Infiltrated Ba0.5Sr0.5Co0.8Fe0.2O3-δ-Based Electrodes as Anodes in Solid Oxide Electrolysis Cells

In the last decades, several works have been carried out on solid oxide fuel cell (SOFC) and solid oxide electrolysis cell (SOEC) technologies, as they are powerful and efficient devices for energy conversion and electrochemical storage. By increasing use of renewable sources, a discontinuous amount of electricity is indeed released, and reliable storage systems represent the key feature in such a future energy scenario. In this context, systems based on reversible solid oxide cells (rSOCs) are gaining increasing attention. An rSOC is an electrochemical device that can operate sequentially between discharging (SOFC mode) and charging (SOEC mode); then, it is essential the electrodes are able to guarantee high catalytic activity, both in oxidation and reduction conditions. Ba0.5Sr0.5Co0.8Fe0.2O3-δ (BSCF) has been widely recognized as one of the most promising electrode catalysts for the oxygen reduction reaction (ORR) in SOFC technology because of its astonishing content of oxygen vacancies, even at room temperature. The purpose of this study is the development of BSCF to be used as anode material in electrolysis mode, maintaining enhanced energy and power density. Impregnation with a La0.8Sr0.2MnO3 (LSM) discrete nanolayer is applied to pursue structural stability, resulting in a long lifetime reliability. Impedance spectroscopy measurements under anodic overpotential conditions are run to test BSCF and LSM-BSCF activity as the electrode in oxidation mode. The observed results suggest that BSCF is a very promising candidate as an oxygen electrode in rSOC systems.


Introduction
The increasing penetration of renewable energy sources in the power market, guided by new energy policies to address the climate change, poses new challenges that need to be tackled [1]. The intermittent nature of wind and solar power requires the development of large-scale energy storage as a key to improve the flexibility of the electric grid. Electrical energy storage (EES) is envisioned as the key factor to boost the development of advanced grid-energy management systems [2,3]. In this context, systems based on reversible solid oxide cells (rSOCs) are gaining increasing attention and interest. An rSOC is an electrochemical device that can operate in both power-producing (solid oxide

Materials and Methods
The shape of the electrolyte support must respect appropriate geometric criteria. Namely, the distance between the reference electrode (RE) and working electrode (WE) should be at least three-times the electrolyte thickness in order to avoid artefact formation in impedance experimental spectra [17,32]. According to the relative shrinkage value obtained from previous sintering experience (22%), a weight of Ce 0.8 Sm 0.2 O 2-δ (SDC20) electrolyte powder equal to 2.4 g was uniaxially pressed at 37 MPa to obtain electrolyte supports. The green pellets were sintered at 1500 • C for 5 h, obtaining sintered discs of 20 mm in diameter and 1.1 mm in thickness. Before the electrode deposition, the SDC discs were sanded down (P320 SiC paper) for obtaining a rough surface reliable for an easy adhesion of the BSCF electrode to the electrolyte layer.
Energies 2020, 13, 3659 3 of 10 A mixture of graphite (KS6, TIMREX ® , TIMCAL Graphite & Carbon, Bodio, Switzerland) and BSCF (Treibacher Industry AG, Althofen, Austria) powders, according to the 60/40 v/v% composition, was ball-milled for 40 h at 40 rpm in distilled water at room temperature (R.T.), and zirconia balls (Tosoh) were employed as mixing bodies. This volume concentration was chosen in order to have a proper porosity resulting in an easy evacuation of the oxygen gas phase. After the mixing, a freeze-drying procedure was applied (24 h, at −52 • C and 22 Pa). Finally, after sieving, the BSCF powder was ready to be deposited. Alpha-terpineol (Sigma-Aldrich, >96%) was added to the powder (BSCF graphite) in a mortar to obtain a mixture suitable for deposition. By applying an appropriate tape mask, a WE and a counter electrode (CE) were slurry-coated on the sides of the SDC pellets. An RE, used for the three-electrode impedance measurements, was applied around the WE. The geometry is described in detail elsewhere [33]. The electrodes were co-sintered at 1100 • C for 1 h. After the sintering process, the geometric area for both the WE and the CE was 0.28 cm 2 .
For the infiltration of a porous BSCF electrode backbone, an aqueous solution of hydrated nitrates, namely La(NO 3 )·2H 2 O, Sr(NO 3 ) 2 and Mn(NO 3 ) 3 ·xH 2 O, with x = 4 or 6, was prepared. In order to estimate the actual cation concentration in precursors, gravimetric titration of the above-mentioned nitrates was carried out. About 5 g of each precursor was weighted out in a clean and dry Pt crucible and heated up to 900 • C for 2 h, in order to allow for the formation of oxides, i.e., La 2 O 3 , SrO and Mn 3 O 4 .
The impregnating solution was prepared by adding stoichiometric amounts of hydrated nitrates, reported in Table 1, to 50 mL of water in order to obtain a concentration of 0.6 mol L −1 . Glycine and polyvinylpyrrolidone (PVP) were added, respectively, as a chelating agent and as a surfactant (see Table 1), and the solution was heated at 200 • C under stirring for 10 min to completely dissolve the components. Impregnation was carried out by using different amounts of the nitrate solution, in the range 1.5-6 µL, either as-prepared or diluted. The samples tested in this work were impregnated with 1.5 µL of a solution with a concentration of 0.06 mol L −1 . In order to eliminate air in the pores, infiltrated samples were placed under a vacuum at 94 Pa for 2 min. The infiltrated cells (both in the WE and the CE) were finally heated at 800 • C for 3 h, with a 1 h dwelling step at 300 • C.
LSM-BSCF/SDC/LSM-BSCF half-cells (with an RE) were tested inside a ProboStat TM (NorECs Norwegian Electro Ceramics AS, Oslo, Norway) setup system. Two Pt nets were placed on the surface of each electrode as current collectors. Electrochemical impedance spectroscopy (EIS) measurements were carried out through a Solartron Analytical potentiostat (SI 1286) coupled with a frequency response analyzer (SI 1255) and the ZPLOT software (version 14.1.3, 45485, Scribner Associates Inc., Southern Pines, NC, USA) Impedance tests were run between 475 and 650 • C in a frequency range 10 4 -10 −2 Hz. An amplitude varying between 5 and 20 mV was applied, according to the system response. Measurements were performed at open circuit voltage (OCV) conditions as well as under anodic overpotential load, up to +0.2 V. An N 2 /O 2 = 80/20 v/v% mixture was fed at both sides of the cell, corresponding to a flow of 40 NmL min −1 and 10 NmL min −1 of N 2 and O 2 , respectively. To test the system durability, a 48 h long-lasting test was run under an anodic potential of +0.150 V, collecting an OCV impedance test every 6 h. Experimental impedance spectra were analyzed by ZView-Impedance Software (version 14.1.3, 45485, Scribner Associates Inc., Southern Pines, NC, USA).
To investigate the morphology and the adhesion between the layers, observations were taken by scanning electron microscopy (SEM, Phenom Pro-X) equipped with Energy-dispersive X-ray spectroscopy (EDXS, Bruker) on the as-sintered and tested samples in order to check the distribution of LSM on the top and cross the section of the BSCF electrode.

Microstructural
In Figure 1, the surface SEM images of the pristine BSCF and of the 1.5 µL, 0.06 mol L −1 LSM-infiltrated electrode are reported, before and after the electrochemical characterization. In the following, the LSM-infiltrated BSCF will be named just LSM-BSCF electrode.

Microstructural 139
In Figure 1, the surface SEM images of the pristine BSCF and of the 1.5 μL, 0.06 mol L -1 LSM-140 infiltrated electrode are reported, before and after the electrochemical characterization. In the 141 following, the LSM-infiltrated BSCF will be named just LSM-BSCF electrode. 142

145
After the sintering process at 1100 °C, the electrodes showed a homogeneous structure, with 146 open porosity and an appreciable coarsening of the grains, resulting in the formation of a good 147 interconnected ceramic network (see Figure 1). 148 The comparison of images obtained before (Figure 1a,b) and after the electrochemical testing 149 (Figure 1c,d) shows that the porosity of structures was not affected by testing. By contrast, 150 transformations were evident at the surface of both electrodes after the electrochemical investigation. 151 On the surface of the blank BSCF electrode, a dot-shaped minor phase appeared ( Figure 1c); likewise, 152 in previous findings [31], it was supposed that this phase could result from the aggregation, during 153 testing time between 475 and 650 °C, of the distributed phase already present on the BSCF starting 154 electrode surface (Figure 1a, red arrow and oval). Concerning the infiltrated electrode, the LSM was 155 distributed as a continuous and homogeneous layer at the beginning (Figure 1b), while, after the 156 testing, several extended defects with sharp edges and micropores appeared ( Figure 1d). 157 Figure 2 shows an SEM surface image of the as-prepared LSM-infiltrated layer and the 158 corresponding atomic concentration obtained through a map elemental analysis. Further EDXS 159 observations performed through the cross-section of the infiltrated electrode confirmed the 160 penetration of the LSM layer up to the electrode/electrolyte interface. 161 After the sintering process at 1100 • C, the electrodes showed a homogeneous structure, with open porosity and an appreciable coarsening of the grains, resulting in the formation of a good interconnected ceramic network (see Figure 1).
The comparison of images obtained before (Figure 1a,b) and after the electrochemical testing ( Figure 1c,d) shows that the porosity of structures was not affected by testing. By contrast, transformations were evident at the surface of both electrodes after the electrochemical investigation. On the surface of the blank BSCF electrode, a dot-shaped minor phase appeared ( Figure 1c); likewise, in previous findings [31], it was supposed that this phase could result from the aggregation, during testing time between 475 and 650 • C, of the distributed phase already present on the BSCF starting electrode surface (Figure 1a, red arrow and oval). Concerning the infiltrated electrode, the LSM was distributed as a continuous and homogeneous layer at the beginning (Figure 1b), while, after the testing, several extended defects with sharp edges and micropores appeared (Figure 1d). Figure 2 shows an SEM surface image of the as-prepared LSM-infiltrated layer and the corresponding atomic concentration obtained through a map elemental analysis. Further EDXS Energies 2020, 13, 3659 5 of 10 observations performed through the cross-section of the infiltrated electrode confirmed the penetration of the LSM layer up to the electrode/electrolyte interface.    Figure 3 reports an optical image (Figure 3a) of the partial side of SDC electrolyte with the slurry-coated WE (in the center) and RE (ring-shaped), while the geometry features of the three-electrode configuration are reported in Figure 3b, following the indications of [32,33], as aforementioned. The measurements were performed using an RE in order to have accurate control of the WE potential, especially when an overpotential was applied.    Impedance spectra were modeled through an equivalent circuit in order to obtain the different parameters, such as polarization resistance. The equivalent circuit consisted of a resistance (R e ) in First, impedance measurements were carried out at OCV, then tests were run by applying an anodic overpotential, up to +0.2 V.

Electrochemical
In Figure 4, impedance spectra of blank BSCF (Figure 4a) and LSM-BSCF (Figure 4b) are shown. It is evident that the values of polarization resistance (R p ), extracted as the difference between the low and high frequency intercepts with the real axis, are larger for the infiltrated electrodes than for pristine BSCF. At OCV and 650 • C, the value of the area specific resistance (ASR), as obtained by modeling, was 0.121 Ω cm 2 for the blank electrode and 0.160 Ω cm 2 for the infiltrated one. These results agree with other works, dealing with both blank and LSM-infiltrated electrodes [34][35][36]. The polarization resistance values obtained for the blank BSCF and LSM-BSCF infiltrated electrode were in contrast with previous findings obtained by the authors on LSM-infiltrated BSCF electrodes. In that case [31], the LSM infiltration had a positive effect in terms of R p at OCV. The reason for this discrepancy is likely due to the different morphology of the LSM-infiltrated layer. In the previous work, we dealt with a nano-sized discrete layer, resulting in an increased electrode-gas surface exchange area and consequent improved electrode activity; in this study, a continuous LSM layer was obtained, whose exchange properties with the gas phase are very different from those of the nano-distributed one. First, impedance measurements were carried out at OCV, then tests were run by applying an 180 anodic overpotential, up to +0.2 V. 181 In Figure 4, impedance spectra of blank BSCF (Figure 4a) and LSM-BSCF (Figure 4b) are shown. 182 It is evident that the values of polarization resistance (Rp), extracted as the difference between the low 183 and high frequency intercepts with the real axis, are larger for the infiltrated electrodes than for 184 pristine BSCF. At OCV and 650 °C, the value of the area specific resistance (ASR), as obtained by 185 modeling, was 0.121 Ω cm 2 for the blank electrode and 0.160 Ω cm 2 for the infiltrated one. These 186 results agree with other works, dealing with both blank and LSM-infiltrated electrodes [34][35][36]. The 187 polarization resistance values obtained for the blank BSCF and LSM-BSCF infiltrated electrode were 188 in contrast with previous findings obtained by the authors on LSM-infiltrated BSCF electrodes. In 189 that case [31], the LSM infiltration had a positive effect in terms of Rp at OCV. The reason for this 190 discrepancy is likely due to the different morphology of the LSM-infiltrated layer. In the previous 191 work, we dealt with a nano-sized discrete layer, resulting in an increased electrode-gas surface 192 exchange area and consequent improved electrode activity; in this study, a continuous LSM layer 193 was obtained, whose exchange properties with the gas phase are very different from those of the 194 nano-distributed one. 195

199
Measurements were performed in three-electrode configuration, using the ring-shaped RE.

200
Considering the effect of anodic overpotential (Figure 4c,d), it had a positive effect in terms of 201 Rp values, both in case of blank and infiltrated samples. It was already found in previous study that 202 cathodic overpotential had, on the contrary, a negative influence on the electrocatalytic performance 203 of BSCF-based electrodes [17]. The positive effect of anodic overpotential was further confirmed 204 through a sweep voltammetry measurement. Although not reported in the paper, the current density 205 Considering the effect of anodic overpotential (Figure 4c,d), it had a positive effect in terms of R p values, both in case of blank and infiltrated samples. It was already found in previous study that cathodic overpotential had, on the contrary, a negative influence on the electrocatalytic performance of Energies 2020, 13, 3659 7 of 10 BSCF-based electrodes [17]. The positive effect of anodic overpotential was further confirmed through a sweep voltammetry measurement. Although not reported in the paper, the current density measured under anodic overpotential was slightly higher than that under cathodic overpotential, indicating a higher activity in anodic conditions (namely, when the electrode is involved as anode in the electrolysis mode of operation).
The presence of LSM did not improve the trend of impedance when η was increased. At the temperature of 475 • C, R p at OCV was 4.43 Ω cm 2 for LSM-BSCF, while it decreased until 1.72 Ω cm 2 for the blank electrode. For the latter, an ASR decrease of 32% was observed under increasing anodic overpotential; the LSM-BSCF displayed a 24% decrease between OCV and +0.100 V, becoming 34.5% between OCV and +0.150 V. Unfortunately, beyond +0.1 and +0.15 V, the overpotential applied to blank and infiltrated electrodes, respectively, was destroyed, with the infiltrated one appearing more resistant.
In Figure 5, the inverse of polarisation resistance (1/R p ) is reported as a function of temperature. An Arrhenius-like behaviour allowed for obtaining the values of apparent activation energy (E a ) for both the systems: 85.3 and 110.6 kJ mol −1 for blank BSCF and LSM-BSCF, respectively. These activation energy values concern the activity during the dynamic equilibrium state at OCV. It has been proven in the literature that LSM activity drops under anodic conditions [37], and, actually, the impedance spectra extracted under anodic overpotential conditions confirm that the presence of the LSM hinders the activity of the electrode (see Figure 4c,d). Despite this detrimental effct due to LSM, the increase in the catalytic activity of the BSCF was confirmed when a net anodic current flowed through the cell. As observable in Figure 4d, the LSM hindering effect was not able to counterbalance the enhanced activity of BSCF. Moreover, the LSM-infiltrated layer resulted in a positive effect on the long-lasting performance of the SDC/LSM-BSCF system. In previous work, it was found that LSM contributed to maintaining the polarization stability of the BSCF electrode, likely because of a positive influence on the electrochemical potential gradient at the interface LSM/BSCF, with this stabilizing effect indeed also observed for LSCF-based and other electrodes [14,17].
Energies 2019, 12, x FOR PEER REVIEW 7 of 10 measured under anodic overpotential was slightly higher than that under cathodic overpotential, 206 indicating a higher activity in anodic conditions (namely, when the electrode is involved as anode in 207 the electrolysis mode of operation In Figure 5, the inverse of polarisation resistance (1/Rp) is reported as a function of temperature. 216 An Arrhenius-like behaviour allowed for obtaining the values of apparent activation energy (Ea) for 217 both the systems: 85.3 and 110.6 kJ mol -1 for blank BSCF and LSM-BSCF, respectively. These activation 218 energy values concern the activity during the dynamic equilibrium state at OCV. It has been proven 219 in the literature that LSM activity drops under anodic conditions [37], and, actually, the impedance 220 spectra extracted under anodic overpotential conditions confirm that the presence of the LSM hinders 221 the activity of the electrode (see Figure 4c,d). Despite this detrimental effct due to LSM, the increase 222 in the catalytic activity of the BSCF was confirmed when a net anodic current flowed through the cell. 223 As observable in Figure 4d, the LSM hindering effect was not able to counterbalance the enhanced 224 activity of BSCF. Moreover, the LSM-infiltrated layer resulted in a positive effect on the long-lasting 225 performance of the SDC/LSM-BSCF system. In previous work, it was found that LSM contributed to 226 maintaining the polarization stability of the BSCF electrode, likely because of a positive influence on 227 the electrochemical potential gradient at the interface LSM/BSCF, with this stabilizing effect indeed 228 also observed for LSCF-based and other electrodes [14,17]. 229 An anodic load of 0.150 V was run for 48 h at 600 °C, and an impedance test was performed 230 every 6 h. The result is shown in Figure 5. It is observed that the activity behavior, in terms of 231 polarization resistance, was stable for the duration of the test. During the voltage load, a current of 232 14 mA cm 2 was observed at 600 °C. This value should certainly be improved, but optimization of the 233 electrode morphology and performance under different gas partial pressures will be the subject of 234 further investigations. 235 An anodic load of 0.150 V was run for 48 h at 600 • C, and an impedance test was performed every 6 h. The result is shown in Figure 5. It is observed that the activity behavior, in terms of polarization resistance, was stable for the duration of the test. During the voltage load, a current of 14 mA cm 2 was observed at 600 • C. This value should certainly be improved, but optimization of the electrode morphology and performance under different gas partial pressures will be the subject of further investigations.

Conclusions
In this study, blank BSCF and LSM-infiltrated BSCF electrodes are analyzed, in view to be used as possible anode electrodes in solid state electrolysis (SOEC) devices. Electrochemical impedance spectroscopy investigation, coupled with an aging test at +0.15 V current load, was performed on a three-electrode system with SDC20 electrolyte.
It was found that polarization resistance decreased in both systems when an increasing anodic current load was applied. The blank BSCF electrode showed lower resistance (R p ) than the infiltrated one; then, it appeared that LSM did not give the chance to improve performance. This is possibly due to the detrimental effect of the continuous LSM layer covering most of the BSCF backbone. An LSM nano-sized discrete layer could result in increasing the surface exchange area, without subtracting appreciable BSCF surface, and consequently making the gas-electrode activity faster. This was actually found out previously by the authors.
On the other hand, the LSM layer had a positive effect on the impedance stability, since no sign of degradation appeared in the first 48 h under a voltage load of +0.150 V. Performance degradation in the first 100 h of operation is one of the major problems to be faced in SOFC/SOEC devices, and LSM infiltration was demonstrated to be a promising approach in overcoming this issue. Further attempts must be carried out in order to find the optimized parameters for a nano-sized discrete layer of LSM to be deposited. Most importantly, it was confirmed that BSCF-based electrodes are very promising candidates to be used as oxygen electrodes in rSOC.