JE, Stoots CM, Sharma VI, Yildiz B, Virkar A (2012) Degradation issues in solid oxide cells during high temperature electrolysis. J Fuel Cell Sci Technol 9:011017 Song J, Zhu T, Chen X, Ni W, Zhong Q (2020) Cobalt and Titanium substituted SrFeO3 based perovskite as efficient symmetrical electrode for solid oxide fuel cell. J Materiomics 6(2):377– 384 Suboti´c V, Futamura S, Harrington GF, Matsuda J, Natsukoshi K, Sasaki K (2021) Towards understanding of oxygen electrode processes during solid oxide electrolysis operation to improve simultaneous fuel and oxygen generation. J Power Sour 492:229600 Virkar AV (1991) Theoretical analysis of solid oxide fuel cells with two-layer, composite electrolytes: electrolyte stability. J Electrochem Soc 138(5):1481–1487 Virkar AV (2001) Transport of H2 , O2 and H2 O through single-phase, two-phase and multi-phase mixed proton, oxygen ion, and electron hole conductors. Solid State Ionics 140(3):275–283 Virkar AV (2005) Theoretical analysis of the role of interfaces in transport through oxygen ion and electron conducting membranes. J Power Sour 147(1–2):8–31

34

M. A. Laguna-Bercero et al.

Virkar AV (2007) A model for solid oxide fuel cell (SOFC) stack degradation. ECS Trans 7(1):443– 454 Virkar AV (2012) Transport through mixed proton, oxygen ion and electron/hole conductors: analysis of fuel cells and electrolyzer cells using Onsager equations. Int J Hydrogen Energy 37(17):12609–12628 Virkar AV (2010) Mechanism of oxygen electrode delamination in solid oxide electrolyzer cells. Int J Hydrogen Energy 35(18):9527–9543 Virkar AV, Tao G (2015) Reversible high temperature cells for power generation and hydrogen production using mixed ionic electronic conducting solid electrolytes. Int J Hydrogen Energy 40(16):5561–5577 Vøllestad E et al (2019) Mixed proton and electron conducting double perovskite anodes for stable and efficient tubular proton ceramic electrolysers. Nat Mater 18(7):752–759 Wan TH, Saccoccio M, Chen C, Ciucci F (2015) Influence of the discretization methods on the distribution of relaxation times deconvolution: implementing radial basis functions with DRTtools. Electrochim Acta 184:483–499 Wang Y, Leung DYC, Xuan J, Wang H (2017) A review on unitized regenerative fuel cell technologies, part B: unitized regenerative alkaline fuel cell, solid oxide fuel cell, and microfluidic fuel cell. Renew Sustain Energy Rev 75:775–795 Wang Y et al (2020) Degradation of solid oxide electrolysis cells: phenomena, mechanisms, and emerging mitigation strategies—a review. J Mater Sci Technol 55:35–55 Wang Q, Hu Z, Xu L, Li J, Gan Q, Du X, Ouyang M (2021) A comparative study of equivalent circuit model and distribution of relaxation times for fuel cell impedance diagnosis. Int J Energy Res 45:15948–15961 Yager WA (1936) The distribution of relaxation times in typical dielectrics. J Appl Phys 7(12):434– 450 Zhu L, Zhang L, Virkar AV (2017) Role of electronic conduction in stability of solid oxide electrolyzer cells (SOEC). ECS Trans 80(9):81–89

Solid-State Electrolytes for Solid Oxide Electrolysis Cells Sivaprakash Sengodan

1 Introduction Solid oxide electrolysis cells (SOECs) are made up of three main components: two porous electrode layers (anode and cathode) and one sandwiched electrolyte layer. The electrolyte employed in SOEC might be an oxygen ionic conductor or a proton conductor; Fig. 1 depicts cell architectures based on the different electrolyte materials mentioned above. The electrode reactions for each cell arrangement are heavily influenced by the electrolyte materials used. The electrolyte should ideally be a pure ionic conductor and an electronic insulator and a dense separator for the oxidative and reductive gases and half-reactions. The electrolyte must meet a set of parameters relating to the multilayer structure of SOECs as well as economic considerations: (Minh et al. 1993). (1) High ionic conductivity (0.01 S cm−1 at operating temperature) (Etsell and Flengas 1970) with an ionic transport number close to or equal to one; (2) Chemical compatibility with electrode materials; (3) Chemical stability in both oxidising and reducing atmospheres for long exposure times; (4) Processing by ceramic methods compatible with mass production of large cell areas; and (5) Thermo-mechanical stability during thermal or redox cycling. To date, only a few ionic conducting materials have been truly applied, despite the fact that numerous oxide ionic conducting materials, such as Bi2 O3 -based materials (Takahashi et al. 1977), Y2 O3 /Sc2 O3 doped zirconia (Mori et al. 2008), doped ceria (Mogensen et al. 2000b), Ca12 Al14 O33 (Lacerda et al. 1988), doped LaGaO3 S. Sengodan (B) Department of Materials, Imperial College, London, UK e-mail: [email protected]; [email protected] Department of Mechanical Engineering, Khalifa University, Abu Dhabi, United Arab Emirates © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M. A. Laguna-Bercero (ed.), High Temperature Electrolysis, Lecture Notes in Energy 95, https://doi.org/10.1007/978-3-031-22508-6_3

35

36

S. Sengodan

Fig. 1 Schematic illustration of cell configurations of a solid oxide electrolysis cell with different ionic conducting electrolyte materials: a oxygen ionic conducting electrolyte and b proton-conducting electrolyte

perovskite-type oxide (Ishihara et al. 1994), Ln10 (SiO4 )6 O3 (Nakayama et al. 1995), and La2 Mo2 O9 (Lacorre et al. 2000), based fast oxide ion conductors have been developed, proposed, and extensively studied for both solid oxide fuel cell (SOFC) and SOEC applications over the last decades. Even though oxide ion conductors are now frequently employed as the electrolytes in SOECs, high-temperature proton conductors are also gaining attention as the electrolytes in SOECs (Choi et al. 2019). The fact that protons are among the lowest ionic sizes of all the elements means that they have both excellent mobility and low activation energy for conduction. Because of this, if we merely compare ionic conductivity in low-temperature areas, proton-conducting materials demonstrate much better ionic conductivity in intermediate-temperature regions and are thus suited as the electrolyte for SOECs operating in low/intermediate-temperature zones. Oxygen ionic conductors made of stabilised zirconia, acceptor-doped ceria, doped lanthanum gallate, and proton conductors made of barium cerate and barium zirconate are common choices in academic and industrial electrolytic cell deployment. Because of its excellent ionic conductivity at high temperatures (up to 700 °C) and wide oxygen partial pressure range, yttrium-stabilised zirconia (YSZ) is a stateof-the-art material. On the other hand, for low-temperature SOEC operation, the ionic conductivity YSZ is not sufficient. For low/intermediate-temperature SOEC operation, other electrolyte materials, such as doped ceria and perovskite doped LaGaO3 oxide, are proposed. However, they always suffer inherent disadvantages, such as the reduction-induced internal short circuit in doped ceria and high reactivity with the commonly used electrode materials in the doped LaGaO3 electrolyte. In fact, doped ceria is not suitable for practical SOEC operation. Although the use of ceria led to lowering both hydrogen electrode and oxygen electrode overvoltages compared to YSZ-based cells, the high applied voltages lead to the reduction of Ce4+ to Ce3+ and deteriorating the ionic transference number. In addition, the recently produced rare-earth ortho-niobates and tantalates (Haugsrud and Norby 2006b), Sr1−x Nax SiO3−0.5x (Singh and Goodenough 2013), as well as Na0.5 Bi0.5 TiO3 (Li et al. 2014), require more and in-depth research before they can be extensively employed in SOEC electrolytes.

Solid-State Electrolytes for Solid Oxide Electrolysis Cells

37

Global research and industry initiatives are underway to decrease the SOEC operational temperature from a high-temperature range (1000–850 °C) to an intermediate temperature (750–500 °C). The lower temperature provides for longer material and system lifespans and lower investment, which are two critical issues for productmarket penetration. The ohmic resistance of the electrolyte should be decreased when the operating temperature is dropped from 1000–800 to 750–500 °C. Two distinct strategies are employed. One is the design of novel electrolyte materials with high ionic conductivity. The second is the lowering of the electrolyte layer thickness from the present millimetre level to the micrometre or even nanoscale level because the ohmic resistance is proportional to the electrolyte layer thickness according to ohmic laws. Specific considerations of the different electrolytes when operating under SOEC mode were already described in the fundamentals chapter. This section will be mainly focused on describing the general aspect of each type of electrolyte, mainly discussing aspects related with their conductivity, effect of type, and amount of dopants and processing conditions.

2 Oxygen Ion-Conducting Electrolyte Stabilised Zirconia At room temperature, pure zirconia adopts a monoclinic crystal form. At elevated temperatures, it undergoes two phase transitions: tetragonal and cubic at 1170 and 2370 °C, respectively (Scott 1977). The phase transitions are linked with a significant volume shift, which results in mechanical damage to the ceramic material. Adding alkaline or rare-earth oxides stabilises the cubic phase, preventing phase transitions and further preventing mechanical damaging. Zirconium oxide is categorised primarily as an electrical insulator. To introduce oxide ion conductivity in ZrO2 , it is necessary to provide an oxygen vacancy. Hence, ZrO2 is often substituted with a lower valence cation such as yttrium and scandium. Oxygen vacancies are formed when Y is doped with ZrO2 , as stated by Eq. 1 (Goodenough 2000). '

Y2 O3 → Y Zr + 3O O× + VO··

(1)

In this case, the high-temperature stable phase of the cubic structure is stabilised down to room temperature by substituting the lattice position with a lower valence cation. Thus, oxygen-vacancy-introduced ZrO2 is referred to as stabilised ZrO2 , which includes the tetragonal (partially stabilised) and cubic (fully stabilised) phases. Yashima et al. (1996) conducted a thorough study of the phase diagrams of doped zirconia, including ZrO2 –Y2 O3 (YSZ) (Heraeus 1899) and ZrO2 –Sc2 O3 (ScSZ), (Yashima et al. 1996) as well as the temperature dependency of their electrical conductivity (Yamamoto. 2000). At 1000 °C, YSZ has a conductivity of 0.14 S cm−1 , and ScSZ has a higher conductivity than YSZ; at 780 °C, ScSZ has the

38

S. Sengodan

same conductivity (0.14 S cm−1 ) as YSZ. Additionally, the ionic conductivity of stabilised zirconia is shown to be significantly reliant on the dopant ionic radius and dopant concentration. The ionic conductivity of substituted zirconia changes with substitutional concentration and delivers a maximum ionic conductivity at a particular concentration (Claussen et al. 1983). The relationship is shown in Fig. 2. The electrical conductivity of ZrO2 is obviously influenced by the dopant element and its concentration. Conductivity increases monotonically with increasing dopant amount in the small doping amount. Several researchers have attempted to explain ionic conductivity in doped zirconia, mainly in terms of a size mismatch between the aliovalent substitutional ion and the host cation. In the ZrO2 -M2 O3 system, the relationship between the substitutional concentration with the maximum conductivity at 1000 °C and the substitutional ionic radius is shown in Fig. 3 (Arachi et al. 1999). It is clear that ionic conductivity increases with doping concentration at first, reaches a maximum value at a given doping concentration, and then falls as the dopant concentrations increase. At maximum conductivity, is observed for Sc3+ doping, which has the closest ion radius to the host ion, Zr4+ , has the highest conductivity and the highest substitutional content. Even though the YSZ with 8 mol % Y2 O3 has the highest possible ionic conductivity, the ionic conductivity diminishes when it is used for extended periods of time (Lina et al. 2019). After 5000 min, the ionic conductivity of YSZ doped with 7.7 mol Fig. 2 Oxide ion conductivity of stabilised ZrO2 at 1080 °C as a function of the dopant amount (Claussen et al. 1983)

Solid-State Electrolytes for Solid Oxide Electrolysis Cells

39

Fig. 3 Composition dependence of the electrical conductivity at 1000 °C for ZrO2 -Ln2 O3 (Ln = Sc, Yb, Er, Y, Dy, Gd, and Eu) (Arachi et al. 1999)

% Y2 O3 drops from 16 to 13.7 S cm−1 . However, if the doping concentration is raised very slightly, this issue can be avoided (9 mol% or 10 mol% Y2 O3 ). Ionic conduction of ScSZ degrades more severely than YSZ does due to phase transformation (Yamamoto et al. 1995). As can be seen in Fig. 4, after 60,000 min, the resistivity of 9ScSZ exhibits an increase that is greater than a 50% increase from its initial value (Haering et al. 2005). In a similar vein, the ageing effect of ScSZ can be mitigated by increasing the amount of Sc2 O3 present in the material, or by co-doping it with Al2 O3 or TiO2 , both of which have the ability to inhibit the phase transformation of ScSZ. Another advantage of co-doping Al2 O3 is that it can increase the mechanical strength of the material. This is possible due to the fact that Al prevents grain formation during the sintering process. The oxygen ion conductivity, on the other hand, drops as the amount of Al2 O3 or TiO2 in the material increases. The influence of temperature on the oxygen ion conductivity of 9YSZ, 9YbSZ, and 9ScSZ is illustrated in Fig. 5 (Strickler and Carlson 1964; Ni et al. 2008) The link between temperature and conductivity is described using an Arrhenius form as follows:   E (2) σ = A exp − kT where A is a pre-exponential factor, E is the activation energy, k is the Boltzmann constant, and T is the absolute temperature. Within the temperature range that is of importance, it is clear that the oxygen ion conductivity declines in the following order: ScSZ > YbSZ > YSZ. At lower temperatures, the differences between them become more pronounced, which suggests that ScSZ has the potential to be utilised

40

S. Sengodan

Fig. 4 Variation of 9ScSZ resistivity with time (Haering et al. 2005)

Fig. 5 Temperature effect on oxygen ionic conductivity of 9YSZ, 9YbSZ, and 9ScSZ-experimental results from Ref. (Strickler and Carlson 1964)

as an electrolyte operating at an intermediate temperature, provided that its price is able to be significantly decreased. When it comes to determining its oxygen ion conduction, the grain boundaries of the doped ZrO2 are also quite essential (Guo and Waser 2006). When heated to a high temperature, the grain boundary resistance of YSZ and ScSZ is rendered essentially meaningless. The grain boundary resistance becomes relevant at a lower temperature (< 800 °C for ScSZ), and it is the grain boundary resistance that determines the oxygen ion conduction in the electrolyte. Based on this discovery, it appears that the ionic conductivity of ScSZ at an intermediate temperature can be increased by lowering grain boundary resistance in order to achieve a higher conductivity. In a separate piece of research, Kosacki et al. found that the thickness of the electrolyte was an important factor in determining the oxygen ion conductivity (Kosacki et al.

Solid-State Electrolytes for Solid Oxide Electrolysis Cells

41

2005, 2002). In the range of film thicknesses between 60 and 2000 nm, the conductivity of YSZ electrolyte diminishes as the thickness increases. When the YSZ film thickness is less than 60 nm, the conductivity of the material improves as the film thickness decreases. The transition from lattice controlled diffusivity to interface controlled diffusivity can be seen in these occurrences. Research has been conducted to investigate the effect of co-doping with another oxide. Densification of the YSZ ceramics was significantly aided by the introduction of ZnO into 8YSZ (Liu and Lao 2006). It is possible to raise the ionic conductivity of 8YSZ by adding a trace amount of ZnO as a dopant. The conductivity of 8YSZ at 800 °C increases from 0.0131 to 0.0289 S cm−1 when doped with 0.5 weight per cent ZnO. When Al2 O3 is added to YSZ, space charge regions are made, which can improve the way ions move through the material (Liu and Lao 2006). On the other hand, Al2 O3 also has a blocking effect that slows decrease in ionic conductivity. Al2 O3 doping has little effect on the conductivity of YSZ because of all of these things. It was discovered that the addition of Bi2 O3 might stabilise the cubic phase of ScSZ at lower temperatures. With 2 mol% of Bi2 O3 , the conductivity of ScSZ was measured to be 0.18 S cm−1 at 600 °C (Liu and Lao 2006). According to the research that were discussed above, co-doping may be an efficient method for increasing either the mechanical strength or the ionic conductivity of a YSZ electrolyte material. Table 1 is a summary of some of the experimental data that was collected on the ionic conductivity of doped ZrO2 . The mechanical stability of the electrolyte is most crucial for electrolyte-supported planar cell structure, while all other cell designs rely on a separate layer for mechanical support. However, in the case of electrolyte-supported planar cell structures, there is a trade-off between the mechanical robustness of the cell, which increases with electrolyte thickness, and the ohmic resistance of the cell, which decreases with electrolyte thickness. This is because the mechanical robustness of the cell depends on the electrolyte thickness. In addition, lower dopant concentrations frequently result in improved mechanical qualities, although this improvement frequently comes at the expense of conductivity. For instance, the mechanical properties of (Y2 O3 )0.03 (ZrO2 )0.97 (3YSZ) are superior to those of (Y2 O3 )0.08 (ZrO2 )0.92 (8YSZ), despite the fact that (3YSZ) has a significantly lower ionic conductivity. Therefore, the minimum thickness of the electrolyte membrane is determined by the mechanical stability of the material. The mechanical properties of ScSZ are quite good and comparable to those of YSZ. Because of its strong oxide ion conductivity, phase stability, and outstanding mechanical properties, ZrO2 with 11 mol% Sc2 O3 and 1 wt% Al2 O3 looks to be one of the finest options for an intermediate-temperature electrolysis cells.

3 Doped Ceria Doped ceria has been proposed as a potential alternative electrolyte for SOFCs operating at intermediate/low temperatures (500–650 °C). Mogensen et al. (2000a) and

42

S. Sengodan

Table 1 Ionic conductivity of different doped zirconia materials Material composition

Conductivity (S cm−1 )

Temperature (°C)

Remarks

References

9.5YSZ

0.057

900

15−25 μm thick film prepared by magnetic pulse compaction of tapes cast of nanopowders

(Ni et al. 2008)

8YSZ

0.083

900

Spark plasma sintering

(Dahl et al. 2007)

MgO–ZrO2 with 13.7 mol% MgO

0.098

1000

Produced by the traditional steps of ceramic manufacturing, which include wet mixing, pressing, sintering, and machining

(Muccillo and Kleitz 1995)

Sc2 O3- ZrO2 with 9-11 mol% Sc2 O3

0.28–0.34

1000

Co-precipitation powders

(Badwal et al. 2000)

Sc2 O3 -ZrO2 with 6 mol% Sc2 O3

0.18

1000

Treated by hot isostatic pressing to improve the mechanical strength

(Hirano et al. 2000)

10.5YSZ

0.034

800

Prepared by (Jiang et al. aerosol-assisted 2007) metal–organic chemical vapour deposition

8YSZ

0.13

100

Pechini process

(Prabhakaran et al. 2007)

Steel et al. (2000) have presented detailed reviews of the electrical conductivity and conduction mechanism in ceria-based electrolytes. It has gained increased attention in recent years as a result of the extensive research and development of SOEC for low-temperature applications, owing to its superior ionic conductivity, favourable chemical compatibility with both conventional and newly developed electrode materials, and self-catalytic activity towards hydrogen oxidation and oxygen reduction reactions. (Mogensen et al. 2000a; Tuller and Nowick 1975; Singman 1966; Andersson et al. 2006; Eguchi et al. 1992). As a result, ceria is probably the most promising electrolyte for SOFCs operating at intermediate/low temperatures. Ceria is a fluorite structured material similar to stabilised zirconia. By substituting trivalent rare-earth ions for Ce4+ , mobile oxygen vacancies are introduced in a similar way to ZrO2 . The conductivity of doped-ceria systems is determined by the type of dopant and its

Solid-State Electrolytes for Solid Oxide Electrolysis Cells

43

Fig. 6 Dependence of ionic conductivity for (CeO2 )0.8 (LnO1.5 )0.2 at 800 °C on the radius of dopant cation (Yahiro et al. 1989; Butler et al. 1983)

concentration, as is the case with stabilised ZrO2 . As shown in Fig. 6, the doping element ionic radius has also had a significant impact on the ionic conductivity of ceria (Yahiro et al. 1989). As illustrated in this Fig. 6 (dotted line), the binding energy calculated by Butler et al. (1983) has a strong relationship with ionic conductivity; clearly, the dopant with low binding energy has high conductivity. Due to its stronger ionic conductivity and lower binding energy, ceria is generally doped with Gd or Sm to form gadolinium-doped ceria (GDC) or samarium-doped ceria (SDC) than other dopants such as Dy, Ca, Sr, La, and Nd. CeO2 –Gd2 O3 and CeO2 -Sm2 O3 have ionic conductivities of 6.3 × 10–2 S cm−1 at 750 °C, which is much higher than stabilised zirconia electrolyte (Kudo and Obayashi 1975). As a result, these systems are very attractive for electrolytes in intermediate/low-temperature SOECs and have been extensively studied. There has been a lot of research into the chemical reduction of doped-ceria electrolytes. At high oxygen partial pressures, ceria-based oxide ion conductors have pure ionic conductivity, and when oxygen partial pressures are low, the materials undergo partial reduction due to reduction of Ce4+ to Ce3+ , which promotes electronic conductivity in the electrolyte (Gödickemeier and Gauckler 1998). In any case, as previously mentioned, doped ceria is not suitable for practical SOEC operation. Although the use of ceria led to lowering both hydrogen electrode and oxygen electrode overvoltages compared to YSZ-based cells, the high applied voltages lead to the reduction of Ce4+ to Ce3+ and deteriorating the ionic transference number (Eguchi et al. 1996). Another issue with doped ceria is the chemical expansion of the electrolyte, which is a function of the oxygen activity μO2 inside the electrolyte and therefore inhomogeneous across the electrolyte thickness, as explained in the fundamentals section. Lenser et al. developed a one-dimensional model that enables the calculation of elastic stresses in the cell on the basis of the cell geometry and materials properties (Lenser et al. 2022). Ceria is subjected to a decreased oxygen activity at the interface to the fuel electrode during electrolysis operation, leading to substantial mechanical stresses in the electrolyte layers. They found that cell failure is observed under high electrolysis currents for a high-performance cell based on a doped-ceria electrolyte. They also modelled the electro-chemo-mechanicallyinduced elastic stresses in the multilayer system of the cell and show that it peaks

44

S. Sengodan

for operation at 750 °C and high current densities, in agreement with the observed onset of cell failure. As explained in the fundamentals chapter, based on the theoretical predictions from Virkar on simple equivalent circuits for SOEC and SOFC, it was concluded that during both modes of operation, the electronic current flowing through the cell was significant (Virkar and Tao 2015). They also found that it is not necessary that the electrolyte be a purely ionic conductor (transference number close to 1) to realise efficient operation in both the SOFC and the SOEC modes. In fact, they observed that cells with some electronic conduction exhibited more stable operation than cells made with purely ionic conducting electrolyte cells. Specifically, the incidence of degradation of cells by oxygen electrode delamination is substantially reduced when using mixed ionic and electronic conductor (MIEC) electrolyte cells. According to these findings, it was shown that the electronic conductivity in predominantly ionic conductors such as YSZ is important concerning the thermodynamics of the electrolyte under transport, mainly the spatial distribution of oxygen chemical potential inside the electrolyte. As a consequence, if the electronic conductivity is very low (such as in YSZ) and electrode polarisations are large, also large variations can occur in the oxygen chemical potential at the oxygen electrode/electrolyte interface, leading to electrode delamination, electrolyte cracking, voids formation, and local electrolyte decomposition, as widely reported in the literature (see fundamentals section). It is then concluded that if the electronic conductivity is relatively high, being the ionic conductivity still dominant, could be beneficial for electrolysis operation, thus avoiding electrolyte degradation. For example, Virkar and Tao (2015) reported that cells made with 8YSZ electrolyte doped with 8 mol.% CeO2 electrolytes (8CYSZ) electrolyte had an ionic transference number of ~ 0.9 and could be operated stably. Many other electrolytes and dopants are possible for the development of stable cells for electrolysis operation. Similar findings were concluded by Chatzichristodoulou et al. (2016). In this sense, several bilayer electrolytes for SOEC applications were proposed, most of them based on YSZ/ceria (Shen and Ni 2015; Li 2022, Kim et al. 2021, Heidari et al. 2017).

4 Doped Lanthanum Gallates Oxygen ion conductivity can be seen in a variety of perovskite materials. The perovskite-structured lanthanum gallate doped with strontium and magnesium (La0.9 Sr0.1 Ga0.8 Mg0.2 O2.85 , commonly referred to as LSGM), is one of the most promising oxides, was first reported in 1994 by Ishihara et al. (1994) and Goodenough et al. (1994). Ishihara et al. (1994) conducted the preliminary research on the LaGaO3 based electrolyte. They discovered that doping Sr for the La sites may increase the conductivity of LaGaO3 (Fig. 7a), and that doping Mg for the Ga sites could further increase the conductivity (Fig. 7b). It was discovered that La0.9 Sr0.1 Ga0.8 Mg0.2 O3 had a conductivity that was greater than that of traditional YSZ and ScSZ. For the composition La0.9 Sr0.1 Ga0.8 Mg0.2 O3 , this material has a pure ionic conductivity of

Solid-State Electrolytes for Solid Oxide Electrolysis Cells

45

Fig. 7 Oxygen ionic conductivity of doped LaGaO3 -a Sr doped LaGaO3 and b Sr and Mg co-doped LaGaO3 Ishihara et al. (1994)

about 0.14 S cm−1 at 800 °C, throughout a wide range of oxygen partial pressures (10–20 < pO2 < 1), with values higher than stabilised zirconia and doped ceria even at low temperatures. La3+ takes up the centre of the cube in perovskite oxide crystal structure, while Ga3+ takes up eight vertices, and oxygen ions take up the centres of each edge. The oxygen vacancy is created by doping Sr and Mg into the crystal phase. The relatively open structure facilitates oxygen vacancy diffusion within the lattice structure, resulting in an oxygen ion-conducting path. The optimised composition of LSGM oxide ion conductivity is slightly higher than stabilised zirconia and very comparable to ceria-based materials. Unlike doped ceria, LSGM has significantly better electrolytic stability in reducing atmospheres and can remain an electronic insulator over a broader range of operating conditions. Table 2 is a summary of some conductivity data that has been published in the past. At both high and intermediate temperatures, it is clear that LSGM has a high ionic conductivity. At 800 °C, LSGM can have a conductivity of about 0.17 S cm−1 . YSZ, on the other hand, only has a conductivity of about 0.026 S cm−1 (Cong et al. 2003). At 600 °C, the difference is more dramatic, with 0.03 S cm−1 for LSGM and 0.00173 S cm−1 for YSZ. According to this statistics, LSGM has the potential to be an effective electrolyte for steam electrolysis performed at intermediate temperatures. In addition, the conductivity of LSGM can change depending on the method of its synthesis. This is because the grain size and grain boundaries can be changed in different synthesis methods. Conductivity of LSGM, like that of YSZ, is proportional to the amount of dopants present in the material. The relationship between the doping content of Sr and Mg and the conductivity of the electrolyte is illustrated in Fig. 8 (Cong et al. 2003). It can be observed that the conductivity is highly sensitive to the doping content, and the doping content that results in the best performance is approximately 15 mol%. It is important to note that the optimal doping concentration that has been reported

46

S. Sengodan

Table 2 Ionic conductivity of lanthanum gallate doped with strontium and magnesium oxide materials Material composition

Conductivity Temperature Remarks (S cm−1 ) (°C)

References

La0.8 Sr0.2 Ga0.8 Mg0.2 O3-δ

0.17

800

Prepared by a combustion method

(Stevenson et al. 1998)

La0.8 Sr0.2 Ga0.83 Mg0.17 O2.815 0.17

800

Prepared by solid-state reactions method

(Huang and Goodenough 2000)

La0.9 Sr0.1 Ga0.8 Mg0.2 O3-δ

0.1193

800

Prepared by solid-state reactions method

(Zhang et al. 2000)

La0.8 Sr0.2 Ga0.85 Mg0.15 O3-δ

0.0606

800

Prepared by (Cong et al. 2003) glycine-nitrate combustion method

La0.9 Sr0.1 Ga0.9 Mg0.1 O2.9

0.051

700

Prepared by citrate sol–gel method

La0.85 Sr0.15 Ga0.85 Mg0.15 O2.8 0.015

600

Prepared by (Liu et al. 2006) acrylamide polymerisation technique

(Polini et al. 2004)

by multiple groups is not the same (Shi et al. 2006; Gorelov et al. 2001). This is because these groups use a variety of different starting materials and apply a variety of different preparation processes. In the course of the preparation process, this can result in the production of a secondary phase. In most cases, a doping concentration of between 15 and 20 mol% of Sr and Mg is considered to be appropriate. The amount of doping and the formation of secondary phase was studied by Zheng et al. (2004) No secondary phases were observed to develop with Sr and Mg doping concentrations below 20 mol%. However, the conductivity of LSGM electrolyte was reduced when the doping concentrations above 20 mol% due to the formation of the secondary phase of SrLaGaO4 . The secondary phase, LaSrGa3 O7 , developed in another work investigated by Liu et al., with doping levels of Sr and Mg of 15 mol%, resulted in relatively poor conductivity (Zheng et al. 2004). Co-doping with other metal ions has been studied in an effort to further improve the conductivity of LSGM or its other features, such as its mechanical strength and thermal expansion characteristics. Stevenson et al. studied at the impact of doping LSGM electrolyte with Co or Fe (Stevenson et al. 2000). It was observed that by doping LSGM with a little quantity (< 10 mol%) of Co or Fe at the Ga site, the electrical conductivity may be improved at low temperatures with no change in conductivity at high temperatures. While the addition of electronic charge carriers to

Solid-State Electrolytes for Solid Oxide Electrolysis Cells

47

Fig. 8 Relationship between the doping content of Sr and Mg and the oxygen ionic conductivity of LSGM (Cong et al. 2003)

the lattice increased conductivity at low temperatures, this had no favourable effect on the conductivity of oxygen ions. The conductivity of the electrolyte could shift to electronic conduction if Co or Fe were added. Khorkounov et al. studied the ionic and electronic conductivity of co-doped LSGM electrolyte (Khorkounov et al. 2006). Both the ionic and electronic conductivities were found to increase upon codoping. Although oxygen ion conductivity improved, the enhancement in electronic conductivity was noticeably larger. Besides the benefits listed above, there are a few drawbacks to LSGM electrolytes. (1) A high temperature is normally required for the sintering process, which results in a small amount of Ga oxide evaporation. This significantly affects the ionic conductivity (Matraszek et al. 2004) (2) The chemical compatibility of LSGM with common electrode materials is poor. During high-temperature processing, nickel oxide, which is a precursor to metallic nickel used in the steam electrode, forms an impurity phase. Therefore, effective diffusion prevention is required when fabricating LSGM electrolyte film using conventional fabrication methods. Huang et al. demonstrated that Ce0.6 La0.4 O2 (LDC) serves as an effective barrier against Ni diffusion into LSGM (Huang and Goodenough 2000)

5 Proton-Conducting Electrolyte The use of proton-conducting oxides as electrolyte materials has the potential to provide a viable solution to the problems associated with conventional SOECs

48

S. Sengodan

(Stevenson et al. 2000). Due to their better ionic conductivity properties, protonconducting solid oxide electrolysis cells (PCEC) are able to work at lower temperatures, such as 500 °C. When compared to SOEC systems based on oxygen-ion electrolytes, the SOEC system that uses proton-conducting electrolytes exhibits a number of distinct advantages. First, only pure and dry hydrogen is generated on the hydrogen electrode side of PCEC, and no additional gas separation is required. Second, due to the absence of steam in the hydrogen electrode, Ni-based fuel electrode materials will not undergo oxidation.

5.1 BaCeO3 -BaZrO3 Mixed Systems Iwahara et al. (Hibino et al. 1993; Iwahara et al. 1988; Hibino et al. 1992) demonstrated proton conduction in SrCeO3 , BaCeO3 , and other perovskite-related oxides at relatively high temperatures in a humidified hydrogen atmosphere in the early 1980s. At a moderate temperature, proton conduction is based on the presence of proton defects and dissociative absorption of water, which takes place in the presence of oxygen vacancies. Protonic defects are formed when water is dissociated into a proton and a hydroxide ion; the proton forms a covalent bond with the lattice oxygen, while the hydroxyl ion fills the oxygen vacancy, based on the following Eq. 2. H2 O + O O× + VO·· ↔ 2OH··O

(3)

Proton diffusion in perovskite-type oxides is widely accepted to have two basic steps (Geneste 2018). As shown in Fig. 9, in a reorientation step (Step 1), the OH group is bent towards the adjacent oxygen ion, lowering the energetic barrier for proton transfer and allowing protons to be transferred via hydrogen bond breaking/formation. In Step 2, the proton migrates by proton hopping between neighbouring oxide ions due to the presence of protonic defects. Protons can hop between oxygen atoms by breaking oxygen-hydrogen bonds and forming new oxygen-hydrogen bonds with neighbouring oxygen atoms (Kreuer 1996). Among all the high-temperature proton conductors, doped barium cerate perovskite-type oxides demonstrated the highest proton conductivity. However, they are highly reactive with water vapour and acidic gases such as CO2 and SO2 , forming hydroxides and carbonates, respectively (Gopalan and Virkar 1993; Bhide and Virkar 1999; Magrasó et al. 2011). Because H-SOECs operate in harsh conditions, barium base electrolyte degrades, limiting its use. Dopants have been introduced to stabilise the barium cerate, which should not affect the high proton conductivity. Matsumoto et al. established a correlation between ionic conductivity and chemical stability by doping cations of various sizes (Matsumoto et al. 2007) They discovered that substituting larger ionic radius dopants for Ce4+ in barium cerate improves chemical stability. On the other hand, doped barium-zirconates exhibit excellent chemical stability in both water vapour and CO2 (Sun et al. 2011). The main disadvantage of BaZrO3 -based proton-conducting electrolytes is their poor sinterability, making it difficult to make

Solid-State Electrolytes for Solid Oxide Electrolysis Cells

49

Fig. 9 Schematic illustration of proton conduction in perovskite oxide. Step 1: the reorientation and Step 2 proton transfer pathways (Kim et al. 2019)

dense membranes and their high resistance to proton conduction at grain boundaries. Several studies have focused on the development of chemically stable and highly conductive electrolytes through the preparation of solid solutions of BaZrO3 and BaCeO3 (Kreuer 1997). It is possible to prepare a chemically stable electrolyte with the desired proton conductivity by controlling the fraction of Zr in BaZrO3 with Ce due to the mutual soluble nature of both BaZrO3 and BaCeO3 . This approach resulted in designing a proton-conducting oxide BaCe0.9–x Zrx Y0.1 O3 (BZCY) with increased chemical stability against CO2 and H2 O under experimental conditions when the Zr content is greater than 0.4 (LÜ et al. 2008). Liu et al. reported Yb, Y, and Ce triple doped BaZrO3 (BaZr0.1 Ce0.7 Y0 Yb0.1 O3 (BZCYYb)) proton conductors with the ionic conductivity of higher than those of BZCY, GDC, and YSZ below 700 °C (Lei et al. 2009). Chemical stability and ionic conductivity have been reported to improved significantly for BaZr0.4 Ce0.4 Y0.1 Yb0.1 O3 (BZCYYb4411) fabricated with a Zr:Ce ratio of 1:1, composition compared to previous works that used a 1:7 composition. In a 100% CO2 atmosphere, the BZCYYb4411 demonstrates excellent chemical stability of the BZCYYb electrolyte (Choi et al. 2018).

5.2 Ba2 In2 O5 -Based Materials Over the past two decades, it has been evident that perovskite-like compounds with structural oxygen vacancies are also capable of realising proton transport. Materials based on brownmillerite are another type of perovskite-related material. These materials have oxygen vacancies and have the formula A2 B2 O5 or A2 BB' O5 . In contrast to simple perovskites, wherein vacancies are formed by acceptor doping (impurity oxygen vacancies), these compounds contain oxygen vacancies that are structural components (structural oxygen vacancies). Along the (101) axis, the oxygen vacancies in the brownmillerite structured materials are arranged in alternating layers. This structure accommodates the greatest number of possible oxygen vacancies: 16.7%

50

S. Sengodan

oxygen vacancies for each formula unit, in contrast to the content of oxygen vacancies in simple doped perovskites, which is often less than 4% (referred to as the total oxygen amount). Ba2 In2 O5 is one of the brownmillerite materials that has been studied the most. It was first thought to be a good anionic conductor, with a conductivity equal to or even higher than zirconia at temperatures above 925 °C (Rolle et al. 2004). A few years ago, scientists also investigated at brownmillerite-based material as protonic conductor when they were exposed to a wet atmosphere near 300 °C. At high temperatures, the good conductivity of oxides is noticed, along with a phase change from an ordered structure of oxygen vacancies to a disordered tetragonal and cubic phase. At low temperatures, the reaction between Ba2 In2 O5 and H2 O produces Ba2 In2 O5• H2 O. The hydrated phase begins to dehydrate around about 400 °C. As the temperature rises above this point, the brownmillerite material transforms into an anionic conductor. The hydrated phase Ba2 In2 O5· H2 O shows reasonable proton conductivity of 10–5 S cm−1 at 400 °C (Zhang and Smyth 1995b, 1995a). The main issue with Ba2 In2 O5 applications is the occurrence of the phase transition about 930 °C. In the literature, a number of different strategies have been proposed and tested in an effort to stabilise the quadratic and cubic forms of barium indate at lower temperatures. When heated to temperatures higher than 930 °C, the ionic conductivity of Ba2 In2 O5 is comparable to that of YSZ. However, when the temperature is lowered below the transition temperature, the conductivity drops considerably (by 1–1.5 orders of magnitude). The primary objective of the doping strategy is to ensure that the cubic phase is maintained at low temperatures as possible. To improve the ionic and protonic conductivities of oxides, partial cationic substitution with Ba and/or In was carried out. Several investigations have shown that, the conductivity of the YSZ is almost identical to that of the cerium-doped Ba2 In2 O5 , which has the chemical formula Ba2 In1.75 Ce0.25 O5.1 . Doping Ba2 In2 O5 with ceria causes the order-to-disorder transition to be eliminated, which ultimately results in the structure becoming more stable at lower temperatures. Therefore, it was not possible to see a jump at 900 °C; however, Ce-doped Ba2 In2 O5 results in lower conductivity for high temperatures. An increase in the proton conduction was seen when Ba was replaced by La in (Ba1-x Lax )In2 O5+x , and this increase was proportional to the amount of La present. At a 400 °C, the ideal conductivity is found to be equal to 1.12 × 10–5 S cm−1 when x = 0.1. It is important to keep in mind that systems based on Ba2 In2 O5 will break down in strongly reducing atmospheres, particularly when temperatures are high (Zhang and Smyth 1995a). Research needs to be done on both their short-term and long-term stability in hydrogen atmospheres for this reason.

5.3 LaNbO4 -Based Materials Norby et al. first reported a relatively new class of proton conductors that are based on LaNbO4 . These materials are the representatives of a relatively new class of proton conductors, (Haugsrud and Norby 2006b, 2006a). Analyses of LaNbO4 show

Solid-State Electrolytes for Solid Oxide Electrolysis Cells

51

that it goes through a phase change from a monoclinic fergusonite-type structure at low temperatures to a tetragonal scheelite-type structure at high temperatures (Norby and Magrasó 2015). This change in phase transition has a big effect on both the unit cell volume and the linear parameters. The low-temperature LaNbO4 phase can have a thermal expansion that is 1.2–2.5 times greater than the hightemperature phase (Fig. 10) (Skilbred and Haugsrud 2012). The stability of the LaNbO4 tetragonal phase at low temperatures is a challenging problem to solve due to the fact that such a system has only a limited range of solid-state solutions that can exist. In most cases, the quantity of acceptor dopants that causes the production of oxygen defects as well as the appearance of oxygen ion and proton conductivities is less than 1–5 mol % (Mokkelbost et al. 2009). For higher concentrations, impurity phases (La1.56 Ca0.96 Nb1.48 O7 , Zn2 Nb2 O7 and LaNbTiO6 ) or other niobates (La3 NbO7 , LaNb3 O9 ) are formed (Mokkelbost et al. 2009; Syvertsen et al. 2012; Cao et al. 2015; Huse et al. 2012) It is possible to generate oxygen vacancies in the basic structure by partially substituting La3+ with alkaline-earth elements such as Ca2+ , Nb5+ , Ti4+ , and Al3+ . This will allow for the creation of oxygen vacancies. At the same time, this kind of doping changes the ionic and electronic properties of the material (Fig. 11) (Huse et al. 2012). The ionic and electronic conductivities of nominally pure LaNbO4 are mixed in both oxidising and reducing environments. When La3+ is replaced with Ca2+ or Nb5+ is replaced with Ti4+ , there is a significant increase in the ionic conductivity, but there is a drop in the electronic n-type component under reducing conditions. It is noteworthy to note that a joint replacement has a detrimental effect on the electrical properties of LaNbO4 , including a decrease in the ionic conductivity and an increase in the electronic conductivity (Huse et al. 2012). Recently, it was demonstrated that a substituted-LaNbO4 –based oxide, known as LaNb0.84 W0.16 O4.08 , could serve as an alternate electrolyte for solid oxide cells. This was accomplished by adopting a superstructure that leads to interstitial ionconducting channels inside the material (Bayliss 2013). Electrolysis measurements at single cell level were performed for this electrolyte, obtaining comparable results with an YSZ-based cell (Laguna-Bercero 2014). In La1-x Ax NbO4 , partial substitution of La3+ for lower valence cations like Ca2+ resulted in excellent protonic conductivity (Magrasó et al. 2010). The oxygen excess is incorporated into the structure by doping the B-site with a W6+ cation, mimicking the structural behaviour of CeNbO4+δ superstructures. The W-doped LaNb0.84 W0.16 O4.08 material has a negligible electronic conductivity at a pO2 as low as 10–22 atm and an ionic conductivity of roughly 0.1 S cm−1 at 1000 °C, which is comparable to that of 8 mol% YSZ (Bayliss 2013).

6 Concluding Remarks Since the 1960s, many oxide formulations have been studied extensively for SOEC electrolyte materials, and a significant amount of effort has been devoted to understanding the ionic conducting mechanism. There will be an ongoing search for new materials; however, the number of potential candidates will remain limited, making

52

S. Sengodan

Fig. 10 Thermal expansion behaviour of Sandvik Sanergy HT compared to LaNbO4 , a candidate electrolyte material for protonic ceramic solid oxide fuel cells (PCFCs) (Skilbred and Haugsrud 2012)

the development of existing materials essential. Zirconia-based compositions are excellent electrolytes due to their low electronic conductivity and acceptable oxide ion conductivity above 800 °C. SOECs are now being developed to operate at lower temperatures, and the lowest operating temperature of SOECs is 700 °C based on thin YSZ with acceptable conductivity and mechanical properties. If the costs of scandia are acceptable, then scandia-doped zirconia may be preferred at temperatures below 700 °C. There have also been developments and proposals for alternative electrolyte materials, such as perovskite (La, Sr)(Ga, Mg)O3 oxides. However, they still have their own problems, such as the high chemical reactivity with other components of the SOECs and instability under reducing conditions. Proton-conducting oxides like BZCY or BZCYYb may also be effective SOEC electrolytes, but the stability/reactivity of many of these materials with H2 O/CO2 must be resolved. Proton-conducting solid oxide electrolysis cells are still in an early stage of development. So far, the largest known proton-conducting solid oxide electrolysis cells that have been reported in the literature are tubular proton-conducting solid oxide electrolysis cells with an air electrode area of 10 cm2 . In order to demonstrate the advantages of proton-conducting solid oxide electrolysis cells in future, it will be necessary to scale up the footprint of the cells and demonstrate multicell stacks. The

Solid-State Electrolytes for Solid Oxide Electrolysis Cells

53

Fig. 11 Oxygen partial pressure dependencies of the total conductivity of LaNbO4 -based materials at 800 and 100 °C (Huse et al. 2012)

electrochemical performance of all the electrolytes described in this chapter will be presented and discussed in detail in subsequent sections.

References Ai N (2006) Preparation of Sm0.2 Ce0.8 O1.9 membranes on porous substrates by a slurry spin coating method and its application in IT-SOFC. J Membr Sci 286:255–259 Andersson DA (2006) Optimization of ionic conductivity in doped ceria. In: Proceedings of the national academy of sciences of the United States of America, vol 103. pp 3518 LP–3521 Arachi Y (1999) Electrical conductivity of the ZrO2 -Ln2 O3 (Ln=lanthanides) system. Solid State Ionics 121:133–139 Badwal SPS, Ciacchi FT, Milosevic D (2000) Scandia–zirconia electrolytes for intermediate temperature solid oxide fuel cell operation. Solid State Ionics 136:91–99 Banerjee S (2007) Enhanced ionic conductivity in Ce0.8 Sm0.2 O1.9 : unique effect of calcium Codoping. Adv Func Mater 17:2847–2854 Bayliss RD (2013) Synthetic, structural and electrochemical studies of superstructured rare-earth niobates. Imperial College London Bhide SV, Virkar AV (1999) Stability of BaCeO3 -based proton conductors in water-containing atmospheres. J Electrochem Soc 146:2038–2044 Bi Z (2004) A high-performance anode-supported SOFC with LDC-LSGM bilayer electrolytes. Electrochem Solid-State Lett 7:A105 Butler V (1983) Dopant ion radius and ionic conductivity in cerium dioxide. Solid State Ionics 8:109–113

54

S. Sengodan

Cao Y (2015) Electrical conductivity of Zn-doped high temperature proton conductor LaNbO4 . Solid State Ionics 278:152–156 Chatzichristodoulou C, Chen M, Hendriksen PV, Jacobsen T, Mogensen MB (2016) Understanding degradation of solid oxide electrolysis cells through modeling of electrochemical potential profiles. Electrochim Acta 189:265–282 Choi S (2018) Exceptional power density and stability at intermediate temperatures in protonic ceramic fuel cells. Nat Energy 3:1–9 Choi S, Davenport TC, Haile SM (2019) Protonic ceramic electrochemical cells for hydrogen production and electricity generation: exceptional reversibility, stability, and demonstrated faradaic efficiency. Energy Environ Sci 12:206–215 Claussen N, Ruhle M, Heuer AH (1983) Science and technology of zirconia II. American Ceramic Society Inc., United States Cong L (2003) Synthesis and characterization of IT-electrolyte with perovskite structure La0.8 Sr0.2 Ga0.85 Mg0.15 O3-δ by glycine-nitrate combustion method. J Alloy Compd 348:325– 331 Dahl P (2007) Densification and properties of zirconia prepared by three different sintering techniques. Ceram Int 33:1603–1610 Dutta A, Kumar A, Basu RN (2009) Enhanced electrical conductivity in Ce0.79 Gd0.20 Co0.01 O2-δ for low temperature solid oxide fuel cell applications. Electrochem Commun 11:699–701 Eguchi K (1992) Electrical properties of ceria-based oxides and their application to solid oxide fuel cells. Solid State Ionics 52:165–172 Eguchi K, Hatagishi T, Arai H (1996) Solid State Ionics 86–88:1245–1249 Etsell TH, Flengas SN (1970) Electrical properties of solid oxide electrolytes. Chem Rev 70:339– 376 Feng M, Goodenough JB (1994) A superior oxide-ion electrolyte. Eur J Solid State Inorg Chem 31:663–672 Gao L (2010) Effect of zinc oxide on yttria doped ceria. J Power Sour 195:3130–3134 Geneste G (2018) Proton transfer in barium zirconate: lattice reorganization Landau-zener curvecrossing approach. Solid State Ionics 323:172–202 Gödickemeier M, Gauckler LJ (1998) Engineering of solid oxide fuel cells with ceria-based electrolytes. J Electrochem Soc 145:414–421 Goodenough JB (2000) Oxide-ion conductors by design. Nature 404:821–823 Gopalan S, Virkar AV (1993) Thermodynamic stabilities of SrCeO3 and BaCeO3 using a molten salt method and galvanic cells. J Electrochem Soc 140:1060–1065 Gorelov VP (2001) The effect of doping and processing conditions on properties of La1-x Srx Ga1-y MgyO3-α . J Eur Ceram Soc 21:2311–2317 Guo X, Waser R (2006) Electrical properties of the grain boundaries of oxygen ion conductors: acceptor-doped zirconia and ceria. Prog Mater Sci 51:151–210 Haering C (2005) Degradation of the electrical conductivity in stabilised zirconia system: part II: Scandia-stabilised zirconia. Solid State Ionics 176:261–268 Haugsrud R, Norby T (2006a) High-temperature proton conductivity in acceptor-doped LaNbO4 . Solid State Ionics 177:1129–1135 Haugsrud R, Norby T (2006b) Proton conduction in rare-earth ortho-niobates and ortho-tantalates. Nat Mater 5:193–196 Heidari D, Javadpour S, Chan SH (2017) An evaluation of electrochemical performance of a solid oxide electrolyzer cell as a function of co-sintered YSZ-SDC bilayer electrolyte thickness. Energy Convers Manage 150:567–573 Heraeus WC (1899) Über die elektrolytische Leitung fester Körper bei sehr hohen Temperaturen. Zeitschrift Für Elektrochemie 6:41–43 Hibino T (1992) Evaluation of proton conductivity in SrCeO3 , BaCeO3 , CaZrO3 and SrZrO3 by temperature programmed desorption method. Solid State Ionics 57:303–306 Hibino T, Mizutani K, Iwahara H (1993) H/D isotope effect on electrochemical pumps of hydrogen and water vapor using a proton-conductive solid electrolyte. J Electrochem Soc 140:2588–2592

Solid-State Electrolytes for Solid Oxide Electrolysis Cells

55

Hirano M (2000) Mechanical and electrical properties of Sc2 O3 -doped zirconia ceramics improved by postsintering with HIP. Solid State Ionics 133:1–9 Huang K, Goodenough JB (2000) A solid oxide fuel cell based on Sr- and Mg-doped LaGaO3 electrolyte: the role of a rare-earth oxide buffer. J Alloy Compd 303:454–464 Huse M, Norby T, Haugsrud R (2012) Effects of A and B site acceptor doping on hydration and proton mobility of LaNbO4 . Int J Hydrogen Energy 37:8004–8016 Ishihara T, Matsuda H, Takita Y (1994) Doped LaGaO3 perovskite type oxide as a new oxide ionic conductor. J Am Chem Soc 116:3801–3803 Iwahara H (1988) Proton conduction in sintered oxides based on BaCeO3 . J Electrochem Soc 135:529–533 Jiang Y (2007) Fabrication and characterization of Y2 O3 stabilized ZrO2 films deposited with aerosol-assisted MOCVD. Solid State Ionics 177:3405–3410 Kharton VV, Marques FMB, Atkinson A (2004) Transport properties of solid oxide electrolyte ceramics: a brief review. Solid State Ionics 174:135–149 Khorkounov BA, Näfe H, Aldinger F (2006) Relationship between the ionic and electronic partial conductivities of co-doped LSGM ceramics from oxygen partial pressure dependence of the total conductivity. J Solid State Electrochem 10:479–487 Kim J (2019) Proton conducting oxides: a review of materials and applications for renewable energy conversion and storage. Renew Sustain Energy Rev 109:606–618 Kim J (2021) Naturally diffused sintering aid for highly conductive bilayer electrolytes in solid oxide cells. Sci Adv 7:40 Kosacki I (2002) Nonstoichiometry and electrical transport in Sc-doped zirconia. Solid State Ionics 152:431–438 Kosacki I (2005) Nanoscale effects on the ionic conductivity in highly textured YSZ thin films. Solid State Ionics 176:1319–1326 Kreuer K-D (1996) Proton conductivity: materials and applications. Chem Mater 8:610–641 Kreuer KD (1997) On the development of proton conducting materials for technological applications. Solid State Ionics 97:1–15 Kudo T, Obayashi H (1975) Oxygen ion conduction of the fluorite-type Ce1-x Lnx O2-x/2 ( Ln = Lanthanoid Element ). J Electrochem Soc 122:142–147 Lacerda M (1988) High oxide ion conductivity in Ca12 Al14 O33 . Nature 332:525–526 Lacorre P (2000) Designing fast oxide-ion conductors based on La2 Mo2 O9 . Nature 404:856–858 Laguna-Bercero MA, Bayliss RD, Skinner SJ (2014) LaNb0.84W0.16O4.08 as a novel electrolyte for high temperature fuel cell and solid oxide electrolysis applications. Solid State Ionics 262:298–302 Le S (2013) Densification of Sm0.2 Ce0.8 O1.9 with the addition of lithium oxide as sintering aid. J Power Sour 222:367–372 Lei Y (2009) Enhanced sulfur and coking tolerance of a mixed ion conductor for SOFCs: BaZr0.1 Ce0.7 Y0.2-x Ybx O3-δ . Science 326:126–129 Lenser C, Zhang J, Russner N, Weber A, Guillon O, Menzler NH (2022) Electro-chemo-mechanical analysis of a solid oxide cell based on doped ceria. J Power Sour 541:231505 Li M (2014) A family of oxide ion conductors based on the ferroelectric perovskite Na0.5 Bi0.5 TiO3 . Nat Mater 13:31–35 Li S (2012) Feasibility and mechanism of lithium oxide as sintering aid for Ce0.8 Sm0.2 O2-δ electrolyte. J Power Sour 205:57–62 Li M, Hou J, Fan Y, Xi X, Fu XZ, Luo JL (2022) Interface modification of Ru-CeO2 co-infiltrated SFM electrode and construction of SDC/YSZ bilayer electrolyte for direct CO2 electrolysis. Electrochimica Acta 426:140771 Lina N (2019) Review on zirconate-cerate-based electrolytes for proton-conducting solid oxide fuel cell. Ceram Int 45:6605–6615 Liu N (2006) Synthesis of La0.85 Sr0.15 Ga0.85 Mg0.15 O2.85 materials for SOFC applications by acrylamide polymerization. Mater Res Bull 41:461–468

56

S. Sengodan

Liu Y, Lao LE (2006) Structural and electrical properties of ZnO-doped 8 mol% yttria-stabilized zirconia. Solid State Ionics 177:159–163 LÜ J (2008) Chemical stability of doped BaCeO3 -BaZrO3 solid solutions in different atmospheres. J Rare Earths 26:505–510 Magrasó A (2010) Novel fabrication of Ca-Doped LaNbO4 thin-film proton-conducting fuel cells by pulsed laser deposition. J Am Ceram Soc 93:1874–1878 Magrasó A (2011) Development of proton conducting SOFCs based on LaNbO4 electrolyte-status in Norway. Fuel Cells 11:17–25 Marques FMB, Navarro LM (1996) Performance of double layer electrolyte cells part I: model behavior. Solid State Ionics 90:183–192 Matraszek A (2004) Phase diagram study in the La2 O3 -Ga2 O3 -MgO-SrO system in air. Solid State Ionics 166:343–350 Matsumoto H (2007) Relation between electrical conductivity and chemical stability of BaCeO3 based proton conductors with different trivalent dopants. Electrochem Solid-State Lett 10:B77 Mehta K, Xu R, Virkar AV (1998) Two-layer fuel cell electrolyte structure by Sol-Gel processing. J Sol-Gel Sci Technol 11:203–207 Minh NQ (1993) Ceramic fuel cells. J Am Ceram Soc 76:563–588 Mogensen M, Sammes NM, Tompsett GA (2000a) Physical, chemical and electrochemical properties of pure and doped ceria. Solid State Ionics 129:63–94 Mogensen M, Sammes NM, Tompsett GA (2000b) Physical , chemical and electrochemical properties of pure and doped ceria. 129:63–94 Mokkelbost T (2009) Thermal and mechanical properties of LaNbO4 -based ceramics. Ceram Int 35:2877–2883 Mori M (2008) Investigation of Li dopant as a sintering aid for ScSZ electrolyte for IT-SOFC. J Korean Ceram Soc 45:760 Mori T (2002) Oxide ionic conductivity and microstructures of Sm- or La-doped CeO2 -based systems. Solid State Ionics 154:461–466 Muccillo ENS, Kleitz M (1995) Ionic conductivity of fully stabilized ZrO2 : MgO and blocking effects. J Eur Ceram Soc 15:51–55 Nakayama S, Aono H, Sadaoka Y (1995) Ionic conductivity of Ln10 (SiO4 )6 O3 (Ln = La, Nd, Sm, Gd and Dy). Chem Lett 24:431–432 Ni M, Leung MKH, Leung DYC (2008) Technological development of hydrogen production by solid oxide electrolyzer cell (SOEC). Int J Hydrogen Energy 33:2337–2354 Norby T, Magrasó A (2015) On the development of proton ceramic fuel cells based on Ca-doped LaNbO4 as electrolyte. J Power Sour 282:28–33 Pérez-Coll D (2007) Re-examination of bulk and grain boundary conductivities of Ce1-x Gdx O2-δ ceramics. Electrochim Acta 52:2001–2008 Polini R, Pamio A, Traversa E (2004) Effect of synthetic route on sintering behaviour, phase purity and conductivity of Sr- and Mg-doped LaGaO3 perovskites. J Eur Ceram Soc 24:1365–1370 Prabhakaran K (2007) Characteristics of 8mol% yttria stabilized zirconia powder prepared by spray drying process. J Mater Process Technol 189:178–181 Rolle A (2004) Oxide ion conduction in oxygen rich doped Ba2 In2 O5+δ brownmillerite. In: MRS Proceedings. vol 835. K2.4 01 Feb 2011 Scott HG (1977) Phase relationships in the yttria-rich part of the yttria-zirconia system. J Mater Sci 12:311–316 Seo DJ (2006) Synthesis and properties of Ce1-x Gdx O2-x /2 solid solution prepared by flame spray pyrolysis. Mater Res Bull 41:359–366 Sha X (2007) Study on La and Y co-doped ceria-based electrolyte materials. J Alloy Compd 428:59–64 Shen S, Ni M (2015) 2D segment model for a Bi-layer electrolyte solid oxide fuel cell. J Electrochem Soc 162:F340–F347 Shi M (2006) Synthesis and characterization of Sr-and Mg-doped LaGaO3 by using glycine–nitrate combustion method. J Alloy Compd 425:348–352

Solid-State Electrolytes for Solid Oxide Electrolysis Cells

57

Singh P, Goodenough JB (2013) Monoclinic Sr1-x Nax SiO3-0.5x : new superior oxide ion electrolytes. J Am Chem Soc 135:10149–10154 Singman D (1966) Preliminary evaluation of ceria-Lanthana as a solid electrolyte for fuel cells. J Electrochem Soc 113:502 Skilbred AWB, Haugsrud R (2012) Sandvik Sanergy HT-A potential interconnect material for LaNbO4 -based proton ceramic fuel cells. J Power Sour 206:70–76 Steele BCH (2000) Appraisal of Ce1-y Gdy O2-y/2 electrolytes for IT-SOFC operation at 500 °C. Solid State Ionics 129:95–110 Stevenson JW (1998) Effect of A-site cation nonstoichiometry on the properties of doped lanthanum gallate. Solid State Ionics 113:571–583 Stevenson JW (2000) Influence of cobalt and iron additions on the electrical and thermal properties of (La, Sr)(Ga, Mg)O3 -δ. J Electrochem Soc 147:3213 Strickler DW, Carlson WG (1964) Ionic conductivity of cubic solid solutions in the system CaOY2 O3 -ZrO2 . J Am Ceram Soc 47:122–127 Sun Z (2011) Lowering grain boundary resistance of BaZr0.8 Y0.2 O3-δ with LiNO3 sintering-aid improves proton conductivity for fuel cell operation. Phys Chem Chem Phys 13:7692–7700 Syvertsen GE (2012) The effect of cation non-stoichiometry in LaNbO4 materials. Int J Hydrogen Energy 37:8017–8026 Takahashi T, Esaka T, Iwahara H (1977) Conduction in Bi2 O3 -based oxide ion conductor under low oxygen pressure. II. determination of the partial electronic conductivity. J Appl Electrochem 7:303–308 Tianshu Z (2002) Ionic conductivity in the CeO2 -Gd2 O3 system (0.05≤Gd/Ce≤0.4) prepared by oxalate coprecipitation. Solid State Ionics 148:567–573 Tuller HL, Nowick AS (1975) Doped ceria as a solid oxide electrolyte. J Electrochem Soc 122:255– 259 Virkar AV (1991) ‘Theoretical analysis of solid oxide fuel cells with two-layer composite electrolytes: electrolyte stability. J Electrochem Soc 138:1481–1487 Virkar AV, Tao G (2015) Reversible high temperature cells for power generation and hydrogen production using mixed ionic electronic conducting solid electrolytes. Int J Hydrogen Energy 40:5561–5577 Wang F-Y, Wan B-Z, Cheng S (2005) Study on Gd3+ and Sm3+ co-doped ceria-based electrolytes. J Solid State Electrochem 9:168–173 Wang Y (2003) Low-temperature fabrication and electrical property of 10mol% Sm2 O3 -doped CeO2 ceramics. Sci Technol Adv Mater 4:229–238 Yahiro H, Eguchi K, Arai H (1989) Electrical properties and reducibilities of ceria-rare earth oxide systems and their application to solid oxide fuel cell. Solid State Ionics 36:71–75 Yamamoto O (1995) Electrical conductivity of stabilized zirconia with ytterbia and scandia. Solid State Ionics 79:137–142 Yamamoto O (2000) Solid oxide fuel cells: fundamental aspects and prospects. Electrochim Acta 45:2423–2435 Yashima M, Kakihana M, Yoshimura M (1996) Metastable-stable phase diagrams in the zirconiacontaining systems utilized in solid-oxide fuel cell application. Solid State Ionics 86:1131–1149 Zha S, Xia C, Meng G (2003) Effect of Gd (Sm) doping on properties of ceria electrolyte for solid oxide fuel cells. J Power Sour 115:44–48 Zhang GB, Smyth DM (1995a) Defects and transport of the brownmillerite oxides with high oxygen ion conductivity—Ba2 In2 O5 . Solid State Ionics 82:161–172 Zhang GB, Smyth DM (1995b) Protonic conduction in Ba2 In2 O5 . Solid State Ionics 82:153–160 Zhang TS (2004) Aging behavior and ionic conductivity of ceria-based ceramics: a comparative study. Solid State Ionics 170:209–217 Zhang X (2000) Interface reactions in the NiO-SDC-LSGM system. Solid State Ionics 133:153–160 Zhang X (2006) A study on sintering aids for Sm0.2 Ce0.8 O1.9 electrolyte. J Power Sour 162:480–485 Zheng F, Bordia RK, Pederson LR (2004) Phase constitution in Sr and Mg doped LaGaO3 system. Mater Res Bull 39:141–155

Oxygen Electrode Materials for Solid Oxide Electrolysis Cells (SOECs) Vaibhav Vibhu, Amir Reza Hanifi, Thomas H. Etsell, and Jean-Marc Bassat

Keywords Solid oxide electrolysis cell (SOEC) · Steam electrolysis · Oxygen electrode materials · Perovskite-type oxides · K2 NiF4 -type oxides · Electrode forming methods · Electrode performance · Degradation

1 Introduction As explained in previous sections, SOECs generally operate at high temperature. A high operating temperature increases the efficiency of SOECs, but it also accelerates the degradation rate. This is mainly caused by the chemical reactivity between the SOEC constituents, resulting in the formation of secondary phases that either block the active electrocatalytic sites or completely transform into inactive local phases. These secondary phases block the reaction and diffusion paths in the electrodes, leading to a decrease in their conductivity and electrocatalytic activity. Therefore, most of the research has been done for better understanding of SOFC degradation which has helped to control the degradation rate below 2% per 1000 h. SOEC investigations about the degradation mechanisms are still under progress. The most recent studies reveal that the degradation mechanism of SOEC stacks is related to oxygen electrode delamination, Cr vapor poisoning, microstructure V. Vibhu Institute of Energy and Climate Research (IEK-9), Forschungszentrum Jülich GmbH, 52425 Jülich, Germany A. R. Hanifi · T. H. Etsell Department of Chemical and Materials Engineering, Donadeo Innovation Centre for Engineering, University of Alberta, Edmonton T6G 1H9, Canada J.-M. Bassat (B) CNRS, Univ. Bordeaux, Bordeaux INP, ICMCB, UMR 5026, F-33600 Pessac, France e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M. A. Laguna-Bercero (ed.), High Temperature Electrolysis, Lecture Notes in Energy 95, https://doi.org/10.1007/978-3-031-22508-6_4

59

60

V. Vibhu et al.

degradation and seal leakage. The oxygen electrode delamination occurs normally due to high oxygen partial pressure in the electrode–electrolyte interface. Cr vapor poisoning is mainly caused by the stack interconnects which are usually made of Cr-rich alloys, e.g., ferritic stainless steel. Microstructure degradation is generally caused by interactions among material constituents, which results in a decrease in the conductivity and electrochemical activity. Seal leakage is also a severe problem, which occurs during thermal cycling or due to the mismatch between the thermal expansion coefficient of the sealant and other components of the cell. However, the dominating degradation mechanism is still unclear. During SOEC operation, the electrode polarizations induce large voltage losses especially at the oxygen electrode compared to the fuel electrode (Park et al. 2018; Yamamoto et al. 1987; Yamamoto 2000; Patcharavorachot et al.2008). Therefore, it is mandatory to find a good oxygen electrode material for SOEC that reduces the overpotential, is compatible with other cell components and chemically shows a high stability during operation. Another key parameter is the design of the electrode, mainly including the microstructural parameters (the specific surface area, the mean phase diameter, the volume fraction of the percolated phase, the tortuosity factor…) which themselves depend on the shaping methods. Therefore, the electrochemical performance of the cell highly depends on the synthesis methods of the electrodes. There are different methods to produce the oxides. This particular chapter is focused on the recent advances in oxygen electrode materials for SOECs, the associated forming methods and future prospects for their development.

2 Oxygen Evolution Reactions in SOECs The electrochemical oxygen evolution reaction (OER) takes place at the oxygen electrode, i.e., anode and can be ascribed as follows: O2− (electrolyte) → 1/2 O2 (gas) + 2e−

(1)

It can also be described using Vink-Kröger notation: Oox →

1 O2 + Vo·· + 2e− 2

(2)

When the electrode material is an electronic conductor, this reaction can take place only at the triple phase boundary (TPB). As a result, the oxygen evolution reaction is restricted to a one- dimensional region; hence, such electrodes show poor electrochemical performance, e.g., La1−x Srx MnO3 (LSM). The most common way to increase the reaction zone of these materials is to prepare composite electrodes by including oxide ion conducting materials such as YSZ.

Oxygen Electrode Materials for Solid Oxide Electrolysis Cells (SOECs)

61

Another approach is to consider a material which has both electronic and ionic conductivity, i.e., a mixed ionic and electronic conductor (MIEC). In this case, the electrode reactions take place throughout the bulk of the electrode in addition to the TPB. Commonly, the MIEC oxides can be formulated as (A1-x A’x )m (Mn+ M(n +1)+ )p Oz±δ , where A and A’ are lanthanide or alkaline earth cations and M is a transition metal cation. In this case, the mixed valence of the metallic cations M gives rise to electronic conductivity σe ; the oxygen nonstoichiometry δ-value, in terms of either oxygen vacancies (Vo.. ) or excess oxygen atoms (Oi ) induces ionic conductivity σion . The oxygen non-stoichiometry δ-value as well as the type of structure are generally determined by lanthanide or alkaline earth cations as well as the m/n ratio.

3 Basic Requirements of Oxygen Electrodes Various requirements must be met to consider any material as an SOEC oxygen electrode. • High electronic conductivity (σ e > 100 S cm−1 ); • High ionic conductivity (σ i > 0.001 S cm−1 ) rather similar to that of the electrolyte; • The thermal expansion coefficient (TEC) should match with the other components of the cell, typically in the range 10 −13 × 10–6 °C−1 to avoid excessive thermomechanical stresses; • Chemical stability at the high sintering temperature of the electrode during cell preparation as well as chemical compatibility with the electrolyte; • High oxide ion diffusivity (D) as well as surface exchange kinetics (k); • High catalytic activity for the OER; • Electrode microstructure must have adequate porosity in order to allow oxygen molecules to easily diffuse through the electrode; • Long-term chemical stability at operating conditions of SOECs; • Low cost and ease of recycling. Relationship between transport properties and electrocatalytic activity: The electrochemical activity of the material is generally recognized by polarization resistance (Rp ). The Rp is highly dependent on the basic material electrode properties, although the OER comprises several steps including O-ion transfer across the electrode–electrolyte interface, O-ion diffusion on the oxide surface or in the bulk, charge transfer and association process and surface exchange processes such as O-desorption and O2 gas diffusion. However, the rate-limiting steps are typically oxygen surface exchange (characterized by the surface exchange coefficient, k*) and bulk/surface diffusion (characterized by the oxygen diffusion coefficient, D*). Normally, good MIEC materials have high values of k* and D*. These two parameters can be determined by experimental techniques such as oxygen diffusion measurements, such as isotopic exchange (18 O/16 O) depth profiling (IEDP) using secondary ion mass

62

V. Vibhu et al.

spectroscopy (SIMS) (based on a gradient of isotopic oxygen), oxygen permeation measurements and electrical conductivity relaxation measurements. Especially with IEDP-SIMS measurements, an 18 O concentration profile is obtained. The oxygen bulk diffusion normally described by the oxygen diffusion coefficient (D*), as well as surface exchange kinetics described by the surface exchange coefficient (k*), is obtained by fitting the 18 O concentration profile with the Crank model. Furthermore, the ionic conductivity of the material is also calculated from the D* values using the Nernst relation (Eq. 1). σi = (z o F)2 D ∗ Co /RT

(3)

where z o F represents the electrical charge of one mole of oxygen ions. D* and C o are the diffusion coefficient and the concentration of oxygen ions, respectively. The Adler-Lane-Steele (ALS) model provides a good mathematical illustration of how these oxygen transport coefficients are linked to the electrochemical properties (Adler et al. 1996). The model describes the conversion of the electronic current into ionic current over the thickness of the electrode. At zero bias, the impedance of a symmetrical cell is given by the formula: Z = Relectrolyte + Z interfaces + Z chem

(4)

Relectrolyte is related to the O2− ionic solid-state transport in the electrolyte, Z interfaces describes the electron transfer at the current collector–electrode interface and the O2− diffusion at the electrode/electrolyte interface, and Z chem is a convoluted impedance including the following steps: (i) the solid-state diffusion of oxygen electro-active species in the electrode; (ii) the oxygen surface exchange between the electrode and its surrounding atmosphere; and (iii) the diffusion of molecular O2 in this atmosphere (Adler 2000). In the limit where only steps (i) and (ii) would be limiting the oxygen reduction reaction, Z chem should reduce to Eq. (5) (Adler 2004): / Z chem = Rchem

/ 1 ) = Rchem ( 1 + j w Rchem Cldiff

1 1 + j w(tchem )

(5)

In this case, Z chem is mathematically equivalent to the Gerischer impedance (Gerischer 1951). Rchem and Cldiff are chemical resistance and capacitance reflecting co-limitation of the oxygen reduction reaction by the exchange surface coefficient R0(MIEC) and the ionic conductivity σi(MIEC) of the MIEC. tchem is the relaxation time of the Gerischer impedance. Expression of Rchem , tchem and .Cldiff has been defined by Adler et al (2000) through Eqs. (6, 7 and 8): / Rchem =

RT τ × 2F 2 (1 − ε) · SA .R0(MIEC) · σi (MIEC)

(6)

Oxygen Electrode Materials for Solid Oxide Electrolysis Cells (SOECs)

(1 − ε) Vm · f thermo · S A · R0(MIEC) / 2·σ i(MIEC) F · (1 − ε)1.5 = × f thermo · Vm RT · τ · SA · R0(MIEC)

tchem = Cldiff =

tchem Rchem

63

(7)

(8)

From these equations, R0(MIEC) and σi(MIEC) can be determined by Eqs. (9) and (10), respectively: (1 − ε) Vm . f thermo .S A .tchem

(9)

RT τ.Vm . f thermo .tchem × 2 2F ((1 − ε).Rchem )2

(10)

R0(MIEC) = σi (MIEC) =

The parameters used in these equations are the electrode porosity ε, the electrode tortuosity τ, the molar volume of the MIEC material Vm and its surface area SA . f thermo is a thermodynamic factor expressing the ease with which oxygen stoichiometry of the material is changed upon variation of the oxygen partial pressure. Rchem and tchem are obtained experimentally via fitting the experimental impedance data using the Gerischer impedance. Finally, R, T and F are the gas constant, absolute temperature and Faraday constant. The thickness of the electrode involved in the oxygen reduction reaction can be also calculated using the ALS model, which can be considered as the electrode active region. The thickness of this region is given by the diffusion length ldiff parameter in Eq. (11): ldiff = Cldiff

Vm . f thermo RT = × × 4F 2 (1 − ε)

/

RT (1 − ε).σi(MIEC) × 8F 2 τ.SA .R0(MIEC)

(11)

Some important conditions must be fulfilled in order to use the ALS model. (i) The first point is that the material should be a MIEC. (ii) The thickness of the electrode must be larger than the theoretical diffusion length as given in Eq. (6). Indeed, in the case where the oxygen reduction reaction should be limited by the electrode thickness, one should use a finite-length Gerischer impedance expression, as formulated by Boukamp et al. (2003) for instance. (iii) The Gerischer impedance derived in the ALS model can only be applied to an electrode in which the limiting steps to the oxygen reduction reaction are the solid-state diffusion of O2− and oxygen exchange at the gas–electrode interface. (iv) Finally, the thickness of the electrode involved in the oxygen electrode reaction must be larger than the average diameter of the electrode grains.

64

V. Vibhu et al.

4 Types of Oxygen Electrode Materials There are mainly three families of MIEC oxide materials which have been already proposed for SOC oxygen electrodes: perovskites (ABO3 ), the layered perovskites (or double perovskites) (AA’B2 O6 ) and the Ruddlesden-Popper An+1 Bn O3n+1 phases. All these materials possess an oxygen sixfold coordinated transition metal scaffold (BO6 ) where alkaline earth or rare earth ions (A) are located on the vertices of the cube containing the octahedron (see Fig. 1). Also, the type and the location of the O-defects (sub-stoichiometry and over-stoichiometry in the case of vacancies and oxygen excess, respectively) are responsible for MIEC behavior. In perovskites, the lanthanide (Ln = La, Pr or Nd) and/or alkaline earth (Ca, Sr and Ba) cations are randomly distributed on A-sites, and O-vacancy defects are also randomly distributed on the O-sublattice (Fig. 1a). If there is a large difference between the ionic radii of the alkaline earth and lanthanide cation (e.g., Ba2+ and Ln3+ ), then it induces a cationic ordering in alternate layers AO − BO2 − A’O − BO2 and results in the formation of layered double perovskites (Fig. 1b). In case of double perovskites, the Ln3+ cations tend to decrease their coordination number, locating the O-vacancies mainly on the LnO layers. As a result of cation ordering, the migration path for O-ion diffusion in AA’B2 O6 is two dimensional, unlike the three-dimension migration path in perovskites.

Fig. 1 Crystal structures of the relevant MIEC cathode materials: a perovskite, b double perovskite and c layered perovskite (Gao et al. 2016)

Oxygen Electrode Materials for Solid Oxide Electrolysis Cells (SOECs)

65

In case of low-order (n = 1) Ruddlesden-Popper compounds A2 BO4 , the difference between A-O and B-O bond distances produces a mismatch between two consecutive perovskite layers and results in a different AO-ABO3 -layered structure (Fig. 1c). In A2 BO4 -type compounds, again the diffusion of O-ions occurs in AO layers that allows the insertion of O-interstitial defects on tetrahedral sites, inducing a change of the O-migrating species from O-vacancies to O-interstitials. All these MIEC materials are generally characterized by their oxygen surface exchange coefficient (k*) and oxygen bulk diffusion coefficient (D*). For a good oxygen electrode, the values of k* and D* should be high. Figure 2 shows the k* and D* of different Co- and Fe-containing compounds, which generally show the fastest oxygen transport coefficients among the perovskites. There is a wide dispersion of the reported transport coefficients of the same compositions, probably arising from the sensitivity of O-transport kinetics to sample thermal history, more especially the k* via surface segregation, surface reconstruction and contamination. Moreover, it is not easy to accurately determine both k* and D* from isotopic exchange and conductivity relaxation measurements. Among all perovskites, the compound with Ba and Sr on the A-sites (BSCF) exhibits the fastest O-ion transport kinetics. Also, Sm-containing compounds generally show faster kinetics than La-containing perovskites. When the A-site of perovskite is doped with lanthanides (Sm or La) and Sr, the Sr-rich compositions exhibit fast kinetics: for example, La1-x Srx CoO3-δ (LSC), La1-x Srx FeO3-δ (LSF) and

Fig. 2 Literature isotope exchange data for perovskite cobalt ferrite materials, including a k* versus temperature and b D* versus temperature (Gao et al. 2016)

66

V. Vibhu et al.

Fig. 3 Literature isotope exchange data for cobalt-rich layered perovskite materials, a k* versus temperature and b D* versus temperature (Gao et al. 2016)

Sm1-x Srx CoO3-δ (SSC). The B-site composition is also important; normally, Co-rich compositions show faster kinetic coefficients than Fe-rich compositions. The plots of k* and D* values versus temperature for double perovskites are shown in Figure 3 These compounds show fast O-reduction kinetics. The highest value is shown by PrBaCo2 O5+δ (PBCO). The best layered perovskites show kinetic coefficients comparable to the best perovskites other than BSCF, which is in good agreement with density functional theory (DFT) calculations. One advantage of the layered perovskites over cobalt barium perovskites is phase stability, since Ba-rich hexagonal phase segregation is not reported for these compounds. Figure 4 shows the literature data of k* and D* versus T for K2 NiF4 -type Ruddlesden-Popper phases. Some compounds have high values especially Pr2 NiO4+δ and La2 NiO4+δ , similar to best the perovskite materials. Normally, A-site doping with Sr decreases D* and k*, contrary to that of perovskite materials. Further doping with (Ca, Ga, Cu) enhances the electronic conductivity, and oxygen diffusion or thermodynamic stability (Cu, Ga) can increase their utility as an oxygen electrode.

5 Forming Methods of Oxygen Electrodes Oxygen electrodes are manufactured from single materials or composites with high electronic and ionic conductivity. The electrochemical performance of the cell depends highly on the synthesis methods of the electrodes. There are different methods to produce the oxides such as reaction of nitrates or acetates in acid and later calcination and milling. Another method can be solid-state reactions between oxides

Oxygen Electrode Materials for Solid Oxide Electrolysis Cells (SOECs)

67

Fig. 4 Literature isotope exchange data for Ruddlesden-Popper nickelate materials, a k* versus temperature and b D* versus temperature (Gao et al. 2016)

or carbonates and similarly milling to obtain finer particles of the phase formed (Gómez et al. 2016). The fabrication process parameters such as sintering temperature, composition and electrode thickness significantly impact the polarization resistance. For instance, Patro et al. made composites of Pr0.58 Sr0.4 Fe0.8 Co0.2 O3-δ (PSCF) and GDC with different weight ratios using screen printing of the composite ink. They found that the PSCF50–GDC50 composite has the lowest polarization resistance. Using the dilatometric shrinkage curve and polarization resistance, they identified 1100 °C as the suitable sintering temperature. By analyzing the impedance curve, they found the optimal thickness of the oxygen electrode to be 8 to 9 μm for which the cell showed a resistance of 0.12 T. cm2 at 800 °C in an air atmosphere (Patro et al. 2010). The forming method depends on the desired microstructure of the electrode and its thickness. There are two main forming methods for thin-film ceramics: particulate and deposition methods. In the particulate methods such as tape casting, pressing, tape calendering, slip casting and extrusion, the ceramic powder is compacted into the desired shape and then sintered at high temperatures. In the deposition methods such as plasma spraying, chemical vapor deposition (CVD), dip coating or spray pyrolysis, a ceramic suspension is coated onto a substrate which can be ceramic or metallic before sintering.

68

V. Vibhu et al.

5.1 Particulate Method Among the particulate processes, tape casting, tape calendering, screen printing and pressing are used to fabricate planar SOECs, while slip casting and extrusion are used to fabricate tubular SOECs. Pressing, slip casting and extrusion are among the methods used to fabricate fuel electrodes which are the cell support and much thicker than the oxygen electrode. Tape casting is used to produce thin sheets of capacitor dielectrics and solid oxide cell layers. In this method, a doctor blade controls the thickness of a ceramic paste. The formed layer is later stripped from the support and sintered. Multilayers can be formed by continuous casting layers onto each other. In tape calendering, ceramic powder and organic binder are mixed in a high shear mixer, and the resulting paste is rolled into a tape of desired thickness and sintered at high temperatures. By a rolling technique, multiple tapes can be laminated together. In the screen-printing method, the ceramic paste is forced through holes in a wire screen using a rubber squeegee. The ceramic paste should have a low viscosity in order to be transferred onto the substrate. The screen then snaps off the substrate, and the printed film becomes level to fill in the unevenness from the screen wires. The viscosity of the paste then increases to stabilize the formed film and to prevent spreading of the deposited film.

5.2 Deposition Method Deposition techniques are used for the fabrication of both planar and tubular SOECs. In this method, a ceramic suspension is applied onto the support using different techniques. By adjusting the solid loading of the suspension or by multiple depositions, a desired thickness of the coated layer can be achieved. As a first approximation, one can quote the following deposition techniques: dip coating, spray pyrolysis, plasma spraying and electrophoretic deposition. The dip coating process involves immersing the substrate in a ceramic slurry and withdrawing at rates of 10–30 cm/min. The slurry should wet the substrate well in order to have a uniform coating meaning the contact angle should be low. In spray pyrolysis, the ceramic slurry is sprayed onto a substrate at high temperatures and then the coated substrate is sintered. In plasma spraying, the ceramic powder is injected into a plasma jet where the powder is accelerated, melted and deposited on the substrate. In electrophoretic deposition, charged particles move through a liquid under the influence of an external electric field. The substrate forms one of the electrodes in an electrochemical cell (Singhal and Kendall 2003; Carlson et al. 2012). Table 1 presents several examples of oxygen electrodes, their manufacturing process and thickness reported in the literature. Among different forming methods discussed so far, tape casting, dip coating and screen printing are preferred to fabricate the thin oxygen electrode (Gamble 2011).

Oxides

LSCN

Calcination (1100 °C, 18 h)

–

Spray pyrolysis

Nitrates

–

−

Calcination

–

Acid-nitrates

−

BSCF

LSC

Screen printing

Annealing (1000 °C, 2 h)

Ball milling

Nd, Ni in nitric acid

70SSC:30BCZY

Nd2 NiO4+δ

SSC/BCZY

LSCF-GDC

Screen printing

–

–

–

Screen printing

Dip coating

Screen printing

Screen printing

Screen printing

Plasma spraying

Screen printing

LSCF

Calcination

Ball milling

Acid-nitrates

Dip coating

–

Calcination

Dip coating

BSCF-SDC

Nitrates-oxides

Co/CeO2

Calcination (900 °C)

Spray painting

Dip coating

−

Ball milling

Dip coating Dip coating

−

−

LSM-YSZ

50LSM:50YSZ

Nitrate salts

LSM-YSZ

Nitrate salts

LSM-YSZ

LSF-YSZ

Nitrate salts

Nitrate salts

LSF-YSZ

LSC-YSZ

Screen printing

Spray pyrolysis (1200 °C)

–

Dip coating Dip coating

LCM

Calcination (900 °C)

Calcination (900 °C)

Acetates mixture

Acetates mixture

Processing

LSM

Powder synthesis

LSC

Precursors

Material

Table 1 Different processing methods for fabrication of oxygen electrodes

10

30

36

30

70

40

30

30

50

–

15

300

10

300

300

300

–

–

–

Thickness (μm)

1000 (2 h)

–

1000 (1 h)

1000 (1 h)

1150 (5 h)

1100 (2 h)

1100

–

1180

1000

–

900

–

1250

1250

1250

1300

1200

1200

Sintering temperature (°C)

(Laguna-Bercero et al. 2011)

(Wang et al. 2010)

(Prestat et al. 2010)

(Kim-Lohsoontorn et al. 2010)

(Liu et al. 2010b)

(He et al. 2010)

(Chauveau et al. 2010)

(Schiller et al. 2009)

(Liang et al. 2009)

(Bo et al. 2008)

(Wang et al. 2008)

(Bidrawn et al. 2008)

(Jensen et al. 2007)

(Wang et al. 2006)

(Wang et al. 2006)

(Wang et al. 2006)

(Kaiser eta l. 1997)

(Eguchi et al. 1996)

(Eguchi et al. 1996)

References

Oxygen Electrode Materials for Solid Oxide Electrolysis Cells (SOECs) 69

70

V. Vibhu et al.

5.3 Infiltration Method Infiltration has been an important method for fabrication of high-performing electrodes for fuel cells which provides unique microstructures in terms of the particle size and distribution. It not only improves the catalytic activity (through the presence of dispersed nanoparticles), but also increases the ionic and electronic conductivity (through an interconnected network of nanoparticles) which overall leads to an increase of TPB length and reduction of the cell resistance. The nano-sized particles form due to a lower heat treatment temperature used to decompose the nitrates of constituents (350–450 ºC) and form the final phases (600–1000 ºC) compared with the conventional sintering process (1100–1400 ºC). Infiltration also prohibits any reaction of phases with the electrolyte and formation of non-conducting phases due to the reduced temperature needed to form the desired phases. In addition, due to their enhanced electrochemical performance, infiltrated cells can be operated at much lower temperatures, improving the long-term stability of the cell (Vohs and Gorte 2009; Sholklapper et al. 2008; Jiang 2012). Infiltration has been widely used to fabricate novel electrode structures such as LSM (Sholklapper et al. 2006; Laguna-Bercero et al. 2015), LSCF (Shah and Barnett 2008; Burye et al. 2016), LSBT (Vincent et al. 2013, 2012) and Ruddlesden-Popper materials such as Lan+1 Nin O3n+1 (n = 1, 2, 3)-YSZ (Li et al. 2016; Choi et al. 2011) or Nd2 NiO4+δ (Laguna-Bercero et al. 2014; Chen et al. 2015). These studies show that the infiltrated electrodes provided superior performance compared with the conventionally sintered electrodes. It is possible to modify the microstructure of a conventional LSM-YSZ composite oxygen electrode by infiltrating LSM nanoparticles in order to increase the electrical performance of the cell (Torabi et al. 2011; Sholklapper et al. 2007). This is due to increased TPB length of infiltrated LSM-YSZ compared with the standard LSMYSZ composite. In addition, no sintering process for the LSM is needed which inhibits the formation of non-conducting secondary phases such as La2 Zr2 O7 and also coarsening of the catalyst during sintering. The infiltration of Ruddlesden-Popper materials into YSZ porous structures was first proposed by Choi et al. They fabricated Lan+1 Nin O3n+1 (n = 1, 2, 3)-YSZ composites as cathodes for planar SOFCs, obtaining maximum power densities of 0.717, 0.754 and 0.889 W/cm2 at 750 °C for La2 NiO4 , La3 Ni2 O7 and La4 Ni3 O10 , respectively (Choi et al. 2011). Song et al. found that RuO2 infiltration to LSM-YSZ improves the OER in solid oxide CO2 electrolyzers (Song et al. 2018). After anchoring 6 wt.% RuO2 into LSM-YSZ, the current density improved from 0.46 A/cm2 to 0.74 A/cm2 at 1.2 V and 800 °C. The cell showed reasonable stability, and RuO2 particles maintained their size and did not agglomerate following a stability test (Choi et al. 2011). Samson et al. (2011) reported for La0.6 Sr0.4 CoO3−δ (LSC) infiltrated into Gddoped ceria (so-called GDC), a polarization resistance, Rp = 0.044 T.cm2 at 600 °C. Concerning the nickelate materials, Nicollet et al. (2015) obtained for La2 NiO4+δ infiltrated into GDC, a polarization resistance value as low as 0.15 T.cm2 at 600 °C.

Oxygen Electrode Materials for Solid Oxide Electrolysis Cells (SOECs)

71

The values of Rp obtained for infiltrated electrodes are far better than the screenprinted electrode. Zheng et al. impregnated La0.8 Sr0.2 Co0.8 Ni0.2 O3-δ (LSCN) into a GDC/LSM composite oxygen electrode to form a high-performance electrode for H2 O/CO2 co-electrolysis. The cell showed a current density of 1.60 A/cm2 at 1.5 V at 800 °C with a H2 O/CO2 ratio of 2/1 and was stable following a 100 h test (Zheng et al. 2018). Shahrokhi et al. studied the electrochemical performance and stability of LSM oxygen electrodes which were infiltrated with La2 Ni0.5 Co0.5 O4 (RuddlesdenPopper structure) and LaNi0.5 Co0.5 O3 (perovskite structure) nanoparticles under cyclic SOEC and SOFC modes. They found that the addition of La2 Ni0.5 Co0.5 O4 nanoparticles was more effective in reducing the polarization resistance compared with LaNi0.5 Co0.5 O3 and attributed that to the facile oxygen transport in the rock salt layer of the Ruddlesden-Popper structure. In addition, Ruddlesden-Popper structure infiltrated electrodes showed a better stability when the polarization resistances were analyzed (Shahrokhi et al. 2018). Zhang et al. infiltrated SrTi0.3 Fe0.6 Co0.1 O3-δ (STFC) into LSM-YSZ oxygen electrodes which resulted in improvements of the oxygen reduction reaction (ORR) and OER. Their cell showed a high stability under SOEC mode (Zhang et al. 2019). Infiltration of praseodymium oxide has also been shown to improve the performance of LSM-GDC oxygen electrodes (Navarrete et al. 2015). In addition, when Pr was infiltrated into the GDC backbone, the polarization resistance was 0.028 T. cm2 at 600 °C which is a low resistance at this temperature. They also achieved the remarkable power density of 0.825 W/cm2 at 600 °C, and at a current load of 0.5 A/cm2 , the voltage showed a degradation rate lower than 1% per 1000 h (Nicollet et al. 2016).

5.3.1

Development of Porous Scaffold for Infiltration

Infiltration of porous YSZ or GDC substrates can provide oxygen electrodes with superior electrochemical performance compared with the traditionally prepared composite electrodes. However, developing a suitable porous scaffold is the key for a successful infiltration. Hanifi et al. developed different types of porous YSZ microstructures for infiltration by calcination of as-received YSZ at 1300–1500 ºC and milling the coarsened powder in an aqueous suspension for 72 h (Hanifi et al. 2012a, b, 2011a, 2011a, 2012a, 2016). Calcination is an essential method for development of a porous substrate since as-received YSZ powder normally has fine particles (250 nm), a high surface area (12–13 m2 g−1 ) and thus high sintering ability. Even with incorporation of pore-formers inside the YSZ layer, there are dense YSZ regions which increase the tortuosity for gas diffusion. YSZ crystals and particles enlarge upon increasing the calcination temperature followed by milling compared with the as-received powder. Thus, their ability to shrink is lowered during the sintering process. Calcination and subsequent milling lead to the formation of particles with larger size (500–750 nm) and a lower surface area (3–5 m2 g−1 ). The particle size correlates with the calcination temperature (Fig. 5) (Hanifi et al. 2012a, b). The

72

V. Vibhu et al.

Fig. 5 TEM images of YSZ powders after 72 h milling. a As-received Tosoh YSZ, b Calcined at 1300, milled for 72 h, c Calcined at 1400, milled for 72 h and d Calcined at 1500, milled for 72 h (Hanifi et al. 2012a, b)

reduced surface area of YSZ generates a porous structure following sintering at 1350– 1400 ºC which contains interconnected pores (~20–25%) even without incorporation of any pore-formers (Hanifi et al. 2012a, b).

5.3.2

Effect of Pore-Former on the Microstructure of the Porous Scaffold

Addition of pore-formers during electrode fabrication leads to the generation of higher amounts of porosity which facilitates gas diffusion. Pore-formers such as graphite, poly methyl methacrylate (PMMA), carbon black and starch have been widely used in oxygen electrodes. The porosity has an optimal amount (50–60 vol.%) for oxygen electrodes above which contact between the particles is weakened which can affect the electrode mechanical properties as well as the cell resistance (Tao et al. 2009). Pore-formers can decrease the shrinkage of the coated layer, and the resulting stresses lead to cell cracking (Ni 2009). It has been observed that when a high surface area YSZ is used in the slip, in order to generate a sufficient amount of porosity (50 vol.%) for infiltration and gas diffusion,

Oxygen Electrode Materials for Solid Oxide Electrolysis Cells (SOECs)

73

Fig. 6 SEM images of the YSZ substrate sintered at 1350 °C. a Tosoh YSZ + 20 vol.% graphite, b Tosoh YSZ + 20 vol.% PMMA, c Calcined 1500 °C-milled YSZ + 20 vol.% graphite and d Calcined 1500 °C-milled YSZ + 20 vol.% PMMA (Hanifi et al. 2012a, b)

60–70 vol.% pore-former (graphite or PMMA) is required. At low contents of poreformers, mainly closed pores are formed which are not desirable for the electrode. However, when calcined and milled YSZ powder is used in the suspension, a low content of pore-former can still create sufficient open porosity (Fig. 6) (Hanifi et al. 2012a, b).

5.3.3

Infiltration of Porous YSZ to Form Electrodes

Hanifi et al. showed that LSM infiltrated into a YSZ scaffold (Fig. 7) boosts the SOEC performance in comparison with standard LSM-YSZ composite electrodes (Hanifi et al. 2014a, b). They also found that the morphology of the porous YSZ network significantly affects the surface area of the YSZ scaffold, TPB length and, therefore, electrochemical performance of an infiltrated cell. Porous YSZ scaffolds fabricated using calcined/milled YSZ showed a uniform microstructure before and after infiltration with nano-sized particles of LSM and led to a homogenous distribution of infiltrates and superior performance compared with the conventional LSM-YSZ cathodes (Torabi et al. 2011; Hanifi et al. 2014a, b).

74

V. Vibhu et al.

Fig. 7 YSZ coverage by LSM near the interface of the cathode and electrolyte (Hanifi et al. 2014a, b)

Figure 8a shows the microstructure of Nd2 NiO4+δ infiltrated into a porous YSZ scaffold which forms the oxygen electrode. The nanostructure of the nickelate phase enhanced its performance significantly. Preliminary SOEC performance using 3% steam in the fuel electrode chamber (Fig. 8b) shows better performance under SOEC mode compared with SOFC mode (Laguna-Bercero et al. 2014).

Fig. 8 a SEM image showing the interface between the YSZ electrolyte and porous YSZ infiltrated with the Nd–nickelate and b Electrochemical performance of the cell under both SOFC and SOEC modes at 800 °C (Laguna-Bercero et al. 2014)

Oxygen Electrode Materials for Solid Oxide Electrolysis Cells (SOECs)

75

6 Performance Under SOEC Operation The most commonly used oxygen electrode materials are LSM, LSFC, LSC, LSF, LSCN, etc. Some composites, for example, LSM-YSZ, LSC-GDC and LSCF-GDC were also investigated as oxygen electrodes. In addition, several K2 NiF4 -type nickelates materials have also been investigated. However, only a few double perovskites have been studied as SOEC oxygen electrodes. The performance of different oxygen electrode materials (in terms of current density) is summarized in Table 2. (La,Sr)MnO3 (LSM) is a typical oxygen electrode material which has been studied from the last few decades. However, this material shows fast degradation during SOEC operation. Chen et al. have reported the failure mechanism of LSM oxygen electrodes under SOEC mode (Chen and Jiang 2011). The authors investigated the behavior of an LSM//8YSZ//LSM half-cell under 500 mA cm−2 anodic current at 800 °C in air. Delamination of the electrode was observed after 48 h of operation; as a result, a large increase in ohmic and electrode polarization resistance was observed. The delaminated electrode and electrolyte interface is characterized by the formation of nanoparticles within LSM and the electrolyte surface, as illustrated in Fig. 9. The formation of nanoparticles is most likely due to the migration of or incorporation of oxide ions from the YSZ electrolyte into the bulk of LSM grains, leading to shrinkage of the LSM lattice due to the oxidation of manganese ions and the formation of manganese cation vacancies. The shrinkage of the LSM lattice creates local tensile strain, resulting in micro-cracks and subsequent formation of nanoparticles within LSM particles at the electrode–electrolyte interface. The formation of nanoparticle clusters weakens the electrode–electrolyte interface, leading to the delamination and failure of the LSM oxygen electrode under high internal partial pressure of oxygen at the interface. To avoid such kind of degradation for LSM electrodes, the reversible operation of solid oxide electrochemical cells could be effective. Graves et al. have reported the reversible behavior of Ni-YSZ//YSZ//LSM-YSZ single cells (Graves et al. 2015). Two types of electrochemical tests were performed at 800 °C, first one at constant current steam electrolysis at − 1 A cm−2 and another one under cycling between steam electrolysis at − 1 A cm−2 and fuel cell operation at + 0.5 A cm−2 . Initially, the −1.0/ + 0.5 A cm−2 cycling was first performed with time periods of 1 h in electrolysis mode and 5 h in fuel cell mode. Later, it was increased up to 5 h electrolysis and 5 h fuel cell mode. The authors have demonstrated the severe electrolysis-induced degradation, which was previously believed to be irreversible. Such a degradation could be completely eliminated by reversibly cycling the cell between electrolysis and fuel cell modes. During continuous steam electrolysis, operation at high current density (−1 A cm−2 ) led to severe microstructure deterioration near the oxygen electrode–electrolyte interface and a corresponding large increase in ohmic resistance. After 4000 h of reversible operation, no microstructural damage was observed and the ohmic resistance even showed slight improvement. Daroukh et al. investigated the performance of Ni-YSZ//YSZ//GDC//LSCF for three different single cells at 770 °C with a current density − 1 A cm−2 and 80%

YSZ YSZ

LSM

BSCF-SDC composite

LSCF1982-GDC composite 0.9 0.3 –

NiO–YSZ 50% H2 + 50% H2 O NiO–YSZ 40% H2 + 60% H2 O NiO–YSZ 50% H2 + 50% H2 O

YSZ/GDC

Sm0.5 Sr0.5 CoO3 –Sm0.2 Ce0.8 O1.9 (SSC-SDC) YSZ/SDC YSZ/GDC YSZ/GDC YSZ

YSZ/GDC

LSCF-GDC

LSC-GDC

LSM-YSZ

SrCo0.8 Fe0.1 Ga0.1 O3−δ

–

–

NiO–YSZ 50% H2 + 50% H2 O NiO–YSZ 60% H2 + 40% H2 O

NiO–YSZ 50% H2 + 50% H2 O

–

NiO–YSZ –

–

–

NiO–YSZ 50% H2 + 50% H2 O

YSZ/GDC

LSCF6482 NiO–YSZ –

0.5

NiO–YSZ 50% H2 + 50% H2 O

Fuel gas i (A composition cm−2 ) 1.5 V/ 700 °C

YSZ/GDC

Electrolyte/barrier Fuel layer electrode

LSCF1982

O2 -electrode

Table 2 Current density obtained with different oxygen electrodes

1.3 at 1.25 V

1.35 at 1.25 V

0.7

0.3

–

–

1.5

1.48

i (A cm−2 ) 1.5 V/ 800 °C

1.45

–

1.0

0.58

–

–

i (A cm−2 ) 1.5 V/ 850 °C

–

–

–

–

–

–

0.87

1.2

2.22

–

(continued)

Cathode supported, (Meng et al. 2016)

Cathode supported, (Ebbesen et al. 2014; Sun et al. 2013)

Cathode supported, (Ebbesen et al. 2014)

Cathode supported, (Ebbesen et al. 2014)

Cathode supported, (Jiang et al. 2014)

Cathode supported, (Choi et al. 2013)

Cathode supported, (Bo et al. 2008)

Cathode supported, (Bo et al. 2008)

Cathode supported, (Vibhu et al. 2019a, b)

Cathode supported, (Choi et al. 2013)

i (A Cell type/Ref cm−2 ) 1.5 V/ 900 °C

0.7 at 0.9 at 1.4 at – 1.25 V 1.25 V 1.25 V

-

0.5

0.6

–

–

–

0.85

i (A cm−2 ) 1.5 V/ 750 °C

76 V. Vibhu et al.

–

–

–

–

–

– – –

NiO–YSZ 50% H2 + 50% H2 O NiO–YSZ 50% H2 + 50% H2 O NiO–YSZ 50% H2 + 50% H2 O NiO–YSZ 50% H2 + 50% H2 O NiO–YSZ 50% H2 + 50% H2 O NiO–YSZ 50% H2 + 50% H2 O NiO–YSZ 50% H2 + 50% H2 O

YSZ/GDC

YSZ/GDC

YSZ/GDC

YSZ/GDC

YSZ/GDC YSZ/GDC YSZ/GDC

La2 NiO4+δ (LNO)

La2 Ni0.8 Co0.2 O4+δ (LNCO20)

Pr2 NiO4+δ (PNO)

Pr2 Ni0.8 Co0.2 O4+δ (PNCO20)

Nd2 NiO4+δ (PNO)

Nd2 Ni0.8 Co0.2 O4+δ (NNCO20)

La1.5 Pr0.5 NiO4+δ (LPNO)

–

–

–

–

–

–

0.5

i (A cm−2 ) 1.5 V/ 750 °C

NiO-GDC 90% H2 + 10% H2 O

Fuel gas i (A composition cm−2 ) 1.5 V/ 700 °C

TZ3Y

Electrolyte/barrier Fuel layer electrode

Nd2 NiO4+δ

O2 -electrode

Table 2 (continued)

1.60

1.80

1.62

1.90

1.62

1.6

1.51

0.9

i (A cm−2 ) 1.5 V/ 800 °C

–

–

–

–

–

–

–

1.45

i (A cm−2 ) 1.5 V/ 850 °C

–

–

–

3.0

2.1

2.3

2.0

–

(continued)

Cathode supported, (Vibhu et al. 2021b)

Cathode supported, (Vibhu et al. 2021b)

Cathode supported, (Vibhu et al. 2021b)

Cathode supported, (Vibhu et al. 2021b, 2021a)

Cathode supported, (Vibhu et al. 2021b, 2021a)

Cathode supported, (Vibhu et al. 2021b, 2019a, b)

Cathode supported, (Vibhu et al. 2021b, 2019a, b)

Electrolyte supported, (Chauveau et al. 2010)

i (A Cell type/Ref cm−2 ) 1.5 V/ 900 °C

Oxygen Electrode Materials for Solid Oxide Electrolysis Cells (SOECs) 77

– – 0.8

NiO–YSZ 50% H2 + 50% H2 O NiO–YSZ 50% H2 + 50% H2 O NiO–YSZ 60% H2 + 40% H2 O

YSZ/GDC YSZ

LSC infiltrated LSCF6482

LSCN infiltrated GDC

Fuel gas i (A composition cm−2 ) 1.5 V/ 700 °C

YSZ/GDC

Electrolyte/barrier Fuel layer electrode

La1.5 Pr0.5 Ni0.8 Co0.2 O4+δ (LPNCO20)

O2 -electrode

Table 2 (continued)

1.2

–

–

i (A cm−2 ) 1.5 V/ 750 °C

1.403

2.0

1.79

i (A cm−2 ) 1.5 V/ 800 °C

–

–

i (A cm−2 ) 1.5 V/ 850 °C

–

–

Cathode supported, (Tan et al. 2016)

Cathode supported, (Vibhu et al. 2019a, b)

Cathode supported, (Vibhu et al. 2021b)

i (A Cell type/Ref cm−2 ) 1.5 V/ 900 °C

78 V. Vibhu et al.

Oxygen Electrode Materials for Solid Oxide Electrolysis Cells (SOECs)

(a)

(d)

(b)

(e)

79

(c)

(f)

Fig. 9 Schematic illustrations of the microstructural change of the LSM oxygen electrode–YSZ electrolyte interface under SOEC operation conditions: a Oxygen migration from YSZ electrolyte to LSM grain bulk at the interface, b Local tensile strains within LSM particles due to the shrinkage of LSM lattice, c Local tensile strains induced microcrack formation, d Formation of individual nanoparticles, e Propagation and continuous nanoparticles formation, and LSM grain is bonded to YSZ electrolyte through bridges along edge of the convex contact ring and f Formation of complete nanoparticle layers and delamination of the LSM oxygen electrode under high internal oxygen partial pressure at the interface (Chen and Jiang 2011)

H2 O of absolute humidity for 9000, 1770 and 1460 h, respectively (Daroukh et al. 2014). Cells 1 and 2 had a 5 μm-thick screen-printed diffusion barrier layer of GDC which was applied between the oxygen electrode and electrolyte, whereas for, cell 3 the GDC layer was applied by PVD having a thickness of 0.75 μm. It was found that in cell 1, the degradation rate was about 2.2% per 1000 h; in cell 2, the degradation rate increased to 3.4% per 1000 h; and in cell 3, the degradation rate was 2.6% per 1000 h. After post-mortem analyses, it was confirmed that there was no change in the microstructure of the electrode material. Some small diffraction peaks were observed (except cell 3) in the XRD patterns after the electrolysis test, indicating the formation of Co3 O4 . The formation of SrZrO3 was also observed between 8YSZ had GDC in cells 1 and 2, as shown in Fig. 10. This formation was not observed in cell 3. SrZrO3 is clearly formed because of the diffusion of Zr from the electrolyte and Sr from the oxygen electrode through the pores of the GDC layer. The formation of SrZrO3 could be the possible reason for the increase of ohmic resistance and as such the degradation of cells 1 and 2. The formation of SrZrO3 was not observed in cell 3, in which the PVD-coated GDC layer seems therefore the most efficient. Cell 1 shows pore formation in the electrolyte parallel to the layer surface particularly close to the 8YSZ–GDC interface and grain boundaries after 9000 h of operation (Fig. 11). These phenomena also contribute to the increase of ohmic resistance and further lead to the degradation of the cell. In the case of physical vapor deposition (PVD)-coated GDC, pore formation was not observed.

80

V. Vibhu et al.

Fig. 10 Images of fracture surfaces of diffusion barrier layers of cells 1, 2 and 3 (a, b and c, respectively) (Daroukh et al. 2014)

Fig. 11 Images of fracture surfaces of electrolytes of cells 1, 2 and 3 (a, b and c, respectively) (Daroukh et al. 2014)

The stability of the LSCF electrode with GDC interlayer has been reported by Kim et al. (2014). The single cells were prepared by the green tapes of the GDC interlayer, YSZ electrolyte, NiO-YSZ fuel electrode and NiO-YSZ support. The layers were laminated under a high pressure (650 MPa) at 60 °C for ~ 15 min. The laminated tapes were cut into a circular shape and then co-fired at 1300, 1350, 1400 and 1500 °C for 5 h. LSCF paste was then applied by screen printing and fired at 1040 °C for 2 h. The single cells were tested with an anodic polarization of − 800 mA cm−2 in 80% H2 O + 20% H2 at 800 °C. The cell without an interlayer showed a large increase in ohmic area specific resistance (ASR) from 0.09 to 0.16 T.cm2 with little delamination during 50 h of operation, possibly due to the delamination between LSCF and YSZ. On the other hand, the cell with a GDC interlayer showed a slower increase in ohmic ASR from 0.14 to 0.153 T.cm2 with little delamination. The electrode polarization ASR of the cell with the interlayer also had a similar slower change, from 0.16 to 0.185 T.cm2 compared to that of the cell without an interlayer from 0.143 to 0.35 T.cm2 after 50 h. The cell co-fired at 1400 °C had the smallest and most stable value; it increased from ~ 0.41 to 0.50 T.cm2 for 100 h (degradation rate ~ 22% for 100 h), which was attributed to an increase in the electrode polarization resistance. A clear change in the XRD pattern was also observed. The LSCF before the SOEC test shows mostly a rhombohedral phase, but after the SOEC test, LSCF had mostly

Oxygen Electrode Materials for Solid Oxide Electrolysis Cells (SOECs)

81

a cubic phase. The increase in electrode polarization resistance could be attributed to the phase change of LSCF from rhombohedral to cubic. The degradation phenomenon of Ni-YSZ//YSZ//GDC//LSCF single cells was also studied by Tietz et al. up to 9000 h under electrolysis mode with a current density of − 1 A cm−2 (Tietz et al. 2013). An increase in cell voltage of about 40 mV kh−1 , i.e., an overall voltage degradation rate of 3.8% kh−1 was observed. Further, post-mortem analysis revealed several changes in the cell materials, for example, (a) structuring of grain surfaces in the electrolyte, void formation along grain boundaries, (b) horizontally aligned pores at the 8YSZ–GDC interface, (c) mass transport and formation of a dense layer at the 8YSZ–GDC interface, (d) compositional change in the LSCF electrode, (e) re-crystallization of LSCF in the electrochemically active area, (f) destabilization of the 8YSZ–Ni-YSZ interface and g) varying surface roughness of Ni particles. In view of these degradation phenomena, a small increase in the cell voltage (3% kh−1 ) was observed. A few authors have also studied the composite electrode. For example, Zheng et al. have recently reported the quantitative electrochemical contribution of cells and stacked interfacial contacts in SOECs with an H2 O/H2 ratio of 90/10 at 750 °C (Zheng et al. 2016). The single cell was prepared using a LSCF-GDC composite oxygen electrode in the following configuration Ni-YSZ//YSZ//GDC//LSCF-GDC. Further, the single-cell stack (only one ceramic cell (electrolyte + two electrodes), also including interconnects and sealing) was prepared using an interconnect, air electrode current collector and hydrogen electrode current collector. It was observed that the ohmic resistances of the cell, hydrogen electrode contact and air electrode contact were 63.6%, 4.8% and 30.9%, respectively, of those of the stack under OCV condition. Furthermore, their polarization resistance accounted for 93%, 2.6% and 3.5% of those of the stack, respectively. When the stack was operated at a current density of 0.8 A cm−2 for 384 h, the voltage degradation of the cells, hydrogen electrode contact and air electrode contact were 71.5%, 8.9% and 19.6%, respectively, of the total voltage degradation of the stack. Here, the increase in the stack resistance is due to the increase in ohmic resistance of the cell caused by the air electrode delamination from GDC. Moreover, the applied voltage of the stack was more sensitive to the cell polarization resistance caused by Ni particle agglomeration in the hydrogen electrode. The agglomeration of Ni resulted in a decrease in the activity of Ni-YSZ hydrogen electrode. Therefore, delamination of the LSCF-GDC oxygen electrode and Ni particle agglomeration in the hydrogen electrode were mainly responsible for the significant increase in the cell degradation. The effect of air electrode contact resistance and hydrogen electrode contact resistance showed minimal effect on the cell degradation. The comparison of performance and degradation of solid oxide electrolysis cells in stacks with different composite air electrodes (LSM-YSZ, LSC-GDC and LSCFGDC) was previously reported by Zheng et al. (2015). The three-fuel electrodesupported, large-scale (10 × 10 cm) SOECs, namely LSM-YSZ, LSC-GDC and LSCF-GDC, were compared for performance and degradation in the same three cell stack with an H2 O/H2 ratio of 90/10. The three cells were operated at a constant current density of −0.5 A cm−2 for 640 h during hydrogen production at 750 °C.

82

V. Vibhu et al.

When an electrolysis voltage of 1.1 V was applied, the LSC cell showed a hydrogen production rate of 218 NmL cm−2 ·h−1 , whereas the LSCF and LSM cells showed hydrogen production rates of 152.2 and 99.5 NmL cm−2 ·h−1 , respectively. The degradation rates were 4.89, 1.27 and 4.33 kh−1 for the LSM, LSC and LSCF cells, respectively. The LSC cell showed the highest, whereas the LSM cell showed the least electrolysis efficiency. The electrolysis efficiency decreased from 93.3% to 86.7%, 116.8% to 115% and 109.7% to 102.6% for the LSM, LSC and LSCF cells, respectively, after 640 h. After operation, the large single cells were equally divided into subparts from the two diagonal parts with the steam or air inlet area and steam or air outlet area for post-mortem analysis. Delamination mainly was observed at the steam and air inlet areas of the LSM and LSC cells, and delamination in the steam inlet area was more severe than that in the air inlet area. No delamination occurred at the interface of the LSCF cell. Agglomeration of Ni in the Ni-YSZ hydrogen electrode was also observed in the LSM and LSCF cells, which reduced the concentration of TPB sites for electron transfer. Oxygen electrodes were also prepared with some other perovskite material, for example, Sm0.5 Sr0.5 CoO3 (SSC) composite electrode with SDC (Jiang et al. 2014). It was observed that at 1.3 V applied electrolysis voltage, the maximum consumed current density increased from 86.9 mA cm−2 at 700 °C to 806 mA cm−2 at 850 °C. The hydrogen production rate was 327 mL cm−2 .h−1 at 850 °C with the electrolysis voltage of 1.3 V and 40% steam concentration. Among various weight ratios of SSC and SDC (7:3, 6:4, 5:5, 4:6), the most promising result was observed at a 7:3 ratio. Apart from the perovskite materials, a few double perovskite materials were also investigated as oxygen electrodes for SOECs. Liu et al. have reported the behavior of a single cell, the anode and cathode prepared by the same Sr2 Fe1.5 Mo0.5 O6-δ (SFM) material (Liu et al. 2010a). LSGM was used as the electrolyte material. They achieved an electrolysis current of −0.88 A.cm−2 and a hydrogen production rate of 380 mL cm−2 h at 900 °C with an electrolysis voltage of 1.3 V and 60 vol % AH (absolute humidity). Cell behavior during long term was even better than the cell prepared with an LSM electrode. Li et al. have reported YSZ-SFM composite oxygen electrodes fabricated by impregnation of SFM into a YSZ scaffold (Li et al. 2014). YSZ-SFM 20% loading showed the best result. For a voltage of 1.2 V, the electrolysis current was −223, −327 and −310 mA cm−2 at 750 °C for YSZ-SFM10, YSZ-SFM20 and YSZ-SFM30 oxygen electrodes, respectively. A hydrogen production rate of 11.46 NL h−1 has been achieved for SOEC with the YSZ-SFM20 electrode at 750 °C. A single cell with SmBaCo2 O5+δ oxygen electrode material showed an area specific resistance of 0.13 T.cm2 and 0.024 T.cm2 at 800 and 900 °C, respectively (Wei et al. 2013). Also, this material showed better behavior under anodic polarization. A relatively stable polarization was obtained at 900 °C for 20 h with an anodic current density of − 200 mA cm−2 . In addition to the above-mentioned materials, lanthanide nickelates (Ln2 NiO4+δ , Ln = La, Pr or Nd) have also shown promising electrochemical properties under SOEC operation. Chauveau et al. have tested Nd2 NiO4+δ material on electrolytesupported single cells (Chauveau et al. 2010). For an electrolysis voltage of 1.3 V,

Oxygen Electrode Materials for Solid Oxide Electrolysis Cells (SOECs)

83

current densities of 0.40, 0.64 and 0.87 A cm−2 were measured at 750, 800 and 850 °C, respectively. These values were much higher than the ones measured in the same conditions for the LSM containing cells. The current density was 1.7, 3 and 4.2 times higher than LSM at 850, 800 and 750 °C, respectively. Ogier et al. have also reported the behavior of Ln2 NiO4+δ (Ln = La, Pr or Nd) as the oxygen electrode for SOEC and compared it with LSCF (Ogier et al. 2015). These materials were tested as half-cells under anodic polarization. Pr2 NiO4+δ -based cells exhibited the highest performances at 600–800 °C temperature range. At the anodic overpotential of 0.15 V, the current densities measured using Pr2 NiO4+δ -based cells were more than ten times higher than those obtained with an LSCF cell. These nickelates are very promising materials, but chemical stability is an issue, especially for Pr2 NiO4+δ . Accordingly, to enhance the chemical stability, researchers have tried doping at A and B sites of the A2 BO4 structure, for example, Sr-doping at the A-site and Co-doping at the B site of La2 NiO4+δ , La2-x Srx Co0.5 Ni0.5 O4±δ (LSCN). Laguna-Bercero et al. have reported the performance of LSCN as an oxygen electrode and found that the material is a good candidate for both SOFC and SOEC (LagunaBercero et al. 2011). Stability tests were performed under SOEC conditions with single cells containing several nickelate electrodes at 800 °C with 50% H2 O and 50% H2 at a current density of − 1.0 A cm−2 up to 250 h (Vibhu et al. 2021b). Remarkably, the single cells containing Co-doped nickelates as the oxygen electrode show less degradation compared to parent nickelates. A large degradation rate is observed for LNO (113 mV/kh) and NNO (140 mV/kh) cells. The PNO cell shows similar degradation rate as that of a commercial LSCF cell; however, it exhibits a lower degradation rate compared to LNO and NNO cells. The most stable performance is observed with PNCO20 and LPNO single cells with an estimated degradation rate of ~ 20 mV/kh. The performance of a La0.8 Sr0.2 Co0.8 Ni0.2 O3-δ (LSCN) impregnated electrode in a GDC scaffold has been recently reported (Tan et al. 2016). The best result was obtained by 30% LSCN loading. A high hydrogen production rate of 484 mL cm−2 ·h−1 was obtained at 750 °C at 1.5 V with 60% AH. Also, the single cell showed stable behavior up to 100 h without degradation.

7 Conclusion and Future Prospects In summary, this chapter highlighted the recent advances in oxygen electrode materials for SOECs. First of all, there is clearly a lack of knowledge in this area in the SOEC community, and most publications are limited to the works performed on perovskite-type materials, e.g., LSM, LSCF, LSC, LSF, etc. Few double perovskites and Ruddlesden-Popper phases have been also investigated. In general, the perovskite-type material cells show degradation in a few hours, mainly due to mass transport, compositional change of electrodes and delamination. Therefore, long-term operation is one of the challenging issues. Using existing oxygen electrode materials, further progress on the chemical stability can be made

84

V. Vibhu et al.

by applying dense GDC, LDC or SDC or PrDC (La-, Sm- or Pr-doped ceria) barrier layers between electrode and electrolyte, to inhibit the formation of detrimental phases. Delamination of electrodes mainly occurs due to high oxygen partial pressure at TPB sites and in the bulk of oxygen electrodes (Chen and Jiang 2011; Virkar 2010). If one succeeds to remove such excess oxygen from the oxygen electrode, then it is possible to avoid delamination. One efficient way could be reversible operation between electrolysis and fuel cell modes (Graves et al. 2015). Infiltration is a promising method to form high-performing oxygen electrodes for SOECs. Ionic and electronic conductivity as well as the catalytic activity of the electrode significantly improve due to the presence of nanoparticles of high surface area. The fine particles are formed following decomposition of nitrate phases and single phases form at low heat treatment temperature. Conventional sintering requires much higher temperatures which leads to particle growth and reduction of TPBs. In addition, using infiltration, unfavorable phases which form due to the reaction of electrolyte and electrodes at high temperatures are avoided, giving the flexibility of more materials choice. Finally, due to the improved cell performance, the operation temperatures can be lowered which can improve cell stability. However, in this technique, the stability issue due to particle growth during sintering needs to be addressed. One effective method can be co-infiltration of multiple components where the minor phase covers the major phase and inhibits its growth. This method has shown its effectiveness in controlling the particle size of the nickel phase coinfiltrated with doped ceria. (Hanifi et al. 2014a, b; Klemensø et al. 2010). Another effective method to improve the stability of the infiltrated cells is to heat treat them at temperatures higher than the operating temperature while still maintaining the nanostructure of the infiltrates. During infiltration, some complex phases such as nickelate (Pr2 NiO4+δ and Nd2 NiO4+δ ) require heat treatment at high temperatures (>1000 °C) to form a single phase. However, when the calcination is carried out at lower temperature (600–800 °C), multiple phases including higher-order RP phases are obtained. Such a multiple-phase electrode does not show the expected performance and stability of the single phase. High-temperature heat treatment to get the single phase can lead to reaction between nickelates and Zr and formation of nonconductive phases such as Nd2 Zr2 O7 . It is believed that future efforts by the fuel cell community will focus on developing infiltrated oxygen electrodes particularly nickelates with high performance and stability at lower temperatures. In addition, it will be worth trying other oxygen electrode materials which have not been used as oxygen electrodes in SOEC, for example, double perovskites with high k* and D* value (e.g., PrBaCo2 O5+δ , GdBaCo2 O5+δ ), chemically stable Ruddlesden-popper phase (e.g., Pr4 Ni3 O10+δ ) (Vibhu et al. 2016) in order to eliminate the electrochemical cell degradation.

Oxygen Electrode Materials for Solid Oxide Electrolysis Cells (SOECs)

85

References Adler SB (2000) Limitations of charge-transfer models for mixed-conducting oxygen electrodes. Solid State Ionics 135:603–612 Adler SB, Lane JA, Steele BCH (1996) Electrode kinetics of porous mixed-conducting oxygen electrodes. J Electrochem Soc 143:3554–3564 Adler SB (2004) Factors governing oxygen reduction in solid oxide fuel cell cathodes. Chem Rev 104:4791–4844 Bidrawn F, Kim G, Corre G, Irvine JTS, Vohs JM, Gorte RJ (2008) Efficient reduction of CO[sub 2] in a solid oxide electrolyzer. Electrochem Solid-State Lett 11:B167 Bo Y, Wenqiang Z, Jingming X, Jing C (2008) Microstructural characterization and electrochemical properties of Ba0.5Sr0.5Co0.8Fe0.2O3−δ and its application for anode of SOEC. Int J Hydrogen Energy 33:6873−77 Boukamp BA, Bouwmeester HJM (2003) Interpretation of the Gerischer impedance in solid state ionics. Solid State Ionics 157:29–33 Burye TE, Tang H, Nicholas JD (2016) The effect of precursor solution desiccation or nano-ceria pre-infiltration on various La0.6Sr0.4FeyCo1-yO3-xinfiltrate compositions. J Electrochem Soc 163:F1017–F1022 Carlson RL, Kardomateas GA, Craig JI (2012) Ceramic Materials Chauveau F, Mougin J, Bassat JM, Mauvy F, Grenier JC (2010) A new anode material for solid oxide electrolyser: the neodymium nickelate Nd2NiO4 +δ. J Power Sour 195:744–749 Chen K, Jiang SP (2011) Failure mechanism of (La, Sr)MnO3 oxygen electrodes of solid oxide electrolysis cells. Int J Hydrogen Energy 36:10541–10549 Choi S, Yoo S, Shin J-Y, Kim G (2011) High Performance SOFC cathode prepared by infiltration of Lan + 1NinO3n + 1 (n = 1, 2, and 3) in Porous YSZ. J Electrochem Soc 158:B995 Choi M-B, Singh B, Wachsman ED, Song S-J (2013) Performance of La0.1Sr0.9Co0.8Fe0.2O3−δ and La0.1Sr0.9Co0.8Fe0.2O3−δ–Ce0.9Gd0.1O2 oxygen electrodes with Ce0.9Gd0.1O2 barrier layer in reversible solid oxide fuel cells. J Power Sour 239:361–373 Chen T, Liu M, Yuan C, Zhou Y, Ye X, Zhan Z, Xia C, Wang S (2015) High performance of intermediate temperature solid oxide electrolysis cells using Nd2NiO4+δ impregnated scandia stabilized zirconia oxygen electrode. J Power Sour 276:1–6 Daroukh M, Tietz F, Sebold D, Buchkremer HP (2014) Post-test analysis of electrode-supported solid oxide electrolyser cells. Ionics 21:1039–1043 Ebbesen SD, Jensen SH, Hauch A, Mogensen MB (2014) High temperature electrolysis in alkaline cells, solid proton conducting cells, and solid oxide cells. Chem Rev 114:10697–10734 Eguchi K, Hatagishi T, Arai H (1996) Power generation and steam electrolysis characteristics of an electrochemical cell with a zirconia- or ceria-based electrolyte. Solid State Ionics 86–88:1245– 1249 Gamble S (2011) Fabrication–microstructure–performance relationships of reversible solid oxide fuel cell electrodes–review. Mater Sci Technol 27:1485–1497 Gao Z, Mogni LV, Miller EC, Railsback JG, Barnett SA (2016) A perspective on low-temperature solid oxide fuel cells. Energy Environ Sci Gerischer H (1951) Alternating-current polarization of electrodes with a potential-determining step for equilibrium potential. Z Phys Chem 198:286 Gómez SY, Gurauskis J, Øygarden V, Hotza D, Grande T, Wiik K (2016) Synthesis and oxygen transport properties of La2−ySryNi1−xMoxO4+δ. Solid State Ionics 292:38–44 Graves C, Ebbesen SD, Jensen SH, Simonsen SB, Mogensen MB (2015) Eliminating degradation in solid oxide electrochemical cells by reversible operation. Nat Mater 14:239–244 Hanifi AR, Torabi A, Shinbine A, Etsell TH, Sarkar P (2011a) Fabrication of thin porous electrolytesupported tubular fuel cells using slip casting. J Ceram Process Res 12:336–342 Hanifi AR, Torabi A, Etsell TH, Yamarte L, Sarkar P (2011b) Porous electrolyte-supported tubular micro-SOFC design. Solid State Ionics 192:368–371

86

V. Vibhu et al.

Hanifi AR, Shinbine A, Etsell TH, Sarkar P (2012a) Development of monolithic YSZ porous and dense layers through multiple slip casting for ceramic fuel cell applications. Int J Appl Ceram Technol 9:1011–1021 Hanifi AR, Zazulak M, Etsell TH, Sarkar P (2012b) Effects of calcination and milling on surface properties, rheological behaviour and microstructure of 8mol% yttria-stabilised zirconia (8 YSZ). Powder Technol 231:35–43 Hanifi AR, Laguna-Bercero MA, Etsell TH, Sarkar P (2014a) The effect of electrode infiltration on the performance of tubular solid oxide fuel cells under electrolysis and fuel cell modes. Int J Hydrogen Energy 39:8002–8008 Hanifi AR, Paulson S, Torabi A, Shinbine A, Tucker MC, Birss V, Etsell TH, Sarkar P (2014b) Slip-cast and hot-solution infiltrated porous yttria stabilized zirconia (YSZ) supported tubular fuel cells. J Power Sour 266:121–131 Hanifi AR, Zahiri B, Mitlin D, Vincent AL, Etsell TH, Sarkar P (2016) Effects of washing and calcination–milling on ionic release and surface properties of yttria stabilized zirconia. Ceram Int 42:6755–6760 He F, Song D, Peng R, Meng G, Yang S (2010) Electrode performance and analysis of reversible solid oxide fuel cells with proton conducting electrolyte of BaCe0.5Zr0.3Y0.2O3−δ. J Power Sour 195:3359–3364 Jensen SH, Larsen PH, Mogensen M (2007) Hydrogen and synthetic fuel production from renewable energy sources. Int J Hydrogen Energy 32:3253–3257 Jiang SP (2012) Nanoscale and nano-structured electrodes of solid oxide fuel cells by infiltration: advances and challenges. Int J Hydrogen Energy 37:449–470 Jiang W, Lü Z, Wei B, Wang ZH, Zhu XB, Tian YT, Huang XQ, Su WH (2014) Sm0.5Sr0.5CoO3Sm0.2Ce 0.8O1.9 composite oxygen electrodes for solid oxide electrolysis cells. Fuel Cells 14:76–82 Kaiser A, Monreal E, Stolten D (1997) Preparation techniques and materials for long term stable SOFC—single cell membranes. Ionics 3:143–148 Kim-Lohsoontorn P, Brett DJL, Laosiripojana N, Kim YM, Bae JM (2010) Performance of solid oxide electrolysis cells based on composite La0.8Sr0.2MnO3 −δ–yttria stabilized zirconia and Ba0.5Sr0.5Co0.8Fe0.2O3−δ oxygen electrodes. Int J Hydrogen Energy 35:3958–3966 Kim SJ, Choi GM (2014) Stability of LSCF electrode with GDC interlayer in YSZ-based solid oxide electrolysis cell. Solid State Ionics 262:303–306 Klemensø T, Thydén K, Chen M, Wang H-J (2010) Stability of Ni–yttria stabilized zirconia anodes based on Ni-impregnation. J Power Sour 195:7295–7301 Laguna-Bercero MA, Hanifi AR, Etsell TH, Sarkar P, Orera VM (2015) Microtubular solid oxide fuel cells with lanthanum strontium manganite infiltrated cathodes. Int J Hydrogen Energy 40:5469–5474 Laguna-Bercero MA, Kinadjan N, Sayers R, El Shinawi H, Greaves C, Skinner SJ (2011) Performance of La2–xSrxCo0.5Ni0.5O4±δ as an oxygen electrode for solid oxide reversible cells. Fuel Cells 11:102–107 Laguna-Bercero MA, Hanifi AR, Monzón H, Cunningham J, Etsell TH, Sarkar P (2014) High performance of microtubular solid oxide fuel cells using Nd2NiO4+δ-based composite cathodes. J Mater Chem A 2:9764–9770 Li J, Zhong C, Meng X, Wu H, Nie H, Zhan Z, Wang S (2014) Sr2Fe1.5Mo0.5O6δ−Zr0.84Y0.16O2-δ materials as oxygen electrodes for solid oxide electrolysis cells. Fuel Cells 14:1046–1049 Li Y, Zhang S-L, Li C-X, Wei T, Yang G-J, Li C-J, Liu M (2016) La2NiO4+δ infiltration of plasma-sprayed LSCF coating for cathode performance improvement. J Therm Spray Technol 25:392–400 Liang M, Bo Y, Wen M, Chen J, Jingming X, Zhai Y (2009) Preparation of LSM–YSZ composite powder for anode of solid oxide electrolysis cell and its activation mechanism. J Power Sour 190:341–345

Oxygen Electrode Materials for Solid Oxide Electrolysis Cells (SOECs)

87

Liu Q, Yang C, Dong X, Chen F (2010a) Perovskite Sr2Fe1.5Mo0.5O 6-δ as electrode materials for symmetrical solid oxide electrolysis cells. Int J Hydrogen Energy 35:10039–10044 Liu Q, Yang C, Dong X, Chen F (2010b) Perovskite Sr2Fe1.5Mo0.5O6−δ as electrode materials for symmetrical solid oxide electrolysis cells. Int J Hydrogen Energy 35:10039–10044 Meng X, Shen Y, Xie M, Yin Y, Yang N, Ma Z-F, João C, da Costa D, Liu S (2016) Novel solid oxide cells with SrCo0.8Fe0.1Ga0.1O3−δ oxygen electrode for flexible power generation and hydrogen production. J Power Sour 306:226–232 Navarrete L, Solís C, Serra JM (2015) Boosting the oxygen reduction reaction mechanisms in IT-SOFC cathodes by catalytic functionalization. J Mater Chem A 3:16440–16444 Ni M (2009) Computational fluid dynamics modeling of a solid oxide electrolyzer cell for hydrogen production. Int J Hydrogen Energy 34:7795–7806 Nicollet C, Flura A, Vibhu V, Rougier A, Bassat JM, Grenier JC (2015) La2NiO4 +δ infiltrated into gadolinium doped ceria as novel solid oxide fuel cell cathodes: electrochemical performance and impedance modelling. J Power Sour 294:473–482 Nicollet C, Flura A, Vibhu V, Rougier A, Bassat J-M, Grenier J-C (2016) An innovative efficient oxygen electrode for SOFC-Pr6O11 infiltrated into Gd-doped ceria backbone. Int J Hydrogen Energy Ogier T, Mauvy F, Bassat JM, Laurencin J, Mougin J, Grenier JC (2015) Overstoichiometric oxides Ln2NiO4+δ (Ln = La, Pr or Nd) as oxygen anodic electrodes for solid oxide electrolysis application. Int J Hydrogen Energy 40:15885–15892 Park J, Kim D, Baek J, Yoon Y-J, Pei-Chen S, Lee S (2018) Numerical study on electrochemical performance of low-temperature micro-solid oxide fuel cells with submicron platinum electrodes. Energies 11:1204 Patcharavorachot Y, Arpornwichanop A, Chuachuensuk A (2008) Electrochemical study of a planar solid oxide fuel cell: role of support structures. J Power Sour 177:254–261 Patro PK, Delahaye T, Bouyer E (2010) Development of Pr0.58Sr0.4Fe0.8Co0.2O3−δ–GDC composite cathode for solid oxide fuel cell (SOFC) application. Solid State Ionics 181:1378– 1386 Prestat M, Morandi A, Heel A, Holzer L, Holtappels P, Graule TJ (2010) Effect of graphite pore former on oxygen electrodes prepared with La0.6Sr0.4CoO3−δ nanoparticles. Electrochem Commun 12:292–295 Samson A, Søgaard M, Knibbe R, Bonanos N (2011) High performance cathodes for solid oxide fuel cells prepared by infiltration of La0.6Sr0.4CoO3−δ into Gd-Doped Ceria. J Electrochem Soc 158:B650–B659 Schiller G, Ansar A, Lang M, Patz O (2009) High temperature water electrolysis using metal supported solid oxide electrolyser cells (SOEC). J Appl Electrochem 39:293–301 Shah M, Barnett SA (2008) Solid oxide fuel cell cathodes by infiltration of La0.6Sr0.4Co0.2Fe0.8O3−δ into Gd-Doped Ceria. Solid State Ionics 179:2059–2064 Shahrokhi S, Babaei A, Zamani C (2018) Reversible operation of La0·8Sr0·2MnO3 oxygen electrode infiltrated with Ruddlesden-Popper and perovskite lanthanum nickel cobaltite. Int J Hydrogen Energy 43:23091–23100 Sholklapper TZ, Jacobson CP, Visco SJ, De Jonghe LC (2008) Synthesis of dispersed and contiguous nanoparticles in solid oxide fuel cell electrodes. Fuel Cells 8:303–312 Sholklapper TZ, Chun L, Jacobson CP, Visco SJ, De Jonghe LC (2006) LSM-infiltrated solid oxide fuel cell cathodes. Electrochem Solid-State Lett 9:A376 Sholklapper TZ, Radmilovic V, Jacobson CP, Visco SJ, De Jonghe LC (2007) Synthesis and stability of a nanoparticle-infiltrated solid oxide fuel cell electrode. Electrochem Solid-State Lett 10:B74 Singhal SC, Kendall K (2003) Introduction to SOFCs. In: Singhal SC, Kendall K (eds) High temperature and solid oxide fuel cells, Elsevier Science, Amsterdam Song Y, Zhang X, Zhou Y, Jiang Q, Guan F, Lv H, Wang G, Bao X (2018) Promoting oxygen evolution reaction by RuO2 nanoparticles in solid oxide CO2 electrolyzer. Energy Storage Mater 13:207–214

88

V. Vibhu et al.

Sun X, Chen M, Liu Y-L, Hjalmarsson P, Ebbesen SD, Jensen SH, Mogensen MB, Hendriksen PV (2013) Durability of solid oxide electrolysis cells for syngas production. J Electrochem Soc 160:F1074–F1080 Tan Y, Duan N, Wang A, Yan D, Chi B, Wang N, Pu J, Li J (2016) Performance enhancement of solution impregnated nanostructured La0.8Sr0.2Co0.8Ni0.2O3-δ oxygen electrode for intermediate temperature solid oxide electrolysis cells. J Power Sour 305:168–174 Tao Y, Nishino H, Ashidate S, Kokubo H, Watanabe M, Uchida H (2009) Polarization properties of La0.6Sr0.4Co0.2Fe0.8O3-based double layer-type oxygen electrodes for reversible SOFCs. Electrochim Acta 54:3309–3315 Tietz F, Sebold D, Brisse A, Schefold J (2013) Degradation phenomena in a solid oxide electrolysis cell after 9000 h of operation. J Power Sour 223:129–135 Torabi A, Hanifi AR, Etsell TH, Sarkar P (2011) Effects of porous support microstructure on performance of infiltrated electrodes in solid oxide fuel cells. J Electrochem Soc 159:B201–B210 Vibhu V, Vinke IC, Eichel RA, Bassat JM, de Haart LGJ (2019a) La2 Ni1−xCoxO4 +δ (x = 0.0, 0.1 and 0.2) based efficient oxygen electrode materials for solid oxide electrolysis cells. J Power Sour 444:227292 Vibhu V, Vinke IC, Eichel RA, de Haart LGJ (2021a) Cobalt substituted Pr2Ni1-xCoxO4+δ (x = 0, 0.1, 0.2) oxygen electrodes: impact on electrochemical performance and durability of solid oxide electrolysis cells. J Power Sour 482:228909 Vibhu V, Yildiz S, Vinke IC, Eichel R-A, Bassat J-M, de Haart LGJ (2019b) High performance LSC infiltrated LSCF oxygen electrode for high temperature steam electrolysis application. J Electrochem Soc 166:F102–F108 Vibhu V, Rougier A, Nicollet C, Flura A, Fourcade S, Penin N, Grenier J-C, Bassat J-M (2016) Pr4Ni3O10+δ: a new promising oxygen electrode material for solid oxide fuel cells. J Power Sour 317:184–193 Vibhu V, Vinke IC, Eichel R-A, De Haart LGJ (2021b) Performance and stability of nickelates based oxygen electrodes for solid oxide cells. ECS Trans 103:1505–1515 Vincent AL, Hanifi AR, Luo J-L, Chuang KT, Sanger AR, Etsell TH, Sarkar P (2012) Porous YSZ impregnated with La0.4Sr0.5Ba0.1TiO3 as a possible composite anode for SOFCs fueled with sour feeds. J Power Sour 215:301–306 Vincent AL, Hanifi AR, Zazulak M, Luo J-L, Chuang KT, Sanger AR, Etsell T, Sarkar P (2013) Preparation and characterization of an solid oxide fuel cell tubular cell for direct use with sour gas. J Power Sour 240:411–416 Virkar AV (2010) Mechanism of oxygen electrode delamination in solid oxide electrolyzer cells. Int J Hydrogen Energy 35:9527–9543 Vohs JM, Gorte RJ (2009) High-performance SOFC cathodes prepared by infiltration. Adv Mater 21:943–956 Wang W, Gorte RJ, Vohs JM (2008) Analysis of the performance of the electrodes in a natural gas assisted steam electrolysis cell. Chem Eng Sci 63:765–769 Wang W, Huang Y, Jung S, Vohs JM, Gorte RJ (2006) A comparison of LSM, LSF, and LSCo for solid oxide electrolyzer anodes. J Electrochem Soc 153:A2066 Wang Z, Mori M, Araki T (2010) Steam electrolysis performance of intermediate-temperature solid oxide electrolysis cell and efficiency of hydrogen production system at 300 Nm3 h−1. Int J Hydrogen Energy 35:4451–4458 Wei B, Chen K, Zhao L, Ai N, Lü Z, Jiang SP (2013) SmBaCo2 O5 -δ as high efficient oxygen electrode of solid oxide electrolysis cells. In: ECS Transactions, pp 3189–96 Yamamoto O, Takeda Y, Kanno R, Noda M (1987) Perovskite-type oxides as oxygen electrodes for high temperature oxide fuel cells. Solid State Ionics 22:241–246 Yamamoto O (2000) Solid oxide fuel cells: fundamental aspects and prospects. Electrochim Acta 45:2423–2435 Zheng Y, Li Q, Chen T, Xu C, Wang WG (2015) Quantitative contribution of resistance sources of components to stack performance for solid oxide electrolysis cells. J Power Sour 274:736–740

Oxygen Electrode Materials for Solid Oxide Electrolysis Cells (SOECs)

89

Zheng Y, Xu C, Wang WG, Guo L (2016) Quantitative electrochemical contributions of cells and stacked interfacial contacts in solid-oxide electrolysis cells. Int J Hydrogen Energy 41:4538– 4545 Zheng H, Tian Y, Zhang L, Chi B, Jian P, Jian L (2018) La0.8Sr0.2Co0.8Ni0.2O3-δ impregnated oxygen electrode for H2 O/CO2 co-electrolysis in solid oxide electrolysis cells. J Power Sour 383:93–101 Zhang S-L, Wang H, Lu MY, Li C-X, Li C-J, Barnett SA (2019) Electrochemical performance and stability of SrTi0.3Fe0.6Co0.1O3-δ infiltrated La0.8Sr0.2MnO3 Zr0.92Y0.16O2-δ oxygen electrodes for intermediate-temperature solid oxide electrochemical cells. J Power Sour 426:233–241

Fuel Electrode Materials for Solid Oxide Electrolysis Cells (SOECs) Muhammad Shirjeel Khan and Ruth Knibbe

1 Introduction During high-temperature electrolysis, the fuel electrode is where H2 O (and/or CO2 ) is reduced to H2 (and/or CO), respectively. As will be discussed in Sect. 2, this becomes a little more complicated for co-electrolysis. The first part of this chapter explores the fabrication of Ni-based cermet fuel electrodes. This is followed by a detailed discussion on H2 O electrolysis reaction mechanisms and associated degradation and CO2 and co-electrolysis reaction mechanism and associated degradation. In the third and fourth part of this chapter, alternative cermet and perovskite fuel electrodes are discussed. Finally, the chapter ends with touching on established and emerging tools for characterization of SOEC fuel electrodes. The fuel electrode (cathode) is an important component of SOECs where the electrochemical reduction of steam (H2 O) and/or CO2 takes place. Therefore, the basic requirement of a fuel electrode is to provide sufficient catalytic sites and extended triple phase boundaries (TPBs) for the reduction reaction. In addition, the fuel electrode must possess high ionic and electronic conductivity to allow the diffusion of oxide ions and electrons, must be chemically stable in reducing (H2 and CO) as well as oxidizing environment (H2 O and CO2 ), must be mechanically compatible with other components such as electrolyte and interconnect to minimize stresses, must exhibit good redox stability to minimize damage (fracture) and must be cost effective for commercialization. The simplest reduction reactions occurring at the fuel electrode are

M. S. Khan Institute of Energy and Climate Research: Materials Synthesis and Processing (IEK-1), Forschungszentrum Jülich GmbH, 52425 Jülich, Germany R. Knibbe (B) School of Mechanical and Mining Engineering, University of Queensland, Brisbane, Australia e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M. A. Laguna-Bercero (ed.), High Temperature Electrolysis, Lecture Notes in Energy 95, https://doi.org/10.1007/978-3-031-22508-6_5

91

92

M. S. Khan and R. Knibbe

H2 O + 2e− → H2 + O−2

(1)

CO2 + 2e− → CO + O−2

(2)

Given the stringent requirements, there are limited materials that can be used as fuel electrodes. These can be broadly classified into Ni-based cermet fuel electrodes, Ni-free metal-based cermet fuel electrodes and perovskite-type oxide fuel electrodes. The first part of this chapter explores the fabrication of Ni-based cermet fuel electrodes. This is followed by a detailed discussion on H2 O electrolysis reaction mechanisms and associated degradation and CO2 and co-electrolysis reaction mechanism and associated degradation. In the third and fourth part of this chapter, alternative cermet and perovskite fuel electrodes are discussed. Finally, the chapter ends with touching on established and emerging tools for characterization of SOEC fuel electrodes.

1.1 Basic Requirements of Fuel Electrodes Various requirements in regards to catalytic activity, stability, conductivity, compatibility and microstructure must be met for a SOEC fuel electrode. • High electronic conductivity (σ e > 100 S·cm−1 ); • High ionic conductivity (σ i > 0.01 S·cm−1 ); • The thermal expansion coefficient (TEC) should match with the other components of the cell, typically in the range 10 − 13 × 10–6 °C−1 ; • Chemical stability during high sintering temperature and with electrolyte; • Long term stability over SOEC operating conditions—wide pO2 range and relatively high temperatures (700–850 °C); • Sufficient porosity to supply gas reactions and remove gas products; • High catalytic activity for appropriate reduction reactions—CO2 and/or H2 O; • High triple phase boundary (TPB) density to ensure high catalytic activity; • Low cost.

1.2 Oxygen Partial Pressures in SOCs The electrochemical operation of SOCs impacts the oxygen partial pressures (pO2 ) in the electrodes and more specifically at the electrode–electrolyte interface. At open circuit voltage, the pO2 in the electrodes is defined by the gas composition, pressure and temperature. Assuming no gas leaking across, in or out of the SOC and no electronic short-circuiting through the electrolyte, the cell voltage is defined by the Nernst equation. However, as a current is drawn or applied across the SOC, the electrodes experience an overvoltage (η), which changes the average oxygen partial pressure

Fuel Electrode Materials for Solid Oxide Electrolysis Cells (SOECs)

93

Fig. 1 SOC schematics showing the electrical and oxygen partial pressure gradients in SOEC, OCV and SOFC mode (Knibbe et al. 2010)

inside, through and along the porous electrodes [36]. The new oxygen activity at the electrode–electrolyte interface can be estimated using the Nernst equation (Eq. 3, Fig. 1). pO2 ,interface =

pO2 ,bulk exp 4ηF RT

(3)

where η is the electrode overpotential (V), pO2,bulk is the oxygen partial pressure supplied to the electrode, pO2,interface is the pO2 at the electrode/electrolyte interface and F, R, T have the usual meaning. This means that during SOEC operation the pO2 at the fuel electrode–electrolyte interface becomes more negative, i.e., more reducing. The pO2 along with the elevated temperatures is key driving forces for degradation. The different oxygen activity drives degradation by impacting stable valence state, oxygen vacancy concentration, cation interdiffusion and destabilizing compounds or facilitating the formation of new phases. Not only does the pO2 change from the bulk gas composition to the interface, but also from the inlet to the outlet. As such, the pO2 distribution across and through the fuel electrode can be quite complex depending on many factors including initial gas composition, gas flow rate, temperature, electrode material and microstructure and current density. These changes make replicating work from other groups difficult if the cell setup is not identical. As such, these differences along with impurity differences (discussed in Sect. 2) can also have an impact of the ability to reproduce results between different laboratories.

2 Nickel-Based Cermet Fuel Electrodes Nickel is an excellent candidate for H2 O and CO2 reduction reactions because of its high electrocatalytic activity. However, it does not possess any oxide ion conductivity (McEvoy 2003; Couper et al. 1990; Cassidy et al. 2016). Therefore, it is usually mixed with oxide ion conducting materials such as yttria stabilized zirconia (YSZ),

94

M. S. Khan and R. Knibbe

gadolinium doped ceria (GDC) and scandium doped ceria (SDC) with a typical volume ratio of ca. 50–60 vol% Ni and 40–50 vol% YSZ, GDC or SDC. Such electrodes are known as Ni-based cermet electrodes. In addition to extending the TPB length, adding the oxide ion conducting material also improves the electrode– electrolyte thermomechanical compatibility and prevents Ni agglomeration (McEvoy 2003; Cassidy et al. 2016). Although Ni-YSZ was first used as a fuel electrode by Westinghouse electric in the 1990s, Ni-YSZ is still the most commonly used fuel electrode. This is due to its low cost, good chemical and mechanical compatibility with other components, good electrolysis current density, excellent long-term durability and good electronic conductivity. For example, the results from Dönitz et al. (Dönitz et al. 1988; Dönitz and Erdle 1985) have shown Faradaic efficiency for steam electrolysis to be close to 100% for a period of 1000 h. The Ni-YSZ fuel electrode has been used in high-temperature electrolysis either as a thin electrode (20–30 μm) on a dense and thick YSZ electrolyte (electrolytesupported configuration) or as a support material, where it is made several hundred micrometers thick (fuel electrode-supported configuration). For the electrolytesupported configuration, the thin fuel electrode is deposited on the YSZ electrolyte as a slurry by spray coating, dip coating or screen printing. The slurry used in these processes is a mixture of NiO and YSZ powders, binder, pore former and solvent. After deposition, the layer is sintered at 1300–1500 °C for 4–5 h. For the fuel electrode-supported configuration, NiO and YSZ powders along with the pore former, binder and solvents are mixed by ball milling followed by drying (Khan et al. 2018). For lab scale fuel electrode-supported cells, the dried powder is then compacted in the form of pellets, which are then sintered around 1000–1200 °C for up to 2 h. The cells are then coated with a cathode functional layer (CFL) (a NiO and YSZ mixture), followed by the YSZ electrolyte and then sintered at 1300– 1500 °C for 4–5 h. Once a full cell is produced, the NiO is reduced to Ni at high temperatures (more than 700 °C) by introducing H2 . To fabricate larger/commercial planar cells, the mixture is not completely dried, but applied as paste in the form layers using the tape casting technique (Schafbauer et al. 2014). The electrolyte (YSZ) is also applied as layers and the green cells are sintered as mentioned above. For tubular or flat-tubular cells, an extrusion method is used to fabricate the fuel electrode supports. These cells are then coated with the CFL and YSZ either by screen printing or dip coating, followed by sintering at high temperature (Khan et al. 2021). The Ni-YSZ fuel electrode contains a network of Ni particles, YSZ particles and pores. At the TPBs (i.e., the sites where these three phases meet each other), the electrolysis reaction takes place (Fig. 2). The gas supplied and produced during electrolysis are transferred to and from the TPBs through the interconnecting porous network, whereas the oxide ions and electrons travel through YSZ and Ni, respectively. Therefore, the size, fraction, distribution and percolation of the Ni, YSZ and pores particles affect the ionic and electronic conductivity as well as the gas flow and are consequently crucial for SOEC performance (e.g., power output and stability). Generally, the porosity of the fuel electrode support should be around 30–35%, while

Fuel Electrode Materials for Solid Oxide Electrolysis Cells (SOECs)

95

Fig. 2 Ni-YSZ fuel electrode after operation at − 2.0 A.cm−2 at 850 °C. top) micrograph of inlet region showing SiO2 deposits in the and around the Ni grains bottom) LV SEM reconstructed image. White—percolating Ni grains; light gray—non-percolating Ni; black—porosity and dark gray—YSZ. (Knibbe et al. 2010)

that of CFL should be around 15–20% after the NiO reduction to Ni (Ebbesen et al. 2015; Hauch et al. 2016).

2.1 Steam Electrolysis The overall steam reduction reaction occurs at the TPB sites in the fuel electrode (Eq. 1). However, this reaction is completed through a series of steps as described by Pan and co-workers (Pan et al. 2015): H2 O(g) → H2 O(ad)

(4)

+ H2 O(ad) → OH− (ad) + H(ad)

(5)

−2 + OH− (ad) → O(ad) + H(ad)

(6)

2− O−2 (ad) → OTPB

(7)

96

M. S. Khan and R. Knibbe − − O2− TPB → e + OTPB

(8)

− O− TPB → e + O(TPB)

(9)

2O(TPB) → O2

(10)

+ H+ (ad) → H(TPB)

(11)

− 2H+ (TPB) + 2e → H2

(12)

Here, Eqs. 4–7 illustrate the dissociative adsorption of steam at the Ni surface to Oad 2− and Had + , Eqs. 7–10 illustrate the formation of O2 , whereas Eqs. 11 and 12 describe the formation of H2 . The dissociative adsorption of H2 O(g) is slow, while its desorption is a very fast process. Therefore, the residential time is extremely small, which leads to its inadequate supply to the TPB region for the reduction reaction. This leads to an electrode polarization resistance increase as the current is increased. Other studies have also shown excellent performance for Ni-YSZ fuel electrodesupported cells with a YSZ electrolyte and LSM electrode. Replacing the YSZ in the fuel electrode with scandium stabilized zirconia (ScSZ), GDC and SDC has been investigated by different groups. This change is often accompanied by a change in YSZ electrolyte to ScSZ, GDC and SDC electrolyte. As these electrolytes have a higher electronic conductivity than YSZ, during steam electrolysis at high current densities, the current can leak through the electrolyte decreasing cell efficiency and leading to cell degradation. For the cells where the electrolyte is not changed, an increased performance and durability have been reported compared to Ni-YSZ (Uchida et al. 2004, 2007).

2.1.1

Degradation During Steam Electrolysis

Although the initial cell performance is similar in SOFC and SOEC modes, the NiYSZ electrode degrades faster under SOEC conditions. This is despite observations that applying a cathodic polarization, for a short period, can create nano structuring at the electrode–electrolyte interface which has been found to be beneficial to Ni-YSZ durability in SOFC applications (Hauch et al. 2018; Klotz et al. 2011). Depending on the strength of the cathodic bias, the overpotential can be a driver for multiple fuel electrode degradation mechanisms including changes in microstructure, composition and consequently TPB length (Hoerlein et al. 2018; Trini et al. 2020). This includes the redistribution and coarsening of Ni, Ni oxidation, ZrO2 nanoparticle formation and impurity accumulation. Among the various fuel electrode degradation phenomena, Ni transport is the most common. Under SOEC conditions, the oxygen partial pressure (pO2 ) increases

Fuel Electrode Materials for Solid Oxide Electrolysis Cells (SOECs)

97

as H2 O/H2 ratio is increased in the incoming gas. A switch in the H2 O/H2 ratio from 3%/97% to 50%/50% increases the pO2 at 800 °C (and the open circuit voltage) from 3.87 × 10–17 to 4.01 × 10–14 Pa. This pO2 increase leads to the formation of gaseous Ni species such as Ni(OH)2 , which in turn leads to Ni evaporation and agglomeration (Chen et al. 2013; Gubner 1997; Hagen et al. 2006; Holzer et al. 2011). Hauch et al. proposed the migration of Ni and its deposition according to the following reaction (Hauch et al. 2008): Ni(OH)2(g) + 4e− → Ni(s) + 2O2− + H2 (g)

(13)

This deposition has been explained in terms of the electric potential gradient through the Ni-YSZ cermet. Due to the electric potential gradient, the TPBs close to the electrolyte are more reducing (as they are producing H2 ) compared to the TPBs far from the electrolyte. As a result, gaseous Ni species will be reduced to metallic Ni deposits close to the fuel electrode/electrolyte interface. This phenomenon is enhanced at high temperatures, high p(H2 O) and high current density. At lower temperatures such as 850 °C, low p(H2 O) such as 0.5 atm and lower current densities such as − 0.5 A cm−2 , the cell can be operated successfully without any significant degradation for up to 1300 h. To improve both initial performance and durability of Ni-YSZ fuel electrodes, it is recommended to start with a relatively dense initial fuel electrode microstructure with a fine and well-dispersed particle distribution. (Hauch et al. 2016). In contrast, many recent studies found Ni depletion at the fuel electrode/electrolyte interface and have proposed the migration of Ni away from the interface during electrolysis operation (Monaco et al. 2019; Sciazko et al. 2021). For example, a study by Trini et al. observed Ni depletion near the electrode/electrolyte interface when operating the cell at 800 °C under − 0.5 A cm−2 and p(H2 O)/p(H2 ) = 0.5/0.5. The Ni/(Ni + Zr + Y) atomic ratio decreased from ~ 0.49 to ~ 0.28 after 1000 h SOEC operation (Trini et al. 2020). Similarly, Hoerlein et al. also observed Ni depletion and an extension in the electrolyte, resulting in an increased cell ohmic resistance. The phenomenon of Ni depletion has been found to be strongly dependent on various operating parameters such as current density, anode overpotential, humidity and temperature. Particularly, high p(H2 O) such as 0.8 atm and high temperature such as > 800 °C would result in enhanced Ni depletion (Hoerlein et al. 2018). The increase in the grain size during operation has a detrimental impact of the TPB length. Lay-Grindler et al. found that the Ni grain size increased from around 1.6 μm for a pristine cell to 2.7 and 3.3 μm for cells operated at 0.5 and 0.8 A cm−2 , respectively (800 °C for 1000 h). This corresponded to a TPB decrease from 10.5 μm−2 to 7.1 and 6.2 μm−2 , respectively. However, this agglomeration only contributed to a small degradation in the full cell with most of the degradation still attributed to the oxygen electrode degradation. The effect of steam content on the electrochemical performance of Ni-YSZ fuel electrode has been demonstrated by various researchers (Dasari et al. 2013; Ebbesen et al. 2014). The electrode polarization resistance decreases remarkably as the steam content is increased from 3 to 50%, which is due to the increase of the reactant

98

M. S. Khan and R. Knibbe

concentration. However, a further increase in the steam content can induce Ni oxidation, which in turn would affect the fuel electrode conductivity and electrocatalytic performance. However, it has been demonstrated (Hauch 2007) that a SOEC can operate in 99% steam without Ni oxidation. Nevertheless, it is well established that the fuel electrode degradation rate increases by increasing the H2 O concentration. For example, a study by Tietz and co-workers has shown relatively higher voltage degradation rate of 3.8% kh−1 for a period of 9000 h, when the cell was tested in 80% H2 O (Tietz et al. 2013). It was observed that the Ni particles near the electrolyte were rough, while the particles at the center of fuel electrode support were smooth and had nano-sized dendrites on the surface after SOEC operation. However, the mechanism responsible for these phenomena is still not clear. In comparison, a study by Fang et al. which tested NiYSZ fuel electrode-supported cells for up to 21,000 h under 50% H2 O at − 0.5 A cm−2 , observed a voltage degradation rate as low as 0.4% kh−1 (Fang et al. 2017). These studies clearly indicate that the fuel electrode experiences less degradation when operated in intermediate pO2 with a p(H2 O) ≤ 0.5, at current densities of − 0.5 to − 1 A cm−2 and temperatures around 800 °C (Hauch 2007; Hauch et al. 2008; Jensen 2006). However, a study by Schefold et al. observed a voltage degradation rate of 0.57% kh−1 during an SOEC operation of 23,000 h in 75% H2 O under − 0.9 A cm−2 (Schefold et al. 2017). This excellent fuel electrode durability was attributed to the Ni-GDC fuel electrode, which did not experience any degradation. Other Nibased fuel electrodes such as Ni-SDC and Ni-Ru-GDC have also shown excellent initial performance and short-term durability (Chen et al. 2015; Kim-Lohsoontorn et al. 2010; Tan and Ishihara 2020). However, long-term testing is still necessary to determine the long-term degradation behavior. Apart from Ni migration, there are two key degradation phenomena observed in H2 O electrolysis that are not observed during SOFC operation—ZrO2 nanoparticle formation (Chen et al. 2013; Hansen et al. 2016) and YSZ reduction (Hansen et al. 2016). The formation of ZrO2 nanoparticles has also been observed in the cell tested under − 1 to − 1.5 A cm−2 at or above 800 °C. This has been ascribed to the atomic Zr dissolution in the Ni due to the low interfacial pO2 . The Zn then diffuses through the Ni bulk or surface and reprecipitates as ZrO2 between the two Ni grains or at Ni surface. The observed ZrO2 reprecipitation is dominated in the high pO2 regions within the fuel electrode (Chen et al. 2013). Under strong cathode polarization YSZ has also been observed to reduce, which is subsequently associated with a Ni-Zr intermetallic formation (Hansen et al. 2016). However, the yttria from the YSZ is not reduced, but remains as nanoparticle inclusions in the reduced intermetallic phase. During long-term operation, this intermetallic phase may experience multiple oxidation–reduction cycles, which impacts the durability of the Ni-cermet fuel electrode. In addition to the degradation due to morphological changes in Ni-YSZ, impurity induced degradation is a common issue. Impurities are considered to be a major factor leading to large differences in experimental results of nominally similar materials in different labs. These impurities originate from sealants, raw fuel cell materials,

Fuel Electrode Materials for Solid Oxide Electrolysis Cells (SOECs)

99

balance of plant components and gas impurities. In 2005, Liu et al. investigated impurity segregation at the Ni-YSZ electrode–electrolyte interface using a combination of FIB and TEM (Liu and Jiao 2005). The group identified large amorphous regions throughout the electrode which were identified as SiO2 , but Na and Al have also been found in these glassy regions. It is proposed that SiO2 migrates as gaseous Si(OH)4 , which can deposit as glassy phases at TPBs and in Ni grains (Fig. 2). The Si(OH)4 partial pressure over silicates at high steam concentration and high temperatures can be quite high due to vapormediated volatilization (Jacobson et al. 2005). The covalent bonding from these SiO2 compounds has a chemical affinity for the ionic oxide surfaces and is often observed at the interface (Irvine et al. 2016). During H2 O electrolysis, it is also possible that Si forms at the TPBs if the cell is subjected to the strong cathodic current (Chatzichristodoulou et al. 2016; Tao, Shao, and Cheng 2016). The accumulation of SiO2 or Si around Ni can also lead to rim formation around the Ni particle, which may remove the entire Ni particle from the percolating Ni network—decreasing the active TPB length (Knibbe et al. 2010). In addition, it is possible for SiO2 to form as inclusions in the Ni grains. (Chen and Jiang 2016; Couper et al. 1990; Ebbesen et al. 2010, 2012; Rinaldi et al. 2017). This impurity induced degradation leads to a performance loss that in some cases is often only partially recoverable.

2.2 CO2 and Co-Electrolysis After the onset of the industrial revolution, the atmospheric CO2 concentration has increased from 280 ppm in 1750, to 338 ppm in 1980 and to 415 ppm in 2021 (Duan et al. 2017). The increased CO2 level has resulted in anthropogenic climate change which impacts everyone. We need to develop and implement solutions to reduce atmospheric CO2 concentration, through a combination of CO2 emission reduction, capture, storage and utilization. Among the various electrochemical technologies being developed for CO2 utilization, SOECs offer the highest current density, energy efficiency and stability (Song et al. 2019). As such, this technology can play a vital role in resolving environmental and energy related issues. Beyond this, CO2 is the dominant gas in the Martian atmosphere—making CO2 conversion a critical technology for production of O2 for future Mars missions (Song et al. 2019). However, the technology is not mature enough and there are significant challenges for commercialization such as high capital, operating and maintenance costs, low CO2 conversion rates as well as degradation of the cells and stacks. The electrochemical reaction which takes place when feeding CO2 to the SOEC fuel electrode is provided in Eq. 2. Detailed mechanism description of direct CO2 reduction has only been described in limited studies and has been demonstrated with the help of in-situ XPS, Raman spectroscopy and DFT calculations (Song et al. 2019). For example, a study by Yang et al. (Yang et al. 2018) proposed the following reactions for CO2 electrolysis when using Lax Sr1-x FeO3-δ fuel electrode using Raman spectroscopy and DFT calculations.

100

M. S. Khan and R. Knibbe 2− CO2(g) + O−2 (latt) → CO3

(ad)

(14)

2− CO2− + e− → CO− 3 2(bent) + O(latt)

(15)

2− − CO− 2(bent) + e → CO(ad) + O(ad)

(16)

CO(ad) → CO(gas)

(17)

(ad)

These steps indicate that the CO2 is converted into adsorbed carbonate species with the help of lattice oxygen ions, followed by the decomposition of carbonate species, which in turn is followed by the desorption of the CO molecule. On the other hand, Opitz and co-workers used in-situ XPS methods to investigated CO2 reduction mechanisms on (La,Sr)FeO3-δ and (La,Sr)CrO3-δ . The ferrite-based perovskite performs better than the chromite-based perovskite, but with similar reaction mechanisms. They also propose that a radical carbonate intermediate forms, but with slightly different reaction steps. In the first step, Opitz et al. propose that CO2 combines with an oxygen vacancy and an electron to produce the CO2 − radical (Opitz et al. 2017). The authors propose that the species further reacts with oxygen ions to produce a carbonate radical ((CO3 )•3– ), which finally decomposes to CO upon accepting an additional electron. According to this study, the following reactions take place: CO2,(g) + V ac + e− → (CO2 )·− (ad)

(18)

2− (CO2 )·− → (CO2 )·3− (ad) + O (ad)

(19)

2− − (CO2 )·3− (ad) + e → CO(g) + 2O

(20)

In the case of co-electrolysis, both CO2 and H2 O molecules diffuse to the TPBs through the porous fuel electrode, where they are converted into H2 and CO according to the reactions provided in Eqs. 1 and 2. In addition, the water–gas shift (WGS) reaction (Eq. 21) and methanation reaction (Eq. 22) can also take place. Water−gas shift(WGS)reaction : H2 O + CO ↔ H2 + CO2

(21)

Methanation reaction : CO + 3H2 ↔ CH4 + H2 O

(22)

However, optimized reactant gas composition and flow rates as well as improved catalysts are required with temperatures < 700 °C to convert syngas to methane. Due to relatively low selectivity and the unclear reaction mechanisms, only limited studies have reported on methane formation (Xie et al. 2011). Therefore, further studies will greatly help to understand and improve methanation reaction during co-electrolysis. A recent study by Lo Faro and co-workers demonstrated a high methane yield (67%)

Fuel Electrode Materials for Solid Oxide Electrolysis Cells (SOECs)

101

with a selectivity of 94% at 525 °C using a Ni–Fe bimetallic catalyst. Moreover, the study also observed delamination because of carbon deposition at the fuel electrode support/electrolyte interface (Lo Faro et al. 2020).

2.2.1

Degradation During CO2 and Co-Electrolysis

Similar to H2 O electrolyte, for CO2 and co-electrolysis, the gas is transported through the porous electrode to the TPBs where the electroreduction process takes places. However, due to the very stable carbon–oxygen double bond, the overpotential associated with CO2 electrolysis is higher than that for H2 O electrolysis (Song et al. 2019). In addition, the CO2 mole shows sluggish diffusivity in comparison with the H2 O molecule (Ebbesen, Graves, and Mogensen 2009; Isenberg 1981). In the first part of this section, we will discuss degradation during CO2 electrolysis before exploring co-electrolysis degradation mechanisms. To avoid Ni oxidation during pure CO2 electrolysis, a reducing gas such as H2 or CO needs to be added to the CO2 . This mixed gas is referred to as safe gas. However, using a mixture of CO2 and H2 results in the formation of CO and H2 O by the R-WGS reaction (Eq. 21). Therefore, the electrochemical reaction is partially transferred from CO2 to H2 O electrolysis. As H2 O is both thermodynamically and kinetically favorable in comparison with CO2 electrolysis, the overpotential decreases (Ebbesen et al. 2009; Isenberg 1981; Song et al. 2019). Similarly, using a mixture of CO2 and CO can result in the formation of carbon especially at high electrolysis voltages and high CO concentrations (Li et al. 2013). The formed carbon blocks the active TPBs which in turn degrades the performance (Li et al. 2015a). Therefore, being able to eliminate safe gas and using pure CO2 is critical not only in terms of increasing the durability but also decreasing the associated costs (Dong et al. 2017). Nevertheless, most studies use safe gas to operate SOECs and only limited studies are available for pure CO2 electrolysis on Ni-based cermet fuel electrodes. Carbon formation during CO2 electrolysis is a critical degradation mechanism which can rapidly degrade the cell performance (Skafte et al. 2018). Carbon formation is a wellknown phenomenon in steam reforming (Rostrup-Nielsen and Alstrup 1999; Tavares et al. 1994), CO2 and H2 O co-electrolysis (Tao et al. 2014, 2016) and hydrocarbonfueled fuel cell operation (Li et al. 2015b; Suboti´c et al. 2015, 2016). However, there are limited studies which address carbon deposition during CO2 electrolysis (Duboviks et al. 2015; Duboviks et al. 2014; Li et al. 2015b; Maher et al. 2013). Carbon formation during CO2 electrolysis can occur due to various pathways, but is commonly attributed to CO disproportionation—known as the Boudouard reaction (Eq. 23). CO → CO2 + C

(23)

Some researchers have proposed alternative pathways including direct electrochemical deposition of CO (Li et al. 2015a).

102

M. S. Khan and R. Knibbe

CO + 2 e− → C + O2−

(24)

However, coking is often observed earlier than anticipated due to temperature and gas composition variations throughout the cell (Tao et al. 2014, 2016). Close to the TPB sites the local the local temperature drops due to the endothermic nature of the CO2 electrolysis reaction which are prime conditions for carbon deposition. Furthermore, large fuel electrode overpotential (such as those originating from impurity degradation) favors carbon formation at the reaction sites. In 2019, Skafte et al. presented a potential solution to this carbon formation issue with the direct use of ceria as the fuel electrode (Fig. 3, Skafte et al. 2019). The ceria electrode showed increased durability through improved tolerance to carbon deposition. Furthermore, it is possible to operate this fuel electrode at higher fuel conversion rates—increasing fuel and O2 yields. Through a combination of ambientpressure XPS and DFT calculations, the group demonstrated the two key mechanisms to avoid carbon deformation—rich oxygen vacancies and carbonate coverage on the ceria surface. The oxygen vacancies trap the carbon atoms, but suppress carbon deposition. The carbonate coverage encourages a reverse Boudouard reaction, facilitating CO production. To mitigate degradation during pure CO2 electrolysis, some other fuel electrode modifications have been explored. Dong et al. used a hierarchically ordered porous Ni-YSZ electrode along with a Pd/GDC catalyst to prevent carbon formation and Ni oxidation, while Li and co-workers used a carbon powder and poly(vinyl alcohol) (PVA) to increase the porosity of the Ni-YSZ fuel electrode to overcome these issues (Dong et al. 2017; Li et al. 2019a). However, the cells in these studies were tested for less than 100 h and therefore need further investigations for long-term operation. CO2 electrolysis degradation studies typically use a safe gas (either CO2 /H2 (typically in the form of H2 O) or CO2 /CO mixture) for long-term testing. In these studies,

Fig. 3 Carbon nanoparticle growth on dispersed nanoparticles, b proposed reaction pathway for CO formation on Sm-doped ceria substrate showing intermediate steps of oxygen vacancy formation, carbonate formation, carboxylate formation and carbon bonding on surface. (Skafte et al. 2019)

Fuel Electrode Materials for Solid Oxide Electrolysis Cells (SOECs)

103

the amount of CO has been varied from more than 50% to 0% (pure CO2 ) while monitoring the Ni-YSZ fuel electrode degradation at different temperatures for several thousand hours. Many studies using a CO2 /CO mixture attribute the Ni-YSZ fuel electrode degradation to carbon deposition (Ebbesen et al. 2014; Wang et al. 2013). However, earlier studies by Ebbesen et al. ascribed the degradation to gas stream impurities, as the cell performance could not be reactivated by using cell operating conditions that would oxidize carbon (Ebbesen et al. 2010, 2009; Ebbesen and Mogensen 2010). Co-electrolysis (electrolysis using a mixture of CO2 and H2 O) is an important technique for syngas (mixture of CO and H2 ) production. Like CO2 and H2 O electrolysis, the co-electrolysis reaction product (syngas) is produced at the Ni-YSZ fuel electrode while oxygen is produced at the oxygen electrode. The process has greater commercial potential as it is more energy efficient than two separate electrolysis processes and the product (syngas) can be used to produce various chemicals. There is increasing interest in co-electrolysis which can be attributed to the possibility of using this technology in both terrestrial (chemical production) and extra-terrestrial applications. Thermodynamically, at the cell level, apart from H2 O and CO2 electrolysis reactions, the WGS as well as the reverse water–gas shift (R-WGS) reactions can also take place. However, the extent of WGS and R-WGS has been found to be different by different researchers. This is attributed to cell and operation differences including temperature, pressure, gas flow rate, cell materials and microstructures—making it difficult to directly compare different degradation tests. At temperatures below 816 °C, the WGS reaction is favored (WGS ]G < 0), whereas above 816 °C, the RWGS is favored (R-WGS ]G < 0). However, at higher temperatures, the degradation of the materials is also enhanced. Other parameters can also promote degradation, for example, using a mixture of CO2 and H2 O at high pressures and high conversion may induce carbon formation. Some research groups have found that the measured area specific resistance (ASR) for co-electrolysis lies between CO2 electrolysis (highest) and H2 O electrolysis (lowest)—suggesting that both H2 O and CO2 electrolysis take place and both the R-WGS and CO2 electrolysis are responsible for the CO produced (Graves et al. 2011; Li et al. 2013). However, some other studies have shown a comparable performance for steam and co-electrolysis operations and proposed that the CO2 reduction during co-electrolysis is due to the R-WGS and not due to CO2 electrolysis (KimLohsoontorn and Bae 2011; Kim-Lohsoontorn et al. 2010). In this scenario, the H2 O is reduced to H2 and then through R-WGS reaction CO will be formed—as such the gas output would be a mixed gas composition similar to what would be produced through direct co-electrolysis. Assuming a comparable ASR for steam electrolysis and co-electrolysis, the conversion through co-electrolysis is still considered superior as it has fewer electrolysis steps and can help in the production of sustainable fuels and chemicals. Moreover, co-electrolysis is an effective way of reducing CO2 without significant degradation of Ni-based cermets compared to pure CO2 electrolysis. Therefore, both short-term and long-term studies are available on the topic of co-electrolysis with

104

M. S. Khan and R. Knibbe

various CO2 /H2 O compositions and in some cases along with the addition of CO. For example, a study by Zhan and co-workers (Zhan et al. 2009) have shown an ASR value as low as 0.22 [ cm2 at 800 °C while using a mixture of 25% H2 , 25% CO2 and 50% H2 O and a syngas production rate of 7 sccm/cm2 at 1.3 V. Another study using 10% H2 , 45% CO2 and 45% H2 O have shown ASR values of 0.26 [ cm2 and 0.37 [ cm2 at 850 and 800 °C, respectively (Ebbesen et al. 2009). The same study has also characterized Ni-YSZ fuel electrodes in 10% H2 , 45% CO2 and 45% H2 O at 850 °C under -0.25 A cm−2 for up to 1300 h. Methane formation has also been reported by Li et al. in 2013, however, the methane formation was only observed under specific conditions (Li et al. 2013). For methane formation, a 2:1 H2 O/CO2 ratio was used and a relatively high operating voltage of 2 V was required. The methane production was increased by increasing the H2 reactant content, but suppressed by including a Ru catalyst. It was proposed that the Ru catalyst suppressed carbon formation in the fuel electrode eliminating a potential methane formation pathway (Eq. 25). C(s) + 2H2 → CH4

(25)

3 Alternative Cermet Fuel Electrodes Ni-YSZ (or GDC and SDC replacing YSZ) has been studied by many groups to investigate its performance and durability for H2 O, CO2 and co-electrolysis operations. However, alternative fuel electrode cermets have also been developed in parallel to increase performance and durability. Most of this research focuses on replacing Ni with other metals such as Cu, Fe and Ru and YSZ with GDC or SDC. Some studies use alloys, such as Ni–Cu, while some others have studied the effect of infiltrated metals on an already sintered cermet fuel electrode. Cu-YSZ is considered an important alternative fuel electrode material for steam electrolysis. Lee et al. reported a small increase in hydrogen production rate while supplying water vapor at room temperature at an applied power density of 2.6 W cm−2 and 800 °C (Lee et al. 2009). The high hydrogen production rate was attributed to the increased catalytic effect and large number of reaction sites of Cu-YSZ compared to Ni-YSZ. Although no evidence was provided for this claim, Park et al. did show good electrocatalytic activity for hydrocarbon oxidation when using a Cu-ceria cermet in an SOFC application (Park et al. 2000). Another study by Kim and co-workers also found an increased hydrogen production rate with a Cu-YSZ. However, the CuYSZ microstructure was relatively coarse, which allows for an increased gas flow (Kim et al. 2010). Therefore, the results from these studies need further investigation. Furthermore, the use of Cu is considered limited due to its low melting point (Gaudillere et al. 2014). Gu and Nikolla (Gu and Nikolla 2015) investigated potential Ni alloying elements. The group used a combination of DFT calculations with microkinetic modeling to

Fuel Electrode Materials for Solid Oxide Electrolysis Cells (SOECs)

105

investigate the factors affecting the water electrolysis process. A volcano-type relationship was determined between the electrochemical rates and the binding energies of O. Furthermore, the barrier for water dissociation increases as Ni was alloyed with metals from the left to the right in the periodic table. The increased water dissociation barrier is related to the decrease of the binding energies for the common intermediates, OH and O. The group concluded that alloying Ni with Fe can lead toward higher electrolysis rates compared to pure Ni, Ni–Cu or Ni–Co. Lee et al. practically demonstrated this superior performance by testing a Ni–FeYSZ bimetallic cermet at 700 °C in 50% H2 O/15% H2 /35% Ar (Lee et al. 2021). Another study utilizing an 0.5 wt. % Fe–Ni-GDC fuel electrode observed a performance enhancement compared to pure Ni-GDC (Neofytidis et al. 2020). However, such electrodes exhibited increased polarization losses with increased steam content as well as a high susceptibility toward oxidation. Moreover, the study did not report on long-term durability. Several other studies have reported the positive impact of Fe addition to Ni-based cermet fuel electrodes for steam electrolysis operation (Ishihara et al. 2010; Simonsen et al. 2019). Nevertheless, most of the studies have focused only on performance enhancement and there is ample space to work on understanding the long-term operation and related degradation mechanisms. The electrodes replacing Ni with other metals such as Cu, Fe or Ru and YSZ with GDC or SDC have also been used for CO2 electrolysis and co-electrolysis operations. Ni-Ru-GDC shows a high performance and good coking tolerance compared to NiYSZ. This is attributed to the combination of the excellent ability of Ru to reduce carbon deposition and the high oxide ion conductivity of GDC (Kim-Lohsoontorn and Bae 2011). Similarly, Cu shows an improved coking tolerance compared to Ni. However, as previously mentioned, the Cu use may be limited due to its low melting point (Gaudillere et al. 2014). To overcome the microstructural degradation, other transition and noble metalbased cermets have been used for co-electrolysis operations. These cermets have the potential for long-term stability without significantly blocking TPBs and causing passivation. To this regard, Ni-GDC with Au, Mo and Fe was designed and tested in co-electrolysis (Ioannidou et al. 2019; Neofytidis et al. 2019). For all compositions, carbon deposition was avoided, but this was attributed to the high temperatures > 750 °C where the Boudouard reaction is not thermodynamically favorable (Stoots et al. 2008). The Au addition inhibited CO production through encouraging the RWGS reaction, whereas the addition of Mo and Fe tended to enhance CO production. Jeong et al. also showed that additions of Fe or Pd can improve the stability of the Ni-YSZ electrode during co-electrolysis. Both electrodes showed a degradation of less than 1% degradation per 100 h, with the Pd-infiltrated electrode showing only 0.8% degradation over 350 h. This was attributed to the excellent coking stability of Pd (Jeong et al. 2018). Despite interesting work in this area, there is still a need to expand both modeling and long-term durability testing to gain deeper insights into these alternate cermet electrodes so that they can be used in commercial SOECs.

106

M. S. Khan and R. Knibbe

4 Perovskite-Type Fuel Electrodes Although Ni-based cermet fuel electrodes have excellent activity and conductivity, there are a number of issues with these fuel electrodes including Ni agglomeration and migration, structural damage due to NiO formation and carbon deposition. These mechanisms impact the SOEC performance. Therefore various perovskitetype fuel electrode materials have gained attention due to their high coking resistance, redox stability, electronic and ionic conductivities and improved tolerance toward impurities (Chen and Jiang 2016). However, there are some major disadvantages including their inferior catalytic activity, reduced electrical conductivity, more complex fabrication and thermomechanical incompatibility. Many SOEC fuel electrode perovskite materials have been developed for H2 O, CO2 and co-electrolysis. The general formula of the perovskite material is ABO3 , where A is a large alkaline or rare earth metal ion, B is a small transition element and O is the oxygen ion. By doping different elements into the A or B-site, which depending on the dopant, can introduce oxygen vacancies and electronic conductivity—introducing a mixed ion and electron conduction (MIEC) behavior. Therefore, doping is considered one of the most effective methods to improve the activity and durability of the parent perovskite. Furthermore, aliovalent cation doping can tune the oxygen vacancies and electron–ion conductivities to increase the principle active sites for H2 O and CO2 electrochemical dissociation (Zhou et al. 2018). A number of doped perovskite materials have shown excellent activity and stability for H2 O, CO2 and co-electrolysis operations. Typical materials used for CO2 electrolysis fuel electrodes are La0.75 Sr0.25 Cr0.5 Mn0.5 O3-δ (LSCM) (Ye et al. 2017), La0.7 Sr0.3 Cr0.5 Fe0.5 O3−δ (LSCF) (Zhang et al. 2016), Sr2 Fe1.5 Mo0.5 O6-δ (SFM) (Li et al. 2017a), La0.2 Sr0.8 TiO3+δ (LST) (Li et al. 2012), Sr0.95 Ti0.9 Nb0.1 O3 (STN) (Zhang et al. 2014) and LaSrFeO3-δ (LSF) (Zhou et al. 2018; Liu, Liu, and Luo 2017). Several groups have demonstrated methods to improve perovskite performance for CO2 electrolysis have been recently demonstrated with modifications to LSF, STN and SFM perovskite materials. Although LSF is a promising fuel electrode material, it shows a relatively poor catalytic activity. However, Zhou et al. demonstrated that through V-doping, the performance can increased dramatically (Zhou et al. 2018). The group propose that this is due to the increased oxygen vacancy concentration, which improves catalytic activity and long-term stability. Furthermore, the performance of LST was enhanced through Sc-doping. The Sc-doped LST showed a 20% enhanced ionic conductivity and electrocatalytic activity for CO2 electrolysis (Yang et al. 2014). Similarly, the addition of Mn or Cr at the B-sites of STN also demonstrates improved electrochemical performance compared to pristine STN (Zhang et al. 2014; Qi et al. 2014). SFM shows a high electrochemical activity because of its relatively high chemical bulk diffusivity and a high CO2 reduction surface reaction rate constant (Li et al. 2017a; Hu et al. 2015). However, this performance was further improved by Li et al. through fluorine doping, which facilitated an oxyfluoride formation—Sr2 Fe1.5 Mo0.5 O6-δ F0.1 (Li et al. 2019b).

Fuel Electrode Materials for Solid Oxide Electrolysis Cells (SOECs)

107

Besides doping, infiltration is another important technique that can improve the perovskites’ catalytic and electrical properties. The increased performance is due to the increased surface oxygen vacancies and TPBs, both of which are associated with the introduction of nano-sized particles. Using infiltration techniques also helps lowering the processing temperatures, thus avoiding the initial electrode degradation. Infiltration has been used in various studies in the past. For example, infiltration of Cu nanoparticles into the LSCM fuel electrode helps in decreasing its ohmic and polarization resistances (Xing et al. 2015). Other catalysts such as Pd, GDC and Ce0.9 Mn0.1 Oδ (CMO) have also been infiltrated to improve the electrochemical performance of pristine LSCM electrodes (Zhang et al. 2018; Yue and Irvine 2017; Jiang 2012). Similarly, infiltration of GDC into Sr2 Fe1.5 Mo0.5 O6-δ has also been found to increase the current density increased more than 50% for CO2 electrolysis from 0.28 to 0.45 A cm−2 at 1.6 V and 800 °C (Lv et al. 2019). Various other infiltration studies have also shown promising results which have been attributed to the accelerated CO2 adsorption and dissociation processes (Song et al. 2019). To avoid the issue with the Ni agglomeration and issues with long-term stability of infiltrated nanoparticles, Neagu et al. demonstrated the use of direct electrocatalyst exsolution from a parent perovskite phase (Fig. 4, Neagu et al. 2015; Neagu et al. 2013). The exolution mechanism will be further explained in detail in Chap. 12. This phenomenon occurs when the metal cations on the B-site of an A-site deficient perovskite exsolve as metallic nanoparticles (either as a pure metal or metal alloy) during reduction. The exsolved nanoparticles act as highly active electrocatalysts to promote the electrochemical performance of the electrode. If the perovskite is placed under oxidizing conditions, these nanoparticles can be oxidized and the cations can be returned to their original sites (Tsekouras et al. 2013; Neagu et al. 2013). However, the exsolved particles are expected to remain socketed as long as the fuel electrode is under reducing conditions. In early work by Neagu et al., Ni was exsolved from a La0.4 Sr0.4 Ni0.03 Ti0.9 7O3-γ parent oxide substrate. Due to the mismatch between the parent perovskite phase and the Ni particle, dislocations are observed at the interface (Fig. 5). It is proposed by the researchers that the strained and embedded nanoparticle mitigates long-term agglomeration and also hydrocarbon coking. He et al. extended the work by Neagu and presented a method of co-exsolution to produce both CeO2 and Ni particles from a titanate perovskite substrate— La0.8 Ce0.1 Ni0.4 Ti0.6 O3-δ (LCeNT) (He et al. 2021). The strained faceted CeO2 cubes are produced after an air-annealing process. It is shown that the CeO2 morphology can be tuned by varying the annealing temperature, while exsolution of Ni nanoparticles form subsequently following reduction. The CeO2 forms a socketed particle— similar to the Ni observed by Neagu with a strong crystallographic orientation with the oxide LCeNT substrate. This co-exsolved microstructure shows a good electrocatalytic performance with stable behavior when operating in a H2 /CH4 environment, however, has not been tested in under electrolysis conditions. Several studies have observed increased electrolysis current densities and high Faradaic efficiencies as well as good short and long-term stabilities for the electrode materials with exsolved metallic and alloy nanoparticles (Li et al. 2017b; Liu et al. 2016; Lu et al. n.d.). Ruan et al. demonstrated that exsolved Ni nanoparticles on a

108

M. S. Khan and R. Knibbe

Fig. 4 Exsolutions from the A-site-deficient, O-stoichiometric La0.52Sr0.28Ni0.06Ti0.94O3 after reduction at 930 °C (20 h) in 5%H2/Ar. b, A-site-stoichiometric, O-excess La0.3Sr0.7Ni0.06Ti0.94O3.09 sample reduced at 930 °C (20 h) in 5%H2/Ar indicates that no exsolution has occurred. (Neagu et al. 2013)

Fig. 5 Ni particle exsolved from the La0.4 Sr0.4 Ni0.03 Ti0.97 O3-γ parent oxide substrate showing crystallographic alignment. (Neagu et al. 2015)

lanthanum chromite (La0.75 Sr0.25 )0.9 (Cr0.5 Mn0.5 )0.9 Ni0.1 O3-δ. This electrode showed both an increase current density and an increase Faradaic efficiency for CO2 electrolysis (Ruan and Xie, 2015). Fe–Ni, Fe-Co and Ni-Cu alloy nanoparticles have also been testing as exsolved nanoparticles with various perovskite parent phases (Li et al. 2017b; Liu, Liu, and Luo 2016; Lu et al. n.d.). Of these exsolved nanoalloys, the exsolved Fe–Ni nanoparticles from the La0.6 Sr0.4 Fe0.8 Mn0.2 O3-δ showed impressive performance with a current density of 1.78 A.cm−2 at 1.6 V with a Faradaic efficiency of over 98% for CO2 electrolysis. Furthermore, the electrode showed good long-term stability for 100 h of testing with no noticeable coking.

Fuel Electrode Materials for Solid Oxide Electrolysis Cells (SOECs)

109

5 Characterization of Fuel Electrodes and Product Analysis The characterization of the Ni-based fuel electrodes can be carried out using a variety of electrochemical and physical techniques. Current–voltage (I-V) curve is the most fundamental electrochemical method and can be used to characterize electrochemical performance at different operating conditions such as temperature, pressure, humidity and gas composition. Similarly, electrochemical impedance spectroscopy (EIS) is invaluable in providing insights into different electrode processes. The major products of the electrolysis processes described above are CO and H2 along with some others such as CO2 , H2 O. The product analysis for hightemperature electrolysis is relatively easy because of better selectivity compared to low-temperature electrolysis processes. For simple CO2 or H2 O electrolysis, the degree of conversion from inlet to outlet can be calculated based on the pO2 at the inlet and the outlet of the cell. However, it is more complicated for co-electrolysis. For coelectrolysis, gas chromatography (GC) and mass spectrometry (MS) are important techniques to measure inlet and outlet gas composition. The operating parameters for the electrolysis process can be changed, and their effect of the relative amounts of the gases released can be analyzed. The combination of GC with MS is integral in investigating and understanding the reaction mechanisms for co-electrolysis. Among the physical characterization techniques, X-ray diffraction (XRD), Xray photoelectron spectroscopy (XPS) and scanning electron microscopy (SEM) and transmission electron microscopy (TEM) coupled with energy-dispersive spectroscopy (EDS) are standard tools for characterizing fuel electrodes before and after operation. High resolution work with both SEM and TEM has been important in revealing the exsolved nanoparticles and understanding the texturing, strain and socketing of the nanoparticles in the parent perovskite phase. There have been particular techniques including low-voltage SEM and FIB-SEM tomography which have grown to be fundamental tools for fuel electrode characterization. Low-voltage SEM imaging is a simple, but powerful, imaging technique that allows for a quick assessment of percolating vs. non-percolating Ni (Fig. 2). FIB-SEM tomography is a more complex tool, but has been central in building an understanding of grain size, electrode tortuosity, TPB lengths and Ni percolation before and after electrolysis testing. With the improvement of SEM cameras and image processing speeds, it is anticipated that in the future the analytical volumes possible with FIB-SEM will increase. The electrochemical processes descried in this chapter are highly dynamic. As such, the possibility of exploring the fuel electrode in-situ provides researchers with a more realistic insight into the dynamic processes. In the past decade, there has been developments in regards to in-situ and in-operando tools. Due to the complex gas mixture and the high temperatures, studying the fuel electrode in-situ/in-operando is challenging. However, in the future, we expect to see more work investigating high-temperature electrolysis processes under in-situ/in-operando conditions using a range of tools including ambient-pressure XPS, in-situ Raman and in-situ TEM.

110

M. S. Khan and R. Knibbe

6 Summary The fuel electrode is where the valuable fuel is produced. Fuel electrodes are broadly classified into Ni-based cermet fuel electrodes, Ni-free metal-based cermet fuel electrodes and perovskite-type oxide fuel electrodes. The Ni-YSZ electrode is the fuel electrode of preference. It predominately degrades through Ni relocation and impurity segregation. Under more extreme electrolysis operating conditions, it can produce ZrO2 nanoparticles, oxidize to NiO or be reduced at the Ni-YSZ interface. CO2 electrolysis and co-electrolysis presents more challenges as carbon can form and quickly degrade the performance. Recently, some of these key issues have been solved through the engineering of this fuel electrode. In the future, it is anticipated that research will expand on increasing yields of C1 and C2 hydrocarbons. Furthermore, from a characterization and modeling perspective, it would be interesting to expand both DFT and in-situ work on the fuel electrode to open up new novel materials with reduced overpotential losses and improved durability.

References Cassidy M, Connor PA, Irvine JTS, Savaniu CD, Kendall K, Kendall M (eds) (2016) Anodes. In: High-temperature solid oxide fuel cells for the 21st century—fundamentals, design and applications, chap 5. Elsevier Chatzichristodoulou C, Chen M, Hendriksen PV, Jacobsen T, Mogensen MB (2016) Understanding degradation of solid oxide electrolysis cells through modeling of electrochemical potential profiles. Electrochim Acta 189:265–282 Chen K, Jiang SP (2016) Review—materials degradation of solid oxide electrolysis cells. J Electrochem Soc 163:F3070–F3083 Chen M, Liu Y-L, Bentzen JJ, Zhang W, Sun X, Hauch A, Tao Y, Bowen JR, Hendriksen PV (2013) Microstructural degradation of Ni/YSZ electrodes in solid oxide electrolysis cells under high current. J Electrochem Soc 160:F883–F891 Chen T, Zhou Y, Liu M, Yuan C, Ye X, Zhan Z, Wang S (2015) High performance solid oxide electrolysis cell with impregnated electrodes. Electrochem Commun 54:23–27 Couper AM, Pletcher D, Walsh FC (1990) Electrode materials for electrosynthesis. Chem Rev 90:837–865 Dasari HP, Park S-Y, Kim J, Lee J-H, Kim B-K, Je H-J, Lee H-W, Yoon KJ (2013) Electrochemical characterization of Ni–yttria stabilized zirconia electrode for hydrogen production in solid oxide electrolysis cells. J Power Sources 240:721–728 Dong D, Xu S, Shao X, Hucker L, Marin J, Pham T, Xie K, Ye Z, Yang P, Yu L, Parkinson G, Li CZ (2017) Hierarchically ordered porous Ni-based cathode-supported solid oxide electrolysis cells for stable CO2 electrolysis without safe gas. J Mater Chem A 5:24098–102 Dönitz W, Erdle E (1985) High-temperature electrolysis of water vapor—status of development and perspectives for application. Int J Hydrogen Energy 10:291–295 Dönitz W, Dietrich G, Erdle E, Streicher R (1988) Electrochemical high temperature technology for hydrogen production or direct electricity generation. Int J Hydrogen Energy 13:283–287 Duan, X, Xu J, Wei Z, Ma J, Guo S, Wang S, Liu H, Dou S (2017) Metal-free carbon materials for CO2 electrochemical reduction. Adv Mater 29:1701784

Fuel Electrode Materials for Solid Oxide Electrolysis Cells (SOECs)

111

Duboviks V, Maher RC, Kishimoto M, Cohen LF, Brandon NP, Offer GJ (2014) A Raman spectroscopic study of the carbon deposition mechanism on Ni/CGO electrodes during CO/CO2 electrolysis. Phys Chem Chem Phys 16:13063–13068 Duboviks V, Lomberg M, Maher RC, Cohen LF, Brandon NP, Offer GJ (2015) Carbon deposition behaviour in metal-infiltrated gadolinia doped ceria electrodes for simulated biogas upgrading in solid oxide electrolysis cells. J Power Sources 293:912–921 Ebbesen SD, Mogensen M (2010) Exceptional durability of solid oxide cells. Electrochem SolidState Lett 13:B106 Ebbesen SD, Graves C, Mogensen M (2009) Production of synthetic fuels by Co-electrolysis of steam and carbon dioxide. Int J Green Energy 6:646–660 Ebbesen SD, Knibbe R, Mogensen M (2012) Co-electrolysis of steam and carbon dioxide in solid oxide cells. J Electrochem Soc 159:F482–F489 Ebbesen SD, Jensen SH, Hauch A, Mogensen MB (2014) High temperature electrolysis in alkaline cells, solid proton conducting cells, and solid oxide cells. Chem Rev 114:10697–10734 Ebbesen SD, Sun X, Mogensen MB (2015) Understanding the processes governing performance and durability of solid oxide electrolysis cells. Faraday Discuss 182:393–422 Ebbesen SD, Graves C, Hauch A, Jensen SH, Mogensen M (2010) Poisoning of solid oxide electrolysis cells by impurities. J Electrochem Soc 157:B1419 Eguchi K, Hatagishi T, Arai H (1996) Power generation and steam electrolysis characteristics of an electrochemical cell with a zirconia- or ceria-based electrolyte. Solid State Ionics 86–88:1245– 1249 Fang Q, Blum L, Menzler NH, Stolten D (2017) Solid oxide electrolyzer stack with 20,000 h of Operation. ECS Meeting Abstracts, MA2017-03:119–19 Gao Z, Mogni LV, Miller EC, Railsback JC, Barnett SA (2016) A perspective on low-temperature solid oxide fuel cells. Energy Environ Sci Gaudillere C, Navarrete L, Serra JM (2014) Syngas production at intermediate temperature through H2O and CO2 electrolysis with a Cu-based solid oxide electrolyzer cell. Int J Hydrogen Energy 39:3047–3054 Graves C, Ebbesen SD, Mogensen M (2011) Co-electrolysis of CO2 and H2O in solid oxide cells: performance and durability. Solid State Ionics 192:398–403 Gu X-K, Nikolla E (2015) Fundamental insights into high-temperature water electrolysis using Ni-based electrocatalysts. J Phys Chem C 119:26980–26988 Gubner A (1997) Investigations into the degradation of the cermet anode of a solid oxide fuel cell. ECS Proce Volumes 1997–40:844–850 Hagen A, Barfod R, Hendriksen PV, Liu Y-L, Ramousse S (2006) Degradation of anode supported SOFCs as a function of temperature and current load. J Electrochem Soc 153:A1165 Hansen KV, Chen M, Jacobsen T, Thydén K, Simonsen SB, Koch S, Mogensen MB (2016) Effects of strong cathodic polarization of the Ni-YSZ interface. J Electrochem Soc 163:F1217–F1227 Hauch A, Ebbesen SD, Jensen SH, Mogensen M (2008) Solid oxide electrolysis cells: microstructure and degradation of the Ni/Yttria-stabilized zirconia electrode. J Electrochem Soc 155:B1184 Hauch A, Brodersen K, Chen M, Mogensen MB (2016) Ni/YSZ electrodes structures optimized for increased electrolysis performance and durability. Solid State Ionics 293:27–36 Hauch A, Marchese M, Lanzini A, Graves C (2018) Re-activation of degraded nickel cermet anodes—nano-particle formation via reverse current pulses. J Power Sources 377:110–120 Hauch A (2007) Solid oxide electrolysis cells : performance and durability. Technical University of Denmark He S, Li M, Hui J, Yue X (2021) In-situ construction of ceria-metal/titanate heterostructure with controllable architectures for efficient fuel electrochemical conversion. Appl Catal B: Environ, 298 Hoerlein MP, Riegraf M, Costa R, Schiller G, Andreas Friedrich K (2018) A parameter study of solid oxide electrolysis cell degradation: microstructural changes of the fuel electrode. Electrochim Acta 276:162–175

112

M. S. Khan and R. Knibbe

Holzer L, Iwanschitz B, Hocker Th, Münch B, Prestat M, Wiedenmann D, Vogt U, Holtappels P, Sfeir J, Mai A, Graule Th (2011) Microstructure degradation of cermet anodes for solid oxide fuel cells: quantification of nickel grain growth in dry and in humid atmospheres. J Power Sources 196:1279–94 Hu B, Wang Y, Zhu Z, Xia C, Bouwmeester HJM (2015) Measuring oxygen surface exchange kinetics on mixed-conducting composites by electrical conductivity relaxation. J Mater Chem A 3:10296–10302 Ioannidou E, Neophytides S, Niakolas DM (2019) Experimental clarification of the RWGS reaction effect in H2O/CO2 SOEC Co-Electrolysis Conditions. Catalysts 9 Irvine JTS, Neagu D, Verbraeken MC, Chatzichristodoulou C, Graves C, Mogensen MB (2016) Evolution of the electrochemical interface in high-temperature fuel cells and electrolysers. Nature Energy 1 Isenberg AO (1981) Energy conversion via solid oxide electrolyte electrochemical cells at high temperatures. Solid State Ionics 3–4:431–437 Ishihara T, Jirathiwathanakul N, Zhong H (2010) Intermediate temperature solid oxide electrolysis cell using LaGaO3 based perovskite electrolyte. Energy Environ Sci 3:665–672 Jacobson NS, Opila EJ, Myers DL, Copland EH (2005) Thermodynamics of gas phase species in the Si–O–H system. J Chem Thermodyn 37:1130–1137 Jensen, SH (2006) Solid oxide electrolyser cell. Technical University of Denmark Jeong H-Y, Yoon KJ, Lee J-H, Chung Y-C, Hong J (2018) Long-term stability for co-electrolysis of CO2/steam assisted by catalyst-infiltrated solid oxide cells. J Korean Ceram Soc 55:50–54 Jiang SP (2012) Nanoscale and nano-structured electrodes of solid oxide fuel cells by infiltration: advances and challenges. Int J Hydrogen Energy 37:449–470 Khan MZ, Iltaf A, Ishfaq HA, Khan FN, Tanveer WH, Song R-H, Mehran MT, Saleem M, Hussain A, Masaud Z (2021) Flat-tubular solid oxide fuel cells and stacks: a review. J Asian Ceram Soc 9:745–770 Khan, MS, Xu X, Knibbe R, Zhu Z (2018) Porous scandia-stabilized zirconia layer for enhanced performance of reversible solid oxide cells. ACS Appl Mater Interfaces 10:25295–302 Kim J, Lee S, Kang K-H, Hong HS (2010) Preparation and characterization of Cu/YSZ cathode for high-temperature electrolysis. Int J Energy Res 34:438–444 Kim-Lohsoontorn P, Bae J (2011) Electrochemical performance of solid oxide electrolysis cell electrodes under high-temperature coelectrolysis of steam and carbon dioxide. J Power Sources 196:7161–7168 Kim-Lohsoontorn P, Yim HB, Bae JM (2010) Electrochemical performance of Ni-YSZ, Ni/RuGDC, LSM-YSZ, LSCF and LSF electrodes for solid oxide electrolysis cells. 623–29 Klotz D, Butz B, Leonide A, Hayd J, Gerthsen D, Ivers-Tiffée E (2011) Performance enhancement of SOFC anode through electrochemically induced Ni/YSZ nanostructures. J Electrochem Soc 158:B587 Knibbe R, Traulsen ML, Hauch A, Ebbesen SD, Mogensen M (2010) Solid oxide electrolysis cells: degradation at high current densities. J Electrochem Soc 157:B1209 Lee S, Kim J-M, Hong HS, Woo S-K (2009) Fabrication and characterization of Cu/YSZ cermet high temperature electrolysis cathode material prepared by high-energy ball-milling method: II. 700 °C-sintered. J Alloy Compd 467:614–621 Lee M-J, Lee K-J, Shin J-H, Hwang H (2021) Fabrication of durable Ni–YSZ hydrogen electrode for high-temperature solid electrolyzer cells. J Nanosci Nanotechnol 21:3842–3846 Li Y, Zhou J, Dong D, Yan Wang JZ, Jiang HX, Xie K (2012) Composite fuel electrode La0.2Sr0.8TiO3−δ–Ce0.8Sm0.2O2−δ for electrolysis of CO2 in an oxygen-ion conducting solid oxide electrolyser. Phys Chem Chem Phys 14:15547–15553 Li W, Wang H, Shi Y, Cai N (2013) Performance and methane production characteristics of H2O– CO2 co-electrolysis in solid oxide electrolysis cells. Int J Hydrogen Energy 38:11104–11109 Li X, Liu M, Lee J-P, Ding D, Bottomley LA, Park S, Liu M (2015b) An operando surface enhanced Raman spectroscopy (SERS) study of carbon deposition on SOFC anodes. Phys Chem Chem Phys 17:21112–21119

Fuel Electrode Materials for Solid Oxide Electrolysis Cells (SOECs)

113

Li Y, Li Y, Wan Y, Xie Y, Zhu J, Pan H, Zheng X, Xia C (2019b) Perovskite oxyfluoride electrode enabling direct electrolyzing carbon dioxide with excellent electrochemical performances. Adv Energy Mater 9:1803156 Li, Wenying, Shi Y, Luo Y, Wang Y, Cai N. 2015a. Carbon deposition on patterned nickel/yttria stabilized zirconia electrodes for solid oxide fuel cell/solid oxide electrolysis cell modes. J Power Sources 276:26–31 Li Y, Chen X, Yang Y, Jiang Y, Xia C (2017a) Mixed-conductor Sr2Fe1.5Mo0.5O6−δ as robust fuel electrode for pure CO2 reduction in solid oxide electrolysis cell. ACS Sustain Chem Eng 5:11403–12 Li Y, Hu B, Xia C, Xu WQ, Lemmon JP, Chen F (2017b) A novel fuel electrode enabling direct CO2 electrolysis with excellent and stable cell performance. J Mater Chem A 5:20833–42 Li X, Wang Y, Liu W, Wilson JA, Wang J, Wang C, Yang J, Xia C, Zhou XD, Guan W (2019a) Reliability of CO2 electrolysis by solid oxide electrolysis cells with a flat tube based on a composite double-sided air electrode. Compos Part B Eng 166:549–54 Liu YL, Jiao C (2005) Microstructure degradation of an anode/electrolyte interface in SOFC studied by transmission electron microscopy. Solid State Ionics 176:435–442 Liu S, Liu Q, Luo J-L (2016) CO2-to-CO conversion on layered perovskite with in situ exsolved Co–Fe alloy nanoparticles: an active and stable cathode for solid oxide electrolysis cells. J Mater Chem A 4:17521–17528 Liu S, Liu Q, Luo J-L (2017) The excellence of La(Sr)Fe(Ni)O3 as an active and efficient cathode for direct CO2 electrochemical reduction at elevated temperatures. J Mater Chem A 5:2673–2680 Lo Faro M, Oliveira da Silva W, Valenzuela Barrientos W, Saglietti GGA, Zignani SC, Antonucci V, Ticianelli EA, Aricò AS (2020) Enhanced production of methane through the use of a catalytic Ni–Fe pre-layer in a solid oxide co-electrolyser. Int J Hydrogen Energy 45:5134–5142 Lu J, Zhu C, Pan C, Lin W, Lemmon John P, Chen F, Li C, Xie K (n.d.) Highly efficient electrochemical reforming of CH4/CO2 in a solid oxide electrolyser. Sci Adv 4:eaar5100 Lv H, Zhou Y, Zhang X, Song Y, Liu Q, Wang G, Bao X (2019) Infiltration of Ce0.8Gd0.2O1.9 nanoparticles on Sr2Fe1.5Mo0.5O6-δ cathode for CO2 electroreduction in solid oxide electrolysis cell. J Energy Chem 35:71–78 Maher RC, Duboviks V, Offer GJ, Kishimoto M, Brandon NP, Cohen LF (2013) Raman spectroscopy of solid oxide fuel cells: technique overview and application to carbon deposition analysis. Fuel Cells 13:455–469 McEvoy A (2003) Anodes. In: Singhal SC, Kendall K (eds) High temperature and solid oxide fuel cells, chap 6. Elsevier Science: Amsterdam Monaco F, Hubert M, Vulliet J, Ouweltjes JP, Montinaro D, Cloetens P, Piccardo P, Lefebvre-Joud F, Laurencin J (2019) Degradation of Ni-YSZ electrodes in solid oxide cells: impact of polarization and initial microstructure on the Ni evolution. J Electrochem Soc 166:F1229–F1242 Neagu D, Tsekouras G, Miller DN, Ménard H, Irvine JTS (2013) In situ growth of nanoparticles through control of non-stoichiometry. Nat Chem 5:916–923 Neagu D, Oh TS, Miller DN, Ménard H, Bukhari SM, Gamble SR, Gorte RJ, Vohs JM, Irvine JTS (2015) Nano-socketed nickel particles with enhanced coking resistance grown in situ by redox exsolution. Nature Commun 6 Neofytidis C, Ioannidou E, Neophytides SG, Niakolas DK (2019) Transition metal elements as Ni/GDC dopants for the H2O electrolysis process in SOECs: Fe-Ni vs Au-Mo-Ni Interaction. ECS Trans 91:2359–2368 Neofytidis C, Ioannidou E, Kollia M, Neophytides SG, Niakolas DK (2020) The promoting effect of Fe on Ni/GDC for the Solid Oxide H2O electrolysis. Int J Energy Res 44:10982–10995 Opitz AK, Nenning A, Rameshan C, Kubicek M, Götsch T, Blume R, Hävecker M, Knop-Gericke A, Rupprechter G, Klötzer B, Fleig J (2017) Surface chemistry of perovskite-type electrodes during high temperature CO2 electrolysis investigated by operando photoelectron spectroscopy. ACS Appl Mater Interfaces 9:35847–35860

114

M. S. Khan and R. Knibbe

Pan W, Chen K, Ai Na, Lü Z, Jiang SP (2015) Mechanism and Kinetics of NiY2O3-ZrO2hydrogen electrode for water electrolysis reactions in solid oxide electrolysis cells. J Electrochem Soc 163:F106–F14 Park S, Vohs JM, Gorte RJ (2000) Direct oxidation of hydrocarbons in a solid-oxide fuel cell. Nature 404:265–267 Qi W, Gan Y, Yin D, Li Z, Wu G, Xie K, Wu Y (2014) Remarkable chemical adsorption of manganese-doped titanate for direct carbon dioxide electrolysis. J Mater Chem A 2:6904–15 Rinaldi G, Diethelm S, Oveisi E, Burdet P, Van herle J, Montinaro D, Fu Q, Brisse A (2017) Post-test analysis on a solid oxide cell stack operated for 10,700 hours in steam electrolysis mode. Fuel Cells 17:541–49 Rostrup-Nielsen JR, Alstrup I (1999) Innovation and science in the process industry steam reforming and hydrogenolysis. Catal Today 53:311–316 Ruan C, Xie K (2015) A redox-stable chromate cathode decorated with in situ grown nickel nanocatalyst for efficient carbon dioxide electrolysis. Catal Sci Technol 5:1929–1940 Schafbauer W, Menzler NH, Buchkremer HP (2014) Tape casting of anode supports for solid oxide fuel cells at Forschungszentrum Jülich. Int J Appl Ceram Technol 11:125–135 Schefold J, Brisse A, Poepke H (2017) 23,000 h steam electrolysis with an electrolyte supported solid oxide cell. Int J Hydrogen Energy 42:13415–13426 Sciazko A, Shimura T, Komatsu Y, Shikazono N (2021) Ni-GDC and Ni-YSZ electrodes operated in solid oxide electrolysis and fuel cell modes. J Thermal Sci Technol 16:JTST0013–JTST13 Simonsen SB, Muhl TT, Thydén KTS, Chatzichristodoulou C, Nielsen J, Sudireddy BR (2019) Effect of Fe on high performing nanostructured Ni/Gd-doped ceria electrocatalysts. Solid State Ionics 340:115019 Skafte TL, Blennow P, Hjelm J, Graves C (2018) Carbon deposition and sulfur poisoning during CO2 electrolysis in nickel-based solid oxide cell electrodes. J Power Sources 373:54–60 Skafte TL, Guan Z, Machala ML, Gopal CB, Monti M, Martinez L, Stamate E, Sanna S, Garrido JA, Torres EJ, Crumlin M-M, Bajdich M, Chueh WC, Graves C (2019) Selective high-temperature CO2 electrolysis enabled by oxidized carbon intermediates. Nat Energy 4:846–855 Song Y, Zhang X, Xie K, Wang G, Bao X (2019) High-temperature CO2 electrolysis in solid oxide electrolysis cells: developments, challenges, and prospects. Adv Mater 31:1902033 Stoots CM, O’Brien JE, Stephen Herring J, Hartvigsen JH (2008) Syngas production via hightemperature coelectrolysis of steam and carbon dioxide. J Fuel Cell Sci Technol 6 Suboti´c V, Schluckner C, Mathe J, Rechberger J, Schroettner H, Hochenauer C (2015) Anode regeneration following carbon depositions in an industrial-sized anode supported solid oxide fuel cell operating on synthetic diesel reformate. J Power Sources 295:55–66 Suboti´c V, Schluckner C, Schroettner H, Hochenauer C (2016) Analysis of possibilities for carbon removal from porous anode of solid oxide fuel cells after different failure modes. J Power Sources 302:378–386 Tan Z, Ishihara T (2020) Effect of Ni-based cathodic layer on intermediate temperature tubular electrolysis cell using LaGaO3-based electrolyte thin film. J Phys Energy 2:024004 Tao Y, Ebbesen SD, Mogensen MB (2014) Carbon deposition in solid oxide cells during co-electrolysis of H 2O and CO2. J Electrochem Soc 161:F337–F343 Tao Y, Shao J, Cheng S (2016) Electrochemically scavenging the silica impurities at the Ni–YSZ triple phase boundary of solid oxide cells. ACS Appl Mater Interfaces 8:17023–17027 Tavares MT, Alstrup I, Bernardo CA, Rostrupnielsen JR (1994) CO disproportionation on silicasupported nickel and nickel-copper catalysts. J Catal 147:525–534 Tietz F, Sebold D, Brisse A, Schefold J (2013) Degradation phenomena in a solid oxide electrolysis cell after 9000 h of operation. J Power Sources 223:129–135 Trini M, Hauch A, De Angelis S, Tong X, Vang Hendriksen P, Chen M (2020) Comparison of microstructural evolution of fuel electrodes in solid oxide fuel cells and electrolysis cells. J Power Sources 450:227599

Fuel Electrode Materials for Solid Oxide Electrolysis Cells (SOECs)

115

Tsekouras G, Neagu D, Irvine JTS (2013) Step-change in high temperature steam electrolysis performance of perovskite oxide cathodes with exsolution of B-site dopants. Energy Environ Sci 6:256–266 Uchida H, Osada N, Watanabe M (2004) High-Performance electrode for steam electrolysis. Electrochem Solid-State Lett 7:A500 Uchida H, Watanabe S, Tao Y, Osada N, Watanabe M (2007) Double layer-type electrodes for reversible solid oxide fuel cells. ECS Trans 7:365–371 Wang S, Inoishi A, Hong J, Ju Y, Hagiwara H, Ida S, Ishihara T (2013) Ni–Fe bimetallic cathodes for intermediate temperature CO2 electrolyzers using a La0.9Sr0.1Ga0.8Mg0.2O3 electrolyte. J Mater Chem A 1:12455–6 Xie K, Zhang Y, Meng G, Irvine JTS (2011) Direct synthesis of methane from CO2/H2O in an oxygen-ion conducting solid oxide electrolyser. Energy Environ Sci 4:2218–2222 Xing R, Wang Y, Zhu Y, Liu S, Jin C (2015) Co-electrolysis of steam and CO2 in a solid oxide electrolysis cell with La0.75Sr0.25Cr0.5Mn0.5O3−δ –Cu ceramic composite electrode. J Power Sources 274:260–264 Yang L, Xie K, Wu L, Qin Q, Zhang J, Zhang Y, Xie T, Wu Y (2014) A composite cathode based on scandium doped titanate with enhanced electrocatalytic activity towards direct carbon dioxide electrolysis. Phys Chem Chem Phys 16:21417–28 Yang Y, Li Y, Jiang Y, Zheng M, Hong T, Wu X, Xia C (2018) The electrochemical performance and CO2 reduction mechanism on strontium doped lanthanum ferrite fuel electrode in solid oxide electrolysis cell. Electrochimica Acta 284:159–67 Ye L, Zhang M, Huang P, Guo G, Hong M, Li C, Irvine JTS, Xie K (2017) Enhancing CO2 electrolysis through synergistic control of non-stoichiometry and doping to tune cathode surface structures. Nat Commun 8:14785 Yue X, Irvine JTS (2017) Modification of LSCM–GDC cathodes to enhance performance for high temperature CO2 electrolysis using solid oxide electrolysis cells (SOECs). J Mater Chem A 5:7081–7090 Zhan Z, Kobsiriphat W, Wilson JR, Pillai M, Kim I, Barnett SA (2009) Syngas production by coelectrolysis of CO2/H2O: the basis for a renewable energy cycle. Energy Fuels 23:3089–3096 Zhang Y-Q, Li J-H, Sun Y-F, Hua B, Luo J-L (2016) Highly active and redox-stable Ce-doped LaSrCrFeO-based cathode catalyst for CO2 SOECs. ACS Appl Mater Interfaces 8:6457–6463 Zhang X, Song Y, Guan F, Zhou Y, Lv H, Wang G, Bao X (2018) Enhancing electrocatalytic CO2 reduction in solid oxide electrolysis cell with Ce0.9Mn0.1O2−δ nanoparticles-modified LSCM-GDC cathode. J Catal 359:8–16 Zhang J, Xie K, Wei H, Qin Q, Qi W, Yang L, Ruan C, Wu Y (2014) In situ formation of oxygen vacancy in perovskite Sr0.95Ti0.8Nb0.1M0.1O3 (M = Mn, Cr) toward efficient carbon dioxide electrolysis. Scientific Reports 4:7082 Zhou Y, Zhou Z, Song Y, Zhang X, Guan F, Lv H, Liu Q, Miao S, Wang G, Bao X (2018) Enhancing CO2 electrolysis performance with vanadium-doped perovskite cathode in solid oxide electrolysis cell. Nano Energy 50:43–51

Ceramic Coatings for Metallic Interconnects and Metal Alloys Support for Solid Oxide Electrolysis Applications Elisa Zanchi, Antonio Gianfranco Sabato, Hassan Javed, Agnieszka Drewniak, Damian Koszelow, Sebastian Molin, and Federico Smeacetto

1 Introduction In a solid oxide electrolysis cell (SOEC) system, the metallic interconnect generally provides the electrical connection for a SOEC stack and gas separation between the cells, as well as gas distribution across the cells. Furthermore, the role of the metallic interconnect to support ceramic element layers with a low-cost and robust material, which is expected to lower the cost of solid oxide cells (SOCs) devices, has growth of interest for these materials in the field of metal-supported solid oxide electrolysis cells (MS-SOECs). Aluminum-containing alloys cannot be used as a support in MS-SOEC devices due to material requirements, since the alumina scale has too poor electrical conductivity, despite proving extremely good corrosion properties. Therefore, the best candidates for MS-SOEC applications seem to be chromia-forming ferritic alloys (Shaigan et al. 2010). Stainless steels are widely considered the most promising candidates as stateof-the-art interconnect materials due to their electrically-conducting oxide scale

E. Zanchi · F. Smeacetto (B) Department of Applied Science and Technology, Politecnico di Torino, Corso Duca Degli Abruzzi 24, 10129 Torino, Italy e-mail: [email protected] A. G. Sabato Institut de Recerca en Energia de Catalunya (IREC), Jardins de Les Dones de Negre, 1, 2ª Pl., 08930 Sant Adrià de Besòs, Barcelona, Spain H. Javed Sunfire GmbH, Gasanstaltstraße 2, 01237 Dresden, Germany A. Drewniak · D. Koszelow · S. Molin Faculty of Electronics, Telecommunications and Informatics, Gdansk University of Technology, Ul. G. Narutowicza 11/12, 80-233 Gdansk, Poland © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M. A. Laguna-Bercero (ed.), High Temperature Electrolysis, Lecture Notes in Energy 95, https://doi.org/10.1007/978-3-031-22508-6_6

117

118

E. Zanchi et al.

and an appropriate thermal expansion coefficient (TEC). Specifically, chromiaforming stainless steels are commonly used as interconnect materials because of TEC matching with adjacent materials, low cost, and good formability. Good oxidation resistance is provided by the formation of a continuous Cr2 O3 scale with a relatively low Cr diffusion and an acceptable electronic conductivity at the operating temperatures (Quadakkers et al. 2003). The deposition of specific ceramic protective coatings onto the interconnects has been demonstrated as an effective method to limit the degradation issues and to increase the lifetime of the SOEC stack by lowering the corrosion rate and blocking the chromium evaporation. There is a growing body of literature that recognizes the importance of ceramic coatings as key materials in determining the durability and performance of SOEC stacks. An effective coating for SOEC metallic interconnect must specifically fulfill the following requirements: • Reduce the oxidation rate of the steel substrate by limiting the oxygen inward diffusion as well as the Cr outward diffusion • Limit the Cr evaporation rate and the consequent electrodes poisoning • Ensure high electrical conductivity to minimize the ohmic losses • Good thermo-chemical stability with respect to the other stack components • Good thermo-mechanical compatibility with the steel substrate Enhanced efficiency of SOEC stacks can be reached only by appropriate material choice with good performance and functional requirements. In addition to a high electrical conductivity, another key requirement for the coating is their TEC, which should match the stainless steel, in order to avoid spallation or delamination phenomena. The spinel family has attracted great attention as reliable coating materials; some of the spinel compositions investigated are Co–Mn, Cu–Mn, Cu–Fe, Mn–Fe, and Ni– Fe. To date, several studies have shown that the Co–Mn spinel family has demonstrated the best performances in terms of functional requirements and properties; the existing literature on Mn–Co spinel is extensive and focuses particularly on deposition methods, tests in relevant conditions, etc. Modified Mn–Co-based spinels have started attracting significant attention as alternative compositions, and recent studies dive further into the feasibility of using different spinel compositions. The first section of the subchapter is focused on the proposed solutions as ceramic coatings and their compositional and functional modifications, describing the most common deposition methods, the influence on their chemical composition and the corrosion and the electrical behavior in relevant conditions during long-term test assessment. The results that are presented here will show how spinel coatings play a key role in the interconnect performance, while pointing out some unanswered questions about their functional properties and some future perspectives for additional research.

Ceramic Coatings for Metallic Interconnects and Metal Alloys Support …

119

The first part of this section will examine different materials proposed as coatings for interconnects, starting from reactive elements oxides, rare earth perovskites to spinels. The second part deals mainly with different methods of modification of the Mn– Co spinel composition, specifically focusing on the effect of Cu and Fe addition and describing the optimization of the coating processing by the electrophoretic codeposition method. Most of the techniques related to the modifications of Mn–Cobased material by electrophoretic co-deposition methods are critically presented, and the new horizons that are opening up, to improve the performances of spinel coatings for SOEC interconnects application, are discussed. This subsection will also discuss how the implementation of electrophoresis technique can offer innovative solutions when adapted with other slurry-based deposition techniques (i.e., dip coating or electrolytic deposition) and perspectives for an industrial upscale of the electrophoretic deposition technique. The second section of this subchapter concernes with an overview of porous metal alloys and their relevant role as metal-supported solid oxide electrolysis cells (MS-SOECs), from methods of production, moving to their corrosion properties and potential innovative tests to assess their interface stability, i.e., with glass-based sealings. The conclusion section presents and summarizes the findings of some specific aspect of this research, focusing on the two key themes that have been reviewed and discussed, thus making some recommendations for future research work and progress on durability issues of solid oxide electrolysis cells devices.

2 Ceramic Coatings for Metallic Interconnects in Solid Oxide Electrolysis Cells: Deposition Methods, Compositional, and Functional Modifications 2.1 Materials Different protective coatings and deposition techniques have been developed in the recent years in order to limit the Cr-poisoning and an excessive oxide scale growth. The characteristics of each material used as protective coating have to be considered also in relation with the properties of the metallic substrate under the point of view of possible chemical interactions (especially with Cr) and coefficient of thermal expansion. In Table 1, the most used ferritic stainless steels up to now are summarized with their Cr content and their TEC. The protective materials typically used today for this application can be classified in three main categories: reactive elements oxides, rare earth perovskites, and spinelbased, as shown in Table 2, where an overview of the main properties is given.

120

E. Zanchi et al.

Table 1 Cr content and TEC of common ferritic stainless steels used as interconnect for SOC stacks (Thyssenkrupp 2001a, b; VDM-Metals 2010; ThyssenKrupp VDM GmbH 2012)

Steel

Cr (wt.%)

TEC (10–6 K−1 )

Crofer22APU

22

12.6

Crofer22H

22

12.3

AISI430

16.5

12

AISI441

18

10.5

Table 2 Classification and properties of materials for protective coatings Coating material

Electronic conductivity

Cr migration inhibition

Oxidation rate reduction

Simplicity of deposition

Reactive elements oxide

Fair

Poor

Good

Good

Rare earth perovskites

Good

Fair

Poor

Fair

Spinels

Good

Good

Fair

Good

2.1.1

Reactive Element Oxides (REO)

In the past, the effect of deposition of reactive elements (i.e., Y, La, Ce, etc.) oxides (REO) or their combination on the surface of chromia-forming stainless steels has been widely investigated (Hou 2000; Qi and Lees 2000; Chevalier and Larpin 2002). The presence of these elements has been demonstrated to be effective in improving the adhesion of the oxide scale to the steel and to limit its oxidation, thus reducing the area specific resistance (ASR) which is directly related to the oxide scale thickness. Many studies reported the deposition of these materials on metallic interconnects for SOC applications (Qu et al. 2004; Fontana et al. 2007, 2012; Piccardo et al. 2007; Shaigan et al. 2010; Mah et al. 2017a, b). Fontana et al. (2012) demonstrated the effectiveness of La2 O3 and Y2 O3 deposition on Crofer22APU in oxidizing atmosphere up to more than 23 kh at 800 °C in air. In Fig. 1, it is possible to see how the application of this REO coatings sensibly reduced the oxidation of the underlying steel even during a significant aging (30 months) especially in the case of La2 O3 . REO coatings are generally very thin (< 1 μm), and for this reason, despite their excellent effect on the oxidation, they are considered not to be effective in limiting the Cr evaporation. Indeed, in recent years, the trend has been to couple reactive elements layer with Co, to reduce the Cr evaporation. In this sense, the group of Jan Froitzheim and co-workers extensively studied the application of Ce/Co coatings on different stainless steels for SOC application, investigating the effects of these coatings on ASR and Cr evaporation in different relevant conditions (Magrasó et al. 2015; FalkWindisch et al. 2017; Goebel et al. 2020; Goebel et al. 2021). Moreover, Goebel et al. (2020) recently conducted a very extended study in which Ce/Co coatings were applied to AISI 441, and their performances were investigated in oxidizing atmosphere at 800 °C up to 36 kh (4 years). In Fig. 2, the results in terms of ASR and Cr evaporation are reported. The application of Ce/Co was strongly effective under

Ceramic Coatings for Metallic Interconnects and Metal Alloys Support …

121

Fig. 1 Oxide scale thickness measured after different aging period in air at 800 °C for uncoated Crofer22APU, La2 O3 coated Crofer22APU, and Y2 O3 coated Crofer22APU. Reproduced from Fontana et al. (2012)

both points of view in comparison with uncoated AISI 441. The Cr evaporation resulted to be at least one order of magnitude lower in comparison with the uncoated steel while the ASR resulted to be 34 m] cm2 after the long aging, well below what is considered the limit value after 40 kh of aging (100 m] cm2 ) (Steele and Heinzel 2001). They recently reported also the self-healing effect of these coatings when exposed at 750 °C; the self-healing was effective after only 71 h of exposition at this temperature (Goebel et al. 2021).

Fig. 2 Long-term performances of Ce/Co coated AISI441 stainless steel in terms of Cr evaporation retention (a) and ASR (b). Reproduced from Goebel et al. (2020)

122

E. Zanchi et al.

Table 3 TEC of typical rare earth perovskites used as protective coatings for solid oxide cells metallic interconnects (Tan et al. 2019)

2.1.2

Material

CTE (10–6 k−1 )

LSM (La1-x Srx MnO3 )

12.5

LSCF (La1-x Srx Co1-x Fex O3)

15.4

SSCF (Sm1-x Srx Co1-x Fex O3 )

16

BSCF (Ba1-x Srx Co1-x Fex O3 )

16.6

LSCM (La1-x Srx Co1-x Mnx O3 )

12.5

LSCN (La1-x Srx Co1-x Nix O3 )

14.6

Rare Earth Perovskites

Perovskites are widely used in solid oxide cells technology, typically as cathode materials. Perovskites are characterized by ABO3 structure with the A site occupied by a trivalent cation (i.e., La), while B is occupied by a transition metal (i.e., Cr, Co, Fe, Mn, etc.). In oxidizing atmosphere and at high temperatures, these perovskites show p-type electronic conductivity. This, in addition to their TEC (see Table 3 TEC of typical rare earth perovskites used as protective coatings for solid oxide cells metallic interconnects (Tan et al. 2019).), makes these materials potentially suitable to act as protective coatings for the metallic interconnects. Furthermore, it is possible to modify the doping in order to tune their conductivity or TEC. Indeed, it is possible to dope the A site with large radii divalent cations (i.e., Sr, Ca) and B site with electron acceptors (i.e., Ni, Fe, Cu). The main perovskites investigated up to now as protective layer are based on: lanthanum chromates (LaCrO3 ) (Johnson et al. 2004; Zhu et al. 2004; Shaigan et al. 2008; Jeong et al. 2020), Sr-doped lanthanum chromate (La1-x Srx CrO3 ) (Belogolovsky et al. 2007; Paknahad et al. 2014), lanthanum strontium manganite (La1-x Srx MnO3 ) (Chu et al. 2009; Pyo et al. 2011) cobaltite (La1-x Srx CoO3 ) (Zhu et al. 2004; Persson et al. 2012), or ferrite (La1-x Srx FeO3 ) (Lee et al. 2010; Tsai et al. 2010). Despite their good conductivity in general, perovskite-based coatings demonstrated poor densification ability. This does not provide a sufficient barrier against Cr evaporation and thus against cathode poisoning (Shaigan et al. 2010; Mah et al. 2017a, b; Tan et al. 2019).

2.1.3

Spinels

In the recent years, the attention of the scientific community has been more focused on the development of spinel-based materials as protective coatings for metallic interconnects. Indeed, these materials demonstrated superior performances in comparison with both reactive elements oxides and rare earth perovskites, in terms of ASR and densification (Cr-retention ability). In addition, the spinel composition can be modified by proper doping, in order to slightly modify the properties of the coating (see next section “modification of the spinel composition”).

Ceramic Coatings for Metallic Interconnects and Metal Alloys Support … Table 4 TEC and electrical conductivity values at high temperatures of typical spinel systems used as protective coatings for SOC applications (Shaigan et al. 2010; Mah et al. 2017a, b; Tan et al. 2019; Zhu et al. 2021)

Spinel system

TEC (10–6 K−1 )

123 σ (S cm−1 )

Mn–Co

7.4–14

21–157

Cu–Fe

11.2

2.3–9.1

Ni–Fe

11.7–12.1

0.05–15

Cu–Mn

11–12.6

51–100

Different binary spinels were considered for this application with different cations in the A and B sites of the spinel generic formula AB2 O4 in order to reach the suitable values of conductivity and TEC (Shaigan et al. 2010; Mah et al. 2017a, b; Tan et al. 2019; Zhu et al. 2021). The main investigated systems are reported in Table 4 with the corresponding values of TEC and electrical conductivity. These values can vary in a range depending from the specific composition of the spinel, synthesis, and sintering methods as well as operating temperature. In the case of Mn–Co-based spinels, a wide range of compositions (Ax B3-x O4 with 1 < x < 2) have been investigated and resulted to be suitable as protective material (Chesson and Zhu 2020), in terms of ASR and Cr-retention ability even after very long aging at high T in air. Among the different considered spinels, the ones resulted more suitable for typical SOC applications were Mn2 CoO4 , MnCo2 O4, and their combination (Zhu et al. 2021). Smeacetto et al. (2015); (Molin et al. 2017; Sabato et al. 2021) extensively studied these protective coatings, especially deposited by the electrophoretic deposition (EPD) method and tested for long aging periods (> 2000 h) in air at operating temperatures (750–850 °C). In Figs. 3 and 4, two relevant results are reported. Mn1.5 Co1.5 O4 coatings demonstrated excellent performances in terms of ASR (Fig. 3) up to 2500 h at 800 °C, in comparison with uncoated Crofer22APU. The overall ASR value after 2500 h of aging at 800 °C in air resulted to be ≈20 m]•cm2 versus the one of uncoated sample of ≈ 30 m] cm2 . In addition, the degradation rate in case of bare Crofer22APU is much higher (≈ 5.3 m] cm2 /kh) in comparison with the coated Crofer22APU (≈ 0.51 m] cm2 /kh). In Fig. 4, it possible to observe the Cr retaining effect of the protective layer after 3000 h stack test at 850 °C. Energy dispersive X-ray spectroscopy (EDS) maps did not detected relevant Cr outward diffusion from the steel substrate to the coating during this period. Furthermore, the Cr2 O3 oxide scale appears to be ~ 2 μm thick after the aging, thus highlighting the effect of the spinel coating on limiting its growth (in accordance with a low ASR). Mn–Co spinels can be produced in tetragonal structure (Mn2 CoO4 ) or in cubic structure (MnCo2 O4 ), depending on the relative amount of Mn and Co. The tetragonal phase is not stable at the SOEC operating conditions, as it undergoes a tetragonal-tocubic phase transformation around 450–600 °C. Different compositions of Mn–Co spinel were investigated with different Mn/Co ratio trying to clarify the effect of compositional changes on the electrical conductivity and TEC; a mixture of the tetragonal and cubic phase (referred with the formula Mn1.5 Co1.5 O4 ) can be used. Concerning the electrical conductivity, a strong discrepancy is reported in literature

124

E. Zanchi et al.

Fig. 3 ASR of Crofer22APU bare and coated with Mn1.5 Co1.5 O4 in air at 800 °C for 2500 h. Reproduced from Smeacetto et al. (2015)

with very different values ranging from 20 to more than 100 S cm−1 at 800 °C (Zhu et al. 2021). This is related not only to the specific composition of the spinel (Mn/Co) but also to the powder synthesis and to the sintering treatment as well. Despite this huge range, all these values are considered more than enough for a protective coating since is commonly accepted that the mayor contribution to the resistance of the interconnects is given by Cr2 O3 (σ ~ 0.01 S cm−1 ) growth and formation of resistive reaction products with the coatings (Goebel et al. 2018). A recent study by Chesson et al. (2020) gave a significant contribution to this topic clarifying the properties of different Mn–Co spinels with different compositions. Their results (summarized in Fig. 5) showed that with the increasing content of Mn, both the TEC and the conductivity decrease. For this reason, a suitable compromise for SOC application can be identified in intermediate compositions (Mnx Co3-x O4 , with 1.2 < x < 1.5). The application of Mn–Co spinel-based coatings on chromia-forming ferritic stainless steel often leads to the formation of a (Mn,Co,Cr)2 O3 reaction layer at the interface with the Cr2 O3 scale due to interdiffusion of elements (Magdefrau et al. 2013; Gambino et al. 2015; Talic et al. 2019). A schematic mechanism is reported in Fig. 6 together with a schematization of the different layers and the corresponding equivalent circuit. Due to the lower conductivity (σ = 0.05 S cm−1 ) of this reaction layer in comparison with the coating itself the excessive growth of this reaction layer can have a detrimental effect on the ASR of the system. The thickness of the reaction layer is directly related to the sintering treatment of the coatings, and to the oxygen diffusion, it is crucial to have a good densification of the protective coating also to limit this phenomenon. However, the main contribution to the resistance is still represented by the Cr2 O3 which has a conductivity of ~ 0.01 S cm−1 , which is five times lower than the typical conductivity reported for the reaction layer (Zhu et al. 2021). In recent years, also different Co-free spinels have been investigated as well, since Co is considered a critical raw material for energy applications. In particular, spinels based on the systems Ni–Fe (You et al. 2018, 2020; Chesson and Zhu 2020), CuFe (Pan et al. 2021a; Pan et al. 2021b) and Mn–Cu (Waluyo et al. 2014; Hosseini

Ceramic Coatings for Metallic Interconnects and Metal Alloys Support …

125

Fig. 4 EDS elemental mapping of a thin lamella collected at the interface Crofer22APU/Mn1.5 Co1.5 O4 after an aging of 3000 h in air at 800 °C. Reproduced from Sabato et al. (2021)

Fig. 5 Variation of TEC (a) and conductivity (b) in Mn-Co spinel systems with different Mn and Co content. Reproduced from Chesson and Zhu (2020)

126

E. Zanchi et al.

Fig. 6 Schematization of interdiffusion of elements at the interface of a chromia-forming stainless steel coated with Mn–Co spinel in oxidizing atmosphere at high temperatures

Fig. 7 ASR performance of Mn–Cu spinel protective coatings on Crofer22APU (a) and AISI430 (b). Reproduced from Waluyo et al. (2014) and Hosseini et al. (2015)

et al. 2015; Spotorno et al. 2015; Sun et al. 2018; Wang et al. 2018) have attracted more attention. Among these, the Mn–Cu spinel-based coatings have been extensively studied with promising results (Fig. 7). Despite this, there is still a lack in the evaluation of their behavior (mass gain, ASR, reactions, cathode poisoning) for long aging periods (>1000 h).

2.2 Modification of the Spinel Composition The Mn-Co spinel structure can accommodate a wide range of different cations: The position occupied by each cation in the spinel lattice defines the degree of distortion of the crystal structure, with major effects on the functional properties of the spinel. For

Ceramic Coatings for Metallic Interconnects and Metal Alloys Support …

127

these reasons, performing modification of the composition of the Mn–Co spinel has been identified as a promising route to improve the functional properties of the base spinel structure. This spinel coating modification is normally referred as “doping” and can be achieved either by substitution of part of Mn or Co of the base spinel or by simple addition, depending on the synthesis method. Modifications of the Mn–Co spinel composition have been focused on doping by transition metals such as Cu, Fe, Ni, Ag, or rare earth elements, such as Y, La, and Ce. The effect of the variation of the functional properties of the spinel depends on the specific doping element, as well as on its amount; however, doping the spinel structure does not generally cause the formation of new crystalline structures. Indeed, the doping element is accommodated in the spinel structure replacing either Co or Mn in the tetrahedral or octahedral sites; this substitution causes a distortion of the crystalline lattice. To this purpose, doping by addition of transition elements is reported to bring a significant impact on the electrical conductivity and on the coefficient of thermal expansion. Results found in literature generally present a relevant scattering of properties, due to different experimental setup and testing conditions. However, most of the studies focus on doping by copper or iron. There is general accordance on Cu increasing both electrical conductivity and TEC of the Mn–Co spinel. The reason found is the stabilization of the cubic spinel structure lead by copper addition. Indeed, the base Mn–Co spinel exhibits a tetragonal-cubic transition that depends on the specific Mn to Co ratio in the composition; the highest electrical conductivity is achieved by increasing the cobalt content (MnCo2 O4 spinel with a full cubic structure) (Brylewski et al. 2014). When copper is introduced in the spinel, even higher conductivity values have been reported; moreover, copper addition also leads to an enhanced sinterability, allowing to reduce sintering time and temperature to obtain satisfactory densification levels (Masi et al. 2016; Mah et al. 2017a, b; Masi et al. 2017). Following these considerations, together with the fact that cobalt is a critical raw material, the implementation of cobalt-free Mn–Cu spinel coatings is recently gaining a lot of attention in research. However, compared to the Mn–Co spinel, the stability range of the Mn–Cu spinel is significantly narrower, meaning that CuO tends to segregate and that higher sintering temperature is needed to obtain densification suitable for application as a protective coating. For these reasons, still few studies focus on Mn–Cu spinel coatings, and further research needs to assess their protective properties (Ignaczak et al. 2020). On the other hand, iron addition to the Mn–Co spinel has an opposite effect on the mentioned properties compared to copper. Indeed, Fe addition is responsible for increasing the lattice parameter of the Mn–Co spinel cubic structure resulting in lower electrical conductivity due to the lower possibility of polaron hopping (Liu et al. 2013; Talic et al. 2018a). Similarly, the lower electrical conductivity is associated with a decrease in the TEC with increasing Fe content in the spinel (Masi et al. 2016). Regarding the spinel densification, Fe addition is reported to have small influence or to partially reduce the sinterability of the coating; results reported in literature do not fully agree due to different synthesis, deposition, and sintering methods. However, the

128

E. Zanchi et al.

Fig. 8 SEM backscatter images of unmodified (MC), Fe-doped (MCFe), and Cu-doped (MCCu) spinel coatings on Crofer22APU. Reproduced from Talic et al. (2017)

major influence of iron addition lays on the improvement of the oxidation resistance, represented by the development of a thinner oxide scale during aging. Indeed, it is apparent that Fe presence in the coating layer next to the oxide scale partially reduces the diffusion rate of chromium (Zanchi et al. 2019; Zanchi et al. 2020). Figure 8 compares coating morphology of the unmodified spinel coating (MC) with the Fe-doped (MCFe) and Cu-doped (MCCu) coatings, highlighting the different residual porosity degree (Talic et al. 2017). A high degree of densification is essential to guarantee satisfactory protection properties of the coating, in order to limit oxygen inward diffusion to the steel interconnect (Bobruk et al. 2018; Zanchi et al. 2021). However, coating densification depends not only on the coating composition but also on the deposition method. To this purpose, various deposition techniques can be exploited to deposit copper or iron-doped Mn–Co-based coatings. Dense Cu-doped coatings have been obtained by high-energy micro-arc alloying using targets of metallic alloys produced by argon arc melting; however, many hours of oxidation are required to form homogeneous oxide layers with this technique (Guo et al. 2018; Guo et al. 2020). Also highly dense Fe-doped spinel coatings have been produced by atmospheric plasma spray, but thermal decomposition of the spinel was encountered during deposition from the target (Puranen et al. 2014). In general, majority of the studies focuses on slurrybased deposition methods. Dip coating followed by two-step sintering of Cu-doped Mn–Co spinel synthetized by citric acid nitrate process is quite reported in literature (Park et al. 2013; Chen et al. 2015; Xiao et al. 2016; Li et al. 2017); on the other hand, slurry painting followed by repeated stages of infiltration and sintering has been used for Fe-doped spinels previously synthetized by sol–gel method (Molin et al. 2016). An innovative deposition technique is represented by the inkjet printing from water-based inks of iron-modified spinel; despite the good potential of the technique, the performance of the obtained coatings still needs to be assessed (Pandiyan et al. 2020). Apart from copper and iron, fewer studies have been focused on Ni- and Ag-doped Mn–Co spinel. Although available results are limited, nickel is suggested to improve the electrical conductivity and sinterability compared to the unmodified spinel, but

Ceramic Coatings for Metallic Interconnects and Metal Alloys Support …

129

by a lower degree than copper (Park et al. 2013). Doping with silver can significantly increase the electrical conductivity of the Mn–Co spinel; however, the long-term stability of the coating, as well as the sustainability of using a precious metal in a large-scale production of the SOC device, still needs to be assessed (Lee et al. 2017). As anticipated in the previous section, rare earth elements can reduce the steel oxidation rate and favor better adhesion of the oxide scale (Hou and Stringer 1995; Seo et al. 2008; Hassan et al. 2020). In recent years, some studies have demonstrated that rare earth elements can be employed to modify the Mn–Co spinel. For example, lanthanum-doped Mn–Co spinel has been produced by high-energy microarc alloying (HEMAA) process using targets obtained by argon arc melting (Guo et al. 2018; Guo et al. 2020). Similarly, La- and Ce-doped Mn–Co spinel coatings have been deposited by magnetron sputtering from commercial sputter targets consisting of the doped spinels: In this case, doping with lanthanum was proved to be strongly effective in reducing oxidation rate and chromium vaporization from the steel substrate in comparison with the unmodified spinel and Ce-doping as well (Tseng et al. 2020). A Ce-doped Mn–Co synthesized by glycine-nitrate combustion synthesis method and deposited by screen printing on AISI 441 showed similar behavior than the undoped Mn–Co spinel coating in the study of Stevenson et al. (Stevenson et al. 2013). The possible application of Y-doped Mn–Co spinel processed by sol–gel method has been investigated by Xin et al. (Xin et al. 2011): the modified coating applied through a spinel powder reduction coating technique expressed satisfactory properties as protective coating. More recently, a Cu-Y-doped Mn–Co spinel obtained by glycine-nitrate process has been proposed: the coating was deposited by screen printing and demonstrated promising behavior with reference to the uncoated steel substrate, but no comparison with performance of the unmodified coating is provided in the study (Thaheem et al. 2021).

2.3 Optimization of the Coating Deposition by Electrophoretic Co-Deposition Method As reviewed in the previous paragraph, the doped spinels have been synthetized before coating deposition: Spray pyrolysis, citric acid nitrate process, Pechini method, and solid-state reaction are some of the synthesis techniques reported in literature. After the synthesis, doped spinels have been deposited through dip coating, slurry painting, screen and inkjet printing, plasma spray, and magnetron sputtering. Other studies focused on the use of high-energy micro-arc alloying (HEMAA) technique to deposit from targets of modified metallic alloys obtained from argon arc melting. All the mentioned techniques require multi-step processes, involving first the spinel synthesis or the metallic alloy preparation and then the coating deposition and can be grouped as “ex situ” spinel modification routes. Electrophoretic deposition (EPD) allows to deposit well adherent layers of ceramic particles in few seconds and by means of a simple and adaptable setup.

130

E. Zanchi et al.

Thanks to the versatility for materials employed and coatings morphology, as well as the low-energy demand, EPD is considered a suitable technique for industrial applications in SOC technology (Kalinina and Pikalova 2021; Zhu et al. 2021). Electrophoretic deposition of base spinel was first proposed in 2015 as a viable and effective deposition technique for SOC metallic interconnect coatings (Smeacetto et al. 2015), then followed by numerous studies highlighting the advantages of EPD in reducing processing time and cost of the interconnect (Zanchi et al. 2021). Talic et al. (2017) have demonstrated the feasibility to process Cu or Fe modified Mn–Co spinel with controlled amount of doping elements via spray pyrolysis from aqueousbased nitrate solutions; the obtained powders have been successfully deposited by electrophoretic deposition using a fully organic solvent mixture. More recently, commercial Cu–Mn–Co commercial powder was deposited by EPD, and various organic solvent media were compared (Aznam et al. 2021). In both cases, the green layers were stabilized by a two-step sintering, consisting of a reduction treatment followed by a re-oxidation: Indeed, this approach is generally accepted as the optimal post-deposition treatment to densify protective spinel coatings. Furthermore, some attempts must be made to optimize the sintering procedure, toward faster and cheaper methods. For example, Javed et al. (Javed et al. 2021) have proposed an innovative sintering procedure involving the use of pressure-assisted flash sintering, requiring just few minutes to obtain highly densified Mn–Co spinel coating. EPD of base and modified spinel powders has already demonstrated promising results, in terms of both processing and performances of the coatings, and it is now raising interest, thanks to the possibility of obtaining doped spinel coatings by a co-deposition method. This upgrade of the technique allows to modify the coating composition “in situ”, i.e., during the deposition process, reducing processing time and cost compared to ex situ synthesis methods. This process does not require any modification of the deposition setup or instrument configuration compared to the electrophoretic deposition of a single element; however, special attention must be paid to the preparation of the EPD suspension and choice of deposition parameters. Indeed, the first and simplest kinetic model of electrophoretic deposition proposed by Hamaker in 1940 correlates the deposited mass m [g] to the concentration Cs [g cm−3 ] and the mobility μ [cm2 s−1 V−1 ] of the solid particles, the electric field E [V cm−1 ], the deposition area S [cm2 ], and deposition time t [s] through the following equation (Hamaker 1940): m = Cs μE St However, when a co-deposition is performed, the reported formula can result further complicated by the addition of factors for each material inserted in the suspension. Moreover, parameters regarding the mutual interaction between the different colloidal particles used should be considered as well. Indeed, the deposition kinetic can significantly vary depending on the material employed, and the possibility of deposit single powders alone is no guarantee of success of the co-deposition process.

Ceramic Coatings for Metallic Interconnects and Metal Alloys Support …

131

On practical level, the following parameters must be taken into consideration when optimizing a suspension for the electrophoretic co-deposition process: • Selection of the powders: As for electrophoresis of single element, selected powders to co-deposit must have a suitable particle size distribution and develop a surface charge high enough to avoid flocculation and sedimentation. Although there is not a general cut-off value of particle size for electrophoretic deposition, the diameter of employed powders is commonly < 1 μm. Moreover, in the case of co-deposition, it is important to consider interactions between the different particles • Formulation of the suspension: The powders to co-deposit must be all stable (not decompose) in the solvent or the mixture of liquid media selected for the suspension. Moreover, the environmental impact of the liquid medium must be considered, giving priority to water-based solutions. When a stable suspension cannot be reached, making use of stabilizers or dispersant agents is necessary. • Post-deposition treatment: Electrophoretic deposition allows to deposit well packed layers of particles; however, performing a post-deposition sintering treatment is always necessary in order to stabilize and densify the green coating and to achieve good protective performances. When electrophoretic co-deposition is performed, the optimized sintering profile usually foresees two subsequential thermal treatments. A first treatment in reducing atmosphere (a mixture of Ar/H2 or N/H2 ) ensures a partial reduction of Mn–Co spinel in metallic Co and MnO; in the case or iron doping, Fe2 O3 is also reduced, forming an intermetallic Co–Fe compound. The subsequent heat treatment is performed in air and leads to the re-oxidation of the reduced coating, with the formation of the final spinel structure. Depending on the doping element, the spinel structure can stabilize either the cubic or the tetragonal crystalline phase. In the case of iron doping, the cubic phase is partially stabilized, together with a distortion of the lattice (Zanchi et al. 2019): As shown in Fig. 9, the degree of lattice distortion depends on the amount of Fe addition to the spinel, and it is visible as a shift of the diffraction peak toward lower 2Theta angles. Cu doping is reported to stabilize the cubic phase without lattice distortion appreciable by XRD analysis. Up to now, the electrophoretic co-deposition technique has been successfully exploited to produce Cu-doped (Molin et al. 2018; Sabato et al. 2019) and Fe-doped Mn-Co spinel coatings (Zanchi et al. 2019; Zanchi et al. 2020). Coating processing approach as well as the evaluation of performances at SOC relevant conditions are reviewed and discussed in the following section. As schematically represented in Fig. 10, the electrophoretic deposition mechanism through which Cu or Fe doping has been performed is opposite, although both the same suspension and deposition parameters have been employed. Indeed, in the case of Cu doping, zeta potential measurements proved both coating precursors developed a positive surface charge in the suspension: + 13 mV for Mn1.5 Co1.5 O4 and + 6 mV for CuO. Despite the lower surface charge developed by CuO, deposition performed at 50 V for 30 s was effective to co-deposit both precursors preserving their relative amount in the EPD suspension in the deposited coating as well. On the other hand, Fe2 O3 develops a negative zeta

132

E. Zanchi et al.

Fig. 9 X-ray diffraction patterns of as-sintered Fe-doped Mn–Co spinel coatings, reproduced form Zanchi et al. (2019). The diffraction pattern of the unmodified Mn1.5 Co1.5 O4 (MCO) is compared with those of Mn1,43 Co1,43 Fe0,14 O4 (5FeMCO) and Mn1,35 Co1,35 Fe0,30 O4 (10FeMCO)

potential of − 10 mV; however, a fully cathodic co-deposition is preserved up to 10 wt.% addition of iron oxide to the Mn1.5 Co1.5 O4 suspension. Indeed, the electrostatic interactions between opposite charges and the smaller dimension of the iron precursor in comparison with Mn–Co spinel resulted in the deposition mechanism reported in the figure. Coatings obtained by electrophoretic co-deposition were processed and tested at slightly different conditions, allowing to reveal properties brought by various additions of Cu or Fe. Both Cu- and Fe-doped coatings were sintered by two-step treatment, with reducing and re-oxidizing steps performed at 900 °C. Moreover, only Fe-doped coatings were reduced also at 1000 °C, to evaluate whether the higher sintering temperature could improve the sinterability of the spinel.

Fig. 10 Schematic diagram showing the electrophoretic deposition process of Mn–Co spinel-based coatings. a ex situ-doped spinel; b in situ copper doping; c in situ iron doping. Reproduced form Zanchi et al. (2021)

Ceramic Coatings for Metallic Interconnects and Metal Alloys Support …

133

Copper doping leads to great densification, as clearly visible in Fig. 11. The residual porosity was found to be dependent on both the amount of copper addition and oxidation time due to a continuous sintering effect. However, coatings obtained with the highest doping level (labeled as 10CuMCO in Fig. 11) underwent pores coalescence during 3000 h aging at 800 °C: This effect can be detrimental for protection properties as it can facilitate cracks propagation and suggest that a lower amount of copper addition would be favorable. In terms of oxidation resistance and area specific resistance measured at 800 °C on Crofer22APU substrate, Cu-doped Mn-Co coatings demonstrated similar results than the undoped spinel; indeed, as suggested by the review of the previously published works, doping seems to have smaller influence on coatings properties at highest working temperature, due to more sever oxidation phenomena (Zanchi et al. 2021). To this purpose, Fe-doped coatings tested at 750 °C, expressed clearer trends depending on doping level, as visible from oxidation resistance test results reported in Fig. 12 (Zanchi et al. 2019). Indeed, despite similar values of residual porosity, iron addition was proved to significantly reduce the growth rate of the oxide scale compared to the base spinel coating (MCO), bringing beneficial effects on the oxidation rate. Furthermore, Fe-doped coatings reduced at higher temperature (named 10FeMCO_R1000)

Fig. 11 Cross-section SEM images of the coated samples after different aging periods: as-sintered (a, b, and c), after 1000 h (d, e, and f) and after 3000 h (g, h, and i) reproduced from Sabato et al. (2019). The picture compares the densification level of undoped coating (MCO) and Cu-doped coatings (5CuMCO and 10CuMCO) with aging time

134

E. Zanchi et al.

exhibited slightly better densification, which resulted essential to limit the degree of internal oxidation and obtain the best performances. Conductivity measurements performed on sintered pellets confirmed higher resistivity for Fe-doped spinel at higher temperatures as well as thermal activated electronic conduction, in agreement with the results reported in various studies (Masi et al. 2016; Talic et al. 2018a). Results are reported in Fig. 13. However, areas specific resistance test monitored for more than 3000 h at 750 °C for both Crofer22APU and AISI 441 coated steel revealed that the composition of the alloy substrate has a major influence on defining the degradation mechanism and the performance of the interconnect (Zanchi et al. 2020). Indeed, the minor presence of residual impurities in the cheaper AISI 441 alloy is believed to modify diffusion mechanisms with benefits for coating densification and conductivity of the oxide scale. The discussion on the influence of the steel substrate results even more relevant considering the current attempts in progressively reducing the operating temperature

Fig. 12 a Mass gain and b parabolic rate plot of coated and bare Crofer22APU during cyclic oxidation at 750 °C in air, 2000 h. Reproduced from Zanchi et al. (2019)

Fig. 13 Results of electrical conductivity measurements (a) and microscopy images of samples microstructure: (b, c) unmodified spinel (MCO); (d, e) Fe-doped Mn–Co spinel (MCFeO)

Ceramic Coatings for Metallic Interconnects and Metal Alloys Support …

135

of SOC stacks. This effort will open to the possibility of using cheaper alloys as interconnect, but also to the need of developing more efficient coatings to prolong the stack working life. To this purpose, optimizing the coating deposition method through electrophoretic co-deposition could allow to modify in-house coating composition and properties, acting on EPD suspension precursors and sintering treatment, depending on the requirement of the stack. Discussed results on Cu and Fe spinel doping set the groundwork for future research on the implementation of spinel composition. Considering the better sinterability and higher conductivity of the Cu-doped Mn–Co spinel, as well as the improvement of oxidation resistance related to Fe doping, it is proposed that a simultaneous Cu-Fe doping of the base Mn–Co spinel is viable approach to improve the performance of the coating. A search of the literature revealed a lack of a comprehensive evaluation in this direction, despite the already suggested possibility of tuning spinel functional properties by Fe and Cu doping into the Mn–Co spinel (Masi et al. 2016). In this framework, the electrophoretic co-deposition plays a key role. Apart from co-deposition, the implementation of electrophoresis technique can offer innovative solutions when adjusted with other slurry-based deposition techniques such as dip coating or electrolytic deposition (ELD), in order to produce multilayer or composite coatings as suggested by few recent publications. For example, a hybrid ELD-EPD approach has been proposed to produce CeO2 composite spinel coatings with the aim of coupling the benefits of rare earth addition with those of Mn–Co spinel coating. Firstly, Zhu et al. (Zhu et al. 2015) prepared coatings from electrolyte solution of cobalt salts, to which Mn3 O4 and CeO2 powders were added; a CeO2 –(Co,Mn)3 O4 composite coating was obtained after thermal conversion treatment at 600–800 °C for a total of 10 h. More recently, Mosavi and Ebrahimifar (2020) worked on electrolyte solution containing manganese sulfate and cobalt sulfate with further addition of CeO2 powders deposited on AISI 430 substrate. After oxidation of 200 h, MnCo2 O4 and CeO2 phases were detected; however, the formation of MnFe2 O4 spinel due to relevant diffusion of iron from the steel substrate defines the need of further investigations on the long-term stability of the coating. Furthermore, Brylewski et al. (2021) reported on a multilayer system consisting of a coating of gadolinium oxide obtained by dip coating followed by EPD of MnCo2 O4 . Dip coating allows to deposit a thin layer of the rare earth oxide on the steel surface, where it is more effective in reducing the oxide scale growth by segregation at the steel grain boundaries; coupled with the spinel coating, this innovative approach provided improved protection at 800 °C. A natural progression of this work is to analyze the possibility to upscale to industrial relevant size of the coating deposition technique. Indeed, all the studies and deposition methods mentioned in the previous paragraphs are related to the deposition and test of protective coatings on small lab-scale samples of few square centimeters. However, real-size interconnects can present larger dimensions (some hundreds of square centimeters) as well as corrugated and channeled surfaces. To this purpose, the investigation on feasibility and sustainability for the industrial application is a fundamental requirement for the validation of the proposed deposition technique. For example, Blum et al. (2020) have recently reported a study on long-term stack

136

E. Zanchi et al.

test including metallic interconnects coated by atmospheric plasma spray and wet powder spray (Bianco et al. 2019). Similarly, atmospheric plasma spray coating is mentioned in the study of Menzler et al. (2018) on post-test characterization of 30 kh stack operation; on the other hand, Bianco et al. (2019) studied the degradation of a stack with interconnect coated by wet powder spray. In all the few mentioned publications, authors have never focused on the evaluation of the coating deposition technique on the stack performance. To this purpose, electrophoretic deposition has been validated as a valuable deposition technique to scale up to industrial relevant conditions (Sabato et al. 2021). The study proved the feasibility to adapt the EPD setup to coat 20 × 20 cm2 steel interconnect, including complex shapes and channeled surfaces; furthermore, the stack test performed for 3000 h at 850 °C confirmed excellent performances of the coating, setting a significant step forward in the use of EPD as versatile, low cost, easy, and reproducible method in the solid oxide technology.

3 Porous Metal Alloys Support in Solid Oxide Electrolysis Cells In the last decade, advanced alloys have been discovered to be an interesting material in a variety of applications such as engines, turbines, and supercritical reactors (Park et al. 2018; Talic et al. 2018b; Bianco et al. 2020). However, the most noticeable growth of interest for these materials was in the field of metal-supported solid oxide electrolysis cells (MS-SOECs) (Tucker 2020). For this particular application, a lot of specific material requirements have to be fulfilled. First of all, the new material should be relatively cheap and easy to manufacture into complex shapes because the typically used SOEC interconnects are ceramic components, which are expensive and hard to produce into specific shapes. Furthermore, the material should be able to withstand very large thermal shocks (> 100 °C/cm (Tucker 2020)) and extremely fast thermal cycling as SOEC support. Moreover, the main role of the SOEC interconnect is to collect a charge and transport it to the electrodes. Therefore, the support material has to have high electronic conductivity. Taking all of the conditions into account, the best material appears to be chromiaforming ferritic stainless steel. It has high enough thermal conductivity and resistance to rapid thermal shock and is cheaper than the ceramic components that are used in commercial SOEC systems. The biggest problem with applying alloys as support in MS-SOEC is the corrosion process at operating conditions. This process causes oxidation of the alloy, which significantly decreases the electrical conductivity of the support and interconnect. In spite of the fact that Al2 O3 has very high corrosion resistance, it is also an insulator, so alumina forming alloys cannot be used as interconnects in SOECs. However, among metal oxides, chromia (Cr2 O3 ) has one of the highest electrical conductivities (~ 1–10 mS cm−1 at 600–800 °C (Holt and Kofstad 1994)); thus, it seems to be the best candidate for interconnect in SOEC systems.

Ceramic Coatings for Metallic Interconnects and Metal Alloys Support …

137

In order to even decrease more the cost and weight of the whole device, the porous form of chromia-forming stainless steel is being considered as an alternative for heavy ceramic components. The basic difference between dense and porous alloys is their specific surface area (SSA), and for the alloy of ~ 30 % porosity produced using − 53 fractioned powder (screened with a 53 μm opening sieve), this parameter is even 28 times bigger compared to the dense form (Koszelow et al. 2021). In Fig. 14, the comparison of the morphology of dense and porous alloys was illustrated. A significantly higher specific surface area of porous alloy implies a larger possible area of the oxygen-alloy interface, which has a strong influence on the corrosion behavior of the porous components, as detailed in the following section. The most common method of producing the porous alloy sheet is the powder metallurgy. In this technique, the powder is mixed with binders and lubricants and then compressed into the specified shape form. The as-prepared sample is sintered in a protective atmosphere to avoid the oxidation process. Powder metallurgy is widely applied for producing different alloys and composites of metals and ceramics. Furthermore, this method enables the production of a high purity product in a complex shape while controlling its density and porosity (i.e., by adjusting the sintering temperature) (Rauscher et al. 2008). The alternative method to obtain the porous alloy sheet is tape-casting. In this technique, the powder of the desired alloy has to be dispersed in an aqueous or

Fig. 14 Comparison of Fe22Cr in a dense and b porous form

138

E. Zanchi et al.

organic medium. Next, binders and plasticizers are added to the mixture to give the tape sufficient strength and flexibility. To prevent big holes and discontinuities within the tape, a defoamer agent is applied or the slurry is degassed. The most important parameter about the slurry is that the powder particles do not drop down by themselves. The as-prepared slurry is cast on the substrate, depending on the used compounds. The most common substrates are metallic foils or special polymer tapes. Then, the tape-casting is performed utilizing the doctor-blade method, which allows for controlling the thickness of the obtained tape. The next step is drying the tape to remove the solvent, and the plastic green tape with connected metallic particles was obtained. In order to remove the binder, the green tape is burned by a thermal treatment at a high enough temperature (usually ~ 200–500 °C) to decompose the binder. The final step is the sintering at reducing conditions to create neck connections between alloy particles and reverse the surface oxidation of the sample, which occurs during the debinding step. The process flow of tape-casting is shown in Fig. 15. The final porosity of the as-obtained sheet depends on the used particle fraction (for larger particle size, the porosity of the final product is larger) and the temperature of reducing sintering. In the case of higher temperatures, the porosity is lower because of grain growth, so the probable pore area is occupied by larger metallic grains (Rauscher et al. 2008). Moreover, if the applied temperature is high enough, the full densification occurs, so the tape-casting process can also be used to produce dense metallic sheets. Another technique to obtain porous alloys or metals is the dealloying process. In order to use this method, an alloy that contains the desired elements and one additional element is necessary. For instance, Han et al. (2020) obtained pure In and Sn porous metals by the dealloying method. They used pure metal powders to create a pellet by uniaxial pressing. The pellets were then sandwiched with Li foil (molar ratio of Li: In or Li: Sn was 5:1) and sintered at 220–230 °C in a protective Ar atmosphere to form the Li–In or Li–Sn alloy. After alloying, the samples were immersed in dry methanol to extract Li, and a porous foil of In (or Sn) was formed. The schematic diagram of the dealloying process with their corresponding SEM images and XRD patterns of as-obtained products is presented in Fig. 16. Dealloying is a more precise process compared with powder metallurgy or tape-casting because it allows for the obtaining of even nano-porous alloys, which are impossible in the case of other techniques. Mokhtari et al. (2020) obtained the nano-porous Fe–Cr alloy by the dealloying process, so they employed a Fe–Cr–Ni alloy as the precursor. The different compositions of the precursor were immersed for 1 h in a molten Mg bath to obtain the Fe–Cr–Mg structure. Then, highly concentrated nitric acid was applied in order to dissolve the Mg-based solid-state solution, and a Fe–Cr microporous alloy occurred. Except for the corrosion process, Cr evaporation from the high chromium alloy limits its application in SOEC systems. These volatile chromium species migrate and deposit on the air electrode, which in turn leads to rapid degradation of the electrochemical performance of the cell. Moreover, chromium evaporation accelerates the corrosion due to decreasing the Cr reservoir, which is a critical factor for the lifespan of the support in MS-SOEC. To prevent this process, protective coatings on the steel

Ceramic Coatings for Metallic Interconnects and Metal Alloys Support …

139

Fig. 15 Process flow of tape-casting for the metallic sheet production

interconnects of the SOC stack are generally applied. Such coatings slow down the corrosion process and reduce Cr evaporation.

4 Corrosion Properties 4.1 Introduction to High-Temperature Oxidation The basic idea behind using metal support in SOCs is to support ceramic element layers with a well-known, low-cost, and robust material, which is expected to lower the cost of SOC devices. Furthermore, alloy support provides thermal strength as well as an electrical connection. However, the oxidation of alloy elements occurs due to the harsh working conditions of HT-SOC devices, which typically operate in the 500–900 °C range. This phenomenon has been observed in ferritic stainless steel alloyed with a high Cr content, which quickly transforms to form a chromia layer. Most metals will inevitably oxidize under a variety of conditions; therefore, the practical challenges of material lifetimes rely on corrosion protection. The strategies revolve around decreasing the oxidation reaction rates, as well as controlling

140

E. Zanchi et al.

Fig. 16 Synthesis and characterization of porous metals by chemical dealloying. (a) schematic showing the synthesis of porous In or Sn foil via the dealloying of Li–In or Li–Sn alloy (a Li:In or Li:Sn molar ratio of 5:1 was used). (b) SEM image of a dealloyed In sample. (c) SEM image of a dealloyed Sn sample. (d, e) XRD patterns of (d) In and (e) Sn samples at different steps of the synthetic procedure (Han et al. 2020)

alloy morphology. Controlling the oxidation phenomena requires a fundamental understanding of different aspects of solid–gas reactions. Oxidation rate depends on steel chemistry, temperature, and gas atmosphere (Young 2008). Therefore, chemical composition and geometrical factor of metal support should be chosen wisely. The surface layer formed by oxidation process dictates the nature of the metal’s corrosion rate and has a significant influence on the material. The oxide is protective if it is continuous and efficient in isolating the alloy from the environment. However, if the oxide does not work as a separator, corrosion problems occur. The dense chromium oxide scale decreases the oxidation rate, while accelerated corrosion may lead to support degradation, such as delamination or cracking. Rapid temperature changes cause acceleration of damaging process. In addition, an undesirable chemical reaction between metallic substrate with other cells’ elements could be expected. The mechanical state of any solid determines its performance. The phenomenon of oxidation occurs as a result of the metal reacting with oxygen. In SOC devices, oxygen is acquired from air or steam. Steam, in general, oxidizes quicker than air. Steam-formed oxide layers are more iron-rich and porous and hence provide less protection. Similarly, air with a high water content accelerates oxidation. The use of certain alloying compounds, i.e., Cr, Si, and Al improves the resistance

Ceramic Coatings for Metallic Interconnects and Metal Alloys Support …

141

to oxidation. Addition of rare earth metals and reactive elements such as Ti, Zr, or Y increases oxidation resistance significantly. Several processes were described, i.e., mass transport through the oxide scale or oxide species evaporation. The considerable role in oxidation process is mechanical stress on the scales and interaction between alloys elements and microstructure. The oxidation rates are typically controlled by diffusion and interfacial mechanisms, where crystallographic and geometrical effects are involved in those processes (Samal 2016). A scheme of oxidation process, resulting in a growing oxide scale, is shown in Fig. 17 and can be described by following steps: Mass transfer in the gas phase causes delivery of oxidant to the gas-scale phase, while the oxygen is incorporated into oxide scale. Meanwhile, oxidized metal from the alloy is transferred into alloy and scale interface, which causes incorporation of metal into scale. As a consequence, metal and/or oxygen transport through the scale is possible. In general, with two metals in the alloy, one of the formed oxides will be more stable than the other oxide. In the most common Fe–Cr alloys, chromium has a greater affinity for oxygen than iron; thus, during initial oxidation stage, a thin layer of protective chromia scale is formed. However, exceeding oxidation conditions, such as temperature or time, lead to breakaway corrosion. Damaging effect of breakaway corrosion is caused by depletion of chromium in alloy structure as well as formation of iron oxides.

Fig. 17 Reactions and transport process involved in growing oxide scale (Young 2008)

142

E. Zanchi et al.

4.2 Oxidation Kinetics Laws The step of reaction is determined by metal consumption, oxygen usage, and postoxidation product formation. Several laws of the kinetics of oxidation are known, i.e., linear wherein surface or interferential process toward the reactive gases is rate controlling, and the metal oxidation proceeds at constant rate or logarithmic wherein oxidation process is subjected only for thin oxide scales (2–4 nm). In-depth considerations are explained in the second chapter of high-temperature oxidation and corrosion of metals book by Young (2016). In this chapter, parabolic kinetic law is discussed, since it is found that alloys used in MS-SOCs applications obey to. The rate controlling mechanism in parabolic law is the diffusion of metal cations through the oxide scale. This approach started from Wagners’ theory, in which the materials corrosion behavior is quantified by parabolic rate constant k p . 

\W A

2 = k p t,

where the \W is weight gain, the A is surface area, and the t is time. k p is usually given in g2 m4 s unit. The value of k p may achieve different values in different temperatures and environments for any material. The target parabolic rate constant for oxidation in green energy devices is 5 × 10–13 g2 m4 s.

4.3 Predictive Methods—Experimental Techniques 4.3.1

Oxidation test

For multi-component systems, the requisite thermodynamic, kinetic, and mechanical data are not always available, necessitating further experimental examination. HTSOCs lifetime should be thousands of hours; therefore, the experimental verification of the predicted results is obligatory before implementation in a real cell. The oxidation can be investigated by measuring metal or oxygen consumption or by observing oxides growth as a function of time. To predict the lifetime of the alloy, most widely used method is the calculation of oxidation kinetic rates by weight gain measurement, which can be performed by continuous or discontinuous method. • Continuous Weight Gain Measurement In continuous, thermogravimetry measurement of the alloy sample is attached on the sensitive microbalance and heated to selected temperature in stable or flowing gas environment. Then, in iso-thermal conditions, the weight gain data of alloy sample is collected.

Ceramic Coatings for Metallic Interconnects and Metal Alloys Support …

143

Fig. 18 Weight gain data presented as a linear weight change, b log–log plot, c square weight change with respect to the initial surface area, and d Arrhenius plot (Koszelow et al. 2021)

• Non-continuous Weight Gain Measurement In discontinuous method, the samples are placed in the furnace and weighted in interval time of exposition on high temperature. This method requires a series of samples, subjected to different aging times. The advantage of discontinuous method is the possibility of taken out samples in interval steps. Those samples can be used for post-mortem analysis of range alloys’ oxidation stages. From acquired data, the information about the oxidation rate constants and activation energies can be calculated. An exemplary results of the oxidation kinetic tests were proposed by Koszelow et al. and are presented in Fig. 18 (Koszelow et al. 2021). Recently, Taylor and Tossey have collected and tabulated parabolic rate constants for dozens of alloys by machine learning method, and they found that Ni, Cr, and Al are the most protective elements. Addition of Mo and Co likewise increases alloys reliability (Taylor and Tossey 2021).

4.3.2

Samples examination—post-mortem analysis

Actual changes in the alloy compositional, microstructural, and phase constitution information can be observed in post-mortem analyzes.

144

E. Zanchi et al.

Fig. 19 SEM–EDS post-mortem analysis of glass–ceramic and alloy powder interface before (a) and after (b) aging (Drewniak et al. 2021)

• 2D Microstructural Analysis Both surface and polished cross-section of the sample morphology could be examined via SEM/EDS analysis. Drewniak et al. (2021) proposed accelerated chemical reactivity test by simulating glass–ceramics sealant and metal interconnect interface by using alloy powder. In Fig. 19, EDS analysis of their samples before (Fig. 6a) and after (Fig. 6b) aging at 850 °C in static air atmosphere for 500 h is shown. The analysis showed that after aging, a thin layer of chromium oxide had formed on the glass/alloy powder interface and changes in crystallization of the glass matrix. • 3D Microstructural Analysis In addition to SEM/TEM microscopy, tomographic studies of alloy samples could be performed. Although there are no tomographic studies in the literature of metallic interconnects for SOC applications, these studies will be very relevant especially for porous metal supports. Thermographic and tomography analysis will allow the investigation of 3D microstructures of the alloys quantitatively, which strongly supplements the weight gain and microscopic analyzes. • Phase Identification by XRD Another complementary method of post-mortem analysis is XRD. Information about the phases (or phase changes) in the investigated alloy can be obtained by matching the resulted diffractogram with tabulated standard. This method allows the identification of post-reaction products. Moreover, XRD measurements can be performed by in situ mode at the selected temperature. • Chemical State Measurements by XPS XPS analysis can also be used to determine the chemical state of an alloy and its oxides. However, it should be noticed that this method, due to its low penetration, allows only for the surface analysis. Nevertheless, using the XPS technique, it is possible to determine the degree of oxidation of the material, so in some cases, this approach could be valuable.

Ceramic Coatings for Metallic Interconnects and Metal Alloys Support …

145

5 Conclusions This chapter contributes to the existing knowledge on metallic interconnects and relative corrosion issues, by providing a comprehensive overview of their different roles and systems, processing and deposition methods to assess their functionality in SOEC devices. Even if it is not possible to unquestionably outline the optimal solution, these findings provide the following insights for coating development: • The modification of the composition of the base Mn–Co spinel is a promising route to improve the functional properties of the base spinel structure • A high degree of densification is fundamental to guarantee reasonable protection properties of the coating, in order to limit oxygen inward diffusion to the steel interconnect • A possible approach to a composite spinel coating (with a proper selection of the powders, formulation of the suspension and post-deposition treatments choice), with the aim of coupling the benefits of rare earth addition with those of Mn–Co spinel, might be explored. The preliminary findings on the role of the porous metal alloys support show that the MS-SOEC design has promises, but further work and recommendations are required to properly evaluate the MS feasibility. Continued development of MSSOEC will be beneficial if performance, durability, or cost can be enhanced beyond that of other SOEC types. The best candidate for SOC applications seems to be chromia-forming ferritic alloys. At SOC operating conditions, these alloys passivate, providing sufficient electrical conductivity and corrosion resistance. The porous forms of the alloys are also promising; however, the corrosion process of them has to be researched in detail. It is found that Ni, Cr, and Al are the most corrosion protective elements. To prevent chromium evaporation and poisoning the electrodes, the functional coatings, for instance Mn-Co spinel, are applied to the steel components. Theoretical bases for predictive methodologies include compositional, microstructure, and phase changes. Wagner’s kinetic model is the most common approach to predict high-temperature oxidation resistance. However, parabolic rate constants may hinder a multiplicity of processes in addiction to mass gain. For a better understanding of high-temperature corrosion effects in metal supports, both kinetic and further structural investigations are necessary.

References Aznam I„ Mah JCW, Muchtar A, Somalu MR, Ghazali MJ (2021) Electrophoretic deposition of (Cu,Mn,Co)3O4 spinel coating on SUS430 ferritic stainless steel: Process and performance evaluation for solid oxide fuel cell interconnect applications. J Eur Ceram Soc 41:1360–1373. Elsevier Ltd Belogolovsky I, Zhou X-D, Kurokawa H, Hou PY, Visco S, Anderson HU (2007) Effects of surfacedeposited nanocrystalline chromite thin films on the performance of a ferritic interconnect alloy. J Electrochem Soc 154:B976

146

E. Zanchi et al.

Bianco M, Jan Pieter Ouweltjes, and Jan Van herle. 2019. ‘Degradation analysis of commercial interconnect materials for solid oxide fuel cells in stacks operated up to 18000 hours. Int J Hydrogen Energy 44:31406–31422. Elsevier Ltd Bianco M, Poitel S, Hong JE, Yang S, Wang ZJ, Willinger M, Steinberger-Wilckens R, Van herle J (2020) Corrosion behaviour of nitrided ferritic stainless steels for use in solid oxide fuel cell devices. Corros Sci 165:108414 Blum L, Fang Q, Groß-Barsnick SM, de Haart LB, Malzbender J, Menzler NH, Quadakkers WJ (2020) Long-term operation of solid oxide fuel cells and preliminary findings on accelerated testing. Int J Hydrogen Energy:1–10 Bobruk M, Molin S, Chen M, Brylewski T, Hendriksen PV (2018) Sintering of MnCo2O4 coatings prepared by electrophoretic deposition. Mater Lett 213:394–398. Elsevier B.V. Brylewski T, Kucza W, Adamczyk A, Kruk A, Stygar M, Bobruk M, D˛abrowa J (2014) Microstructure and electrical properties of Mn1+xCo2−xO4 (0≤x≤1.5) spinels synthesized using EDTA-gel processes. Ceram Int 40:13873–13882 Brylewski T, Molin S, Marczy´nski M, Mazur K, Domaradzki OK, Gil A (2021) Influence of Gd deposition on the oxidation behavior and electrical properties of a layered system consisting of Crofer 22 APU and MnCo2O4 spinel. Int J Hydrogen Energy 46:6775–6791 Chen G, Xin X, Luo T, Liu L, Zhou Y, Yuan C, Lin C, Zhan Z, Wang S (2015) Mn1.4Co1.4Cu0.2O4 spinel protective coating on ferritic stainless steels for solid oxide fuel cell interconnect applications. J Power Sources 278:230–234. Elsevier B.V. Chesson DA, Zhu JH (2020) Effect of off-stoichiometry on electrical conductivity in Ni–Fe and Mn–Co spinel systems. J Electrochem Soc 167:124515. IOP Publishing Chevalier S, Larpin JP (2002) Formation of perovskite type phases during the high temperature oxidation of stainless steels coated with reactive element oxides. Acta Mater 50:3107–3116 Chu CL, Lee J, Lee TH, Cheng YN (2009) Oxidation behavior of metallic interconnect coated with La-Sr-Mn film by screen painting and plasma sputtering. Int J Hydrogen Energy 34:422–434. International Association for Hydrogen Energy Drewniak A, Koszelow D, Błaszczak P, Górnicka K, Jurak K, Javed H, Sabato AG, Jasi´nski P, Molin S, Smeacetto F (2021) Glass-ceramic sealants and steel interconnects: accelerated interfacial stability and reactivity tests at high temperature. Mater Des 212 Falk-Windisch H, Claquesin J, Sattari M, Svensson J-E, Froitzheim J (2017) Co- and Ce/Co-coated ferritic stainless steel as interconnect material for Intermediate Temperature Solid Oxide Fuel Cells. J Power Sources 343:1–10 Fontana S, Amendola R, Chevalier S, Piccardo P, Caboche G, Viviani M, Molins R, Sennour M (2007) Metallic interconnects for SOFC: characterisation of corrosion resistance and conductivity evaluation at operating temperature of differently coated alloys. J Power Sources 171:652–662 Fontana S, Chevalier S, Caboche G (2012) Metallic interconnects for solid oxide fuel cell: performance of reactive element oxide coating during 10, 20 and 30 months exposure. Oxid Met 78:307–328 Gambino LV, Magdefrau NJ, Aindow M (2015) Microstructural effects of the reduction step in reactive consolidation of manganese cobaltite coatings on Crofer 22 APU. Mater High Temp 32:142–147 Goebel C, Fefekos AG, Svensson JE, Froitzheim J (2018) Does the conductivity of interconnect coatings matter for solid oxide fuel cell applications?J Power Sources 383:110–114. Elsevier Goebel C, Berger R, Bernuy-Lopez C, Westlinder J, Svensson JE, Froitzheim J (2020) Long-term (4 year) degradation behavior of coated stainless steel 441 used for solid oxide fuel cell interconnect applications. J Power Sources 449 Goebel C, Asokan V, Khieu S, Svensson JE, Froitzheim J (2021) Self-healing properties of Ce/Cocoated stainless steel under simulated intermediate temperature solid oxide fuel cell conditions. Surf Coat Technol 428:127894. Elsevier B.V.

Ceramic Coatings for Metallic Interconnects and Metal Alloys Support …

147

Guo P, Lai Y, Shao Y, Zhang Y, Sun H, Wang Y (2018) Oxidation characteristics and electrical properties of doped Mn–Co spinel reaction layer for solid oxide fuel cell metal interconnects. Coatings 8 Guo PY, Sun H, Shao Y, Ding JT, Li JC, Huang MR, Mao SY, et al (2020) The evolution of microstructure and electrical performance in doped Mn–Co and Cu–Mn oxide layers with the extended oxidation time. Corros Sci 172 Hamaker HC (1940) Formation of a deposit by electrophoresis. Trans Faraday Soc 35:279 Han SY, Lewis JA, Shetty PP, Tippens J, Yeh D, Marchese TS, McDowell MT (2020) Porous metals from chemical dealloying for solid-state battery anodes. Chem Mater 32:2461–2469 Hassan MA, Mamat OB, Mehdi M (2020) Review: influence of alloy addition and spinel coatings on Cr-based metallic interconnects of solid oxide fuel cells. Int J Hydrogen Energy 45:25191– 25209. Elsevier Ltd Holt A, Kofstad P (1994) Electrical conductivity and defect structure of Cr2O3. II. reduced temperatures ( 12·10–6 K−1 . For mixed components such as glass-ceramics, where the Fig. 7.6 Thermal expansion coefficient as function of temperature for metallic interconnects and solid electrolytes together with TCE target regions for corresponding glass-ceramic sealants

162

J. Schilm et al.

Fig. 7.7 Liquidus temperatures of binary R2 O-SiO2 - and RO-SiO2 -melts; Blue: alkaline oxides; black: earth alkaline oxides; orange: FeO & ZnO as conditional network formers; red: PbO as a network former (data re-drawn from binary phase diagrams)

respective phases have a known coefficient of thermal expansion, it is possible to calculate the thermal expansion of the composite from the addition of all individual phases according to eq. (7.3): aglass ceramic =

 (ai Ki Wi /ρi ) (Ki Wi /ρi )

(7.3)

with a = CTE, K = Modulus factor, W = mass fraction and r = density and the index “i” indicating the individual phases (Donald 1993). A summary of the most widespread crystalline phases in residual glass matrix used in SOC sealants according to Fergus (2005) and Mahapatra (2010b) and Theerapapvisetpong et al. (2011) is given in Table 7.1, where the phases containing barium oxide show tendentially the highest CTE values. However, most of the listed earth alkaline silicates show rather high CTE values. For this reason, glass solders with high-expansion coefficients are developed to a major extent based on glass systems that contain BaO to be able to crystallize their corresponding phases. In general, it is possible to melt binary (RO-SiO2 ) or ternary (R1O-R2O-SiO2 ) earth alkaline glasses, but these are not suitable for SOC application as they crystallize quite fast especially when oxide mixtures with a stoichiometry near to crystalline phase compositions are chosen. Furthermore, instead of simple binary glasses more complex compositions are needed to adjust wetting, viscosity as well as crystallization kinetics and crystalline phase content. In this sense, Al2 O3 and B2 O3 are important oxides that can stabilize the amorphous glass phase and help to control the formerly mentioned parameters. By adding these oxides, ternary or quaternary glasses can be created, which, however,

Glass Ceramic Sealants for Solid Oxide Cells

163

Table 7.1 Overview on earth alkaline silicate phases and their thermal expansion coefficients (Fergus 2005; Mahapatra and Lu 2010b, Theerapapvisetpong et al. 2011) Earth alkaline oxide

Crystalline phase

CTE 10–6 K−1

MgSiO3 (Clinoenstatite)

7.8…13.5

MgSiO3 (Protoenstatite)

9.8

Mg2 SiO4 (Forsterite)

9.4

CaSiO3 (Wollastonite)

9.4

Ca2 SiO4 (Orthosilicate)

10.8…14.4

MgO + CaO

Ca2 MgSiO7 (Akermanite)

10…11.3

BaO

BaSiO3 (Barium silicate)

9…13

BaSi2 O5 (Barium orthosilicate)

14…14.4

Ba2 Si3 O8

12.6

Ba5 Si8 O21

12.5

Ba3 CaSi2 O8

12.2…13.8

MgO

CaO

CaO + BaO

do not generally offer usability as sealants (see Sect. 7.3.1). While B2 O3 is a real network-forming oxide, Al2 O3 is an intermediate oxide and can participate in crystallization processes of silicates, if it is present in larger concentrations in the glass composition. Possible Al2 O3 -containing crystalline phases in such glasses are not so favorable for application in SOC as their CTE values can be rather low (Table 7.2). Alkaline earth borates as phases are only hardly observed and with BaB2 O4 only one phase with a strong crystallographic anisotropy but large CTE value is known in this context (Mahapatra and Lu 2010b). This combination of properties can lead to considerable thermomechanical stresses in glass ceramic microstructures at temperatures below Tg of the residual glassy phase. Based on this information, it is now obvious to look at the binary BaO-SiO2 and, as an extension, the ternary BaO-Al2 O3 -SiO2 phase diagram. Figure 7.8 shows the section of the phase diagram in which the crystalline phases suitable for our application are existing (Shabanova et al. 2003). A number of binary barium silicates exist along the BaO-SiO2 junction, and some of those properties have been described above. For the selection of glass-forming compositions, it is necessary Table 7.2 Overview on an alkaline earth-aluminosilicates and their expansion coefficients (Fergus 2005) Alkaline earth oxide

Al2 O3 -containing crystalline phase

TEC 10–6 K−1

MgO

Mg2 Al4 Si5 O18 (Cordierite)

2

BaO

BaAl2 Si2 O8 (Hexacelsian)

7…8

BaO

BaAl2 Si2 O8 (Monocelsian)

2…3

BaO

BaAl2 Si2 O8 (Orthocelsian)

5…7

164

J. Schilm et al.

that sufficiently high proportions of the network former SiO2 are present. According to the literature, the BaO:SiO2 molar ratios used for more complex glass–ceramic sealants range from 1:1 to 1:3, within the binary compounds with high CTE values are formed (Lara et al. 2004b, 2004a; Da Silva et al. 2016; Schilm et al. 2018). At lower BaO:SiO2 ratios, the fraction of network-forming oxides is too low to form a coherent glass structure, and at higher ratios, crystalline SiO2 phases are formed, which are considered undesirable due to phase transformations with large, discontinuous volume changes causing tensile stresses (Schilm et al. 2018). The fundamental ability of this binary system to form glasses over a quite large compositional range has already been confirmed much earlier (Shelby 1979). Additionally, simple binary glasses, where the BaO-SiO2 -system represents an example, are often prone to phase separations in the molten status, which means that two separated melts with different properties are existing at the same time (Gueguen et al. 2017). So, a technical applicability of such simple glasses is not given for any reason. As described before, additions of, e.g., Al2 O3 , stabilize the glass formation, prevent phase separations and give better control over crystallization processes. However, there is also the possibility of crystallization of additional Al2 O3 -containing phases. According to Fig. 7.8, in the composition range of BaO:SiO2 that we consider relevant, only the Celsian phases can be considered as a further co-existing phase, which are deemed to be disadvantageous due to their low CTE values. Complementary information on the domains of co-existing crystal phases is provided by studies on glass-forming regions in such ternary systems. For comparison, Fig. 7.8b marks different compositions that according to the work of Cleek et al. can be synthesized as glasses (x = glass; square = partially crystalline) (Cleek and Babcock 1973). In contrast to studies on the formation of crystalline phases, which consider the system in the state of a thermodynamic equilibrium, such studies are carried out away from any equilibrium considerations to avoid the formation of crystalline phases. Usually, such investigations are performed in a way that small amounts of melts in systematically varying compositions are cooled down rapidly, and the phase composition of the resulting solid is investigated. Since the conditions for these investigations are not defined, the results in the literature can vary considerably. Despite this, such results help to identify suitable compositional ranges for the development of glass-ceramic sealants. It is observed from the comparison of both phase diagrams in Fig. 7.8 that the glass-forming region in the ternary BaO-Al2 O3 SiO2 system is close to the binary BaO-SiO2 line and covers SiO2 contents between 30 and 45 wt.%. Furthermore, the permissible Al2 O3 content in the glass-forming range is limited at about 20 wt.%. Higher SiO2 contents lead to very high viscosity and the crystallization of SiO2 phases such as quartz and cristobalite (Schilm et al. 2018). It is also noticeable that glass formation is also not possible or is more difficult with very Al2 O3 -poor compositions. We have already explained above that those binary glasses tend to devitrify/crystallize faster than more complex compositions. In addition to the crystal phases already mentioned, the crystalline SiO2 phases have a special role. Both cristobalite and quartz exist in high-temperature and lowtemperature modifications whose phase transformation temperatures are below 300 and 600 °C respectively and are associated with considerable volume changes. This

Glass Ceramic Sealants for Solid Oxide Cells

165

Fig. 7.8 Schematic ternary phase diagram of the BaO-Al2 O3 -SiO2 -system with crystalline phases (Thomas 1950) together with the compositional range of the glass-forming region both indicated in wt.% (data taken from Cleek et al. (1973))

behavior leads to considerable mechanical stresses in SOC stacks, and so their formation in glass-ceramic sealants has to be avoided. Especially glass compositions with SiO2 contents higher than 45 wt.% and molar ratios of SiO2 :BaO > 2 are sensitive to the formation of these phases. Up to this stage, it has been demonstrated how to approach a material system that can fundamentally serve as a basis for the development of glass–ceramic sealants for high-temperature applications. Similarly, other ternary phase systems have been investigated for their suitability in developing glass–ceramic sealants for SOC. Wanko et al. for example, considered the system BaO-MgO-SiO2 with regard to co-existing binary barium silicates and ternary barium-magnesium silicates (Wanko et al. 2007). While the CTE values for the barium silicates correspond to the high values of other authors, the BaMg2 Si2 O7 -phase showed TEC values around 11·10–6 K−1 . Silica-rich compositions lead to the simultaneous formation of a MgBaSi3 O8 -phase and of crystalline silica phases together with the unfavorable phase transformations. This system also shows useful prerequisites for the development of glass–ceramic sealants. In a similar manner, but with a direct focus on the development of glass-ceramic sealants, Lara et al. investigated the systems MgO-BaO-SiO2 and ZnO-BaO-SiO2

166

J. Schilm et al.

with regard to their glass-forming properties (Lara et al. 2004a). Figure 7.9 illustrates the glass-forming regions by means of the ternary phase diagrams. Along the BaOSiO2 line, the glass-forming region is limited by the formation of crystalline silica phases in the case of high SiO2 contents. High amounts of BaO do not lead to the formation of a stable glass network; a multitude of possible crystalline compounds is possible (cf. Fig. 7.8a). In fact, both systems show quite similar glass-forming regions, with the areas marked in gray indicating compositions for potential glass– ceramic sealants. Therefore, one must be conscious of the fact that such ternary compositional ranges, even if they can in principle be synthesized as glasses, are not suitable for technological applicability as glass ceramic sealants. A suitable composition must first soften amorphously under joining conditions, form a cohesive bond with the metallic and ceramic sealing partners and subsequently form the property-creating crystalline phases. For this purpose, Fig. 7.10 compares the shrinkage and softening behavior measured by hot stage microscopy in combination with the corresponding DTA curves of two crystallizing glasses. Also, the images of the measured powder compacts taken at the maximum sintering and under potential joining conditions are plotted. While the glass in Fig. 7.10a crystallizes strongly at temperatures only slightly higher than the maximum sintering and shows no softening even at 950 °C (cf. Fig. 7.3), the glass in Fig. 7.10b crystallizes at significantly higher temperatures above 900 °C and softens in a temperature range between 870 °C and 950 °C suitable for joining processes. However, it is possible to expand these glass-forming regions in the respective ternary and also more complex systems by adding further oxides. The effect of further oxides, which ideally are not involved in the formation of crystalline phases, can be visualized in a first approximation as that of disturbing elements. They prevent the elements required for the formation of crystalline phases in the glass melts from arranging themselves in exactly this way when the crystallization temperatures are

Fig. 7.9 Glass-forming area, between solid lines, in the MgO-BaO-SiO2 a and the ZnO-BaO-SiO2 b composition diagrams (. glassy, ◻ partially crystalline, o crystalline). Eutectic points in the binary systems are indicated with a large circle (Lara et al. 2004a)

Glass Ceramic Sealants for Solid Oxide Cells

167

Fig. 7.10 Superposition of the shrinkage and DTA curves of a fast a and a slow b crystallizing glass; the shrinkage curves were measured by means of hot stage microscopy so that, in addition, the images of characteristic shape changes with the corresponding temperatures are shown

present. In this way, crystallization processes can be influenced both kinetically in their rates and thermodynamically in their proportions. In the case of glass–ceramic sealants, this effect is used to adapt the softening and crystallization behavior to technological joining processes.

3.1 Single Component, Crystallizing Sealing Glasses With the expansion of ternary basic compositions, which essentially consist of the components for setting the relevant properties such as sufficiently high softening temperatures and coefficients of thermal expansion, in the direction of quaternary and even more complex glass compositions, one arrives at the glass compositions, which have a technological relevance. The main task of the additional oxides in the glass

168

J. Schilm et al.

ceramic sealants is to adjust the relevant properties such as crystallization behavior, expansion coefficient, and softening temperatures to technological processes that enable the production and operation of SOC stacks. The options for developing such glasses using different oxides are very diverse and are reflected in many publications, which cannot be presented in an overview within the context of this section. If one is interested in a corresponding overview, please refer to the available review articles (Fergus 2005; Lessing 2007; Mahapatra and Lu 2010b; Donald et al. 2011; Singh and Walia 2021). So, the question arises which kind of oxides make sense and have which function in the parental glass and the final glass-ceramic sealant. It is generally accepted that additional components in glasses stabilize the glassy state and can retard or even completely suppress crystallization processes. This fact certainly helps in the development of applicable glass-ceramic sealants in that the crystallization behavior of the glasses can be adapted to the technological joining processes of fuel cells. The introduction of at least one further alkaline earth oxide or other divalent oxides represents a simple way of integrating network modifiers into the glasses since alkali oxides are to be avoided for reasons described in Sects. 7.3 and 7.5. Lara et al. studied the effects of the additions of MgO, CaO, and ZnO on the molar base composition 50 SiO2 -(45-x) BaO–x RO–5 Al2 O3 with (with R = Ca, Mg, Zn and x = 0,15) which corresponds basically to the crystalline phase BaSiO3 (see Fig. 7.8) (Lara et al. 2004b). They investigated their softening behavior by hot stage microscopy. The ternary composition 50 SiO2 -45 BaO–5 Al2 O3 showed a softening temperature of 1217 °C after sintering due to strong and rapid crystallization of BaSiO3 together with BaAl2 Si2 O8 (Table 7.3). This temperature lies well above sealing and operating temperatures of SOC, and no joining process is possible with this composition. In contrast, all quaternary compositions have softening temperatures in the region of 900 °C with subsequent crystallization which are suitable temperatures for SOFC technologies. Another possibility to extend ternary base systems with additional oxides is the addition of further network-forming oxides such as B2 O3 or P2 O5 (Lahl et al. 2000; Da Silva et al. 2016). These oxides stabilize the glass formation and reduce the softening temperature, but in high portions they are too unstable against the aggressive atmospheres and show higher vapor pressures, especially in steam-rich atmospheres. In small amounts, they can have a positive effect on the stabilization of the glass Table 7.3 Temperatures (°C) for the viscosity fix point according to hot stage microscopy, glass transition temperature (Tg), and crystallization temperature Tc (DTA) all in °C after (Lara et al. 2004b) Glass Initial Maximum Softening Half-ball Flow Tg 50 SiO2 -(45-x) BaO–5 Al2 O3 shrinkage shrinkage

Tc

+ 5 CaO

765

862

896

1200

1245 680 1000

+ 5 MgO

765

862

931

1211

1250 745 970

+ 5 ZnO

751

838

907

1175

1200 740 940

x=0

751

863

1217

1230

1260 750 975

Glass Ceramic Sealants for Solid Oxide Cells

169

phase, control the crystallization and obtain a residual glass phase that gives the glass-ceramic sealant plastic formability even after joining or a partial crystallization has finished. (Ley et al. 1996; Larsen et al. 1999; Kuang and Easler 2007). Lahl et al. added amounts of 5 mol-% B2 O3 to ternary RO-Al2 O3 -SiO2 glasses (with R = MgO, CaO, BaO, and SiO2 :RO = 1) to effectively reduce the glass transition temperature Tg (Lahl et al. 2000). They received RSiO3 and the RAl2 SiO8 as crystalline phases in case of CaO- and BaO-based glasses. Da Silva et al. investigated the effect of B2 O3 additions in BaO-Al2 O3 -SiO2 -glasses with molar SiO2 :BaO ratios of ca. 0.65. Contents of 6,5 mol% B2 O3 and less did not create any boroncontaining phase while higher contents up to 15.6 mol% lead to the formation of BaB2 O4 , which has already been mentioned to cause thermomechanical stresses (Da Silva et al. 2016). A further effect of the addition of B2 O3 (change of B2 O3 :SiO2 ratio from 0 to 0,1) can also be the decrease of the crystallization temperature (Tc) of more than 100 K as shown by Sohn et al. for BaO-Al2 O3 -B2 O3 -SiO2 -glasses and by Pascual et al. for MgO-BaO-B2 O3 -SiO2 -glasses (Sohn et al. 2004; Pascual et al. 2007). This effect can be useful for the adjustment of sealing temperatures for SOC concepts operating at lower temperatures (< 700 °C). A dedicated effect of Al2 O3 additions in BaO-Al2 O3 -B2 O3 -SiO2 glasses with a BaO:SiO2 ratio between 1,0 and 1,2 and B2 O3 contents between 16, 7 and 20 mol% was investigated by Sun et al. (2010). An addition of less than 2 mol% does not retard the crystallization while additions between 2 and 5 mol% of Al2 O3 have been proven to stabilize the glass structure and allow a good wetting of the glass on 8YSZ-surfaces before crystallization processes start. Even though the number of publications on ternary and quaternary glass systems represents only a minority of the known literature on this topic, fundamental approaches and principles can be demonstrated on the basis of the available data. It is obvious that the glass–ceramic sealants considered here are not conventional glass compositions such as container glass or flat glass, as they contain significantly higher proportions of network modifiers. Glasses with such high proportions of network modifiers lose their structural integrity because the low SiO2 proportions mean that a coherent glass network can no longer be formed. From a structural point of view, they consist of discontinuous silicate tetrahedron groups tied together with modifying cations. In case of small polyvalent cations such as Mg2+ , Ca2+ , Zn2+ , which exert strong cohesion between separate silicate groups, so-called invert glasses are formed. The silicate tetrahedron in invert glasses can be bonded together as metasilicates if the [O]:[Si] corresponds to 3 or as polysilicates (short chains) if this ratio is larger than 3 (Lessing 2007; Donald et al. 2011). Due to this absence of a coherent network consisting of network formers, these glasses show a strong tendency to crystallize when heated to temperatures above Tg. As this crystallization tendency is often too strong or to fast with regard to the time scale of technological relevant sealing processes, also the use of quaternary glasses as real sealant is only hardly possible with a low reliability. The way to control the designated crystallization processes is to add further oxides with different ionic radii which do not fit into these crystalline lattice structures and therefore hinder the crystallization to some extent. This is the main reason for the development and investigation of more,

170

J. Schilm et al.

multicomponent complex glasses, which also make up the majority of the known literature on this subject (Meinhardt et al. 2008). This behavior is one of the reasons why many glass–ceramic sealants based on BaO are blended with the oxides of the elements Mg + , Ca2+ , Zn2+ . Besides the extension of the glass-forming range and the control of the crystallization processes, there is another motivation to add certain oxides (e.g., La2 O3 ) to the glasses, namely the formation of further suitable crystalline phases with high-expansion coefficients (Ley et al. 1996; Lahl et al. 2000; Heinrich and Aldinger 2001; Sohn et al. 2004). Up to this present section, we have essentially explained, based on simple glass systems and model glasses, how the glass compositions that are potentially suitable as glass–ceramic sealants can be identified. However, almost all glass sealants published in the literature have more than four components. This results in quite many glass compositions, whose nominal properties may be suitable for joining SOC components. It should be noted that only very few of the published glass sealants have been tested and qualified in real stacks. Therefore, within the scope of this overview, a systematic classification of the known glass ceramic sealants or the corresponding systems is not possible. Thus, only examples of suitable sealing glasses will be briefly explained here for demonstration purpose. A good example for a glass ceramic sealant based on five components is the BaO-CaO-Al2 O3 -B2 O3 -SiO2 system, which results from additions of CaO to the quaternary BaO-Al2 O3 -B2 O3 -SiO2 system. On the one hand, CaO causes the desired delay of crystallization processes and, in a suitable content range, can also cause the formation of Ba-Ca silicates (Table 7.1). A known glass sealant of this type, which has been claimed for practical use in stacks, is the G18 of the Pacific Northwestern National Laboratories (PNNL) with about 70 vol.-% crystalline phase content, which main crystal phase is BaSiO3 . Other phases formed are Ba3 CaSi2 O8 and BaAl2 Si2 O8 (hexagonal modification) and referred as secondary phases (Yang et al. 2003; Meinhardt et al. 2008). The CTE of G18 is reported to be as high as 10.5·10–6 K−1 . However, the transformation of hexagonal BaAl2 Si2 O8 into the monoclinic modification in long-term aging up to 1000 h at 750 °C was observed. Theerapapvisetpong et al. investigated the effect of B2 O3 additions (5 Ma.%) on crystal phase formation in BaO-free glasses in the CaO-MgO-B2 O3 -Al2 O3 -SiO2 system, where MgO and CaO are present in similar proportions and reported CTE’s ranging from 10.4 to 11.7·10–6 K−1 (Theerapapvisetpong et al. 2011). In all their B2 O3 -free base glasses they found Åkermanite (Ca2 MgSi2 O7 ) as the only crystalline phase. In contrast, the B2 O3 -containing glasses formed additional crystalline phases beside Åkermanite. This observation indicates that also additions of network-forming oxides can affect the type of formed crystalline phases in each base glass composition. The addition of La2 O3 to BaO-Al2 O3 -B2 O3 -SiO2 -system has already been mentioned as a further useful option to increase the CTE of the glass due to the large ion radius of lanthanum but moreover for the introduction of other high-expansion crystalline phases (i.e., Ba4 La6 O(SiO4 )4 ) as demonstrated by Sohn et al. (2004). It was shown clearly that the CTE increases from 7.7·10–6 K−1 for a BaO content of 20 mol% to 11.1·10–6 K−1 for 40 mol% of BaO while Tg and Tc remained on a nearly constant level of 670 and 740 °C in selected glass system (Sohn et al. 2002).

Glass Ceramic Sealants for Solid Oxide Cells

171

Table 7.4 Oxide additives and their function in glass ceramic sealants (Meinhardt et al. 2008; Goel et al. 2010; Mahapatra and Lu 2010b) Additive

Function in glass ceramic microstructure

Ga2 O3

Hinders crystallization and reduction of glass viscosity

La2 O3 , Nd2 O3

Increase of thermal expansion and modification of viscosity

Y2 O3

Increases the glass viscosity

ZnO, PbO, Bi2 O3

Conditional network formers, improve flowability of glass by reduction of viscosity

Cr2 O3 , V2 O5

Reduction of surface tension

CuO, NiO, CoO, MnO Improves adhesion of glass sealants to cell components TiO2 , ZrO2

Nucleates or introduces crystallization centers

In case of even more complex glasses with six and more components, publications are often limited to the variation of individual components or ratios of two components in each system and to the presentation of the resulting effects on glass properties. Accordingly, only incremental knowledge is gained from individual publications. However, such studies also reveal whether the treated glasses are appropriate for joining SOC components or not, as these complex glasses are usually preceded by extensive fundamental investigations into the selection and evaluation of the base systems. Table 7.4 lists the known additives of glass sealants and the effects assigned to them in glass structures. Regarding transition metal oxides, it should be noted that these are often added to conventional glass solders and enamels to improve wetting on metallic surfaces. The reason given for this is the presence of small amounts of these elements in the metallic alloys which has to be coated or joined. Hence the addition of transition metal oxides is sometimes also adopted for the development of SOFC sealants to achieve improved adhesion and wetting on the interconnect materials. It remains questionable if this leads to any favorable effect, as the high sealing temperatures lead to the formation of oxidation layers during the sealing process on metallic interconnects. This effect facilitates the wetting by the glass melts in a similar way as intended by the added transition metal oxides (Li et al. 2021). There are also studies demonstrating that additions of polyvalent oxides or transition metal oxides lead to increased degradation reactions of the glass sealants (see Sect. 7.5). In summary, it can be said that most of the published glass-ceramic sealant compositions have probably not been validated in real stacks. There are also products for this application offered from commercial suppliers, which need validation under customized sealing conditions. Finally, it is important to understand that glass properties are closely related to development of joining process and the stack design and cannot be considered as a completely stand-alone task in the field of SOC.

172

J. Schilm et al.

3.2 Multicomponent Systems Including Fillers Due to the complex requirements of sealants for SOC and the consequent limited choice of materials, further approaches exist to utilize sealants in the form of composite materials where a glass phase can be both the matrix phase and the filler. So, mica gaskets known for a better thermal shock resistance can be infiltrated with glass solders, which are intended to block the leakage paths between the individual mica platelets. For example, by infiltrating mica with boric acid, hydrate silicates, talc, or bismuth nitrate, the impermeability can be improved. The reason is that the two compounds react with the mica during the joining process and transform into glass phases, thus closing the leakage paths between the individual mica platelets (Chou and Stevenson 2004; Le et al. 2006; Wiener et al. 2007). However, the original advantages of an elastic and non-bonding mica-based seal are lost (Wiener et al. 2007). In a similar way, ceramic fiber mats can also be used for this approach instead of mica (Yeong et al. 2003). Finally, this results in a composite seal that is significantly more complex in its reactive behavior than the single phase crystallizing base glasses. More systematically, the technique of modifying glass solder matrices with particle fillers has been investigated and developed. The predominant motivation for this approach is to increase the coefficient of thermal expansion of the resulting composites and to match them better to the components to be joined. Therefore, ceramic and metallic fillers are used that have a higher CTE than the glass matrix. Another effect of the fillers is an increased apparent viscosity of the glass melt, which is viscous under joining and operating conditions, giving the SOC stacks increased stability (Lejeune and Richet 1995). An analogous effect is obtained with glass–ceramic microstructures based on single-component, crystallizing glasses (see Sect.7.2). A prerequisite for the stability of such composites over the lifetime of SOC stacks is a constant phase content. Thus, the fillers should neither provoke crystallization of undesired phases nor should their dissolution in the glass melt lead to uncontrolled phase formation. One approach to composite sealants is the use of ceramic particles as fillers in a matrix of a glass sealant to adjust the CTE, viscosity, and also mechanical strength. The resulting microstructures can also be classified as glass–ceramic structures. The addition of NiO in a SrO-B2 O3 -Y2 O3 SiO2 -CaO-based glass matrix could linearly increase the sealant CTEs in agreement with Turner’s equation for the rule of mixture, and also the mechanical properties were improved up to a content of 15 vol.% (Chou et al. 2007b, 2007a). However, no data are given on the long-term stability of NiO in a viscous glass matrix at high temperatures with respect to the formation of new crystalline phases or reactions with metal interconnects. In a similar way, the additions YSZ particles and YSZ-fibers to barium-calcium-silicate glass matrix enhances the mechanical properties of the resulting composite seal in comparison to the pure glass ceramic material (Gross et al. 2011). In case of using MgO particles as a filler in a sodium aluminosilicate glass matrix, a considerable increase of the CTE close to 12 ppm K−1 was observed, but also the dissolution of the MgO particles into the glass matrix and consequently

Glass Ceramic Sealants for Solid Oxide Cells

173

a change of the properties is reported (Nielsen et al. 2007). Crystalline barium disilicate particles (BaSi2 O5 ) added to a CaO-MgO-Al2 O3 -SiO2 base glass phase to form a composite sealant transformed at least partially into the Celsian phase which has a lower CTE (compare Tables 7.1 and 7.2). Due to interactions at temperatures above Tg of the glass phase, the positive effect that certain oxides with large CTE values can bring about in composite solders is canceled out in the long term. However, ternary BaO-MgO-SiO2 glasses without Al2 O3 give stable composites, where crystalline silicate phases create a stable multi-phase microstructure allowing to tailor of the properties of sealants (Gross et al. 2005; Wanko et al. 2007, 2008a, b). The influence of silver particles as metallic fillers on glass or glass–ceramic composite seals has been investigated to a much lesser extent (Beatty 2005; Corral et al. 2008; Gross et al. 2011). Despite the proven effects of increased CTE and apparent viscosity up to now such concepts are not followed intensively. The reactivity of the materials used at the operating temperature of the SOCs is a major issue of composite seals. The more components are in contact with each other at high temperatures, the higher the probability that undesirable reactions will occur driven by gradients of chemical potentials. Consequences are the formation of new phases, which can negatively affect the sealing properties. For this reason, sealants for high-temperature fuel cells should contain as few components as possible, and their reaction characteristics should be well-balanced.

4 Technological Aspects of Application of Glass Ceramic Sealants It was already explained in the introduction that glass solders are used normally as powders. Pastes and casted tapes are the technologically mostly relevant semifinished products, which are applied to the interconnects and electrolytes prior/during the assembling of stacks. Pastes can be processed efficiently, scalable and almost lossfree using screen / stencil printing technologies to single layers with thickness up to approx. 120 μm. For larger thickness, several printing steps are required. Likewise, pastes can be printed with significantly greater heights via dispensing technologies, depending on the paste properties. The use of casted tapes enables the realization of planar sealing parts in any individual geometry via cutting and punching. They can be produced in heights of up to about 500 μm as single-layer components. This technology enables easy handling and application of glass gaskets when stacks are built up manually or automatically. For the correct design of the height of the sealant to be applied, it is necessary either to know the porosity of the paste or tape in the as-produced state, or to know its respective shrinkage by means of a dilatometric measurement (hot stage microscopy). From these data, it is possible to calculate the volume by which the sealing compound will shrink during the joining process and the volume it will take up in the joined state. A slightly different situation is given in case of tubular stack concepts. Here, the glass sealants are not arranged in a sandwich

174

J. Schilm et al.

configuration but more as an open butt weld or filet located between the outer tubular cell walls and the openings of an interconnector plate. Such a configuration requires a good wetting by the glass sealant, and under joining conditions, it should be ensured that a wetting situation accordingly at least to the half-ball-point or even better the flow point (compare Fig. 7.3) is realized. These conditions are necessary because in this configuration there is no external force or dead load to ensure a complete filling of the brazing zone. The joining process of stacks itself is divided into several steps. In the first step, the organic binder components in the pastes or tapes must be completely burnt out (debindered) at temperatures between 300 °C and approx. 500 °C. In this temperature range, holding times of up to several hours and sufficient air supply must be provided to the assembled stacks. It must be considered that the glass sealant layers are sandwiched between the individual stack components and the binder components can only outgas along the lateral dimensions via narrow gaps at the inner and outer edges of the stacks. According to Fig. 7.3, the debinding process is followed by the sintering or densification of the glass particles to a dense micro structure. The sintering behavior of vitreous structures differs significantly from crystalline phases. Glasses do not have a specific melting temperature but soften above Tg and then, as temperatures rise, pass through a logarithmically scaled viscosity range between 1013 poise and approximately 10 poise usual for a typical glass melt. In contrast to crystalline particles (ceramics and metals), the material flow in sintering glass powders is not restricted to the grain boundaries. The particles flow as a whole, with the material flow being significantly greater near the surfaces than inside the particles due to the interfacial tensions acting. It is reported that compacted glass powders, largely independent of their composition, densify in a viscosity range between 109 and 108 poise (Lara et al. 2004b). Therefore, the densification of glass powder compacts is usually possible at temperatures well below the sintering regimes of crystalline phases. In case of glass compositions, which are sensitive to crystallization processes as explained in Sect. 7.3.1, sintering and crystallization might occur in the same temperature range what in turn hinders the densification and also the joining process (Prado and Zanotto 2002). Normally, such glass compositions are excluded from the development process before testing in stacks. Also, unlike the sintering characteristics of crystalline materials, glasses do not need long holding times at their maximum sintering temperatures. Usually, the viscous flow of the whole particles leads to a densified microstructure except a small fraction of residual pores coming from enclosed gases (Moritz et al. 2019). When the maximum possible densification of the glass solder is reached, it is not yet possible to join the components because the viscosity of the glass sealant is too high for wetting or adhesion with the joining partners to take place. Higher temperatures and lower viscosities are necessary for this, at which the molten glass forms at least a ball-point or ideally a half-ball-point on contacted material according to Fig. 7.3. Ideally, the formation of the propertydetermining crystal phases in the glass-ceramic seals should not take place until the joining temperature has been reached and the glass phase has wetted or bonded to the surface of the contacted material (see Fig. 7.10). After joining, the temperatures are lowered again to a level that is suitable for operating the stack.

Glass Ceramic Sealants for Solid Oxide Cells

175

5 Interaction of Sealing Glasses During SOC Operation In the section above, the design of glass-ceramic sealants and joining of SOCs were described. To some extent, also the operating conditions of SOE devices and requirements have been discussed. In this section, these aspects will be covered in detail. The glass-ceramic materials shall ensure reliable long-term operation for about 40.000 to more than 80.000 h of SOC service. The service conditions are very harsh: temperatures between 700 to 850 °C, oxidizing atmosphere in regions exposed to air or oxygen, and reducing or even oxidizing atmosphere in regions containing steam with respect to the actual SOC operation mode. Additionally, electrical voltage across the sealants and the contact zone to metallic interconnectors increase the number of possible adverse interactions. Against every single influential factor as well as their superposition, a high stability of the sealant materials is required. It is comprehensible that under these conditions both monolithic materials and rather complex composites of different materials are subject to degradation reactions. Especially in relation to the long-term operation, interface reactions between the sealant glasses and adjacent components must be considered (Fig. 7.11).

5.1 Interaction with Interconnects One of the following material types is usually used as SOC interconnect: ferritic chromium steel, chromium-based alloys or the ceramic lanthanum chromite (LaCrO3 ) (Horita et al. 1993, Sarkardeh et al. 1999). Due to high bill of materials (high La2 O3 content) and manufacturing costs for LaCrO3 , the interest in the ceramic interconnect material continuously decreases (Quadakkers et al. 1994). On

Fig. 7.11 Electrochemical and chemical conditions across sealing glasses in SOC stacks

176

J. Schilm et al.

the other hand, multiple SOCs have been realized using metallic interconnect and its reliable operation has been demonstrated. For ferric steels and chromium-based alloys, the CTE matches well the thermal expansion of the ceramic electrolyte (see Fig. 7.6). Chromium-based alloys like CFY offer high tensile strength, high stiffness, and good creep resistance even at high temperatures. Their disadvantage is, however, their poor machinability. In stateof-the-art production, interconnects made of chromium-based alloys are produced by net-shaped technology (Venskutonis et al. 2009). This technique is much more expensive as the casting and rolling of ferritic chromium steels. The thermophysical properties of steels are not as superior as the properties of CFY but acceptable for their use in SOC stacks. The ferritic steel can be machined easier, and the interconnect sheets can be made as thin as 300–500 μm, which in turn saves material and weight. For both materials the formation of dense, conductive, and continuous oxide scales at elevated temperatures is important. For electrolyzers as well as for power generation in fuel cell mode, low internal electric resistance of this oxide scale is required. That is why alumina-forming steels with insulating oxide scales are not suitable for SOCs. In this sense, chromium bases oxide scale formers are preferred due to the conductive properties of the formed crystalline phases. Dense oxide scales with a passivating character prevent further access of oxygen and so the oxidation of metallic bulk material. Furthermore, the oxide scales have to be stable in air atmosphere but also in steam/hydrogen gas mixtures (Folgner et al.). One of the challenges of conductive oxide scales is the chromium release. This not only affects the sealing glasses by the formation of detrimental Chromate containing crystalline phases, but also leads to poisoning effects at the SOC cathode. Several protective layers are known to minimize the evaporation of chromium (Sauchuck et al. 2008, 2020). For the whole lifetime of SOC stacks, the glass ceramic sealing material is in contact with chromium-based oxide scales. So, interaction of both must be balanced in terms of a necessary but non-detrimental interface formation. In principle, four different reactions between glass and metallic interconnect can occur: o chromium diffusion into the glass-ceramic sealant, which might result in the formation of conducting phases o pore formation due to release of volatile species from the microstructure and increase of the leakage rate o formation of alkaline or earth alkaline chromates which might lead to short circuits and leakage for the sealing o internal oxidation or break away oxidation of the metallic interconnect after reaction with components of the sealant. The most commonly described interaction is the formation of reaction layers between the interconnect respectively the formed oxide scale and the glass-ceramic sealant. Zhenguo et al. described the development of different kinds of interlayers in one sealing rim based on a BaO containing glass-ceramic sealant (Zhenguo et al. 2003). On the chromia forming alloy 442 as well as on Crofer 22 APU, the formation of BaCrO4 is detected in air. At the interface but more inside the sealing glass with

Glass Ceramic Sealants for Solid Oxide Cells

177

no access of oxygen, a change to chromium-rich solid solutions of Cr and the used BACS glasses (BaO-Al2 O3 -CaO-SiO2 -glass) is reported. The dissolution of Cr into the sealing glass also can cause the pore formation. According to Zhenguo et al. pores can be formed by H2 after the reaction of metal with water dissolved in the sealing glassy phase. X M(alloy) + y H2 O(sealing glass) → Mx Oy (sealing glass) + H2 ↑

(7.4)

The comparison of the Gibbs energy of the reactions of Fe, Cr, and Ni reveals, that chromium is the most reactive species. So, Cr leads to the formation of H2 which is captured between glass and metallic interconnect and results in the formation of pores in the glass. Systematic investigations on interfacial reactions between interconnector alloys and glass–ceramic sealants were carried out by Rost, Schilm, and Kusnezoff on partially crystallizing barium aluminosilicate glasses to which certain amounts of ZnO were added to control the crystallization process (Rost et al. 2009; Schilm et al. 2012, 2018). First to say, in any cases they confirmed the formation of Cr-containing interlayers between sealing glass and interconnects. At the interface exposed to air, usually a scale of BaCrO4 is formed. The exposure time at elevated temperatures, the BaO content of the glass, and the presence of Cr-containing gas species rule the formation of BaCrO4 on the surface of sealant exposed to air, marked with (1) in Fig. 7.12. The chromate formation affects the surface, but not the hermeticity of the sealing. As marked with the number (2) in Fig. 7.12a) and depicted in Fig. 7.12b), the formation of BaCrO4 is detected about 1 mm from the surface exposed to the air. Due to the strong anisotropy and high CTE value, the presence of such a phase at the interface between sealing glass and interconnect is detrimental to the hermeticity of the sealing during thermal cycling. Crack propagation at the interface can form open cracks and pathways in SOC stacks during thermal cycling and lead to leakages. It has been found that the growth of BaCrO4 into the glass is only possible with the access of oxygen. For suitable sealing glasses, the formation of chromates is limited to the surface of the sealing exposed to air and chromates are not present at the interface (see Fig. 7.12c and d). At the central region of the joining zone, a 2–3 μm thick and dense oxide scale has formed at the interface between Crofer 22 APU and the sealing glass, as observed in Fig. 7.12c) as a dark gray phase. According to EDS analysis, not only Cr, Mn, and O are present, but also Zn. The composition varies in the amount of Zn, Cr, and Mn. A solid solution of Zn in (CrMn)3 O4 is the assumed reaction product between the sealing glass and metal interconnect. More severe is the presence of crystals of the same composition grown up to 30 μm into the glassceramic microstructure of the (see nearly black appearing crystals in Fig. 7.12c). Due to the electric conductive properties of these metal oxides, the resistivity of the sealing glass is affected adversely. Especially in combination with a voltage applied across the sealant, the formation of such conducting phases must be avoided. At the interface between the sealing glass and Crofer 22 APU interconnects, a further change in the microstructure is present at the fuel side of a sealant (see Fig. 7.12d). Near the glass surface exposed to reducing atmosphere, the thin oxide

178

J. Schilm et al.

Fig. 7.12 Cross sections of the interface of sealing glasses and metallic interconnects from a air side to d fuel side, (Rost et al. 2009)

scale can be observed on top of the metallic interconnect. At the interface between glass and interconnect, a bonding layer of Ba, Cr, Si, and O is formed. According to EDS analysis, a composition suggesting BaCrx Si4 O10+y as a possible stoichiometry was identified (the index x in the sum formula summarizes Chromium in the valence states 2 and 3). According to Miletich et al. a gillespite type (ABSi4 O10 ) component is formed, which can be synthesized at temperatures above 840 °C under strongly reducing conditions, where chromium has the rare valence of 2 (Miletich et al. 1997). The temperature range is typical for joining and service conditions of SOC and so the formation of the indicated compound is very likely. However, as the reducing conditions might vary in a SOC stack, is more likely that a composition with mixed chromium valences of 2 and 3 exists at the interface. Summarizing the findings reported in the literature, several reactions between sealing glasses and metallic interconnect materials are known. These reactions not only occur between both joint components but also are affected by the present atmospheres. Across the cross section of the sealed components, starting with the oxidizing air side to the reducing steam/fuel gas atmosphere, three different areas can be identified. At the air side, BaCrO4 is formed at the surface of the glass-ceramic sealant and also at the interface of the glass-ceramic to the metallic interconnect.

Glass Ceramic Sealants for Solid Oxide Cells

179

This phenomenon will be discussed in detail in the next section. In the center region of the sealing, a thin oxide scale between sealing glass and metallic interconnect sheet is present. The oxide scale consists of Cr and Mn of the metal and Zn from the sealing glass, forming a solid solution. The separation of individual crystals of the oxide scale and their presence in the sealing glass must be regarded as critical due its electric conductive nature. Under reducing conditions, a bonding layer of pure chromium oxide between the sealing glass and the metallic interconnect is not obligatory and can have a more complicated composition like BaCrx Si4 O10+y or other phases.

5.2 Interaction with Different Atmospheres The interaction of the surface of glass-ceramic sealants and the surrounding atmospheres clearly depends on the oxidizing or reducing properties of the gas and thus can be separated into air side and fuel or steam side reactions. At the air side, the formation of chromates is widely reported in the literature. Due to the presence of BaO in sealants, the formation of BaCrO4 is the most likely one. The replacement of Ba by Sr or Na often leads to the formation of the respective SrCrO4 (Mahapatra and Lu 2010a; Zhang et al. 2012) or Na2 CrO4 (Sabato et al. 2019). In Fig. 7.13, the formation of BaCrO4 is clearly observed as white appearing phase at the air side. In case of well-adapted sealants for chromium-containing steels, the formation of BaCrO4 is limited to a narrow region at the surface. For SOCs fabricated from interconnects made of chromium base alloys, the utilization of sealants suitable for ferritic steels is not possible. Here, the available amount of chromium at the surface results in high gas phase concentrations of chromium species and the diffusion from the bulk of the interconnect leads to massive formation of chromates on the glass surface exposed to air. For these interconnect materials glass-ceramic sealants with reduced BaO content are required. In Fig. 7.14, the air and fuel side of a sealing glass suitable for CFY (Chromium content > 90 wt.-%) is depicted. Despite the much

Fig. 7.13 Air and fuel side of a SOFC sealing after 1.000 at 850 °C without voltage (Rost 2013)

180

J. Schilm et al.

Fig. 7.14 Chromate formation of an adapted sealing glass applied on the chromium base alloy CFY (Schilm et al. 2018)

higher chromium content of the interconnect, the formation of bright appearing chromates at the air side is in the same range as in Fig. 7.13 for Crofer 22 APU. This result is obtained by profound changes of the glass composition especially the reduction of the BaO content as chromate-forming species. At the fuel or steam side, much less interactions are reported. Measurements on boron volatility have been performed by Zhang et al. (2008). They described a diffusion-controlled mechanism with higher evaporation of hydrated B2 O3 species in wet atmosphere. In contrast to reactions at the air side, in common SOC literature, this is not regarded as limiting factor for the lifetime of SOCs. This fact is confirmed by investigations published by Rost et al. as seen in Figs. 7.13 and 7.14, where no additional phases or decomposition of glass-ceramic sealants under reducing conditions were observed.

5.3 Interaction with Electrolytes In previous sections, it was shown that depending on composition of the sealing glass and interconnects made of ferritic steel and sintered chromium alloys, more or less intensive interaction can take place. For the interface of sealing glasses to ceramics (electrolyte), this is not the case. This combination is much less reactive, and reaction products are formed to a much lower extent. Smeacetto et al. reported no interaction between a SACN sealing glass (SACN: SiO2 -Al2 O3 -CaO-Na2 O) and YSZ after 200 h at 800 °C in a H2 –3% H2 O atmosphere (Smeacetto et al. 2008a) (Fig. 7.15). In earlier studies, the diffusion of Y into glasses was reported, which results in the formation of monoclinic ZrO2 (Lahl et al. 2002). Due to an expansion in volume, the transition of cubic or tetragonal ZrO2 into the monoclinic configuration has to be prevented to maintain crack-free membrane-electrolyte-assembly (MEA) in SOC stack.

Glass Ceramic Sealants for Solid Oxide Cells

181

Fig. 7.15 EDS scan at a glass YSZ interface after 200 h treatment at 800 °C in H2 -3%H2 O atmosphere (Smeacetto et al. 2008a)

In our own investigations, Rost et al. found reaction products between YSZ (MEA and ceramic spacers) and the sealing glass described in the context of Fig. 7.12. Crystalline phases have been formed at the interface between YSZ and the glassy matrix of the sealant (see Fig. 7.16). Direct from the interface, they have been grown into the glass. In principle, two types of morphologies can be detected: sharp needlelike structures with length up to 20 μm length and very short needles up to 2 μm. The second type of observed crystallites has length of about 3—5 μm and width of around 3 μm. For both types, the volume is very small, and for EDS analysis, high contribution of the surrounding matrix has to be considered. Due to the small dimension of the crystals, the formation of recrystallized ZrO2 as stated by Lahl et al. cannot be excluded. The crystallites are embedded in a glassy matrix consisting of Si, Ba, and O. In EDS spot analysis, Zr, Ba, Si, and O are detected with different ratios of Zr to Ba and Si. These data imply the existence of Ba-Zr-silicates at the interface. One of the reported possible crystalline compositions matching these phases is BaZrSi2 O7 (Kolitsch et al. 2003). The formation bonding layers of Ba-Zr-Si crystallites are therefore likely. Schilling intensively investigated the interaction of YSZ with BaO-Al2 O3 -SiO2 type glass-ceramic sealants and reported the formation of a barium zirconate layer with a low amount of silica (Schilling and Roosen 2012). They described a dense layer of approx. 2 μm which has been formed on the surface the YSZ electrolyte. In the glassy matrix in contact to the barium zirconate layer, no further zirconia

182

J. Schilm et al.

Fig. 7.16 Interface of glass and YSZ with reaction zones consisting of Zr, Ba, and Si

was detected. For more detailed analysis, TEM investigations are necessary. Gou et al. performed STEM investigation at YSZ interfaces of two glass seals of different compositions (Guo et al. 2022). In the case of a Ca-Ba-Si glass, he also found Zr-Ba-Si bonding layers. These results agree well with the results of Rost et al.: between YZS and BaAl-(Zn)-silicate glasses a bonding layer of Zr and components of glass Ba and Si is formed due to the limited interdiffusion of elements. In contrast to reactions of interconnect metals and glass, where extensive reaction is widely reported, the glassto-electrolyte interfaces are much less reactive and can be assumed as long-term stable.

5.4 Interaction Induced by Electric Voltage Even if the compatibility glass-ceramic sealants with interconnects, electrolytes or surrounding atmosphere is important, the most important property is to withstand effects caused by electric voltage. Although glasses are known to be electric insulators, not having free electrons for conducting electricity, two principles can cause a real or fictitious electric current. Sealing glasses in SOC stacks operate above the glass transition temperature T g , where the ions in the glass have a certain and measurable mobility. In the electric field of a SOC, the ions in the glassy matrix can migrate. This migration depends on the size of the ions and their charge. Large ions with low charge are less mobile in such glasses as vice versa. The second principle is the change of oxidation state of components by redox reactions. By electron-consuming reactions at cathodic polarized interfaces and electronreleasing reactions at anodic polarized interfaces, it is not necessary that the sealing glass conducts electrons but yet an electric current is detected.

Glass Ceramic Sealants for Solid Oxide Cells

183

Due to the consequence of both effects, migration of ions and redox reactions, the composition of the sealing glasses can change and so the thermal, physical, and chemical properties. For the development of glass-ceramic sealants, the comprehension of internal processes is mandatory. Very much literature is available for the development of sealing glasses focusing on optimized of intrinsic properties (Nielsen et al. 2007; Smeacetto et al. 2008c; Goel et al. 2009; Mahapatra and Lu 2010c; Da Silva et al. 2016). Many groups are working on the mechanical properties (Zhenguo et al. 2003; Malzbender et al. 2007; Chou et al. 2008a; Smeacetto et al. 2008a). The number of publications, describing interactions with oxidizing or reducing atmospheres at 600–850 °C also is quite high (e.g., (Eichler et al. 2000; Zhenguo et al. 2003; Menzler et al. 2005; Yang et al. 2005; Smeacetto et al. 2008a). Much lower is the number of references, where electric properties of sealing glasses in contact with interconnects are investigated (Li et al. 2010; Rodríguez-López et al. 2016). In real SOC stacks, all the single effects appear simultaneously: contact to interconnect at elevated temperatures, two different atmospheres and electric voltage. To investigate the result of the superposition of all impacts, tests in SOFC stacks in combination with so-called post-operation inspection or post-test analysis are the best option (Bram et al. 2004; Batfalsky et al. 2006). Testing a whole stack for glass development and characterization is quite expensive. More effective is to test only the sealing glass in real SOC conditions in so-called dual atmosphere tests with an applied voltage. For this topic, only very few literature is available (Haanappel et al. 2005; Chou et al. 2010; Rodríguez-López et al. 2016; Javed et al. 2020). Main advantages are that in a small furnace several samples can be tested simultaneously and the resistance of the sealing glass can be recorded. In Fig. 7.17, a photograph and scheme of a sample are depicted. Two 3 × 6 cm metal sheets are joined by the desired sealing glass. In the lower metal sheet, two holes allow the flush of the inner volume with fuel or steam gas according to SOFC or SOEC conditions. After more than 500 measurements in our group using different glass-ceramics, it can be found that the development of the resistance strictly follows the same pattern. After applying voltage, the resistance is rising up to a maximum value, followed by a constant region before a decrease is measured, as observed in Fig. 7.18. The

Fig. 7.17 a photography of a dual atmosphere sample and b scheme of the sample at testing

184

J. Schilm et al.

Fig. 7.18 Development of resistivity of sealing glass at a voltage of 0.7 V

overall height of the resistance and the duration of every step strongly depends on the individual glass and glass-ceramic compositions and the height of the voltage applied. The first step, increasing the resistance is measured between 40 and 100 h. This is driven by the migration of mobile ions in the glass to the interfaces. Initially, the ions migrate freey in the applied electric field and a low resistance is the result. This results in a charge separation of anions and cations which leads to an increase of the measured resistance. The following decrease of the resistance is usually connected with reactions between the glass-ceramic sealants and the metal interconnects. One consequence of of the ongoing reactions can be immediately recognized by observation of the microstructures after aging sealants with voltage. Without voltage, the BaCrO4 scale at the glass surface exposed to the air is distributed homogeneously, as seen in Fig. 7.13. With an applied voltage, the BaCrO4 formation is shifted to the side with anodic polarization (Figs. 7.19 and 7.12). The term “anodic polarization” here refers to the positively charged electrode, where oxidation reaction takes place. The reaction mechanism in this case is the following: Chromium in the Cr2 O3 oxide scale of the metallic interconnect is oxidized from Cr3+ or Cr6+ correspondingly with the release of electrons which flow into the anodically polarized interconnector sheet. 6+ − Cr3+ (Cr2 O3 ) → Cr (BaCrO4 ) + 3 e

(7.5)

By Cr6+ , BaO of the glass, and oxygen from the air, BaCrO4 can be formed: Cr6+ + BaO +

3 O2 → BaCrO4 2

(7.6)

As shown, oxygen is necessary for the described reaction, which is available at the surface exposed to the air atmosphere. This is the reason why the reaction starts at anodic polarized interconnector at the air side. Additionally, the growth of BaCrO4 along the anodically polarized interface can be detected up to a certain depth in the sealant (see Fig. 7.19a). This is only possible if a permeability for oxygen through

Glass Ceramic Sealants for Solid Oxide Cells

185

Fig. 7.19 Microstructure of a glass-ceramic sealant after 300 h at 850 °C with applied voltage of 0.7 V. The upper side is at cathodic polarization, the bottom side at anodic

glass-ceramic sealant or at the interface of sealant and interconnect metal is given. As shown in Fig. 7.12, for well-adapted sealants, a depth of 1 mm is not exceeded. At cathodically polarized interfaces, the polarization prevents the formation of BaCrO4 . This leads to the described non-symmetric formation of BaCrO4 . Embedded between the BaCrO4 crystals are additionally SiO2 crystals detected. In the SEM images in BSE mode, the crystals appear black. SiO2 is only formed in combination with Crofer 22 APU as interconnect material. In cross-check experiments with gold as inert interconnect material, no formation of SiO2 can be detected (Schilm et al. 2010). This leads to the conclusion that not SiO2 of the sealing glass, but residual silicon from the Crofer 22 APU is responsible for this phenomenon. Similar to the formation of BaCrO4 , SiO2 is also detected near the air side and up to a certain depth into the sealing glass only at the anodically polarized interface. Also, for this reaction, the anodic polarization promotes the following by drawing electrons: − Si (Crofer 22 APU) → Si4+ (SiO2 ) + 4 e

(7.7)

In contact with the glass, Si4+ reacts with surrounding O2− of the glass structure to form SiO2 . To balance the oxygen deficit and ensure electrical neutrality, O2 from the air side is incorporated into the glass structure. So, the formation of SiO2 is limited to the airside region. The whole anodically polarized interconnect is covered by a thin Cr and Mn containing oxide scale of less than 2 μm thickness (see Fig. 7.20d–f). For Zncontaining sealing glasses, the presence of metallic Zn was additionally detected. The formation of continuous oxide scale is the main difference to the cathodically polarized interface. At air and fuel side, oxide scales of up to 2 μm thickness were also detected. In contrast to the anodically polarized interface, no oxide scale is present in the central region of the sealing. The sealing glass is in contact with bare Crofer 22 APU bonded by a thin (< 1 μm) oxide layer (see Fig. 7.20b). At the interface, the already described bonding layer of BaCrx Si4 O10+y has been formed.

186

J. Schilm et al.

Fig. 7.20 Interface of sealing glass and interconnect at different regions (Rost 2013)

In ZnO-containing sealing glasses, an additional reaction can be detected. In the central region at cathodically polarized interfaces, spherical inclusions, partially filled with a bright appearing phase are formed close to the metallic interconnect (see Fig. 7.20b and 7.22). According to EDS analysis, the composition is 83 at% Zn and 17 at% Fe. Phase diagrams for these compositions are described by Mita et al. and Kubaschewski et al. (2001, 2010). For the described composition, the first liquid phase occurs above 665 °C. Under the operating temperature of SOCs, the inclusions consist of Zn-Fe melt. The presence of a metallic melt in a SOC joining is a serious risk for the hermeticity and insulating properties of the sealing (Donald et al. 2008). The formation of Zn also can be detected without applying any voltage near both joining interfaces of sealing glasses (Rost 2013). If voltage is applied, the reaction is suppressed at anodic polarized surfaces and only occurs at cathodic polarization. The excess of electrons supports the following reaction: − Zn2+ (Glas) + 2 e → Zn

(7.8)

If voltage is applied at a sealing glass, the deficit respective excess of electrons causes distinct reactions between metallic interconnect materials and components of the glass as described above. So, one contribution to the measured electric current are redox reactions with turnover of electrons. Another contribution is the presence of electric conductive species in the sealing glasses. After treating sealing glasses at elevated temperatures, crystals with the composition of the oxide scale of Crofer

Glass Ceramic Sealants for Solid Oxide Cells

187

22 APU can be detected in the matrix of the glass, consisting of Cr-Mn–O and for Zn-containing glasses also Zn. Without voltage, the crystals can be detected at all interfaces between thee sealants and the interconnect. With applied voltage, they concentrate at cathodically polarized interfaces, if they are not covered with a SiO2 layer and also at anodically polarized interfaces. With SiO2 interlayer, no such crystals are detectable at cathodically polarized interfaces. This leads to the following assumption: positively charged chromium and manganese ions migrate from the metallic interconnect into the glassy matrix and toward the anodic polarized interface. Without SiO2 layer, further chromium and manganese are provided by the Crofer 22 APU. In this case, the Cr-Mn-Zn–O crystals can be detected anywhere in the sealing glass. With SiO2 interlayer, the migration of further chromium and manganese is blocked. In this case, in the cathodically polarized region a depletion of the described phases is detected and at anodic polarized interfaces an enrichment is detected (Rost et al. 2009). The described process is a second contribution to an electric current inside the microstructure of a sealing glass. Not only redox reactions with electron turnover affect the resistance of sealing glasses but also the migration of ions in the electric field. In the investigated samples, ions of the metallic interconnect chromium and manganese and zinc from the sealing glass itself are the migrating and reacting species. The described effects depend on electric voltage. During 300 h of testing, only limited results on degradation effects can be obtained. Even 1000 h of testing is very short in contrast to 40,000 h of a proposed lifetime of SOC systems. The possibility of accelerated testing by increasing the applied voltage is of interest. The evolution of the resistance at different voltages of up to 30 V is depicted in Fig. 7.21, in logarithmic scale. The increase of voltage from 0.7 to 1.3 V has only minor effects on the resistivity. Voltages of 5 V and higher clearly affect the resistivity values. With increasing the voltage, the maximum resistivity is increased and the maximum resistance is achieved

Fig. 7.21 Development of resistivity of sealing glasses at increased voltages (Rost et al. 2012)

188

J. Schilm et al.

Fig. 7.22 Interface of sealing glass to Crofer 22 APU samples after testing for 300 h in dual atmosphere and 850 °C with different electric potentials applied (Rost et al. 2012)

after shorter testing times. As the electrical voltage rises, however, the rates at which the resistance decreases after passing through the maximum also increase. At 0.7 V of applied voltage, this drop is only 38%. For a voltage of 30 V, this drop is as large as 90%, meaning a resistivity degradation to a value of only 10%. The reasons can be found by analyzing the cross sections after testing. In Fig. 7.22, interfaces of tested samples after different applied voltages between 0.7 and 30 V are presented. For 0.7 V, thin and smooth interlayers are formed at both polarized interfaces with no considerable porosity. At the anodic polarized interface, single SiO2 crystals (black in Fig. 7.22) and (ZnCrMn)3 O4 -crystals (dark gray) are detected. The morphology of this interface considerably changes for higher voltages. For 5 to 30 V, the oxide scale of the Crofer 22 APU at anodic polarized interfaces gets more and more rugged and porous. On top, a dense and ca. 10 μm thick layer of SiO2 has been formed. Behind this layer, no more (ZnCrMn)3 O4 -crystals can be detected. This supports the assumption that the SiO2 -layer is a barrier to ions migrating from the Crofer 22 APU steel and the oxide scale into the sealing glass. Also, at cathodically polarized interfaces, the increase of voltage leads to more severe reactions at the interface. At 0.7 V, crystallites of BaCrx Si4 O10+y are formed. At higher voltages, continuous layers of this composition with a thickness of 5 to 10 μm are present. Between this layer and the steel, pore formation has been also detected. Due to this pore formation, disintegration of the Crofer 22 APU and metallic droplet-like inclusions in the sealing glass are formed. These metallic inclusions in the sealing glass extremely affect the resistivity. After testing at the respective voltage, leak detection by helium leakage rate measurements (sample type from Fig. 7.17) were performed. For 0.7 and 5.0 V, the treated samples are still gas tight sufficient with leakage rates below 10–10 mbar l s−1 cm−1 . When applying 10 V, leakage rates of around

Glass Ceramic Sealants for Solid Oxide Cells

189

10–8 mbar l s−1 cm−1 are generated. This is a common accepted maximum for the leakage rates in SOC systems due to the fact that hydrogen-containing gases are applied and leakages at the operation temperature lead to immediate combustion and temperature increase at the respective spot. For even higher voltages, He leakage rates between 10–7 and 10–5 mbar l s−1 cm−1 have been measured, which are inadequate for SOC systems. Applying voltage to glass-ceramic sealants lead to distinct reaction, depending on the polarization. Reactions like the formation of BaCrO4 , which also occur without voltage, are drawn to the interface with the cathodic polarization. Under the influence of voltage, also additional reactions are observable. In ZnO-containing sealing glasses, the reduction of zinc to the metallic state was detected, which leads to molten metallic inclusions at the operating temperature. Investigations on degradation effects are time-consuming with treatment times of 300–1000 h. Shortening these times by accelerated testing will be of high interest. The intensity of the described reactions depends on the height of the voltage: increasing voltage leads to more intense reactions. For voltages higher than 5 V, changes in the microstructure of the sealing glasses and the interface to metallic interconnects are detected, which do not occur at operating voltages.

6 Mechanical Properties of Glass Ceramic Sealants Glass-ceramic sealants in solid oxide cell stacks are exposed to mechanical stresses. The type (compressive, tensile, or/and shear) and magnitude of stresses are defined by the sealing geometry and stack operation conditions. The stresses in glass-ceramics are relaxed and can even disappear during long-term annealing at temperatures above T g because of possibility of stress relaxation by viscous flow of softened glassy phase in the glass ceramic microstructure (Liu et al. 2011). There are three main drivers for stresses in SOC stack sealants: – mismatch of thermal expansion coefficients between the glass-ceramics and sealed materials (unavoidable stresses, which are cumulated during the cooling stack from Tg (or operating temperature, if Tg is less than operating temperature) down to the ambient conditions) – thermal gradients in the stack during heat up/cooling down process (stresses can be minimized by optimizing the cooling down/heating up procedure) – thermal gradients in the stack during operation caused by current density and resulting temperature distribution (stresses can be minimized by appropriate design (i.e., co-flow, counter-flow, or cross-flow of oxidant and steam/fuel) and operating conditions (inlet temperatures of gases, etc.). The real stresses appearing in sealants of complex geometry in the stacks are not measured directly but estimated using finite element simulation (Lin et al. 2009),

190

J. Schilm et al.

which results are compared with critical stress values obtained from tests on simplified geometries allowing calculation of specific values (Lin et al. 2012). Under optimized operating conditions of SOC systems, the highest magnitude of stresses in the sealants is reached at room temperature. The simulation results of Lin et al. (2009) for planar stack utilizing Ni/YSZ cermet supported cells, Crofer 22 APU interconnects, and glass seals resulted in maximum in-plane principal stresses in glass ceramic sealant after cooling down of stack to room temperature of 26 to 39 MPa (depending on applied compression force). For calculation of stresses in sealants, the material physical properties like Young’s modulus (E) and Poisson’s ratio (n) are needed. Further physical important characteristic for mechanical strength of sealant is fracture toughness (K Ic ), which can be calculated from Vicker’s indentation technique utilizing the measured Vicker’s hardness (H V ) (Bhattacharya and Shashikala 2019). The measurement of physical materials properties is performed by indentation technique with diamond tips on polished flat glass ceramic surfaces, which allow the calculation of specific material parameters (Malzbender et al. 2002). However, for measurement of elastic modulus and Poisson’s ratio on samples with well-defined dimensions, the ultrasonic pulse technique described in international standards like ASTM E1876-09 can be applied (Malzbender and Zhao 2012). Although the physical properties (E, n, and HV ) are temperature-dependent, the material characteristics mostly reported in the literature are related to ambient conditions (room temperature). Physical properties of quenched sealing glasses in dependence on BaO content in BaO-SiO2 -CaO-Al2 O3 system have been investigated by Bhattacharya and Shashikala (2019), which are summarized in Table 7.5. It was found that the elastic modulus of glasses depends on composition (increasing with BaO content) as well as fracture roughness, where the correlation with BaO content is not so evident. The comparison of physical properties of barium/calcium (Bhattacharya and Shashikala 2019) and sodium-based silicate glasses (Kilinc and Hand 2015), mostly used as sealants, shows that all glasses have an elastic modulus in the range of 60–85 GPa and corresponds to Poisson’s ratios of 0.20–0.22 (Kilinc and Hand 2015). In general, the material properties of glass in non-crystallized and crystallized states are different. Even porosity has some influence on elastic modulus, which decreases with increasing porosity (Selçuk and Atkinson 1997). Table 7.6 summarizes the reported deterioration of physical properties of glass-ceramic sealants during annealing/crystallization. For a fully crystallized material, no changes on the properties with annealing time were observed. For samples with high amount of glassy phase in the initial state, the elastic modulus increases during progress of the crystallization. This effect can be explained by a higher elastic modulus of crystallized phases compared to glass matrix, resulting also in a higher elastic modulus of the composite material. The increase of fracture toughness in glass-ceramics with increasing crystal phase content is probably connected with crack deflection by crystalline phases, which hinders the crack propagation in the material.

Glass Ceramic Sealants for Solid Oxide Cells

191

Table 7.5 Physical properties of quenched glasses measured at room temperature reported by Bhattacharya and Shashikala (2019) E GPa

KIc MPa m1/2

HV GPa

BaO-(SiO2 :B2 O3 )-CaO-Al2 O3 + 6 wt.% (ZnO:MgO) (40–50-5–5)

60.4

1.21

6.36

5.29

BaO-(SiO2 :B2 O3 )-CaO-Al2 O3 + 6 wt.% (ZnO:MgO) (35-50-10-5)

62.5

0.7

6.34

9.11

BaO-(SiO2 :B2 O3 )-CaO-Al2 O3 + 6 wt.% (ZnO:MgO) (30-50-15-5)

64.4

0.83

6.48

7.78

BaO-(SiO2 :B2 O3 )-CaO-Al2 O3 + 6 wt.% (ZnO:MgO) (25-50-20-5)

67

1.63

6.45

3.96

BaO-(SiO2 :B2 O3 )-CaO-Al2 O3 + 6 wt.% (ZnO:MgO) (20-50-25-5)

69.7

0.65

6.26

9.56

BaO-(SiO2 :B2 O3 )-CaO + 6 wt.% (ZnO:MgO) (35-50-10)a

58.5

0.22

6.14

BaO-(SiO2 :B2 O3 )-CaO + 6 wt.% (ZnO:MgO) (30-50-15)a

60.4

0.66

6.22

BaO-(SiO2 :B2 O3 )-CaO + 6 wt.% (ZnO:MgO) (20-50-25)a

65.9

0.59

6.3

BaO-(SiO2 :B2 O3 )-CaO + 6 wt.% (ZnO:MgO) (10-50-35)a

72

0.76

6.8

8.97

BaO-(SiO2 :B2 O3 )-CaO + 6 wt.% (ZnO:MgO) (0-50-45)a

80

0.95

7.1

7.48

Composition mol%

Brittleness μm1/2

BaO-(SiO2 :B2 O3 )-CaO-Al2 O3

BaO-(SiO2 :B2 O3 )-CaO

a

28.1

9.44

10.8

Composition is not normalized to 100%, numbers provide from the ratio of glass constituents

For operating conditions of solid oxide cell stacks, the properties of glass-ceramic sealants in the crystallized state are most important. After high-temperature crystallization of initial glass composition, the crystalline phases with higher thermal expansion coefficient compared to glass matrix are formed to adjust the CTE of composite material closely to that of the electrolyte and the interconnect. Such a composite microstructure results in tension stresses in crystalline phases and compression stresses in the glassy phase after cooling down to room temperature. The magnitude of the stresses in two-phase microstructure (only one precipitate phase (p) and matrix phase (m)) can be estimated using the approach of Serbena et al. applied to Li2 O–Al2 O3 –SiO2 glass ceramics (Serbena et al. 2011). Mechanical stresses (negative for compression and positive for tension) in glass matrix with a small amount

192

J. Schilm et al.

Table 7.6 Physical properties of BaO-SiO2 -CaO glass ceramic before and after annealing (crystallization) at 800 °C at room temperature reported by Zhao et al. (2011) Composition wt.%

State

Glassy phase vol. %

E GPa

KIc MPa m1/2

HV GPa

BaO-SiO2 -CaO (48.2-29.8-6.1)

Initial

75

72 ± 5

0.99 ± 0.21

5.6 ± 0.7

BaO-SiO2 -CaO (48.2-29.8-6.1)

Annealed

41

80 ± 9

1.12 ± 0.15

5.5 ± 1.3

BaO-SiO2 -CaO + YSZa (48.2-29.8-6.1)

Initial

83

73 ± 8

0.92 ± 0.22

5.5 ± 1.2

BaO-SiO2 -CaO + YSZa (48.2-29.8-6.1)

Annealed

63

79 ± 6

1.5 ± 0.12

5.1 ± 0.9

BaO-SiO2 -CaO (36.7-46.8-15.8)

Initial

6

77 ± 3

0.64 ± 0.04

7.2 ± 0.5

BaO-SiO2 -CaO (36.7-46.8-15.8)

Annealed

5

77 ± 3

0.69 ± 0.05

7.5 ± 0.5

a

20 wt.% of YSZ particles added to initial glass powder

of spherical crystalline precipitates increase with an increasing difference in thermal expansion coefficients between glass and crystalline phase according to Selsing in Eq. 7.9 (Selsing 1961):    α p − αm Tg − TRT   σp = (1 + νm )/2E m + 1 − 2ν p /E p

(7.9)

with α p and αm as thermal expansion coefficients of precipitates and matrix, E p and E m as elastic moduli of precipitates and matrix and ν p and νm as Poisson’s ratios of precipitates and matrix. In the case of samples with considerable amount of inclusions (is valid for crystalline matrix and glass inclusions or glass matrix and crystalline inclusions), the stresses can be estimated by the relationship according to Eq. 7.10 derived by Hsueh and Becher (1996):    α p − αm Tg − TRT σp = 1/3K p + 1/[4(1 − f )G m ]+1/[3(1 − f )K m ]

(7.10)

with K p and K m as bulk moduli of precipitates and matrix, f as volume fraction of inclusions and G m as shear modulus of matrix. The average stress in glass and crystalline phase at room temperature depends on the volume content of both phases and satisfies eq. (7.11): f σ p = (1 − f )σm

(7.11)

Glass Ceramic Sealants for Solid Oxide Cells

193

The mechanical stresses depend strongly on differences in thermal expansion coefficients between glass and crystalline phase and glass transition temperature Tg. The shown relationships contain strong simplification assuming the spherical particle shape and the presence of only one crystalline phase in the microstructure. Several crystalline phases in crystallized glasses with needle-like shapes are often present, and the calculation of internal stresses in the composite microstructure becomes very complicated leading also to shear stress constituents. Nevertheless, this simplified approach helps to understand the reasons for microcrack formation in sealants after cooling down to room temperature, as observed by several groups (Zhao et al. 2011). Although the microcracks were observed during the thermal cycling of crystallized sealants, the self-healing ability of materials with a sufficient amount of residual glassy phase during heating to temperatures above T g has been recognized to be able to restore the initial crack-free condition (Wenning et al. 2010). This ability allows to restore the gastight and mechanically inherent microstructure, even if small microcracks appear in the sealant after cooling stack down to the room temperature. Macroscopic mechanical strength testing of glass sealants is a topic of continuing research. Especially, the presence of interfaces between glass and sealed materials makes very questionable the consideration of intrinsic material properties of sealants for tests where joints are used (see Sect. 7.5). Mechanical strength for the following types of mechanical load was reported in the literature: • • • •

compression load bending load tensile load shear load.

The tests under compressive load were exclusively reported at high temperatures (Liu et al. 2011) to investigate the creep of crystallized glass and clarify its ability to sustain long-term compressive forces. It was found that the creep rate of glasses containing considerable amount of glassy phase is relatively high and sealants cannot keep the initial shape over the time of long-term stack operation. The creep of sealants allows considerable reduction of the mechanical stresses during hold time and cooling down from sealing temperature to operating conditions (Govindaraju et al. 2009). The compression strength of glass ceramic sealants at room temperature is assumed to be high enough to easily sustain the compression forces needed for stack tension. Several approaches were used for the measurement of bending, tensile, and shear strength at room temperature under different load conditions. The results of different mechanical strength values reported in the literature are classified and summarized in Table 7.7. For measurement of bending strength, the four-point bending method with two types of samples has been reported: head-to-head sealed specimens (see Fig. 7.23a) and sintered glass ceramic samples with polished surfaces (see Fig. 7.23b). Whereas the measurement of bending strength on sintered glass ceramics depicts the bulk properties of material, the head-to-head sealed specimen contains multilayer structure consisting of a thin sealant layer and oxidized surface layers of metals. For this reason, the crack propagation along the interface glass/Crofer 22 APU is likely to

75 6 0.3

0.8

13.7

BaO-SiO2 -CaO (Malzbender and Zhao 2012) (48.2-29.8-6.1) wt.%

BaO-SiO2 -CaO (Malzbender and Zhao 2012) (36.7-46.8-15.8) wt.%

CaO-MgO-BaO-SrO-Al2 O3 -La2 O3 -SiO2 -B2 O3 -NiO (Reddy et al. 2013) (22.14-24.6-0.52-0-1.23-1.23-47.79-1.69-0.79)

CaO-MgO-BaO-SrO-Al2 O3 -La2 O3 -SiO2 -B2 O3 -NiO (Reddy et al. 2013) (17.19-24.56-0.54-4.91-1.23-1.23-47.75-1.76-0.82)

CaO-MgO-BaO-SrO-Al2 O3 -La2 O3 -SiO2 -B2 O3 -NiO (Reddy et al. 2013) (12.26-24.52-0.56-9.81-1.23-1.23-47.71-1.83-0.85) 75 N/A N/A

BaO-SiO2 -CaO (Gross et al. 2011) (48.2-29.8-6.1) wt.%

SrO-CaO-Y2 O3 -B2 O3 -SiO2 (Chou et al. 2008b) (42.5-9-6-8.5-34) mol.% (Chou et al. 2007c)

BaO-B2 O3 -SiO2 -Al2 O3 (Lin et al. 2012)

Tensile load

75

Glassy phase vol. %

BaO-SiO2 -CaO (Malzbender and Zhao 2012) (48.2-29.8-6.1) wt.%

Bending load

Composition

Figure 7.24c

Figure 7.24b

Figure 7.24a

Figure 7.23b

Figure 7.23b

Figure 7.23b

Figure 7.23 b

Figure 7.23b

Figure 7.23a

Test setup

Table 7.7 Mechanical strength of glass ceramic sealant materials reported in the literature

Glass ceramic Mixed/adhesive

2.5 × 0.25 × 0.044

Glass ceramic

0.3 × 0.4 × 4.0

1.25 × 1.25 × 0.02

Glass ceramic

0.3 × 0.4 × 4.0

Glass ceramic

Glass ceramic

0.3 × 0.4 × 4.0

D3.2 × D1.2 × 0.6

Glass ceramic

0.3 × 0.3 × 4.5

(continued)

22 ± 4

6.3 ± 0.9

3.8 ± 0.03

115 ± 3

144 ± 5

156 ± 8

91 ± 12

55 ± 7

Glass ceramic

0.3 × 0.3 × 4.5

Strength MPa 52 ± 1

Crack propagation pathway

0.3 × 0.3 × 0.01 Glass ceramic

Tested volume cm × cm × cm

194 J. Schilm et al.

N/A

CaO-SiO2 -Na2 O-Al2 O3 (Smeacetto et al. 2008b) (24.0.26-53.0.58-10.0.12-6.0.8) mol.% 75 N/A N/A

BaO-SiO2 -CaO (Faidel 2012) (48.2-29.8-6.1)

BaO-B2 O3 -SiO2 -Al2 O3 (Lin et al. 2012)

BaO-SiO2 -B2 O3 -CaO-Al2 O3 (Ferraris et al. 2020) (26-55-8-7-4) mol.%

Shear load

Glassy phase vol. %

Composition

Table 7.7 (continued)

Figure 7.25c

Figure 7.25b

Figure 7.25a

Figure 7.24b

Test setup

Glass ceramic Adhesive Mixed

2.5 × 0.6 × 0.044 D0.5 × 0.02

Glass ceramic

8 × 6 × 0.02

2 × 2 × 0.02

Crack propagation pathway

Tested volume cm × cm × cm

49 ± 10

6.2 ± 2.2

12

6

Strength MPa

Glass Ceramic Sealants for Solid Oxide Cells 195

196

a

J. Schilm et al.

b

45mm

40mm

3mm

3mm

3mm

4mm

1 mm

Fig. 7.23 Head-on-head samples for bending strength measurement accordingly (Malzbender and Zhao 2012) versus sintered bulk glass samples with polished surfaces used in (Reddy et al. 2013)

Fig. 7.24 Proposed setups for tensile stress measurement: ring-on-ring test (Gross et al. 2011) a, plate-on-plate test (Chou et al. 2008b) and head-on-head test (Lin et al. 2012)

a

Shear tool

b

c

Force

Plates

Glass sealant

Sample

Fig. 7.25 Proposed setups for shear stress measurement: pull- and shear test (Faidel 2012) a, sandwich shear test (Lin et al. 2012) and torsion test setup (Ferraris et al. 2020)

happen. Even if the values obtained by Malzbender et al. on both types of samples are very similar, the larger tested volume of the sintered samples should lead to much higher bending strength as in case of the small glass volumes used in head-to-head specimens in comparison with the observed one. This fact shows that interfaces in head-to-head specimens play an important role and cannot be neglected. Tensile load stress is a very uncommon load situation for seals in compressed planar stack environments. Nevertheless, the measurement of tensile strength is

Glass Ceramic Sealants for Solid Oxide Cells

197

important for ability of stack handling without compression as well as for sealing of tubular single cells in ceramic or metallic gas manifolds (Sumi et al. 2019) (Schilling and Roosen 2012). For measurement of tensile stress, similar experimental setup with minor modifications / improvements is used (see Fig. 7.24). Except of the values reported by Lin et al. for its GC9 material (Lin et al. 2012), the glass-ceramic sealant cannot sustain high tensile stresses. The values reported in (Lin et al. 2012) are probably strongly related to small volumes tested and specific setup selected. Shear load is the most important characteristic and the development of a reliable methodology for its measurement is in focus of experimental efforts. The following setups have been proposed for the realization of this type of mechanical load conditions: – pull-and-shear test setup (Faidel 2012) (Fig. 7.25a) – sandwich setup (Lin et al. 2012) (Fig. 7.25b) – torsion setup (Ferraris et al. 2020) (Fig. 7.25c). In all proposed setups, the crack propagation pathway is decisive for specific values obtained from the corresponding measurements. The conditions and magnitude of shear strength measured for material combination Crofer 22 APU|Sealant|Crofer 22 APU are summarized in Table 7.7. Especially, the torsion test (Cela Greven et al. 2018; Ferraris et al. 2020) provides the options for temperaturedependent investigations and delivers the reproducible values. It allows the allocation of corresponding crack propagation pathways and assignment of obtained values to glass ceramics, oxide layer, or mixed mode (Ferraris et al. 2020). The detailed investigation of stress distribution in sealants of radial symmetric samples (see Fig. 7.24a) revealed strong dependence on edge effects like glass underfilled gap (free-edge) or glass squeezed out of gap (Fakouri Hasanabadi et al. 2017), which is often neglected in stack simulation (Lin et al. 2009) due to the necessity of detailed imperfections description and fine calculation mesh application in these areas. In both cases, it was found that maximal shear stresses in the sealant are considerably higher in comparison to ideal conditions and especially free-edge condition is critical. The situation of glass squeezed out of the sealing gap is common for planar stacks and also results in higher mechanical stresses in relatively small glass volume in the close-to-the-edge regions. Especially, shear strength values reported by different groups show high scattering and reveal the challenge for interpretation of obtained data, as they are strongly influenced by the adhesion strength of sealant to metal, mechanical strength of oxide layers beneath the sealant, adhesion strength to electrolyte surface, flaw shape, and dimensions and the crack propagation pathway. To overcome this challenge, Frandsen et al. (2019), Ritucci et al. (2019) have optimized the approach originally proposed by Malzbender et al. (2003) and applied the four-point bending strength method for measurement of critical energy release rate during the crack propagation (see Fig. 7.26). Using critical energy release rate and effective elastic constant, the fracture toughness of sealant can be calculated. By the proposed approach, also the separation of glass and interface mechanical properties in case of control of crack propagation pathway can be done. The sensitivity analysis of the effect of

198

J. Schilm et al.

a

Stiffener

Sealant

Substrate

b

Fig. 7.26 Modified four-point bending test for measurement of crack propagation energy: initial setup (Malzbender et al. 2003) and modified symmetrical setup (Ritucci et al. 2019)

microstructure inhomogeneities (i.e., size of defects) on measurement result showed only limited influence on experimentally measured values. The obtained values of critical energy release done by Ritucci et al. (2019) are close to the results of earlier work of Malzbender et al. (2003) for similar glass composition, indicating that it is a promising approach for a more reliable description of mechanical properties of joints and sealants. However, the application of critical energy release values for evaluation of crack formation, probability in stack sealants by modeling, requires novel approaches for definition of crack formation based on critical energy release rate as crack initiation criteria, instead of description based on critical stresses, fracture toughness, and flaw dimensions (Miao et al. 2021). Measured mechanical strength values at room temperature show that especially tensile stresses are critical for glasses and stack design. Therefore, the CTE of glassceramic sealants should be correspondingly adjusted to keep the sealant material under compression and allow crack-free thermal cycling.

7 Concluding Remarks Due to the almost unique requirements for glass–ceramic sealants defined by the operating conditions and material combinations of SOC stacks, a complex field of research arises with regard to the glass systems under investigation and the resulting

Glass Ceramic Sealants for Solid Oxide Cells

199

glass–ceramic materials. The explanations in this section cannot do justice to the entirety of the published research on this topic and have therefore limited their scope to relevant aspects and the explanation of possible development strategies exemplified by popular glass systems. Of course, glass–ceramic compositions of other glass systems can also be suited for the joining of SOC stacks. This is ultimately determined by their intrinsic material properties (CTE, softening behavior, crystallization characteristics) and even more so by their operational behavior in the stacks themselves. Since direct performance tests of glass sealants in SOC stacks are very complex, time-consuming, and cost-intensive, additional characterization methods have been established parallel to the actual glass development. These are intended to provide information on the chemical stability and mechanical properties of the glass–ceramic sealants in combination with stack materials at the level of joined composite samples. According to the authors, the development of glass–ceramic sealants for SOC can only succeed if the intrinsic material development is accompanied by technologically relevant characterization methods.

References Abd El-Moneim A, Youssof IM, Abd El-Latif L (2006) Structural role of RO and Al2 O3 in borate glasses. Acta Materialia 54:3811–3819 Batfalsky P, Haanappel VAC, Shemet V, Vinke IC, Steinbrech RW (2006) Chemical interaction between glass-ceramic sealants and interconnect steels in SOFC stacks. J Power Sour 155(2):128–137 Beatty CC (2005) Compliant glass-silver seals for SOFC applications electrochemical society proceedings 2005-07, pp 1949–1955 Bhattacharya S, Shashikala HD (2019) Effect of BaO on thermal and mechanical properties of alkaline earth borosilicate glasses with and without Al2 O3 . Physica B: Condensed Matter 571:76–86. https://doi.org/10.1016/j.physb.2019.06.065 Bram M, Reckers S, Drinovac P, Mnch J, Steinbrech RW, Buchkremer HP, Stver D (2004) Deformation behavior and leakage tests of alternate sealing materials for SOFS stacks. J Power Sour 138:111–119 Cela Greven B, Gross-Barsnick S, Koppitz T, Conradt R, Smeacetto F, Ventrella A, Ferraris M (2018) Torsional shear strength of novel glass-ceramic composite sealants for solid oxide fuel cell stacks Int. J Appl Ceram Technol 15(2):286–295. https://doi.org/10.1111/IJAC.12819 Chou YS, Stevenson JW (2004) Novel infiltrated Phlogopite mica compressive seals for solid oxide fuel cells. J Power Sour 135:72–78 Chou YS, Stevenson JW, Gow NR (2007a) Novel alkaline earth silicate sealing glass for SOFC Part I the effect of nickel oxide on the thermal and mechanical properties. J Power Sour 168:426–433 Chou YS, Stevenson JW, Gow NR (2007b) Novel alkaline earth silicate sealing glass for SOFC part II sealing and interfacial microstructure. J Power Sour 2007(170):395–400 Chou YS, Stevenson JW, Singh P (2007c) Novel refractory alkaline earth silicate sealing glassesfor planar solid oxide fuel cells. J Electrochem Soc 7(154):644–651 Chou Y, Stevenson JW, Singh P (2008a) Effect of pre-oxidation and environmental aging on the seal strength of a novel high-temperature solid oxide fuel cell (SOFC) sealing glass with metallic interconnect. J Power Sour 184(1):238–244. https://doi.org/10.1016/j.jpowsour.2008.06.020 Chou YS, Stevenson JW, Singh P (2008b) Effect of pre-oxidation and environmental aging on the seal strength of a novel high-temperature solid oxide fuel cell (SOFC) sealing glass with metallic interconnect. J Power Sour 184:238–244

200

J. Schilm et al.

Chou Y, Stevenson JW, Choi J-P (2010) Alkali effect on the electrical stability of a solid oxide fuel cell sealing glass. J Electrochem Soc 157(3):B348. https://doi.org/10.1149/1.3275744 Cleek GW, Babcock CI (1973) Properties of glasses in some ternary systems containing BaO and SiO2 . National Bureau of Standards Monograph, vol 135 Corral EL, Gauntt BD, Loehman RE (2008) Properties of particle-filled glass composites used for sealing solid oxide fuel cells. In: Ceramic engineering and science proceedings band 28 (2008) 4, Seite 315–323 (9 Seiten, 11 Quellen), Westerville: American Ceramic Society (ACS) Da Silva MJ, Bartolomé JF, Aza AH de, Mello-Castanho S (2016) Glass ceramic sealants belonging to BAS (BaO–Al2 O3 –SiO2 ) ternary system modified with B2 O3 addition: a different approach to access the SOFC seal issue. J Europ Ceramic Soc 36(3):631–644. https://doi.org/10.1016/j. jeurceramsoc.2015.10.005 Donald IW (1993) Preparation, properties and chemistry of glass- and glass-ceramic-to-metal seals and coatings. J Mater Sci 28(11):2841–2886. https://doi.org/10.1007/BF00354689 Donald IW, Metcalfe BL, Gerrard LA (2008) Interfacial reactions in glass-ceramic-to-metal seals. J Am Ceram Soc 91(3):715–720 Donald IW, Mallison PM, Metcalfe BL, Gerrard LA, Fernie JA (2011) Recent developments in the preparation, characterization and applications of glass and glass-ceramic-to-metal- seals and coatings. J Mater Sci 46:1975–2000. https://doi.org/10.1007/s10853-010-5095-y Eichler K, Solow G, Otschik P, Schaffrath W (2000) Degradation effects at sealing glasses for the SOFC European solid oxide fuel cell. Forum 4:899–906 Faidel D (2011) Laserstrahllöten mit glas-keramischen Zusatzwerkstoffen. Zugl.: Aachen, Techn. Hochs., Dissertion, Aachener Berichte Fügetechnik, 2012, 2. Shaker, Aachen Fakouri Hasanabadi M, Faghihi-Sani MA, Kokabi AH, Malzbender J (2017) The analysis of torsional shear strength test of sealants for solid oxide fuel cells. Ceram Int 43(15):12546–12550. https://doi.org/10.1016/j.ceramint.2017.06.128 Fergus JW (2005) Sealants for solid oxide fuel cells. J Power Sour 147:46–57 Ferraris M, de La Pierre S, Sabato AG, Smeacetto F, Javed H, Walter C, Malzbender J (2020) Torsional shear strength behavior of advanced glass-ceramic sealants for SOFC/SOEC applications. J Eur Ceram Soc 40(12):4067–4075. https://doi.org/10.1016/j.jeurceramsoc.2020. 04.034 Folgner C, Sauchuck V, Kusnezoff M, Michaelis A (2022) Influence of temperature and steam content on degradation of metallic interconnects in reducing atmosphere, pp 17–33. https://doi. org/10.1002/9781119474746.ch3 Frandsen HL, Ritucci I, Khajavi P, Talic B, Kiebach R, Hendriksen PV (2019) Enhancing the robustness of brittle solid oxide cell stack components. ECS Trans 91(1):2201–2211. https:// doi.org/10.1149/09101.2201ecst Frieser RG (1975) A review of solder glasses. Electrocomponent Sci Technol 2:163–199 Gent AN (1960) Theory of the parallel plate viscosimeter. British J Appl Phys 11:85–87 Goel A, Tulyaganov DU, Kharton VV, Yaremchenko AA (2009) Optimization of La2 O3 -containing diopside based glass-ceramic sealants for fuel cell applications. J Power Sour 189:1032–1043 Goel A, Pascual MJ, Ferreira JM (2010) Stable glass-ceramic sealants for solid oxide fuel cells: influence of Bi2 O3 doping. Int J Hydrogen Energy 35(13):6911–6923. https://doi.org/10.1016/ j.ijhydene.2010.04.106 Govindaraju N, Liu WN, Sun X, Singh P, Singh RN (2009) A modeling study on the thermomechanical behavior of glass-ceramic and self-healing glass seals at elevated temperatures. J Power Sour 190(2):476–484 Gross SM, Koppitz T, Reisgen U, Verlotski V, Conradt R (2005) Glass-ceramic Composites as a new sealing material for SOFCs. In: Electrochemical society proceedings 2005-07, pp 1924–1931 Gross SM, Federmann D, Remmel J, Pap M (2011) Reinforced composite sealants for solid oxide fuel cell applications. J Power Sour 196(17):7338–7342. https://doi.org/10.1016/j.jpowsour. 2011.02.002

Glass Ceramic Sealants for Solid Oxide Cells

201

Gueguen Y, Houizot P, Célarié F, Chen M, Hirata A, Tan Y, Allix M, Chenu S, Roux-Langlois C, Rouxel T (2017) Structure and viscosity of phase-separated BaO-SiO2 glasses. J Amer Ceramic Soc 100(5):1982–1993. https://doi.org/10.1111/jace.14642 Guo Q, Feng T, Lance MJ, Unocic KA, Pantelides ST, Lara-Curzio E (2022) Evolution of the structure and chemical composition of the interface between multi-component silicate glasses and yttria-stabilized zirconia after 40,000 h exposure in air at 800 °C. J Europ Ceramic Soc 42(4):1576–1584. https://doi.org/10.1016/j.jeurceramsoc.2021.11.013 Haanappel VAC, Shemet V, Gross SM, Koppitz T, Menzler NH, Zahid M, Quadakkers WJ (2005) Behaviour of various-glass-ceramic sealants with ferritic steels under simulated SOFC stack conditions. J Power Sour 150:86–100 Heinrich JG, Aldinger F (2001) Ceramic materials and components for engines. Wiley-VCH, Weinheim, New York Horita T, Sakai N, Kawada T, Yokokawa H, Dokiya M (1993) Reaction of SOFC components with sealing materials. Denki Kagaku1961. 61(7):760–762. https://doi.org/10.5796/electroch emistry.61.760 Hsueh C-H, Becher PF (1996) Residual thermal stresses in ceramic composites. Part i: with Ellipsoidal Inclusions Mater Sci Eng: A 212(1):22–28. https://doi.org/10.1016/0921-5093(96)101 76-3 Javed H, Sabato AG, Mansourkiaei M, Ferrero D, Santarelli M, Herbrig K, Walter C, Smeacetto F (2020) Glass-ceramic sealants for SOEC: thermal characterization and electrical resistivity in dual atmosphere. Energies 13(14):3682.https://doi.org/10.3390/en13143682 Kilinc E, Hand RJ (2015) Mechanical properties of soda–lime–silica glasses with varying alkaline earth contents. J Non-Crystalline Solids 429:190–197. https://doi.org/10.1016/j.jnoncrysol. 2015.08.013 Kolitsch U, Lengauer CL, Krause W, Bernhardt H-J, Medenbach O, Blaß G (2003) BaZrSi2 O7 , a new mineral from the Eifel volcanic area Germany. Mitt Österr Mineral Ges 148:199–200 Kuang K, Easler K (eds) (2007) Fuel cell electronics packaging. Springer, New York, NY Kubaschewski O (2010) Iron-binary phase diagrams. Springer, Berlin, Heidelberg, New York, Verlag Stahleisen m.b.H.Dsseldorf, pp 172–175 Lahl N, Singh K, Singheiser L, Hilpert K, Bahadur D (2000) Crystallisation kinetics in AO-Al2 O3 SiO2 -B2 O3 glasses (A = Ba, Ca, Mg). J Mater Sci 35(12):3089–3096 Lahl N, Bahadur D, Singh K, Singheiser L, Hilpert K (2002) Chemical interactions between aluminosilicate base sealants and the components on the anode side of solid oxide fuel cells. J Electrochem Soc 149. https://doi.org/10.1149/1.1467945 Lara C, Pascual MJ, Durn A (2004a) Glass-forming ability, sinterability and thermal properties in the systems RO-BaO-SiO2 (R = Mg, Zn). J Non-Crystalline Solids 348:149–155 Lara C, Pascual MJ, Prado MO, Durn A (2004b) Sintering of glasses in the system RO-Al2 O3 BaO-SiO2 (R=Ca, Mg, Zn) studied by hot-stage microscopy. Solid State Ionics 170:201–208 Larsen PH, Poulsen FW, Berg RW (1999) The influence of SiO2 addition to 2MgO-Al2 O3 –3.3P2 O5 glass. J Non-Crystalline Solids Band 244:Heft 1, Seite 16–24 (9 Seiten, 6 Bilder, 6 Tabellen, 21 Quellen) (1999) Le S, Sun K, Zhang N, An M, Zhou D, Zhang J, Li D (2006) Novel compressive seals for solid oxide fuel cells. J Power Sour 161:901–906 Lejeune A-M, Richet P (1995) Rheology of crystal-bearing silicate melts: an experimental study at high viscosities. J Geophys Res 100(B3):4215–4229. https://doi.org/10.1029/94JB02985 Lessing PA (2007) A review of sealing technologies applicable to solid oxide electrolysis cells. J Mater Sci 42:3465–3476 Ley KL, Krumpelt M, Kumar R, Meiser JH, Bloom I (1996) Glass-ceramic sealants for solid oxide fuel cells: part I. Phys Propert J Mater Res 11(6):1489–1493. https://doi.org/10.1557/JMR.1996. 0185 Li R, Tao M, Wang P, Yang J, Ma B, Chi B, Pu J (2021) Effect of interconnect pre-oxidation on high-temperature wettability and mechanical properties of glass seals in SOFC. J Am Ceram Soc 104(12):6172–6182. https://doi.org/10.1111/jace.18024

202

J. Schilm et al.

Li JH, Jiang YL, Wu JW, Pineault R, Gemmenn R, Liu XB (2010) Effect of electrical current on solid oxide fuel cells metallic interconnect oxidation in syngas SO. Int J Appl Ceram Technol 7(1):41–48. https://doi.org/10.1111/j.1744-7402.2009.02430.x Lin C-K, Huang L-H, Chiang L-K, Chyou Y-P (2009) Thermal stress analysis of planar solid oxide fuel cell stacks: effects of sealing design. J Power Sour 192(2):515–524. https://doi.org/10.1016/ j.jpowsour.2009.03.010 Lin C-K, Chen J-Y, Tian J-W, Chiang L-K, Wu S-H (2012) Joint strength of a solid oxide fuel cell glass–ceramic sealant with metallic interconnect. J Power Sour 205:307–317. https://doi.org/ 10.1016/j.jpowsour.2012.01.048 Liu WN, Sun X, Koeppel B, Stephens E, Khaleel MA (2011) Creep behavior of glass/ceramic sealant and its effect on long-term performance of solid oxide fuel cells. Int J Appl Ceram Technol 8(1):49–59. https://doi.org/10.1111/j.1744-7402.2009.02455.x Mahapatra MK, Lu K (2010a) Effect of atmosphere on interconnect-seal glass interaction for solid oxide fuel/electrolyzer cells. J Am Ceram Soc 94(3):875–885 Mahapatra MK, Lu K (2010b) Glass-based seals for solid oxide fuel and electrolyzer cells—a review. Mater Sci Eng: R: Reports 67(5–6):65–85. https://doi.org/10.1016/j.mser.2009.12.002 Mahapatra MK, Lu K (2010) seal glass for solid oxide fuel cells. J Power Sour 195(21):7129–7139 Malzbender J, Steinbrech RW, Singheiser L (2003) Determination of the interfacial fracture energies of cathodes and glass ceramic sealants in a planar solid-oxide fuel cell design. J Mater Res 18(4):929–934. https://doi.org/10.1557/JMR.2003.0127 Malzbender J, Zhao Y (2012) Flexural strength and viscosity of glass ceramic sealants for solid oxide fuel cell stacks. Fuel Cells 12(1):47–53. https://doi.org/10.1002/fuce.201100116 Malzbender J, Toonder J den, Balkenende AR, With G de (2002) Measuring mechanical properties of coatings: a methodology applied to nano-particle-filled sol–gel coatings on glass. Mater Sci Eng: R: Reports 36(2–3):47–103. https://doi.org/10.1016/S0927-796X(01)00040-7 Malzbender J, Mnch J, Steinbrech RW, Koppitz T, Gross SM, Remmel J (2007) Symmetric shear test of glass-ceramic sealants at SOFC operation temperature. J Mater Sci 42:6297–6301 Meinhardt KD, Kim DS, Chou YS, Weil KS (2008) Synthesis and properties of a barium aluminosilicate solid oxide fuel cell glass-ceramic sealant. J Power Sour 182(2):188–196 Menzler NH, Sebold D, Gross SM, Shemet V, Zahid M (2005) Interaction of metallic SOFC interconnect and glass ceramic sealing under various atmospheric conditions at 800 °C. Electrochem Soc Proc 2005–07:1903–1912 Messler RW (2004) Joining of materials and structures: from pragmatic process to enabling technology. Elsevier, Amsterdam, Heidelberg Miao X-Y, Rizvandi OB, Navasa M, Frandsen HL (2021) Modelling of local mechanical failures in solid oxide cell stacks. Appl Energy 293:116901.https://doi.org/10.1016/j.apenergy.2021. 116901 Miletich R, Allan DR, Angel RJ (1997) The synthetic Cr2+ silicates BaCrSi4 O10 and SrCrSi4 O10 : the missing links in the gillespite-type ABSi4O10 series. American Mineralogist 82(697) Mita K, Ikeda T, Maeda M (2001) Phase diagram study of Fe-Zn intermetallics. J Phase Eqilibria 22(2):122–125 Moritz T, Schilm J, Rost A, Schwarzer E, Weingarten S, Scheithauer U, Wätzig K, Wagner D, Michaelis A (2019) Ceramic additive manufacturing methods applied to sintered glass components with novel properties CMT 1(2):111–119.https://doi.org/10.29272/cmt.2019.0008 Nielsen KA, Solvang M, Nielsen SBL, Dinesen AR, Beaff D, Larsen PH (2007) Glass composite seals for SOFC application. J Eur Ceram Soc 27(2–3):1817–1822 Niewolak L, Tietz F, Quadakkers WJ (2016) Interconnects. In: High-temperature solid oxide fuel cells for the 21st century, Elsevier, pp 195–254 Pascual MJ, Durán A (2001) Determination of the viscosity-temperature curve for glasses on the basis of fixed viscosity points determined by hot stage microscopy. Phys Chem Glasses 42(1):61– 66 Pascual MJ, Guillet A, Durn A (2007) Optimization of glass-ceramic sealant compositions in the system MgO-BaO-SiO2 for solid oxide fuel cells (SOFC). J Power Sour 169:40–46

Glass Ceramic Sealants for Solid Oxide Cells

203

Prado MO, Zanotto ED (2002) Glass sintering with concurrent crystallization. C R Chim 5(11):773– 786. https://doi.org/10.1016/S1631-0748(02)01447-9 Quadakkers WJ, Greiner H, Köck W (1994) Metals and alloys for high temperature SOFC applications. Europ Solid Oxide Fuel Cell Forum 535–541 Reddy AA, Tulyaganov DU, Pascual MJ, Kharton VV, Tsipis EV, Kolotygin VA, Ferreira JM (2013) Diopside–Ba disilicate glass–ceramic sealants for SOFCs: enhanced adhesion and thermal stability by Sr for Ca substitution. Int J Hydrogen Energy 38(7):3073–3086. https://doi.org/ 10.1016/j.ijhydene.2012.12.074 Ritucci I, Kiebach R, Talic B, Han L, Zielke P, Hendriksen PV, Frandsen HL (2019) Improving the interface adherence at sealings in solid oxide cell stacks. J Mater Res 34(07):1167–1178. https://doi.org/10.1557/jmr.2018.459 Rodríguez-López S, Haanappel V, Durán A, Muñoz F, Mather GC, Pascual MJ, Gross-Barsnick SM (2016) Glass–ceramic seals in the system MgO BaOB2 O3 SiO2 operating under simulated SOFC conditions. Int J Hydrogen Energy 41(34):15335–15345. https://doi.org/10.1016/j.ijh ydene.2016.07.051 Rost A, Schilm J, Kusnezoff M (2009) Influence of electrical load on the stability of glass sealings. ECS Trans 25(2):1509–1518 Rost A, Schilm J, Kusnezoff M, Michaelis A (2012) Degradation of sealing glasses for SOFC under electrical load and dual atmosphere. J Ceramic Sci Technol 3(2):69–80. https://doi.org/10.4416/ JCST2012-00002 Rost A (ed) (2013) Competence in ceramics: Entwicklung degradationsstabiler Glaslote für keramische Hochtemperaturbrennstoffzellen, vol 17. FRAUNHOFER VERLAG, Stuttgart Sabato AG, Rost A, Schilm J, Kusnezoff M, Salvo M, Chrysanthou A, Smeacetto F (2019) Effect of electric load and dual atmosphere on the properties of an alkali containing diopside-based glass sealant for solid oxide cells. J Power Sour 415:15–24. https://doi.org/10.1016/j.jpowsour. 2019.01.051 Sarkardeh FB, Meschke F, Hilpert K, Singheiser L (eds) (1999) Berichte des Forschungszentrums Jülich: Stabilität und Kompatibilität des keramischen Interkonnektors der HochtemperaturBrennstoffzelle SOFC, Jülich Sauchuck V, Kusnezoff M, Trofimenko N, Megel S, Baldus H-P, Reinert A (2008) Development of effective protective materials for SOFC metallic interconnects. In: 8th European solid oxide fuel cell forum, vol 09 Sauchuck V, Trofimenko N, Megel S, Rothe S, Schilm J, Andritschky M, Hiller M, Goebel C, Froitzheim J, Kusnezoff M (2020) Application of composite coatings as protection/contacting layers for metallic highchromium-content SOFC interconnect material. In: 14th European SOFC & SOE Forum, pp 601–611 Schilling C, Roosen A (2012) Fügen von keramischen Komponenten der HochtemperaturBrennstoffzellen mittels Glas-und Glaskeramikloten. Friedrich Alexander University ErlangenNuremberg, PdD Schilm J, Rost A, Kusnezoff M, Michaelis A (2010) Sealing glasses for SOFC—degradation behaviour. In: Advances in solid oxide fuel cells V, Wiley, Ltd, pp 183–193 Schilm J, Rost A, Poenicke A, Kusnezoff M, Michaelis A (2012) Ceramic integration technologies for solid oxide fuel cells. Int J Appl Ceram Technol 9(4):688–699. https://doi.org/10.1111/j. 1744-7402.2012.02789.x Schilm J, Rost A, Kusnezoff M, Megel S, Michaelis A (2018) Glass ceramics sealants for SOFC interconnects based on a high chromium sinter alloy Int. J Appl Ceram Technol 15(2):239–254. https://doi.org/10.1111/ijac.12811 Scholze H (1962) Der Einfluss Von Viskositaet Und Oberflchenspannung Auf Erhitzungsmikroskopische Messungen an Glaesern Berichte Der Deutschen Keramischen Gesellschaft 59:63–68 Scholze H (1977) Glas, natur, struktur und eigenschaften, Zweite, neubearbeitete. Springer, Berlin Heidelberg New York

204

J. Schilm et al.

Selçuk A, Atkinson A (1997) Elastic properties of ceramic oxides used in solid oxide fuel cells (SOFC). J Europ Ceramic Soc 17(12):1523–1532. https://doi.org/10.1016/S0955-2219(96)002 47-6 Selsing J (1961) Internal stresses in ceramics. J Am Ceram Soc 44(8):419. https://doi.org/10.1111/ j.1151-2916.1961.tb15475.x Serbena FC, Soares VO, Peitl O, Pinto H, Muccillo R, Zanotto ED (2011) Internal residual stresses in sintered and commercial low expansion Li2 O-Al2 O3 -SiO2 glass-ceramics. J Amer Ceramic Soc 94(4):1206–1214. https://doi.org/10.1111/j.1551-2916.2010.04220.x Shabanova GN, Taranenkova VV, Korogodskaya AN, Khristich EV (2003) Structure of the BaOAl2 O3 -SiO2 system (a review). Glass and Ceramics 60(1–2):43–46 Shelby JE (1979) Effect of morphology on the properties of alkaline earth silicate glasses. J Appl Phys 50(12):8010–8015. https://doi.org/10.1063/1.325986 Singh K, Walia T (2021) Review on silicate and borosilicate-based glass sealants and their interaction with components of solid oxide fuel cell. Int J Energy Res 45(15):20559–20582. https://doi.org/ 10.1002/er.7161 Smeacetto F, Salvo M, Ferraris M, Casalegno V, Asinari P, Chrysanthou A (2008a5a) Characterization and performance of glass-ceramic sealant to join metallic interconnects to YSZ and anode-supported-electrolyte in planar SOFCs. J Europ Ceramic Soc 28:2521–2527 Smeacetto F, Salvo M, Ferraris M, Casalegno V, Asinari P, Chrysanthou A (2008b) Characterization and performance of glass–ceramic sealant to join metallic interconnects to YSZ and anodesupported-electrolyte in planar SOFCs. J Europ Ceramic Soc 28(13):2521–2527. https://doi. org/10.1016/j.jeurceramsoc.2008.03.035 Smeacetto F, Salvo M, Ferraris M, Casalegno V, Asinari P (2008c) Glass and composite seals for the joining of YSZ to metallic interconnect in solid oxide fuel cells. J Eur Ceram Soc 28(3):611–616 Sohn S-B, Choi S-Y, Kim G-H, Song H-S, Kim G-D (2002) Stable sealing glass for planar solid oxide fuel cell. J Non-Cryst Solids 297(2–3):103–112. https://doi.org/10.1016/S0022-3093(01)010 42-0 Sohn SB, Choi SY, Kim GH, Song HS, Kim GD (2004) Suitable glass-ceramic sealant for planar solid-oxide fuel cells. J Am Ceram Soc 87(2):254–260 Sumi H, Shimada H, Yamaguchi T, Fujishiro Y, Awano M (2019) Development of portable solid oxide fuel cell system driven by hydrocarbon and alcohol fuels. In: Salem J, Koch D, Mechnich P, Kusnezoff M, Bansal N, LaSalvia J, Balaya P, Fu Z, Ohji T (eds) Proceeding of the 42nd international conference on advanced ceramics and composites. Ceramic engineering and science proceedings, Wiley, Inc, Hoboken, NJ, USA, pp 159–163 Sun T, Xiao H, Guo W, Hong X (2010) Effect of Al2 O3 content on BaO–Al2 O3 –B2 O3 –SiO2 glass sealant for solid oxide fuel cell. Ceramics Int 36(2):821–826. https://doi.org/10.1016/j.ceramint. 2009.09.045 Theerapapvisetpong A, Jiemsirilers S, Thavorniti P, Conradt R (2011) Barium-free glass-ceramic sealants from the system CaO-MgO-B2 O3 -Al2 O3 -SiO2 for application in the SOFC. MSF 695:1–4. https://doi.org/10.4028/www.scientific.net/MSF.695.1 Thomas RH (1950) Phase equilibrium in a portion of the ternary system Ba OA-I2,O3,-SiO2 . J Amer Ceramic Soc 33(2):35–44. https://doi.org/10.1111/j.1151-2916.1950.tb14163.x Venskutonis A, Brandner M, Kraussler W, Sigl L (2009) High volume fabrication of ready-to-stack components for planar SOFC concepts ECS. Trans 25(2):1353–1359. https://doi.org/10.1149/ 1.3205664 Wanko E, Gross SM, Koppitz T, Remmel J, Reisgen U, Wilsmann W, Conradt R (eds) (2007) Glaskeramische Materialien des Systems BaO-MgO-SiO2 zum Lten der HochtemperaturBrennstoffzelle (SOFC). In: Disserion. Brazing, High Temperature Brazing and Diffusion Welding, Aachen Wanko E, Gross SM, Remmel J, Reisgen U, Conradt R (2008a) Composits from the system BaOMgO-SiO2 as sealing material for the high temperature fuel cell (SOFC) DVS-Berichte(250), 474–477 (2008a)

Glass Ceramic Sealants for Solid Oxide Cells

205

Wanko E, Gross SM, Remmel J, Reisgen U, Conradt R (eds) (2008b) Verbundwerkstoffe aus dem System BaO-MgO-SiO2 als Fügematerial für die Hochtemperaturbrennstoffzelle (SOFC). In: Dissertion Groe Schweitechnische Tagung, Dresden Weil KS, Deibler JE, Hardy JS, Kim DS, Xia GG, Chick LA, Coyle CA (2004) Rupture testing as a tool for developing planar solid oxide fuel cell seals. J Mater Eng Perform 13(3):316–326. https://doi.org/10.1361/10599490419306 Wenning NL, Xin S, Brian K, Mohammad K (2010) Experimental study of the aging and selfhealing of the glass/ceramic sealant used in SOFCs SO. Int J Appl Ceram Technol 7(1):22–29. https://doi.org/10.1111/j.1744-7402.2009.02417.x Wiener F, Bram M, Sebold D (2007) Chemical interaction between Crofer 22 APU and mica-based gaskets under simulated SOFC conditions. J Mater Sci 42(8):2643–2651 Yang Z, Meinhardt KD, Stevenson JW (2003) Chemical compatibility of barium-calciumaluminosilicate-based sealing glasses with the ferritic stainless steel interconnect in SOFCs. J Electrochem Soc 150(8):1095–1101. https://doi.org/10.1149/1.1590325 Yang Z, Xia GG, Meinhardt J, Weil KS, Stevenson JW (2005) Glass sealing in planar SOFC stacks and chemical stability of seal interfaces. Ceram Trans 158:135–146 Yeong SC, Stevenson JW, Chick LA (2003) Novel compressive mica seals with metallic interlayers for solid oxide fuel cell applications. J Am Ceram Soc 86(6):1003–1007 Zhang T, Fahrenholtz WG, Reis ST, Brow RK (2008) Borate volatility from SOFC sealing glasses. J Am Ceram Soc 91(8):2564–2569. https://doi.org/10.1111/j.1551-2916.2008.02479.x Zhao Y, Malzbender J, Gross SM (2011) The effect of room temperature and high temperature exposure on the elastic modulus, hardness and fracture toughness of glass ceramic sealants for solid oxide fuel cells. J Europ Ceramic Soc 31(4):541–548. https://doi.org/10.1016/j.jeurceram soc.2010.10.032 Zhang T, Brow RK, Fahrenholtz WG, Reis ST (2012) Chromate formation at the interface between a solid oxide fuel cell sealing glass and interconnect alloy. J Power Sour 205:301–306 Zhenguo Y, Stevenson JW, Meinhardt KD (2003) Chemical interactions of barium-calciumaluminosilicate-based sealing glasses with oxidation resistant alloys. Solid State Ionics 160:213–225

Modeling of Solid Oxide Electrolysis Cells Yang Wang, Chengru Wu, Kui Jiao, Qing Du, and Meng Ni

Abstract The practical operation of solid oxide electrolysis cell (SOEC) involves complex physicochemical coupling processes between “multi-physics fields” at “multiple scales”. Mathematical simulation and modeling can explain the inherent connections and influencing mechanisms of multi-physics fields at different scales, which are crucial for the study of SOEC’s basic electrochemical characteristics and the development of engineering applications. In this chapter, we mainly summarize different simulation techniques for SOEC from the perspective of spatial scale categories. Models related to single cells and stacks are mainly based on the continuum hypothesis, and the macroscopic characteristics such as the distribution of multiphysics fields, input/output power, and cell efficiency inside single cells/stacks are obtained through traditional computational fluid dynamics using finite volume method or finite element method. This article first introduces the relevant conservation equations and modeling methods of macroscopic models based on the continuum hypothesis. Then, numerical simulation methods for heterogeneous electrode structures at the electrode scale are introduced, including the lattice Boltzmann method, kinetic Monte Carlo method, and phase field method. Finally, we also introduce the application of machine learning methods in SOEC simulation and provide prospects for future research.

Nomenclature C Ci dp

Conserved order parameters Molar concentration of species i (mol m-3 ) Average diameter of pore (µm)

Y. Wang · M. Ni (B) Department of Building and Real Estate, Research Institute for Sustainable Urban Development (RISUD) and Research Institute for Smart Energy (RISE), The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong, China e-mail: [email protected] C. Wu · K. Jiao · Q. Du State Key Laboratory of Engines, Tianjin University, Tianjin, China © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M. A. Laguna-Bercero (ed.), High Temperature Electrolysis, Lecture Notes in Energy 95, https://doi.org/10.1007/978-3-031-22508-6_8

207

208

Di,k Di,j E act F F total f0 J j0 K K pr K ps k rf k sf M MC ne P p rp T Vi W ca

Y. Wang et al.

Knudsen diffusion coefficients (m2 s-1 ) Binary diffusion coefficients (m2 s-1 ) Activation energy (J mol-1 ) The Faraday constant (C mol-1 ) Total free energy of the system Bulk free energy density Electrochemical reaction rate (A m-3 s-1 ) Exchange current density (A m-3 ) Permeability (m2 ) Equilibrium constant of MSR Equilibrium constant of WGSR Forward reaction rate constant for MSR Forward reaction rate constant for WGSR Average molar mass (kg mol-1 ) Mobility of conserved order parameters Number of electrons transferred per reaction Pressure (Pa) Partial pressure (Pa) Pore diameter (m) Temperature (K) Special Fuller diffusion volume (cm-3 mol-1 ) Wettability parameter

Greek Letters ϕ ηi ηact κ κ eff σ ρ μ ε ωi τ τi

Electric potential (V) Non-conserved order parameters Activation overpotential Gradient energy coefficients Effective thermal conductivity (W m-1 K-1 ) Conductivity (S m-1 ) Density (kg m-3 ) Dynamic viscosity (N s m-2 ) Porosity Weight coefficients Tortuosity Relaxation coefficients

Subscripts and Superscripts act bulk

Activation Bulk diffusion

Modeling of Solid Oxide Electrolysis Cells

conc ele ion N ohm Y P

209

Concentration Electron Ion Ni Ohmic YSZ Pore

1 Working Principle The core of a high-temperature solid oxide electrolysis cell (SOEC) is a sandwichshaped positive-electrolyte-negative (PEN). Figure 1a shows the geometry of a planar SOEC, where the PEN is composed of two porous electrodes sandwiched by a dense electrolyte layer (Wang et al. 2020a). The PEN structure is divided into anode diffusion layer (ADL), anode functional layer (AFL), dense electrolyte, cathode functional layer (CFL), and cathode diffusion layer (CDL) from top to bottom. The dense electrolyte layer provides a pathway for the transport of oxygen ions or protons and serves to separate the oxygen and fuel gases from each other. The cathode (fuel electrode) and anode (air electrode) are usually porous layers prepared from two types of ion-conducting and electron-conducting materials. The interface among the two materials and the gas phase constitutes the three-phase boundaries (TPBs), which are active sites of the electrochemical reactions. For oxygen ion-conducting SOEC (OSOEC), the reactants at the fuel electrode (H2 O and CO2 ) diffuse into the porous electrode from the gas channel and undergo electrochemical reactions at TPBs to produce fuels (H2 and CO) and generate oxygen ions. Subsequently, the oxygen ions pass through the dense electrolyte to produce oxygen at the electrochemical reaction sites of the oxygen electrode. As for the proton-conducting SOEC (PSOEC), the incoming steam generates protons and oxygen at the oxygen electrode through electrochemical reactions. And the protons transfer across the electrolyte layer to produce hydrogen directly at the fuel electrode or react with the incoming N2 and NO to generate the mixture of N2 , N2 O, and NH3 . The abovementioned planar SOEC is the most common structure, which has the advantages of low cost, easy assembly, short current path, and high output power. At the same time, the planar configuration suffers from difficult sealing and poor thermal cycling stability. For comparison, the tubular SOEC (Fig. 1b) can well solve the sealing problem well and makes it easy to expand the size of the stack, but its long current path leads to its lower power density (Luo et al. 2015). To gain a fundamental understanding on the transport and reaction processes in the SOEC, models at different levels are available for simulation and optimization.

210

Y. Wang et al.

Fig. 1 Schematics of a planar cathode-supported SOEC (Wang et al. 2020a), and b tubular SOEC (Luo et al. 2015)

2 Macroscale Modeling Usually, the macroscale model of SOFC includes a series of multi-physics coupling processes such as gas flow, heat and mass transfer, chemistry, and electrochemistry. The macroscale modeling can well describe the above multi-physics fields distribution and evaluate the effects of different parameters on the cell performance. The single-cell macroscopic model is the basis for SOEC multi-scale simulations. In this section, the building process of a single-cell model will be detailed based on the required governing equations.

Modeling of Solid Oxide Electrolysis Cells

211

2.1 Momentum Transfer The conservation of momentum within the flow channel and porous electrodes is usually characterized by the Navier–Stokes (NS) equation and Darcy’s law, respectively. In order to avoid the exchange of pressure and velocity data at the interface between the flow channel and the porous electrode, the Brinkman equation can be used to characterize the momentum conservation. The Brinkman equation adds a viscous term compared to Darcy’s law and simplifies the inertial term compared to the NS equation (Durlofsky and Brady 1987). The momentum transfer equation can be written as follows:       2 μ u 1  ρ ∂u T + (u · ∇) = −∇ p + ∇ · μ ∇u + (∇u) − μ(∇ · u) − u ε ∂t ε ε 3 K (1) where ρ, u, p, and μ are the density, velocity vector, pressure, and dynamic viscosity, respectively. The values of porosity (ε) and permeability (K) are set as 1.0 and ∞ in gas channels, respectively.

2.2 Multicomponent Mass Transfer The generalized mass balance equation of species i can be expressed as:   ∇ · ρavg uyi = ∇ Ji + Si

(2)

where ρ avg (kg m−3 ) is the average density of the gas mixture. yi is the mass fraction of species i. J i (kg m−2 s−1 ) is the mass diffusion flux. And S i is the source term. Generally speaking, the gas components for which the above equation is solved in the SOEC fuel electrode include CO, CO2 , H2, and H2 O. The species composition, temperature, pressure, and microstructure can have a great effect on the multicomponent transfer process inside the SOEC porous electrode. There are three main mechanisms for the transport of gas components inside porous electrodes: convection, molecular diffusion, and Knudsen diffusion (Wang et al. 2018). The molecular diffusion occurs due to collisions between gas molecules, and Knudsen diffusion takes place when gas molecules collide with the pore walls frequently. Typically, three models are used to analyze the species transfer within the SOEC porous electrode, namely the Fick’s law, the Stefan–Maxwell model (SMM), and the dust gas model (DGM). Among them, Fick’s law is the simplest model, which takes into account molecular diffusion in binary and dilutes component systems and can be combined with the Knudsen diffusion in pores to correct the effective diffusion coefficient. The SMM ignores the effects of Knudsen diffusion and pressure gradient and thus does not accurately reflect the multicomponent mass transfer within the

212

Y. Wang et al.

SOEC porous electrode. Compared with the first two models, the DGM takes into account both molecular diffusion and Knudsen diffusion as well as the viscous flow, thus it can accurately describe the multicomponent diffusion mechanism inside the SOEC porous electrode (Yuan and Sunden 2014). The one-dimensional DGM can be expressed as: Ni ef f

Di,k

+

n

x j Ni − x i N j ef f

j=1, j/=i

Di j

1 =− RT



 dp d xi Kp + yi p 1 + ef f dz dz Di,k μ

(3)

where N i is the molar diffusion flux (mol m−2 s−1 ), x i is the molar fraction of species ef f ef f i, Di j (m2 s−1 ) is the effective binary diffusion coefficient, and Di,k (m2 s−1 ) is the effective Knudsen diffusion coefficient. The tortuosity (τ ) and porosity are used to correct the binary and the Knudsen diffusion coefficient, which can be expressed as: ef f

ε Di,k τ ε = Di j τ

Di,k = ef f

Di j

(4)

The binary and the Knudsen diffusion coefficients can be calculated by the following equations, respectively: /   3.1976 × 10−4 T 1.75 (1/Mi ) + 1/M j Di j = 2  1/3 1/3 p VF,i + VF, j / 2r p 8RT Di,k = 3 π Mi

(5)

where T (K) is the temperature, M i (kg mol−1 ) is the molecular weight of species i, V F (cm−3 mol−1 ) is the specific Fuller diffusion volume, and r p (µm) is the average pore diameter. The simplest model describing the multicomponent diffusion, Fick’s law, can be formulated as the following equation: ef f

Ni = −Di

p ∇xi + cxi (u + v) RT

(6)

The second term on the right-hand side of the equation represents convective flow and includes the mixture flow velocity (u) as well as the viscous flow velocity (v). And the effective diffusion coefficient is a combined form of the binary and Knudsen diffusion coefficients, and can be evaluated as:

Modeling of Solid Oxide Electrolysis Cells

1 Di,e f f

=

213

n

x j − xi βi j j=1, j/=i

ef f Di j

+

1 ef f

Di,k

(7)

where βi j = N j /Ni .

2.3 Heat Transfer and Thermal Effect Based on theoretical and numerical studies on solid oxide fuel cells, the heat radiation effect can be safely neglected (Damm and Fedorov 2005). Thus, the heat transfer phenomena in the SOEC can be well described by the heat convection and conduction processes, which can be described by the following equation:      ∂ ερc p T + ∇ · ερc p uT = ∇ · κe f f ∇T + ST ∂t

(8)

where cp (J K−1 ) represents the specific heat capacity. The effective thermal conductivity (κ eff , W m−1 K−1 ) of the porous electrodes can be estimated by conducting a volume average of thermal conductivities of solid phase and gas mixture: κe f f = (1 − ε)κs + εκ f

(9)

The heat source term (S T , W m−3 s−1 ) includes several heat generation and consumption processes and can be expressed as the sum of these terms: ST = Sir e + Sr e + Sche + Sohm

(10)

where S ire , S re , S che , S ohm , respectively, represent the irreversible heat, reversible heat, chemical heat, and ohmic heat. When the electrochemical reaction occurs, the reversible heat is generated:  T △Sa =J ne F   T △Sc =J ne F 

Sr ev,a Sr ev,c

(11)

where J (A m−3 s−1 ) is the local electrochemical reaction rate, ne is the number of electrons transferred per reaction, F is the Faraday constant, △S (J K−1 ) represents the entropy change of the electrochemical reaction. Apart from the heat generation due to entropy change, irreversible heat generation due to polarization loss and ohmic loss should be considered as well. In order to overcome the activation barrier, part of the potential will be sacrificed. The irreversible heat is obtained by calculating the product of sacrificed potential (activation

214

Y. Wang et al.

overpotential) and electrochemical reaction rate: Sir ev,a = J ηact,a Sir ev,c = J ηact,c

(12)

where ηact (V ) represents the activation overpotential of the cell electrode. In addition, the chemical reactions and electron/ion transfer will also contribute to the total heat source term. The chemical reactions will be detailed discussed in the next section. The ohmic heat can be calculated from: Sohm =

I2 Ie2 + i σe σi

(13)

where I e and I i (A m−2 ) are the electronic current density and ionic current density, respectively. σ e and σ i (S m−1 ) represent the electron and ion conductivity of electrodes, respectively. For the high-temperature SOEC, the co-electrolysis process is endothermic. There will be a thermoneutral state in which the heat generation from the SOEC is equal to the heat absorption for electrolysis reactions. In terms of the whole system including the electrolysis cell, a thermoneutral state (or slightly above) is beneficial for maintaining high system efficiency and reducing the possibility of thermal fatigue. A large part of SOEC’s application scenarios is in the renewable energy industry. The intermittent supply of renewable energy places high demands on SOEC in terms of thermal cycle loading. Simulation on the thermal management of SOEC, especially in terms of mechanical thermal stress, is also a major tool to improve the performance of SOEC. Figure 2 shows the comparisons of heat transfer rates and heat source terms of two different types of the fluid domain at the cathode side (Wang et al. 2020c). It can be seen that the electrolysis cell is in an endothermic state when the current is below the operating current of the thermoneutral voltage (TNV), and vice versa in an exothermic state. In addition, reversible heat plays a dominant role in the overall cell heat generation and absorption.

2.4 Chemical Reactions One of the advantages of high-temperature SOEC is the co-electrolysis of H2 O and CO2 , and a substantial process has been made in this area over the past decades. The reduction reactions of H2 O and CO2 occur in the electrochemical reaction sites (TPBs). With the existence of Ni, the reactant gas of electrochemical reactions will undergo several chemical reactions, and this is also the pathway of syngas production inside SOEC. The most common chemical reactions are the water gas shift reaction (WGSR), reverse methane steam reforming reaction (MSR) or methanation, and reverse dry reforming reaction (DR), which can be expressed as:

Modeling of Solid Oxide Electrolysis Cells

215

Fig. 2 Schematics of a two different types of fuel electrode-supported SOEC and comparisons between b the heat transfer rates and c the heat source terms (Wang et al. 2020c)

216

Y. Wang et al.

WGSR: CO + H2 O ↔ CO2 + H2 MSR: CH4 + H2 O ↔ CO + 3H2 DR: CH4 + CO2 ↔ 2CO + 2H2

(14)

The Langmuir–Hinshelwood–Hougen–Watson (LHHW) reaction rate equations are commonly used as the expressed the kinetic reaction rates of the above reactions, which can be described as:   pCO2 pH2 (15) RW G S R = ks f pCO pH2 O − K ps  3 pCO pH2 R M S R = kr f pCH4 pH2 O − (16) K pr kCO2 K CO2 pCO2 pCH4  RD R =  1 + K CO2 pCO2 + K CH4 pCH4

(17)

where the RWGSR , RMSR , RDR (mol m−3 s−1 ) are the reaction rates of WGSR, MSR, DR, respectively. The forward rate constants and the equilibrium constants (k sf , k rf , K ps , K pr , kCO2 , K CO2 , K CH4 ) are temperature-dependent functions calculated by fitting experimental data. In addition to syngas production, SOEC can also provide an eco-friendly solution to ammonia production (Li et al. 2021). When the humidified nitrogen is passed through the electrode, the H2 O is firstly split into oxygen ions and protons. The protons will react with nitrogen to produce ammonia. The whole process inside the fuel electrode can be summarized as: H2 O → O2− + 2H+ N2 + 6H+ + 6e− → 2NH3

(18)

But, when H2 O is used as the source of protons, the above ammonia synthesis reaction will be sluggish. Therefore, today’s research on ammonia synthesis using SOEC is mainly focused on the development of advanced electrocatalysts. However, it should be noted that future simulation work is also urgently needed to optimize the ammonia-synthesis SOEC.

2.5 Electrochemistry As mentioned above, the electrochemical reactions inside fuel and oxygen electrodes occur on the surfaces of TPBs. The cell input voltage (V in ) can be calculated by deducting each voltage loss from the open circuit voltage (E OCV ):

Modeling of Solid Oxide Electrolysis Cells

217

Vin = E OC V − ηact,a − ηact,c − ηconc,a − ηconc,c ηohm

(19)

where ηact , ηconc , and ηohm (V) are the concentration loss, activation loss, and ohmic loss, respectively. The open circuit voltage can be calculated by the Nernst equation which relates the reduction potential of an electrochemical reaction to the standard electrode potential, temperature, and concentration approximated activities of species undergoing reduction and oxidation. In the fuel and oxygen electrodes of SOEC, the Nernst equations of hydrogen and carbon monoxide can be expressed as the following equations, respectively. E OCV,H2

E OCV,CO

  0.5  pH2 pO2 △G RT ln =− + ne ne F pH2 O   0.5  pCO pO2 RT △G ln + =− ne ne F pCO2

(20)

(21)

The activation loss is often identified by the Butler–Volmer (B-V) equation:      n F  ne F e H2 O CO2 Jc = j0,c + j0,c ηact,c − exp −(1 − α) ηact,c exp α RT RT      ne F ne F ηact,a − exp −(1 − β) ηact,a Ja = j0,a exp β RT RT

(22) (23)

where j0 , α, and β are the reference exchange current density, transfer coefficient in cathode and anode, respectively. For the co-electrolysis of SOEC, the CO2 reduction reaction is usually sluggish than the rate of the H2 O reduction reaction. The value of H2 O CO2 j0,c is usually set as several times larger than that of j0,c . The concentration overpotential can be calculated as: ηconc,c

bulk r ea

pCO pCO2 pHbulk pHr ea RT RT 2 2O ln ln = + bulk r ea bulk r ea 2F 2F pH2 O pH2 pCO pCO 2

pObulk RT 2 ln ηconc,a = 4F pOr ea 2

(24)

(25)

where pibulk and pir ea (Pa) are the partial pressure of species i in the gas channel and reaction zone. The ohmic loss is determined by Ohm’s law: ηohm =

J ·l σe f f

(26)

218

Y. Wang et al.

where l (m) is the thickness of the related cell component. In the porous electrodes, the effective electronic and ionic conductivities (σ eff , S m−1 ) are obtained considering the tortuosity, phase volume fraction, and the intrinsic conductivity (σ 0 , S m−1 ). σe f f =

(1 − ε) σ0 τ

(27)

Meanwhile, the electron and ion transfer satisfy the following conservation equations:   ef f 0 = ∇ · σele ∇ϕele + Jele

(28)

  ef f 0 = ∇ · σion ∇ϕion + Jion

(29)

In Eqs. (28) and (29), the electric potential (ϕ ele , V) and ionic potential (ϕ ion , V) are used to calculate the activation overpotential: ηact,a = ϕele − ϕion − E OC V ,a

(30)

ηact,c = ϕele − ϕion − E OC V ,c

(31)

2.6 Current Leakage of PSOEC Compared with the OSOEC, the PSOEC has some unique advantages, such as lower operating temperature (773 K-973 K) and lower activation energy, etc. Figure 3 shows the schematic of the working principle for PSOEC (Bi et al. 2014). The macroscopic modeling of proton-conducting SOEC is the same as that of ion-conducting SOEC. However, it is worth noting that the electrolyte materials (e.g., BCZY) of PSOEC are usually mixed conductors, which have a certain conductivity for electron holes while conducting protons. At high oxygen partial pressures, oxygen molecules will combine with oxygen vacancies on the electrolyte thereby creating electron holes, and these electron holes can combine with electrons: 1 O2 + Vo·· ↔ OO + 2h ∗ 2 h ∗ + e− → 0

(32)

This phenomenon leads to current leakage, which means that while some of the electrons entering the cell react with protons to produce hydrogen, others combine with electron holes, leading to a decrease in electrolysis efficiency. The Faraday efficiency (F eff ) is defined as the ratio of the absolute proton current (JH+ , A m−2 )

Modeling of Solid Oxide Electrolysis Cells

219

Fig. 3 Schematics of the working principle of PSOEC (Bi et al. 2014)

and the external input current (Jext , A m−2 ): Fe f f =

JH+ Jext − Jh ∗ = Jext Jext

(33)

where Jh ∗ (A m−2 ) is the current of electron holes. Typically, the Nernst–Planck equation can be used to identify the mixed transfer process of protons and electron holes: Jk = −D∇ck −

σ ∇φ zk q

(34)

where J k is the molar fluxes of protons and electron holes, ck is the mobile ions density, zk is the valence of charges, q is the elementary charge, and ϕ is the electronic potential. The Nernst–Planck equation is used to describe the motion of a charged chemical in a fluid medium. It is essentially a mass conservation equation that extends Fick’s law to the case where the diffusing particles also move relative to the fluid under electrostatic forces. There have been some modeling studies on current leakage in proton-conducting SOFC (Zhang et al. 2018, 2019), as shown in Fig. 4, which illustrates the transport inside the mixed conductor of proton-conducting SOFC. But, there is still a gap in simulation studies on current leakage in PSOEC. An analytical solution based on the Nernst–Planck equation has been derived to describe the relationship between the electron holes current and the cell output current in PSOFC, which can be used as a reference for calculating the leakage current in PSOEC (Zhang et al. 2018):  Jh ∗ =

   pHa 2 RT τele τele τele ln c − Jext + / 2F pH2 σH+ σh ∗ σH+

(35)

220

Y. Wang et al.

Fig. 4 Schematics of protons and electron holes transport processes inside proton-conducting SOFC (Zhang et al. 2019)

where σH+ (S m−1 ) and σh ∗ (S m−1 ) are the conductivities of protons and electron holes, respectively.

3 Powder-to-Power: Full Life Cycle Modeling of SOEC Electrode From preparation to operation, SOEC electrodes are accompanied by several complex physical processes, such as powder sintering during electrode preparation, fuel electrode reduction before the operation, and changes in electrode micromorphology during long-term operation (Zheng et al. 2017). Taking the Ni–YSZ fuel electrode as an example, the actual electrode fabrication consists of the following processes: Ni is first present in the electrode preparation in the form of NiO. The ball milling powders of NiO and YSZ (morphologically nearly spherical) are combined with the binder to form the desired shape and then burned at high temperatures after drying. Generally, the slurry is partially filled with the pore former (e.g., graphite), which serves to be burned off during the high-temperature sintering to form larger pore space. Before the sintered electrode can operate, the NiO needs to be reduced to Ni, and since NiO will occupy more space compared to Ni, this process will also increase the porosity. Figure 5 shows the manufacturing process of Ni–YSZ substrate utilizing Coat-Mix process with subsequent thermal pressing (Birnbaum et al. 2013).

Modeling of Solid Oxide Electrolysis Cells

221

Then, the Ni phase will migrate during the actual operation of the electrolysis cell, resulting in degradation of performance due to phenomena such as Ni coarsening and depletion. With the enrichment and development of simulation methods in recent years, it has been possible to realize the whole life cycle simulation of electrodes from powder to electrochemical performance. These simulation methods include, are not limited to, heterogeneous electrode microstructure reconstruction methods, dynamic Monte Carlo (DMC) method, phase field method (PFM), and lattice Boltzmann method (LBM). Compared with the above macroscopic model, we can visualize the influence of electrode microstructure on the electrochemical performance output and long-term stability, so that we can optimize the heterogeneous electrodes from preparation to operation. This section details the full life cycle modeling process for powder-topower electrodes. Figure 6 shows the heterogeneous electrode simulation flow from electrode reconstruction to electrochemical performance evaluation (Yan et al. 2019). The microstructure reconstruction methods of SOEC heterogeneous electrode mainly include experimental methods such as focused ion beam-scanning electron microscopy (FIB-SEM) (Prokop et al. 2018) and transmission X-ray microscopy (TXM) (Nelson et al. 2012), random reconstruction methods (Hasanabadi et al. 2016; Riazat et al. 2017; Tafazoli et al. 2017), the discrete element method (DEM)based random generation methods (Sanyal et al. 2010), and particle packing methods (Zheng and Ni 2016). Here, we mainly describe the microstructure reconstruction approaches based on numerical methods. The reconstruction of the electrodes can first be achieved by a random numerical method. Hasanabadi et al. (2016) used the decay exponent multiplied by a sinusoidal

Fig. 5 Manufacturing the Ni–YSZ substrate utilizing the Coat-Mix® process with subsequent thermal pressing (Birnbaum et al. 2013)

222

Y. Wang et al.

Fig. 6 Heterogeneous electrode stimulation flow from electrode reconstruction to electrochemical performance evaluation (Yan et al. 2019)

function to initialize a complete set of two-point correlation function (TPCF) that was used to characterize a unique three-dimensional electrode microstructure. Riazat et al. (2017) and Tafazoli et al. (2017) constructed electrodes based on the above approach and through Monte Carlo simulations and combined artificial intelligence methods and genetic algorithms to obtain optimal microstructure of the correlation scheme. The discrete element method predicts the evolution of multiple particle trajectories by solving a set of ordinary differential equations that take into account interparticle collisions and other relevant forces. The general steps of the DEM are as follows: The first step is to define the computational domain. Typically for SOEC electrodes, the computational domain length and width vary from a few microns to tens of microns. And the height is generally slightly larger than the length and width, in order to better predict the drop trajectory and stacking process. Then, spherical particles (electron-conducting and ion-conducting particles in SOEC electrodes) with random initial positions and velocities are introduced in the top plane of the computational domain, and these particles are subjected to gravity and interparticle interaction forces. Afterward, the Lagrangian equations of motion are solved to calculate the particle trajectories. The particles collide with the rest of the particles and walls during their motion, are subjected to the corresponding forces and torques, and gradually accumulate in the computational domain. After the stacking process is completed, an electrode domain without interparticle overlap is formed. It is then possible to simulate the sintering process simply by increasing the particle radius to create an overlap between the particles. Figure 7 shows the DEM-based packing algorithm and the contact areas after the overlap of particles (Zheng and Ni 2016). Another method for synthesizing SOEC electrode microstructures is random particle packing. The algorithm achieves filling the computational domain by randomly generating spherical particles that satisfy certain constraints to complete the particle packing. In this algorithm, if the randomly generated particle is valid

Modeling of Solid Oxide Electrolysis Cells

223

Fig. 7 Array of spheres produced from the DEM packing algorithm. The lower image highlights the contact areas between the spheres after artificial sintering (Sanyal et al. 2010)

(satisfies the corresponding constraints), the particle information (position and size) is fixed and then the next random generation of particles is performed. Conversely, if the particle does not satisfy the corresponding constraints, it is discarded and another particle is randomly generated afterward. The corresponding constraints can be set according to practical needs, e.g., a valid particle needs to be adjacent to or overlap with at least two particles/boundaries; the overlapping part of two particles cannot overlap with a third particle; the overlap angle is within a certain range. As can be expected, the random particle packing method will be extremely timeconsuming when the computational domain is large compared to the particle size. Because the algorithm is always performing trial-and-error in the later stages of random particle generation, it is difficult to generate new particles that satisfy the constraints. However, the algorithm is customizable to process according to your needs and is fully automated. The kinetic Monte Carlo (KMC) method is generally employed as an important modeling reconstruction approach due to its excellent computational efficiency (Wu

224

Y. Wang et al.

et al. 2022). Another advantage of the kinetic Monte Carlo method is that sintering kinetics are fully considered, thereby the convenient control of the sintering process is allowed. The reduction of system free energy is the driving force for the sintering process which is dependent on the interaction between the central pixel and the nearest (common side) or next-nearest neighbors (common corner):     1

 E= J qi , q j wi, j 1 − δ qi , q j 2 i j n

N

(36)

where E means the total free energy of the system, i is the number of pixels, N represents the total number of pixels, q represents the pixel value of different particles, j is the neighbor pixel number, n represents the total number of neighbor pixels, J is the surface energy dependent on pore-grain interaction and grain-grain interaction, δ is a Kronecker delta function which restricts that only interactions between different solid particles or between the solid phase and pores phases contribute to system energy:  δ=

1 qi / = q j 0 qi = q j

(37)

and wi,j is the weighting factor. Two values of wi,j are employed to distinguish the different contributions of the neighbor (common side) and nearest neighbor pixels (common corner) around the center pixel: wi j = 1 for i, j nearest neighbors, wi j = √12 for i, j next-nearest neighbors, and wi j = 0 for otherwise. J is the surface energy with a range of 0.5–2 J m−2 . The system free energy reduction is the driving force for the numerical sintering process, which is mainly realized by four procedures: grain growth, pore migration, vacancy formation, and annihilation. The grain growth is realized by randomly exchanging an arbitrary neighboring grain site with the centered grain site. Firstly, we assume that the exchange attempt has already been accepted. The system free energy change could be obtained by: △E = E initial − E final

(38)

where E initial is the initial system energy calculated by Eq. 36, while E final is the final system energy after the virtual exchanging step. The system free energy change is used within the standard Metropolis algorithm to calculate the probability of exchanging attempts:  P=

1 exp

−△E KBT

 △E < 0 △E ≥ 0

(39)

Modeling of Solid Oxide Electrolysis Cells

225

Whether the exchange attempt is accepted or not is determined by the comparison of P with a specific grain growth frequency. If P is larger than the grain growth frequency, the exchange will be accepted. Otherwise, the system is restored to its original state. The pore migration procedure is simulated by exchanging the central pore pixel at random with a neighboring grain pixel. Similarly, the standard Metropolis algorithm probability and specific pore migration frequency are also adopted here for the acceptance or rejection of exchanging attempts. Vacancy formation is inevitable at a relatively large sintering temperature (>1000 °C). It is assumed in the present study that once the vacancy is generated, it will annihilate in the next KMC step with a specific frequency close to 0. The growth of sintering necks is mainly caused by grain growth and pore migration procedure while the vacancy formation and annihilation contribute to the electrode densification phenomenon. To further improve the computational efficiency of KMC simulation, interface recognition is conducted to realize the efficiency and accuracy of boundary lattice exchange. Firstly, we define a solid pixel as the central pixel. If the neighboring pixels contacted with the central pixel are all solid-phase pixels and the neighboring pixels have at least one different pixel gray value, then this pixel is defined as a solid-phase grain boundary. If the neighboring pixel contains a pore pixel, the central pixel is stipulated as a surface grain boundary. Only the mutual exchange between the above two types of boundary pixels will cause the reduction of the free energy. All the solid phase grain boundaries and surface grain boundaries are firstly identified before the numerical sintering. It should be noted that in a KMC step, each grain or surface boundary pixel has an opportunity to exchange. After completing all the exchange steps, the vacancy annihilation step is performed to confirm that no isolated pixels are generated during the high-temperature sintering procedure. For modeling validation, a similar microstructure appearance cannot ensure the reliability of the model. Figure 8 shows the LSCF electrode reconstructed by the KMC model and the related O2 concentration, ionic and electronic overpotential distribution (Wu et al. 2022). SOCs fuel electrodes are subject to Ni phase migration under long-term operating conditions (Mogensen et al. 2021). For SOFC, the accumulation between smaller Ni particles at the beginning of cell operation results in larger Ni particles, which is called Ni coarsening. Ni coarsening directly leads to a decrease in the density of three-phase boundary length (TPBL), which increases the activation overpotential and reduces the cell output performance. For SOEC fuel electrodes, which are also Nibased composite cermet, in addition to Ni coarsening, there is another Ni migration phenomenon called Ni depletion that affects the durability of SOEC (Monaco et al. 2019). Figure 9 shows the distribution of Ni and YSZ phases of the SOEC fuel electrode after long-term operation taken by experimental approaches (Mogensen et al. 2017). It can be seen that after a period of operation, the Ni phase content decreases significantly near the electrolyte layer, and a porous YSZ layer is formed. Phase field modeling is a well-suited tool to simulate Ni migration at fuel electrodes of SOCs. Previous studies have focused on Ni coarsening in SOFC using PFM (Wang et al. 2021a, b) extended this model to study Ni depletion in SOEC. Figure 10a shows the schematic of Ni migration inside the SOEC fuel electrode,

226

Y. Wang et al.

Fig. 8 O2 concentration, ionic and electronic overpotential distribution of LSCF electrode reconstructed by the KMC model (Wu et al. 2022)

Fig. 9 Representative micrographs from Ni–YSZ electrodes. a pristine electrode. b and c are from a cell tested for 9000 h at 800 °C applying an inlet gas composition of 90% H2 O + 10% H2 . b shows a part of the cell with no current load. c piece of cell tested at −1 A cm−2 (Mogensen et al. 2017)

and Fig. 10b represents the mechanism of Ni depletion. Trini et al. (2020) attributed the driving force of Ni depletion to the chemical potential gradient caused by the change in wettability of the Ni phase on the YSZ in the direction perpendicular to the electrolyte. The wettability is directly reflected by the contact angle between the Ni and YSZ phases, which is also the basis of the study by Wang et al. In this study, the morphologically evolved SOEC electrode was characterized as a computational domain using LBM to investigate the effect of Ni depletion on the electrochemical performance of SOEC. In the following, we will describe the application of PFM in the morphological evolution of SOEC electrodes and the application of LBM in the electrochemical characterization of heterogeneous structured electrodes. As a mathematical method used to solve interfacial problems, the phase field model has been able to effectively simulate the microstructural evolution of various material processes, such as sintering, solidification, solid-state phase transformation, grain dendrite growth, and grain growth. Figure 11 shows the comparison between

Modeling of Solid Oxide Electrolysis Cells

227

Fig. 10 a Schematic of the Ni migration process in the fuel electrode, b mechanism of Ni migration (Wang et al. 2021b)

the sharp interface and the diffuse interface in the phase field model (Nabiollahi et al. 2015). In recent years, the phase field model has been extended for application in modeling the microstructural evolution of SOCs. The physical processes simulated include the migration process of the Ni phase in the fuel electrode, the transformation of the NiO phase in the reduction process (Jiao and Shikazono 2016), and the cubic to tetragonal transformation in YSZ (Da et al. 2021). Here, we will detail the phase field modeling process applied to Ni phase migration. The operating temperature of SOEC is usually higher than the melting point of Ni, which makes the Ni phase tend to migrate during the electrolytic cell operation. While electrolysis phases such as YSZ or GDC can be considered stationary or are given a negligible migration rate, the phase field model is characterized by its diffuse interface. This kind of phase interface is identified by several non-equilibrium thermodynamic theory-based partial differential equations. Two sets of order parameters are used to represent the interface migration of phase volume fraction and grain boundary. For a two-phase composite system of SOEC (e.g., Ni–YSZ), the volume fractions of the pore, Ni, and YSZ are represented by the conserved order parameters for C p , C N , and C Y , respectively. Meanwhile, these three parameters satisfy the following condition: C p + C N + CY = 1. The non-conserved

228

Y. Wang et al.

Fig. 11 Comparison between the sharp interface and the diffuse interface in the phase field model (Nabiollahi et al. 2015)

parameters (ηN and ηY ) are represented the different rain orientations of the solid phases. The range of values for the conserved parameters is a smooth transition from 0 to 1 at the solid- or gas-phase interface. And the non-conserved parameters take values in the range of −1 to 1, while the values within the hole phase are all 0. The total system free energy is determined by two types of parameters together and is expressed by the following equation:  

m n

κiN  N 2 κiY  Y 2 ∇ηi ∇ηi + 2 2 i i ⎤

κi j   C ∇Ci · ∇C j ⎦dv + 2 i=N ,Y j=N ,Y

f 0 ({Ci }, {ηi }) +

Ftotal =

ij

(40)

where κiN , κiY , and κC are the gradient energy coefficients. The bulk free energy, the residual interfacial energy of grain boundaries, and the residual interfacial energy at the phase interfaces are represented as the first term, the second and third terms as well as the last term of Eq. (40), respectively. The bulk free energy density (f 0 ) can be calculated as:

f0 = αi f 1 (Ci ) + α N Y f 1 (C N + CY ) i=N ,Y

+ +

i=N ,Y n

β f 2 (Ci ) +

δ f 3 (CY , ηin )

i

m

δ f 3 (C N , ηim )

i

+

m

m

i=1 j/=i

  ε f 4 ηiN , η Nj

Modeling of Solid Oxide Electrolysis Cells

+

n

n

229

  ε f 4 ηiY , ηYj +γ f 5 (C N , CY ) + f 6 (C N , CY )

(41)

i=1 j/=i

where ⎧ f 1 (C) = C 2 (1 − C)2 ⎪ ⎪ ⎪ ⎪ ⎪ 1 ⎪ ⎪ f 2 (C) = C 4 ⎪ ⎪ ⎪ 4 ⎪ ⎪ ⎪ 1 1 ⎪ ⎪ ⎨ f 3 (C, ηi ) = C 2 ηi2 + ηi4 2 4 ⎪ ⎪ f η , η  = 1 η2 η2 ⎪ 4 i j ⎪ ⎪ 2 i j ⎪ ⎪ ⎪ ⎪ 1 ⎪ ⎪ f 5 (CN , CY ) = CN2 CY2 ⎪ ⎪ 2 ⎪ ⎪   ⎩ f 6 (CN , CY ) = −Wca CN |∇CY |2 + k

(42)

The above relationship equations include the volume fraction-depended doublewell potential function, the coupling relations between volume fractions of different phases and between volume fraction and crystallographic orientation, as well as the energy relations between grain orientations. The wettability of Ni on the YSZ surface is controlled by W ca in f 6 . And k is used as a regulation parameter to fit the Ni depletion experiment. The morphology evolution of electrode is driven by the decreasing trend of total free energy, and the value changes of the order parameters over time are characterized by the following Cahn–Hilliard and Allen–Cahn equations: ⎡ ⎛ ⎞⎤

∂ f ∂Ci 0 ij = ∇ · ⎣ MCi ∇ ⎝ − κC ∇ 2 C j ⎠⎦, i = N , Y ∂t ∂Ci j=N ,Y   ∂ f0 ∂ηi 2 − κ∇ ηi , i = 1, 2, . . . , m/n, = −L j ∂t ∂ηi

j = N /Y

(43)

(44)

where M c and L are the mobilities of volume fraction and grain orientations. The migration of solid phases in SOC electrodes is divided into several processes: the surface diffusion, the volume diffusion, and the grain boundary diffusion, and the variable Mc can be calculated as the following equation:  

Ni Ni Ni ηi η j g(1 − CY ) MCN i (C N ) = Mvol φ(C N ) + Msur f C N (1 − C N ) + M gb (45) During the long-term operation of SOEC, the Ni coarsening is mainly affected by the operating temperature and the initial microstructure parameters, such as volume fraction and initial particle diameter of Ni phase. Except for the above two influence

230

Y. Wang et al.

factors, the Ni depletion is attributed to the combined effects of various factors, such as inlet H2 O fraction and polarization strength. These factors determine the degree of Ni depletion together. For simplification, the value of W ca is used to reflect the degree of Ni depletion. It is worth looking forward to the future simulation study of Ni depletion, in which the abovementioned factors need to be reflected in detail in the model to study the role of multiple influence factors. Figure 12 represents the solid phases and TPB distribution of SOEC fuel electrodes with different Ni depletion degrees (Wang et al. 2021b). After the long-term operation (5000 h), we can visually observe that the particle size of the Ni phase is significantly larger compared to the initial state through the coarsening process. Meanwhile, the TPB network becomes sparser as Ni depletion increases. This is because severe Ni depletion can simultaneously lead to increased Ni coarsening. The bottom surface of the computational domain is the intersection of the electrolyte layer and the electrode, and Ni depletion causes a porous YSZ layer in the electrode near the electrolyte layer, which is caused by the migration of the Ni phase away from the electrolyte layer. And this phenomenon has directly led to a decrease in TPB density. In addition, there is a gap in the study of Ni migration on the thermomechanical properties of electrodes. During the actual operation of the cell and electrolysis cell, they will inevitably suffer from the effects of thermal cycling. The mismatch between the thermal expansion coefficients (TEC) of the electrodes and electrolyte can cause serious faults such as delamination between the two layers. If the abovementioned porous YSZ layer is generated by Ni depletion during the operation of the electrolysis cell, it may have an impact on the thermomechanical properties at the interface, thus accelerating the delamination phenomenon. Further simulations and experimental studies are needed in this regard in the future. Unlike traditional computational fluid dynamic (CFD) methods that solve multiple sets of macroscopic conservation equations (e.g., mass, momentum, and energy), LBM simulates a fluid composed of hypothetical particles and solves for the collision and propagation processes of these particles on discrete lattices (Chen et al. 2014). Due to its particle nature, LBM is very suitable for solving complex boundary systems and mesoscopic physics problems compared to CFD methods, and its algorithmic construction is suitable for parallelized computations. Based on the microstructure of the SOEC electrode after morphology evolution, the lattice Boltzmann method with the multiple-relaxation-time (MRT) collision operator is used to calculate the electrochemical performance. The LBM also considers both mass transfer and electron/ion transfer processes in the macroscopic model described above. Compared to the single-relaxation-time scheme, MRT offers higher numerical stability (Li et al. 2016). The D3Q7 lattice scheme is sufficient to meet the accuracy of electrochemistry and mass transport calculations of SOEC electrodes and also has high computational efficiency. The schematic of the D3Q7 lattice scheme is shown in Fig. 13. And the evolution of the LBM equation with the MRT operator can be expressed as:   eq f i (x + ei △t, t + △t) − f i (x, t) = −M −1 ΛM f i (x, t) − f i (x, t)

(46)

Modeling of Solid Oxide Electrolysis Cells

231

where f i is the density distribution function in the ith direction. The discrete velocity ei can be given as:  ei =

(0, 0, 0) i = 0 (±c, 0, 0), (0, ±c, 0), (0, 0, ±c) i = 1 − 6

(47)

Fig. 12 Effects of Ni depletion on the morphology of fuel electrode: a Ni–YSZ composite; b TPB network (Wang et al. 2021b)

Fig. 13 D3Q7 discrete velocities model for mass transfer

232

Y. Wang et al.

The lattice sound speed c = △x/△t, where △x and △t are the lattice space and the time step, respectively. M and Λ are the transformation matrix and the corresponding diagonal relaxation matrix, respectively, given as: ⎛

⎞ 1 1 1 1 1 1 1 ⎜ 0 1 −1 0 0 0 0 ⎟ ⎜ ⎟ ⎜ 0 0 0 1 −1 0 0 ⎟ ⎜ ⎟ ⎜ ⎟ M = ⎜ 0 0 0 0 0 1 −1 ⎟ ⎜ ⎟ ⎜ 6 −1 −1 −1 −1 −1 −1 ⎟ ⎜ ⎟ ⎝ 0 2 2 −1 −1 −1 −1 ⎠ 0 0 0 1 1 −1 −1  −1 −1 −1 −1 −1 −1 −1  Λ = diag τ0 τ1 τ2 τ3 τ4 τ5 τ6

(48)

(49)

eq

The equilibrium distribution function can be given by: f i (x, t) = ωi C with C ∑ calculated as C= 6i=0 f i (x, t). The weight coefficients ωi are given as: ω0 = k 0 , ω1 −-6 = (1−k 0 )/6. The isotropic diffusion is adopted in this study, and the relaxation coefficients τ i are given as: ⎧ τ0 = 1 ⎪ ⎪ ⎪ ⎪ ⎪ 1 1 ⎨ τ1−3 = + D 2 c△x(1 − k0 )/3 ⎪ ⎪ ⎪ 1 1 ⎪ ⎪ ⎩ τ4−6 = + 2 6(τ1−3 − 1/2)

(50)

During the LBM calculation of coupled electron–ion transport and mass transport, the problem of mismatching the magnitudes of the diffusion coefficients of physical quantities in the two physical fields is usually encountered. For example, the diffusion coefficients of electron and ion transfer in SOEC electrodes are several orders of magnitude larger than those of the gas components. To avoid instability of the calculation while ensuring the robustness of the multi-physics coupling, the following equation is used for the transfer of electrons and ions: 0 = ∇ · (D0 ∇φion/ele ) ± D0 J/σion/ele

(51)

Figure 14 illustrates the distribution of the ionic potential of the Ni depletion electrode (Wang et al. 2021b). It can be seen that there is a large difference in ion potential between the two sides of the electrode at a high Ni depletion level. This is mainly due to the presence of the porous YSZ layer mentioned above which directly leads to a decrease in TPB density on the one hand and an increase in ion transport paths on the other. The increase in ion transport paths is because the electrochemical reactions are mainly concentrated near the electrolyte layer in the absence of Ni depletion, but with the increase in Ni depletion, the porous YSZ layer is created, which makes the electrochemical reaction active sites migrate away from

Modeling of Solid Oxide Electrolysis Cells

233

Fig. 14 Ionic potential distributions with different degrees of Ni depletion (Wang et al. 2021b)

the electrolyte layer and directly increases the ion transport paths, which is also the major reason for the large increase in ohmic overpotential under Ni depletion.

4 Applications of Machine Learning in SOEC Modeling 4.1 Machine Learning Overview Machine learning (ML) has gained unprecedented development in computer vision, image recognition and big data analysis in recent years. In addition, machine learning has been increasingly applied to physics, chemistry, biology, materials science, and engineering (Jordan and Mitchell 2015). Unsupervised learning, supervised learning, and reinforcement learning are the three most common machine learning methods. Unsupervised learning is performed on unlabeled datasets and is often used in clustering and dimensionality reduction problems. In the field of energy, the main unsupervised learning algorithms are the k-mean clustering algorithm, for clustering problems, and principal component analysis (PCA), and singular value decomposition (SVD) for dimensionality reduction problems. Reinforcement learning builds on the principles of Markovian decision processes, training agents to make decisions for the next action that maximizes the cumulative reward. Q-learning is the most commonly used algorithm of reinforcement learning. Compared with the above two machine learning methods, supervised learning is more widely used in the energy field due to its accuracy and efficiency. The two main tasks of supervised learning are regression and classification. Regression means predicting continuous, specific values, while classification means classifying various things into categories for discrete predictions. The following table lists the common algorithms and features of supervised learning and the problems they can solve. Table 1 lists the typical supervised learning algorithms and characteristics.

234

Y. Wang et al.

Table 1 Typical supervised learning algorithms and characteristics Algorithms

Characteristics

Linear regression

Linear regression is one of the Regression most commonly used algorithms to handle regression tasks. The algorithm has a very simple form, and it expects to fit the dataset using a hyperplane (a straight line when there are only two variables)

Types

K-nearest neighbor (KNN)

New data points are predicted by Regression/classification searching the entire training set of the K most similar instances (neighbors) and summarizing the output variables of those K instances

Logistic regression

Logistic regression is used to Classification deal with regression problems in which the dependent variable is a categorical variable, commonly dichotomous or binomial distribution problems, and can also deal with multi-categorical problems, which belong to a classification method

Decision tree

The decision tree is a simple but widely used classifier that builds decision trees from training data to classify unknown data

Classification

Support vector machine (SVM)

Support vector machines transform the classification problem into a problem of finding a classification plane and achieving classification by maximizing the distance of classification boundary points from the classification plane

Classification

Artificial neural network (ANN)

It abstracts the neuronal network of the human brain from the perspective of information processing and builds some simple models to form different networks with different connections

Regression/classification

Modeling of Solid Oxide Electrolysis Cells

235

4.2 Machine Learning in SOEC Modeling The main machine learning methods currently applied on SOEC modeling are ANN and deep learning. For conventional mathematical models based on complex multiphysics fields such as fuel cells, they are usually algorithmically complex and very computationally time-consuming. ANN is a good tool to perform the tasks applied in the above mathematical models. This is due to the fact that ANN is essentially a black-box model, and the computations are usually parallel and do not require the construction of complex algorithms, so ANN can be computationally faster than many conventional mathematical models. Using the properties of ANN, we can also accomplish the corresponding target operations without knowing the algorithm for a particular problem. In addition, ANNs have the ability to generalize knowledge, meaning that ANN can be taught to provide correct solutions to problems that are similar but not identical to known problems. This is one of the reasons why ANN can be widely used in energy research. For the application of ANN to SOEC, on the one hand, a large amount of experimental data can be used to train the neural network to obtain a neural network-based model. However, in general, ANN training requires a large amount of data, especially for deep learning, so it is a promising approach for ANN modeling in the field of fuel cell and electrolysis cell to train neural networks and obtain surrogate models using experimentally fitted mathematical models as parameter generators, as opposed to models based on experimental data (Wang et al. 2020b). The surrogate model obtained by this method can achieve fast convergence of the model (no iterations) and obtain low errors with the high matching of experimental data. This means that the obtained surrogate models can predict SOEC performance under a wide range of operating conditions and can be used for control and monitoring of real SOEC systems and predicting their performance before changing the operating conditions. As a black-box model, an ANN relates specific combinations of input and output data. The process of training a neural network is both a process of obtaining internal relationships between inputs and outputs and predicting the behavior of the system as a whole, without the need for specific physically meaningful equations within this entire process. The whole ANN consists of an aggregation of neurons. Information is passed from dendrites to neurons; activation functions are implemented in neurons, and the modified information is then passed forward to other neurons via axons and through synapses (Fig. 15a). As shown in Fig. 15b, the input parameters are received   W . The the first layer of neurons and multiplied by the corresponding weights k,i,1   above results are summed and added the bias xk,0  and then recomputed using the activation function to obtain the final output yk,i . This forward movement of information is known as forward propagation. If the generated output is far from the actual value, the error is calculated using the forward propagated output. Based on this error value, the weights and biases of the neurons are updated. This process is called backward propagation.

236

Y. Wang et al.

Fig. 15 a Neuron scheme and its b mathematical model

Deep learning has grown rapidly in recent years and has become a frontier area of artificial intelligence compared to many other machine learning methods. Deep learning is built on artificial neural networks, which are artificial neural networks with deep structures or multiple hidden layers, and can be unsupervised or supervised. Deep learning creates a connection between the input and output parameters, which is typically represented by a nonlinear function and uses this to create a black-box model to compute the values of the target parameters. Deep learning has a much simpler and clearer mathematical form than similar traditional algorithms and can achieve very good predictive performance with the support of big data. Deep learning will be an important way for artificial intelligence to learn physical knowledge. Figure 16 shows the relationship between AI, machine learning, and deep learning. In the field of energy batteries, machine learning and deep learning have been widely used. For example, the SVM has been used to predict the remaining life of lithium battery performance (Patil et al. 2015). The grid-long short-term memory (G-LSTM) is used for the prediction of PEMFC performance degradation (Ma et al. 2018). And the generative adversarial network (GAN) has been used to generat the microscopic electrode structures of Li-ion battery and SOFC (Gayon-Lombardo et al. 2020). Although there are not many machine learning applications related to SOEC at this stage, it is expected that future researchers can use a series of open-source machine learning frameworks (e.g., PyTorch, TensorFlow, and Keras)

Modeling of Solid Oxide Electrolysis Cells

237

to perform multi-scale optimization of SOEC, further improve SOEC performance, and accelerate its commercialization. As a result of the neural network properties introduced above, we can apply machine learning methods to many aspects of SOEC simulation. Taking a singlecell level model as an example, structural parameters, operating conditions, etc., can be used as input parameters to the neural network, while target parameters to be optimized, such as electrolysis voltage, efficiency, and cell temperature, are used as outputs to train the neural network to obtain a surrogate model. Figure 17 shows the structure of the neural network which can be applied in SOEC study. Xu et al. (2021) used gas flow rate, components, and operating voltage as inputs to the neural network and electrolysis heat production as outputs to train the neural network and combine it with a genetic algorithm to predict SOEC operating conditions at thermoneutral voltage. The resulting surrogate model takes about 0.6 s to predict an optimization point, which is significantly less than the time required by

Fig. 16 Relationship of AI, machine learning, and deep learning

Fig. 17 Structure of the neural network can be applied in SOEC study

238

Y. Wang et al.

the multi-physics field model (>60 s), and this is one of the reasons why the surrogate model can be quickly coupled into the genetic algorithm to find the optimization. In Fig. 18, a four-dimensional plot of power density, temperature, gas fraction, and applied voltage is shown. It can be seen that the applied voltage and power density required for electrolysis are smaller for higher water content and operating temperature conditions. When the inlet gas is 90% H2 O and 10% CO2 and the operating temperature is 1023 K, only 1.124 V and 0.258 W cm−2 of voltage and power density is required to make the cell operate at a thermoneutral voltage. The operating temperature has a greater impact on reducing the power density for the thermoneutral operation of the electrolysis cell compared to the inlet gas composition. These parameters need to be carefully chosen in operation to weigh the operating conditions under the thermoneutral operation of the cell, and the above process of machine learning-based agent model combined with genetic algorithm provides an effective means for this purpose. In addition to this, there are researchers who use generative adversarial networks in deep learning for the direct generation of SOC electrode microstructures (Hsu et al. 2021; Ma et al. 2018). GANs were invented by Lan Goodfellow in 2014 and were first described in relation to them in the paper Generative Adversarial Nets (Goodfellow et al. 2014). The structure of GANs consists of two deep learning

Fig. 18 a Distribution of required power density and applied voltage at different gas compositions and temperatures of cathode inlet under the thermoneutral condition, b Voltage-temperature-gas composition relationship, c Voltage–power-gas composition relationship, and d Voltage–powertemperature relationship (Xu et al. 2021)

Modeling of Solid Oxide Electrolysis Cells

239

Fig. 19 Schematic diagram of GANs

networks (generally CNNs), called generators and discriminators, respectively. The generator continuously generates “fake” images (G(z)), which are in turn fed into the discriminator. The discriminator acts as a “judge” and discriminates between the images generated by the generator and the real images (X) in the training set. The real image in the training set should have a value (D(X)) close to 1 after passing through the discriminator, while the image generated by the generator should have a value (D(G(z))) close to 0 after passing through the discriminator. During the training process, the generator continuously optimizes and gradually generates images that can fool the discriminator, and the discriminator continuously grows to determine the authenticity of the images. The training process is also the process of finding the equilibrium point, i.e., the discriminator can only determine with 50% probability that the image generated by the generator is false. The schematic diagram of GANs is shown in Fig. 19. The entire loss function of GANs can be represented by the following equation:     min max V (D, G) = E x log D(x) + E z log(1 − D(G(z))) G

D

(51)

where z is the latent space vector to be sampled as the noise. D and G play a min–max game in which D tries to maximize the probability for correctly classifying the real images (log D(x)), and G tries to minimize the probability that D correctly predicts the fake images (log(1 − D(G(z)))). Convolutional neural networks (CNNs) used in GANs are mainly applicable to image data and computer vision. CNNs receive images and apply a series of signal processing steps (typically filter convolution, rectification operations, and pooling or down sampling) to generate feature vectors that contain information about the image features. The CNN architecture can be easily defined by the user to obtain the best performance according to the actual requirements. CNNs can also operate in reverse, taking the feature vectors and converting them to image identities. GANs are based on both convolutional and transposed neural networks for discriminator and generator training. Hsu et al. (2021) used Wasserstein GAN (WGAN) to generate a 3D electrode structure by dividing the training on the original 3D experimental data using a generator after the training was completed, as shown in Fig. 20. WGAN is a variant of GAN

240

Y. Wang et al.

Fig. 20 Volume renderings of representative microstructures cropped from the original experimental data (top) or generated by the WGAN-trained generator using random noise vector inputs (bottom) (Hsu et al. 2021)

and differs from GAN in that its discriminator outputs an approximate difference metric between the synthetic and real images (Wasserstein loss). It can be seen that the 3D structure generated by WGAN is very similar to the original 3D structure in terms of morphology. GAN can generate realistic 3D, topologically complex, multi-phase, grayscale microstructures that closely resemble the original experimental structures in terms of visual appearance, statistical representation of geometric and topological properties, and simulated electrochemical properties. The GAN framework can be applied to arbitrary material systems, microstructure morphologies, and imaging modalities, making it a versatile and flexible tool for microstructure generation. In the future, different GANs can be trained to generate SOEC electrodes with different microstructural parameters (e.g., porosity, volume fraction, and particle size) and accordingly derive optimized solutions to achieve the best electrode performance and durability.

5 Conclusions and Perspectives The above chapter details the simulation process of high-temperature SOEC from the macroscopic cell to the electrode level with a basic framework such as the control equations. Multi-scale simulation tools play a very important role in the development of SOEC technology, where the macroscopic models are the basis for engineering applications of SOEC. For the macroscopic modeling of SOEC, the electrochemistry and multi-physics transport should be fully considered, including the multicomponent mass transfer process in porous electrodes, the global and fundamental kinetics of internal reforming and the electrochemical reactions. Meanwhile, the modeling

Modeling of Solid Oxide Electrolysis Cells

241

process at the electrode scale is also a key topic which is beneficial to improve the cell performance and durability from a more microscopic perspective. Through a series of modeling approaches, the full life-cycling behaviors of SOEC electrodes can be simulated, and thus, SOEC performance can be improved from all stages. In addition, the machine learning algorithms are introduced and can be combined with different models as a complement for simulation optimization tools. In addition to the models detailed above, there are still many important models that are not covered in this chapter for reasons of space, such as system-level models, dynamic models, and more microscopic models at the atomic and molecular levels. These models are also an important part of multi-scale modeling. Next, some perspectives on the multi-scale modeling are discussed below: 1. For the macroscopic model, the large-scale multi-physics field model is also needed to be further improved at the stack level, which is important to optimize the overall stack flow field, distribution and geometry configuration and thus improve the overall performance and efficiency of the stack. Especially for in the aspect of thermal management of SOEC stack, the non-isothermal multi-physics model provides an intuitive distribution of velocity and temperature fields within the stack, which in turn can be combined with structural mechanics and other models to optimize the stack structure and achieve improved long-term thermal stability. In addition, for SOEC applied to renewable energy systems, the SOEC input is usually intermittent, so further dynamic modeling of the SOEC is required to improve the dynamic response characteristics and thermal cycling stability of the cell thermal and electrochemical performance. PSOEC is also the subject of future simulation work that needs to be focused. The analysis of the internal transport mechanism of mixed conductors by means of simulations will be very useful to improve the Faraday efficiency of PSOEC. 2. The system-level model consists of a series of sub-model of SOEC combined system, such as the stack model and the balance of plant (BoP) components. Through system-level process simulation, we can improve control strategies to maximize system efficiency and assess the feasibility of new SOEC system designs in terms of economics and commercial viability. It is foreseen that new SOEC systems integrated with renewable energy sources and near-zero CO2 emissions through co-electrolysis of steam and CO2 will be the target for system-level modeling and performance evaluation. 3. At the microscopic electrode level, the decoupling model combining multiple simulation methods will be one of the research directions. The combination of multiple simulation approaches allows for the optimal design of electrode microstructures from the full electrode life cycle from electrode preparation to long-term electrode operation and to improve the design of different electrode materials and structures such as mixed-conducting and nanoparticle-infiltrated electrodes. As mentioned above, the simulation studies on the long-term stability of electrodes are still very inadequate. The suppression of Ni phase migration through decoupling models to enhance the stability of the cell during thermal and redox cycling would be promising means to achieve this. The long-term stability

242

Y. Wang et al.

of air electrodes is also an aspect that deserves a lot of attention. Modeling studies of relevant degradation mechanisms such as cation diffusion and Cr deposition are still nearly vacant. Atomic-molecular scale simulations such as density functional theory and molecular dynamics can be used for solving such problems. Density functional theory calculations have been widely used to investigate insights for electrochemical materials, which bring insights into the material design that promotes the electrochemical performance of the cells. Molecular dynamics is a computer simulation method for analyzing the physical movements of atoms and molecules. It is foreseeable that these methods will have great potential for microscopic simulations. 4. The artificial intelligence methods such as machine learning provide more options for SOEC simulation research, which in turn can be combined with optimization methods such as genetic algorithms to achieve improvements in SOEC performance at multiple scales and can provide real-time control optimization solutions for SOEC systems. At the same time, machine learning can be used as a support tool for all of the above simulation methods and can be combined with them for different optimization purposes.

References Bi L, Boulfrad S, Traversa E (2014) Steam electrolysis by solid oxide electrolysis cells (SOECs) with proton-conducting oxides. Chem Soc Rev 43:8255–8270 Birnbaum K, Steinberger-Wilkens R, Zapp P (2013) Solid oxide fuel cells, sustainability aspects. Fuel Cells, Springer, pp 731–790 Chen L, Kang QJ, Mu YT, He YL, Tao WQ (2014) A critical review of the pseudopotential multiphase lattice Boltzmann model: methods and applications. Int J Heat Mass Transf 76:210–236 Da Y, Jiao Z, Zhong Z (2021) Phase field modeling of cubic to tetragonal transformation in YSZ and its effect on ionic conductivity degradation and mechanical damages. In: ECS meeting abstracts, IOP Publishing, pp 247 Damm DL, Fedorov AG (2005) Spectral radiative heat transfer analysis of the planar SOFC. J Fuel Cell Sci Technol 2:258–262 Durlofsky L, Brady JF (1987) Analysis of the Brinkman equation as a model for flow in Porousmedia. Phys Fluids 30:3329–3341 Gayon-Lombardo A, Mosser L, Brandon NP, Cooper SJ (2020) Pores for thought: generative adversarial networks for stochastic reconstruction of 3D multi-phase electrode microstructures with periodic boundaries. Npj Comput Mater 6 Goodfellow IJ, Pouget-Abadie J, Mirza M, Xu B, Warde-Farley D, Ozair S, Courville A, Bengio Y (2014) Generative adversarial nets. Adv Neural Info Process Syst 27(Nips 2014), 27:2672–2680 Hasanabadi A, Baniassadi M, Abrinia K, Safdari M, Garmestani H (2016) Optimization of solid oxide fuel cell cathodes using two-point correlation functions. Comput Mater Sci 123:268–276 Hsu T, Epting WK, Kim H, Abernathy HW, Hackett GA, Rollett AD, Salvador PA, Holm EA (2021) Microstructure generation via generative adversarial network for heterogeneous, topologically complex 3D materials. Jom 73:90–102 Jiao ZJ, Shikazono N (2016) Simulation of the reduction process of solid oxide fuel cell composite anode based on phase field method. J Power Sour 305:10–16

Modeling of Solid Oxide Electrolysis Cells

243

Jordan MI, Mitchell TM (2015) Machine learning: trends, perspectives, and prospects. Science 349:255–260 Li Z, Yang M, Zhang YW (2016) Lattice Boltzmann method simulation of 3-D natural convection with double MRT model. Int J Heat Mass Transf 94:222–238 Li R, Liu XM, He GZ, Hu PF, Zhen Q, Liu JBL, Bashir S (2021) Green catalytic synthesis of ammonia using solid oxide electrolysis cells composed of multicomponent materials. Catal Today 374:102–116 Luo Y, Shi YX, Li WY, Cai NS (2015) Dynamic electro-thermal modeling of co-electrolysis of steam and carbon dioxide in a tubular solid oxide electrolysis cell. Energy 89:637–647 Ma R, Yang T, Breaz E, Li ZL, Briois P, Gao F (2018) Data-driven proton exchange membrane fuel cell degradation predication through deep learning method. Appl Energy 231:102–115 Mogensen MB, Hauch A, Sun X, Chen M, Tao Y, Ebbesen SD, Hansen KV, Hendriksen PV (2017) Relation between Ni particle shape change and Ni migration in Ni-YSZ electrodes—a hypothesis. Fuel Cells 17:434–441 Mogensen MB, Chen M, Frandsen HL, Graves C, Hauch A, Hendriksen PV, Jacobsen T, Jensen SH, Skafte TL, Sun XF (2021) Ni migration in solid oxide cell electrodes: review and revised hypothesis. Fuel Cells 21:415–429 Monaco F, Hubert M, Vulliet J, Ouweltjes JP, Montinaro D, Cloetens P, Piccardo P, Lefebvre-Joud F, Laurencin J (2019) Degradation of Ni-YSZ electrodes in solid oxide cells: impact of polarization and initial microstructure on the Ni evolution. J Electrochem Soc 166:F1229–F1242 Nabiollahi N, Moelans N, Gonzalez M, De Messemaeker J, Wilson CJ, Croes K, Beyne E, De Wolf I (2015) Microstructure simulation of grain growth in Cu through silicon vias using phase-field modeling. Microelectron Reliab 55:765–770 Nelson GJ, Grew KN, Izzo JR, Lombardo JJ, Harris WM, Faes A, Hessler-Wyser A, Van Herle J, Wang S, Chu YS, Virkar AV, Chiu WKS (2012) Three-dimensional microstructural changes in the Ni-YSZ solid oxide fuel cell anode during operation. Acta Mater 60:3491–3500 Patil MA, Tagade P, Hariharan KS, Kolake SM, Song T, Yeo T, Doo S (2015) A novel multistage support vector machine based approach for Li ion battery remaining useful life estimation. Appl Energy 159:285–297 Prokop TA, Berent K, Iwai H, Szmyd JS, Brus G (2018) A three-dimensional heterogeneity analysis of electrochemical energy conversion in SOFC anodes using electron nanotomography and mathematical modeling. Int J Hydrogen Energy 43:10016–10030 Riazat M, Tafazoli M, Baniassadi M, Safdari M, Faraji G, Garmestani H (2017) Investigation of the property hull for solid oxide fuel cell microstructures. Comput Mater Sci 127:1–7 Sanyal J, Goldin GM, Zhu HY, Kee RJ (2010) A particle-based model for predicting the effective conductivities of composite electrodes. J Power Sour 195:6671–6679 Tafazoli M, Shakeri M, Baniassadi M, Babaei A (2017) Investigation of the geometric property hull for infiltrated solid oxide fuel cell electrodes. Int J Energy Res 41:2318–2331 Trini M, Hauch A, De Angelis S, Tong X, Hendriksen PV, Chen M (2020) Comparison of microstructural evolution of fuel electrodes in solid oxide fuel cells and electrolysis cells. J Power Sour 450 Wang Y, Du YM, Ni M, Zhan RB, Du Q, Jiao K (2020a) Three-dimensional modeling of flow field optimization for co-electrolysis solid oxide electrolysis cell. Appl Therm Eng 172 Wang Y, Seo B, Wang B, Zamel N, Jiao K, Adroher XC (2020b) Fundamentals, materials, and machine learning of polymer electrolyte membrane fuel cell technology. Energy and AI 1:100014 Wang Y, Wu C, Du Q, Ni M, Jiao K, Zu B (2021a) Morphology and performance evolution of anode microstructure in solid oxide fuel cell: a model-based quantitative analysis. Appl Energy and Combustion Sci 5:100016 Wang Y, Wu C, Zu B, Han M, Du Q, Ni M, Jiao K (2021b) Ni migration of Ni-YSZ electrode in solid oxide electrolysis cell: an integrated model study. J Power Sour 516:230660 Wang Y, Zhan RB, Qin YZ, Zhang GB, Du Q, Jiao K (2018) Three-dimensional modeling of pressure effect on operating characteristics and performance of solid oxide fuel cell. Int J Hydrogen Energy 43:20059–20076

244

Y. Wang et al.

Wang Y, Zu B, Zhan R, Du Q, Ni M, Jiao K (2020c) Three-dimensional modeling and performance optimization of proton conducting solid oxide electrolysis cell. Fuel Cells 20:701–711 Wu C, Wang Y, Hou Y, Li X, Peng Z, Du Q, Ni M, Jiao K (2022) Reconstruction and optimization of LSCF cathode microstructure based on kinetic Monte Carlo method and Lattice Boltzmann method. Chem Eng J 436:132144 Xu HR, Ma JB, Tan P, Wu Z, Zhang YX, Ni M, Xuan J (2021) Enabling thermal-neutral electrolysis for CO2 -to-fuel conversions with a hybrid deep learning strategy. Energy Convers Managem 230 Yan Z, He A, Hara S, Shikazono N (2019) Modeling of solid oxide fuel cell (SOFC) electrodes from fabrication to operation: Microstructure optimization via artificial neural networks and multi-objective genetic algorithms. Energy Convers Managem 198 Yuan JL, Sunden B (2014) On mechanisms and models of multi-component gas diffusion in porous structures of fuel cell electrodes. Int J Heat Mass Transf 69:358–374 Zhang JH, Lei LB, Liu D, Zhao FY, Ni M, Chen FL (2018) Mathematical modeling of a protonconducting solid oxide fuel cell with current leakage. J Power Sour 400:333–340 Zhang QP, Guo YX, Ding JW, Xia SH (2019) Hole conductivity in the electrolyte of protonconducting SOFC: mathematical model and experimental investigation. J Alloy Compd 801:343–351 Zheng KQ, Ni M (2016) Reconstruction of solid oxide fuel cell electrode microstructure and analysis of its effective conductivity. Sci Bulletin 61:78–85 Zheng Y, Wang JC, Yu B, Zhang WQ, Chen J, Qiao JL, Zhang JJ (2017) A review of high temperature co-electrolysis of H2 O and CO2 to produce sustainable fuels using solid oxide electrolysis cells (SOECs): advanced materials and technology. Chem Soc Rev 46:1427–1463

Protonic Ceramic Electrolysis Cells (PCECs) Laura Almar, Sonia Escolástico, Laura Navarrete, David Catalán-Martínez, Jesús Ara, Sonia Remiro-Buenamañana, Imanol Quina, and José M. Serra

1 Introduction 1.1 Brief History Proton-conducting cell technology has been of ever-growing interest since the first studies reported by Iwahara et al. in the early 1980s (Iwahara et al. 1981, 1982; Uchida et al. 1982). In these works, steam electrolysis was firstly performed on a protonconducting oxide electrolyte based on a Sc-doped SrCeO3 composition, reaching very promising Faradaic efficiency values between 50–95% with current densities in the range of −0.1 to −0.8 A cm−2 at 900 °C. In addition, it was observed lower polarization resistances compared to coetaneous oxygen-ion solid oxide technology, but it was necessary to reduce the film thickness to overcome some deficiencies found in terms of protonic conductivity and the synthesis technique employed. Later on, in the 1990s, Iwahara and co-workers followed this demonstration of application with a comprehensive work (Hamakawa et al. 1993), including both experimental and theoretical results, searching for other synthesis routes and materials, with the aim of improving the performance and show new application fields, i.e., fuel cells, membrane reactors, H2 sensors, etc. In this pioneering study, the enormous potential of the proton-conducting ceramic technology could already be envisaged. In the first decade of the XXI century, novel technological advances were focused on enhancing electrolyte and electrodes performance. New proton-conducting oxides were mainly based on ABO3 perovskites as electrolyte materials (Schober and Ringel L. Almar (B) · S. Escolástico · L. Navarrete · D. Catalán-Martínez · J. Ara · S. Remiro-Buenamañana · I. Quina · J. M. Serra (B) Instituto de Tecnología Química (Universitat Politècnica de València – Consejo Superior de Investigaciones Científicas), Av. Los Naranjos, S/N, 46022 Valencia, Spain e-mail: [email protected] J. M. Serra e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M. A. Laguna-Bercero (ed.), High Temperature Electrolysis, Lecture Notes in Energy 95, https://doi.org/10.1007/978-3-031-22508-6_9

245

246

L. Almar et al.

2004). In 2008, Stuart et al. (2008) developed a highly dense electrolyte layer using Ydoped BaZrO3 material, reaching current density values of −0.12 A cm−2 for 750 °C and 2 V, and exceptional chemical stability, although the protonic conductivity at lower temperatures was still insufficient. In addition, significant advances regarding the composition of the electrodes were accomplished. Sakai et al. (2009) reported alternative materials to porous platinum with significantly lower overpotentials and enhanced current efficiency of the H2 production rate. Over the last years, considerable progress has been made boosting the performance of lab-scale reactors, especially on reversible operations. Choi et al. (2019) demonstrated outstanding performance in reversible cells, with remarkable stability under cyclic operation and a long time-on-stream test. Current density values of − 0.18 A cm−2 were observed at 1.3 V and 600 °C. In 2019, Vøllestad et al. (2019) reported a BaZrO3 -based tubular proton ceramic electrolyser with 10 cm2 of active area. An excellent Faradaic efficiency close to 100% for high steam chamber pressures were achieved, and current densities of −0.1 A cm−2 at 1.25 V and 600 °C. At the same time, Duan et al. (2019) proposed a reversible protonic cell with 97% overall electric-to-hydrogen energy efficiency, reaching current densities up to − 1 A cm−2 at 1.28 V and 600 °C, and long-term stable operation over 1000 h.

1.2 Mechanism Figure 1 depicts the schematic diagram of the working principle of protonic ceramic electrolysis cells (PCECs) and solid oxide electrolysis cells (SOECs). The main difference between them lies in the charge carrier transferred through the electrolyte, i.e., when an external voltage is applied to the cell, protons are transported instead of oxide ions. In PCECs, steam is fed to the anode, where it is electrochemically split into protons and oxygen. Then, protons are transferred through the dense electrolyte layer, which is non-permeable to oxygen ions or other molecular gases, to the cathode and finally, they are combined with electrons forming a pure and dry H2 stream. On the other hand, in SOECs, the mechanism works oppositely. Steam needs to be fed to the cathode side, so steam molecules are split on this electrode. A mixture of H2 and unreacted steam is collected at the cathode chamber, whereas oxygen ions migrate through the dense electrolyte to the anode, where they are oxidized forming a high-purity oxygen gas stream. Typically, as it is schematized in Fig. 1, a PCEC is composed by three different layers: (1) a dense protonic ceramic electrolyte, (2) a porous steam electrode (anode) where the oxygen evolution reaction (OER) occurs and (3) a porous hydrogen electrode (cathode) where the hydrogen evolution reaction (HER) occurs. These layers or components must fulfill different requirements: • Protonic electrolyte: the protonic material should possess high protonic conductivity with negligible electronic conductivity under operation conditions.

Protonic Ceramic Electrolysis Cells (PCECs)

247

Fig. 1 Schematics of the working mechanism of PCECs (left) and SOECs (right)

• Steam electrode (anode): the material(s) should possess both high electronic conductivity (p-type conductivity) for current collection and high ionic (protonic and oxygen ion) conductivity in addition to electrocatalytic activity. The electrode should possess an optimized microstructure allowing the diffusion of steam and O2 removal. • Hydrogen/fuel electrode (cathode): the material(s) should present high electronic and protonic conductivity. Nowadays, most of the cathodes are based on Ni-based composite materials. Each of these components are discussed in later sections of this chapter.

1.3 Comparison Between PCECs and SOECs The use of PCECs presents several advantages over the SOECs (Bi et al. 2014; Zhang et al. 2017). Based on their working principle above-mentioned, protonic electrolysers allow the direct production of dry and pure compressed H2 by tuning the operating pressure on the hydrogen-side chamber and the operation current density. Thus, this technology is more cost-efficient and reduces the system complexity compared to a multi-step H2 compression. In SOECs a highly pure O2 stream is produced. Handling an undiluted O2 flow at high temperature and pressure involves safety hazards, but also gives rise to new challenges concerning cell degradation and construction of materials. Furthermore, to obtain dry H2 , an extra step to separate the unreacted steam from H2 is required, resulting in a more sophisticated system, and an increase in the corresponding capital expenditure (CAPEX) and operational expenditure (OPEX). Therefore, two key advantages of the protonic technology are (1) avoiding the risks associated with the undiluted O2 stream and (2) the reduction of the cost-driving downstream H2 separation. Another advantage of PCECs is the possibility of operating at lower temperatures than SOECs. Protons migrate with a lower energy barrier than oxide ions, so that, they

248

L. Almar et al.

are more favorable for operation at intermediate temperatures (300–700 °C), whereas oxide-ion conducting cells require higher temperatures due to the moderate ionic conductivity at this temperature range. Consequently, overpriced interconnections and sealing materials are required in SOECs in order to avoid long-term performance degradation. The intermediate operating temperature range allows a higher efficiency than low-temperature polymer electrolyte membrane cells (PEMs) and its reduced temperature, compared to solid oxide cells (SOCs), enables a more flexible coupling of waste heat sources and less constraint balance of plant. Furthermore, protonic cells are expected to be more durable than oxide-ion conducting because of their working mechanism. On one hand, Ni cermet is one of the most common hydrogen evolution reaction (HER) electrode employed in electrolysers at T > 600 °C, to not be oxidized by steam. In SOECs, an inlet gas containing a considerable concentration of H2 O has to be flow. However, in PCECs, steam is fed on the other electrode side, and the Ni electrode is only exposed to dry H2 protecting the electrode for the oxidation. The oxidation of the cathode would lead to higher resistance or even creation of fractures (Bernadet et al. 2020; Hauch et al. 2016). In addition, the kinetic and thermodynamic regime is favored by the protonic transport for the synthesis of higher-added value products. Theoretical analysis also showed that the reversible operating mode would be more favorable with protonconducting cells in terms of energy conversion efficiency (Ni et al. 2008). The whole device is well-known as reversible protonic ceramic electrochemical cell (RePCEC), so that, it can be operated as an electrolyser when there is an electricity surplus or as a fuel cell when the electricity demand cannot be supplied. The small molecular weight of H2 facilitates its transport leading to low overpotentials. Despite the advantages, several barriers need to be overcome to promote the practical use of PCECs. Over the last decades protonic ceramic electrochemical cells (PCECs) have received significantly lower attention compared to solid oxide cells (SOCs), which can be attributed mainly to the complexity of the devices’ fabrication, involving: (1) poor electrolyte sinterability or low chemical stability, (2) lack of reliable and cost-effective manufacturing routes or (3) large overpotentials of the electrodes, involving poor activity of the steam electrode for the water splitting reaction (Kim et al. 2019). Therefore, many efforts have been focused on the development of new electrolyte and electrode materials to fabricate and scale-up reliable PCECs, leading to success its market potential. Overall, PCEC is seen as a very promising technology for efficient, clean and cost-effective energy conversion and storage.

2 Thermodynamics of High-Temperature Electrolysis 2.1 Reversible Process A basic thermodynamic analysis can be applied to a general water splitting reaction, driven by an external work (W elec ). The treatment assumes a reversible process,

Protonic Ceramic Electrolysis Cells (PCECs)

249

where the inlet stream is pure water at a certain pressure and temperature. The outlet streams are pure hydrogen on the cathode, whereas on the anode chamber a mixture of unreacted steam and oxygen is obtained (see Fig. 1). The total energy demand for the water splitting reaction is given by its enthalpy of reaction (△Hr ). The Eq. 1 is expressed considering the enthalpies of the reactants (△Hi ), where the effect of pressure on the enthalpies is negligible under the usual operating conditions for this process (1–40 bar) (O’Brien 2012). This energy needs to be provided as electrical energy (△G r ) and heat (T · △Sr ). 1 △Hr = △HH2 + △HO2 − △HH2 O = f (T ) 2

(1)

△Hr = △G r + T · △Sr

(2)

From Eq. 2, the reversible electrical work must satisfy the overall demand minus the entropic change of the process. The entropy contribution appears as a heat demand of the process, which depends on the partial pressure of the different compounds:   1 △Sr = △SH2 + △SO2 − △SH2 O = f T , pH2 O , pO2 , pH2 2

(3)

where T is the temperature and pi is the partial pressure of the compound i. In order to perform electrolysis, a certain voltage needs to be applied across the cell. Considering the rate-based version of Eq. 2, combined with Faraday’s law, the thermoneutral voltage (E th ) of the cell can be predicted, which is a voltage sufficient to drive the cell reaction and also maintain a constant temperature in the system. E th =

△Hr (T ) z · F

(4)

where z is the number of electrons transferred in the cell reaction and F is the Faraday’s constant. Assuming an ideal gas, the reversible potential can be expressed in terms of the Nernst equation (Eq. 5), that is the minimal electric potential that must be applied to perform the electrolysis reaction. From this expression, it can be deduced that the larger the partial pressure of steam the lower the electrical work requirement and an even larger fraction of the total energy demand can be delivered as heat.   pH2 · pO0.52 R · T 0 E rev = E rev + (5) ln z · F p H2 O   △G 0r T , p I , p I I 0 (6) E rev = z · F

250

L. Almar et al.

0 where E rev is the reversible voltage at standard conditions, pI and pII the total pressure of the hydrogen and steam chambers of the electrolyser. Therefore, the heat demand could be expressed in terms of energy or voltage as follows:

Heat demand = △Sr · T

(7)

Heat demand = (E th − E rev ) · i

(8)

Figure 2 shows the enthalpies, Gibbs energies, and their respective voltages depending on the temperature and the chambers pressures. Note that the reversible voltage should be corrected by the partial pressures as in Eq. 5, but in this case, the steam conversion should have been considered. The thermodynamic expression from Eqs. 2, 4 and 6 are plotted as a function of temperature and for different pressures on both chambers. These graphs show how △G r decreases with increasing temperature in every case, whereas the total energy demand slightly increases, meaning that there is an important increment of the entropy contribution. Thus, it is more desirable to perform high-temperature electrolysis than low temperature, because, from the thermodynamic point of view, a larger fraction of the energy demand for electrolysis can be supplied in form of heat, reducing the electrical work. Comparing the different steam chamber pressures, a slight decrease in reversible voltage is observed for larger pressures as it is expected from Eq. 5, whereas the hydrogen chamber pressure has the opposite effect. At low hydrogen chamber pressures, the reversible voltage is strongly affected by the chamber pressure, but the influence becomes less pronounced at higher pressure.

Fig. 2 Energies and the respective voltages of the electrolysis reaction in PCECs as function of the temperature, the steam and H2 chamber pressures

Protonic Ceramic Electrolysis Cells (PCECs)

251

2.2 Joule Effect Apart from the reversible potential, the cell performance is strongly influenced by ohmic losses as a consequence of the transport of protons through the electrolyte. The total resistance of the cell consists mainly of the electrolyte resistance to the transport of protons, but also the polarization resistances of the anode and cathode electrodes. The measured resistance is typically normalized to the electrode area, leading to the area-specific resistance (ASR), which is a very useful parameter when comparing different cell types and their performance. The ASR will be lower as the proton conductivity (σH ) of the cell increases, which depends on temperature and steam partial pressure. Thus, the overpotential of the resistance to proton migration through the electrolyte can be estimated as follows:   ASR T , pH2 O =

L electrolyte   σH T, pH2 O

(9)

where L electrolyte is the thickness of the electrolyte. A significant part of the energy demand, previously explained (Eq. 7), can be provided by heat coming from external sources or due to the Joule effect as a consequence of the ohmic losses. Thus, the heat released by the Joule effect can be defined as:   Q Joule = ASR T , pH2 O · i 2

(10)

Therefore, the overall heat balance of the electrolysis (Heatbalance ) could be expressed in terms of the operating cell potential (E cell ):   E cell = E rev + ASR T , pH2 O · i  Heatbalance =

demand = (Eth − E rev) · i source = ASR T , pH2 O · i 2

(11)

 = (E cell − E th ) · i

(12)

For operating voltages above thermoneutral, Eq. 12 predicts a positive heat flux to the electrolyser which would result in a local increment of temperature, whereas below thermoneutral voltage a negative heat transfer would take place, cooling down the system. Operating with a fixed current density, an increase in hydrogen partial pressure leads to a higher reversible voltage (see Eq. 5), and consequently, a decrease of the heat demanded while the heat source remains constant, so the heat balance would be more exothermic. Under this scenario, the temperature would increase up to a point where the ASR is small enough to compensate the heat balance, since the higher the temperature, the lower the ASR value. As shown in the above equations, the current density applied to the cell not only determines the amount of steam that is being split, but also plays an important role in the heat balance. Since the electrolysis reaction is endothermic, cell temperature would decrease at low current densities unless an external source is used to provide the

252

L. Almar et al.

thermal energy required to accomplish the heat balance. As current density increases, more heat is demanded for the reaction, but the thermal energy released due to the Joule effect is also higher. Once the thermoneutral voltage is overcome, a further increase in current density leads to an overheating of the cell, which is not desirable since a too high temperature might degrade the cell materials. As stated before, protonic electrolysers allow the efficient production of compressed H2 by tuning the operating pressure on the hydrogen-side chamber and the current density. Currently, H2 pressurization is achieved by using mechanical compressors such as diaphragm pumps, piston pumps, screws, and turbo compressors. In these devices, the H2 compression follows an isentropic pathway, and thus, a multi-step operation is required to reduce temperature rises and to improve the energy efficiency (Bouwman 2014; Zou et al. 2020). Otherwise, in a protonic electrolyser, the H2 pressurization occurs when it is released in the cathode at a certain pressure. The H2 cannot increase its internal energy as in a conventional compressor because the reaction takes place on the electrode surface, and due to the low thickness of the cell, it may be considered a quasi-isothermal phenomenon. Consequently, the energy efficiency of the system significantly increases, especially when the heat released associated to the Joule effect is utilized to compensate the energy demand of the endothermic electrolysis reaction thus, resulting a more cost-efficient technology than a multi-step H2 compression.

2.3 Electronic Leakage The Faradaic efficiency (ηFar ) has an important influence on the cell performance and thus, some strategies to lead to large Faradaic efficiency values should be considered to reduce as much as possible the electronic leakage. Protonic ceramics are typically not pure proton conductors; they possess mixed electron and proton conductivity. Therefore, the transport mechanism across the cell can be reduced to an electric circuit analogy, where protons and electrons are parallel rails with different resistive contributions that are competing for being the charge transported. From the heat balance point of view, both charge movements generate the Joule effect according to the next equation:   Q Joule = ASRH+ T, pH2 O · (i · ηFar )2 + ASRe− (T ) · (i · (1 − ηFar ))2

(13)

Therefore, considering the parallel distribution of the equivalent electrochemical circuit, Eq. 11 and 12 become:   E cell = E rev + ASRH+ T , pH2 O · i · ηFar  Heatbalance =

demand = (E th − E rev ) · i · ηFar source = Q Joule

(14)

= (E cell − E th · ηFar ) · i

(15)

Protonic Ceramic Electrolysis Cells (PCECs)

Electric appied

Protonic and electronic

Pure protonic

Ideal Process

253

Heat

Process

Electric appied

Heat

reversible

Process

Electric appied

Heat

reversible

Charge tranport

Charge tranport

Fig. 3 Energy balance in electrolysis operating under endothermic conditions, considering ideal (left), pure protonic (middle) and protonic-electronic (right) systems

The transference number of each charge carrier depends on the operating conditions and material composition (Duan et al. 2020). On one hand, electronic leakage in electrolysis is correlated with the electron hole transference number, and many efforts have been made over the last years trying to develop electrolyte materials that exhibit lower transference numbers (Saqib et al. 2019; Zhu et al. 2018). On the other hand, operating temperatures below 600 °C suppress electronic charge carriers, but it is not a desirable strategy from the thermodynamic point of view (larger electric work contribution and ohmic resistances). In addition, high steam concentration should be maintained at the anode side to increase proton concentration and hydration, suppressing electronic charge carriers. Figure 3 shows the energy balance in the electrolysis under different assumptions. These examples are endothermic scenarios, i.e., Joule heat does not suppose enough energy to overcome the energy demands of the process and, consequently, the temperature of the process would decrease. On the left, it is assumed an ideal ionic and electronic conductor, or, in other words, there is no Joule effect. Therefore, the entire heat demand needs to be supplied from an external source. Next, a pure protonic scenario is illustrated, where protons are being transported through the electrolyte, where the corresponding heat source may be estimated with Eq. 10. Finally, a more realistic case, involving both proton and electron transport through the cell is depicted.

3 Transport Properties of High-Temperature Electrolysis 3.1 Protonic Transport and Conductivity Transport of protons in ceramic oxides occurs via the Grotthuss mechanism through defects induced by water, based on proton rotational diffusion and hopping between neighboring oxide ions (Kreuer 1996; Marx 2006). The incorporation of protons in

254

L. Almar et al.

ceramic oxides requires the presence of oxygen vacancies, since protons are not native to their lattice structure. Protons are dissolved in the oxide lattice of the materials forming positively charged defects, following the hydration reaction (Eq. 16): H2 O + Oox + Vo.. = 2OH.o

(16)

where Vo.. represents an oxygen vacancy (+2 charged), Oox is an oxygen atom in its original lattice position (neutral charged) and OH.o is a hydroxide or proton defect (+1 charged) at a lattice oxygen site, in Kröger-Vink notation. Oxygen vacancies are commonly generated by an acceptor dopant, which is a cation substitution with a lower oxidation state than the original host cation (Marrony 2015; Meulenberg et al. 2011). The concentration of proton defects implies a weight gain and thus, can be measured by thermogravimetric analysis (TGA) as a function of temperature or water partial pressure (Duan et al. 2020; Yang et al. 2021). Thus, the thermodynamics of the hydration reaction and operating conditions determine the proton concentration as follows (Duan et al. 2020): K hyd

 0   

. 2 0 Shyd Hhyd OHo −1 = x  ..  · pH2 O = exp exp − R RT Oo Vo

(17)

0 where K hyd corresponds to the equilibrium constant of the hydration reaction and Shyd 0 and Hhyd are the standard hydration entropy and enthalpy, respectively. From Eq. 16 is stemmed a competition between oxidation and hydration for oxygen vacancies. Enhancing the hydration (large K hyd ), promotes the proton transport decreasing the electronic leakage and the suppression of parasitic oxidation reactions. Increasing oxygen vacancies concentration would promote both oxidation and hydration. On the other hand, materials with low K hyd present lower proton conductivity and thus, more oxygen vacancies for oxidation, so that greater electronic leakage. In conclusion, higher K hyd results in much higher Faradaic efficiencies.

3.2 Electronic Conductivity Proton ceramic materials present mixed conductivity between protonic and electronic conductivity. Electronic conductivity is a complex phenomenon that occurs by the electrons in the conduction band (n-type conductivity) and holes in the valence band (p-type conductivity). The electronic conductivity changes drastically across the cell, from p-type at the anode side to n-type at the cathode side. Assuming a fully hydrated electrolyte, the p-type conductivity can be expressed as (Vøllestad et al. 2019): σp =

σp0

·



O H .o



 1 4

· pO2 ·

−1 pH22O

=

σp0

· exp −

  F E − E O0 2 RT

(18)

Protonic Ceramic Electrolysis Cells (PCECs)

255

where σp0 is the standard p-type conductivity, E O0 2 the standard potential for the oxygen reduction reaction, and E the cell potential. Alternatively, the n-type conductivity can be expressed in terms of the standard potential of the hydrogen reaction:    F E − E H0 2 0 σn = σn · exp − (19) RT

4 Protonic Electrolyte Materials A wide range of crystal structures have been considered as potential proton conductors, including ABO3 -based perovskites, brownmillerite A2 B2 O5 -based materials and LnBO4 -based ortho-phosphates, ortho-niobates and ortho-tantalates. Among them, ABO3 perovskite structures have showed the highest proton conductivity and the lowest activation energy for proton transport due to the minimal distortion from the ideal cubic structure and the large crystallographic unit cell volume (Kreuer 1999). The state-of-the-art proton-conducting oxides are materials formed by barium in the A-site and cerium and/or zirconium as primary cations in the B-site (BaCeO3 and BaZrO3 -based materials).

4.1 BaCeO3 and BaZrO3 -Based Materials Iwahara et al. pioneering reported a wide variety of B-site acceptor dopants for BaCeO3 -based oxides and demonstrated its superior protonic conductivity and good sinterability (Iwahara et al. 1988, 1994). A proton conductivity of 0.05 S cm−1 at 600 °C was reported for the Gd-doped BaCeO3 compound, with a proton transport number close to 1 (Taniguchi et al. 1992). The transport numbers in BaCe0.9 M0.1 O3−δ (M=Nd, Sm, Gd, Dy, Y or Yb) were compared showing a trend of the proton transport number that increased with decreasing temperature and decreased with increasing ionic radius of trivalent dopant M, since the lattice of orthorhombic structure became more asymmetric and thus, the contribution of the oxide ions to the total conduction grows. Yttrium showed the least migration-restricting distortion, yielding the smallest lattice parameter and the highest transport number of protons (Iwahara et al. 1994). However, BaCeO3 -based oxides are chemically unstable in H2 O and CO2 containing atmospheres at the typical operation conditions (Bhide and Virkar 1999; Tanner and Virkar 1996). Ce appears to be more basic than Zr, more basic oxides tend to react with acidic gases like CO2 or even amphoteric gases like H2 O.

256

L. Almar et al.

Hence, many works started to focus on BaZrO3 -based oxides and particularly on Y-doped barium zirconate, due to its large bulk proton conductivity and chemical stability in H2 O and CO2 containing atmospheres (Katahira et al. 2000). The good bulk conductivity of BaZr0.8 Y0.2 O3−δ (BZY20) was reported decades ago measured by electrochemical impedance spectroscopy (EIS), where the bulk and grain boundary conductivities might be discerned (Kreuer 1999). In spite of this, BaZrO3 -based oxides have a refractory nature and need hightemperature sintering, resulting in significant manufacturing challenges and a smallgrained microstructure, which leads to an inconvenience due to the high density number of grain boundaries and associated large resistances, leading to a lower total proton conductivity (bulk + grain boundary), if compared to BaCeO3 -based oxides. Ricote et al. detailed these findings with total conductivity values in wet 9% H2 at 600 °C of ~10–3 and ~8 × 10–3 S cm−1 for BCZY09 and BCZY45, respectively (Ricote et al. 2012).

4.1.1

BaCe1-x-y Zry Yx O3−δ (BCZY) Related Systems

The advantages of Y-doped barium cerate (BCY) and Y-doped barium zirconate (BZY) were successfully combined, by preparing a solid solution with both of them. BaCe1−x−y Zry Yx O3−δ (BCZY) presents high proton conductivity, good chemical stability and has been widely used for protonic ceramic electrochemical devices. Systematic studies varying the Ce/Zr ratio of BCZY demonstrated that increasing the Ce content improved the proton conductivity and sinterability, while increasing the Zr content improved the chemical stability at the cost of the performance (Ricote et al. 2012; Sawant et al. 2012). During the last decades, many works focused on further improving the properties of BaCe1−x−y Zry Yx O3−δ by B-site doping strategies. Remarkable examples were achieved with Yb-doped compounds in BaCe1−x–y−z Zry Yx Ybz O3−δ , more specifically for both BaCe0.7 Zr0.1 Y0.1 Yb0.1 O3−δ (BCZYYb7111) (Yang et al. 2009) and BaCe0.4 Zr0.4 Y0.1 Yb0.1 O3−δ (BCZYYb4411) (Choi et al. 2018) compositions. Yang et al. reported for the first time very high conductivity of BCZYYb7111 at relatively low temperatures (500–700 °C) and resistance to resist deactivation by sulfur and coking (Yang et al. 2009). A conductivity of 3.3 × 10−2 S cm−1 at 600 °C was measured in oxygen with 3% H2 O. However, with higher concentrations of H2 O and CO2 , decomposition of BCZYYb were still found as reported Duan et al. (2019). Choi et al. reported enhanced CO2 tolerance, attributed to the higher Zr amount for BCZYYb4411 (Choi et al. 2018). Experimental results showed that Yb-doping improves the total conductivity compared to BZY20 and chemical stability to HCs and H2 S making sulfur and coking-tolerance cells. A conductivity of 0.03 S/cm at 600 °C was reported under humidified N2 atmosphere. Deeper experimental analysis and computational analysis is still required to fully understand the relationship between composition, structure, stability and performance of this family of materials.

Protonic Ceramic Electrolysis Cells (PCECs)

4.1.2

257

Influence of the Material Synthesis and Fabrication Process

The electrolyte materials sintering has been challenged the fabrication of easy scaleup and reproducible protonic ceramic devices. Typically, very high temperatures (1700–2000 °C) and long sintering times are required to process BZY-based materials with high density (>95%). The extreme conditions often cause Barium evaporation and as a result a detrimental lower proton conductivity of the materials have been reported (Schober and Bohn 2000), since the lower occupancy of the A-site favors the dopant substitution and consequent reduction of oxygen vacancies. The fabrication process has a strong impact in the bulk and the grain boundary resistances, explaining the large scattering of conductivity values reported in literature and well documented in (Fabbri, Pergolesi, and Traversa 2010). For instance, for BZY materials values of total proton conductivities varying from 1 × 10−6 to 1 × 10−2 S cm−1 at 600 °C for nominally identical compositions can be found (Duval et al. 2007; Fabbri et al. 2010; Pergolesi et al. 2010). Sintering the materials covered with a powder mixture of the material composition and excess of BaCO3 (~10 mass%) is usually employed as an effective method to avoid Ba evaporation leading to higher bulk conductivities (Babilo et al. 2007). On the other hand, lower sintering temperatures can be achieved by wet chemical routes or addition of nano-sized powders (Cervera et al. 2007; Khani et al. 2009; Magrez and Schober 2004; Stuart et al. 2009; Yamazaki et al. 2009) at the expense of an increase complexity and costs. Further advance of protonic ceramic materials toward an easy scalable sintering approach was achieved by solid-state reactive sintering (SSRS), which involved the sintering of simple and cost-effective precursors in a single sintering step (Duan et al. 2015; Tong et al. 2010). A total conductivity as large as 3.3 × 10−2 S cm−1 at 600 °C was measured under wet argon for the BZY20 pellets sintered using SSRS and NiO as a sintering aid.

4.1.3

Sintering Aids

Diverse sintering aids have been used as a strategy to lower the sintering temperature. Many authors used additives like CuO, NiO, ZnO, SnO, MgO, Al2 O3 TiO2 , Mo or Bi2 O3 in barium cerates, barium zirconates and doped mixtures to assess the influence of those sintering aids on the material performance (Babilo and Haile 2005; Baral and Tsur 2019; Lee et al. 2019; Wang et al. 2018). The most effective transition metal oxides improving the densification seemed to be CuO, ZnO and NiO (Duval et al. 2008; Park et al. 2015). More in detail, Park et al. (2015) compared the effect of different sintering aids concluding that the effect of sintering aids on the bulk conductivity is insignificant, while has a bigger influence on the grain boundary. Similar conclusions were achieved by other authors for the incorporation of NiO (Lee et al. 2019; Wang, Bi, and Zhao 2018) and ZnO (Babilo and Haile 2005; Baral and Tsur 2019) on BCZY electrolytes or in BZY systems (Ricote et al. 2014). Further studies investigated the effect of the NiO content on the BCZY electrolyte microstructure and fuel cell

258

L. Almar et al.

performance, demonstrating the importance of the appropriate NiO addition (Lee et al. 2019). The optimized content of NiO promotes the electrolyte densification and grain growth, while higher contents give rise to NiO segregation and fuel cell performance deterioration. Overall, the use of the appropriate content of transition metal oxides commonly employed as sintering aids produces the reduction of the sintering temperature, but it has detrimental effect on the electrical conductivity, being more significant in reducing atmospheres than in oxidizing conditions, which is ascribed to the reduction of the effective concentration on protons (Park et al. 2015). Besides the sintering aids may introduce minor secondary phases at grain boundaries or small levels of electronic conductivity, which may impact the performance of PCECs. Despite the challenges, NiO has been widely used and established as the sintering aid of choice for fabrication of PCEC devices (An et al. 2018; Duan et al. 2015, 2018).

4.2 Other Materials Other families of crystal structures exhibit proton conduction at intermediate temperatures (300–700 °C), including A2 B2 O5 -based materials, rare earth ortho-niobates, tantalates, tungstates, phosphates and pyrochlores (Bielecki et al. 2014; Haugsrud and Norby 2006; Islam et al. 2001; Jankovic, Wilkinson, and Hui 2012; Murugaraj et al. 1997; Othman et al. 2007; Toyoura et al. 2012). Few of these works addressed the fabrication and characterization of these materials for fuel cell applications (Magrasó et al. 2014, 2010; Solís et al. 2012; Solís and Serra 2011; Song et al. 2019). Incompatibilities with other cell components or lower conductivity than ABO3 -based perovskites led to reduced efforts to develop their applicability as potential electrolyte candidates for electrolysis cells (Bi et al. 2014).

5 Electrodes for PCECs The materials usually employed for electrodes in PCEC are neither purely ionic (proton and/or oxygen ion) nor electronic conductors and possess mixed H+ , O2− and e− conductivities, being the transport number of the species dependent on the operation conditions, i.e., temperature, atmosphere, overpotential and current density applied (Duan et al. 2020). One of the main handicaps in the development of electrodes for PCECs is the lack of materials with the appropriate mixed protonic and electronic conductivity (MPEC). Therefore, efforts have been focused on the development of different cell configurations able to perform electrolysis with a high efficiency. Electrodes should accomplish the following requirements in order to reduce the electrodes’ overpotential and obtain low polarization resistances:

Protonic Ceramic Electrolysis Cells (PCECs)

259

• Stability, electrodes should be chemically, morphologically and dimensionally stable in the operation conditions and along the fabrication process. • Conductivity, high electronic and ionic and/or protonic conductivity is required. • Porosity, enough porosity is desired (>30%) to allow a fast gas transport from/to the active reaction sites. • Thermal expansion coefficient (TEC), the TEC of electrodes should match the TEC of the electrolyte and adjacent components. • Compatibility, electrodes should be compatible with other components in the cell in the operation and fabrication conditions. • Catalytic activity, electrodes must have sufficient catalytic activity (low polarization resistance) for the redox reactions.

5.1 State-of-the-Art Steam Electrodes (Anodes) Normally, the activity of the anode toward the steam electroreduction reaction limits the PCEC overall efficiency leading to a high anode overpotential. The sluggish kinetics at the operation temperatures ( 1 V and an increased power output owing to minimized electrolyte thickness (Jang et al. 2017). Schlupp et al. have successfully implemented AACVD for the fabrication of GDC (150 nm)-YSZ bilayers, integrating them in anode-supported SOFCs and demonstrating high power density above 850 mW cm−2 at 650 °C. (Schlupp et al. n.d.) An alternative way for the fabrication of thin film-based SOCs is based on the development of ad-hoc supports, including silicon micromachined platform or porous anodized aluminum oxide (AAO). Such a strategy has been tested for the fabrication of energy microsystems (microSOCs). However, while microSOCs have been explored as fuel cells (Lee et al. 2018) microSOEC remain practically unexplored to date. In the case of Si support, Si microelectromechanical processes (MEMS) are employed in order to fabricate a free-standing membrane of the active structure. The resulting device is characterized by minimized electrolyte thickness and has been proven to be advantageous for enhanced power densities: Prinz et al. demonstrated the feasibility of a Si-based microSOFC device having up to 1.3 W cm−2 at 450 °C, employing a thin-film 3D free-standing electrolyte architecture and nanogranular catalytic interlayer at the cathode/electrolyte interface. A compilation of other results can be found, e.g., in ref. (Evans et al. 2009). Among the main drawbacks of Sibased microSOFCs, one should include the complexity of Si manufacturing based on cleanroom microfabrication processes and the possibility of device shortage due to electrolyte pinholes. Moreover, while a strongly reduced electrolyte thickness can ensure sufficient ionic conductance even a low temperature, electrode activity limitations come into play. Typical solutions employing metal-based electrodes make the long-term stability of microSOFC devices questionable. In this sense, the work by Garbayo et al. assumes relevance as it proposes a full-ceramic device employing LSC and SDC-based electrodes (Garbayo et al. 2014). A total area-specific resistance of 3.5 Ω cm−2 , resulting in a SOFC power density of 100 mW cm−2 at 750 °C, has been reported. AAO support represents a cost-effective solution which does not require MEMS (unlike Si-based microSOFCs) (Kerman et al. 2013) and that can be more easily fabricated at the industrial level. AAO is characterized by a highly porous structure with straight channels (tunable pore size in the range of tens to hundreds of nanometers) across the support thickness. Moreover, the similar thermal expansion coefficient of Al2 O3 (≈ 9 × 10–6 K−1 ) and YSZ (≈ 10.5 × 10–6 K−1 ) limits potential issues caused by thermal stresses. The major technological challenges here are related to fabricating

Emerging Trends in Solid Oxide Electrolysis Cells

355

fully gas-tight electrolytes—which typically require an increased electrolyte thickness and a smaller pore size—and to obtaining a suitable anode architecture with adequate TPB density. Kwon et al. have for example fabricated a high-quality cell (OCV 1.02 V) with a 900 nm electrolyte, resulting in a peak cell power density of 110 mW cm−2 at 450 °C (Kwon et al. 2011). More recently, Cha et al. have introduced a combinatorial method to deposit a dense, pinhole-free nanoscale thin-film YSZ electrolyte by combining atomic layer deposition (ALD) and sputtering (Ha et al. 2013). Full gas tightness (OCV 1.14 V) was achieved for electrolyte thickness < 400 nm. Using Pt electrodes, a peak power density of 180 mW at 450 °C was obtained. In order to achieve a reduction of the electrolyte thickness, Ji et al. have implemented plasma-enhanced ALD, in which the reactants are excited by a plasma source for higher reactivity (Ji et al. 2015). Following this approach, a fully dense YSZ layer was obtained with a thickness of only 70 nm (OCV 1.17 V at 500 °C). The so-obtained full AAO-supported cell shows a power peak density > 150 mW cm−2 at 550 °C. An et al. (Oh et al. 2018) have recently reported a high peak power density of 562 mW cm−2 at 450 °C for an AAO-supported cell comprising an SDC-based electrolyte fabricated by sputtering. In order to improve the gas tightness, as well as to limit the reduction of the ceria, thin YSZ interlayer (deposited by ALD) has been embedded in the electrolyte structure, ensuring a constant OCV > 1 V. In a later work by same authors, a CeO2 cathode overcoating by ALD allowed further boost of the performance, resulting in a power density of 800 mW cm−2 at 500 °C (Shin et al. 2019).

6 High Entropy Oxides for SOCs The high entropy oxides (HEOs) are part of a new class of materials, so-called high entropy materials (HEMs), that is gaining interest for the past fifteen years in order to obtain properties that allow an improvement of existing technologies and the development of new ones (Cantor et al. 2004). Those materials are based on a structural stabilization of a highly doped or substituted structure, where five or more elements with equal or close to equal molar fractions are introduced, due to the attained high entropy configuration. Apart from the phase stability, the combination of various elements of different sizes and valence electron distribution leads to severe lattice distortion impacting on some properties such as diffusion, catalytic activity, conductivity, magnetism, and thermal conductivity (Akrami et al. 2021; Ma et al. 2021). High entropy alloys were the first type of HEMs to be studied before Rost et al. proved the concept of HEOs in 2015 by stabilizing a rock salt structure above 900 °C, mixing MgO, NiO, ZnO, CuO, and CoO in equimolar proportions (Rost et al. 2015). In the field of SOEC, other types of HEO phase like fluorite, perovskite, or spinel HEO, which will be presented in the following sections, can be of high interest to be used as electrolyte or electrode materials. Components for SOEC must be stable upon temperature and polarization and present high ionic or electronic or

356

A. Tarancón et al.

mixed conductivities. State-of-the-art materials are multi-doped structures like yttriastabilized zirconia (ZrO2 Y2 O3 , YSZ), Ce1x Gdx O2 (CGO), or La1−x Srx Ga1−y Mgy O3 (LSGM) as electrolytes and La1x Srx MnO3 (LSM), La1−x Srx Co1−y Fey O3 (LSCF), La1−x Srx CoO3 (LSC), or Sr2 Fe1.5 Mo0.5 O6 (SFM) oxygen electrodes, among others, which can present some stability issues by decomposing, reacting at the interlayers or being deactivated by external pollutants (Chen et al. 2016). In this sense, HEOs present a high potential to overcome such degradation issues as already shown by recent works where Sr segregation could be avoided (Baiutti et al. 2021; Shen et al. 2021; Yang et al. 2021b). This section aims to present from a theoretical point of view the HEOs structures that could be of interest for SOECs and the results already obtained in terms of electrochemical performance and durability.

6.1 Theoretical Concepts and Principles of HEOs The stability of a solid solution is defined by the minimization of the Gibbs free energy (△Gmix ) of the mixture that is composed by the enthalpy (△H mix ) and the entropy (△S mix ) of the mixture according to Eq. 3. G mix = Hmix − T Smix

(3)

where T is the absolute temperature. It is clear that △H mix will stabilize the solid solution at low temperature whereas the entropy can predominate the free energy at elevated temperatures. The ideal entropy of a solid solution is a function of the atomic fraction of the composing elements x i given by Eq. 4 (Murty et al. 2019). Smix = −R

N ∑

xi ln xi

(4)

i=1

where R is the ideal gas constant and N is the number of components. In the case of a ceramic structure with anion and cation sites, Eq. 2 can be simply re-written into the following Eq. 5 (Sarkar et al. 2018, 2020).

Smix

⎡( ) M ∑ = −R ⎣ xi ln xi i=1

cation−site

⎛ +⎝

N ∑ j=1

⎞

⎤

x j ln x j ⎠

⎦

(5)

anion−site

Materials are then considered HEMs when their entropy △S mix is equal or higher than 1.5R. This value can be easily reached on systems with 5 or more elements and is maximized for equimolar distribution as seen on Fig. 15a. It is worth mentioning that entropy stabilization can also occur in medium-entropy ceramics (compositions of 3 or 4 cations with 1.0R ≤ △S mix < 1.5R) (Gao et al. 2016a).

Emerging Trends in Solid Oxide Electrolysis Cells

357

Fig. 15 a Configurational entropy calculated as a function of the molar fraction of the Nth component, the N−1 components being equimolar (Rost et al. 2015), b Maximum entropy obtained for N elements in equimolar distribution (Akrami et al. 2021), and c XRD patterns of an equimolar mixture of 5 elements that stabilizes in a rock-salt structure above 850 ºC (Rost et al. 2015)

As previously mentioned, the single-phase stabilization is one criterion to define HEOs. This property can be studied by high temperature X-ray diffraction (HTXRD) or by quenching the samples after being heated up to the desired temperature. Latter was used on Fig. 15c where an oxide containing five elements was heated up to 1000 °C and cooled down to 750 °C in order to observe any reversibility in the stabilization process. The results clearly evidence that a single rock salt structure is obtained and stable above 850 °C due to the entropy contribution and is then destabilized when decreasing the temperature (Rost et al. 2015). Temperature of stabilization will be different for each compound, and pure phase can even be obtained at room temperature like shown by Sarkar et al. with a (5A0.2 )(5B0.2 )O3 perovskite (Sarkar et al. 2018b). The value of this stabilization temperature will define the suitability of using HEOs as materials for SOEC that are expected to operate in a range of 650–900 °C.

358

A. Tarancón et al.

Fig. 16 High-entropy ceramics crystal structures used in energy-related fields (Ma et al. 2021)

6.2 HEOs with Relevant Structures for SOCs Typical SOCs material structures have already shown their ability to be stabilized by entropy as presented in Fig. 16. Several examples of fluorites and perovskites and even on spinel hosting five or more cations are reported in the following sections. Structures like rock-salts or layered O3-type have been giving promising results in the energy field, as materials for batteries (Ma et al. 2021).

6.2.1

Fluorite

Fluorite HEOs were successfully synthetized during the past years by solid-state reactions (Chen et al. 2018a, 2021a; Gild et al. 2018; D˛abrowa et al. 2020a; Bonnet et al. 2021; Zhang et al. 2022) and wet chemical synthesis (Anandkumar et al. 2019; Spiridigliozzi et al. 2020; Bonnet et al. 2021; Wen et al. 2021). In both type of synthesis route, the parent oxides for solid-state reactions or the salts for wet chemical synthesis are simply mixed in appropriate amounts according to the desired final composition before following with the regular synthesis and sintering steps. Those fluorite HEOs consist of three to seven cationic elements mixed in equimolar and non-equimolar proportions, including tetravalent elements such as Zr, Hf, and Ce and trivalent ones like Gd, Y, Yb, La, Nd, or Sm in order to maximize the number of oxygen vacancies. They crystallize into a Fm-3 m cubic symmetry with some slight changes in their lattice parameter depending on the composition, especially when trivalent ions are present. Unfortunately, at the light of the published results none of the synthesized fluorite HEOs that were electrically characterized (c.a. 25 components) presented an ionic

Emerging Trends in Solid Oxide Electrolysis Cells

359

Fig. 17 Conductivity of some high entropy fluorite oxides compared with 8YSZ and CGO fluorites. Data reproduced from the following references (Zhang et al. 2017, 2022; Gild et al. 2018; D˛abrowa et al. 2020a; Bonnet et al. 2021)

conductivity higher than the state-of-the-art electrolytes such as YSZ or CGO. They are actually one or several orders of magnitude lower than expected at the targeted operating temperature for SOEC as seen on Fig. 17 (Gild et al. 2018; D˛abrowa et al. 2020a; Bonnet et al. 2021; Zhang et al. 2022). To understand this behavior and try to maximize the ionic conductivity, Bonnet et al. synthetized new compositions based on their entropy and the ionic size difference. Those compounds are also reported on Fig. 17 and did not reach the values of the classical fluorite materials used in SOEC. It appears that neither entropy nor the ionic size difference seems to have a linear relation with the ionic conductivity. The reason of a decrease in performance might be related to the disorder that would provoke local crystal network perturbations that decrease the oxygen transport (Bonnet et al. 2021). Further work will be required to find combinations that might be suitable for SOEC. In this sense, models like Kinetic Monte Carlo (KMC) or Density Functional Theory (DFT) can be useful tools to predict the energy of the defects and screen a large amount of composition.

6.2.2

Perovskite

Different combination of multiple cations into the A-site and B-site of ABO3 perovskite has been tried since 2018 by placing in some cases 5 elements on the

360

A. Tarancón et al.

A-site (5ABO3 ) (Sarkar et al. 2018; Shi et al. 2021; Yang et al. 2021b), in some other studies 5 elements on the B-site (A5BO3 ) (Sarkar et al. 2018; D˛abrowa et al. 2020; Han et al. 2021) and finally, 5 types of atoms on each site (5A5BO3 ) reaching a entropy value of 3.22R (Sarkar et al. 2018b). Some medium entropy perovskites were also studied (Baiutti et al. 2021; Shen et al. 2021). Most of them were produced by modified Pechini sol–gel and combustion method (D˛abrowa et al. 2020b; Han et al. 2021; Shen et al. 2021; Shi et al. 2021; Yang et al. 2021a, b). Nebulized spray pyrolysis also allowed to fabricate some of those compounds (Sarkar et al. 2018; Witte et al. 2019) while one of them resulted from the interdiffusion between vertical aligned grains of La0.8 Sr0.2 MnO3 and Ce0.8 Sm0.2 O2 obtained by pulsed lased deposition from a target made by mixing the powders (Baiutti et al. 2021). Crystal structure of obtained compounds was found to vary from orthorhombic (Pbnm), rhombohedral (R 3 c), or hexagonal symmetry, having a Goldschmidt’s tolerance factor t comprised between 0.78 and 1.05, which corresponds to perovskite structures (Sarkar et al. 2018b). Sarkar et al. found out that for systems with smaller cations on the A-site and/or larger cations on the Bsite, a large metric distortion from cubic symmetry was happening, as shown in Fig. 18a, along with strong tilting of the BO6 polyhedra and a lowering of the effective coordination number of the A-site cation. Nevertheless, this lower symmetry perovskite phase could be stabilized. Especially, in the case of the 10-cationic system ((Gd0.2 La0.2 Nd0.2 Sm0.2 Y0.2 )(Co0.2 Cr0.2 Fe0.2 Mn0.2 Ni0.2 )O3 ), pure orthorhombic phase was observed from room temperature to 1000 °C with random and homogeneous element distribution (Sarkar et al. 2018b). Up to now no clear rules have been determined to predict the stability of those new compounds. In order to determine the suitability of those new compounds to be used as electrodes for Solid Oxide cells, they must be compatible with the electrolyte materials, present good conductivity and stability to obtain high power density and long-term durability under operating conditions. Table 3 presents the compounds characterized in the literature as oxygen electrode materials. Note that only few compounds have been fully characterized and none of them have been yet tested under operation for more than 100 h. However, some interesting results can already be pointed out. The highest conductivity was obtained for a 6A4BO3 compound with a value of 635 S·cm−1 at 800 °C, which is comprised between typical LSCF and LSC electrode materials (Yang et al. 2021a). Han et al. obtained better results with the La5BO3 HEO (La(Mn0.2 Fe0.2 Co0.2 Ni0.2 Cu0.2 )O3 ) than with the LSM electrode manufactured and operated in the same conditions (Han et al. 2021). Actually, the measured conductivity and the one obtained by Shi et al. and Yang et al. with 5AMnO3 compounds are in the range of typical LSM values (c.a. 102 S·cm−1 ) (Jiang 2008; Shi et al. 2021, Yang et al. 2021b). The resulting performance in full cell configuration is promising for the tested materials, reaching up to 0.95 A cm−2 at 0.7 V and 800 °C in fuel cell mode for the (La0.2 Pr0.2 Nd0.2 Sm0.2 Sr0.2 )MnO3 compound (Yang et al. 2021b). An improvement of the stability of some of the studied HEO perovskites toward Sr segregation was also observed (Baiutti et al. 2021; Shen et al. 2021; Yang et al. 2021b). Shen et al. found that an equimolar medium-entropy oxide is more efficient to block the migration of Sr to the surface than an equivalent compound with

Emerging Trends in Solid Oxide Electrolysis Cells

361

(a)

(b)

Fig. 18 (a) Crystal structure of an ideal perovskite (Pm3m) and resulting orthorhombic structure with tilted polyhedra for different high-entropy perovskites (Sarkar et al. 2018b)

non-equimolar proportion of the cations, which agrees with an entropy stabilization (Shen et al. 2021). This might be due to the lattice distortion promoted by the presence of various elements of different sizes, which results in a disordered stress field formed around Sr that can limit his transport rate (Baiutti et al. 2021; Yang et al. 2021b). Those results might open the way of simplifying the fuel cells architecture by avoiding the use of the barrier layer placed between the Zr-based electrolyte and the Sr-containing oxygen electrode to block the formation of SrZrO3 insulating phase.

6.2.3

Spinel

Up to date, only one high entropy spinel has been recently reported at the date of writing this chapter by Shijie et al. (2022). The (5A)2 BO4 spinel with the composition (La0.2 Pr0.2 Nd0.2 Sm0.2 Gd0.2 )2 CuO4 was synthetized by sol–gel method and crystalized in a single tetragonal phase. On the contrary of La2 CuO4 , this compound presented high compatibility with CGO after 24 h at 1000 °C and high conductivity (64.5 S cm−1 at 800 °C). Good electrochemical performances in SOFC mode were obtained corresponding to 0.6 A cm−2 at 0.7 V and 700 °C.

La0.7 Sr0.3 (Co0.2 Cr0.2 Fe0.2 Mn0.2 Ni0.2 )O3

(La0.2 Pr0.2 Nd0.2 Sm0.2 Sr0.2 )MnO3

D˛abrowa et al. (2020b)

Yang et al. (2021b)

Yang et al. (2021a)

(La0.2 Pr0.2 Nd0.2 Sm0.2 Ba0.1 Sr0.1 )(Co0.2 Fe0.6 Ni0.1 Cu0.1 )O3

Han et al. La(Mn0.2 Fe0.2 Co0.2 Ni0.2 Cu0.2 )O3 (2021)

Compound

Refs.

–

No reaction with YSZ and CGO below 900 °C

–

No reaction with LSGM below 1100 °C

Reactivity with electrolyte

635 S·cm−1 at 800 °C

171 S·cm−1 at 800 °C

Rp = Rp (LSM) at 800 °C

12.6 S·cm−1 at 800 °C

Conductivity

–

–

Stable over 70 h under 200 mA·cm−2 at 800 °C

–

Material stability

Table 3 Characteristics of the high entropy perovskite oxides that were studied as potential oxygen electrode materials Long-term operation

0.9 A·cm−2 – at 800 °C 0.35 A·cm−2 at 700 °C (continued)

0.7 A·cm−2 50 h at 0.27 A·cm−2 and at 800 °C 0.5 A·cm−2 700 °C at 700 °C

0.95 100 h at 0.7 V and A·cm−2 at 700 °C 800 °C 0.4 A·cm−2 at 700 °C

0.5 A·cm−2 – at 850 °C 0.2 A·cm−2 at 700 °C

Current density at 0.7 V (SOFC)

362 A. Tarancón et al.

Compound

(La0.2 Nd0.2 Pr0.2 Sr0.2 Ba0.2 )MnO3

(La0.2 Nd0.2 Sm0.2 Ca0.2 Sr0.2 )MnO3

Shi et al. (2021)

Shi et al. (2021)

Shen et al. Sr(Fe0.5 Ti0.2 Co0.2 Mn0.1 )O3 (2021)

Shen et al. Sr(Fe0.25 Ti0.25 Co0.25 Mn0.25 )O3 (2021)

Refs.

Table 3 (continued) Long-term operation

145 S·cm−1 at 800 °C

No reaction 200 S·cm−1 with YSZ up at 800 °C to 1400 °C

No reaction with YSZ below 1200 °C

–

–

Sr segregation after 100 h

–

–

–

–

1.6 A·cm−2 – at 850 °C 0.8 A·cm−2 at 700 °C

Current density at 0.7 V (SOFC)

No reaction 18 S·cm−1 at with LSGM 800 °C No reaction with SDC up to 1000 °C Reaction with YSZ

Material stability

Stable over 140 h 1.3 A·cm−2 – at 850 °C 0.6 A·cm−2 at 700 °C

Conductivity

No reaction 18 S·cm−1 at with LSGM 800 °C No reaction with SDC up to 1000 °C Reaction with YSZ

Reactivity with electrolyte

Emerging Trends in Solid Oxide Electrolysis Cells 363

364

A. Tarancón et al.

7 Outlook and Future Perspective This chapter was devoted to present different emerging trends in the field of Solid Oxide Cells that could give rise to the next generation of SOECs. The recent development of new approaches and techniques that allow ceramics structuring at different length scales (from the atomic to the cm-scale) is considered a major opportunity to enhance SOC technologies in the short- to mid-term future. In this regard, 3D printing is the most promising candidate to define new architectures that will increase the power density of cells and stacks while thin film-based intermediate layers and exsolution of nanocatalysts have already proved to be very efficient in solving currently existing limitations of SOCs. Although several pioneering works showing a successful application of novel methodologies for fabricating multiscale structured cells are presented in this chapter, the field is still in its infancy and further developments will undoubtedly come in the next years. Among other unsolved challenges that will foster the implementation of such methodologies, one can find the multi-material 3D printing of complete devices, the full understanding of the stability of thin films under operation conditions, the exsolution of nanoparticles at the oxygen electrode, or the impact of the use of HEOs on the long-term stability of SOECs. Facing these and other existing challenges will drive to relevant breakthroughs in the field of SOCs increasing their performance and durability while reducing the use of critical raw materials. Acknowledgements The authors want to acknowledge the Generalitat de Catalunya (2017 SGR 1421, NANOEN) for the received funding. The authors also received funding from the national project 3DPASSION (RETOS INV, PID2019-107106RB-C31) and from the NewSOC Project (ref. 874577) supported through the European Commission’s Fuel Cells Hydrogen Joint Undertaking.

References Acosta M, Baiutti F, Tarancón A, MacManus-Driscoll JL (2019) Nanostructured materials and interfaces for advanced ionic electronic conducting oxides. Adv Mater Interfaces 6:1900462 Adler SB, Lane JA, Steele BCH (1996) Electrode kinetics of porous mixed-conducting oxygen electrodes. J Electrochem Soc 143:3554–3564 Adler SB (2004) Factors governing oxygen reduction in solid oxide fuel cell cathodes. Chem Rev 104:4791–4843 Aguiar P, Adjiman CS, Brandon NP (2005) Anode-supported intermediate-temperature direct internal reforming solid oxide fuel cell: II. Model-based dynamic performance and control. J Power Sources 147:136–147 Akrami S, Edalati P, Fuji M, Edalati K (2021) High-entropy ceramics: review of principles, production and applications. Mater Sci Eng R Rep 146:100644 Ali A, Wen X, Nandakumar K, Luo J, Chuang KT (2008) Geometrical modeling of microstructure of solid oxide fuel cell composite electrodes. J Power Sources 185:961–966 Ali S, Sørensen K, Nielsen MP (2020) Modeling a novel combined solid oxide electrolysis cell (SOEC)—biomass gasification renewable methanol production system. Renew Energy 154:1025–1034

Emerging Trends in Solid Oxide Electrolysis Cells

365

Almar L, Andreu T, Morata A, Tarancón A (2011) Mesoporous NiO-CGO obtained by hard template as high surface area anode for IT-SOFC. ECS Trans 35:1647–1654 Almar L, Andreu T, Morata A, Torrell M, Yedra L, Estradé S, Peiró F, Tarancón A (2014) Highsurface-area ordered mesoporous oxides for continuous operation in high temperature energy applications. J Mater Chem A 2:3134–3141 Almar L, Morata A, Torrell M, Gong M, Liu M, Andreu T, Tarancón A (2016) Synthesis and characterization of robust, mesoporous electrodes for solid oxide fuel cells. J Mater Chem A 4:7650–7657 Almar L, Morata A, Torrell M, Gong M, Andreu T, Liu M, Tarancón A (2017) A durable electrode for solid oxide cells: mesoporous Ce0.8 Sm0.2 O1.9 scaffolds infiltrated with a Sm0.5 Sr0.5 CoO3-δ catalyst. Electrochim Acta 235:646–653 Ambrosi A, Shi RRS, Webster RD (2020) 3D-printing for electrolytic processes and electrochemical flow systems. J Mater Chem A 8:21902–21929 Anandkumar M, Bhattacharya S, Deshpande AS (2019) Low temperature synthesis and characterization of single phase multi-component fluorite oxide nanoparticle sols. RSC Adv 9:26825–26830 Anelli S, Baiutti F, Hornés A, Bernadet L, Torrell M, Tarancón A (2019) Improved mesostructured oxygen electrodes for highly performing solid oxide cells for co-electrolysis of steam and carbon dioxide. J Mater Chem A 7:27458–27468 Anelli S, Moreno-Sanabria L, Baiutti F, Torrell M, Tarancón A (2021) Solid oxide cell electrode nanocomposites fabricated by inkjet printing infiltration of ceria scaffolds. Nanomaterials 11:3435 Anelli S, Rosa M, Baiutti F, Torrell M, Esposito V, Tarancón A (2022) Hybrid-3D printing of symmetric solid oxide cells by inkjet printing and robocasting. Addit Manuf 51:102636 Arrivé C, Delahaye T, Joubert O, Gauthier G (2013) Exsolution of nickel nanoparticles at the surface of a conducting titanate as potential hydrogen electrode material for solid oxide electrochemical cells. J Power Sources 223:341–348 Baiutti F, Chiabrera F, Acosta M, Diercks D, Parfitt D, Santiso J, Wang X, Cavallaro A, Morata A, Wang H, Chroneos A, MacManus-Driscoll J, Tarancon A (2021) A high-entropy manganite in an ordered nanocomposite for long-term application in solid oxide cells. Nat Commun 12:1–11 Berges C, Wain A, Andújar R, Naranjo JA, Gallego A, Nieto E, Herranz G, Campana R (2021) Fused filament fabrication for anode supported SOFC development: towards advanced, scalable and cost-competitive energetic systems. Int J Hydrogen Energy 46:26174–26184 Bietsch A, Zhang J, Hegner M, Lang HP, Gerber C (2004) Rapid functionalization of cantilever array sensors by inkjet printing. Nanotechnology 15:873–880 Blank DHA, Dekkers M, Rijnders G (2013) Pulsed laser deposition in twente: from research tool towards industrial deposition. J Phys D Appl Phys 47:034006 Bonnet E, Grenier JC, Bassat JM, Jacob A, Delatouche B, Bourdais S (2021) On the ionic conductivity of some zirconia-derived high-entropy oxides. J Eur Ceram Soc 41:4505–4515 Brady GA, Halloran JW (1997) Stereolithography of ceramic suspensions. Rapid Prototyping J 3:61–65 Brisse A, Schefold J, Zahid M (2008) High temperature water electrolysis in solid oxide cells. Int J Hydrogen Energy 33:5375–5382 Brown KA, Brittman S, Jariwala, D, Celano U (2020) Machine learning in nanoscience: big data at small scales Browne MP, Redondo E, Pumera M (2020) 3D printing for electrochemical energy applications. Chem Rev 120:2783–2810 Burnat D, Kontic R, Holzer L, Steiger P, Ferri D, Heel A (2016) Smart material concept: reversible microstructural self-regeneration for catalytic applications. J Mater Chem A 4:11939–11948 Cai JW, Zhang BC, Zhang MH, Wen YJ, Qu XH (2021) Indirect 3D printed ceramic: a literature review. J Cent South Univ 28:983–1002 Cai Q, Adjiman CS, Brandon NP (2011) Modelling the 3D microstructure and performance of solid oxide fuel cell electrodes: computational parameters. Electrochim Acta 56:5804–5814

366

A. Tarancón et al.

Cantor B, Chang ITH, Knight P, Vincent AJB (2004) Microstructural development in equiatomic multicomponent alloys. Mater Sci Eng A 375–377:213–218 Capel AJ, Edmondson S, Christie SDR, Goodridge RD, Bibb RJ, Thurstans M (2013) Design and additive manufacture for flow chemistry. Lab Chip 13:4583–4590 Carolan MF, Michaels JN (1987) Chemical vapor deposition of yttria stabilized zirconia on porous supports. Solid State Ionics 25:207–216 Cavallaro A, Pramana SS, Ruiz-Trejo E, Sherrell PC, Ware E, Kilner JA, Skinner SJ (2018) Amorphous-cathode-route towards low temperature SOFC. Sustain Energy Fuels 2:862–875 Cebollero JA, Lahoz R, Laguna-Bercero MA, Larrea A (2017) Tailoring the electrode-electrolyte interface of solid oxide fuel cells (SOFC) by laser micro-patterning to improve their electrochemical performance. J Power Sources 360:336–344 Cebollero JA, Laguna-Bercero MA, Lahoz R, Silva J, Moreno R, Larrea A (2019) Optimization of laser-patterned YSZ-LSM composite cathode-electrolyte interfaces for solid oxide fuel cells. J Eur Ceram Soc 39:3466–3474 Cell3Ditor (n.d.) Cell3Ditor project on 3D printing tech for SOFC stacks (2016) Fuel Cells Bull 2016:12 Chang I, Ji S, Park J, Lee MH, Cha SW (2015) Ultrathin YSZ coating on Pt cathode for high thermal stability and enhanced oxygen reduction reaction activity. Adv Energy Mater 5:1402251 Chanthanumataporn M, Hui J, Yue X, Kakinuma K, Irvine JTS, Hanamura K (2019) Electrical reduction of perovskite electrodes for accelerating exsolution of nanoparticles. Electrochim Acta 306:159–166 Chelmehsara ME, Mahmoudimehr J (2018) Techno-economic comparison of anode-supported, cathode-supported, and electrolyte-supported SOFCs. Int J Hydrogen Energy 43:15521–15530 Chen XJ, Chan SH, Khor KA (2004) Simulation of a composite cathode in solid oxide fuel cells. Electrochim Acta 49:1851–1861 Chen K, Ai N, Jiang SP (2012a) Reasons for the high stability of nano-structured (La,Sr)MnO3 infiltrated Y2 O3 –ZrO2 composite oxygen electrodes of solid oxide electrolysis cells. Electrochem Commun 19:119–122 Chen Y, Bunch J, Li T, Mao Z, Chen F (2012b) Novel functionally graded acicular electrode for solid oxide cells fabricated by the freeze-tape-casting process. J Power Sources 213:93–99 Chen K, Ai N, Jiang SP (2014) Performance and structural stability of Gd0.2 Ce0.8 O1.9 infiltrated La0.8 Sr0.2 MnO3 nano-structured oxygen electrodes of solid oxide electrolysis cells. Int J Hydrogen Energy 39:10349–10358 Chen K, Jiang SP (2016) Review—materials degradation of solid oxide electrolysis cells. J Electrochem Soc 163:F3070–F3083 Chen K, Pei X, Tang L, Cheng H, Li Z, Li C, Zhang X, An L (2018a) A five-component entropystabilized fluorite oxide. J Eur Ceram Soc 38:4161–4164 Chen Z, Ouyang J, Liang W, Yan ZC, Stadler F, Lao C (2018b) Development and characterizations of novel aqueous-based LSCF suspensions for inkjet printing. Ceram Int 44:13381–13388 Chen Z, Li Z, Li J, Liu C, Lao C, Fu Y, Liu C, Li Y, Wang P, He Y (2019) 3D printing of ceramics: a review. J Eur Ceram Soc 39:661–687 Chen K, Jiang SP (2020) Surface segregation in solid oxide cell oxygen electrodes: phenomena, mitigation strategies and electrochemical properties. Electrochem Energy Rev 3:730–765 Chen K, Ma J, Tan C, Li C, An L (2021a) An anion-deficient high-entropy fluorite oxide with very low density. Ceram Int 47:21207–21211 Chen Z, Sun X, Shang Y, Xiong K, Xu Z, Guo R, Cai S, Zheng C (2021b) Dense ceramics with complex shape fabricated by 3D printing: a review. J Adv Ceram 10:195–218 Cheng Y, Raman AS, Paige J, Zhang L, Sun D, Chen MU, Vojvodic A, Gorte RJ, Vohs JM (2019) Enhancing oxygen exchange activity by tailoring perovskite surfaces. J Phys Chem Lett 10:24 Cho S, Kim YN, Lee J, Manthiram A, Wang H (2012) Microstructure and electrochemical properties of PrBaCo2 O5+δ /Ce0.9 Gd0.1 O1.95 vertically aligned nanocomposite thin film as interlayer for thin film solid oxide fuel cells. Electrochim Acta 62:147–152

Emerging Trends in Solid Oxide Electrolysis Cells

367

Choi S, Sengodan S, Park S, Ju YW, Kim J, Hyodo J, Jeong HY, Ishihara T, Shin J, Kim G (2016) A robust symmetrical electrode with layered perovskite structure for direct hydrocarbon solid oxide fuel cells: PrBa0.8 Ca0.2 Mn2 O5+δ . J Mater Chem A 4:1747–1753 Coddet P, Amany ML, Vulliet J, Caillard A, Thomann AL (2019) YSZ/GDC bilayer and gradient barrier layers deposited by reactive magnetron sputtering for solid oxide cells. Surf Coat Technol 357:103–113 Connor PA, Yue X, Savaniu CD, Price R, Triantafyllou G, Cassidy M, Kerherve G, Payne DJ, Maher RC, Cohen LF, Tomov RI, Glowacki BA, Kumar RV, Irvine JTS (2018) Tailoring SOFC electrode microstructures for improved performance. Adv Energy Mater 8:1800120 Coppola N, Polverino P, Carapella G, Sacco C, Galdi A, Ubaldini A, Vaiano V, Montinaro D, Maritato L, Pianese C (2018) Structural and electrical characterization of sputter-deposited Gd0.1 Ce0.9 O2−δ thin buffer layers at the Y-stabilized zirconia electrolyte interface for IT-solid oxide cells. Catalysts 8 Coppola N, Polverino P, Carapella G, Sacco C, Galdi A, Montinaro D, Maritato L, Pianese C (2020) Optimization of the electrical performances in solid oxide fuel cells with room temperature sputter deposited Gd0.1 Ce0.9 O1.95 buffer layers by controlling their granularity via the in-air annealing step. Int J Hydrogen Energy 45:12997–13008 Coppola N, Polverino P, Carapella G, Ciancio R, Rajak P, Dario M, Martinelli F, Maritato L, Pianese C (2021) Large area deposition by radio frequency sputtering of Gd0.1 Ce0.9 O1.95 buffer layers in solid oxide fuel cells: structural, morphological and electrochemical investigation. Materials 14:5826 Cost-effective and flexible 3D printed SOFC stacks for commercial applications (n.d.) D˛abrowa J, Szymczak M, Zajusz M, Mikuła A, Mo´zdzierz M, Berent K, Wytrwal-Sarna M, Bernasik ´ A, Stygar M, Swierczek K (2020a) Stabilizing fluorite structure in ceria-based high-entropy oxides: influence of Mo addition on crystal structure and transport properties. J Eur Ceram Soc 40:5870–5881 D˛abrowa J, Olszewska A, Falkenstein A, Schwab C, Szymczak M, Zajusz M, Mo´zdzierz M, Mikuła ´ A, Zieli´nska K, Berent K, Czeppe T, Martin M, Swierczek K (2020b) An innovative approach to design SOFC air electrode materials: high entropy La1−x Srx (Co,Cr,Fe,Mn,Ni)O3−δ (x = 0, 0.1, 0.2, 0.3) perovskites synthesized by the sol–gel method. J Mater Chem A De Vero JC, Develos-Bagarinao K, Matsuda H, Kishimoto H, Ishiyama T, Yamaji K, Horita T, Yokokawa H (2018) Sr and Zr transport in PLD-grown Gd-doped ceria interlayers. Solid State Ionics 314:165–171 del-Mazo-Barbara L, Ginebra MP (2021) Rheological characterisation of ceramic inks for 3D direct ink writing: a review. J Eur Ceram Soc 41:18–33 Deng X, Chen K, Tü Ysü H (2017) Protocol for the nanocasting method: preparation of ordered mesoporous metal oxides. Chem Mater 29:40–52 Derby B (2010) Inkjet printing of functional and structural materials: fluid property requirements, feature stability, and resolution. Annu Rev Mater Res 40:395–414 Derby B (2015) Additive manufacture of ceramics components by inkjet printing. Engineering 1:113–123 Desai S, Yang M, Xu Z, Sankar J (2014) Direct write manufacturing of solid oxide fuel cells for green energy. J Environ Res Dev 8:477–483 Despeisse M, Ford S (2015) The role of additive manufacturing in improving resource efficiency and sustainability. IFIP Adv Inf Commun Technol 460:129–136 Develos-Bagarinao K, De Vero J, Kishimoto H, Ishiyama T, Yamaji K, Horita T, Yokokawa H (2018) Multilayered LSC and GDC: an approach for designing cathode materials with superior oxygen exchange properties for solid oxide fuel cells. Nano Energy 52:369–380 Develos-Bagarinao K, Budiman RA, Liu SS, Ishiyama T, Kishimoto H, Yamaji K (2020) Evolution of cathode-interlayer interfaces and its effect on long-term degradation. J Power Sources 453:227894

368

A. Tarancón et al.

Develos-Bagarinao K, Celikbilek O, Budiman RA, Kerherve G, Fearn S, Skinner SJ, Kishimoto H (2022) On the role of surfaces and interfaces in electrochemical performance and long-term stability of nanostructured LSC thin film electrodes. J Mater Chem A 10:2445–2459 Ding D, Li X, Lai SY, Gerdes K, Liu M (2014) Enhancing SOFC cathode performance by surface modification through infiltration. Energy Environ Sci 7:552–575 Dong D, Shao X, Hu X, Chen K, Xie K, Yu L, Ye Z, Yang P, Parkinson G, Li CZ (2016) Improved gas diffusion within microchanneled cathode supports of SOECs for steam electrolysis. Int J Hydrogen Energy 41:19829–19835 Dong D, Xu S, Shao X, Hucker L, Marin J, Pham T, Xie K, Ye Z, Yang P, Yu L, Parkinson G, Li CZ (2017) Hierarchically ordered porous Ni-based cathode-supported solid oxide electrolysis cells for stable CO2 electrolysis without safe gas. J Mater Chem A 5:24098–24102 Dudek M, Tomov RI, Wang C, Glowacki BA, Tomczyk P, Socha RP, Mosiałek M (2013) Feasibility of direct carbon solid oxide fuels cell (DC-SOFC) fabrication by inkjet printing technology. Electrochim Acta 105:412–418 Esposito V, Gadea C, Hjelm J, Marani D, Hu Q, Agersted K, Ramousse S, Jensen SH (2015) Fabrication of thin yttria-stabilized-zirconia dense electrolyte layers by inkjet printing for high performing solid oxide fuel cells. J Power Sources 273:89–95 Esposito V, Tarancón A (eds) (2021) 3D printing for energy applications. John Wiley & Sons Inc., Hoboken, New Jersey Evans A, Bieberle-Hütter A, Rupp JLM, Gauckler LJ (2009) Review on microfabricated micro-solid oxide fuel cell membranes. J Power Sources 194:119–129 Evans A, Martynczuk J, Stender D, Schneider CW, Lippert T, Prestat M (2015) Low-temperature micro-solid oxide fuel cells with partially amorphous La0.6 Sr0.4 CoO3-δ cathodes. Adv Energy Mater 5:1–9 Fan H, Zhang Y, Han M (2017) Infiltration of La0.6 Sr0.4 FeO3-δ nanoparticles into YSZ scaffold for solid oxide fuel cell and solid oxide electrolysis cell. J Alloy Compd 723:620–626 Fan L, Zhu B, Su PC, He C (2018) Nanomaterials and technologies for low temperature solid oxide fuel cells: recent advances, challenges and opportunities. Nano Energy 45:148–176 Farahani RD, Dubé M, Therriault D (2016) Three-dimensional printing of multifunctional nanocomposites: manufacturing techniques and applications. Adv Mater 5794–5821 Farandos NM, Kleiminger L, Li T, Hankin A, Kelsall GH (2016) Three-dimensional inkjet printed solid oxide electrochemical reactors I. Yttria-stabilized zirconia electrolyte. Electrochim Acta 213:324–331 Farandos NM, Li T, Kelsall GH (2018) 3-D inkjet-printed solid oxide electrochemical reactors. II LSM–YSZ electrodes. Electrochim Acta 270:264–273 Fasaki I, Siamos K, Arin M, Lommens P, Van Driessche I, Hopkins SC, Glowacki BA, Arabatzis I (2012) Ultrasound assisted preparation of stable water-based nanocrystalline TiO2 suspensions for photocatalytic applications of inkjet-printed films. Appl Catal A 411–412:60–69 Feilden E, Blanca EGT, Giuliani F, Saiz E, Vandeperre L (2016) Robocasting of structural ceramic parts with hydrogel inks. J Eur Ceram Soc 36:2525–2533 Feng Z, Liu L, Li L, Chen J, Liu Y, Li Y, Hao L, Wu Y (2019) 3D printed Sm-doped ceria composite electrolyte membrane for low temperature solid oxide fuel cells. Int J Hydrogen Energy 44:13843–13851 Feng Z, Pu J, Liu M, Zhang W, Zhang X, Cui L (2022) Facile construction of hierarchical Co3 S4 /CeO2 heterogeneous nanorod array on cobalt foam for electrocatalytic overall water splitting. J Colloid Interface Sci 613:806–813 Ferreira VJ, Wolff D, Hornés A, Morata A, Torrell M, Tarancón A, Corchero C (2021) 5 kW SOFC stack via 3D printing manufacturing: an evaluation of potential environmental benefits. Appl Energy 291 Franco T, Brandner M, Rüttinger M, Kunschert G, Venskutonis A, Sigl L (2009) Recent development aspects of metal supported thin-film SOFC. ECS Trans 25:681–688

Emerging Trends in Solid Oxide Electrolysis Cells

369

Gaiselmann G, Neumann M, Holzer L, Hocker T, René M (2013) Stochastic 3D modeling of La0.6 Sr0.4 CoO3−δ cathodes based on structural segmentation of FIB–SEM images. Comput Mater Sci 67:48–62 Gan L, Ye L, Tao S, Xie K (2016) Titanate cathodes with enhanced electrical properties achieved via growing surface Ni particles toward efficient carbon dioxide electrolysis. Phys Chem Chem Phys 18:3137–3143 Gao MC, Liaw PK, Yeh JW, Zhang Y (2016a) High-entropy alloys: fundamentals and applications. High-Entropy Alloys Fundam Appl 1–516 Gao Y, Chen D, Saccoccio M, Lu Z, Ciucci F (2016b) From material design to mechanism study: nanoscale Ni exsolution on a highly active A-site deficient anode material for solid oxide fuel cells. Nano Energy 27:499–508 Gao Z, Mogni LV, Miller EC, Railsback JG, Barnett SA (2016c) A perspective on low-temperature solid oxide fuel cells. Energy Environ Sci 9:1602–1644 Garbayo I, Tarancón A, Santiso J, Peiró F, Alarcón-LLadó E, Cavallaro A, Gràcia I, Cané C, Sabaté N (2010) Electrical characterization of thermomechanically stable YSZ membranes for micro solid oxide fuel cells applications. Solid State Ionics 181:322–331 Garbayo I, Pla D, Morata A, Fonseca L, Sabate N, Tarancon A (2014) Full ceramic micro solid oxide fuel cells: towards more reliable MEMS power generators operating at high temperatures. Energy Environ Sci 7 Garbayo I, Baiutti F, Morata A, Tarancón A (2019) Engineering mass transport properties in oxide ionic and mixed ionic-electronic thin film ceramic conductors for energy applications. J Eur Ceram Soc 39:101–114 Gelfond NV, Bobrenok OF, Predtechensky MR, Morozova NB, Zherikova KV, Igumenov IK (2009) Chemical vapor deposition of electrolyte thin films based on yttria-stabilized zirconia. Inorg Mater 45(6):659–665 George SM (2009) Atomic layer deposition: an overview. Chem Rev 110:111–131 Gild J, Samiee M, Braun JL, Harrington T, Vega H, Hopkins PE, Vecchio K, Luo J (2018) Highentropy fluorite oxides. J Eur Ceram Soc 38:3578–3584 Golbert J, Adjiman CS, Brandon NP (2008) Microstructural modeling of solid oxide fuel cell anodes. Ind Eng Chem Res 47:7693–7699 Greer JA, Tabat MD (1998) Large-area pulsed laser deposition: techniques and applications. J Vac Sci Technol A Vac Surf Films 13:1175 Grew KN, Chiu WKS (2012) A review of modeling and simulation techniques across the length scales for the solid oxide fuel cell. J Power Sources 199:1–13 Ha S, Su PC, Cha SW (2013) Combinatorial deposition of a dense nano-thin film YSZ electrolyte for low temperature solid oxide fuel cells. J Mater Chem A 1:9645–9649 Han GD, Neoh KC, Bae K, Choi HJ, Park SW, Son JW, Shim JH (2016) Fabrication of lanthanum strontium cobalt ferrite (LSCF) cathodes for high performance solid oxide fuel cells using a low price commercial inkjet printer. J Power Sources 306:503–509 Han J, Wang X, Yan L, Dahlak A (2019) Modelling the performance of an SOEC by optimization of neural network with MPSO algorithm. Int J Hydrogen Energy 44:27947–27957 Han GD, Bae K, Kang EH, Choi HJ, Shim JH (2020) Inkjet printing for manufacturing solid oxide fuel cells. ACS Energy Lett 1586–1592 Han X, Yang Y, Fan Y, Ni H, Guo Y, Chen Y, Ou X, Ling Y (2021) New approach to enhance Srfree cathode performance by high-entropy multi-component transition metal coupling. Ceram Int 47:17383–17390 Hanifi AR, Laguna-Bercero MA, Etsell TH, Sarkar P (2014) The effect of electrode infiltration on the performance of tubular solid oxide fuel cells under electrolysis and fuel cell modes. Int J Hydrogen Energy 39:8002–8008 Hartings MR, Ahmed Z (2019) Chemistry from 3D printed objects. Nat Rev Chem 3:305–314 Haslam JJ, Pham AQ, Chung BW, DiCarlo JF, Glass RS (2005) Effects of the use of pore formers on performance of an anode supported solid oxide fuel cell. J Am Ceram Soc 88:513–518

370

A. Tarancón et al.

Haugen AB, Gurauskis J, Kaiser A, Søgaard M (2017) Graphite and PMMA as pore formers for thermoplastic extrusion of porous 3Y-TZP oxygen transport membrane supports. J Eur Ceram Soc 37:1039–1047 Hedayat N, Du Y, Ilkhani H (2017) Review on fabrication techniques for porous electrodes of solid oxide fuel cells by sacrificial template methods. Renew Sustain Energy Rev 77:1221–1239 Hedayat N, Du Y, Ilkhani H (2018) Pyrolyzable pore-formers for the porous-electrode formation in solid oxide fuel cells: a review. Ceram Int 44:4561–4576 Herbstritt D, Weber A, Ivers-Tiffée E (2001) Modelling and DC-polarisation of a three dimensional electrode/electrolyte interface. J Eur Ceram Soc 21:1813–1816 Hernández E, Baiutti F, Morata A, Torrell M, Tarancón A (2018) Infiltrated mesoporous oxygen electrodes for high temperature co-electrolysis of H2 O and CO2 in solid oxide electrolysis cells. J Mater Chem A 6:9699–9707 Hill TY, Reitz TL, Rottmayer MA, Huang H (2015) Controlling inkjet fluid kinematics to achieve SOFC cathode micropatterns. ECS J Solid State Sci Technol 4:P3015–P3019 Hjelm J, Soegaard M, Knibbe R, Hagen A, Mogensen M (2008) Electrochemical characterization of planar anode supported SOFC with strontium-doped lanthanum cobalt oxide cathodes. ECS Trans 13:285–299 Hoath SD (ed) (2015) Fundamentals of inkjet printing: the science of inkjet and droplets. Wiley-VCH Verlag, Weinheim, Germany Horlick SA, Huang YL, Robinson IA, Wachsman ED (2021) Controlling exsolution with a chargebalanced doping approach. Nano Energy 87:106193 Horri BA, Selomulya C, Wang H (2012) Characteristics of Ni/YSZ ceramic anode prepared using carbon microspheres as a pore former. Int J Hydrogen Energy 37:15311–15319 Huang H, Nakamura M, Su P, Fasching R, Saito Y, Prinz FB (2007) High-performance ultrathin solid oxide fuel cells for low-temperature operation. J Electrochem Soc 154:B20–B24 Hubert M, Laurencin J, Cloetens P, Morel B, Montinaro D, Lefebvre-Joud F (2018) Impact of nickel agglomeration on solid oxide cell operated in fuel cell and electrolysis modes. J Power Sources 397:240–251 Huijben M, Koster G, Kruize MK, Wenderich S, Verbeeck J, Bals S, Slooten E, Shi B, Molegraaf HJA, Kleibeuker JE, Van Aert S, Goedkoop JB, Brinkman A, Blank DHA, Golden MS, Van Tendeloo G, Hilgenkamp H, Rijnders G (2013) Defect engineering in oxide heterostructures by enhanced oxygen surface exchange. Adv Func Mater 23:5240–5248 Infortuna A, Harvey AS, Gauckler LJ (2008) Microstructures of CGO and YSZ thin films by pulsed laser deposition. Adv Func Mater 18:127–135 Iora P, Aguiar P, Adjiman CS, Brandon NP (2005) Comparison of two IT DIR-SOFC models: impact of variable thermodynamic, physical, and flow properties. Steady-state and dynamic analysis. Chem Eng Sci 60:2963–2975 Irvine JTSS, Neagu D, Verbraeken MC, Chatzichristodoulou C, Graves C, Mogensen MB (2016) Evolution of the electrochemical interface in high-temperature fuel cells and electrolysers. Nat Energy 1:1–13 Ishihara T, Kusaba H, Kim HH, Kang BS (2019) Preparation of La0.9 Sr0.1 Ga0.8 Mg0.2 O3 film by pulse laser deposition (PLD) method on porous Ni–Fe metal substrate for CO2 electrolysis. ISIJ Int 59:613–618 Ishizaki K, Komarneni S, Nanko M (1998) Porous materials. Med J Zambia, Materials Technology Series, vol 4. Springer US, Boston, MA Islam QA, Paydar S, Akbar N, Zhu B, Wu Y (2021) Nanoparticle exsolution in perovskite oxide and its sustainable electrochemical energy systems. J Power Sources 492:229626 Janardhanan VM, Deutschmann O (2011) Modeling diffusion limitation in solid-oxide fuel cells. Electrochim Acta 56:9775–9782 Janardhanan VM (2021) Microkinetic modeling of CO2 and H2 O electrolysis on Ni in a solid oxide electrolysis cell: a critical evaluation. J Electrochem Soc

Emerging Trends in Solid Oxide Electrolysis Cells

371

Jang DY, Kim M, Kim JW, Bae K, Son JW, Schlupp MVF, Shim JH (2017) High performance anodesupported solid oxide fuel cells with thin film yttria-stabilized zirconia membrane prepared by aerosol-assisted chemical vapor deposition. J Electrochem Soc 164:F484–F490 Januschewsky J, Ahrens M, Opitz A, Kubel F, Fleig J (2009) Optimized La0.6 Sr0.4 CoO3-δ thin-film electrodes with extremely fast oxygen-reduction kinetics. Adv Func Mater 19:3151–3156 Jardiel T, Caldes MT, Moser F, Hamon J, Gauthier G, Joubert O (2010) New SOFC electrode materials: the Ni-substituted LSCM-based compounds (La0.75 Sr0.25 )(Cr0.5 Mn0.5−x Nix )O3−δ and (La0.75 Sr0.25 )(Cr0.5−x Nix Mn0.5 )O3−δ . Solid State Ionics 181:894–901 Jha S, Velhal M, Stewart W, Amin V, Wang E, Liang H (2021) Additively manufactured electrodes for supercapacitors: a review. Appl Mater Today 101220 Ji S, Cho GY, Yu W, Su PC, Lee MH, Cha SW (2015) Plasma-enhanced atomic layer deposition of nanoscale yttria-stabilized zirconia electrolyte for solid oxide fuel cells with porous substrate. ACS Appl Mater Interfaces 7:2998–3002 Jia K, Zheng L, Liu W, Zhang J, Yu F, Meng X, Li C, Sunarso J, Yang N (2022) A new and simple way to prepare monolithic solid oxide fuel cell stack by stereolithography 3D printing technology using 8 mol% yttria stabilized zirconia photocurable slurry. J Eur Ceram Soc 42:4275–4285 Jiang SP (2008) Development of lanthanum strontium manganite perovskite cathode materials of solid oxide fuel cells: a review. J Mater Sci 43:6799–6833 Jiang SP (2012) Nanoscale and nano-structured electrodes of solid oxide fuel cells by infiltration: advances and challenges. Int J Hydrogen Energy 37:449–470 Jiang SP (2019) Development of lanthanum strontium cobalt ferrite perovskite electrodes of solid oxide fuel cells—a review. Int J Hydrogen Energy 44:7448–7493 Jo S, Sharma B, Park DH, Myung JH (2020) Materials and nano-structural processes for use in solid oxide fuel cells: a review. J Korean Ceram Soc 57:135–151 Jordan N, Assenmacher W, Uhlenbruck S, Haanappel VAC, Buchkremer HP, Stöver D, Mader W (2008) Ce0.8 Gd0.2 O2−δ protecting layers manufactured by physical vapor deposition for IT-SOFC. Solid State Ionics 179:919–923 Jung W, Dereux JO, Chueh WC, Hao Y, Haile SM (2012) High electrode activity of nanostructured, columnar ceria films for solid oxide fuel cells. Energy Environ Sci 5:8682–8689 Kaiser A, Foghmoes SP, Pe´canac G, Malzbender J, Chatzichristodoulou C, Glasscock JA, Ramachandran D, Ni DW, Esposito V, Søgaard M, Hendriksen PV (2016) Design and optimization of porous ceramic supports for asymmetric ceria-based oxygen transport membranes. J Membr Sci 513:85–94 Kamecki B, Miruszewski T, Górnicka K, Klimczuk T, Karczewski J (2019) Characterization methods of nickel nano-particles obtained by the ex-solution process on the surface of Pr, Ni-doped SrTiO3 perovskite ceramics. SN Appl Sci 1:1–9 Kao WX, Lee MC, Chang YC, Lin TN, Wang CH, Chang JC (2010) Fabrication and evaluation of the electrochemical performance of the anode-supported solid oxide fuel cell with the composite cathode of La0.8 Sr0.2 MnO3−δ -Gadolinia-doped ceria oxide/La0.8 Sr0.2 MnO3–δ . J Power Sources 195:6468–6472 Kawale S, Jang I, Farandos N, Kelsall GH (2022) Inkjet 3D-printing of functional layers of solid oxide electrochemical reactors: a review. React Chem Eng Kelly PJ, Arnell RD (2000) Magnetron sputtering: a review of recent developments and applications. Vacuum 56:159–172 Kerman K, Tallinen T, Ramanathan S, Mahadevan L (2013) Elastic configurations of self-supported oxide membranes for fuel cells. J Power Sources 222:359–366 Khan MZ, Song RH, Mehran MT, Lee SB, Lim TH (2021) Controlling cation migration and inter-diffusion across cathode/interlayer/electrolyte interfaces of solid oxide fuel cells: a review. Ceram Int 47:5839–5869 Kiebach R, Zielke P, Høgh JVT, Thydén K, Wang HJ, Barford R, Hendriksen PV (2016) Infiltration of SOFC stacks: evaluation of the electrochemical performance enhancement and the underlying changes in the microstructure. Fuel Cells 16:80–88

372

A. Tarancón et al.

Kim JD, Kim GD, Moon JW, Park YI, Lee WH, Kobayashi K, Nagai M, Kim CE (2001) Characterization of LSM-YSZ composite electrode by ac impedance spectroscopy. Solid State Ionics 143:379–389 Kim EH, Jung HJ, An KS, Park JY, Lee J, Hwang ID, Kim JY, Lee MJ, Kwon Y, Hwang JH (2014) Degradation of La0.6 Sr0.4 CoO3 -based cathode performance in solid oxide fuel cells due to the presence of aluminum oxide deposited through atomic layer deposition. Ceram Int 40:7817–7822 Kim KJ, Han H, Defferriere T, Yoon D, Na S, Kim SJ, Dayaghi AM, Son J, Oh TS, Jang HM, Choi GM (2019) Facet-dependent in situ growth of nanoparticles in epitaxial thin films: the role of interfacial energy. J Am Chem Soc 141:7509–7517 Kim M, Kim DH, Han GD, Choi HJ, Choi HR, Shim JH (2020) Lanthanum strontium cobaltiteinfiltrated lanthanum strontium cobalt ferrite cathodes fabricated by inkjet printing for highperformance solid oxide fuel cells. J Alloy Compd 843:155806 Kim S, Joh DW, Lee DY, Lee J, Kim HS, Khan MZ, Hong JE, Lee SB, Park SJ, Song RH, Mehran MT, Rhee CK, Lim TH (2021a) Microstructure tailoring of solid oxide electrolysis cell air electrode to boost performance and long-term durability. Chem Eng J 410:128318 Kim S, Kim G, Manthiram A (2021b) A review on infiltration techniques for energy conversion and storage devices: from fundamentals to applications. Sustain Energy Fuels 5:5024–5037 Knibbe R, Hjelm J, Menon M, Pryds N, Søgaard M, Wang HJ, Neufeld K (2010) Cathode-electrolyte interfaces with CGO barrier layers in SOFC. J Am Ceram Soc 93:2877–2883 Kobsiriphat W, Madsen BD, Wang Y, Shah M, Marks LD, Barnett SA (2010) Nickel-and Rutheniumdoped lanthanum chromite anodes: effects of nanoscale metal precipitation on solid oxide fuel cell performance. J Electrochem Soc 157:279–284 Koo B, Kim K, Kim JK, Kwon H, Han JW, Jung WC (2018) Sr Segregation in perovskite oxides: why it happens and how it exists. Joule 2:1476–1499 Kostretsova N, Pesce A, Nuñez M, Morata A, Torrell M, Tarancón A (2021) Self-supported solid oxide fuel cells by multimaterial 3D printing. ECS Trans 103:59 Kostretsova N, Anelli S, Nuñez M, Morata A, Torrell M, Tarancón A (n.d.) Monolithic solid oxide cells by 3D printing. In: 15th European SOFC & SOE forum. Lucerne, Switzerland, p A1207 Krishnan VV (2017) Recent developments in metal-supported solid oxide fuel cells. Wiley Interdisc Rev Energy Environ 6 Kuhn M, Napporn T, Meunier M, Vengallatore S, Therriault D (2008) Direct-write microfabrication of single-chamber micro solid oxide fuel cells. J Micromech Microeng 18 Kwon CW, Son JW, Lee JH, Kim HM, Lee HW, Kim KB (2011) High-performance micro-solid oxide fuel cells fabricated on nanoporous anodic aluminum oxide templates. Adv Func Mater 21:1154–1159 Kwon O, Sengodan S, Kim K, Kim G, Jeong HY, Shin J, Ju YW, Han JW, Kim G (2017) Exsolution trends and co-segregation aspects of self-grown catalyst nanoparticles in perovskites. Nat Commun 8:1–7 Kwon O, Joo S, Choi S, Sengodan S, Kim G (2020a) Review on exsolution and its driving forces in perovskites. J Phys Energy 2:032001 Kwon O, Joo S, Choi S, Sengodan S, Kim G (2020b) Review on exsolution and its driving forces in perovskites. J Phys Energy 2 Laguna-Bercero MA, Hanifi AR, Menand L, Sandhu NK, Anderson NE, Etsell TH, Sarkar P (2018) The effect of pore-former morphology on the electrochemical performance of solid oxide fuel cells under combined fuel cell and electrolysis modes. Electrochim Acta 268:195–201 Laha SC, Ryoo R (2003) Synthesis of thermally stable mesoporous cerium oxide with nanocrystalline frameworks using mesoporous silica templates. Chem Commun 2138–2139 Larrea A, Sola D, Laguna-Bercero MA, Peña JI, Merino RI, Orera VM (2011) Self-supporting thin yttria-stabilised zirconia electrolytes for solid oxide fuel cells prepared by laser machining. J Electrochem Soc 158:B1193

Emerging Trends in Solid Oxide Electrolysis Cells

373

Laurencin J, Hubert M, Couturier K, Le Bihan T, Cloetens P, Lefebvre-Joud F, Siebert E (2015) Reactive mechanisms of LSCF single-phase and LSCF-CGO composite electrodes operated in anodic and cathodic polarisations. Electrochim Acta 174:1299–1316 Lee W, Han JW, Chen Y, Cai Z, Yildiz B (2013) Cation size mismatch and charge interactions drive dopant segregation at the surfaces of manganite perovskites. J Am Chem Soc 135:7909–7925 Lee YH, Chang I, Cho GY, Park J, Yu W, Tanveer WH, Cha SW (2018) Thin film solid oxide fuel cells operating below 600 °C: a review. Int J Precis Eng Manuf-Green Technol 5:441 Lee YL, Kleis J, Rossmeisl J, Yang SH, Morgan D (2011) Prediction of solid oxide fuel cell cathode activity with first-principles descriptors. Energy Environ Sci 4:3966–3970 Lee C, Shin SS, Choi J, Kim J, Son JW, Choi M, Shin HH (2020) A micro-patterned electrode/electrolyte interface fabricated by soft-lithography for facile oxygen reduction in solid oxide fuel cells. J Mater Chem A 8:16534–16541 Lewis JA, Smay JE, Stuecker J, Cesarano J (2006) Direct ink writing of three-dimensional ceramic structures. J Am Ceram Soc 89:3599–3609 Li W, Shi Y, Luo Y, Cai N (2014) Theoretical modeling of air electrode operating in SOFC mode and SOEC mode: the effects of microstructure and thickness. Int J Hydrogen Energy 39:13738– 13750 Li C, Chen H, Shi H, Tade MO, Shao Z (2015) Green fabrication of composite cathode with attractive performance for solid oxide fuel cells through facile inkjet printing. J Power Sources 273:465–471 Li X, Wang Y, Liu W, Wilson JA, Wang J, Wang C, Yang J, Xia C, Zhou XD, Guan W (2019) Reliability of CO2 electrolysis by solid oxide electrolysis cells with a flat tube based on a composite double-sided air electrode. Compos B Eng 166:549–554 Li H, Song Y, Xu M, Wang W, Ran R, Zhou W, Shao Z (2020) Exsolved alloy nanoparticles decorated Ruddlesden-Popper perovskite as sulfur-tolerant anodes for solid oxide fuel cells. Energy Fuels 34:11449–11457 Li J, Huang Q, Lei Q, Maxim T, Chen D (2021a) The effects of microstructural parameters on the electrochemical properties of LSM-LSCF composite cathode by the particle-based discrete element method 2901–2907 Li Z, Zhang H, Xu H, Xuan J (2021b) Advancing the multiscale understanding on solid oxide electrolysis cells via modelling approaches: a review. Renew Sustain Energy Rev 141:110863 Lichtner AZ, Jauffrès D, Roussel D, Charlot F, Martin CL, Bordia RK (2015) Dispersion, connectivity and tortuosity of hierarchical porosity composite SOFC cathodes prepared by freeze-casting. J Eur Ceram Soc 35:585–595 Lim Y, Lee H, Hong S, Kim YB (2019) Co-sputtered nanocomposite nickel cermet anode for high-performance low-temperature solid oxide fuel cells. J Power Sources 412:160–169 Ling Y, Liu L, Zhou S, Yang Y, Zhu H, Chen H, Paydar MH, Wang S (2021) Leveraging in-situ formation of Ni–Fe nanoparticles to promote the catalytic performance of Ruddlesden-Popper based electrode for symmetrical solid oxide fuel cells. Int J Hydrogen Energy 46:27149–27155 Linne DL, Kuczmarski MA, Johnston JC, Farmer SC, Green RD, Mital SK, Setlock JA (2018) Advanced manifolds for improved solid oxide electrolyzer performance. In: 2018 AIAA SPACE and astronautics forum and exposition, pp 1–6 Liu PS, Chen GF (2014) Porous materials—processing and applications. Butterworth-Heinemann, Elsevier Liu X, Tarn TJ, Huang F, Fan J (2015) Recent advances in inkjet printing synthesis of functional metal oxides. Particuology 19:1–13 Liu T, Ren C, Zhang Y, Wang Y, Lei L, Chen F (2017) Solvent effects on the morphology and performance of the anode substrates for solid oxide fuel cells. J Power Sources 363:304–310 Lu Z, Zhou XD, Fisher D, Templeton J, Stevenson J, Wu N, Ignatiev A (2010) Enhanced performance of an anode-supported YSZ thin electrolyte fuel cell with a laser-deposited Sm0.2 Ce0.8 O1.9 interlayer. Electrochem Commun 12:179–182

374

A. Tarancón et al.

Lu X, Heenan TMM, Bailey JJ, Li T, Li K, Brett DJL, Shearing PR (2017a) Correlation between triple phase boundary and the microstructure of solid oxide fuel cell anodes: the role of composition, porosity and Ni densification. J Power Sources 365:210–219 Lu Z, Darvish S, Hardy J, Templeton J, Stevenson J, Zhong Y (2017b) SrZrO3 formation at the interlayer/electrolyte interface during (La1–x Srx )1−δ Co1−y Fey O3 cathode sintering. J Electrochem Soc 164:F3097–F3103 Lv H, Lin L, Zhang X, Song Y, Matsumoto H, Zeng C, Ta N, Liu W, Gao D, Wang G, Bao X (2020) In situ investigation of reversible exsolution/dissolution of CoFe alloy nanoparticles in a Co-doped Sr2 Fe1.5 Mo0.5 O6−δ cathode for CO2 electrolysis. Adv Mater 32:1–9 Ma Y, Ma Y, Wang Q, Schweidler S, Botros M, Fu T, Hahn H, Brezesinski T, Breitung B (2021) High-entropy energy materials: challenges and new opportunities. Energy Environ Sci 14:2883– 2905 Mahato N, Banerjee A, Gupta A, Omar S, Balani K (2015) Progress in material selection for solid oxide fuel cell technology: a review. Prog Mater Sci 72:141–337 Mai A, Haanappel VAC, Tietz F, Stöver D (2006) Ferrite-based perovskites as cathode materials for anode-supported solid oxide fuel cells. Part II. Influence of the CGO interlayer. Solid State Ionics 177:2103–2107 Manogharan G, Kioko M, Linkous C (2015) Binder jetting: a novel solid oxide fuel-cell fabrication process and evaluation. JOM J Minerals Metals Mater Soc 67:660–667 Marchand P, Hassan IA, Parkin IP, Carmalt CJ (2013) Aerosol-assisted delivery of precursors for chemical vapour deposition: expanding the scope of CVD for materials fabrication. Dalton Trans 42:9406–9422 Marquardt T, Hollmann J, Gimpel T, Schade W, Kabelac S (2020) Femtosecond laser-induced surface modification of the electrolyte in solid oxide electrolysis cells. Energies 13:1–22 Masciandaro S, Torrell M, Leone P, Tarancón A (2019) Three-dimensional printed yttria-stabilized zirconia self-supported electrolytes for solid oxide fuel cell applications. J Eur Ceram Soc 39:9–16 Meng X, Liu Y, Yang N, Tan X, Liu J, Diniz da Costa JC, Liu S (2017) Highly compact and robust hollow fiber solid oxide cells for flexible power generation and gas production. Appl Energy 205:741–748 Mitchell-Williams TB, Tomov RI, Saadabadi SA, Krauz M, Aravind PV, Glowacki BA, Kumar RV (2017) Infiltration of commercially available, anode supported SOFC’s via inkjet printing. Mater Renew Sustain Energy 6:1–9 Mohammed Hussain A, Høgh JVT, Jacobsen T, Bonanos N (2012) Nickel-ceria infiltrated Nbdoped SrTiO3 for low temperature SOFC anodes and analysis on gas diffusion impedance. Int J Hydrogen Energy 37:4309–4318 Monaco F, Ferreira-Sanchez D, Hubert M, Morel B, Montinaro D, Grolimund D, Laurencin J (2021) Oxygen electrode degradation in solid oxide cells operating in electrolysis and fuel cell modes: LSCF destabilization and interdiffusion at the electrode/electrolyte interface. Int J Hydrogen Energy 46:31533–31549 Mooraj S, Qi Z, Zhu C, Ren J, Peng S, Liu L, Zhang S, Feng S, Kong F, Liu Y, Duoss EB, Baker S, Chen W (2021) 3D printing of metal-based materials for renewable energy applications. Nano Res 14:2105–2132 Morales M, Miguel-Pérez V, Tarancón A, Slodczyk A, Torrell M, Ballesteros B, Ouweltjes JP, Bassat JM, Montinaro D, Morata A (2017) Multi-scale analysis of the diffusion barrier layer of gadolinia-doped ceria in a solid oxide fuel cell operated in a stack for 3000 h. J Power Sources 344:141–151 Morales M, Pesce A, Slodczyk A, Torrell M, Piccardo P, Montinaro D, Tarancón A, Morata A (2018) Enhanced performance of gadolinia-doped ceria diffusion barrier layers fabricated by pulsed laser deposition for large-area solid oxide fuel cells. ACS Appl Energy Mater 1:1955–1964 Moussaoui H, Laurencin J, Gavet Y, Delette G, Hubert M, Cloetens P, Bihan TL, Debayle J (2018) Stochastic geometrical modeling of solid oxide cells electrodes validated on 3D reconstructions. Comput Mater Sci 143, 262–276

Emerging Trends in Solid Oxide Electrolysis Cells

375

Moussaoui H, Sharma RK, Debayle J, Gavet Y, Delette G, Laurencin J (2019) Microstructural correlations for specific surface area and triple phase boundary length for composite electrodes of solid oxide cells. J Power Sources 412:736–748 Murty BS, Yeh JW, Ranganathan S, Bhattacharjee PP (2019) High-entropy alloys. Elsevier Myung JH, Neagu D, Miller DN, Irvine JTS (2016) Switching on electrocatalytic activity in solid oxide cells. Nature 537:528–531 Nakajo A, Rinaldi G, Caliandro P, Jeanmonod G, Navratilova L, Cantoni M, Van herle J (2020) Evolution of the morphology near triple-phase boundaries in Ni–yttria stabilized zirconia electrodes upon cathodic polarization. J Electrochem Energy Conv Storage 17:1–13 Neagu D, Tsekouras G, Miller DN, Ménard H, Irvine JTS (2013) In situ growth of nanoparticles through control of non-stoichiometry. Nat Chem 5:916–923 Neagu D, Oh TS, Miller DN, Ménard H, Bukhari SM, Gamble SR, Gorte RJ, Vohs JM, Irvine JTS (2015) Nano-socketed nickel particles with enhanced coking resistance grown in situ by redox exsolution. Nat Commun 6:8120 Neagu D, Papaioannou EI, Ramli WKW, Miller DN, Murdoch BJ, Ménard H, Umar A, Barlow AJ, Cumpson PJ, Irvine JTS, Metcalfe IS (2017) Demonstration of chemistry at a point through restructuring and catalytic activation at anchored nanoparticles. Nature Commun 8:1–8 Nenning A, Fleig J (2019) Electrochemical XPS investigation of metal exsolution on SOFC electrodes: controlling the electrode oxygen partial pressure in ultra-high-vacuum. Surf Sci 680:43–51 Ngo TD, Kashani A, Imbalzano G, Nguyen KTQ, Hui D (2018) Additive manufacturing (3D printing): a review of materials, methods, applications and challenges. Compos B Eng 143:172– 196 Nielsen J, Jacobsen T, Wandel M (2011) Impedance of porous IT-SOFC LSCF:CGO composite cathodes. Electrochim Acta 56:7963–7974 Nishihata Y, Mizuki J, Akao T, Tanaka H, Uenishi M, Kimura M, Okamoto T, Hamada N (2002) Self-regeneration of a Pd-perovskite catalyst for automotive emissions control. Nature 418:164– 167 Niu B, Lu C, Yi W, Luo S, Li X, Zhong X, Zhao X, Xu B (2020) In-situ growth of nanoparticlesdecorated double perovskite electrode materials for symmetrical solid oxide cells. Appl Catal B 270:118842 Niu Y, Zhou Y, Lv W, Chen Y, Zhang Y, Zhang W, Luo Z, Kane N, Ding Y, Soule L, Liu Y, He W, Liu M (2021) Enhancing oxygen reduction activity and Cr tolerance of solid oxide fuel cell cathodes by a multiphase catalyst coating. Adv Funct Mater 31 Noh HS, Hong J, Kim H, Yoon KJ, Kim BK, Lee HW, Lee JH, Son JW (2016) Scale-up of thin-film deposition-based solid oxide fuel cell by sputtering, a commercially viable thin-film technology. J Electrochem Soc 163:F613–F617 Nurk G, Vestli M, Möller P, Jaaniso R, Kodu M, Mändar H, Romann T, Kanarbik R, Lust E (2016) Mobility of Sr in gadolinia doped ceria barrier layers prepared using spray pyrolysis, pulsed laser deposition and magnetron sputtering methods. J Electrochem Soc 163:F88–F96 Oh S, Park J, Shin JW, Yang BC, Zhang J, Jang DY, An J (2018) High performance low-temperature solid oxide fuel cells with atomic layer deposited-yttria stabilized zirconia embedded thin film electrolyte. J Mater Chem A Opitz AK, Nenning A, Vonk V, Volkov S, Bertram F, Summerer H, Schwarz S, Steiger-Thirsfeld A, Bernardi J, Stierle A, Fleig J (2020) Understanding electrochemical switchability of perovskitetype exsolution catalysts. Nat Commun 11:1–11 Osada N, Uchida H, Watanabe M (2006) Polarization behavior of SDC cathode with highly dispersed Ni catalysts for solid oxide electrolysis cells. J Electrochem Soc 153:A816 Pandiyan S, El-Kharouf A, Steinberger-Wilckens R (2020) Formulation of spinel based inkjet inks for protective layer coatings in SOFC interconnects. J Colloid Interface Sci 579:82–95 Park JH, Hong WS, Yoon KJ, Lee JH, Lee HW, Son JW (2014) Physical and electrochemical characteristics of pulsed laser deposited La0.6 Sr0.4 CoO3−δ Ce0.9 Gd0.1 O2−δ nanocomposites as a function of the mixing ratio. J Electrochem Soc 161:F16–F22

376

A. Tarancón et al.

Park JH, Han SM, Yoon KJ, Kim H, Hong J, Kim BK, Lee JH, Son JW (2016) Impact of nanostructured anode on low-temperature performance of thin-film-based anode-supported solid oxide fuel cells. J Power Sources 315:324–330 Park JH, Lee JH, Yoon KJ, Kim H, Ji HI, Yang S, Park S, Han SM, Son JW (2021) A nanoarchitectured cermet composite with extremely low Ni content for stable high-performance solid oxide fuel cells. Acta Mater 206 Parra-Cabrera C, Achille C, Kuhn S, Ameloot R (2018) 3D printing in chemical engineering and catalytic technology: structured catalysts, mixers and reactors. Chem Soc Rev 47:209–230 Patro PK, Delahaye T, Bouyer E, Sinha PK (2012) Microstructural development of Ni-1Ce10ScSZ cermet electrode for solid oxide electrolysis cell (SOEC) application. Int J Hydrogen Energy 37:3865–3873 Pei K, Zhou Y, Xu K, He Z, Chen Y, Zhang W, Yoo S, Zhao B, Yuan W, Liu M, Chen Y (2020) Enhanced Cr-tolerance of an SOFC cathode by an efficient electro-catalyst coating. Nano Energy 72:104704 Peng E, Zhang D, Ding J (2018) Ceramic robocasting: recent achievements, potential, and future developments. Adv Mater 30:1–14 Pesce A (2021) 3D printing of ceramic-based solid state energy conversion devices. Autonomus University of Barcelona Pesce A, Lira M, Santos D, Kostretsova N (2020a) B1308 new concepts for solid oxide cells manufacturing: the use of 3D printing technologies 1–10 Pesce A, Hornés A, Núñez M, Morata A, Torrell M, Tarancón A (2020b) 3D printing the next generation of enhanced solid oxide fuel and electrolysis cells. J Mater Chem A 8:16926–16932 Ping JS, Liu J (2017) Mesoporous materials for advanced energy storage and conversion technologies. CRC Press, Taylor & Francis Group Primdahl S, Mogensen M (1999) Gas diffusion impedance in characterization of solid oxide fuel cell anodes Qu P, Xiong D, Zhu Z, Gong Z, Li Y, Li Y, Fan L, Liu Z, Wang P, Liu C, Chen Z (2021) Inkjet printing additively manufactured multilayer SOFCs using high quality ceramic inks for performance enhancement. Addit Manuf 48:102394 Rahmanipour M, Cheng Y, Onn TM, Donazzi A, Vohs JM, Gorte RJ (2017) Modification of LSFYSZ composite cathodes by atomic layer deposition. J Electrochem Soc 164:F879–F884 Ren Y, Bruce PG, Ma Z (2011) Solid-solid conversion of ordered crystalline mesoporous metal oxides under reducing atmosphere. J Mater Chem 21:9312–9318 Reolon RP, Sanna S, Xu Y, Lee I, Bergmann CP, Pryds N, Esposito V (2018) Effects of accelerated degradation on metal supported thin film-based solid oxide fuel cells. J Mater Chem A 6:7887– 7896 Riegraf M, Han F, Sata N, Costa R (2021a) Intercalation of thin-film Gd-doped ceria barrier layers in electrolyte-supported solid oxide cells: physicochemical aspects. ACS Appl Mater Interfaces Riegraf M, Bombarda I, Dömling F, Liensdorf T, Sitzmann C, Langhof N, Schafföner S, Han F, Sata N, Geipel C, Walter C, Costa R (2021b) Enhancing the mechanical strength of electrolytesupported solid oxide cells with thin and dense doped-ceria interlayers. ACS Appl Mater Interfaces Rispoli N, Vitale F, Califano F, Califano M, Polverino P, Rosen MA, Sorrentino M (2020) Constrained optimal design of a reversible solid oxide cell-based multiple load renewable microgrid. J Energy Storage 31:101570 Rosa M, Gooden PN, Butterworth S, Zielke P, Kiebach R, Xu Y, Gadea C, Esposito V (2017) Zirconia nano-colloids transfer from continuous hydrothermal synthesis to inkjet printing. J Eur Ceram Soc 0–6 Rost CM, Sachet E, Borman T, Moballegh A, Dickey EC, Hou D, Jones JL, Curtarolo S, Maria JP (2015) Entropy-stabilized oxides. Nature Commun 6

Emerging Trends in Solid Oxide Electrolysis Cells

377

Ruiz-Morales JC, Tarancón A, Canales-Vázquez J, Méndez-Ramos J, Hernández-Afonso L, AcostaMora P, Marín Rueda JR, Fernández-González R (2017) Three dimensional printing of components and functional devices for energy and environmental applications. Energy Environ Sci 10:846–859 Sachs E, Cima M, Williams P, Brancazio D, Cornie J (1992) Three dimensional printing: rapid tooling and prototypes directly from a CAD model. J Manuf Sci E T ASME 114:481–488 Sadeghi-chahardeh A, Mollaabbasi R, Picard D, Taghavi SM (2021) Discrete element method modeling for the failure analysis of dry mono-size coke aggregates 1–30 Samson AJ, Søgaard M, Hendriksen PV (2017) Model for solid oxide fuel cell cathodes prepared by infiltration. Electrochim Acta 229:73–95 Saranya AMAM, Morata A, Pla D, Burriel M, Chiabrera F, Garbayo I, Hornés A, Kilner JAJA, Tarancón A (2018) Unveiling the outstanding oxygen mass transport properties of Mn-rich perovskites in grain boundary-dominated La0.8 Sr0.2 (Mn1−x Cox )0.85 O3±δ nanostructures. Chem Mater 30:5621–5629 Sarikaya A, Dogan F (2013) Effect of various pore formers on the microstructural development of tape-cast porous ceramics. Ceram Int 39:403–413 Sarkar A, Velasco L, Wang D, Wang Q, Talasila G, de Biasi L, Kübel C, Brezesinski T, Bhattacharya SS, Hahn H, Breitung B (2018a) High entropy oxides for reversible energy storage. Nature Commun 9 Sarkar A, Djenadic R, Wang D, Hein C, Kautenburger R, Clemens O, Hahn H (2018b) Rare earth and transition metal based entropy stabilised perovskite type oxides. J Eur Ceram Soc 38:2318–2327 Sarkar A, Breitung B, Hahn H (2020) High entropy oxides: the role of entropy, enthalpy and synergy. Scripta Mater 187:43–48 Sažinas R, Hansen KK (2019) Silver exsolution-enhanced electrical properties of lanthanum-based perovskites. J Mater Sci Eng A 9:116–129 Schiller G, Ansar A, Lang M, Patz O (2009) High temperature water electrolysis using metal supported solid oxide electrolyser cells (SOEC). J Appl Electrochem 39:293–301 Schlupp MVF, Scherrer B, Ma H, Grolig JG, Martynczuk J, Prestat M, Gauckler LJ (2012) Influence of microstructure on the cross-plane oxygen ion conductivity of yttria stabilized zirconia thin films. Phys Status Solidi (A) 209:1414–1422 Schlupp MVF, Kurlov A, Hwang J, Yμng Z, Döbeli M, Martynczuk J, Prestat M, Son JW, Gauckler LJ (n.d.) Gadolinia doped ceria thin films prepared by aerosol assisted chemical vapor deposition and applications in intermediate-temperature solid oxide fuel cells Seo H, Iwai H, Kishimoto M, Ding C, Saito M, Yoshida H (2020a) Microextrusion printing for increasing electrode–electrolyte interface in anode-supported solid oxide fuel cells. J Power Sources 450:227682 Seo HG, Ji S, Seo J, Kim S, Koo B, Choi Y, Kim H, Kim JH, Kim TS, Jung WC (2020b) Sinteringresistant platinum electrode achieved through atomic layer deposition for thin-film solid oxide fuel cells. J Alloy Compd 835:155347 Seo J, Tsvetkov N, Jeong J, Yoo Y, Ji S, Kim JH, Kang JK, Jung W (2020c) Gas-permeable inorganic shell improves the coking stability and electrochemical reactivity of Pt toward methane oxidation. ACS Appl Mater Interfaces 12:4405–4413 Seo J, Kim S, Jeon SH, Kim S, Hwan Kim J, Jung WC (2022) Nanoscale interface engineering for solid oxide fuel cells using atomic layer deposition. Nanoscale Adv Shah M, Barnett SA (2008) Solid oxide fuel cell cathodes by infiltration of La0.6 Sr0.4 Co0.2 Fe0.8 O3−δ into Gd-doped ceria. Solid State Ionics 179:2059–2064 Shah MA, Lee DG, Lee BY, Hur S (2021) Classifications and applications of inkjet printing technology: a review. IEEE Access 9:140079–140102 Shen L, Du Z, Zhang Y, Dong X, Zhao H (2021) Medium-entropy perovskites Sr(FeαTiβCoγMnζ)O3−δ as promising cathodes for intermediate temperature solid oxide fuel cell. Appl Catal B 295:120264 Shi Y, Ni N, Ding Q, Zhao X (2021) Tailoring high-temperature stability and electrical conductivity of high entropy lanthanum manganite for solid oxide fuel cell cathodes. J Mater Chem A 21–28

378

A. Tarancón et al.

Shijie Z, Na L, Liping S, Qiang L, Lihua H, Hui Z (2022) A novel high-entropy cathode with the A2 BO4 -type structure for solid oxide fuel cells. J Alloy Compd 895:162548 Shimada H, Ohba F, Li X, Hagiwara A, Ihara M (2012) Electrochemical behaviors of nickel/yttriastabilized zirconia anodes with distribution controlled yttrium-doped barium zirconate by ink-jet technique. J Electrochem Soc 159:F360–F367 Shin JW, Oh S, Lee S, Yu JG, Park J, Go D, Yang BC, Kim HJ, An J (2019) Ultrathin atomic layer-deposited CeO2 overlayer for high-performance fuel cell electrodes. ACS Appl Mater Interfaces Shin SS, Kim JH, Bae KT, Lee KT, Kim SM, Son JW, Choi M, Kim H (2020) Multiscale structured low-temperature solid oxide fuel cells with 13 W power at 500 °C. Energy Environ Sci 13:3459– 3468 Sholklapper TZ, Jacobson CP, Visco SJ, De Jonghe LC (2008) Synthesis of dispersed and contiguous nanoparticles in solid oxide fuel cell electrodes. Fuel Cells 8:303–312 Shri Prakash B, Senthil Kumar S, Aruna ST (2014) Properties and development of Ni/YSZ as an anode material in solid oxide fuel cell: a review. Renew Sustain Energy Rev 36:149–179 Singh M, Zappa D, Comini E (2021) Solid oxide fuel cell: decade of progress, future perspectives and challenges. Int J Hydrogen Energy 46:27643–27674 Singhal SC (2003) High temperature and solid oxide fuel cells, vol 16. Elsevier, Hoboken, NJ, USA Singhal SC, Kendall K (2003) High temperature and solid oxide fuel cells. Elsevier Skafte TL, Hjelm J, Blennow P, Graves C (2018) Reactivating the Ni-YSZ electrode in solid oxide cells and stacks by infiltration. J Power Sources 378:685–690 Sobolev A, Stein P, Borodianskiy K (2020) Synthesis and characterization of NiO colloidal ink solution for printing components of solid oxide fuel cells anodes. Ceram Int 46:25260–25265 Solovyev AA, Lebedynskiy AM, Shipilova AV, Ionov IV, Smolyanskiy EA, Lauk AL, Remnev GE, Maslov AS (2017) Scale-up of solid oxide fuel cells with magnetron sputtered electrolyte. Fuel Cells 17:378–382 Sønderby S, Klemensø T, Christensen BH, Almtoft KP, Lu J, Nielsen LP, Eklund P (2014) Magnetron sputtered gadolinia-doped ceria diffusion barriers for metal-supported solid oxide fuel cells. J Power Sources 267:452–458 Sønderby S, Christensen BH, Almtoft KP, Nielsen LP, Eklund P (2015) Industrial-scale high power impulse magnetron sputtering of yttria-stabilized zirconia on porous NiO/YSZ fuel cell anodes. Surf Coat Technol 281:150–156 Song Y, Zhang X, Zhou Y, Jiang Q, Guan F, Lv H, Wang G, Bao X (2018) Promoting oxygen evolution reaction by RuO2 nanoparticles in solid oxide CO2 electrolyzer. Energy Storage Mater 13:207–214 Spiridigliozzi L, Ferone C, Cioffi R, Accardo G, Frattini D, Dell’Agli G (2020) Entropy-stabilized oxides owning fluorite structure obtained by hydrothermal treatment. Materials 13:1–12 Steele BCH, Heinzel A (2001) Materials for fuel-cell technologies. Nature Su PC, Prinz FB (2012) Electrochemistry communications nanoscale membrane electrolyte array for solid oxide fuel cells. Electrochem Commun 16:77–79 Sukeshini MA, Cummins R, Reitz TL, Miller RM (2009) Ink-jet printing: a versatile method for multilayer solid oxide fuel cells fabrication. J Am Ceram Soc 92:2913–2919 Sukeshini AM, Meisenkothen F, Gardner P, Reitz TL (2013) Aerosol Jet® Printing of functionally graded SOFC anode interlayer and microstructural investigation by low voltage scanning electron microscopy. J Power Sources 224:295–303 Sun C, Sun J, Xiao G, Zhang H, Qiu X, Li H, Chen L (2006a) Mesoscale organization of nearly monodisperse flowerlike ceria microspheres. J Phys Chem B 110:13445–13452 Sun C, Xie Z, Xia C, Li H, Chen L (2006b) Investigations of mesoporous CeO2 -Ru as a reforming catalyst layer for solid oxide fuel cells. Electrochem Commun 8:833–838 Sun C, Li H, Chen L (2007) Study of flowerlike CeO2 microspheres used as catalyst supports for CO oxidation reaction. J Phys Chem Solids 68:1785–1790 Sun C, Hui R, Roller J (2010) Cathode materials for solid oxide fuel cells: a review. J Solid State Electrochem 14:1125–1144

Emerging Trends in Solid Oxide Electrolysis Cells

379

Sun J, Binner J, Bai J (2020) 3D printing of zirconia via digital light processing: optimization of slurry and debinding process. J Eur Ceram Soc 40:5837–5844 Sun C, Li H, Chen L (2012a) Nanostructured ceria-based materials: synthesis, properties, and applications. Energy Environ Sci 5:8475–8505 Sun W, Zhang N, Mao Y, Sun K (2012b) Preparation of dual-pore anode supported Sc2 O3 -stabilizedZrO2 electrolyte planar solid oxide fuel cell by phase-inversion and dip-coating. J Power Sources 218:352–356 Szász J, Wankmüller F, Wilde V, Störmer H, Gerthsen D, Menzler NH, Ivers-Tiffée E (2018) Nature and functionality of La0.58 Sr0.4 Co0.2 Fe0.8 O3−δ /Gd0.2 Ce0.8 O2−δ /Y0.16 Zr0.84 O2−δ interfaces in SOFCs. J Electrochem Soc 165:F898–F906 Tabei SA, Sheidaei A, Baniassadi M, Pourboghrat F, Garmestani H (2013) Microstructure reconstruction and homogenization of porous Ni-YSZ composites for temperature dependent properties. J Power Sources 235:74–80 Tai XY, Zhakeyev A, Wang H, Jiao K, Zhang H, Xuan J (2019) Accelerating fuel cell development with additive manufacturing technologies: state of the art, opportunities and challenges. Fuel Cells 19:636–650 Takahashi K, Fujita H, Ishikawa Y, Nakagaki T (2021) Microfabrication of anode functional layer in SOFC by 3D printer. MATEC Web Conf 333:17001 Tanaka H, Uenishi M, Taniguchi M, Tan I, Narita K, Kimura M, Kaneko K, Nishihata Y, Mizuki J (2006) The intelligent catalyst having the self-regenerative function of Pd, Rh and Pt for automotive emissions control. Catal Today 117:321–328 Tanasini P, Cannarozzo M, Costamagna P, Faes A, Van Herle J, Hessler-Wyser A, Comninellis C (2009) Experimental and theoretical investigation of degradation mechanisms by particle coarsening in SOFC electrodes. Fuel Cells 9:740–752 Tang C, Kousi K, Neagu D, Metcalfe IS (2021) Trends and prospects of bimetallic exsolution. Chem Eur J 27:6666–6675 Tarancón A (2009) Strategies for lowering solid oxide fuel cells operating temperature. Energies 2:1130–1150 Tarancón A, Esposito V, Torrell M, Vece MD, Son JS, Norby P (2022) 2022 roadmap on 3D printing for energy. J Phys Energy 4:011501 The D, Grieshammer S, Schroeder M, Martin M, Al Daroukh M, Tietz F, Schefold J, Brisse A (2015) Microstructural comparison of solid oxide electrolyser cells operated for 6100 h and 9000 h. J Power Sources 275:901–911 Tian Y, Zhang L, Liu Y, Jia L, Yang J, Chi B, Pu J, Li J (2019) A self-recovering robust electrode for highly efficient CO2 electrolysis in symmetrical solid oxide electrolysis cells. J Mater Chem A 7:6395–6400 Tietz F, Sebold D, Brisse A, Schefold J (2013) Degradation phenomena in a solid oxide electrolysis cell after 9000 h of operation. J Power Sources 223:129–135 Timurkutluk B, Timurkutluk C, Mat MD, Kaplan Y (2016) A review on cell/stack designs for high performance solid oxide fuel cells. Renew Sustain Energy Rev 56:1101–1121 Timurkutluk B, Altan T, Toros S, Genc O, Celik S, Korkmaz HG (2021a) Engineering solid oxide fuel cell electrode microstructure by a micro-modeling tool based on estimation of TPB length. Int J Hydrogen Energy 46:13298–13317 Timurkutluk C, Altan T, Yildirim F, Onbilgin S, Yagiz M, Timurkutluk B (2021b) Improving the electrochemical performance of solid oxide fuel cells by surface patterning of the electrolyte. J Power Sources 512:230489 Tomov RI, Krauz M, Jewulski J, Hopkins SC, Kluczowski JR, Glowacka DM, Glowacki BA (2010) Direct ceramic inkjet printing of yttria-stabilized zirconia electrolyte layers for anode-supported solid oxide fuel cells. J Power Sources 195:7160–7167 Tomov RI, Dudek M, Hopkins SC, Krauz M, Wang H, Wang C, Shi Y, Tomczyk P, Glowacki BA (2013) Inkjet printing of direct carbon solid oxide fuel cell components. ECS Trans 57:1359– 1369

380

A. Tarancón et al.

Tomov RI, Mitchel-Williams TB, Maher R, Kerherve G, Cohen L, Payne DJ, Kumar RV, Glowacki BA (2018) The synergistic effect of cobalt oxide and Gd-CeO2 dual infiltration in LSCF/CGO cathodes. J Mater Chem A 6:5071–5081 Tomov RI, Mitchel-Williams TB, Venezia E, Kawalec M, Krauz M, Kumar RV, Glowacki BA (2021) Inkjet printing infiltration of the doped ceria interlayer in commercial anode-supported SOFCs. Nanomaterials 11:1–14 Torrell M, Morata A, Kayser P, Kendall M, Kendall K, Tarancón A (2015a) Performance and long term degradation of 7 W micro-tubular solid oxide fuel cells for portable applications. J Power Sources 285:439–448 Torrell M, Almar L, Morata A, Tarancón A (2015b) Synthesis of mesoporous nanocomposites for their application in solid oxide electrolysers cells: microstructural and electrochemical characterization. Faraday Discuss 182:423–435 Travitzky N, Bonet A, Dermeik B, Fey T, Filbert-Demut I, Schlier L, Schlordt T, Greil P (2014) Additive manufacturing of ceramic-based materials. Adv Eng Mater 16:729–754 Tsai T, Barnett SA (1997) Effect of LSM-YSZ cathode on thin-electrolyte solid oxide fuel cell performance. Solid State Ionics 93:207–217 Tsekouras G, Neagu D, Irvine JTS (2013) Step-change in high temperature steam electrolysis performance of perovskite oxide cathodes with exsolution of B-site dopants. Energy Environ Sci 6:256–266 Tsoga A, Gupta A, Naoumidis A, Nikolopoulos P (2000) Gadolinia-doped ceria and yttria stabilized zirconia interfaces: regarding their application for SOFC technology. Acta Mater 48:4709–4714 Tucker MC (2020) Progress in metal-supported solid oxide electrolysis cells: a review. Int J Hydrogen Energy Uhlenbruck S, Moskalewicz T, Jordan N, Penkalla HJ, Buchkremer HP (2009) Element interdiffusion at electrolyte-cathode interfaces in ceramic high-temperature fuel cells. Solid State Ionics 180:418–423 Varghese G, Moral M, Castro-García M, López-López JJ, Marín-Rueda JR, Yagüe-Alcaraz V, Hernández-Afonso L, Ruiz-Morales JC, Canales-Vázquez J (2018) Fabricación y caracterización de cerámicas medinate impresión 3D DLP de bajo coste. Bol Soc Esp Ceram Vidrio 57:9–18 Venezia E, Viviani M, Presto S, Kumar V, Tomov RI (2019) Inkjet printing functionalization of SOFC LSCF cathodes. Nanomaterials 9:654 Virkar AV (2010) Mechanism of oxygen electrode delamination in solid oxide electrolyzer cells. Int J Hydrogen Energ 35:9527–9543 Vohs JM, Gorte RJ (2009) High-performance SOFC cathodes prepared by infiltration. Adv Mater 21:943–956 Wan Y, Zhao D (2007) On the controllable soft-templating approach to mesoporous silicates. Chem Rev 107:2821–2860 Wang S, Jiang V, Zhang Y, Van J, Li W (1998) The role of 8 mol% yttria stabilized zirconia in the improvement of electrochemical performance of lanthanum manganite composite electrodes. J Electrochem Soc 145:621 Wang HB, Xia CR, Meng GY, Peng DK (2000) Deposition and characterization of YSZ thin films by aerosol-assisted CVD. Mater Lett 44:23–28 Wang XM, Li CJ, Li CX, Yang GJ (2010) Microstructure and electrochemical behavior of a structured electrolyte/LSM-cathode interface modified by flame spraying for solid oxide fuel cell application. J Therm Spray Technol 19:311–316 Wang H, Ji W, Zhang L, Gong Y, Xie B, Jiang Y, Song Y (2011) Preparation of YSZ films by magnetron sputtering for anode-supported SOFC. Solid State Ionics 192:413–418 Wang F, Chen D, Shao Z (2012) Sm0.5 Sr0.5 CoO3−δ -infiltrated cathodes for solid oxide fuel cells with improved oxygen reduction activity and stability. J Power Sources 216:208–215 Wang F, Brito ME, Yamaji K, Cho DH, Nishi M, Kishimoto H, Horita T, Yokokawa H (2014a) Effect of polarization on Sr and Zr diffusion behavior in LSCF/GDC/YSZ system. Solid State Ionics 262:454–459

Emerging Trends in Solid Oxide Electrolysis Cells

381

Wang XM, Li CX, Huang JY, Yang GJ, Li CJ (2014b) Deposition mechanism of convex YSZ particles and effect of electrolyte/cathode interface structure on cathode performance of solid oxide fuel cell. Int J Hydrogen Energy 39:13650–13657 Wang C, Tomov RI, Mitchell-Williams TB, Kumar RV, Glowacki BA (2017) Inkjet printing infiltration of Ni-Gd:CeO2 anodes for low temperature solid oxide fuel cells. J Appl Electrochem 47:1227–1238 Wang Y, Li W, Ma L, Li W, Liu X (2020) Degradation of solid oxide electrolysis cells: phenomena, mechanisms, and emerging mitigation strategies—a review. J Mater Sci Technol 55:35–55 Wang Y, Wu C, Zu B, Han M, Du Q, Ni M, Jiao K (2021a) Ni migration of Ni-YSZ electrode in solid oxide electrolysis cell: an integrated model study. J Power Sources 516:230660 Wang Y, Jia C, Lyu Z, Han M, Wu J, Sun Z, Iguchi F, Yashiro K, Kawada T (2021b) Performance and stability analysis of SOFC containing thin and dense gadolinium-doped ceria interlayer sintered at low temperature. J Materiomics Wei L, Zhang J, Yu F, Zhang W, Meng X, Yang N, Liu S (2019) A novel fabrication of yttriastabilized-zirconia dense electrolyte for solid oxide fuel cells by 3D printing technique. Int J Hydrogen Energy 44:6182–6191 Wen Y, Liu Y (2021) Processing and microstructure of a fluorite high-entropy oxide (Zr0.2 Ce0.2 Hf0.2 Y0.2 Al0.2 )O2−δ . Ceram Int 48:2546–2554 Wen K, Lv W, He W (2015) Interfacial lattice-strain effects on improving the overall performance of micro-solid oxide fuel cells. J Mater Chem A 3:20031–20050 Wen Y, Yang T, Lee D, Lee HN, Crumlin EJ, Huang K (2018) Temporal and thermal evolutions of surface Sr-segregation in pristine and atomic layer deposition modified La0.6 Sr0.4 CoO3−δ epitaxial films. J Mater Chem A 6:24378–24388 Williford RE, Chick LA, Maupin GD, Simner SP, Stevenson JW (2003) Diffusion limitations in the porous anodes of SOFCs. J Electrochem Soc 150:A1067 Witte R, Sarkar A, Kruk R, Eggert B, Brand RA, Wende H, Hahn H (2019) High-entropy oxides: an emerging prospect for magnetic rare-earth transition metal perovskites. Phys Rev Mater 3:1–8 Wu H, Liu W, He R, Wu Z, Jiang Q, Song X, Chen Y, Cheng L, Wu S (2017) Fabrication of dense zirconia-toughened alumina ceramics through a stereolithography-based additive manufacturing. Ceram Int 43:968–972 Wu T, Zhang W, Li Y, Zheng Y, Yu B, Chen J, Sun X (2018) Micro-/nanohoneycomb solid oxide electrolysis cell anodes with ultralarge current tolerance. Adv Energy Mater 8:1–7 Wu P, Tian Y, Lü Z, Zhang X, Ding L (2022) Electrochemical performance of La0.65 Sr0.35 MnO3 oxygen electrode with alternately infiltrated Sm0.5 Sr0.5 CoO3−δ and Sm0.2 Ce0.8 O1.9 nanoparticles for reversible solid oxide cells. Int J Hydrogen Energy 47:747–760 Xing B, Yao Y, Meng X, Zhao W, Shen M, Gao S, Zhao Z (2020a) Self-supported yttria-stabilized zirconia ripple-shaped electrolyte for solid oxide fuel cells application by digital light processing three-dimension printing. Scripta Mater 181:62–65 Xing B, Cao C, Zhao W, Shen M, Wang C, Zhao Z (2020b) Dense 8 mol% yttria-stabilized zirconia electrolyte by DLP stereolithography. J Eur Ceram Soc 40:1418–1423 Xu M, Wang W, Zhong Y, Xu X, Wang J, Zhou W, Shao Z (2019) Enhancing the triiodide reduction activity of a perovskite-based electrocatalyst for dye-sensitized solar cells through exsolved silver nanoparticles. J Mater Chem A 7:17489–17497 Yan Z, Hara S, Shikazono N (2017) Effect of powder morphology on the microstructural characteristics of La0.6 Sr0.4 Co0.2 Fe0.8 O3 cathode: a kinetic Monte Carlo investigation. Int J Hydrogen Energy 42:12601–12614 Yang H, Zhao D (2005) Synthesis of replica mesostructures by the nanocasting strategy. J Mater Chem 15:1217–1231 Yang L, Miyanaji H (2017) Ceramic additive manufacturing: a review of current status and challenges. Solid Freeform Fabrication 2017:652–679 Yang J, Ma W, Yu J, Chen X, Sun H, Xie Y (2014) Study on the pore-formers for porous anode substrates of solid oxide fuel cell. Xiyou Jinshu Cailiao Yu Gongcheng/rare Metal Mater Eng 43:269–273

382

A. Tarancón et al.

Yang SM, Lee S, Jian J, Zhang W, Lu P, Jia Q, Wang H, Noh TW, Kalinin SV, MacManus-Driscoll JL (2015) Strongly enhanced oxygen ion transport through samarium-doped CeO2 nanopillars in nanocomposite films. Nat Commun 6:8588 Yang Q, Wang G, Wu H, Beshiwork BA, Tian D, Zhu S, Yang Y, Lu X, Ding Y, Ling Y, Chen Y, Lin B (2021a) A high-entropy perovskite cathode for solid oxide fuel cells. J Alloy Compd 872:159633 Yang Y, Bao H, Ni H, Ou X, Wang S, Lin B, Feng P, Ling Y (2021b) A novel facile strategy to suppress Sr segregation for high-entropy stabilized La0.8 Sr0.2 MnO3−δ cathode. J Power Sources 482:228959 Ye L, Zhang M, Huang P, Guo G, Hong M, Li C, Irvine JTS, Xie K (2017) Enhancing CO2 electrolysis through synergistic control of non-stoichiometry and doping to tune cathode surface structures. Nat Commun 8:1–10 Yoon J, Cho S, Kim JH, Lee J, Bi Z, Serquis A, Zhang X, Manthiram A, Wang H (2009) Vertically aligned nanocomposite thin films as a cathode/electrolyte interface layer for thin-film solid oxide fuel cells. Adv Func Mater 19:3868–3873 Yoon KJ, Biswas M, Kim HHJ, Park M, Hong J, Kim HHJ, Son JW, Lee JH, Kim BK, Lee HW (2017) Nano-tailoring of infiltrated catalysts for high-temperature solid oxide regenerative fuel cells. Nano Energy 36:9–20 Young D, Sukeshini AM, Cummins R, Xiao H, Rottmayer M, Reitz T (2008) Ink-jet printing of electrolyte and anode functional layer for solid oxide fuel cells. J Power Sources 184:191–196 Zhang H, Zhao F, Chen F, Xia C (2011) Nano-structured Sm0.5 Sr0.5 CoO3−δ electrodes for intermediate-temperature SOFCs with zirconia electrolytes. Solid State Ionics 192:591–594 Zhang W, Zheng Y, Yu B, Wang J, Chen J (2017) Electrochemical characterization and mechanism analysis of high temperature Co-electrolysis of CO2 and H2 O in a solid oxide electrolysis cell. Int J Hydrogen Energy 42:29911–29920 Zhang J, Wei L, Meng X, Yu F, Yang N, Liu S (2020a) Digital light processing-stereolithography three-dimensional printing of yttria-stabilized zirconia. Ceram Int 46:8745–8753 Zhang X, Wu X, Shi J (2020b) Additive manufacturing of zirconia ceramics: a state-of-the-art review. J Market Res 9:9029–9048 Zhang F, Cheng F, Cheng C, Guo M, Liu Y, Miao Y, Gao F, Wang X (2022) Preparation and electrical conductivity of (Zr,Hf,Pr, Y,La)O high entropy fluorite oxides. J Mater Sci Technol 105:122–130 Zhao D, Wan Y, Zhou W (2013) Ordered mesoporous materials. Wiley-VCH Verlag GmbH & Co, KGaA, Weinheim, Germany Zheng Y, Wang J, Yu B, Zhang W, Chen J, Qiao J, Zhang J (2017) A review of high temperature co-electrolysis of H2 O and CO2 to produce sustainable fuels using solid oxide electrolysis cells (SOECs): advanced materials and technology. Chem Soc Rev 46:1427–1463 Zhong Y, Fang H, Ma Q, Dong X (2018) Analysis of droplet stability after ejection from an inkjet nozzle. J Fluid Mech 845:378–391 Zhou J, Shin TH, Ni C, Chen G, Wu K, Cheng Y, Irvine JTS (2016) In Situ growth of nanoparticles in layered perovskite La0.8 Sr1.2 Fe0.9 Co0.1 O4-δ as an active and stable electrode for symmetrical solid oxide fuel cells. Chem Mater 28:2981–2993 Zhu K, Yang D, Yu Z, Ma Y, Zhang S, Liu R, Li J, Cui J, Yuan H (2020) Additive manufacturing of SiO2 –Al2 O3 refractory products via direct ink writing. Ceram Int 46:27254–27261 Zhu Z, Gong Z, Qu P, Li Z, Rasaki SA, Liu Z, Wang P, Liu C, Lao C, Chen Z (2021) Additive manufacturing of thin electrolyte layers via inkjet printing of highly-stable ceramic inks. J Adv Ceram 10:279–290 Zouridi L, Garagounis I, Vourros A, Marnellos GE, Binas V (2022) Advances in inkjet-printed solid oxide fuel cells. Adv Mater Technol 2101491:1–32

Stack/System Development for High-Temperature Electrolysis Hamza Moussaoui, Vanja Suboti´c, Jan Van herle, Ligang Wang, Xinyi Wei, and Hangyu Yu

Abstract In order to design a high performing and durable SOE stack or system, all the relevant length-scales should be addressed using different approaches and tools, and all the components of the Balance-of-Plant need to be properly dimensioned and integrated. This chapter adresses the requirements of building an SOE stack and system with an optimised performance and durability.

1 Introduction In order to design a high performing and durable SOE stack or system, all the relevant length-scales should be addressed using different approaches and tools (c.f. Fig. 1). For instance, on the atomic scale, density functional theory (DFT) allows to screen the most energetically favourable materials configuration. These materials are then used to manufacture electrodes and cells with different methods, like screen printing and tape casting. On the cell microstructural level, different properties are sought, for example increasing the density of active sites (triple phase boundaries, TPB), providing sufficient porosity for gas diffusion, and ensuring chemical and mechanical compatibility between the different layers (Moussaoui et al. 2018, 2020). These cells are then stacked by using interconnects to modulate the electrolyser power. Finally, the produced stack is integrated with the balance of plant (BoP) components, described further in this chapter, to obtain the final system. Several approaches and tools are employed to ensure highly performing and durable SOE systems and stacks. First, experimental testing is one of the main pillars. Indeed, tests of symmetrical-cells, button-cells, segmented-cells, short-stacks, and full-stacks allow to acquire complementary insights. Performance mapping, longterm durability tests, and accelerated stress tests (AST) are used in this case to better understand the performance and durability of the tested objects. During these tests, advanced online monitoring and diagnostic tools, like electrochemical H. Moussaoui (B) · V. Suboti´c · J. Van herle · L. Wang · X. Wei · H. Yu Group of Energy Materials, École Polytechnique Fédérale de Lausanne (EPFL), 1951 Sion, Switzerland e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M. A. Laguna-Bercero (ed.), High Temperature Electrolysis, Lecture Notes in Energy 95, https://doi.org/10.1007/978-3-031-22508-6_12

383

384

H. Moussaoui et al.

Fig. 1 Illustration of the multi-scale and multi-approach methodology

impedance spectroscopy (EIS) and total harmonic distortion (THD) are used for optimal, fault-free operation. Post-test characterization, for instance via scanning electron microscopy (SEM) and energy-dispersive X-ray spectroscopy (EDS), bring very useful insight into the potential degradation mechanisms. These latest generally manifest as secondary phase formation, morphological change, poisoning or mechanical damage. Complementarily to these experimental aspects, modelling is another fundamental pillar to support the measurements and simulate a large number of tests, thus saving a large amount of time and effort (Moussaoui et al. 2019). Finally, prognosis and lifetime prediction tools are key to foresee performance evolution of SOE stacks and thus ensure optimal operation over time. This chapter addresses the points mentioned above and provides insight into high-temperature electrolysis systems, thus taking into account their core component solid oxide cell (SOC) stack and balance of plant. Moreover, in order to understand their behaviour, the failure modes that can inhibit the SOC performance and morphology are briefly introduced and possibilities for their identification and prevention are discussed. Eventually, the modelling approaches used to predict and improve performance of SOE-based systems are discussed here.

Stack/System Development for High-Temperature Electrolysis

385

2 Stack Components and Requirements The core component of a SOE is a solid oxide electrolysis cell (SOEC). In order to optimize the performance of these cells, different designs have been proposed to date, as it will be discussed in the next chapter; the most obvious distinction being tubular and planar geometries. Because of their manufacturing advantages, planar cells have been established as the most common design. Taking into account the design of single cells, we can distinguish between electrolyte-supported (ESC) and electrodesupported cells, depending on which layer is used as the mechanical cell support. Electrolyte-supported cells, with the higher thickness of the electrolyte layer, provide an advantage of mechanical stability. They can withstand a negative impact of certain electrode failure modes, e.g. Ni-reoxidation that causes strong volume increase and mechanical damage, if the electrolyte thickness is lower than the electrode thickness. The higher electrolyte thickness on the other hand causes higher ohmic losses and thus the need to increase the operation temperature. Considering electrode-supported cells, mainly fuel electrode-supported cells (FESC) are employed. The processes that occur on the air electrode, oxygen reduction or evolution, are related to significant losses. In order to reduce these, the thickness of the air electrode layer is minimized and therefore the air electrode-supported SOC type is not common for commercial applications. FESCs provide higher efficiency even at lower temperatures, since the electrolyte thickness is significantly lower and a higher amount of catalyst is available within the thicker fuel electrode. However, this brings specific risks such as undesired damage while operating the cells, which occurs predominantly as a consequence of Ni-reoxidation degradation. Different cell and stack manufacturers produce cells with different sizes, which influence the cell behaviour, such as the maximum current density and the temperature distribution. The size of cells relevant for industrial application usually ranges between 22 and 128 cm2 as discussed in (Léon et al. 2021; Larsen et al. 2001). Single cells with an active surface area as high as 900 cm2 have been manufactured and successfully tested by TNO in the Netherlands (VoltaChem 2022). Nevertheless, because the cells are mainly made of ceramics, their increasing size can be correlated with increasing temperature gradients and thus higher risk of mechanical damage. Taking into account the typical operating conditions for the electrolysis mode, the maximum voltage of an electrolysis cell lies between 1.3 and 1.5 V. In order to reach higher voltages and thus higher fuel production rates, a number of cells should be stacked together. This is related to the serial connection of cells within a stack, which further requires the presence of additional components. These are interconnects, current collectors, sealing and a compression system. The interconnects enable electrical contact, fuel and air distribution and are often responsible for current distribution. If necessary, additional current collectors are used. Fuel and flow distribution along the cells and stack should be uniform and for this purpose an appropriate flow field is required. Designing the flow field on both feed sides of all cells (fuel and air) is a very important aspect while designing a stack. The flow field design has a particular impact on the reactions that occur and thus on the current and

386

H. Moussaoui et al.

temperature distribution along the cells and stack. This, in turn, impacts the overall stack performance. The main concepts applied are co-flow, counter-flow and crossflow of fuel and air (Suboti´c et al. 2020a, b; Kendall et al. 2003). Different stack manufacturers employ different types of flow fields, which are mainly optimized to the manufacturing procedure. It is, however, very important to take into account the operating environments when making decisions about the flow field design. Air flow is usually used to control the temperature gradient, as it enters the stack at a lower temperature than the fuel. Therefore, if exothermal reactions are expected to be predominant at the stack inlet, co-flow configuration is favourable, since the cooling effect of air at the stack inlet can make the temperature distribution along the stack more uniform. In the opposite case, if endothermal reactions are expected at the stack inlet, the counter-flow configuration can be favourable, since an additional temperature reduction at the stack inlet can be avoided, thus ensuring a more uniform temperature distribution along the cells and stack. Moreover, different flow field patterns can be implemented on the electrodes, e.g. Z-flow, serpentine-flow, radial flow, spiral flow, etc. (Kendall et al. 2003). One single unit containing all the mentioned components is known as a stack repeat unit, whereby a number of repeat units is connected into a stack. Very detailed overviews of the stack components and their requirements are provided in previous chapters. In Fig. 2, five different commercially available stacks are compared with each other, clearly showing that stack design significantly differs between manufacturers. This also includes SOC type and size, flow field design, number of cells in a stack and many other characteristics.

Fig. 2 Pictures of six different stack types prepared for testing on the laboratory scale (Preininger et al. 2020; Sampathkumar et al. 2022; Topsoe 2022)

Stack/System Development for High-Temperature Electrolysis

387

3 Balance of Plant (BoP) and Available Systems High efficiency of SOEs allows the to scale in different fields, such as household applications, on-site application for industrial systems or even centralized hydrogen and syngas production. When scaling-up SOE-based systems, diverse technical challenges and obstacles arise. These are related to both the stack and balance of plant (BoP), thus including steam generators, reforming reactors, electronics, heat exchangers, blowers, pumps and compressors. The major technological improvements over time are expected to be achieved for the SOC stacks, especially to increase their lifespan and to reduce undesired degradation phenomena, while other components belong to more established technologies. Although the BoP components are not considered as technological bottlenecks towards the large-scale SOE systems deployment, their specifications and integration design are crucial to optimize the system’s efficiency and durability. The requirements of the main BoP components are summarized hereafter. Water Treatment System: In order to avoid the poisoning and clogging of the stack and pipelines, water needs to have a sufficient level of purity. For example, the company Elcogen cleans tap water to Type I (ultrapure) before feeding their electrolyser (Lehtinen and Noponen 2021). The quality of water can be checked via its resistivity (measured in M]cm) or conductivity (measured in µScm−1 ), which reveals inorganic impurities. A water treatment system may include the following filters: reverse osmosis filter, demineralization filter, carbon filter, sediment filter, resin filter, etc. Furthermore, a desalinization filter is required for near-shore or off-shore applications where sea water is used. Air filter: Not only water needs to be purified, but also ambient air fed to the positive electrode of the stack. The most common pollutant is sulphur dioxide (SO2 ), with typical concentrations ranging from 10 to 20 ppb in cities, and higher in polluted zones. Moisture is also undesired in air. Absolute humidity commonly ranges from 0 to 4% but could be higher in coastal locations. Air and water filters should be easily accessible in order to facilitate their replacement which is required on a frequent basis. Air blower: Air is fed to the stack for two purposes: to sweep the O2 produced at the positive electrode and to regulate the stack temperature. An air blower is used in this frame to provide sufficient flow rate. Electrical heater: An electrical heater is only used for the system start-up when no sufficient heat is readily available. As soon as the stack reaches its nominal operation point, and produces enough internal heat, the electrical heating is cut off progressively. Burner: A fraction of the produced fuel (up to around 10%) is sometimes burned in a burner to produce heat required for the nominal functioning of the system.

388

H. Moussaoui et al.

Heat Exchange Network (HEN): In general, a dozen of heat exchangers is required to transfer the heat from the hot parts (burner, stack exhaust, etc.) to the relevant locations (water and air feed, etc.). Heat exchangers are characterized by their surface exchange area (expressed in m2 ) and the largest exchange area is typically the one between the burner outlet and the air inlet. Furthermore, a specific heat exchanger is used to dry the produced humidified hydrogen by simply cooling down the gas and thus condensing humidity into water. The condensate is then partly recirculated. Recirculation network: Not only water is recirculated, but also a fraction of the produced H2 is recirculated to the stack inlet to prevent Ni oxidation in the fuel electrode. More details are provided further in this chapter. Compressors: Two kinds of compressors are generally used in series, to compress the produced hydrogen from ambient pressure to typically 30 bar, and up to a standard value of 200 bar or higher. Tanks: Produced hydrogen is stored in adapted tanks rated for a specific pressure and volume. Safety equipment: For safety reasons, a ventilation system is usually used to prevent gas accumulation and thus increased risk of explosion in case of fuel leakage. Moreover, specific sensors are used to monitor the concentration of inflammable gases. Power electronics: Power electronics mainly include AC/DC converters that convert alternative current from the grid to direct current required by the stack. These converters should be designed to have low power losses. Monitoring equipment includes the equipment presented below. Thermocouples: Thanks to their low cost and compactness, K-type thermocouples are abundantly used in different locations to monitor the temperature of the different BoP components. Redundant pairs of thermocouples are used in some key locations, like at the burner, for security reasons. Electrical measuring instruments: These are used for current and voltage measurement. Mass-flow controllers and actuators: Their application is necessary in order to adjust the different gas flow rates. Pressure sensors: These are employed to measure the pressure at different key locations. Programmable logical controller PLC: The control system is required to acquire and process all the measured data. It is also used to control the SOE system. It should be noted that a maintenance schedule is necessary to ensure optimal and continuous operation of the SOE system. The cost of the BoP components and their maintenance are not negligible, especially for small-scale systems. However, it is a necessary investment to maintain high performance and avoid any extrinsic degradation of the stack. When scaling-up the system, the costs expected per kW are

Stack/System Development for High-Temperature Electrolysis

389

significantly reduced as the system size increases, as reported by Garðasson et al. (2016). The mentioned study reported that, e.g. when increasing the system size from 10 kW up to 100 MW, the costs are reduced as follows: 104 $/kW down to 7 $/kW for heat exchangers, 723 $/kW down to 46 $/kW for rectifiers, 193 $/kW down to 12 $/kW for compressors and an assumed 69 $/kW down to 4 $/kW for other BoP. A large stack module (LSM) prototype manufactured by SolydEra, Switzerland, which integrates a new generation of stacks suitable for a reversible operation, reaching 25 kWel in SOFC mode and 75 kWDC in SOE mode, has already been successfully tested at CEA-Liten, France, presenting a steam conversion rate higher than 80% (Lehtinen and Noponen 2021). The operation of such a LSM prototype, as shown in Fig. 3, requires the application of a 60 kW industrial steam generator, a 22 kW air compressor, two 30 kW air preheaters, four reversible 30 kWDC power supplies and a number of different actuators and sensors in order to generate 50 kg of hydrogen per day. As a representative example of a large real-world SOE-based system, operated in an industrial environment, Fig. 4 provides an insight into the design of a reversible SOC system consisting of two identical modules, which is manufactured by Sunfire, installed at the site of the combined cycle gas turbine (CCGT) power plant Mellach in Austria and operated within the Austrian national project HotFlex. The provided system design presumes continuous operation in electrolysis mode to generate 40 m3 /h of hydrogen when supplying it with 150 kWAC . In fuel cell mode, the system is designed to be employed only in emergency cases as an uninterrupted power supply (UPS) providing 20 kWAC . This includes utilization of natural gas, which is firstly pre-reformed in a reforming reactor in order to generate hydrogen-rich syngas to supply the SOC stacks. A desulphurization unit is also required for such kind of systems, in order to remove sulphur impurities from the fuel, thus preventing undesired degradation mechanisms of the system components. Moreover, both operating modes require continuous supply with steam. Thus, the operating environment has

Fig. 3 SolydEra’s large stack module (LSM) (SolydEra 2022; Liten 2022; Aicart et al. 2022)

390

H. Moussaoui et al.

to be adjusted in order to ensure safe operation of the mentioned system components as well as the remaining balance of plant. Moving towards the MW scale, the leading solution worldwide is provided by the Sunfire-HyLink SOE electrolyser, the design of which is shown in Fig. 5. Such large-scale electrolysers are meant to exploit waste heat from different industrial processes, which is freely available, based on which the operating costs can be significantly reduced. This also significantly decreases the overall operational costs of high-temperature electrolysers compared to low temperature electrolysers, thus further increasing their attractiveness for large-scale application. The HyLink SOE electrolysers also enable to utilize the advantage of the hightemperature electrolysers and to produce not only hydrogen, but also syngas as an essential feedstock for industry decarbonization. The hourly produced amount of

Fig. 4 150 kW rSOC module positioned at the power plant site Mellach in Austria (Köningshofer et al. 2021; Verbund 2022)

Fig. 5 Sunfire-HyLink SOE (Sunfire 2022)

Stack/System Development for High-Temperature Electrolysis

391

hydrogen amounts to 750 m3 while utilizing 860 kg/h of steam. Moreover, its dynamic operation ensures to vary the hourly hydrogen generation between 5 and 100% in a very efficient manner. As such, they can be installed near to steel or chemical industries, refineries, etc. In general, because of their modular design, as illustrated in the two previous figures, the flexible system scaling between low- and high-power scale is achievable. Recently, the interest in direct electrochemical utilization of CO2 , either pure or mixed with steam, has distinctly increased. Significant efforts have been put into the development and investigation of co-electrolysis and CO2 electrolysers. Electrochemical conversion of CO2 ad H2 O into syngas enables to subsequently liquidize it and generate e-fuels by employing a Fischer–Tropsch-process. Within the project MegaSyn, the first demonstration plant worldwide for syngas production in an industrial environment on MW scale by employing co-electrolysis will be designed, built and demonstrated. The demonstration period should last 2 years to generate more than 900 tonnes of syngas by utilizing renewable energy. Besides the core system component—SOC stack—a pre-reformer unit presents an unavoidable part of the system, responsible for feed stream preparation (MegaSyn 2022). Solutions for direct CO2 utilization and generation of pure CO are provided by Haldor Topsoe, known as eCOs™ carbon monoxide generator, the core unit of which is a SOEC stack. This system is designed for on-demand CO production, thus excluding the necessity for transport and storage of toxic carbon monoxide. In order to produce 1 Nm3 of CO, 6–8 kWh and 1 Nm3 of CO2 are utilized. The air side can be supplied either by air or CO2 . Nevertheless, it is of crucial importance to consider the presence of different impurities in CO2 , since those vary as a function of the CO2 source and can have a detrimental impact on the SOE performance and morphology, as described below. The standard eCOs™ units available on the market offer a product gas containing 99 vol% CO; the remaining 1 vol% seems to be mainly CO2 , whereby H2 O and CH4 are present in a concentration lower than 5 ppm. The presence of higher hydrocarbons is thus avoided. When performing specific modifications, the higher CO purity of even 99.999 vol% can be reached. The first such demonstration plant was installed at Gas Innovations site at La Porte in Texas, USA in January 2016, to hourly produce 3–5 Nm3 of CO with a purity grade of 99.95% (Topsoe 2022; Küngas et al. 2017). In 2021, Haldor Topsoe reported to invest in yearly manufacturing of SOEC stacks with a capacity of 500 MW, which should further be expanded to 5 GW. The efficiency of the Haldor Topsoe stacks was presented to be higher by 30% in comparison to the efficiency of commercial alkaline or PEM electrolysers.

4 Power-to-X The Paris Agreement firstly introduced the concept of achieving net zero emission by 2050. Thus, the implementation of renewable energies is essential. The high

392

H. Moussaoui et al.

penetration of intermittent renewable energy sources requires large-scale and longterm power storage, e.g. in the form of energy vectors (Wang et al. 2019a, b). The concept of power-to-X, which converts renewable power to various gases (power-togas (Schiebahn et al. 2015; Samsatli and Samsatli 2019), e.g. H2 , CH4 and syngas) or liquids (power-to-liquid, e.g. methanol and dimethyl ether) and other chemicals (power-to-chemical, e.g. formic acid and waxes) (Reddy et al. 2021), has recently gained an increasing interest. Compared with alkaline or proton-exchange membrane electrolysis, solid oxide electrolysis (SOE) is advantageous due to: (1) higher system efficiency, (2) CO2 electrolysis to produce CO (Zhang et al. 2013), (3) direct syngas production by co-electrolysis (CE) of steam and CO2 , (4) heat management opportunity with other industrial processes and (5) its reversible system (SOFC) can be used to balance energy demand. Applying the concept of using renewable energies to produce carbon–neutral synthetic fuels is linked to so-called electro-fuels or e-fuels. Hydrogen is a direct product from steam electrolysis. Water is vaporized and superheated for this purpose, and mixed with the recirculated fuel electrode off-gas to be used as feed stream to SOE. Subsequently, the steam is electrochemically converted to hydrogen with a very high efficiency of above 85% (Arslan et al. 2022). The outlet gas must, however, be cooled to condense water. The remaining gas should be compressed and processed usually by a pressure swing adsorption unit (PSA) to achieve a certain purity of H2 (over 99.999 vol%)) (Yang et al. 2008), which can be further compressed for other applications. However, the main bottleneck of steam electrolysis system design is the significant amount of heat required for steam generation. Moreover, the hydrogen future is still heavily depending on its challenges and innovations in some specific areas such as hydrogen storage and transportation. This requires new solutions for easier storage of the produced hydrogen or rather its further conversion into liquid fuels. Green ammonia seems to be an attractive solution not only as storage medium but also energy carrier, the production of which is made using nitrogen and hydrogen as gases and specifically designed electrocatalytic approaches. Pure nitrogen is, however, required here and it is mainly made available by separating from air employing the air separation unit. The process of green ammonia production is also known as power-to-ammonia (PtA) (Wen et al. 2022). Ammonia can be further utilized for different purposes, either in the agro industry or as a fuel (ammonia-to-power A2P), e.g. directly in solid oxide fuel cells (SOFC). Compared with hydrogen or subsequently ammonia as the main products, fuel synthesis is also a promising downstream processing solution. The first option is a methanation process, during which H2 generated during the electrolysis operation and CO2 are utilized to produce synthetic methane. The methanation process is also known as power-to-methane (PtM). For SOE-based power-to-methane, the heat required can be largely supported by the heat released from the exothermic methanation reaction. Furthermore, properly selecting the SOE design points may allow for thermally self-sufficient PtM that requires no additional heating. However, when making a choice on the best appropriate process pathway when generating gaseous fuel, both regional resources and system boundaries, e.g. gas network tolerance towards hydrogen, gas buffering, heat application or mobility,

Stack/System Development for High-Temperature Electrolysis

393

have to be considered (Suboti´c et al. 2016). At the beginning of 2019, 153 power-togas (PtG) projects were running in 22 countries, as reported by Thema et al. (2019). The gas produced has been injected directly into the gas grid in nearly half (45%) of the projects. However, following the regulations regarding the amount of hydrogen that is allowed to be injected into the gas grid, the majority of those projects (65%) has been devoted to the methanation process and production of synthetic methane. A further possibility is direct electrochemical conversion of different gases in solid oxide electrolysers—either pure CO2 to produce CO or both H2 O and CO2 , which enables to generate syngas. Systems available for both cases are discussed in the previous section. The produced syngas mainly contains H2 and CO, but also low concentrations of CO2 and H2 O, or even CH4 . In the further step, employing the Fischer–Tropsch synthesis, syngas can be processed and converted into a liquid fuel mainly based on hydrocarbons (Mehariya et al. 2019). The main advantage of liquid fuels is their density and ease of storage and transportation, in addition to their advantage of higher calorific value.

5 Reliability and Durability Despite all the advantages of SOE technology, specific issues must still be addressed to prolong its lifetime and to inhibit undesired performance loss over time. In order to increase the competitiveness of SOC technology on the global market, the reliability and durability of the system must be increased. Besides materials improvement, a very important aspect of the system reliability is the operating reliability, which involves monitoring of the state-of-the-health (SoH) employing appropriate monitoring tools, determining different faults, isolate them and taking subsequent counteractions. For the latter, degradation level analysis is required for lifetime inference and design of the appropriate mitigations and/or regeneration strategies.

5.1 Degradation Degradation can occur on different levels: single cells, stacks or BoP. Depending on the operating environment, single cell performance and morphology can be degraded by a number of different mechanisms, e.g. carbon deposition, poisoning by different impurities contained in both conventional and alternative fuels (i.e. sulphur, chlorine, phosphine, silica), fuel starvation and air starvation that cause morphology alternations, and many others. Moving towards the stack and system level, all the accompanying components are mainly fabricated from metallic materials, and are thus subjected to high-temperature corrosion (Reddy et al. 2021). This, in turn, also leads to the chromium evaporation issue and subsequent electrode poisoning. More detailed insight into diverse degradation modes that can occur while operating solid

394

H. Moussaoui et al.

oxide electrolysers and their impact on the cell performance and morphology is given in Chap. 10. A very detailed overview of the degradation rates reported for the SOE stacks operated in steam-, co- and CO2 -electrolysis mode between 2010 and 2019 can be found in the work of Hauch (2020). The author reported the lowest degradation rate in the case of steam electrolysis to be 0.56%/1000 h for a commercially available stack (manufacturer not mentioned) after only 1000 h of operation. This value slightly increased to 0.64%/1000 h, for the Forschungszentrum Jülich stack operated for 21,363 h, thus presenting the most representative value with the longest operating time available in the literature to date. The results presented for CO2 -electrolysis are solely available for Haldor Topsoe stacks and the lowest degradation rate observed amounted to 0.13%/1000 h after 7034 h. Moving towards co-electrolysis, the data are available for Topsoe Fuel Cell, Fraunhofer IKTS and SOLIDpower stacks, reporting the lowest degradation rate for the SOLIDpower stack tested at EPFL to be 0.34%/1000 h after 4445 operating hours.

5.2 State-of-the-Health Diagnosis for Reliable System Operation In order to ensure reliable long-term operation of systems in real-world operating environments and to prevent unexpected system shut-down due to the possible failure modes that can occur in different system components, on-board diagnosis of the system state-of-the-health is necessary. This enables to identify degradation at its preliminary stage, to define degradation causes and eventually take appropriate counteractions to prolong the system lifespan. A list of available online diagnostic tools is provided by Suboti´c and Hochenauer (2022a), and this consists of temperature monitoring using, e.g. thermocouples or thermography, gas analysis tools, mass spectrometry, voltage and current monitoring, polarization curve measurement, electrochemical impedance spectroscopy or total harmonic distortion analysis. However, not all of these tools are suitable for system application, since, e.g. access to the individual cells is not ensured, or even the application of specific tools could significantly increase the system complexity and cost, and cause potential system damage. With this in mind, commercial systems require simple tools that can provide all the necessary information in a time- and cost-efficient manner. The first and unavoidable methodology to be used is voltage and current monitoring. It provides information about the general stack state-of-health at a specific operating point. The changes that occur can be correlated with general performance alternations, but no exact conclusion can be made in this case. More information can be retrieved when measuring the polarization curve, voltage as a function of the operating current, or the opposite. In order to obtain more detailed insight into the processes that occur as a function of the operating environment, alternating current or voltage signal should be superimposed to the running system.

Stack/System Development for High-Temperature Electrolysis

395

The more conventional tool to perform in-operando state-of-health analysis is electrochemical impedance spectroscopy. It is an established non-destructive methodology used to analyse processes that run in different electrochemical systems, during which an alternating current or voltage is superimposed to the running system. The system response is measured and impedance can be calculated. However, in order to apply this methodology, a linear, time-invariant, steady state and causal system is required. The requirements mentioned must be confirmed before starting the measurements. For example, in order to ensure the system linearity, the applied signal amplitude has to be adjusted—it should not be too low in order to maximize the signal/noise ratio and not too high in order to retain the linear response (Barsoukov et al. 2013, Suboti´c et al. 2016). However, since several processes can be overlapped in impedance spectra, the advanced tool distribution of relaxation times (DRT) is often used to separate the measured impedance data into individual peaks and to correlate these with specific processes. Very detailed insight into process mechanisms that can occur while operating solid oxide cells in electrolysis mode and their specific frequencies are represented in Fig. 6. The applicability of these methodologies for solid oxide electrolysers has been verified multiple times in the literature. A representative example is shown in Fig. 7, for which EIS and DRT were applied to identify the impact of chlorine poisoning on SOE operation. Voltage monitoring, presented in the first diagram, shows significant performance loss as the concentration of chlorine was increasing, which is associated with the increasing voltage. A slight increase of polarization resistance and specific DRT peaks, as shown in the second and third diagrams, could be correlated with

Fig. 6 Overview of process mechanisms presented by their specific frequencies based on the literature available (Suboti´c and Hochenauer 2022a)

396

H. Moussaoui et al.

Fig. 7 Applying voltage measurement, electrochemical impedance spectroscopy and distribution of relaxation times to identify SOE poisoning with chlorine (Thema et al. 2019)

charge transfer processes, thus leading to the conclusion that electrochemically active sites were mainly impacted by chlorine. The EIS characterization technique can also be used to determine behaviour of large stacks and systems, the preliminary results of which are available for the 150 kW rSOC module system manufactured by Sunfire (Suboti´c et al. 2022b), as presented in Fig. 4. In order to decrease the measuring time and ensure rapid identification of different failure modes at their preliminary stage, the total harmonic distortion tool seems to be very suitable to achieve this goal. Its application supposes that the linearity criterion is neglected and distortions from the linear state are monitored (Suboti´c et al. 2020a, 2021). This means that the THD tool provides information related to all undesired changes that occur within the system under operation. This principle has already been used in the electrical engineering field with the aim to reduce non-linear loads of power sources and thus does not present a completely new methodology. However, its application to analyse very complex mechanisms that occur within solid oxide cells is still challenging. The preliminary results presented in the literature for SOC technology confirm the THD suitability for application to real-world systems. A representative example is shown in Fig. 8a. In this case, a SOFC single cell was fuelled with humidified methane under conditions specifically designed to initiate carbon deposition on the cell fuel electrode and thus cause performance and morphology degradation. The signal distortions from the linear state have been observed at frequencies of 2, 4 and 8 kHz, as well as for frequencies lower than 100 Hz. As the amount of carbon increased, the height of the peaks observed also increased, thus referring to the intensified effect of the degradation phenomenon observed. Besides the linear scaling of the results observed, a more appropriate way to visualize the measured data and use it for practical application is shown in Fig. 8b. The colours shown present the intensity

Stack/System Development for High-Temperature Electrolysis

397

of the changes, which is correlated to the percentage of the signal distortion. Rising distortion percentage refers to the intensified changes that occur within the system. However, in order to carry out interpretation of the results in a correct manner, it is very important to define a threshold value. In specific cases, like in the present one, the threshold value was defined to be approximately 2.5%. This is because complex mixtures are used and specific time is required in order to reach the equilibrium state. When the 2.5% value is exceeded, the alarming state appears, which can be correlated to the specific degradation mechanisms. Nevertheless, when operating SOFC with pure hydrogen, the linear state is reached rapidly and no distortions are available in the system. In such case, a distortion value of 2.5% could refer to damaging mechanisms within the cell, which would require emergent counteractions in this case. Moreover, THD methodology has also been successfully employed to identify failure modes such as fuel and air starvation when operating the cells in a fuel cell mode (Suboti´c et al. 2020b; Moussaoui et al. 2022), or even to identify carbon removal from the SOFC fuel electrode and effectiveness of the employed regeneration strategies (Suboti´c et al. 2021). The preliminary investigations for the application of the THD methodology in electrolysis mode have been carried out in the Group of Energy Materials at École Polytechnique Fédérale de Lausanne in cooperation with Graz University of Technology. A representative diagram is illustrated in Fig. 8c, which clearly refers to the increasing response signal distortion, even up to approximately 80%, with the rising steam conversion rate. This could be a clear indicator to better adjust the operating conditions (current, steam flow, etc.).

6 System Modelling There is a difference between academic plant design and real systems, since the former one focuses on investigating the theoretical optimal performance via conceptual plant designs by omitting many practical constraints, and then stepping ahead to the real systems by sequentially including various practical design considerations. Thus, for given specific objectives and constraints, system or process modelling becomes a useful and cost-effective way to propose better system designs. To actually guide the practical system design, process modelling requires reliable component models derived from theoretical mathematical formulations and experiment calibration for a wide range of operating conditions. Thus, there is a clear need to closely combine component model and experiments. On the one hand, the proper design of experiments requires the recommendations provided from model predications. The available experimental data enables us to minimize model mismatch, and identify unfeasible solutions generated by models. On the other hand, system modelling with reliable component models can be used in exploring complex systems and predicting real industrial plant performance for different operating conditions, especially for large-scale systems. Therefore, system modelling plays an important role in process

398

H. Moussaoui et al.

Fig. 8 THD-distribution measured while operating a SOC in different operating modes (Barsoukov et al. 2013)

design. It can provide guidance on process components sizing and connection. In this section, some essential aspects from literature will be presented.

Stack/System Development for High-Temperature Electrolysis

399

6.1 Operating Modes Thermodynamically, the endothermic water (steam) splitting reaction requires a minimum potential for a given temperature to break the bonds of the H2 O molecule (or CO2 for co-electrolysis), as explained in detail in Chap. 2. Meanwhile, the electrochemical internal losses (overpotential) to overcome the resistance of, e.g. mass and charge transfer, take up a big share of the power consumption, which is converted to heat and eventually absorbed by the fluid and the electrolyser hardware. For given steam and air electrode feeds, and according to the imposed current and the corresponding voltage, the SOE can be operated at three different modes. These three possible operating modes, namely endothermic, thermoneutral and exothermic, are illustrated in Fig. 9 and discussed below: • Endothermic mode: When the operating voltage is relatively low, the amount of heat caused by internal losses is lower than that absorbed by the H2 O (CO2 ) splitting reaction. Therefore, additional external heat is necessary to maintain the SOE operating temperature. • Thermoneutral mode: At thermoneutral voltage, the amount of heat caused by the overpotential equals that absorbed by the H2 O (CO2 ) splitting reaction, thus the SOE operating temperature can be maintained even without external heat supply, under the assumption of zero heat losses to the surroundings. In theory, thermoneutral mode means that the temperature at the stack outlet flow is high enough to heat up the stack inlet flow, at which the temperature is lower, which enables to achieve the target temperature via heat exchangers and no additional electrical heater would be needed in this case. • Exothermic mode: With an operating voltage beyond thermoneutral voltage, the amount of heat caused by large overpotential is higher than that absorbed by the H2 O (CO2 ) splitting reaction. Thus, excess heat needs to be removed to maintain a sufficiently low thermal gradient along cells.

Fig. 9 Illustration of the endothermic, exothermic and thermoneutral modes: each graph shows the evolution of the current density and the temperature difference between under polarization (i /= 0) and under OCV (i = 0) at three different voltages around the thermoneutral voltage (1.2, 1.3 and 1.4 V). These results have been obtained by using a specific segmented cell setup at EPFL

400

H. Moussaoui et al.

The definition of endothermic/thermoneutral/exothermic operation above is based on a constant cell temperature under isothermal operation, when the tested cells or stacks are placed in a temperature-stabilized oven. However, in practical applications, the stacks are usually well insulated under adiabatic operation but not placed in a temperature-controlled oven. For adiabatic operation, if the stack operates with a large overpotential, the stack temperature will increase along the flow direction with the outlet temperature larger than the inlet temperature. This offers an additional degree of freedom for system design, particularly the design of the heat exchanger network for minimizing the electrical heating.

6.2 Power-to-X Flexibility by Means of Fuel Upgrading In order to optimize the power-to-X systems and to compare their effectiveness for different proposed solutions, available simulation studies are discussed in this section. As shown in Fig. 10, the hydrogen-rich gas mixture (23) is heated up to 240 °C (25), depending on the catalysts employed, to start the reactions. For steam electrolysis, H2 from electrolysis (12) and CO2 (19) are fed stoichiometrically into the methanator. The product (26, a gas mixture of H2 , CO, CO2 , CH4 and H2 O) is then cooled down under the process pressure (4–30 bar) with water knock-out, before entering the purification system, membrane separation for example. Here is worth mentioning that the reversible SOC system is able to provide potential CO2 as feedstock to the methanator. Carbon capture and storage (CCS) technologies can be applied downstream of the combustion chamber, which are able to retrieve most of the produced CO2 from emission intensive activities and store it permanently in nature (e.g. underground). Moreover, when biofuels are used as energy source, the SOFC system integrated with CCS can be considered as a carbon negative technology. Besides the process of generating synthetic natural gas, there are other processes to generate more complex hydrocarbon fuels, for example, dimethyl ether (DME) and gasoline. By consuming the CO2 to produce syngas and utilizing the renewable energy as the power source, the downstream of the SOE system can be used to produce green DME or gasoline which can be burned additional carbon emissions. In this case, the promising DME/gasoline synthesis process integrated with SOE technology is presented in this section. The well-known, mature direct dimethyl ether (DME) synthesis is dehydration of pure MeOH with a specific catalyst. After forming MeOH, normally, a basic twostage process of subsequent MeOH synthesis and dehydration is used for indirect DME synthesis as shown in Fig. 11. The crude MeOH (35) is mixed with recycled DME/MeOH (43) and is separated in a distillation column to obtain high-purity MeOH (37). The subsequent DME reactor performs MeOH dehydration. The mixture of DME/MeOH/H2 O (40) is further separated towards the DME product (over 98 vol%) and unreacted MeOH is recycled by (42).

Stack/System Development for High-Temperature Electrolysis

401

Fig. 10 Schematic of SOE-based power-to-methane system with oxygen as the sweep gas (Wang et al. 2020a)

Fig. 11 Schematic of DME synthesis and upgrading from methanol pathway (Wang et al. 2019b)

For gasoline, as one of the complex product recovery and upgrade processes, methanol synthesis is still considered as the intermediate process, as shown in Fig. 12. After that, the effluent (39) is first separated into water (41), crude gasoline (43) and gas streams (42). The condensed crude gasoline (43) then enters product recovery and is fractionated in the de-ethanizer, where the light hydrocarbons C1-C3 are separated at the top via (45) to the absorber and the C7-rich mixture (46) enters the stabilizer. The stabilizer further separates the LPG-rich stream (47) and stabilized gasoline (48), part of which (49) is recycled back to the absorber and the remaining (60) becomes heavy gasoline. The LPG-rich stream (47) passes through the alkylation reactor and then a column to obtain the final LPG (57) as the main by-product by separating the C4 hydrocarbons out, since too much C4 hydrocarbons, especially isobutene, can

402

H. Moussaoui et al.

Fig. 12 Schematic of gasoline synthesis and purification from methanol pathway (Wang et al. 2019b)

cause security problem and low fuel quality. The fuel quality and security status can be evaluated by measuring the Reid vapour pressure, which provides information about the volatility of gasoline and other petroleum products. Finally, the recovered light gasoline (62) is mixed with the heavy gasoline (61) as the major gasoline product (64). Conclusively, as a hydrogen-generation system working at high temperature, the combination of fuel synthesis and a SOE system will increase the overall powerto-chemical efficiency via plant-wise heat integration and the operating flexibility when dealing with fluctuating renewable power. Moreover, operating the SOE stack in reverse mode (SOFC) or integrating other fuel cell technologies, for example proton-exchange membrane fuel cell (PEMFC) or alkaline fuel cell (AFC), could optimize energy utilization with the aid of storage technologies. Furthermore, the SOE system itself should be flexible to adapt to the fluctuations of the excess power source. Such complex systems can be applied for seasonal energy storage, largescale fuel production, etc. By balancing the power and fuel production using a rSOC system, the highest overall efficiency can be achieved considering different system aspects. To achieve this goal, a multi-objective optimization approach can be used to generate a decision matrix.

6.3 Multi-objective Optimization (MOO Decision Parameters) After screening and selecting the process, the technology will be assisted by the multi-objective optimization (MOO) tool OSMOSE (Wang et al. 2019a), in order to optimize the heat integration and thereby the system efficiency (or the system revenue). The goal of the OSMOSE platform is to enable the combination of flowsheeting tools, process integration, costing tools and life-cycle assessment with multiobjective optimization techniques.

Stack/System Development for High-Temperature Electrolysis

403

OSMOSE is a decision support tool that includes problem definition and model, integration and optimization, visualization and reporting, and it is used to generate a list of solutions for the decision maker. The optimal solutions provided by OSMOSE may have minimum resource and utility consumptions. It is a computer platform, that bridges process models and data flows from different software and tools to aid decision-making in the design and operation of the process and energy systems. It is built upon the knowledge acquired from different processes/projects, heat integration and multi-objective optimization methods and results visualization tools. The OSMOSE platform has been interlinked with Aspen Plus and other flow-sheeting software, where data can be transferred from OSMOSE to Aspen Plus, and vice versa. In order to perform multi-objective optimization, users can develop process models (in Aspen Plus), cost functions, and environmental impact models, and include them into the OSMOSE platform. OSMOSE will provide trade-off solutions for conflicting objectives. Figure 13 presents the common workflow in performing MOO. In SOE systems, with specific values of decision variables related to the process design such as operating pressure, steam flow rate, recirculation ratio and heating temperature, the process models are developed as the first step, to obtain the mass and energy balances of the whole plant. Then, heat cascade utilization is formulated in a mixed integer linear programming problem, which is calculated to select and size hot and cold utilities. With the sized utilities, the minimum number, exchange surface area and cost of the heat exchangers are estimated by vertical heat transfer. Moreover, the values of the objective functions are then evaluated with the investment of major equipment and main economic assumptions from literature studies. Finally, multiobjective optimization (MOO) is carried out using evolutionary algorithms to vary the decision variables and to evaluate the process designs with the objective values. Objective values can also be called as key performance indicators (KPIs) (e.g. efficiency, net hydrogen production). The steps are repeated until a cluster of highquality Pareto-optimal solutions are found, which reveal potential trade-offs between different objective functions.

7 Concluding Remarks Following the goals of green energy and sustainable growth, it is very important to solve the issues of storing excess electricity generated by renewable energy. Since it is not possible to directly store large quantities of electricity, other than pumped hydro storage, appropriate green solutions have to be provided here. The most appealing solution is related to hydrogen generation and its subsequent storage or, better, processing. The most efficient hydrogen generation is proposed by employing solid oxide cell technology. However, as discussed in this chapter, hydrogen storage is not an easy task. With this in mind, when employing the SOC technology simpler solutions are possible. These are related to different feedstock and direct electrochemical utilization of CO2 to generate syngas. Applying additional processes, the generated syngas can be converted into methane and liquid fuels, the storage of which is thus

404

H. Moussaoui et al.

Fig. 13 The employed multi-objective optimization framework. (Zhang et al. 2020)

significantly simplified and the calorific value is increased, in parallel. This seems to be one of the most important approaches for future application. Moreover, ammonia production from renewable hydrogen presents a further possibility. However, in order to find the most appropriate solution and to be able to provide an answer on the question “which X for Power-to-X”, it is critically necessary to take into consideration regional resources and the final user, based on which the system boundaries can be optimized. For this purpose, both numerical and experimental approaches should be combined.

References Aicart J et al (2022) Performance evaluation of a 4-stack solid oxide module in electrolysis mode. Int J Hydrogen Energy 47(6):3568–3579

Stack/System Development for High-Temperature Electrolysis

405

Arslan O, Acikkalp E, Genc G (2022) A multi-generation system for hydrogen production through the high-temperature solid oxide electrolyzer integrated to 150 MW coal-fired steam boiler. Fuel 315:123201 Barsoukov E, Macdonald JR (2013) In: Impedance spectroscopy theory, experiment, and applications. Wiley Inc Blanco H, Faaij A (2018) A review at the role of storage in energy systems with a focus on power to gas and long-term storage. Renew Sustain Energy Rev 81:1049–1086 Brisse A, Schefold J, Zahid M (2008) High temperature water electrolysis in solid oxide cells. Int J Hydrogen Energy 33(20):5375–5382 Dutta S (1990) Technology assessment of advanced electrolytic hydrogen production. Int J Hydrogen Energy 15(6):379–386 Garðarsson JS, Vang P, Højgaard S, Nielsen R (2016) Predicting the price of solid oxide electrolysers (SOECs). DTU Report Hauch A, Küngas R, Blennow P, Hansen AB, Hansen JB, Mathiesen BV, Mogensen MB (2020) Recent advances in solid oxide cell technology for electrolysis. Science 370(6513):eaba6118 Jeanmonod G, Diethelm S, Van herle J (2021) Poisoning effects of chlorine on a solid oxide cell operated in co-electrolysis. J Power Sour 506:230247 Kendall K, Minh NQ, Singhal SC (2003) Cell and stack designs. In: High-temp. solid oxide fuel cells fundamental designs and applications. Elsevier, pp 197–228 Königshofer B et al (2021) Performance assessment and evaluation of SOC stacks designed for application in a reversible operated 150 kW rSOC power plant. Appl Energy 283:116372 Küngas R et al (2017) eCOs—a commercial CO2 electrolysis system developed by Haldor Topsoe. ECS Trans 78(1):2879–2884 Larsen PH et al (2001) Status of the Danish SOFC program. ECS Proc 2001–16(1):28 Lehtinen T, Noponen M (2021) Solid oxide electrolyser demonstrator development at Elcogen. ECS Trans 103:1939 Léon A, Micero A, Ludwig B, Brisse A (2021) Effect of scaling-up on the performance and degradation of long-term operated electrolyte supported solid oxide cell, stack and module in electrolysis mode. J Power Sour 510:230346 Liten (2022) Liten—An experimental platform to test rSOC modules up to 120kWDC set up at CEA. https://liten.cea.fr/cea-tech/liten/english/Pages/Medias/News/Hydrogen-Vector/Anexperimental-platform-to-test-rSOC-modules.aspx. Accessed 25 Jul 2022 MegaSyn (2022) Home—MegaSyn English (2022). https://www.megasyn.eu/. Accessed 25 Jul 2022 Mehariya S et al. (2019) Fischer-Tropsch synthesis of syngas to liquid hydrocarbons. Chapter 7 in Lignocellulosic biomass to liquid biofuels, pp 217–248 Mougin J (2015) Hydrogen production by high-temperature steam electrolysis. In: Subramani V, Basile A, Veziroglu TN (eds) Compendium of hydrogen energy, Woodhead Publishing, pp 225–253 Moussaoui H, Laurencin J, Gavet Y, Delette G, Hubert M, Cloetens P, Le Bihan T, Debayle J (2018) Stochastic geometrical modeling of solid oxide cells electrodes validated on 3D reconstructions. Comput Mater Sci 143:262–276 Moussaoui H, Sharma RK, Debayle J, Gavet Y, Delette G, Laurencin J (2019) Microstructural correlations for specific surface area and triple phase boundary length for composite electrodes of solid oxide cells. J Power Sour 412:736–748 Moussaoui H, Debayle J, Gavet Y, Cloetens P, Laurencin J (2020) Particle-based model for functional and diffusion layers of solid oxide cells electrodes. Powder Technol 367:67–81 Moussaoui H, Hammerschmid G, Van herle J, Suboti´c V (2022) Fast online diagnosis for solid oxide fuel cells: optimisation of total harmonic distortion tool for real-system application and reactants starvation identification. J Power Sour 556:232352 Preininger M, Stoeckl B, Suboti´c V, Hochenauer C (2020) Characterization and performance study of commercially available solid oxide cell stacks for an autonomous system. Energy Convers Managem 203:112215

406

H. Moussaoui et al.

Reddy MJ, Svensson JE, Froitzheim J (2021) Evaluating candidate materials for balance of plant components in SOFC: oxidation and Cr evaporation properties. Corros Sci 190:109671 Sampathkumar SN et al (2022) Degradation study of a reversible solid oxide cell (rSOC) short stack using distribution of relaxation times (DRT) analysis. Int J Hydrogen Energy 47(18):10175– 10193 Samsatli S, Samsatli NJ (2019) The role of renewable hydrogen and inter-seasonal storage in decarbonising heat—comprehensive optimisation of future renewable energy value chains. Appl Energy 233–234:854–893 Schiebahn S, Grube T, Robinius M, Tietze V, Kumar B, Stolten D (2015) Power to gas: technological overview, systems analysis and economic assessment for a case study in Germany. Int J Hydrogen Energy 40(12):4285–4294 SOLIDPower (2022) Successfully tested SOLIDpower large stack module for high efficiency hydrogen generation. SOLIDpower (2022). https://www.solidpower.com/en/news/allnews/details/?tx_news_pi1%5Bnews%5D=722&tx_news_pi1%5Bcontroller%5D=News&tx_ news_pi1%5Baction%5D=detail&cHash=9ba4e28689e8522d308d5c7e2828bc7a. Accessed 25 Jul 2022 Suboti´c V et al (2016) In-situ electrochemical characterization methods for industrial-sized planar solid oxide fuel cells Part I: methodology, qualification and detection of carbon deposition. Electrochim Acta 207:224–236 Suboti´c V et al (2020a) Performance assessment of industrial-sized solid oxide cells operated in a reversible mode: detailed numerical and experimental study. Int J Hydrogen Energy 45(53):29166–29185 Suboti´c V et al (2020b) On the origin of degradation in fuel cells and its fast identification by applying unconventional online-monitoring tools. Appl Energy 277:115603 Suboti´c V, Harter P, Kusnezoff M, Napporn TW, Schroettner H, Hochenauer C (2021) Identification of carbon deposition and its removal in solid oxide fuel cells by applying a non-conventional diagnostic tool. Sustain Energy Fuels 5(7):2065–2076 Suboti´c V, Hochenauer C (2022) Analysis of solid oxide fuel and electrolysis cells operated in a real-system environment: state-of-the-health diagnostic, failure modes, degradation mitigation and performance regeneration. Progress in Energy Combustion Sci 93:101011 Suboti´c V, Königshofer B, Höber M, Hochenauer C, Koroschetz M, Hochfellner M, Brabandt J, Baumgartner J (2022) Comprehensive electrochemical analysis of a 150 kW reversible SOC system installed at a power plant Mellach in Austria. In: EFCF conference Sunfire (2022) Sunfire-Hylink SOEC core advantages Thema M, Bauer F, Sterner M (2019) Power-to-gas: electrolysis and methanation status review. Renew Sustain Energy Rev 112:775–787 Topsoe (2022) SOEC high-temperature electrolysis|Equipment|Products|Topsoe (2022). https:// www.topsoe.com/our-resources/knowledge/our-products/equipment/soec. Accessed 26 Jul 2022 Verbund (2022)Grüner Wasserstoff von VERBUND für die Energiewende (2022). https://www.ver bund.com/de-at/geschaeftskunden/industrie/gruener-wasserstoff. Accessed 12 Jul 2022 VoltaChem (2022) Towards industrial-scale solid oxide electrolyzers: TNO achieves increase in cell size by up to 5 times. VoltaChem (2022). https://www.voltachem.com/news/towards-industrialscale-solid-oxide-electrolyzers-tno-achieves-increase-in-cell-size-by-up-to-5-times. Accessed 24 Aug 2022 Wang L et al (2019a) Power-to-methane via co-electrolysis of H2 O and CO2 : The effects of pressurized operation and internal methanation. Appl Energy 250:1432–1445 Wang L et al (2019b) Power-to-fuels via solid-oxide electrolyzer: operating window and technoeconomics. Renew Sustain Energy Rev 110:174–187 Wen D, Aziz M (2022) Techno-economic analyses of power-to-ammonia-to-power and biomassto-ammonia-to-power pathways for carbon neutrality scenario. Appl Energy 319:119272 Yang S-I, Choi D-Y, Jang S-C, Kim S-H, Choi D-K (2008) Hydrogen separation by multi-bed pressure swing adsorption of synthesis gas. Adsorption 14(4–5):583–590

Stack/System Development for High-Temperature Electrolysis

407

Zhang X, O’Brien JE, O’Brien RC, Hartvigsen JJ, Tao G, Housley GK (2013) Improved durability of SOEC stacks for high temperature electrolysis. Int J Hydrogen Energy 38(1):20–28 Zhang H, Wang L, Pérez-Fortes M, Van herle J, Maréchal F, Desideri U (2020) Techno-economic optimization of biomass-to-methanol with solid-oxide electrolyzer. Appl Energy 258:114071

Concluding Remarks Miguel A. Laguna-Bercero

This book has tried to expose the status of high-temperature electrolysis. We have covered fundamental aspects, summarized the different materials employed in the devices, showed their performances and covered degradation issues and cell, stack and system scale. Important aspects regarding modelling or alternative novel designs have been also discussed. As for the SOEC electrolyte material, although many oxide formulations have been extensively studied, and a significant amount of effort has been devoted to understanding the ionic conducting mechanism, the most used material is still the traditional YSZ due to its low electronic conductivity and acceptable oxide ion conductivity above 800 °C. As discussed, there is still an ongoing search for new materials; however, the number of potential candidates will remain limited, making the development of existing materials essential. If the cost can be sacrificed, scandia-doped zirconia may be preferred at temperatures of about 650–700 °C. There have also been developments and proposals for alternative electrolyte materials, such as perovskite (La,Sr)(Ga, Mg)O3 oxides. However, they still have their own problems, such as the high chemical reactivity with other components of the SOECs and instability under reducing conditions. Regarding the recent advances in oxygen electrode materials for SOECs, most of the studies are performed on perovskite-type materials such as LSM or LSCF. In addition, several double perovskites and Ruddlesden-Popper phases have been also investigated. In general, the perovskite-type material cells show degradation in a few hours, mainly due to mass transport, compositional change of electrodes and delamination. Therefore, long-term operation is one of the challenging issues. Using existing oxygen electrode materials, further progress on the chemical stability can be made by applying dense GDC, LDC or SDC or PrDC (La, Sm or Pr-doped M. A. Laguna-Bercero (B) Instituto de Nanociencia y Materiales de Aragón (INMA), CSIC- Universidad de Zaragoza, C/ María de Luna 3, 50018 Zaragoza, Spain e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M. A. Laguna-Bercero (ed.), High Temperature Electrolysis, Lecture Notes in Energy 95, https://doi.org/10.1007/978-3-031-22508-6_13

409

410

M. A. Laguna-Bercero

ceria) barrier layers between electrode and electrolyte, to inhibit the formation of detrimental phases. Delamination of electrodes mainly occurs due to high oxygen partial pressure at TPB sites and in the bulk of oxygen electrodes. If one succeeds to remove such excess oxygen from the oxygen electrode, then it is possible to avoid delamination. One efficient way could be reversible operation between electrolysis and fuel cell modes. In addition, infiltration is a promising method to form highperforming oxygen electrodes for SOECs. Ionic and electronic conductivity as well as the catalytic activity of the electrode significantly improve due to the presence of nanoparticles of high surface area. By using this methodology, unfavourable phases which form due to the reaction of electrolyte and electrodes at high temperatures are avoided, giving the flexibility of more materials choice. However, the stability issue due to particle growth during sintering needs to be addressed. One effective method can be co-infiltration of multiple components where the minor phase covers the major phase and inhibits its growth. Another effective method to improve the stability of the infiltrated cells is to heat treat them at temperatures higher than the operating temperature while still maintaining the nano-structure of the infiltrates. It is believed that future efforts by the fuel cell community will focus on developing infiltrated oxygen electrodes, particularly double perovskites or Ruddlesden-Popper phases with high performance and stability at lower temperatures, which have not been studied in detail in order to eliminate the electrochemical cell degradation. Regarding the fuel electrode, the state-of-the-art material is the Ni-YSZ. It predominately degrades through Ni relocation and impurity segregation. Under more extreme electrolysis operating conditions, it can produce ZrO2 nanoparticles, oxidize to NiO or be reduced at the Ni-YSZ interface. CO2 electrolysis and co-electrolysis presents more challenges as carbon can form and quickly degrade the performance. Recently, some of these key issues have been solved through the engineering of this fuel electrode. In addition, other alternatives include Ni-free metal-based cermet fuel electrodes and perovskite-type oxide fuel electrodes. Furthermore, from a characterization and modelling perspective it would be interesting to expand both DFT and in-situ work on the fuel electrode to open up new novel materials with reduced overpotential losses and improved durability. Concerning metallic interconnects and their relative corrosion issues, it seems well-established that the modification of the composition of the base Mn-Co spinel is a promising route to improve the functional properties of the base spinel structure. In addition, a high degree of densification is fundamental to guarantee reasonable protection properties of the coating, in order to limit oxygen inward diffusion to the steel interconnect. Furthermore, a possible approach to a composite spinel coating coupling the benefits of rare earth addition with those of Mn-Co spinel should be explored. The preliminary findings on the role of the porous metal alloys support show that the MS-SOEC design has promised, but further work is required to properly evaluate their feasibility. Continued development of MS-SOEC will be beneficial if performance, durability or cost can be enhanced beyond that of other SOEC types. The best candidate for SOC applications seems to be chromia-forming ferritic alloys. The porous forms of the alloys are also promising; however, corrosion issues need to be solved. It is found that Ni, Cr and Al are the most corrosion protective elements.

Concluding Remarks

411

To prevent chromium evaporation and poisoning of the electrodes, the functional coatings, for instance Mn-Co spinel, are applied to the steel components. For a better understanding of high-temperature corrosion effects in metal supports, both kinetic and further structural investigations are necessary. A complex field of research arises with regard to the glass–ceramic sealing materials. Traditional sealing materials include metallic brazing alloys and such based on mica compression seals, but currently they play a marginal role and are only used in individual cases. Glass–ceramic seals are predominantly used such as earth alkaline silicate-based glass systems (RO-SiO2 -type). In addition, multicomponent material composites where a glass phase can be both the matrix phase and the filler are also proposed. In any case, the development of glass–ceramic sealants for SOC can only succeed if the intrinsic material development is accompanied by technologically relevant characterization methods. Multi-scale simulation tools play a very important role in the development of SOEC technology, where the macroscopic models are the basis for engineering applications of SOEC. The simulation process is then a fundamental aspect in order to design highly efficient electrolysers, covering from the macroscopic cell to the electrode level. For the macroscopic modelling of SOEC, the electrochemistry and multiphysics transport should be fully considered, including the multicomponent mass transfer process in porous electrodes, the global and fundamental kinetics of internal reforming and the electrochemical reactions. Meanwhile, the modelling process at the electrode scale is also a key topic which is beneficial to improve the cell performance and durability from a more microscopic perspective. Through a series of modelling approaches, the full life-cycling behaviours of SOEC electrodes can be simulated and thus SOEC performance can be improved from all stages. In addition, the machine learning algorithms are introduced and can be combined with different models as a complement for simulation optimization tools. For the macroscopic model, the large-scale multi-physics field is also needed at the stack level (stack flow field, distribution and geometry configuration), which will contribute to improve the overall performance and efficiency of the stack. Especially for thermal management, the non-isothermal multi-physics model provides an intuitive distribution of velocity and temperature fields, which in turn can be combined with structural mechanics and other models to optimize the stack structure and achieve improved long-term thermal stability. In addition, further dynamic modelling of the SOEC is required to improve the dynamic response characteristics and thermal cycling stability of the cell thermal and electrochemical performance. Furthermore, the analysis of the internal transport mechanism of mixed conductors will be very useful to improve the Faraday efficiency of proton-conducting solid oxide electrolysers. Regarding the microscopic electrode level, the decoupling model combining multiple simulation methods will be one of the research directions. The combination of multiple simulation approaches allows for the optimal design of electrode microstructures from the full electrode life cycle from electrode preparation to long-term electrode operation and to improve the design of different electrode materials and structures such as mixed-conducting and nanoparticle-infiltrated electrodes. Modelling studies of relevant degradation

412

M. A. Laguna-Bercero

mechanisms are still nearly scarce, and they will be very relevant to achieve longterm stability. Atomic-molecular scale simulations such as density functional theory and molecular dynamics can be used for solving such problems. It is foreseeable that these methods will have great potential for microscopic simulations. Finally, the artificial intelligence methods such as machine learning provide more options for SOEC simulation research, which in turn can be combined with optimization methods such as genetic algorithms to achieve improvements in SOEC performance at multiple scales and can provide real-time control optimization solutions for SOEC systems. Concerning solid oxide electrolysers based on proton-conducting ceramic electrolytes, they are emerging as a promising technology with distinctive operation characteristics. Their advantages include the operation at temperatures below 550 °C, the direct delivery of dry hydrogen and the limit risks of nickel electrode re-oxidation. The lower operation temperature enables the integration with different plants that have already pressurized steam in this temperature range, leading to higher plant energy efficiency, as for instance, in the potential integration with ammonia or methanol synthesis processes. In addition, the low temperature will lead to lower costs of housing and auxiliary materials, heat-exchange requirements and potentially lower degradation rates. In any case, several challenges are being tacked to push the maturity and readiness level and, then, making further steps towards its industrial scale-up and deployment. Long-term stability is still unknown since the few PCelectrolysers built at relevant scale (>100 cm2 active electrode) are still relative recent and very scarce stability data is available. Another important aspect is related to the materials composition, especially the need of using barium and some rare-earths’ metals such as Ce or Yb. These special ceramics require notably high manufacturing temperatures above 1500 °C, leading to potential barium loss by evaporation that, in turn, gives rise to loss in proton-conduction properties and interface degradation. In addition, other applications are being explored including co-electrolysis, nonoxidative hydrocarbon conversion, steam reforming of hydrocarbons or intensified ammonia cracking, employing both planar and tubular geometries in different stack arrangements. Durability is a crucial aspect in order to ensure a rapid implementation on the technology in the market. As for fuel electrodes, Ni oxidation, agglomeration and microcrack in Ni-containing hydrogen electrode can dramatically occur at elevated temperatures, under high steam partial pressure and redox cycles. Moreover, impurities in feed gas such as CO2 and H2 S can lead to carbon deposition and sulfur poisoning on the Ni surface, respectively. Nickel cermet is still the most used material, although its optimization is still required in order to present a minimal impact on these issues. Optimized electrode structure, controlled operating conditions and gas-flow configuration will help to mitigate its degradation. In this sense, Ni alloys and perovskite-based hydrogen electrodes have gain increasingly attention. However, thermal expansion behaviour and electrochemical performance of the Ni-alloy electrode remains as a challenge, while chemical instability due to cation segregation is still the main issue in perovskite-based hydrogen electrode. Regarding degradation

Concluding Remarks

413

in the oxygen electrode, various factors including microstructure, TEC, phase segregation, impurities in feed gas and pO2 must be considered. The delamination of an O2 -electrode is another major problem in SOEC operation. Oxygen built-up at the TPB leads to the formation of local tensile strains, microcracks and eventually electrode delamination. Moreover, high pO2 can be a driving force for segregation of Sr in Sr-containing electrode. Therefore, an SOEC oxygen electrode is required to be able to accommodate a large oxygen over-stoichiometry under oxidizing conditions. Electrolytes in SOEC are required to be sufficiently dense and exhibit high ionic conductivity and durability. However, phase transformation and partially reduction occurs in electrolyte during high-temperature densification process and under prolong operation. Therefore, dopants are applied. They can enhance phase stability and introduce oxygen vacancy in the structure leading to an increase in oxygen mobility. As previously mentioned, other important aspects to be considered regarding the durability of SOEC cells are those related with interdiffusion between electrode and electrolyte leading to secondary phase or impurities, as well as interconnect degradation due to chromium poisoning of metallic alloys. Furthermore, the recent development of new approaches and techniques that allow ceramics structuring at different length scales (from the atomic to the cm-scale) is considered a major opportunity to enhance SOC technologies in the short- to midterm future. In this regard, 3D structuring is the most promising candidate to define new architectures that will increase the power density of cells and stacks while thinfilm-based intermediate layers and ex-solution of nanocatalysts have already proved to be very efficient in solving currently existing limitations of SOCs. Although several pioneering works have shown a successful application of novel methodologies for fabricating multiscale structured cells, further developments are still required. Among other unsolved challenges that will foster the implementation of such methodologies, one can find the multi-material 3D printing of complete devices, the full understanding of the stability of thin films under operation conditions, the ex-solution of nanoparticles at the oxygen electrode or the impact of the use of HEOs on the long-term stability of SOECs. Facing these and other existing challenges will drive to relevant breakthroughs in the field of SOCs increasing their performance and durability while reducing the use of critical raw materials. It is then clear that decarbonization of the energy system and increased utilization of renewable energy sources require new solutions to bridge the intermittence of the renewables and differences that occur as a consequence of varying supply and demand time points. SOECs present a very attractive solution for the mentioned issues. Based on electrochemical processes, they enable to directly produce valuable fuels when utilizing electricity and input gaseous species, e.g. steam and/or carbon dioxide. Besides hydrogen, syngas can be produced in a highly efficient manner. Applying additional processes, the generated syngas can be converted into methane and liquid fuels, the storage of which is thus significantly simplified and the calorific value is increased, in parallel. This seems to be one of the most important approaches for future application. Moreover, ammonia production from renewable hydrogen presents a further possibility. However, in order to find the most appropriate solution and to be able to provide an answer on the question “which X for Power-to-X”, it is

414

M. A. Laguna-Bercero

critically necessary to take into consideration regional resources and the final user, based on which the system boundaries can be optimized. To summarize, SOE is significantly advantageous compared with alkaline or proton-exchange membrane electrolysis due to: (1) higher system efficiency, (2) heat management opportunity with other industrial processes, (3) possibility of performing CO2 electrolysis to produce CO, (4) direct syngas production is also possible by the co-electrolysis of steam and CO2 and (5) its reversible system (SOFC) can be used to balance energy demand. These advantages have been recently demonstrated at Gas Innovations (La Porte in Texas, USA) in 2016, producing 3–5 Nm3 per hour of CO with a purity grade of 99.95%. In addition, Haldor Topsoe reported that the efficiency of their stacks is higher by 30% in comparison with the efficiency of commercial alkaline or PEM electrolysers. Furthermore, they have confirmed the use of SOEC systems for green ammonia production, in a process known as power-to-ammonia (PtA). Hydrogen, thanks to high-temperature electrolysis, seems to have finally come to stay in our daily lives. We hope that this book will help contribute to a rapid implementation of the technology.