With increasing concerns on energy shortage and environment pollution over the last few decades, Solid Oxide Fuel Cells (SOFCs) have gained worldwide attention as a new means of power generation due to their coherent characteristics of high electrical efficiencies and low emissions. As the state-of-the-art anode materials, Ni/YSZ cermets exhibit excellent catalytic activity toward hydrogen oxidation reactions, but very poor redox stability with high susceptibility toward Ni coarsening during the long-term operation^[1]. In order to overcome these issues, a series of conductive oxides were investigated as alternative anodes. Some perovskite oxides such as La_1-3x/2Sr_xTiO_3-δ^[2], La_0.75Sr_0.25Cr_0.5Mn_0.5O_3-δ^[3] and Sr₂Fe_1.5Mo_0.5O_6-δ^[4] showed promising anode performance owing to their high thermal and chemical stability as well as high mixed ionic and electronic conductivities. Optimization of the chemical composition and the structural defects may allow for some improvements in the anode performance^[5,6,7], nevertheless, their catalytic activity was still much lower than the typical Ni-cermet anodes. Adding a secondary catalyst to the perovskite oxide by infiltration^[8,9] or vapor deposition^[10] was effective in improving the overall anode performance. It remains a challenge to precisely control their surface morphology with enhanced long-term stability at elevated temperatures.

In-situ exsolution was demonstrated as an elegant and effective approach to obtain nano-scale metal catalysts on the surface of perovskite oxides^{[11,12,13,14]}. Specifically, transitional metals such as Ni, Co and Pt were selected to partially substitute the B-sites of the perovskite oxides during the powder synthesis in air, which were then in-situ segregated from the lattice and precipitated as metal nanoparticulates in the reducing environments at high temperatures. With an ideal catalyst morphology, these exsolved metal nanoparticulates evenly distributed on the surface of the perovskite oxides. Note that the amounts and sizes of these metal nanoparticulates may significantly affect the anode performance^[15]. Moreover, their physical distribution may also change due to the continuous precipitation in the reducing environments at high temperatures^[16]. Therefore, it becomes crucial to control the ion segregation kinetics so as to optimize the dimension and density of the exsolved nanoparticulates, thereby enhancing their performance as the anode catalysts^[17].

Recently, non-stoichiometry^[12] and additional electric field^[18] were used to tailor the metal segregation process. Furthermore, the segregation driving force was analyzed by the distribution of relaxation time (DRT) calculations^[11,19-20], indicating that ion size mismatch and electrostatic interaction played important roles in the segregation process. Altering the dopant valence and the ion size may change the exsolution behaviour. In this work, lanthanum was introduced into the A-site to partially replace Sr²⁺ in Sr₂Fe_1.5Ni_0.1Mo_0.4O_6-δ. A series of samples, La_xSr_2-3x/2Fe_1.5Ni_0.1Mo_0.4O_6-δ (x=0-0.4), were synthesized in air. The chemical compositions were examined before and after reduction in H₂ by X-ray diffraction(XRD), with the surface morphology observed using scanning electron microscope (SEM). It was found that La doping could inhibit in-situ exsolution of Fe-Ni alloys and improve the stability of ceramic substrates. Symmetrical anode fuel cells and functioning fuel cells were fabricated by the impregnation method and tested to verify optimization of the anode performance for La_xSr_2-3x/2Fe_1.5Ni_0.1Mo_0.4O_6-δ at x= 0.3.

La_xSFNM (La_xSr_2-3x/2Fe_1.5Mo_0.4Ni_0.1O_6-δ, x=0, 0.1, 0.2, 0.3, 0.4) powders used in this study were synthesized by the standard solid-state reaction method. The starting materials were La₂O₃(>99%), SrCO₃ (>98%), Fe₂O₃ (>98%), MoO₃ (>99%) and NiO (>99%). La₂O₃ was fired at 800 ℃ for 12 h before being weighed, and SrCO₃ was dried at 150 ℃ to remove the adsorbed water in air. Similar to the synthesis of Sr₂Fe_1.5Mo_0.5O₆(SFM) powders^[21], the first step was to prepare Fe₂Mo₃O₁₂. Fe₂O₃ and MoO₃ oxides were mixed and thoroughly ground by ball milling in anhydrous ethanol at a stoichiometric ratio of 1:3 with 6wt% triethanolamine as the dispersant. The mixture was heat-treated at 750 ℃ for 12 h with the resulting phase pure Fe₂Mo₃O₁₂ oxide confirmed by XRD. Next, a stoichiometric ratio of La₂O₃, SrCO₃, Fe₂O₃, Fe₂Mo₃O₁₂ and NiO were similarly mixed and ball milled for 12 h in anhydrous ethanol. In order to ensure full decomposition of SrCO₃ to SrO and CO₂, the mixture was heated at 3 ℃/min up to 1000 ℃ and held at this temperature for 12 h. After cooling to room temperature, the powders were reground in the mortar, followed by the second heattreatment at 1200 ℃ for 12 h.

The crystalline structure of as-synthesized La_xSFNM powders was examined by XRD using Rigaku Ultima IV with monochromatic Cu Kα radiation. In order to analyze the ceramic surface morphology after in-situ dissolution of metal particles in the reducing environments, ceramic pellets were prepared by uniaxially dry-pressing La_xSFNM powders under 15 MPa, followed by sintering at 1400 ℃ for 5 h in air and reducing at 750 ℃ for 4 h in wet 10vol% H₂/N₂ (3vol% H₂O). SFNM and La_0.3SFNM samples were further reduced at 800 ℃ in pure hydrogen for 24 h. The surface topography was observed by SEM (Hitachi SU8220 and FEI Inspect F50), and the chemical compositions of the precipitated particles were analyzed by EDS.

Symmetrical anode fuel cells were prepared by the impregnation method, and La_0.9Sr_0.1Ga_0.8Mg_0.2O₃ (LSGM) was selected as the electrolyte due to its high oxide ionic conductivity. “Porous|dense|porous” LSGM scaffolds were obtained by laminating the tape-casted ceramic green tapes with rice starch as the fugitive material for the porous layers, as reported in our previous publication^[22]. The La_xSFNM anode catalysts were added by infiltrating an aqueous solution, containing La(NO₃)₃, Sr(NO₃)₂, Fe(NO₃)₃·9H₂O, Ni(NO₃)₂, (NH₄)₆Mo₇O₂₄·4H₂O and citric acid at an appropriate ratio, followed by calcinations at 850 ℃ for 30 min. Such a process was repeated until a catalyst loading of 30wt% was achieved after calcination at 1000 ℃ for 4 h to produce the perovskite oxides. The impedance data were measured on the symmetrical anode fuel cells in 97vol% H₂-3vol% H₂O at 100 sccm using the Electrochemical Workstation (ZAHNER IM6e, Germany) over the frequency range of 0.05 Hz to 100 kHz.

To fabricate the functioning fuel cells, SmBa_0.5Sr_0.5Co₂O_5+δ (SBSCO) was infiltrated as the cathode catalyst into the porous LSGM of the tri-layered scaffolds on one side, with the La_xSFNM anode catalysts infiltrated on the other side. SBSCO was selected as the cathode catalyst owing to its high electrical conductivity and excellent catalytic activity toward oxygen reduction reactions. The effective area of as-prepared fuel cells was 1.05 cm². Silver paste (DAD-87, Shanghai Research Institute of Synthetic Resins) was printed on the electrode surface to collect the electrical current. Single cells were sealed on the alumina tubes using the ceramic glue. The anode was reduced in humidified H₂ for 5 h before any electrochemical measurement.

The crystalline structure of as-synthesized powders was determined by room temperature XRD. Fig. 1(a) summarizes the XRD patterns of a series of La_xSFNM (x=0, 0.1, 0.2 0.3 and 0.4) oxides, which can be well indexed in single perovskite structure with cubic symmetry. Fig. 1(b) shows the XRD patterns of the thermally treated oxides in wet H₂ (3vol% H₂O) at 750 ℃, showing the formation of minor amounts of secondary FeNi₃ (#88-1715) phases at 2θ=44° and 51° for x=0-0.2. Fig. 1(b) also shows that diffraction peaks from FeNi₃ decreased with the La³⁺ dopant increasing, which became difficult to be detected at x=0.3 and 0.4. These results suggest that doping La³⁺ in the A-site seemingly inhibit the in-situ exsolution behavior of Sr₂Fe_1.5Ni_0.1Mo_0.4O_6-δ oxides and thereby enhance the structural stability of La_xSFNM oxides in the reducing environments.

To better understand the influence of lanthanum dopants on the exsolution behavior and the resulting morphology of nanoparticulates, a series of La_xSFNM (x=0, 0.1, 0.2, 0.3, 0.4) pellets were prepared at 1400 ℃ in air and then reduced in 10vol% H₂/N₂ at 750 ℃ for 4 h. Fig. 2 shows the surface morphologies of as-prepared and reduced perovskite oxides. The as-prepared SFNM sample shows clean and smooth grains without any visible secondary phases (Fig. 2(a)). Upon reduction, many nano-scale particles exsolved and homogeneously distributed on the oxide surface (Fig. 2(b-f)). EDS analysis confirms that these precipitated nanoparticulates exist largely as Fe-Ni alloys (Fig. 3), consistent with the XRD peaks of FeNi₃ in Fig.1(b). The diameter of Fe-Ni alloy nanoparticles is 40-60 nm. Fig. 2 also shows that the density of exsolved nanoparticulates on the oxide surface decreases significantly with the La³⁺ dopant increasing, supporting the conclusion from the XRD analysis that doping La³⁺ in the A-site helps to inhibit the exsolution of metal nanoparticulates and thus improve the structural stability of the pristine perovskite oxides.

The polarization resistance provides a quantitative measurement of the catalytic activity of La_xSFNM oxides toward electro-oxidation of hydrogen, which can be readily obtained from impedance measurements on symmetrical anode fuel cells, i.e., Nano-La_xSFNM@LSGM|LSGM|Nano- La_xSFNM@LSGM. Fig. 4(a) summarizes Nyquist plots of impedance spectra measured at 750 ℃ in a homogeneous environment of 97vol% H₂-3vol% H₂O for symmetrical anode fuel cells with varied La³⁺ dopants, where the ohmic resistances due to the electrolytes and the collecting wires were removed and the polarization resistances were divided by two to account for contributions of two symmetrical anodes. The anode specific polarization resistances (R_{P, A}) are taken as the overall widths of depressed arcs in Fig. 4(a). The anode polarization resistance is R_P,A=0.21 W∙cm² for SFNM, decreases gradually with the La³⁺ dopant increasing, and attains a minimal value of 0.16 W∙cm² at x=0.3. Further increasing the La³⁺ dopant yields a slightly increased R_P,A value (0.17 Ω∙cm²) for La_0.4SFNM cell. The Bode plots of impedance data in Fig. 4(b) show that the response difference mainly occurs over the frequency range of 10-1000 Hz. Analysis of Distribution of Relaxation Time (DRT) was performed to effectively separate the multiple time constants^[23,24], with the results summarized in Fig. 4(c). Notably, the impedance spetra consist of three different processes (denoted as R_H, R_M, R_L) in the frequency range of 100-1000, 10-100 and 0.1-10 Hz, respectively. Similar to the SFNM anode, the high-frequency resistance R_H relates to charge transfer across the electrolyte/electode interface, whereas the medium- and low-frequency resistances (R_M and R_L) are associated with the surface reactions^[13,25]. Increasing the La³⁺ dopant yields a pronounced decrease in R_M with R_H and R_L almost unchanged, indicating that doping La³⁺ in the A-site effectively promotes the surface reaction on the perovskite oxides. Since the SEM observations show inhibited exsolution of FeNi₃ nanoparticulates and increased stability of the perovskite oxides with more La³⁺ dopants, the optimal catalytic activity at x=0.3 may be associated with the synergy between the exsolved catalysts and the supporting perovskite oxides.

The anode behavior was further examined on the functioning fuel cells as fabricated from tri-layer structures of “porous|dense|porous” LSGM. The dense LSGM electrolytes are typically ~22 mm thick (Fig. 5(a)). SBSCO and La_0.3SFNM were impregnated into the two porous LSGM scaffolds, respectively. High magnification view of the anode after the fuel cell measurements show the formation of in situ exsolved nanoparticulates on the surface of La_0.3SFNM oxides (Fig. 5(b)). Fig. 6(a) shows the representative polarization curves of the cell voltages (V) and power densities (P) as functions of current densities (J) for the single cells with La_0.3Sr_1.55Fe_1.5Ni_0.1Mo_0.4O_6-δ as anode catalyst. The maximum power densities are 1.26, 1.13, 0.90, 0.74 and 0.52 W∙cm^-2 at 750, 700, 650, 600 and 550 ℃, respectively. Cells with varied La³⁺ doped anodes were also prepared and measured under similar conditions, with the results compared in Fig. 6(b). The cells (x=0.3) exhibit the highest power densities at all temperatures. The present values are also much higher than previously reported power densities, e.g., 0.59 W∙cm^-2 for SFNiM^[13] and 0.85 W∙cm^-2 for SFM^[26] at 750 ℃.

Long-term stability of nano-scale FeNi₃ catalysts as shown in Fig. 2 is very questionable due to their high susceptibility to continuous agglomeration and coarsening at elevated temperatures. Although extended measurements are required for better evaluation of the long- term stability of these nano-scale catalysts, some of the cells were operated for a few hundred hours. Fig. 7 shows a representative plot of the cell voltage as a function of time. The single cell was run at 0.5 A∙cm^-2 under 550 ℃ for 120 h, followed by another 80 h operation at 1.0 A∙cm^-2, 650 ℃, showing negligible decay in the power output at both temperatures. These results demonstrate great promise of La_0.3Sr_1.55Fe_1.5Mo_0.4Ni_0.1O_6-δ as efficient catalyst for electro-oxidation of hydrogen fuels.

In summary, a series of La_xSr_2-3x/2Fe_1.5Ni_0.1Mo_0.4O_6-δ (x=0-0.4) oxides were synthesized, and their structural stability in reducing environments was examined. Both XRD patterns and SEM observations confirm the exsolution of nano-scale FeNi₃ particles on the oxide surface after the thermal treatment in H₂. Increasing the La³⁺ dopant enhances the structural stability of these oxides with inhibited exsolution of FeNi₃ nanoparticulates. Synergy between the exsolved FeNi₃ nanoparticulates and the supporting oxides yields the highest catalytic activities toward hydrogen oxidation reactions with the lowest polarization resistances at x=0.3, e.g., 0.16 W∙cm² at 750 ℃. Thin LSGM electrolyte SOFCs with nano- La_0.3Sr_1.55Fe_1.5Mo_0.4Ni_0.1O_6-δ@LSGM anodes and nano- SBSCO@LSGM cathodes produces peak power densities of 1.26, 0.90 and 0.52 W∙cm^-2 at 750, 650 and 550 ℃, respectively.