Impact of impurities and interface reaction on electrochemical activity.pdf

Impact of impurities and interface reaction on electrochemical activity M. Mogensen and K. V. Hansen Technical University of Denmark, Roskilde, Denmark 1 INTRODUCTION Segregation of impurities and constituents to the interfaces of ceramics has been treated in many early as well as recent reviews. [1–7] The first two of these are full mono- graphs. These references are examples and not in any way meant to cover the whole literature on the subject. Thus, the intention of this short article is not to give a full pic- ture of the phenomena of segregations to interfaces but to give a brief review of the literature relating to the impact of impurities and interface reactions on the electrochemi- cal performance of solid oxide fuel cell (SOFC) electrodes. The concept “electrochemical activity” of an electrode at a given condition means, in the present context, the over- all rate of conversion of gaseous reactant molecules into ions (in the electrolyte), electrons (in the electrode), and gaseous product molecules. This means that the exchange current density, i 0 , may be used as a measure of the electro- chemical activity. The exchange current density is inversely proportional to the area-specific electrode polarization resis- tance, ASR p , at given conditions. ASR p is often a more convenient measure of electrochemical activity. ASR p may be defined as the slope of the i –V curve and can be mea- sured by electrochemical impedance spectroscopy (EIS) at any electrode overpotential. Materials are never completely pure. Trace elements are always present and may result in desirable or detrimental properties of the material. During fabrication and/or oper- ation at high temperatures, some of these elements will segregate to the interfaces of the material driven by forces such as strain relaxation, compensation of space charges, and lowering of surface tension. [3, 4] Another reason for changes in the surface composition may be kinetic demixing due to transportation of cations in a potential gradient. [8–10] The segregation-induced concentration gradients in metals are typically limited to one or a small number of atomic layers. In ionic solids, such as metal oxides, concentration gradients may extend over hundreds of atomic layers. [6, 7] The properties of the surfaces and interfaces are inher- ently of a nature that tends to change the local chemical composition to reach a lower energy level compared to a surface equal to a simple termination of a single crys- tal. This change may happen by a local surface relaxation, by segregation of components and/or impurities to the sur- face, or by adsorbing reactants from the gas phase. These phenomena seem very detrimental both in the context of SOFC electrolytes and electrodes. Usually, the electrode reactions (oxidation of H 2 , reduction of O 2 ) are concen- trated near the three-phase boundary (TPB), [11] where the electrolyte, electrode, and gas phase meet each other (see also O 2 -reduction at high temperatures: SOFC,Vol- ume 2). TPB seems to be an especially favorable site for accumulation of impurities. [12–15] Thus, a blockage of the TPB by segregated species may decrease the electrochemi- cal activity significantly. The actual activity of really clean electrodes is not known, but reported observations indi- cate that the polarization resistance of very clean model Ni electrodes may be orders of magnitude lower than the usually achieved values. [13] Regarding the measures against the deleterious effects of impurities on the electrochemical activity, it may be wise to look into the area of heteroge- neous catalysis, where the use of nanoparticles improves the Handbook of Fuel Cells – Fundamentals, Technology and Applications. Edited by Wolf Vielstich, Hubert A. Gasteiger, Arnold Lamm and Harumi Yokokawa. ? 2010 John Wiley per- ovskites such as (La 1?x Sr x ) s MnO 3 (LSM), (La 1?x Sr x ) s CoO 3 (LSC), or (La 1?x Sr x ) s Fe 1?y Co y O 3 are the most pre- ferred type of SOFC cathode materials. Furthermore, per- ovskite materials (e.g., doped SrTiO 3 ,La 0.75 Sr 0.25 Cr 0.5 Fe 0.5 O 3 ) are now appearing as important SOFC anode and SOEC (solid oxide electrolyser cell) cathode materials. [49–52] Thus, it is of interest to consider the behavior of perovskite surfaces with respect to segregation phenomena. 3.2.1 Segregation to perovskite surfaces As in the case of YSZ and CGO, materials different from the bulk will segregate to the surfaces, but in the case of the mentioned perovskites, it does not seem to be impurities such as SiO 2 ,Al 2 O 3 ,Na 2 O, etc. that segregate. Instead, the A-site oxide (according to the general perovskite formula ABO 3 ) tends to segregate to the perovskite surface form- ing a thin layer of complex compounds like AO(ABO 3 ) n , n = 1, 2,., 10, ∞ according to Szot et al. [7] for BaTiO 3 , SrTiO 3 ,PbTiO 3 , and KNbO 3 . Similar behavior has been reported for LSM by Caillol et al. [53] The actual composi- tion of the perovskite surface seems to be dependent on tem- perature, oxygen partial pressure, and the size of cathodic polarization. Other parameters such as the precise compo- sition, and in particular the A/B ratio, are supposed to be of significance. Such surface layers may have much lower electronic conductivity than the LSM bulk and may exhibit strongly nonlinear current–voltage characteristics. [54] Baumann et al. [55] found that such “equilibrium” surface layers of SrO enrichments on La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 give rise to a very high electrode polarization resistance. The thin-film dense electrodes, which were deposited on a YSZ single crystal, could be activated by a high cathodic overpotential of 2 V or more. The perovskite is not stable at such a high cathodic overpotential. It will depend on the rate of the solid-state reactions and to what extent the electrode material is broken down or reacted with the electrolyte, but it seems fair to assume that at least the regions near surfaces and interfaces of the La 0.6 Sr 0.4 Co 0.8 Fe 0.2 O 3 are transformed into a multiphase mixture of probably extremely small particles during polarization. When the polarization is lifted, the perovskite structure is apparently formed again. The treatment changed the surface composition toward the bulk composition, and the polarization resistance was decreased with a factor of ～300. Figure 2 illustrates the phenomenon. Figure 2(a) shows that beyond a cathodic bias of ～2.5V no further activation takes place, and Figure 2(b) shows that for 1 V the activation follows a square-root-of-time law up to ～2000 s, indicating that a cation diffusion process is behind this. The electrodes relax back toward 6 Materials for high temperature fuel cells 1 0.1 0.01 0.001 0 1 2 3 4 5 10 100 1000 10 000 Cathodic bias (V) Duration of bias application (s) R / R initial (a) (b) ～t ?0.5 Figure 2. (a) Dependence of the resistance reduction on the magnitude of cathodic dc bias applied to the cell with a 60- μm diameter thin-film (0.1 μm) microelectrode at 700 ? C for an activation period of 3 min. Each data point represents two measurements, once before application of the bias and again 30 s after switching off the voltage, on a different microelectrode. (b) Relative change of the electrode resistance by dc bias pulses of different lengths but with a constant bias magnitude of ?1 V. [F. S. Baumann, J. Fleig, M. Konuma, U. Starke, H.-U. Habermeier, and J. Maier, J. Electrochem. Soc., 152, A2074 (2005). Reproduced by permission of ECS - The Electrochemical Society.] the original state when left at the temperature without any polarization. 3.2.2 Effects of interface silica in LSM–YSZ electrodes (La 1?x Sr x ) s MnO 3 (LSM), where typically 0.1 x0.3, 0.9 s1, has been the preferred type of SOFC cath- ode for most developers. It is well established that MnO x may easily diffuse into the glassy phases along the grain boundaries. [56] Therefore, Appel and Bonanos studied the effect of SiO 2 together with Mn oxide (MnO x ). [57] It was found that MnO x and SiO 2 form a glassy phase in which Y 2 O 3 is dissolved. In the glass phase, 4–5 at % Mn was measured, and a Y/Mn ratio of ～0.5 may be derived from their data. Formation of monoclinic YSZ in regions close to the surface and grain boundaries next to such glassy phase might therefore be expected at intermediate temper- atures of 600–900 ? C, i.e., the operating temperature of SOFCs, because the supply of yttria by cation diffusion from the grain interior may be too slow to compensate the yttria leached to the glass phase, similar to the nonequi- librium segregations observed by de Ridder et al. [21] and as sketched in Figure 1(b). This will, in turn, increase the ohmic resistance of the surface layer of the YSZ signif- icantly, as monoclinic zirconia has a conductivity more than 10 times lower than cubic 8YSZ. [58] A lower con- ductivity near the electrolyte surface may decrease the cell performance because the ASR p of the LSM electrode seems to decrease with decreasing conductivity of the YSZ. [59] It should also be noted that above 8–10 mol% Y 2 O 3 the conductivity of the YSZ decreases rapidly with increas- ing Y 2 O 3 concentration (see Low temperatureelectrolytes and catalysts, Volume 4). Thus, the increased yttria con- centrations (reported above in Section 2) probably also mean that the transfer of oxide ions into the electrolyte will be retarded owing to Y 2 O 3 enrichments at the electrolyte surface. No direct evidence of a detrimental effect on the LSM cathode performance of the impurities (SiO 2 ,Na 2 O, Al 2 O 3 , etc.) segregated to the surface was found in the literature, but reaction products of lanthanum manganate (LM) and SiO 2 were found at the interface between LM and YSZ in the form of La silicate “islands” by Kuscer et al. [60] It was speculated that these silicates could be the reason for the separation of the LM layer from the YSZ electrolyte after aging at high temperatures. Indirect evidences of the impeding effect on the elec- trode kinetics were recently reported by Backhaus-Ricoult and Work. [61] Significant improvements of the LSM–YSZ cathodes were observed, when the YSZ–LSM interfaces were cleaned of impurities. The degradation of SOFCs with LSM–YSZ composite cathodes have been studied at various conditions at Ris? National Laboratory, see e.g., Refs [62–64]. It has been found that the degradation of the LSM–YSZ composite cathode is accelerated by high cathodic overpotential, and that the shape and composition of the LSM–YSZ interac- tion “craters” shown in Figure 3 change to a smaller size. Reaction products containing La and Si are found in these structures. This might indicate that segregations of both La oxide and silica play a role in the performance and degra- dation of the LSM electrodes. It is tempting to relate the formation of these “craters” to processes involving a glassy surface layer similar to the reactions reported by Appel and Bonanos [57] and Kuscer Impact of impurities and interface reaction on electrochemical activity 7 200 nm Figure 3. Electrolyte surface of a cell with LSM–YSZ cathode after test in air and removal of the cathode. Test conditions were 0.75 Acm ?2 and 750 ? C. The “craters” are formed by the interaction between the YSZ and LSM during fabrication. The “craters” change shape during testing. Reaction products containing La and Si are found in these structures. [A. Hagen, Y. L. Liu, R. Barfod, and P. V. Hendriksen, in: SOFC-X, ECS Transactions, 7(1), K. Eguchi, S.C. Singhal, H. Yokokawa, J. Mizusaki, (Eds.), p. 301–309, The Electrochem. Soc. Inc., Pennington, NJ, (2007). Reproduced by permission of ECS - The Electrochemical Society.] et al. [60] However, further work is necessary to clarify the details of the “crater” formation process. 3.2.3 LSF and LSC electrodes on doped ceria electrolyte Mixed electron and oxide ion conductors such as (La 1?x Sr x ) s FeO 3 , s is near 1 and 0 x1 (short LSF), (La 1?x Sr x ) s CoO 3 , s is near 1 and 0 x1 (short LSC), or (La 1?x Sr x ) s Fe 1?y Co y O 3 , s is near 1, 0 x1, and 0 y1 (short LSFC) are usually used as cathode material on doped ceria electrolytes. As doped ceria with respect to surface segregation has a behavior similar to YSZ, problems of interaction with the silica surface layer and the cathode might be expected. Bae and Steele [65] investigated the effect on the polarization resistance of LSFC cathodes by cleaning the gadolinia-doped ceria (CGO) electrolyte by etching with hydrofluoric acid. The LSFC was deposited by electrostatic- spray-assisted vapor deposition. Two types of CGO were studied: Ce 0.8 Gd 0.2 O 1.9 (CGO20) with about 200 ppm SiO 2 , and Ce 0.9 Gd 0.1 O 1.95 (CGO10) with about 30 ppm SiO 2 .A significant (by a factor of ～3) lower ASR p was observed for the LSCF on the etched CGO20 compared to the as- sintered CGO20. No effect of etching was found for the CGO10 electrolyte. Dunyushkina and Adler [66] compared the polarization resistance (measured using EIS) of porous La 0.8 Sr 0.2 CoO 3?δ (LSC) electrodes sintered at 1080 ? C onto polished (“clean”) Ce 0.8 Sm 0.2 O 1.9 (CSO) and onto CSO with segregated surface layer (CSO either as-fabricated or polished and resintered in air at temperatures 1450, 1550, and 1650 ? C). A large effect of the surface treatment of the electrolyte was observed, but surprisingly the polished electrolyte gave the biggest ASR p . Apparently, the LSC adhesion to the as-fabricated CSO was better than to the cleaner polished CSO surface. A similar behavior was observed with Pt electrodes on the same type of CSO electrolytes. The explanation of the apparently positive effect of the surface impurities might be that it is difficult to sinter electrodes onto a smoothly polished electrolyte surface. In this case, a surface impurity layer may act as a sintering aid for the electrode/electrolyte interface and thereby improve the electrode adhesion. It is doubtful whether a beneficial effect of the impurities would have been observed if a more appropriate fabrication method had been used. 3.3 Metal electrodes Even though most researches in the area of metal electrodes seem to neglect the effect of the surface segregations on the ceramic electrolyte, it is clearly established that such silica glassy types of impurities are usually present, in particular, in case of model electrodes of Pt or Ni. It is less clear in the case of Ni–YSZ cermets with high purity and high surface area per unit volume. 3.3.1 Single-phase nickel and platinum electrodes There is pronounced discrepancy among the many pub- lished results on SOFC anode kinetics, [67–71] and this is valid for oxidation of hydrogen as well as of CO and of hydrocarbons. [72, 73] Therefore, model electrodes with well- defined geometry (Ni pattern and point electrodes) have been studied with the hope that this would give unambigu- ous results pointing to definite rate-limiting steps, but these 8 Materials for high temperature fuel cells resu