Multi-Fuel oxidation in Solid Oxide Fuel Cells: Model anodes and system studies

(1)

Multi ‐Fuel oxidation in Solid Oxide

Fuel Cells : Model anodes and system

studies

(2)

(3)

iii

Multi -Fuel oxidation in Solid Oxide Fuel Cells : model anodes and system studies

Proefschrift

ter verkrijging van de graad van doctor aan de Technische Universiteit Delft,

op gezag van de Rector Magnificus prof. Ir. K.Ch.A.M. Luyben; voorzitter van het College voor Promoties,

in het openbaar te verdedigen op Maandag 28 September om 15:00 uur

Door

Hrishikesh PATEL

MSc Sustainable Energy Technology geboren te Pune, India

(4)

iv

This dissertation has been approved by the

promotor : Prof. Dr. Ir. B J Boersma copromotor: Dr. P V Aravind

Composition of the doctoral committee:

Rector Magnificus

Prof. Dr. Ir. B J Boersma promotor Dr. P V Aravind

Prof. dr. ir. JHA Kiel

Independent members:

Prof. dr. F M Mulder, Delft University of Technology Prof. dr. D J E M Roekaerts, Delft University of Technology Prof. dr. ir. J E ten Elshof, University of Twente

Dr. J Kiviaho, VTT Finland

Reserve Member

Prof. dr. ir. TJH Vlugt, Delft University of Technology

The author is grateful to ADEM, A green Deal in Energy Materials of the Ministry of Economic Affairs of The Netherlands (www.adem-innovationlab.nl) for their financial support.

(5)

v

Summary

With the evolution of renewable energy technologies it has become necessary that a balance is found between power production with conventional energy sources and other long term solutions. SOFCs offer an alternative for utilising conventional fossil fuels as well as sustainable biomass derived fuels at a high efficiency. For use with such fuels, materials like ceria can offer higher performance and improved carbon tolerance as compared to conventional Ni-YSZ anodes. The mixed ionic electronic conductivity of ceria helps to spread the electrochemically active area beyond the interface between nickel and ceramic which is expected to improve performance. While it is known that H2 and CO oxidation can occur on ceria, there are very few

studies and no consensus about the exact mechanism or the rate limiting steps. This is mainly because of the 3D interpenetrating microstructure of the anode which makes reactions difficult to localise and identify individually.

Pattern anodes can be made with well-defined geometry which helps to localise reactions and study elementary reactions with no interference from phenomena such as diffusion. Deriving kinetics and rate parameters from this data can be used to extend knowledge gained into other studies. This thesis is aimed at studying fundamental electrochemistry of H2 and CO oxidation

using pattern anodes of nickel and ceria and extend this knowledge to macroscopic simulations and system studies. The thesis comprises of four main parts 1) Preparation and testing of Ni and Ce pattern cells with H2 and CO 2) Elementary kinetic modeling of oxidation on ceria 3) Effect

of H₂S on Ni and Ce 4) SOFC-Gas Turbine systems with different biofuels Preparation and testing of Ni and Ce pattern cells with H2 and CO

Fuel oxidation on pattern anodes is studied using individual materials of the cermet anode ie nickel and ceria. Sputtering through steel masks gives a cheaper method of making pattern cells as compared to lithographic methods while still obtaining reasonable results. Hydrogen and CO oxidation on both the materials is studied in particular and the work is extended to hydrogen CO mixtures. Electrochemical Impedance Spectroscopy (EIS) was carried out at different temperatures between 700 to 850 o_{C and equivalent circuit fitting of the obtained spectra is}

carried out to quantify results. It is found that ceria can be a better electrocatalyst than nickel for hydrogen as well as CO oxidation.

CO oxidation in dry as well as wet atmospheres shows a higher polarisation resistance compared to hydrogen on both nickel and ceria pattern cells. While rate limiting mechanism could not be identified accurately, from the activation energies a spillover mechanism for hydrogen is most likely while CO spillover is not considered likely. For wet CO mixtures, addition of water brings about a large drop in the polarisation resistance for both nickel and ceria. Preferential hydrogen oxidation can be inferred from the much lower polarisation resistance of H2/CO mixtures.

Elementary kinetic modeling of oxidation on ceria

A state space model where the charge transfer and adsorption reactions are used to describe electrochemical oxidation of hydrogen and carbon monoxide on ceria has been developed. The model takes into account thermodynamic consistency and microscopic reversibility of the reactions. The rate constants for charge transfer reaction are calculated to match the current produced in experimental data obtained previously. A sensitivity analysis for hydrogen oxidation

(6)

vi shows that the adsorption reaction is not limiting. The limiting charge transfer reaction is found to be between the adsorbed hydrogen and oxygen from ceria. In case of CO oxidation the sensitivity analysis shows that the charge transfer and not adsorption of CO is limiting.

An approach is presented to use the kinetic data obtained on geometrically well-defined electrodes in macroscopic studies. The exchange current density is estimated which is used to calculate the activation overpotential. This can be used in macroscopic simulations for a more accurate description of fuel oxidation kinetics. For example, in a CFD model, the activation overpotential can be calculated from actual species concentration at the interface rather than using semi empirical approximations.

Effect of H2S on Ni and Ce

The effect of H₂S on the performance of model anodes of nickel and ceria is tested for 5 and 20 ppm levels of H2S. While the polarization resistance of both nickel and ceria patterns increases

significantly after exposure to H2S, the change is completely reversible only in case of nickel after

removal of H2S. The poisoning kinetics of nickel and ceria pattern cells is extremely fast, however

only the nickel pattern cells recover quickly while the ceria pattern cells take a very long time to stabilize.

A model with elementary kinetics for H2S adsorption on ceria pattern cells is developed assuming

the adsorbed species do not take part in the charge transfer process. It is observed that the surface coverage of the S species is very high and impedes the H adsorption on ceria. While this is not a comprehensive model, it is representative of the effect of H2S on polarization resistance

and the rate constants obtained here can be used in extending the effect of H2S on pattern

electrodes to working cells by incorporating into other simulations as described previously.

System Studies

Solid Oxide Fuel Cell – Gas Turbine (SOFC-GT) systems provide a thermodynamically high efficiency alternative for power generation from biofuels. In this study, SOFC GT systems are evaluated for their performance with biofuels namely methane, ethanol, methanol, hydrogen and ammonia using the exergetic criteria. Performance at system level and in system components like heat exchangers, fuel cell, gas turbine, combustor, compressor and the stack is studied. It is found that the heat effects play a major role. This also affects the power levels attained in individual components. The per pass fuel utilisation dictates the efficiency of the fuel cell itself, but the system efficiency is not entirely dependent on fuel cell efficiency alone, but depends on the split between the fuel cell and gas turbine powers which in turn depends highly on the nature of the fuel and its chemistry. Counter intuitively it is found that with recycle, the fuel cell efficiency using methane is less than that of hydrogen but the system efficiency using methane is higher.

In summary, the following observations are made from this study. 1) Pattern anodes of nickel and ceria can give important indications about oxidation of H2 and CO and H2-CO

mixtures. Preferential H₂ oxidation on ceria as well as nickel was observed. 2) Elementary

modelling of H2 and CO oxidation on ceria is carried out using fitting of impedance spectra. This

model can be used with macroscopic simulations which require kinetic data 3) H2S has a

deleterious effect on ceria as well as nickel. Elementary modelling can capture the poisoning effect of H2S on ceria 4) System studies indicate that high efficiency SOFC – GT systems need to

(7)

vii be adapted for different fuels after a thorough understanding of exergy losses associated with subsystems – particularly the fuel cell subsystem.

(8)

(9)

1.1 Background

We are on the verge of a significant change in the energy scenario in which a paradigm shift to alternative energy resources is at hand. This has opened up research and development in renewable and sustainable energy sources. The great debate about global warming and hydrogen economy not-withstanding, the incentive of high efficiency and lower emissions (not just CO₂) by itself makes a strong argument in favour of some of the energy technologies. Irrespective of whether there is a drastic change in the way we produce energy in the coming 20 odd years, there is very little doubt that today there is a need to produce energy efficiently and sustainably. Energy from renewable sources like wind, solar etc has already found its way into everyday life in many countries. However, the share of renewable sources is not likely to be close to 20-30% by 2020 considering the present status [1]. Other problems with renewables like intermittency, availability also affect their choice for power generation as a single technology. One may expect that there is a gradual shift in the technologies and while renewables gain more share it is unlikely that fossil fuel based energy are entirely eschewed. It is most likely some intermediate technologies are developed which can provide a cleaner and more efficient means of producing power from renewable as well as conventional sources of energy. A fuel cell is such a device which can convert chemical energy to electrical energy electrochemically. In particular a Solid Oxide Fuel Cell (SOFC) can utilize most of available fuels like natural gas, biogas and even coal and biomass after gasification at very high efficiencies. The following sections describe the SOFC and its microstructure and kinetic aspects, material aspects which leads to fuel oxidation mechanisms and further the gaps this study aims at addressing.

1.2 Solid Oxide Fuel Cells (SOFCs)

Figure 1‐1: Schematic of the Solid Oxide Fuel Cell ("Solid oxide fuel cell" by Sakurambo ‐ Own work, based on w:Image:Fcell_diagram_sofc.gif (public domain). Licensed under Public Domain via Wikimedia Commons )

Figure 1-1 shows a schematic of the SOFC operating principle. The driving force is the difference in oxygen partial pressure between the cathode and anode. Under this potential the oxygen ions migrate to the anode where they oxidise the fuel and release electrons. The electrons

(16)

5 travel the external circuit to produce current. The overall reactions for CO and hydrogen oxidation in an SOFC can be written as

Cathode:

2 → (1)

Anode:

→ 2 (2)

→ 2 (3)

Solid Oxide Fuel Cells operate around 700-1000 0_{C. This very high operating temperature can be}

a disadvantage for materials selection but renders a few distinct advantages. The high operating temperature means that there is a high grade waste heat source available after the SOFC. Thus the SOFC can be combined with a bottoming cycle like a gas or steam turbine cycle to utilise the waste heat and increase efficiency [2, 3]. Also, hydrocarbon fuels can be internally reformed and utilised in the SOFC. Since the catalyst is usually nickel, carbon monoxide is a fuel and not a poison (unlike a PEM) and can be oxidised [2]. Fuels such as methane, ammonia, methanol ethanol synthesis gas etc can be directly oxidised in the SOFC at a very high efficiency [2].

1.2.1 The SOFC anode

Since all the fuel oxidation reactions happen at the anode, it is important to study the anode, its chemistry and interaction with different fuel components. The anode is a porous cermet (ceramic and metal). Many properties such as high catalytic activity, stability (with respect to temperature and reducing conditions), high electronic and ionic conductivity, low cost etc are desirable. The state of art material nowadays is considered to be nickel – yttria stabilised zirconia ( Ni-YSZ) [4, 5]. Nickel becomes a natural choice for the anode catalyst because of its high catalytic activity, stability at high temperatures and reducing atmospheres and low(er) cost. However, it is susceptible to sulphur poisoning, carbon deposition and oxidation under oxidising environments. Many alternatives to Ni-YSZ anodes are being studied. Most of these are related to reducing the operating temperature but another important aspect is to improve tolerance towards sulphur and carbon deposition. This primarily done by replacing either the catalyst or the ionic phase. Typical replacements like cobalt/ ruthenium or copper have been suggested [6-8] while replacements like ceria have been suggested for YSZ. Other alternatives like perovskite materials, chromites or titanates have been discussed [4, 9]. Ceria has been a promising choice for the anode because of its higher tolerance to carbon deposition [10]. This is usually related to the mixed ionic electronic conductivity it exhibits under reducing atmospheres.

1.2.2 Microstructure of the SOFC anode.

The SOFC anode is a interpenetrating network of gas, ionic and electronic phases which should be percolating. The triple phase boundary (TPB) is the place where the ionic, electronic and gas phases meet. This is usually where the reactions are limited to and is referred to as the electrochemically active area. The TPB length is an important parameter in determining the rate chemistry and for model anodes it has been shown that the polarisation resistance scales linearly with the TPB length [11].

(17)

6 Figure1-2 shows the Focused Ion Beam milled Scanning Electron Microscopy (FIB-SEM) image of a Ni-YSZ cermet anode. Clearly the exact TPB length of the anode is difficult to determine as the distribution of grains and pores is heterogeneous. Moreover it is most likely that the TPB changes during operation.

Figure 1‐2: Microstructure of the anode

1.3 Kinetics of fuel oxidation and need for pattern anodes

While overall oxidation mechanisms of both hydrogen and CO oxidation are well known, there is very little consensus on the elementary steps involved especially for CO oxidation. The kinetics would generally include chemical as well as electrochemical steps, surface species and reaction rates. Detailed knowledge of the reaction rates can help in optimising the anode microstructure, identify alternative materials and design anodes based on fuels.

The study of elementary reactions on the SOFC anode has been limited by the operating conditions where catalysis data is not available easily. The microstructure makes it difficult to obtain accurate kinetics because of overlapping chemical, electrochemical and mass transport phenomena which occur in the nickel, ceramic as well as gas phase. One way to circumvent the problem of not being able to localise and study reactions is to use pattern anodes.

Pattern anodes are anodes with a well-defined 2 dimensional geometry and contact area between the anode and electrolyte so that the TPB can be precisely identified. The TPB lengths with pattern anodes are usually smaller than operating cermet anodes. This makes it possible to study reaction mechanisms without interference from other limiting phenomena such as gas phase diffusion which can be a major influence in observed polarisation resistance in case of cermet anodes [12]. The cells are usually tested using AC or DC techniques and the analysis is carried out by equivalent circuit fitting. While equivalent circuit fitting gives a way to predict the electrochemical behaviour of a system, it is not straightforward to interpret the results in terms of mechanistic details.

Ceramic Pore

(18)

7 Previous experimental studies on nickel pattern anodes for hydrogen oxidation conclude that the rate determining step could be adsorption/ desorption [11, 13], removal of O2-_{from the}

electrolyte [11], charge transfer [14]. Similarly for CO oxidation usually a charge transfer step is considered limiting [15]. Clearly using experimental data alone, one cannot propose reaction mechanisms.

1.3.1 Modelling of kinetics

Modelling of the experimental data in order to establish kinetics has been a subject of various studies [16-22]. The models have been simple rate equations to state space models to exhaustive models taking into account surface chemistry, charge transfer reactions, diffusion and global mass and charge transport. These have been instrumental in identifying the various mechanisms of charge transfer, which surfaces are relevant for the charge transfer and rate determining steps. For example for hydrogen oxidation, the most predominant mechanisms discussed are hydrogen and oxygen spillover mechanisms [16, 19, 21]. In the hydrogen spillover, the hydrogen atoms diffuse or spillover to either an oxygen ion or a hydroxyl ion on the YSZ surface. In the oxygen spillover, oxygen ions spillover from the YSZ surface onto the Ni surface and combined with a hydrogen there. Modelling of CO oxidation has received attention recently and detailed mechanisms are modelled for nickel pattern anodes on YSZ similar to that with hydrogen. Usually oxygen spillover mechanisms are discussed for CO [22] which brings in other limitations like diffusion [20] on the surface in the discussion.

1.4 Ceria in SOFC anodes

The preceding discussion is based on the nickel being the active phase for fuel adsorption and electron removal since the YSZ acts as purely an ion conductor. The presence of a separate ionic and electronically conducting phase in principle defines the TPB. Since there is a clear advantage in increasing the triple phase boundary length, it would be advantageous to have the ionic phase to participate in the fuel adsorption/ electron removal thus spreading the TPB length beyond the interface between nickel and ceramic. Mixed Ionic Electronic Conductors (MIECs) have received a lot of attention recently as materials for SOFC electrodes [8, 23, 24]. Primary among them is ceria which has very high ionic conductivity (0.1 Scm-1_{at 800}0_{C) and is among the leading}

candidates for the electrolyte of intermediate temperature SOFCs. One of the main reasons for considering ceria for SOFC anodes is its electronic conductivity under reducing conditions.

Ceria is a well-known catalyst for hydrogen and CO thermal oxidation. In fact it used mainly as a selective CO oxidation catalyst. While ceria has been used extensively as an anode material instead of traditional YSZ, very little is known about the mechanism of oxidation of hydrogen or CO on ceria. While it is well known that the MIEC nature of ceria helps in extending the TPB beyond just the interface, the participation of ceria in fuel oxidation has not been quantified. Further the mechanism of hydrogen oxidation on CO, its rate as compared with traditional catalysts like nickel and the limiting mechanisms have not been studied in detail. Even lesser is known about electrochemical CO oxidation on ceria. Selective hydrogen oxidation is reported for mixtures of H2-CO on nickel [25] but considering the propensity of ceria for CO,

(19)

8

1.5 Aims of the study

A detailed understanding of fuel oxidation on SOFC anodes working on different fuel mixtures helps in selection of suitable materials for particular fuel types. The above discussion indicates that mechanistic details are not clearly understood for oxidation of hydrogen or CO on ceria and to an extent on nickel. Hence, kinetic parameters are not readily available. The complex anode microstructure is one of the major contributors to this as various processes overlap and limiting mechanisms can be other than the reaction itself. Further application of the obtained kinetics from fundamental studies into macroscopic studies remains an area which has not received a lot of attention.

Using pattern anodes, one can obtain kinetic parameters for fuel oxidation. Elementary kinetic modelling helps establish mechanistic details, rate limiting steps and other kinetic parameters. This knowledge can be combined together to obtain an expression for activation losses in a typical operating SOFC as pattern measurements are not usually influenced by gas phase or microstructure effects. Thus the rate expressions or models developed here can be extended to use for cermet anodes of Ni-Ce and even to larger simulations and system studies.

System studies of an SOFC with a bottoming cycle like a Gas Turbine (GT) can give important information about the thermodynamic limits of achievable efficiencies especially while comparing across different fuels. Identification of individual components contributing to exergy losses can be crucial for system design and optimisation of next generation power plants.

From the above discussions the following aims are laid out for this thesis

 To make pattern anodes of known geometry so that the reactive zone can be quantified exactly.

 Quantify and compare the electrochemical response of nickel and ceria pattern cells under different fuel gas mixtures.

 Build an elementary electrochemical model to describe the kinetics of fuel oxidation for H₂ and CO on ceria which can be further used in larger simulations.

 Study the effect of contaminants on the performance of SOFCs using pattern cells of nickel and ceria. Include the effect of contaminant in terms of a kinetic term so that performance can be predicted more accurately with contaminant laden gas.

 Development of thermodynamic concepts for high efficiency energy systems based on SOFCs such as SOFC- gas turbine systems which can produce electricity at very high efficiencies. Identify exergy losses in system and individual components with focus of the SOFC component.

1.6 Thesis outline

The thesis is divided into 6 chapters.

Chapter 1 is the introductory chapter

Chapter 2 provides details about experimental methods. In particular, making of model anodes of

nickel and ceria. Since most of the testing is done with Electrochemical Impedance Spectroscopy (EIS), a short introduction is provided about the technique and the test setup as well. Three different types of pattern anodes (Nickel pattern on Yttria stabilized Zirconia (YSZ) (Ni-YSZ), ceria pattern on YSZ (Ce-YSZ), Nickel pattern on ceria (Ni-Ce)) are prepared and tested in order to gain an insight into the electrochemical oxidation of hydrogen on nickel and ceria.

(20)

9

Chapter 3 discusses mechanisms for electrochemical oxidation of H2 and CO on nickel and ceria

pattern anodes with and without water. Further, H₂/CO mixtures are studied in dry and wet conditions. The comparison is made based on polarisation resistances as well as activation energies by which we can compare the catalysis as well. A preliminary observation is made about selective oxidation of H₂ or CO in syn gas mixtures.

Chapter 4 delves into building an elementary kinetic model considering charge transfer and

adsorption steps for electrochemical H2 and CO oxidation on ceria by fitting of impedance

spectra. A novel method is presented to extend the kinetics obtained from pattern anodes to macroscopic simulations in which the activation overvoltage can be calculated on the basis of elementary kinetics.

Chapter 5 discusses the effect of H2S on pattern cells of nickel and ceria. While it is well known

that cermet anodes are poisoned significantly increasing the polarization resistance, the

mechanism for poisoning is still under discussion. Different aspects of the poisoning such as the reversibility and speed of poising and recovery are presented. The effect of H2S poisoning is

modelled with elementary kinetics for H2S adsorption on ceria pattern cells assuming the

adsorbed species do not take part in the charge transfer process. It is expected that he rate constants obtained here can be used in extending the pattern electrode work to larger macroscopic simulations.

Chapter 6: Solid Oxide Fuel Cell – Gas Turbine (SOFC-GT) systems provide a thermodynamically

high efficiency alternative for power generation from biofuels. In this chapter biofuels namely methane, ethanol, methanol, hydrogen and ammonia are evaluated using exergy with respect to their performance at system level and in system components like heat exchangers, fuel cell, gas turbine, combustor, compressor and the stack. Further, the fuel cell losses are investigated in detail with respect to their dependence on operating parameters such as fuel utilisation, Nernst voltage etc as well as fuel specific parameters.

(21)

10 References

[1] http://www.iea.org/media/statistics/surveys/electricity/mes.pdf.

[2] James Larminie AD. Fuel Cell Systems Explained: John Wiley & Sons Ltd; 2003. [3] Patel HC, Woudstra T, Aravind PV. Thermodynamic Analysis of Solid Oxide Fuel Cell Gas Turbine Systems Operating with Various Biofuels. Fuel Cells. 2012;12:1115-28. [4] Kan WH, Thangadurai V. Challenges and prospects of anodes for solid oxide fuel cells (SOFCs). Ionics. 2015;21:301-18.

[5] Gorte RJ, Vohs JM, McIntosh S. Recent developments on anodes for direct fuel utilization in SOFC. Solid State Ion. 2004;175:1-6.

[6] Cheng Z, Wang J-H, Choi Y, Yang L, Lin MC, Liu M. From Ni-YSZ to sulfur-tolerant anode materials for SOFCs: electrochemical behavior, in situ characterization, modeling, and future perspectives. Energy & Environmental Science. 2011;4:4380-409.

[7] Tsipis EV, Kharton VV. Electrode materials and reaction mechanisms in solid oxide fuel cells: a brief review. III. Recent trends and selected methodological aspects. J Solid State Electrochem. 2011;15:1007-40.

[8] Gorte RJ, Vohs JM. Catalysis in Solid Oxide Fuel Cells. In: Prausnitz JM, editor. Annual Review of Chemical and Biomolecular Engineering, Vol 2. Palo Alto: Annual Reviews; 2011. p. 9-30.

[9] Verbraeken MC, Ramos T, Agersted K, Ma Q, Savaniu CD, Sudireddy BR, et al. Modified strontium titanates: from defect chemistry to SOFC anodes. RSC Adv. 2015;5:1168-80.

[10] Aravind PV, Ouweltjes JP, Woudstra N, Rietveld G. Impact of Biomass-Derived

contaminants on SOFCs with Ni/Gadolinia-doped ceria anodes. Electrochem Solid State Lett. 2008;11:B24-B8.

[11] Bieberle A, Meier LP, Gauckler LJ. The electrochemistry of Ni pattern anodes used as solid oxide fuel cell model electrodes. J Electrochem Soc. 2001;148:A646-A56.

[12] Aravind PV, Ouweltjes JP, Schoonman J. Diffusion Impedance on Nickel/Gadolinia-Doped Ceria Anodes for Solid Oxide Fuel Cells. J Electrochem Soc. 2009;156:B1417-B22. [13] Mizusaki J, Tagawa H, Saito T, Kamitani K, Yamamura T, Hirano K, et al.

PREPARATION OF NICKEL PATTERN ELECTRODES ON YSZ AND THEIR

ELECTROCHEMICAL PROPERTIES IN H2-H2O ATMOSPHERES. J Electrochem Soc. 1994;141:2129-34.

[14] Boer Bd. Hydrogen oxidation at porous nickel and nickel/yttriastabilised zirconia cermet electrodes. Twente: University of Twente,; 1998.

[15] Utz A, Leonide A, Weber A, Ivers-Tiffee E. Studying the CO-CO2 characteristics of SOFC anodes by means of patterned Ni anodes. J Power Sources. 2011;196:7217-24. [16] Bessler WG, Vogler M, Stormer H, Gerthsen D, Utz A, Weber A, et al. Model anodes and anode models for understanding the mechanism of hydrogen oxidation in solid oxide fuel cells. Physical Chemistry Chemical Physics. 2010;12:13888-903.

[17] Bieberle A, Gauckler LJ. State-space modeling of the anodic SOFC system Ni, H2– H2O∣YSZ. Solid State Ion. 2002;146:23-41.

[18] Ciucci F, Chueh WC, Goodwin DG, Haile SM. Surface reaction and transport in mixed conductors with electrochemically-active surfaces: a 2-D numerical study of ceria. Physical Chemistry Chemical Physics. 2011;13:2121-35.

[19] Goodwin DG, Zhu H, Colclasure AM, Kee RJ. Modeling electrochemical oxidation of hydrogen on Ni--YSZ pattern anodes. J Electrochem Soc. 2009;156:B1004-B21.

[20] Hanna J, Lee WY, Ghoniem AF. Kinetics of Carbon Monoxide Electro-Oxidation in Solid-Oxide Fuel Cells from Ni-YSZ Patterned-Anode Measurements. J Electrochem Soc. 2013;160:F698-F708.

(22)

11 [21] Vogler M, Bieberle-Hutter A, Gauckler L, Warnatz J, Bessler WG. Modelling Study of Surface Reactions, Diffusion, and Spillover at a Ni/YSZ Patterned Anode. J Electrochem Soc. 2009;156:B663-B72.

[22] Yurkiv V, Utz A, Weber A, Ivers-Tiffee E, Volpp HR, Bessler WG. Elementary kinetic modeling and experimental validation of electrochemical CO oxidation on Ni/YSZ pattern anodes. Electrochimica Acta. 2012;59:573-80.

[23] Gaudillere C, Vernoux P, Mirodatos C, Caboche G, Farrusseng D. Screening of ceria-based catalysts for internal methane reforming in low temperature SOFC. Catalysis Today. 2010;157:263-9.

[24] Marban G, Fuertes AB. Highly active and selective CuOx/CeO2 catalyst prepared by a single-step citrate method for preferential oxidation of carbon monoxide. Applied Catalysis B-Environmental. 2005;57:43-53.

[25] Sukeshini AM, Habibzadeh B, Becker BP, Stoltz CA, Eichhorn BW, Jackson GS. Electrochemical oxidation of H-2, CO, and CO/H-2 mixtures on patterned Ni anodes on YSZ electrolytes. J Electrochem Soc. 2006;153:A705-A15.

(23)

12

Chapter 2 Model anodes of nickel

and ceria

This chapter have been published in

H C Patel, N Biradar , V Venkataraman and P V Aravind “Ceria Electrocatalysis Compared to Nickel

(24)

13 In this chapter, making of model anodes of nickel and ceria are discussed. The pattern cells are characterised using optical microscopy and SEM imaging and found to be satisfactory. Testing is carried out using Electrochemical Impedance Spectroscopy (EIS). Three different types of pattern anodes (Nickel pattern on Yttria stabilized Zirconia (YSZ) (Ni‐YSZ), ceria pattern on YSZ (Ce‐ YSZ), Nickel pattern on ceria (Ni‐Ce)) are prepared and tested in order to gain an insight into the electrochemical oxidation of hydrogen on SOFC anodes. Patterns of nickel on YSZ demonstrate highest polarisation resistance followed by ceria on YSZ. Polarisation resistance is least with patterns of nickel on ceria indicating the dominance of surface path of hydrogen oxidation on ceria rather than that of spillover from nickel.

(25)

14

2.1 Introduction

Cermet of Nickel (Ni) - Yttria stabilized Zirconia (YSZ) or Gadolina doped Ceria (GDC) are now commercially used as SOFC anodes. The cermet anode is a 3D structure with

interpenetrating ionically conducting phase (YSZ) and an electronic conducting phase (Ni). As a result of this mixing of the two phases the microstructure is difficult to quantify exactly.

Moreover, while operating the microstructure may undergo changes. Thus the reaction

mechanisms are difficult to elucidate without accurate knowledge of the triple phase boundaries (TPBs) – where the ionically conducting phase, electronically conducting phase and gas phase are in contact with each other. The knowledge of reaction mechanisms at the TPBs is essential for understanding the fuel oxidation and refining existing designs. One way to localize and study reactions is to make anodes with well -defined geometry and hence TPBs. Also variations across the third dimension (thickness) can be omitted to provide a much simpler geometry with well-defined TPB lengths. Hence a 2D pattern is chosen where the TPB length can be measured accurately. Impedance measurements on this geometry can give invaluable insight into the reaction mechanisms at the SOFC anode.

2D model electrodes have been studied extensively to elucidate anode electrochemistry[1-5]. Almost all these studies have been dedicated to nickel pattern anodes on yttria stabilised zirconia (YSZ). Previous work on model electrodes has mainly focussed on nickel as the electrode. However, mixed ionic electronic conductors (MIEC) like ceria present an interesting case to study purely (ie without a metal) as anode materials. The inherent advantage offered by MIECs is that the reaction zone is not limited to the metal ceramic interface. The reaction zone can extend to the surface of the ceramic and oxidation can occur independent of metal mediated catalysis. Hitherto most studies relate catalytic activity with the metal and the general viewpoint is that metal catalysis is better than that of ceramics[6, 7]. In this work with the use of pattern anodes we aim to compare (electrochemical) catalysis on ceria and compare it to that of nickel. Recent studies on ceria indicate that ceria based anodes (without metal enhanced catalysts) can have similar polarisation resistance as traditional cermet anodes[8]. Most studies in pattern anode literature use pattern anodes of nickel on YSZ[1, 2] [3, 9]. Studies of patterns on MIECs are rare and anodes with purely MIECs are even more difficult to find. Most studies on MIEC materials have been performed for the cathode[10]. Nickel patterns on Samaria doped ceria (SDC) deposited on YSZ have been studied recently[8] where it was found that the oxidation reaction proceeds on SDC surface rather than the interface. In another study, the polarization resistance of nickel patterns on an MIEC - LSGM ((La,Sr)(Ga,Mg)O3−x) was found to be lesser than that of

(26)

15 MIEC and metal surfaces, pattern anodes of the MIEC material can be made on YSZ. In

addition to metal/ YSZ and MIEC YSZ, metal patterns on MIEC should complete the picture as far as different possible combinations of anodes are concerned. This comparison between

different anode/ substrate combinations can give a clear basis for understanding the hydrogen oxidation mechanism on nickel and ceria. It is aimed in this work to shed some light on the mechanism of hydrogen oxidation on ceria as well as on nickel. While ceria has been studied as a catalyst in the past, most of the studies have been related to kinetics of ceria reduction [12-14]. The use of ceria as an electrocatalyst is not well understood. Also symmetrical cell measurements are very rare in pattern anode literature. The differences between hydrogen oxidation mechanism on nickel and ceria can lead to a more optimized design of the SOFC anode especially when operating with real fuels with contaminants.

2.2 Experiments

Nickel and ceria pattern anodes were made by DC Magnetron sputtering of respective targets on YSZ and ceria disks via stainless steel masks. A schematic of the pattern is shown in Figure 2-1.

Figure 2‐1: Schematic of pattern anodes

Three types of anodes are discussed 1. Nickel pattern on YSZ (Ni-YSZ) 2. Ceria pattern on YSZ (Ce-YSZ) 3. Nickel pattern on ceria (Ni-Ce)

Nickel deposition on YSZ and ceria was carried out at 100 W and 3*10-6_{bar of argon. A growth}

rate of 0.12 nm/s was achieved. The deposition of ceria was carried out at 150 W with all other parameters remaining the same. The deposition rate achieved was 0.035 nm/s. Nickel patterns were deposited with a thickness of 1780 nm on YSZ and ceria. Ceria patterns were deposited on

(27)

16 YSZ at a thickness of 500 nm. The thicknesses were chosen to obtain stable patterns. Light microscopy showed no apparent defect. The patterns were all dense before and after testing and triple phase boundary (TPB) should not be affected by the thickness of the patterns [15, 16]. The sputtering through steel masks gave resonable 2D patterns but at the cost of higher line widths of Ni. The line width obtained was 1mm as compared to micrometer range obtained by other researchers. The TPB length obtained was 150 mm. Figure 2-2 to Figure 2-5 show the Scanning Electron Microscopy (SEM) images of the nickel and ceria patterns. From the figures it is clear that the patterns are not porous.

Figure 2‐2: SEM image of ceria pattern cell

(28)

17

Figure 2‐4 SEM Image of Nickel pattern cell

Figure 2‐5: SEM Image of ceria pattern cell

It is expected that the TPB line is not as sharp as when using more sophisticated techniques like photolithography. However, the above technique is more accessible, much cheaper and reasonable results can be obtained. Current was collected using two gold wires attached to the ends of the patterns using a gold paste as shown in Figure 2-1. We recognize that this method of current collection has a drawback in case of ceria patterns where the electrochemically active zone may be limited to the vicinity of the current collector [17] as the electronic conductivity is limited. However, as will be clear in subsequent sections this should most likely have no effect on the conclusions drawn as only qualitative results are discussed. Ceria in case on Ce-YSZ and Ni-Ce was allowed to reduce under H2-H2O for at least 48 hours prior to measurement.

Electrochemical Impedance Spectroscopy (EIS) was performed on all three cells between 700 to 850 0_{C with 97 % H}

(29)

18 Figure 2‐6: Test Setup BY-PASS HUMIDIFIER HUMIDIFIER OVEN CO N2 H2 CO2 H2 +H2S 50ppm Temperature Control Panel Frequency Response Analyzer (FRA) Flow Control Panel Figure 2‐7: Schematic of test station

(30)

19

2.3 Electrochemical Impedance Spectroscopy (EIS)

Most of the electrochemical characterisation is carried out using EIS in this work. The method and applications are discussed in detail in .. but for the purpose of this work some of the principles are presented here. The most basic measurements would consist of measuring the polarisation on the basis of a perturbation (or overpotential) applied and corresponding signal generated. In the potentiostatic method a small ac voltage signal is applied to the electrochemical cell and then the current response is recorded. The impedance of the system, which is given as the ratio of the voltage to the current is given by the following equation,

(4)

In practice electrochemical cells are not linear, hence electrochemical impedance measurements are done by applying a small excitation signal (10-50 mV) and recording the output response. Impedance plots can be represented in two ways viz Nyquist plots and Bode plots. In the former the imaginary impedance of the electrochemical cell is plotted against the real impedance of the cell, while in the latter the magnitude of the impedance and phase angle are plotted against the frequency. The Nyquist plot is advantageous because it gives information not only on the capacitive and inductive nature of the electrochemical cell but also on the resistive nature of the cell. The activation controlled processes show up as distinct arcs in the Nyquist plot, because they have different time constants and this gives a visual indication of the possible mechanisms or the governing phenomena. The disadvantage however is that the frequency information is implicit and must be indicated for distinct points. With Bode plots the frequency information is explicit and change in magnitude of the impedance can be clearly noticed with frequency.

Both these representations have their own merits and demerits and is usually best to represent them together to provide a more holistic view. However in this work more attention is paid to the shape of the impedance arcs and hence only Nyquist plots are presented.

2.3.1 Equivalent Circuit Models

Equivalent circuit models (ECM) are one of the easiest and quickest ways to interpret the physico-chemical processes recorded by the impedance spectrum. The ECM construction involves the use of circuit elements such as resistors, inductors, capacitors and certain other special elements such as CPE (Constant Phase Element), Warburg elements etc from electrical engineering and putting them together in an equivalent circuit, to represent the impedance. Resistors are generally used to represent the bulk resistance of the material and this could either be resistance of the electrolyte to ion transport or resistance of the electrode to electron transport. Space charge polarization regions and other physic-chemical processes are represented via inductors, capacitors and other special elements.

The circuit elements used in this work is described below.

Polarization resistance, RP :

The polarization resistance is defined as the value of the resistance obtained from the low frequency intercept of the impedance data from the Nyquist plot minus the high frequency intercept of the impedance data on the Nyquist plot. This is attributed to contributions from electrodes and also partly from the electrolyte.

(31)

20 The electrolyte resistance is defined as the value of the resistance obtained from the high frequency intercept of the impedance spectrum. This parameter indicates the ohmic losses occurring within the SOFC and is mainly contributed by the electrolyte.

Relaxation Frequency:

This is defined as the frequency at the maximum imaginary impedance value.

Constant Phase element:

The constant phase element is an equivalent circuit element used to model an imperfect capacitor. The expression for the impedance of a CPE is given by

1 (5)

Where, n ranges from 0 to 1, o signifying that the CPE behaves as an ideal resistor and 1 signifying that the CPE behaves as an ideal capacitor.

2.3.2 Symmetrical cells

The concentration difference of oxygen between the cathode and the anode is the driving force in a normal SOFC for the oxide ions to migrate from cathode to the anode via the electrolyte. However in a symmetrical cell set up a single gas atmosphere is used. In this case the applied AC excitation acts as the driving force, polarizing the electrodes, making one of the electrodes act as the anode and the other as the cathode. The symmetrical SOFC now functions as an electrolyzer splitting H2O on one side (at the Cathode) into H+ and O2- ions and as a fuel

cell on the other where H2 is oxidsed. The use of symmetrical cells has some advantages like no

reference electrode, no sealing, easy setting up and operation etc. However it must be recognised that no DC measurements like current voltage measurements can be performed.

(32)

21

2.4 Results and Discussion

Figure 2-9, Figure 2-10 and Figure 2-11 show impedance spectra obtained for nickel –YSZ (Ni-YSZ), ceria YSZ (Ce-YSZ) and nickel ceria (Ni-Ce) pattern anodes respectively. Equivalent circuit fitting was carried out on all three spectra at different temperatures using the equivalent circuit in Figure 2-12.

Figure 2‐9 Impedance spectra with nickel pattern on YSZ electrolyte[15, 16]

(33)

22

Figure 2‐11: Impedance spectra with nickel pattern on ceria electrolyte

Figure 2‐12 Equivalent circuit model

In order to compare the results obtained with other work on nickel pattern anodes, the

conductivity (inverse of the total polarisation (R1+R2+R3)) is computed at 2.33 kPa and 700 0_C

[18]. Figure 2-13 shows the comparison on the conductivity obtained by different research groups in H2 atmosphere at 2.33 kPa H2O partial pressure and 700 0C at at TPB length of 0.2027

m/cm2_{on nickel pattern anodes. Appropriate assumptions were used to scale the data from}

literature to above mentioned values of partial pressure, temperature and TPB length. The conductivity obtained is well within acceptable range compared to previous work [1-3] showing that pattern anodes prepared by this method can be used to obtain reasonable results.

(34)

23

Figure 2‐13: Comparison of conductivity from literature

The polarization resistance is highest for Ni-YSZ and lowest for Ni-Ce at all temperatures. This is interesting in itself as it shows that even undoped ceria can act as an anode and that ceria is active for hydrogen oxidation. Ceria is a well-known catalyst in shift reactors for its capacity to store oxygen [12-14, 19, 20]. The two oxidation states of ceria (Ce3+_{and Ce}4+_{) are}

stable and intermediate stoichiometries of CeO(2-x) are stable as well[21]. Electrochemical

oxidation of hydrogen on the surface of reduced ceria should take place by the oxygen ions supplied from the electrolyte.

The oxidation of hydrogen on the surface of reduced ceria makes the triple phase boundary of the pattern very difficult to determine as it would seem the reaction is not limited only to the interface between YSZ and ceria but it spreads over the surface of ceria making the three phase boundary concept redundant in this case. A two phase boundary is sufficient for oxidation [8].

Since the surface of ceria is active for hydrogen oxidation it is expected that the polarization of Ni Ce anodes is the least. The polarization resistance obtained at 800 0_{C is 6 ohms which is 3}

orders of magnitude smaller than that of Ni-YSZ and 2 orders of magnitude smaller than Ce-YSZ even though the nominal TPB length is the same. This is a significant result as it implies that the TPB in case of Ni-Ce is not very important and most of the reaction seems to proceed through the surface of ceria. The polarization resistance for Ni-Ce approaches the polarization obtained with a cermet anode. Indeed it has been shown that with nano-structured ceria, the performance of a cermet anode can be achieved at least with regards to the polarization[8].

(35)

24 Results of Ni- YSZ can be compared to existing literature. No such data is available for Ce-YSZ or Ni-Ce anodes as far as the authors could find.

Figure 2‐14 Hydrogen oxidation on nickel pattern anodes on YSZ

(36)

25

Figure 2‐16 Hydrogen Oxidation on nickel pattern anodes on ceria

2.5 Conclusions

Three different types of pattern cells are prepared and tested at different temperatures viz. 1. Nickel pattern on YSZ (Ni-YSZ)

2. Ceria pattern on YSZ (Ce-YSZ) 3. Nickel pattern on ceria (Ni-Ce)

The pattern anodes obtained have been characterized by SEM and light microscopy imaging. While the TPB line is not as sharp as obtained by lithographic methods, the prepared pattern cells are considered to be satisfactory. It is found that the polarization resistance is the least with Ni –Ce cells and the highest with Ni-YSZ cells. This seems to indicate that hydrogen oxidation on ceria can proceed with any metal mediated catalyst and can be indeed faster than the metal catalyzed pathway. A reaction similar to reduction of ceria is most likely even considering the wide range of activation energies reported in literature. The oxidation is unlikely to be limited to the TPB in case of Ce-YSZ and Ni-Ce anodes but can proceed on the surface of ceria.

(37)

26 References [1] Bieberle A, Meier LP, Gauckler LJ. The electrochemistry of Ni pattern anodes used as solid oxide fuel cell model electrodes. J Electrochem Soc. 2001;148:A646-A56. [2] Mizusaki J, Tagawa H, Saito T, Kamitani K, Yamamura T, Hirano K, et al. PREPARATION OF NICKEL PATTERN ELECTRODES ON YSZ AND THEIR ELECTROCHEMICAL PROPERTIES IN H2-H2O ATMOSPHERES. J Electrochem Soc. 1994;141:2129-34. [3] Boer Bd. Hydrogen oxidation at porous nickel and nickel/yttriastabilised zirconia cermet electrodes. Twente: University of Twente,; 1998. [4] Sukeshini AM, Habibzadeh B, Becker BP, Stoltz CA, Eichhorn BW, Jackson GS. Electrochemical oxidation of H-2, CO, and CO/H-2 mixtures on patterned Ni anodes on YSZ electrolytes. J Electrochem Soc. 2006;153:A705-A15. [5] Aaberg RJ, Tunold R, Odegard R. On the electrochemistry of metal-YSZ single contact electrodes. Solid State Ion. 2000;136:707-12. [6] Gorte RJ, Vohs JM. Catalysis in Solid Oxide Fuel Cells. In: Prausnitz JM, editor. Annual Review of Chemical and Biomolecular Engineering, Vol 2. Palo Alto: Annual Reviews; 2011. p. 9-30. [7] McIntosh S, Vohs JM, Gorte RJ. Effect of precious-metal dopants on SOFC anodes for direct utilization of hydrocarbons. Electrochem Solid State Lett. 2003;6:A240-A3. [8] Chueh WC, Hao Y, Jung W, Haile SM. High electrochemical activity of the oxide phase in model ceria-Pt and ceria-Ni composite anodes. Nature Materials. 2012;11:155-61. [9] A U. The Electrochemical Oxidation of H2 and CO at patterned Ni anodes of SOFC. Karlsruhe: Karlsruhe Institute of Technology; 2011. [10] Fleig J, Baumann FS, Brichzin V, Kim HR, Jamnik J, Cristiani G, et al. Thin film microelectrodes in SOFC electrode research. Fuel Cells. 2006;6:284-92. [11] Rao MV, Fleig J, Zinkevich M, Aldinger F. The influence of the solid electrolyte on the impedance of hydrogen oxidation at patterned Ni electrodes. Solid State Ion. 2010;181:1170-7. [12] Al-Madfaa HA, Khader MM. Reduction kinetics of ceria surface by hydrogen. Materials Chemistry and Physics. 2004;86:180-8. [13] Campbell CT, Peden CHF. Chemistry - Oxygen vacancies and catalysis on ceria surfaces. Science. 2005;309:713-4. [14] Laachir A, Perrichon V, Badri A, Lamotte J, Catherine E, Lavalley JC, et al. REDUCTION OF CEO2 BY HYDROGEN - MAGNETIC-SUSCEPTIBILITY AND FOURIER-TRANSFORM INFRARED, ULTRAVIOLET AND X-RAY PHOTOELECTRON-SPECTROSCOPY MEASUREMENTS. Journal of the Chemical Society-Faraday Transactions. 1991;87:1601-9. [15] Patel HC, Biradar N, Venkataraman V, Aravind PV. Ceria Electrocatalysis Compared to Nickel Using Pattern Anodes. International Journal of Electrochemical Science. 2014;9:4048-53. [16] Patel HC, Biradar N, Venkataraman V, Aravind PV. Pattern Electrodes for Studying SOFC Electrochemistry. In: Kawada TS, SC, editor. 13th International Symposium on Solid Oxide Fuel Cells (SOFC-XIII) Oct 06-11, 2013. Okinawa, JAPAN: ECS Transactions; 2013. p. 1613-8.

(38)

27 [17] Zhang CJ, Grass ME, McDaniel AH, DeCaluwe SC, El Gabaly F, Liu Z, et al. Measuring fundamental properties in operating solid oxide electrochemical cells by using in situ X-ray photoelectron spectroscopy. Nature Materials. 2010;9:944-9. [18] Patel HC, Venkataraman V, Aravind PV. Nickel Pattern Anodes for Studying SOFC Electrochemistry. Advances in Solid Oxide Fuel Cells IX: John Wiley & Sons, Inc.; 2013. p. 89-94. [19] Gennari FC, Neyertz C, Meyer G, Montini T, Fornasiero P. Hydrogen adsorption kinetics on Pd/Ce0.8Zr0.2O2. Physical Chemistry Chemical Physics. 2006;8:2385-95. [20] Trovarelli A. Catalytic properties of ceria and CeO2-containing materials. Catalysis Reviews-Science and Engineering. 1996;38:439-520. [21] Sohlberg K, Pantelides ST, Pennycook SJ. Interactions of hydrogen with CeO2. Journal of the American Chemical Society. 2001;123:6609-11.

(39)

28

Chapter 3 Oxidation of H2

, CO and

syngas mixtures on ceria and nickel

pattern anodes

This Chapter has been published as

H.C. Patel, A.N. Tabish, F. Comelli, P.V. Aravind “Oxidation of H2, CO and syngas mixtures on ceria

(40)

29 Understanding synthesis gas (syngas) oxidation on SOFC anodes is an important step towards commercialising SOFCs operating on realistic fuels. Using pattern anodes, we study electrochemical oxidation of H2 and CO on nickel and ceria pattern anodes with and without water. Further, H2/CO mixtures are studied in dry and wet conditions. For all the compositions studied here, the polarisation resistance on ceria pattern anodes is lower than nickel pattern anodes indicating that the reaction zone is not limited to the three phase boundary. Also, in most cases the activation energy is also lower indicating ceria can be a superior catalyst. It is found that the polarisation resistance with CO is higher than that with hydrogen for both ceria and nickel pattern anodes. For wet CO mixtures there is a possibility that water gas shift reaction can play an important role. Addition of H2 brings about a dramatic decrease in the polarisation resistance in case of H2/CO mixtures for both nickel and ceria pattern anodes. This indicates that hydrogen is preferentially oxidised so much so that it is possible that only hydrogen is oxidised with CO acting as a diluent.

(41)

30

3.1 Introduction

Understanding oxidation of syngas mixtures has particular importance for designing optimised electrodes for SOFCs working on biomass derived gases or even natural gas feedstock. Present efforts are directed towards developing anode materials capable of operating with hydrocarbon fuels without carbon deposition[1-3] . As different types of anodes have their own advantages, it is important to compare the anodes for the type of fuel and application under consideration Following our previous work on gasifier SOFC gas turbine systems development[4-10], in this study we compare the performance of different anode materials when fed with H2, CO and

syngas. The traditional cermet anode is a porous three dimensional structure with metallic and ionic phases heterogeneously mixed making reactions difficult to localise. Pattern electrodes are employed in this study as they can help in understanding electrochemical fuel oxidation without being influenced by anode microstructure. Using pattern anodes, individual processes like charge transfer, adsorption diffusion etc. can be relatively easier to localise and identify. In this study pattern anodes of nickel and ceria are studied on Yttria Stabilised Zirconia (YSZ) electrolytes. Previous studies on model electrodes indicate that electrochemical oxidation of hydrogen can proceed on ceria without any metal mediated catalysis[11-14].Hence, it is important to study fuel oxidation on nickel as well as on ceria. While hydrogen oxidation on model electrodes has been studied extensively [15-18], fewer studies have been performed for evaluation of CO oxidation on model electrodes . CO-CO2 mixtures have been studied before on cermet anodes or on

porous metal anodes [19-21]. Model anodes like point, grid and pattern anodes have been used to study CO-CO₂ characteristics and parameter dependencies [22-25].

Mixed ionic electronic conductors like ceria may have the advantage over purely ionic conductors of the extension of the reactive zone into the ceramic phase[13, 26]. The polarisation resistance of hydrogen oxidation on ceria has been shown to be lower than that on nickel pattern anodes and that a spillover step is not required in case of ceria [12, 13]. No such studies of electrochemical oxidation of CO on ceria have been reported to the knowledge of the authors. In fact ceria is traditionally used as a catalyst for preferential CO oxidation in hydrogen CO streams[27-30]. Whether the same behaviour is followed for electrochemical oxidation is not studied.

In H2, CO and H2O mixtures, there is a possibility of water gas shift reaction followed by

hydrogen oxidation. Whether this is preferred or direct CO oxidation is preferred is an important question as well. Hydrogen CO mixtures have been rarely studied with pattern anodes. Sukeshini et al [24] studied H₂ /CO mixtures on nickel pattern anodes and concluded that there is preferential oxidation of hydrogen compared to CO. Whether the same trend is followed on ceria as well is not studied to the knowledge of the authors.

In this work, pattern anodes of nickel and ceria are tested in CO, H₂ and mixtures of CO/H₂ with and without humidification. Activation energies have been calculated for processes on the different pattern anodes and compared with available literature on pattern cells, molecular level calculations and catalyst studies.

3.2 Experiments

Nickel and ceria patterns were made according to the schematic in 2-6. The detailed procedure has been described previously [11, 12, 31]. Since symmetrical cells are used for testing, a single gas atmosphere is employed. Electrochemical Impedance Spectroscopy (EIS) was performed using a GAMRY Reference 600 potentiostat on the pattern cells between 700 and 850 o_{C using}

the following compositions (volume %). Ceria in case on Ce-YSZ was allowed to reduce under H2 for at least 48 hours prior to measurement. Since CO is a good reducing agent as well, the

(42)

31 patterns should remain reduced with CO. The results obtained with humidified hydrogen have been reported previously[12] but have been included here for comparison with CO and H2/ CO

mixtures. 2

1. 50% H2, 50% CO2

2. 50% CO, 50% CO2

3. 20% H2, 30% CO, 50% CO2

4. Humidified H2 ( 96% H2, 4% H2O)- referred as 100 H2 (reported previously)

5. Humidified CO (( 96% CO, 4% H2O)- referred as 100 CO

6. Humidified H2-CO (48% H2, 48% CO 4% H2O) referred as 50 H2 50 CO

The mixture compositions are chosen to firstly compare the oxidation of CO and H2 and

secondly to study the effect of CO addition to H2. Testing in dry as well as wet atmospheres is

also important as this can be an important pointer towards the presence of water gas shift reaction in the oxidation of CO.

3.3 Results and Discussion

Figure 3.1, Figure 3.2, Figure 3.3 and Figure 3.4 shows the impedance spectra obtained at 800 o_C

in dry and wet gas atmospheres for nickel and ceria pattern anodes respectively. Insets in Figures 3.2 and Figure 3.4 depict the high frequency region of the Nyquist plots. Figure 3.5 depicts the equivalent circuit model with two R- CPE (Constant Phase Element) representing two processes and a series resistance representing the ohmic resistance. For the purpose of discussion, the process corresponding to high frequency is referred to as HF and low frequency as LF. The corresponding R-CPE elements are R2-CPE1 and R3-CPE2 for HF and LF respectively. The polarisation resistance for ceria pattern anodes is lower than nickel for all compositions at the same temperatures. This is consistent with previous findings on hydrogen oxidation on ceria and nickel pattern anodes [12]. This indicates that ceria is perhaps better than nickel as a catalyst for hydrogen as well as CO oxidation. Comparing oxidation of CO and H2, the polarisation

resistance is always higher for CO than for H₂ for both ceria and nickel pattern anodes in both dry and wet gas atmospheres. This is a novel result for ceria and has been hitherto not been reported. Ceria is used as a preferential catalyst for CO oxidation in hydrogen mixtures but the same behaviour is not followed for electrochemical oxidation on ceria.

It can be seen that the addition of water causes a large drop in the polarisation resistance for both nickel and ceria pattern anodes. While this effect has been reported previously for nickel pattern anodes [16], with these measurements it shown to be true for CO oxidation as well.

(43)

32

Figure 3‐2: Impedance spectra at different compositions (DRY) at 800 oC on Nickel pattern anodes

(44)

33

Figure 3‐4: Impedance spectra at different compositions (DRY) at 800 0C on Ceria pattern anodes

Figure 3‐5: Equivalent circuit model

3.1 CO oxidation

(45)

34

Figure 3‐7: Arrhenius plots of resistances of humidified carbon monoxide on nickel pattern cells

(46)

35

Figure 3‐9 : Arrhenius plots of resistances of humidified carbon monoxide on ceria pattern cells 3.1.1 CO oxidation (dry) on nickel pattern anodes

Firstly we discuss the oxidation of CO on Ni patterns. The global oxidation reaction can be formulated as follows

6) ↔ 2

3.1.1.1 Previous work on CO oxidation with model electrodes

Limiting mechanisms have been discussed by many using impedance data, Tafel slopes or modelling of mechanisms. Estell [19] performed experiments on porous platinum electrodes on YSZ and concluded that there should be two species taking part in the rate determining step. Similarly, Mizusaki [18] tested CO oxidation on porous platinum on YSZ. They proposed adsorbed species on platinum and YSZ taking part in the charge transfer process. It was concluded that adsorption desorption of CO or CO2 cannot be rate limiting. Lauvstad [23]

studied CO-CO2 point electrodes where the analysis was based on low frequency inductive loops.

[20] and [24] observed considerable scatter or instability while operating with CO-CO2 gas

atmospheres. Yurkiv et al [31] propose an oxygen spillover kind of mechanism where oxidation happens in a Langmuir-Hinshelwood type reaction on the nickel surface involving adsorbed CO and oxygen. The oxygen from the electrolyte gives up an electron and spills over to the nickel surface. The global activation energy was 1.46 eV and 0.87 eV for high and low CO/CO2 ratios

respectively. Hanna et al [32] have considered CO being directly involved in the charge transfer step. An adsorption/ desorption step is followed by charge transfer steps. In the charge transfer step O from the electrolyte gives up an electron and moves to the TPB where it further reacts with the adsorbed CO and finally CO2 is desorbed. They conclude that the surface diffusion

(length scale 10 nm) of adsorbed CO on the nickel to the TPB is the rate determining process. They argue that CO is very strongly adsorbed on the Ni and is not mobile. For CO oxidation, Utz et al [33] also obtained activation energy values for CO oxidation on nickel in the range of 0.85 to 1.42 eV. The oxidation can be formulated according to reactions O2, 7,8 and 9 as shown in Table 1 where ( )_Ni is a vacant site on nickel, CO_Ni is adsorbed CO on nickel, O-1

YSZ is an

oxygen atom on the surface of YSZ and ONi is the corresponding species spilled over to the

(47)

36 Table 3‐1: Reaction mechanisms Activation energies for CO oxidation Reaction Ea (eV) O2 ↔ 1.81 [31] 7) ↔ 0 [31] 8) ↔ 1.275 [31] 9) ↔ 1.85 [32] 3.1.1.2 Discussion of results obtained

The activation energies (Figure 3.6) obtained at HF and LF for dry conditions are 1.8 and 1.66 eV respectively. The high activation energy at HF can be indicative of a charge transfer step. It is expected that adsorption of CO is not limiting [21]. Moreover, CO is strongly adsorbed on nickel. The oxidation can take place at the TPB region near the surface of nickel. If CO is indeed strongly adsorbed on nickel, it makes a CO spillover mechanism unlikely. If oxygen spills over from the YSZ, then charge transfer may occur via the oxygen spillover mechanism as proposed by Yurkiv. However, it may not take place in one step – a second step involving adsorbed CO on nickel can be considered for charge transfer as well (reaction 9 as proposed by Hanna). It is difficult to come to a definite conclusion about the charge transfer mechanism as the activation energies obtained with experiments matches well with either O2 or with reaction 9.

3.1.2 CO oxidation (dry) on ceria pattern anodes

The polarisation resistance for CO oxidation on nickel is much higher than that on ceria(Figures 3.1 -3.4) (almost double for both dry and wet cases). Also the activation energy is lower (Figure 3-8 and Figure 3-6). This indicates that catalysis on ceria is probably better as well. If CO oxidation on nickel proceeds via a Langmuir Hinshelwood type reaction [31]then the site occupancy may be the limiting factor in case of nickel. Also the reaction is supposed to be limited to the TPB region in case of nickel pattern anodes. CO oxidation on ceria has been studied so far only in catalysis where ceria can store the oxygen[34, 35]. This is commercially important for example in shift reactors or automobile exhausts but there have been no electrochemical studies of CO oxidation on ceria to the knowledge of the authors. The HF polarisation resistance is very small and not considered to be limiting. The LF activation energy is 1.22 eV for dry CO. The charge transfer mechanism is likely to resemble the ceria reduction process where CO reacts with the oxygen supplied from the electrolyte. Oxygen spillover mechanisms are like in case of nickel pattern anodes are possible but considering the ceria itself has oxygen vacancies, the oxygen can be supplied from the bulk to the surface of ceria. The CO (adsorbed or gaseous) can be directly involved in the charge transfer reaction. There is a possibility of carbonite species as well from the CO – implying CO disproportionates to C and CO₂[35].

While the above discussion is not sufficient to arrive at a definite conclusion about the rate limiting steps, modelling studies should provide an indication of the same. This is ongoing work.

3.1.3 CO oxidation (wet) on nickel and ceria pattern anodes

The polarisation resistance is lower for wet CO than dry CO for both nickel and ceria pattern anodes. With the addition of water, the possibility of water gas shift reaction cannot be ruled out. A hydrogen content of 3.8% can be expected at equilibrium at 800 0_{C for an inlet concentration}

of 4% water in CO. For the small currents employed in symmetrical cells this is enough for producing the entire current. From Figure 3.1 and 3.3, the polarisation resistance for wet

oxidation is closer to hydrogen CO mixtures especially for nickel pattern anodes to a lesser extent for ceria pattern anodes. Thus there is the possibility that water gas shift reaction and subsequent hydrogen oxidation can occur in case of CO- H2O mixtures. The LF activation energy increases

with addition of water which is counterintuitive. This can be because the rate limiting step in case of wet CO may be different with the addition of water which may facilitate adsorption based

(48)

37 intermediates and even water gas shift reaction. As mentioned above, ongoing modelling studies will be instrumental in addressing such questions.

3.2 Hydrogen oxidation

The results for hydrogen oxidation on ceria and nickel pattern anodes have been presented previously but have been included here with an extended discussion in line with the discussions above on CO.

Figure 3‐10: Arrhenius plots of resistances of hydrogen on nickel pattern cells

Multi-Fuel oxidation in Solid Oxide Fuel Cells: Model anodes and system studies