La0.6Sr0.4Co0.2Fe0.8O3 Perovskite: A Stable Anode Catalyst for Direct Methane Solid Oxide Fuel Cells

Direct methane solid oxide fuel cells, operated by supplying methane to a Ni/YSZ anode, suffer from degradation via accumulation of carbon deposits on the Ni surface. Coating a 40 μm thin film of La0.6Sr0.4Co0.2Fe0.8O3 (LSCF) perovskite on the Ni/YSZ anode surface decreased the amount of carbon deposits, slowing down the degradation rate. The improvement in anode durability could be related to the oxidation activity of LSCF which facilitates oxidation of CH4 and carbon deposits. Analysis of the crystalline structure of LSCF revealed that LSCF was stable in the reducing anode environment under H2 and CH4 flow at 750 °C and retained its perovskite structure throughout the 475 h long-term stability test.


Introduction
Perovskites are mixed metal oxides with the general formula of ABO 3 , in which A and B are two different metal cations. Perovskites have attracted vast interest in the catalysis area owing to the unique crystal structure, oxygen vacancy, redox and acid-base properties, and thermal stability. La 1−x Sr x Co 1−y Fe y O 3 (lanthanum strontium cobalt iron oxide, LSCF) is a perovskite, which has been synthesized by incorporation of strontium and iron cations into a lanthanum cobalt oxide crystal structure.
Different methods have been used to prepare LSCF perovskite including Pechini [14][15][16], glycine nitrate combustion [17], solid state reaction [18], and spray pyrolysis [19]. The Pechini method, a solution technique, has been shown to produce high purity and homogeneous perovskite phase with high surface area at low synthesis temperature without intermediate grinding.
This approach immobilizes the metal ion precursors in a polyester network which is formed by citric acid and ethylene glycol [20,21]. We employed the Pechini method to synthesize LSCF perovskite for this study.
Fuel cells offer a clean technology with high efficiencies for power generation from direct conversion of chemical energy to electrical energy using fuel and an oxidant. A solid oxide fuel cell consists of an anode and a cathode separated by a solid electrolyte. The anode is in contact with the fuel and the cathode is exposed to air. Figure 1 illustrates the operating principle of solid oxide fuel cell along with the reactions in presence of H 2 or CH 4 fuel. Oxygen from air adsorbs on the cathode surface becoming oxygen ion and the oxygen ion passes through the ion conducting electrolyte to meet the fuel on the anode side. Electrochemical oxidation of fuel on the anode/electrolyte interface generates products and releases electrons. The transfer of electrons from the anode to the cathode through an external circuit completes the electrochemical cycle [22]. Among a wide variety of fuels that can be used in fuel cells, natural gas has generated significant interest because of its abundance in the United States. Therefore, development of a stable anode catalyst for direct utilization of methane as a major component of natural gas has been considered. Nickel in a conventional Ni/YSZ anode shows high catalytic activity for C-H bond dissociation and C-C bond formation in hydrocarbons. Formation of C-C bond, known as coking, deactivates the Ni/YSZ anode catalyst through the loss of active sites and microstructural damage [23,24]. A number of approaches were investigated to suppress the coking and delay the anode deactivation including the addition of steam [25], dilution of hydrocarbon fuel with CO 2 [26,27] Air materials such as bimetallic anodes [28], and applying a non-coking barrier layer onto the anode surface [29,30]. This study investigated the effect of LSCF perovskite coating on the Ni/YSZ anode surface on long-term stability of the anode in direct CH 4 SOFC. LSCF, which has been used frequently as a cathode [31][32][33][34][35], for the first time is used as a promoter in direct CH 4 Ni/YSZ anode catalyst. LSCF was selected because of its oxidation activity and mixed conductivity as well as its moderate cost. The cost of LSCF powder is about three times greater than that of conventional Ni/YSZ anode powder.
Although perovskite materials with mixed electronic-ionic conductivity are widely used as cathodes for solid oxide fuel cells, the catalytic and electrocatalytic oxidation activity of perovskites enable the reversible oxygen redox reaction Equation (1) to occur on anodes at the operating temperature of SOFC.
The occurrence of Equation (1) in reverse direction for oxidation of CH 4 and carbon deposits could be one of the key reasons for LSCF to slow down anode degradation. Activity and long-term stability of a Ni/YSZ anode and a Ni/YSZ anode promoted with LSCF (LSCF-Ni/YSZ) was studied in CH 4 by continuous monitoring of electrochemical performances. Figure 2 shows voltage-current (V-I) curves and Nyquist plots of impedance spectra for the Ni/YSZ anode fuel cell recorded in He/H 2 (100 sccm, 50 vol% H 2 ) after 18 h of reduction and in He/CH 4 (25 vol% CH 4 ) after 1 h of switching fuels. The Ni/YSZ anode fuel cell generated a maximum current density of 295 mA/cm 2 in H 2 and CH 4 . The relatively low current density of the fuel cell resulted from high resistance of the thick anode and electrolyte layers. Fuel cell resistance is a function of the thickness of the anode, electrolyte and cathode [36]. The ohmic resistances (the intercept of the high frequency impedance arc with the X-axis) of the Ni/YSZ anode in H 2 and CH 4 are the same. Ohmic resistance resulted from resistance against ion transfer through electrolyte and electrodes, and electron transfer through electrodes and current collectors. Polarization resistances were obtained by subtraction of ohmic contribution from total resistance (the X value at the lowest frequency of impedance arc). The polarization resistances were 7.2 and 6.1 Ω•cm 2 in H 2 and CH 4 , respectively. The absolute value of cathode polarization was assumed to remain constant. Therefore, the change in polarization resistance was ascribed to the change in anode polarization after switching fuels. The change in anode polarization was shown by the ratio of polarization resistance toward oxidation of CH 4 to that of H 2 , which is 0.8 in the current study, compared to 2.5 in our previous work [37]. Values reported for polarization resistance ratios of CH 4 to H 2 vary from 1.8 to 13.6 depending on operating conditions and fuel cell specifications [38][39][40]. The greater polarization resistance of the Ni/YSZ anode fuel cell in H 2 is consistent with the sharper slope of the corresponding V-I curve at low current densities, where voltage loss is dominated by activation polarization [41]. Lower polarization resistance in CH 4 than H 2 could be related to carbon formation from CH 4 dissociation Equation (2).

Results and Discussion
Small amounts of carbon deposits may connect some isolated Ni particles, converting them to active sites for electrochemical reactions and allowing the produced electron to transport to the external circuit [42].  Figure 3 shows V-I curves and Nyquist plots of impedance spectra for the LSCF-Ni/YSZ anode fuel cell in He/H 2 (100 sccm, 50 vol% H 2 ) after 16 h of reduction and in He/CH 4 (25 vol% CH 4 ) after 3 h of switching fuels. Fuel cell operation in H 2 and CH 4 fuel generated a maximum current density of 340 and 355 mA/cm 2 , respectively. The ohmic resistance in CH 4 is lower than that in H 2 . This could be due to the gradual penetration of Ag, which has been used for current collection, into the cathode microstructure leading to an increase in electronic conductivity [43]. The cross-sectional EDX mapping indicated the presence of Ag across the cathode (result not shown). Polarization resistance of the LSCF-Ni/YSZ anode fuel cell in H 2 was 3 Ω•cm 2 , which is greater than the 2.3 Ω•cm 2 in CH 4 . The greater polarization resistance in H 2 was also observed for the Ni/YSZ anode fuel cell.  Figure 4 shows maximum power density of the Ni/YSZ and LSCF-Ni/YSZ anode fuel cells recorded during the long-term stability test in CH 4 . Maximum power density decreased with respect to time for both fuel cells. The rate of decrease in maximum power density is much slower for the LSCF-Ni/YSZ anode than the Ni/YSZ anode fuel cell. The addition of LSCF coating on the Ni/YSZ anode increased the operation life of the direct CH 4 fuel cell to 475 h. Maximum power density profile of the LSCF-Ni/YSZ anode exhibits fluctuation with respect to time. This fluctuation could be attributed to repetitive accumulation and oxidation of carbon deposits, in which oxidation of carbon deposits was catalyzed by LSCF perovskite.   4 . Maximum power density decreased to 22% of its initial value after 38 h. The slopes of the V-I curves at low current densities were decreased after 9 h of operation in CH 4 , matching with the lower polarization resistances in the corresponding impedance spectra. The decrease in polarization resistance can be explained by deposition of moderate levels of carbon on the anode, which may connect isolated metal particles and convert them to active sites for electrochemical reactions [42]. A gradual decrease in open circuit voltage (OCV) from theoretical OCV was observed after 2 h. A significant decrease in OCV (from 0.8 to 0.6 V) was accompanied by a decrease in current density, indicating initiation of a crack in the fuel cell after about 13 h. Oxygen leakage into the anode chamber from the crack would cause a decrease in OCV and a significant increase in ohmic resistance of the fuel cell, shown in the corresponding impedance spectra in Figure 5a. The cell cracking increased ohmic resistance by weakening contacts between electrodes and current collectors and interrupting the current collection pathway [39]. Further exposure to CH 4  anode. Current density gradually decreased to 33% of its initial value over 475 h of operation in CH 4 . Impedance spectra of the LSCF-Ni/YSZ fuel cell recorded at OCV during the long-term stability test did not show a significant increase or decrease, which is indicative of a stable anode that did not undergo severe carbon deposition and structural changes. Noises in impedance spectra that appeared after 440 h operation are likely due to partial detachment of the current collectors from the electrode surfaces. Fracture of fuel cells by deposition and diffusion of high levels of carbon into the bulk of the Ni has long been recognized as a key problem of direct methane fuel cells with Ni-based anodes [39,[44][45][46]. Examination of the scanning electron microscopy (SEM) image and EDX mapping of the fractured Ni/YSZ anode in Figure 6 revealed that the amount of carbon deposit varied from a C/Ni weight ratio of 0.1 at 5 mm from the center Figure 6a to 0.4 near the center Figure 6b and 0.2 at the center Figure 6c, where the inlet CH 4 fuel was directed toward the anode. Figure 6b also shows that the majority of deposited carbon is located on the Ni surface. Previous studies on types of carbon produced on anodes during exposure to hydrocarbon fuels at temperature range of 625-900 °C showed that carbon was deposited in graphitic and disordered form [37,[47][48][49]. Graphitic carbon has been suggested to be responsible for anode degradation by dissolution into the bulk of the Ni/YSZ cermet resulting in volume expansion and breakage of the Ni-Ni network [37,50]. The most obvious effect of LSCF is to delay cracking of the fuel cell from 38 h to 475 h where significant drops in OCV occurred. The SEM image and EDX mapping of the cracked LSCF-Ni/YSZ anode in Figure 7 revealed that deposited carbon was sparse on the Ni particle surface with a C/Ni weight ratio of 0.1, whereas deposited carbon was not observed on LSCF. The absence of carbon deposits on LSCF could be explained by the scheme of the LSCF-Ni/YSZ anode cross-section in Figure 8. This scheme shows that LSCF, a mixed conductor, catalyzes electrocatalytic oxidation of carbon or CH 4 with O 2− anions transported from YSZ and releases electrons produced during reaction to the Ni connected particles. Distribution of LSCF in the LSCF-Ni/YSZ anode and its effect on fuel cell stability and performance will be further studied.
The ability of LSCF to oxidize carbon should slow down degradation of the Ni/YSZ anode. Oxidation of the deposited carbon from CH 4 by LSCF has been viewed as a self de-coking process by Huang et al. [51,52] who suggested that the lattice oxygen of LSCF was responsible for the oxidation. Oxidation by LSCF would produce oxygen vacancies in the LSCF lattice, which may be filled with O 2− anions coming from YSZ in the anode. These oxygen extraction (i.e., LSCF reduction) and refilling (i.e., LSCF re-oxidation) processes are kinetic controlled and could lead to oscillation of electric current and CO/CO 2 product formation [51]. The formation rate of CO/CO 2 oxidation products strongly depends on oxidation activity of the perovskite catalyst. Oxidation activity of perovskites has been related to their reducibility. Highly reducible perovskites such as LaCoO 3 have been shown to be more active than less reducible perovskites such as LaCrO 3 [13,53]. LaCoO 3 has been found to be one of the most stable structures among the Co-based perovskites in reducing environments [54]. LaCoO 3 also showed high propensity toward re-oxidization of its reduced forms back to its initial structure [55]. Incorporation of strontium into LaCoO 3 could enhance its reducibility, leading to high oxidation activity [56][57][58]. Partial substitution of La 3+ with Sr 2+ , a lower oxidation state cation, would convert Co 3+ to Co 4+ and produce oxygen vacancies at low O 2 partial pressure. This partial substitution could decrease the perovskite stability in a reducing environment [59], but increase the oxygen transport capability and oxidation catalytic activity. Oxidation catalytic activities of the most active perovskites, La 1−x Sr x BO 3 (B: Mn, Co, Fe; 0 < x < 0.4), toward CH 4 oxidation are comparable to that of Pt/Al 2 O 3 catalyst [60]. It has also been found that LSCF was able to exhibit high catalytic activity toward CH 4 coupling reaction; however, it can exhibit electrocatalytic activity only for complete oxidation of CH 4 [61].
Comparison of XRD patterns of LSCF perovskite before and after the long-term stability test in Figure 9a,b showed that both X-ray diffractometer (XRD) patterns of LSCF matched well with the characteristic peaks of LSCF perovskite phase [62]. It is unclear whether LSCF had decomposed to separate phases and then re-oxidized with O 2− anions coming from the cathode side during the stability test or whether it was re-oxidized by exposure to ambient environment. Figure 9b also shows Ni, NiO and YSZ crystalline phases from the LSCF-Ni/YSZ anode along with a number of minor peaks that could not be matched with any known phases. Crystallite size of LSCF and YSZ was increased after the long-term stability test from 16 to 25 nm and from 38 to 46 nm, respectively, as determined by the Scherrer equation. Increase in crystallite size of LSCF on the Ni/YSZ anode due to sintering of perovskite crystallites suggests possible occurrence of a decomposition/re-oxidation process for LSCF. Co-and Fe-based perovskites can be decomposed to separate metal and metal oxide phases in reducing atmospheres [55,63,64]. Re-oxidation of separate metal/metal oxide phases has been observed to reproduce perovskites with large crystallite size [65]. Sintering of particles would decrease catalytic activity by lowering the number of surface sites. Sintering was inhibited by re-oxidation of reduced perovskites at low oxidation rates with very dilute oxygen flow [66,67].
The stability of perovskites including LSCF, especially retaining initial catalytic/electrocatalytic activity and electronic-ionic conductivity, during long-term operation in CH 4 has to be considered for the use of these materials as a part of the SOFC anode. Further development will require optimization of electrocatalytic oxidation activity and sintering resistance through fine-tuning perovskite compositions (i.e., adjusting the amounts of dopants to A and B metal cations) and/or searching for proper operating conditions. Studies on stability of LSCF in reducing environments of CH 4 and CO 2 showed that presence of O 2 in the feed gas allowed LSCF to maintain its perovskite structure with the desired electronic-ionic conductivity [68,69]. Perovskites were found to demonstrate reversible redox behavior at temperatures below the sintering temperature of their decomposed phases [13,55,65]. Thus, periodic addition of low concentration of air into CH 4 fuel as well as a systematic study of operating temperature could be the options to further stabilize the perovskite on the anode. 20

Fuel Cell Fabrication
Anode supported fuel cells were prepared by co-tape casting of anode support, anode interlayer and electrolyte slips, and screen-printing of cathode interlayer and cathode current collector layer. The slips of NiO/8YSZ (8 mol% Y 2 O 3 stabilized ZrO 2 , Tosoh, Tokyo, Japan) 65:35 wt% anode support, NiO/8YSZ 63:37 wt% anode interlayer, and 8YSZ electrolyte were produced by dispersing metal oxide powders in ethanol solvent, introducing binder and dispersant, and ball milling the resulted mixture for 24 h. The anode support slip also contained microcrystalline cellulose pore former ( 2 and sintered at 1250 and 1100 °C, respectively. The thickness of layers for fabricated fuel cells, determined from cross-sectional SEM micrograph (result not shown), was 1600 µm for anode support, 30 µm for anode interlayer, 25 µm for electrolyte, 40 µm for cathode interlayer, and 20 µm for cathode current collector layer. The LSCF promoted Ni/YSZ anode fuel cell was prepared by reducing Ni/YSZ anode fuel cell in He/H 2 (100 sccm, 50 vol% H 2 ) at 750 °C for 16 h. Screen-printing was selected as a convenient technique to coat LSCF on the surface of anode support. Two 40 µm thick strips of LSCF with areas of 0.75 cm 2 were screen-printed on the Ni/YSZ surface leaving an uncoated Ni/YSZ surface in between, followed by firing at 1050 °C for 1 h. This fuel cell is labeled as LSCF-Ni/YSZ anode fuel cell versus unmodified fuel cell which is labeled as Ni/YSZ anode fuel cell.

Fuel Cell Testing and Characterization
An anode supported fuel cell was placed and sealed on a metal tube serving as the anode chamber and anode current collector as illustrated in Figure 11. A silver strip cathode current collector was attached to the cathode surface. A thin layer of silver paste (C8728 Heraeus, West Conshohocken, PA, USA) was applied between the electrodes and current collectors to improve current collection. The fuel cell was heated to 750 °C and reduced in a flow of He/H 2 (100 sccm, 50 vol% H 2 ). The flow was then switched to He/CH 4 (100 sccm, 25 vol% CH 4 ) for the long-term stability test at 750 °C. The Ni/YSZ and LSCF-Ni/YSZ anode fuel cells were operating under low current densities of 25-50 mA/cm 2 throughout the long-term stability test. Voltage-current curves and impedance spectra were recorded at different time intervals with a potentiostat and a frequency response analyzer (1470E and 1400 Solartron CellTest System, Solartron Analytical, Oak Ridge, TN, USA). Composition and microstructure of the tested fuel cells were analyzed using scanning electron microscopy (SEM, TM-3000 Hitachi, Tokyo, Japan) and energy-dispersive X-ray (EDX, Bruker, Ewing, NJ, USA). The LSCF-Ni/YSZ anode was characterized with XRD crystallography after the stability test in CH 4 .

Conclusions
The effect of LSCF perovskite coating on the durability of the conventional Ni/YSZ anode was studied by performing the long-term stability test of the Ni/YSZ and LSCF-Ni/YSZ anode in direct CH 4 utilization. Severe carbon deposition resulted in degradation and fracture of the Ni/YSZ anode after 38 h of exposure to CH 4 . The LSCF-Ni/YSZ anode on the other hand exhibited high stability leading to long-term operation of 475 h in CH 4 . Compositional and microstructural analysis of the fuel cell anodes after the test indicated lower carbon deposition on the LSCF-Ni/YSZ anode compared to that of Ni/YSZ. LSCF retained its perovskite structure after the long-term stability test in CH 4 . The results of this study suggest that LSCF is a promising promoter catalyst for stable direct CH 4 fuel cell anodes.