The Identification of Degradation Parameters in SOC Under Load and OCV Aging Approaches

Recent SOC activities focus on upscaling systems to MW scale with target operation of several 10,000 h. These long lifetimes require new approaches for durability testing. In the present study, the influence of different operating parameters on degradation were studied by long-term cell testing in fuel cell and electrolysis mode (In-situ aging). Based on the results, accelerating parameters for degradation were identified and cells were treated/aged under these conditions without applying an external potential (OCV aging). This approach is cheaper and time saving as compared to conventional cell/stack long-term durability testing. Two commercial SOC cell designs, Fuel Electrode supported cells (FSC) and Electrolyte supported cells (ESC) were studied in this work. They were aged according to the two approaches (under operating conditions and with and without current load) and the obtained degradation effects compared to each other. Based on the cell composition and configuration the degradation parameters were observed to affect the cell performance to varying extents.

Solid oxide cells (SOC) have shown immense potential as efficient energy conversion technologies. Cells operated under fuel cell operation converts chemical energy to electrical energy with about 60% electrical efficiency and 90% combined heat and power (CHP) efficiency and in electrolysis operation for hydrogen and synthetic fuel production with efficiencies above 90%, as reported already in the 1980s. [1][2][3][4] However, cost and durability continue to challenge the commercialization of these systems today. 1,2 Although the need for expensive noble metals as catalysts is eliminated, there is a need for expensive balance-of-plant components due to these rather high operating temperatures (>600°C). In addition, the desired operation times for commercial applications are in the range of ∼60,000 h. 5 Different cell configurations have been developed during the past decades, Fuel Electrode supported cells (FSC) and Electrolyte supported cells (ESC) being the most common with different cell compositions. Nickel (Ni) based cermets with Yttria Stabilized Zirconia (YSZ) or Gadolinium Doped Ceria/Cerium Gadolinium Oxide (CGO) fuel electrodes are the most common in both cell configurations. 6,7 One of the common degradation processes that have been reported for the fuel electrodes is with respect to microstructural changes in the Ni cermets. Ni-YSZ has been widely studied in the past preferentially in anode-supported configurations under different operating conditions and gas compositions. [8][9][10][11][12][13] These electrodes have shown excellent performance and durability under long-term operation. However, at high current densities and particularly in presence of high steam, challenges related to loss of triple phase boundary (TPB) sites due to Ni migration and agglomeration under long-term operation have been previously reported. [14][15][16][17] Ni migration has been reported to be more severe in case of electrolysis operation with high steam as compared to fuel cell operation. 14,18 The acceleration in migration has been attributed to an increase in the contact angle between the Ni and YSZ particles at local pO2 values lower than 10 −19 , which is in the range of typical SOEC operation conditions. 19,20 It is suggested that they migrate towards higher pO2 values. In contrast, it was proposed by Mogensen et al. 21 that Ni migration in both fuel cell and electrolysis operation occurs, however in reverse directions. It was reported that the primary driving force in the first 100 h in fuel cell (FC) operation is the Ni coarsening whereas in EC mode, it could be the increase in the contact angles leading to Ni migration in the form of volatile Ni(OH)x species towards lower local pO2 in both modes. It was also reported that the loss in Ni percolation explains the apparent migration towards higher pO 2 values considering global pO 2 values (away from electrolyte-fuel electrode interface in FC and towards the interface in EC mode). The degree to which the different mechanisms may affect the overall degradation remains a challenge. Partial re-oxidation of Ni due to fuel starvation or under redox cycling where there might be periods of air cross-over, results in mechanical fatigue of the cells, which can have a detrimental impact especially for FSC configurations. 22,23 In comparison to conventional FSCs, the ESC configuration results in a more mechanically robust structure due to the thick, dense electrolyte support (3YSZ), however at the expense of a higher ohmic cell resistance, which typically requires higher operating temperatures. Ni-CGO fuel electrodes have been commonly used in the recent years in case of ESCs. These electrodes show a high impurity tolerance and good cell performance also at nominal operating temperatures 24 where the nominal operation parameters are defined by the manufacturers for the specific cell configurations. Ni-CGO electrodes combat the challenges in loss of TPB with the introduction of double phase boundary reaction sites, in which CGO, being a mixed ionic and electronic conductor (MIEC), aids the transport of both O 2− ions as well as electrons ensuring a longer cell operation. 25 The challenges continue to exist with the Ni coarsening and migration as well as in oxidizing environments. Further effect of impurities such as Silica has also been showed to have a severe impact on the cell performance in case of both Ni-YSZ as well as Ni-CGO electrodes. [26][27][28][29] Similar to Ni-YSZ based electrodes, studies have reported that steam has an accelerating effect on the Ni migration and coarsening. 30,31 Microstructural changes in Ni and CGO in the initial h of operation under wet fuel conditions followed by an apparent plateau have been reported. 30,32 Furthermore, in Ni-CGO electrodes, Ni growth and migration is additionally governed by CGO film formation on the Ni 30,33 as well as the dissolutionprecipitation process of CGO on the Ni particles. 34 Rapid degradation was observed during operation under high current densities at high pH 2 O/pH 2 ratios due to the formation of H 2 O or hydroxyl layers on the NiO. 35 As highlighted in the previous works, there is a strong interplay of the different operating parameters (steam, temperature, current density, operation mode) on the degradation processes. Thus, the identification of the impact of isolated parameters on cell performance remains a challenge.
This work aims to identify potential accelerating parameters towards cell degradation in both cell configurations while isolating their individual influence on the cell performance. Degradation studies were carried out on cells with FSC and ESC configurations under different fuel compositions and temperatures and fuel utilization under in situ and OCV aging approaches. The potential of z E-mail: anke@dtu.dk inducing degradation effects through an OCV aging treatment, eventually leading to accelerated aging, was investigated. Cells were therefore operated according to two approaches: (i) Cells operated with high steam compositions under load (in situ aging) in fuel cell or electrolysis mode and (ii) Cells exposed to similar gas compositions without any load (OCV treatment). The cells aged using these two approaches are compared with each other with the aim to establish cheaper (preferentially accelerated) aging approaches.

Experimental Test Plan
Commercial planar solid oxide cells provided by Sunfire (ESC) and Solid Power (FSC) were tested at Technical University of Denmark (DTU). The cells were laser cut to 5.3 cm × 5.3 cm with an active area of 16 cm 2 . FSCs consist of a Ni-YSZ anode (∼250 μm in total), which serves as both, a functional as well as a support layer, a dense YSZ electrolyte layer (8-10 μm), a CGO barrier layer (∼5 μm) and a Lanthanum Strontium Cobalt Ferrite-Gadolinium doped Ceria (LSCF-CGO) air electrode (∼50 μm). ESCs have a thin Ni-CGO fuel electrode (∼25 μm), a CGO layer close to the electrolyte (∼2 μm), a Ni-CGO functional layer (15-20 μm) and a Ni contact layer (∼5 μm), which aids contact with the Ni meshes for better current collection on the fuel electrode side. The electrolyte in the ESC is a dense 3YSZ electrolyte (80-85 μm). The air electrode side consist of a CGO barrier layer and a LSCF-CGO air electrode (30-35 μm) and a Lanthanum Strontium Manganite Cobaltite (LSMC) contact layer (∼5 μm) as shown in Fig. 1. The cells were mounted in alumina test houses described in detail elsewhere. 36 Gold seal was used for the fuel electrode side. As current collector, Nickel meshes (fine and coarse) and Gold mesh were used on the fuel electrode side and on the air electrode side respectively. In some tests (96/4 H2/H2O FSC), the test house consisted of fixed fuel flow channels with only a fine mesh (without a coarse mesh). Steam was generated within the alumina house by the controlled combustion of hydrogen and oxygen within the test house close to the cell inlet. 36 This ensures pure steam and without any fluctuations in the flow for good data quality.
The start-up of the cells followed the recommended protocol for the respective cell configuration. The cells were heated to 750°C for FSCs and to 850°C for ESCs using Argon on the fuel electrode and air on the air electrode as per manufacturer's protocols. The cells were then reduced under diluted hydrogen on the fuel electrode (5% H 2 in N 2 ) for two h followed by one hour under pure hydrogen. Cell voltages were measured both across as well as in-plane at the inlet and outlet. In addition, pO2 measurements were recorded at the inlet and outlet to calculate the corresponding local fuel compositions. These are recorded as the potential across the fuel at the inlet and outlet vs air in mV using in-house built setup with Pt wires. Thermocouples were located at the inlet and outlet of the cell, in addition to temperature measurements at the fuel and air inlet and outlet. Cells were characterized using I-V curves and electrochemical spectroscopy impedance (EIS) measurements at H 2 /H 2 O ratios of 96/4, 80/20 and 50/50 with an oxygen content of 0.21 (air) and 1 atm on the air electrode. These measurements were recorded at temperatures of 750°C (FSC only), 800°C, and 850°C. The characterisations at specific conditions were recorded after the cell reached a steady voltage for 10 min. To ensure high data quality, the EIS measurements were recorded and averaged for 100 cycles at high frequencies up to 6.7 Hz and 50 cycles for the low frequencies.
The measured spectra were also corrected for short circuit and well as a resistance measurement prior to the tests to identify any contributions that may arise from the test setup. This resulted in negligible deviations with a maximum of +/−5 mΩcm 2 in the presence of leaks. The same characterization measurements were recorded both prior to and post durability testing under OCV for all cells. For 1000 h of aging tests, the cells were operated under current (in situ aging) densities of +/−0.4 A cm −2 for FSC and +/−0.2 A cm −2 for ESC following nominal operation protocols for the respective configurations. During this operation, EIS spectra were recorded at intervals of 8 h for these in situ tests at the durability conditions. The aging under OCV conditions was carried out in the same test setup, only without applying a load (current). During OCV treatment, no i-V curves were recorded in order to avoid polarization of the cells. EIS were only recorded at 0 h, 500 h and 1000 h, to minimize impact of current. The long-term conditions under which the cells were tested are summarized in Table I. Relatively low fuel utilization values were chosen for the different long term tests so as to avoid effects of any additional microstructural changes such as elemental segregation or delamination issues in certain conditions.
For the FSCs, the cells were tested in 96/4 H 2 /H 2 O and 60/40 H 2 /H 2 O for fuel cell operation. For the electrolysis operation, a fuel composition of 10/90 H 2 /H 2 O was selected as the nominal operating condition. Since the EC operation was observed to have the highest degradation, the same condition (T and steam content) was chosen also for OCV treatment. In case of ESC cells, the cells were aged under similar fuel compositions as FSCs in FC mode. i.e., at 96/4 and 60/40 H2/H2O compositions. However lower total flows were specified for the ESCs as per the protocols and thus to maintain the similar fuel utilization as FSCs and using the manufacturer specified current load, the total flow was made up with N2 resulting in lower steam partial pressures (pH2O). This would imply that the effective steam composition is lower than desired. However, in this work, the effect of degradation was studied with respect to H2/H2O ratios and these were kept similar tests in both configurations. The effective inlet pH2O values were also kept constant for the different aged ESCs to have a direct comparison of the cells. For the 60/40 H2/H2O FC operation, an additional test was performed with a different batch to check reproducibility of the performance since the first cell, in addition, also experienced some load failure in between fingerprint measurements. In case of the Ni-CGO based ESCs, significant variations were observed at the 40% steam composition and thus EC There is a certain variation of ASR values between the cells (slopes in Fig. 1a), which could be attributed to variations in the cells from different production batches. Further, it is also noteworthy that a different test setup was used for one cell test (Cell A1 in Fig. 1); fixed flow channels (instead of coarse mesh) which could also contribute to some differences in the measured ohmic resistances. The ohmic contributions (see Fig. 2b) differ by ca. 0.026 Ωcm −2 (∼22%). This is within the variation limit observed with different batches for similar cells tested at 750°C. 37 The polarisation resistances were also slightly higher especially at the frequencies of ∼1-4 kh which arises from differences in the available TPB at the fuel electrode 38,39 due to initial cell microstructure variations among the different cell batches.
In case of the ESC configuration shown in Figs at the fuel inlet. Thus, the difference observed at the OCV is quite negligible also in the case of ESC configuration. For the ESC configuration, I-V curves were not recorded for cells to be aged OCV to eliminate effects of overpotential to cell microstructure as they showed higher sensitivity to fuel and temperature variations. In case of the ESCs, the I-V curves were recorded over a shorter current range of up to 0.4 Acm −2 as per the manufacturer protocols. Cells E1 and E2 showed a slightly higher initial resistance as compared to the other cells. The EIS spectra of the different cells confirms this observed variation in the IV curves primarily arises from the ohmic contributions (see Fig. 3b). The cell resistances among the cells differ by 0.08 Ωcm 2 (∼25%). Since the ESCs have a significant contribution from the electrolyte in terms of the ohmic contribution, slight variations in the electrolyte, in terms of thickness variations during processing, operation temperatures at which the measurements were recorded could result in such variations. However, the initial polarisation resistances for the different cells were comparable indicating similar electrode performance (Fig. 3c). The initial performance of the cell E5 showed a higher contribution at the lower frequencies (∼1 Hz), at which the contributions for the Ni-CGO electrode arises. 40 This  could arise due to impurities or some microstructure differences in the layer during processing steps. The iV and EIS characterization proves that the initial performances of the cells were comparable for both configuration and within the experimental error limits. Thus, the long term aging of 1000 h were carried out to study the changes in the cell performance before and after the aging as an effect of different operation atmospheres and overpotentials.
Deconvolution of loss contributions.-To understand the effect of the different parameters on the cell behavior, a detailed study of the different loss contributions to the total area specific resistance was carried out as in authors' previous study. 38 This was performed using the EIS spectra recorded during the initial performance characterization. The peak identification aids to further point out the differences between the two cell configurations (FSC, ESC) as well as the evolution of the different losses over time during aging under in situ operation and OCV treatment.
The EIS spectra recorded at OCV using different gas compositions to the fuel and air electrodes were used to identify the different contributions of both the electrodes to the cell impedance for the two cell configurations. The gas shift was also performed at different temperatures to study further the temperature dependencies for the different processes. Using the Analysis of difference in impedance spectra (ADIS) and the distribution of relaxation times (DRT) deconvolutions, the different loss contributions were identified. For DRT plots, the regularisation parameter used was 0.15. These are listed in Table II. In case of FSC, three distinct peaks were observed for the fuel electrode contributions. The low frequency peak at ∼1 Hz was identified to be gas transport process namely gas conversion and a middle frequency peak ∼30-50 Hz as gas diffusion. Further, a high frequency peak >1 kHz was assigned to the charge transfer process. These are in accordance with previous studies on similar cells. 22 For the ESC, the fuel electrode contributions were shifted to low frequencies owing to the high chemical capacitance of oxygen non-stoichiometry of CGO. 23,24 The low frequency peak between 0.1-1 Hz is assigned to the gas transport losses (in case of ESCs, which have thin electrodes, the gas diffusion losses also arise as a result of the Ni contact meshes). The charge transfer peak in the CGO based fuel electrode is observed at very low frequencies <0.1 Hz. This was also observed using symmetric cells reported by Riegraf et al. 25 The air electrodes of the two cell configurations were composed of similar materials, although with possible different microstructural properties. This would thus result in contributions at similar frequencies. Two distinct peaks were observed for the air electrode, where the low frequency ∼0.1-10 Hz was attributed to the gas diffusion in the bulk. The middle frequency 5-30 Hz in case of ESC and >100 Hz for FSCs was attributed to the oxygen ion transfer kinetics and diffusivity in the bulk. 26 Fuel electrode supported cells: durability studies.-FSC durability was tested as function of the inlet hydrogen to steam ratios, varying from 96/4 to 10/90 and operating mode (FC vs EC). OCV aging was carried out for the highest steam content of 90% to study the isolated effects of steam (without applied overpotential) on the cell performance. The different FSCs were aged under the specified conditions as described in Table I. For both FC as well as EC operation, a nominal current density of 0.4 A cm-2 was applied at low fuel utilizations so as to avoid additional degradation effects due to the higher overpotentials.
The different cell parameters were monitored continuously throughout the test duration. The resulting voltage curves for the different tests are shown in Fig. 4. In case of the EC operated cell, there was a power failure at ∼160 h (red curve in Fig. 4). The  voltage recorded following the power failure indicated a decrease in the OCV value from the initial OCV measured. The value decreased from 840.12 mV to ∼803 mV which corresponds to an effective steam composition increase of 5%. The increase in steam composition was however not associated with any change in measured cell temperatures, thus indicating no signs of significant cracks and combustion at the cell surface. Thus, the change in OCV is likely associated with leak from the edges around the cell sealing. The voltage curve after the power failure was corrected for the appropriate OCV values. The difference in the voltage at the start and end of the durability studies were compared (see Table III). It is noteworthy that these changes are also affected by the fuel compositions and the applied loads and hence only a preliminary analysis is obtained. The effective fuel composition at the inlet was computed based on the measured OCV values at the beginning of the durability test after initial fingerprint measurements. The OCV aged cell showed an effective composition of 10/90 H2/H2O very close to the set values. The voltage in the 96/4 FC operation, i.e., with lowest steam content, changed by ∼6 mV (0.63 V% kh-1) whereas the FC operated cell with higher steam composition of 40% resulted in a voltage decrease of 14.6 mV corresponding to a degradation rate of 1.73 V% kh-1. This shows an increase of about 2.7 times the voltage degradation as in the drier feed operation in FC mode. Furthermore, the EC operation with the 10/90 H 2 /H 2 O resulted in an even higher degradation rate of ∼3 V%kh-1 with ∼30 mV change in the measured voltage. This is also expected from the literature studies, which show an increase of the cell degradation under EC operation and higher steam conditions. 18,41 Under EC operation, the voltage increase is higher in the initial 500 h and the cell performance stabilizes towards the end of the test (see red curve, Fig. 4). It must be noted that in case of the OCV treated cell, the cell voltage measured over time is representative only of the variations in the sealing and gas tightness since it is aged only under OCV.
The change of the resistance contributions were analyzed using the EIS spectra recorded under load and during OCV treatment for the different cells. The Nyquist and Bode plots at 0 h, 100 h, 500 h and 1000 h are shown in Fig. 5. The ohmic resistance of the in situ operated cell with 4% steam (Fig. 5a) increased slightly from 0.12 to 0.127 Ωcm 2 (50% kh −1 ) within the first 50 h, and then stabilized at ∼0.127 Ωcm 2 until 1000 h corresponding to a degradation rate of ∼3.5% kh −1 . In case of the FC operation with 40% steam in hydrogen (Fig. 5b), the Rs increased almost linearly from ∼0.143 Ωcm 2 to ∼0.165 Ωcm 2 corresponding to an increase of ca. 20-25 mΩcm 2 kh-1 (∼12%-17% kh −1 ), which was significantly larger than under less humid fuel conditions. For the electrolysis operated cell ( Fig. 5c), the cell experiences an average Rs increase rate of ca. 5 mΩcm 2 kh −1 (∼2.8%kh −1 overall). Due to the power failure that occurred, the degradation rate has also been computed for initial and later stage of the test. This corresponds to a ohmic degradation in the initial period (first 160 h) of 11.67 mΩcm 2 kh −1 (∼9.6% kh −1 ) and ∼6 mΩcm 2 kh −1 (∼5.6% kh −1 ) at the end of 1000 h. This apparent decrease in the degradation is also an effect of the extended period of operation in the post power failure test. In the 100 h, following the power failure, the ohmic resistance changed at a rate of ∼25 mΩcm 2 kh −1 (∼20% kh −1 ) which stabilized after the initial period. Fig. 5d shows the change in losses over 1000 h for OCV aging under OCV with 10/90 H 2 /H 2 O conditions. The ohmic resistance in this case remained unchanged.
The results indicate that the rate of increase of Rs is highest for the cell operated with 40% steam under FC mode with close to 20 mΩcm 2 kh −1 , followed by the cells operated with 90% steam under EC mode with ca. 6 mΩcm 2 kh −1 and the cell operated in fuel cell under more dry conditions with ca. 4.5 mΩcm 2 kh −1 , i.e. with rather similar Rs degradation rates. Thus, the ohmic resistance does not show any clear dependency on the operation mode or the fuel compositions for the durability testing. A significant contribution was only observed in case of the 40% steam inlet composition FC operated cell, which could also partly arise from the different contact layer on the fuel electrode (fixed channels) used in the test house.
The increase in the ohmic resistance contributions in durability tests of FSCs with Ni/YSZ fuel electrodes has been attributed to various causes such as (i) formation of Kirkendall voids in the electrolyte causing the effective ionic pathway to increase, 41,42 (ii) Ni migration resulting in a higher ohmic resistance due to an increase in the effective electrolyte thickness (longer pathway for the O 2− ions) 43 (iii) loss of contact due to delamination from foreign phases. 44 A qualitative analysis of the polarisation losses of the different cells (both operated under load and aged at OCV) as shown in Figs. 5a-5d reveal a degradation of the high frequency (∼1 kHz) contribution, which has been previously identified to be the charge transfer contributions at the fuel electrode. Furthermore, the contribution from the fuel electrode TPB at high frequencies is lower with the increase in steam composition (see Figs. 5a and 5b) and significantly low for the 10/90 H 2 /H 2 O aged at OCV conditions (see Fig. 5d) highlighting the favorable reaction kinetics under high steam conditions. The fuel conversion and diffusion contributions were identified at ∼1 Hz for the FC operated cells, whereas in case of the EC and the OCV aged cells, with significantly higher steam at the inlet of 90%, the peak was observed to shift to the lower  Fig. 5c), due to the applied cathodic overpotential the initial resistance at the high frequencies was higher. In contrast, the low frequency region pertaining to fuel conversion was lower in comparison to the OCV aged cell. However, it must be noted that unlike the comparison of ohmic resistances under the different operation conditions (at same temperature), polarization resistances must be compared at similar conditions (current density) due to dependency of reaction kinetics on the effective fuel composition.
Fuel electrode supported cells: pre-and post-durability.-To get a deeper understanding of the cell degradation under different modes and fuel compositions, EIS spectra recorded prior and post durability at the same set of conditions were analyzed. With this approach, the end state of the cell as compared to the fresh state is characterized, disregarding changes during aging and the specific conditions, such as the current density. The total cell resistance (as a sum of the ohmic and polarization contributions) recorded at 80/20 Fig. 6. For the 96/4 FC operated cell, a reduction was observed for the resistance measured after 1000 h operation, which would correspond to an activation rather than a degradation. The reduction can be attributed partly to a lower ohmic resistance and the polarization contributions as well. In case of the post-test characterization, the cells showed a better contact resistance which might be indicative of the test setup as well. In this test, the contact for the cells on the fuel side was done using flat flow channels as compared to the Ni meshes used in the other cases. Although this should not cause a significant change due to the high Ni conductivity, the contact seems to have improved slightly over the long-term operation resulting in an "apparent" improvement in the overall cell performance. In terms of the cell operated in 40% steam in hydrogen as fuel cell the resistance increases by 0.06 Ωcm 2 . About ∼67% of this increased resistance (∼0.04 Ωcm 2 ) was observed to arise from the polarisation contributions. As discussed in the long-term operation section, this cell also observed a significant increase in ohmic resistance. This contributes to the increased resistance in addition to the polarization contributions changes. The electrolysis operated and OCV aged cells showed an increase of the total resistance of 0.07 Ωcm 2 and 0.04 Ωcm 2 respectively. The ohmic resistance change under load was small and hence the observed changes should mainly arise from the increased polarization losses (see Figs. 5c and 5d). The cell aged OCV under a higher steam composition of 90% was observed to have similar increase in the cell resistance, in terms of polarisation contributions as the cell degraded in EC mode (in situ) under 90% inlet steam composition.
It is crucial to assign a specific frequency region for the changes of the polarization resistance. It is important to understand if the different operating conditions affected the same cell processes without inducing new degradation mechanisms. The pre-and posttest fingerprints for different aging tests were compared at 80/20 H 2 /H 2 O and air at 750°C. The Bode and DRT deconvolutions for the fuel cell operated cells at high and low steam contents (4% and 40%) are shown in Figs. 7a and 7b respectively. The peaks labelled P1 to P4 have been summarized in Table II. In case of the 96/4 H 2 /H 2 O FC operated cell (Fig. 7a, an increase was observed at the low frequency region below 1 Hz (P1) which could indicate an increase in the gas conversion contributions. This result, coupled with a higher ohmic resistance of 0.15 Ωcm 2 at initial fingerprints compared to the 0.13 Ωcm 2 post durability fingerprints could be associated with non-ideal contact at the initial few h of the fingerprint measurements (although with good gas sealing throughout). A negligible change was observed at the high frequency ∼1 kHz associated to the Ni-YSZ charge transfer losses. The peak became more narrow and steeper and thus the area under the peak for DRT deconvolution remained unchanged. For the FC operated cell, with the higher steam content of 40% (Fig. 7b), degradation at the peaks P3 and P4 was observed. A higher variation was observed at P4 associated with the Ni-TPB contributions. The presence of steam has been shown to accelerate the Ni migration in FC operation. 45,46 This would result in such an increase of the charge transfer contributions. Changes at the air electrode in terms of demixing and degradation in reaction kinetics over long term operation could result in the increase observed at P3 associated to the oxygen surface kinetics. Thus, the fuel cell operation over 1000 h leads to degradation of both electrodes of the FSC.
EIS spectra recorded prior to and post cell operation as electrolysis cell at 10/90 H 2 /H 2 O inlet fuel composition for 1000 h at −0.4 A cm −2 are shown in Fig. 8. The ohmic resistances did not change significantly as a consequence of the durability operation as discussed in the earlier section (see Fig. 5c). Looking at the polarization resistances, the contributions at the low frequencies are fairly unchanged before and after durability operation. The most significant changes are observed at the middle and high frequency (>200 Hz) region similar to the 60/40 H2/H2O FC operated cell (Fig. 7b). The variation observed in peak P4, is larger after the EC operation indicating that the change in this region is accelerated in the EC operation in comparison to the FC operation. This can be attributed to both a higher local overpotential experienced by the EC operation as well as the higher steam content in the fuel feed leading to more significant changes of the TPB sites. In addition, the peak P2 was observed to shift to the right slightly, indicating of a decrease in the diffusion resistance due to suspected higher resulting porosities in the cell with Ni migration and agglomeration. Peak P3, at middle frequency ca. 500 Hz, was seen to increase in resistance owing to the degradation at the air electrode as observed in the case of FC operation. The EIS of the cell aged at OCV with high steam content (under OCV conditions) are shown in Fig. 8b. Similar to the EC operated cell, the low frequency peak P1 remained unchanged. Furthermore, peak P2, associated with gas diffusion also remained unchanged in this case. The variations were observed only in the  high frequency region >1000 Hz. The primary change observed was at the high frequency at ∼10,000 Hz similar to the FC and EC operated cells. This indicates that the degree of degradation occurring in the cells aged at OCV under high steam condition undergo TPB loss in terms of Ni coarsening or migration comparable to operating the cells under load (in situ) as FC or EC. The presence of high steam seems to govern the degradation mechanism that occurs in such Ni-YSZ based fuel electrodes. The associated migration and coarsening effects need to be confirmed with microstructural analysis, which is not within the scope of this work.
For further understanding the different contributions, the EIS spectra were fitted using a suitable equivalent circuit model (ECM) for the processes identified in Table II. The ECM model chosen was the same as the one used for the detailed deconvolutions of the losses reported in author's earlier work. 38 The ECM model comprises of inductance (L), ohmic resistor for the ionic contributions (R) and 4 constant phase elements (RQ) for diffusion and fuel electrode contributions and Gerischer (G) for the air electrode contribution in the following configuration: L-R-R1Q1-R2Q2-G-R4Q4-R5Q5. DRT deconvolutions were used to identify the frequencies for the different contributions in the pre and post aging. The variation in the air electrode contributions were within 1% and thus the changes observed at peak P3 in the Bode and DRT plots were a possible consequence of overlapping effects from the variations in Peak P4. The variations observed in the high frequency peak, P4 are shown in Fig. 9. As explained earlier the 96/4 FC operated cell has a higher initial resistance as compared to the other aged cells. However, the variation observed at the Ni-TPB contributions pre and post durability operation was negligible. In case of the other three cells, that were aged under 60/40 under FC, 10/90 under EC and 10/90 at OCV conditions, the initial charge transfer resistance was comparable. The EC operation showed the highest increase close to 100% (0.057 Ωcm 2 ) as observed also in Fig. 8a followed by 88% (0.046 Ωcm 2 ) increase in case of the OCV aged cell at the same high steam content of 90%. A charge transfer resistance increase of ∼71% (0.038 Ωcm 2 ) was observed for the FC operated cell with 40% steam. This further confirms that the presence of high steam accounts for the primary mechanism of the degradation leading to the increase in the charge transfer contribution in FSC configuration with Ni-YSZ fuel electrodes. A slightly higher degradation on the charge transfer resistance observed in the case of the EC operated cell could be hypothesized with the higher local overpotential in addition to the high steam especially at the inlet of the cell where the current density and the steam content are the highest as seen in case of previous studies of long-term EC operated cells. 8,18,21,39,[47][48][49] The microstructural changes that result in these differences in electrochemical performance need to be studied in detail.
Electrolyte supported cells: durability operation.-The effect of operating modes (FC, EC or OCV), operating temperature and steam content on the durability behavior of ESCs was studied. The aging conditions over 1000 h are listed in Table I. Figure 9 shows the cell voltage over operating time for the different cells. Two cells were operated in fuel cell mode using 96/4 (blue curve) and 60/40 (red curve) H 2 /H 2 O fuel ratios at 850°C under current density of 0.2 A cm −2 . Electrolysis operation was studied with 60/40 H 2 /H 2 O at −0.2 Acm −2 (orange) and 60/40 H 2 /H 2 O at −0.2 Acm −2 with higher fuel utilization of 83% (green). Furthermore, to study the effect of steam without load, two OCV aging tests were carried out under the same steam inlet conditions (60/40 H 2 /H 2 O), at two temperatures (850°C (black) and 950°C (purple) in Fig. 10. The change of the voltage was highest in case of the 60/40 H2/H2O FC operated cell (red curve). The 60/40 H2/H2O FC test was repeated with a cell from a different batch and operated under same fuel ratio indicated only a higher initial ohmic resistance in comparison to the previous test and the performance was comparable. Thus the repeated test for the   Table IV. The voltage drop in case of the 60/40 H2/H2O FC operated cell at a rate of 12.39 mV/kh (1.5 V% kh-1) is twice the drop observed in case of the FC operation aged under lower steam content, which showed a voltage degradation of 6.19 mV/kh (0.67 V% kh-1). In case of the EC operation under similar inlet fuel compositions, the voltage rise corresponded to a change of ∼4 mV/kh (0.37 V% kh-1) and under a higher fuel utilization it resulted in a difference 2 times higher (than the EC under lower fuel utilization) at 8.05 mV/kh (0.76 V% kh-1). The post-test OCV values were compared for the different cells and no significant increase in leaks were observed. Thus the cells need to be further studied under similar conditions at OCV for a clear understanding of the variations and to deconvoluted the different loss contributions. The actual inlet composition differed by a maximum of 8% (corresponding to OCV change of 3 mV deviation between the cells) in case of the cell tested under higher fuel utilization. This could be a result of a higher leak effect at the lower flow rates, thus an additional flow of N2 was introduced to set the total flow rate same as the other tested cells after first 250 h of operation under lower total flows.
A closer look at the overall cell performance was done by plotting the Nyquist and Bode plots to understand the variation observed in the ohmic and total polarization contributions (see Figs. 11a-11d). In comparison to the Nyquist and Bode plots of Ni-YSZ based FSCs (see Fig. 5), the ESC configuration composed of Ni-CGO fuel electrodes, show only one predominant peak primarily attributed to the charge transfer and diffusion contributions at the low frequencies ∼1 Hz. 38,40,50 The degradation of the fuel electrode is expected occur to a different extent in ESC due to the primary difference in the materials. In case of the ESCs based on Ni-CGO, the entire fuel electrode results to be active due to the expansion DPB as compared to the TPB in case of Ni YSZ based FSCs, where the Ni migration and coarsening would result in severe loss of cell performance. There is not a lot of studies done on the long-term degradation of Ni CGO based cells. However some studies have pointed the changes in the electrochemical performance in the presence of high steam at the lower frequencies. 33,35,[51][52][53][54][55] In addition to the EIS for the cells aged during FC operation (Figs. 11a, 11b), the results of EIS recorded on two OCV aged cells over time in 60/40eam/hydrogen fuel are shown in the subplots of Figs. 11c, 11d. The ohmic resistances of the four cells under in situ cells under polarization as fuel cells and electrolysis showed a minor increase over the 1000 h operation. In case of the high steam content in hydrogen, there was a slight increase after ca. 800 h, with a rate of ca. 13 mΩcm 2 kh −1 (∼3.4% kh −1 ). In case of the EC operated cell (see Fig. 11d), slight increase in ohmic resistance was observed after a H 2 cut off for a few h close to 85 h into the long term operation resulting in a change at a rate of ∼14 mΩcm 2 kh −1 (∼5.2% kh −1 ). Similar to the FSCs, the polarization resistance changes are suppressed under load and hence the variations are difficult to observe and compare different aged cells. However, the variation of frequencies at which the processes occur can be observed in the Bode plots in Figs. 11a-11d. In case of the 96/4 H 2 /H 2 O FC operation as well as the EC operation under high fuel utilization, Uf, the low frequency peak was observed ∼0.2 Hz. In the 96/4 H 2 /H 2 O cell, the 4% steam at the inlet was operated with a Uf of 14% resulting in a steam content of ∼15% at the outlet and in the case of the high Uf EC operation, the outlet steam content was reduced from 40% to 6% at the cell outlet. However, the fuel composition measured at the outlet shows fluctuations with a higher value of 10% steam corresponding to a higher local fuel utilisation of ca.70% at the cell outlet. Thus, such reducing atmospheres would have a negative effect on the reaction kinetics since the reaction is known to proceed through the spill over mechanism, shifting it to lower frequencies as seen in the Bode plots. In comparison, this peak occurs at ∼1 Hz in case of 60/40 H2/H2O FC operation as well as 10/90 EC operated cell with a fuel utilization of ∼30% (with steam composition varying from 90% at inlet to 26% at the outlet). This observation is in contrast to the observation for FSC (Fig. 5), where the low frequency contribution increased in case of the higher steam content. This can be explained by the difference in the cell microstructures, where the diffusion contribution in the FSC is more significant as compared to the ESCs with thinner fuel electrodes (∼8-10 times thinner). 38 Furthermore, the contributions at the low frequency in addition to diffusion is also attributed to the charge transfer in case of Ni-CGO based electrodes, thus having an inverse correlation of resistance with steam. From a qualitative analysis, only a slight variation in the peak height was observed in case of the FC operation with 60/40 H 2 /H 2 O at the lower frequencies.
During the OCV treatment at two temperatures (850°C and 950°C ), only three EIS were recorded (at the start, 500 h and at the end of the durability testing over 1000 h) in order to limit the influence of any current. Ohmic contributions obtained from EIS of the cells during both the OCV treatments showed a significant increase at a rate of ca. 92 mΩcm 2 kh −1 for first 500 h in both OCV aged cells (25% kh −1 and 45% kh −1 for 850 ⁰C and 950 ⁰C respectively). The degradation rate of ohmic resistance during the treatment at 950 ⁰C starts to decrease with time, to 72 mΩcm 2 kh −1 at 1000 h (35% kh −1 ). It is difficult to explain the significantly larger degradation of the ohmic resistance of the OCV treated cells as compared to the operated cells under otherwise similar conditions (same steam/ hydrogen and temperature, but without polarization, comparing Figs. 11 and 12). One would suspect that polarization of cells would result in additional degradation or at least the same, but not less. The cells did not show any significant variations in the initial characterization at different conditions. Thus, a detailed microstructural analysis is required along with spectroscopy techniques to analyse further the influence of these conditions under OCV conditions. In terms of polarisation contributions, similar to the in situ FC operated cells under more oxidizing environments the summit frequency of the low frequency charge transfer and diffusion contributions occurs at ∼1 Hz as seen in the Bode plots in (Figs. 11b and 12). The changes over the 1000 h is difficult to observe in the Nyquist and Bode plots (Fig. 12b) due to good reaction kinetics in Ni-CGO fuel electrodes. For the 850°C OCV aged cell (Fig. 12a), the changes  were observed at the middle frequency range ∼100 Hz, identified to arise primarily from the air electrode contributions, whereas for the 950°C OCV aging (Fig. 12b), variations were observed at the low frequencies as well as the middle frequency ranges. Comparing the polarization resistances over 1000 h for a qualitative understanding, the cell with 96/4 H 2 /H 2 O fuel composition (Fig. 11a) shows the highest initial polarization resistance due to the lower steam content and applied fuel cell current. A similar trend of ca.15 mΩcm 2 kh −1 (ca.10% kh −1 ) Rp degradation was observed in the initial 400 h for both the FC operated cells. However, for the fuel cell operation in higher steam content, showed an increase to almost double the dry inlet FC operated cell (ca. 36 mΩcm 2 kh −1 (∼20%kh −1 )) after ca. 400 h operation (Fig. 11b). Both the EC operated cells showed a similar variation as the FC operated cell under 60/40 fuel composition (∼38 mΩcm 2 kh −1 ). The higher voltage drop thus observed in the FC operated cell (Table IV) arises from the higher variation in the ohmic resistance contributions. For the case of OCV treated cells (Figs. 12a and 12b), a rate of ca. 35-37 mΩcm 2 kh −1 (∼45%/kh) was observed in the first 500 h. In the next 500 h, the slope of the OCV operated cells reduced slightly. The absolute change in the polarization contributions of both in situ and OCV aged cells with 40% steam was 0.036-0.038 Ωcm 2 , which indicates that an increase in Rp aging was achieved by treating the cells under higher steam partial pressures (0.3 atm) with or without polarization as compared to the dry steam conditions. Furthermore, for a direct comparison, the fingerprint tests recorded at OCV pre-and post-durability tests were analysed.
To understand the changes of the different cell component contributions upon long-term aging, the EIS of the cells (both in situ operation and OCV treatment) were recorded prior to and after the aging processes. The fingerprint measurements recorded under 80/20 H2/H2O composition with air on the air electrode at the ESC nominal operation temperature of 850°C for the different aged cells are compared in Figs. 14. The EIS recorded prior to the aging treatments should be the same, as they were recorded under the same EIS conditions at OCV. However, this is not the case; both the ohmic and the polarization resistances have slight differences. These variations could be due to different cell batches, an ohmic resistance change of ∼0.02Ωcm 2 for 5 microns in 3YSZ (see Fig. 13, the range variation was further also seen in literature using similar cells [56][57][58] and also with variations in furnace temperature (due to the thicker electrolyte) during the fingerprint measurements as previously mentioned. Overall cell resistance change is highest in case of OCV aged cell under 60/40 H2/H2O at 950°C.
This shows an increase of 0.25 Ωcm 2 followed by the OCV aged cell under same fuel composition at 850°C with 0.15 Ωcm 2 . In case of the in situ operated cells, the highest variation was observed for the EC operated cell under 60/40 H 2 /H 2 O fuel composition under high fuel utilization showed an increase of 0.07 Ωcm 2 . These changes observed in the total cell resistances differ either in ohmic or polarisation contributions for the different aged cells. The OCV aged cells resulted in ohmic resistance increase of about 7-10 times higher than the in situ aged cells. Thus a significant amount of the total contribution in Fig. 14 would arise from the ohmic contributions for the OCV aged cells. The variations observed for the ohmic contributions remain the same as compared to durability operation since it does not depend on the polarisation or gas composition. For the comparison of the polarisation contributions, the Bode as well at the corresponding DRT deconvolutions of the fingerprint measurements were analysed.
The results for the cells operated in FC mode at 0.2 A cm −2 and 96/4 and 60/40 H2/H2O are shown in Figs. 15a, 15b. The cells experience degradation in two different processes, namely at low frequency polarization resistance assigned to the charge transfer resistances and diffusion terms (∼1 Hz) and at a frequency close to 100 Hz which indicates changes in the oxygen electrode. However, the changes at the air electrode occur to a much lesser extent in comparison to the fuel electrode losses. The FC aged cell under 60/ 40 composition showed a higher initial charge transfer contribution (Peak P1) due to some power disruption observed during fingerprint measurements under load prior to the durability testing. These shown initial fingerprints were recorded after resolution of the observed issues. The increase in the polarisation contributions correspond to 24.8 mΩcm 2 kh −1 and 38.04 mΩcm 2 kh −1 (∼30% higher) for the FC operated cell under dry and wet inlet conditions respectively. Due to the significant overlap of losses, as previously reported in literature, 38,50 the deconvolution of losses in the charge transfer, fuel conversion and diffusion contributions is a challenge.
The pre-and post-durability fingerprints recorded for cells tested under EC operation with 60/40 H 2 /H 2 O fuel composition and air at air electrode for 1000 h are shown in Figs. 16a and 16b respectively. It was observed that the changes in case of the EC operation were primarily at even lower frequencies of ∼0.1 Hz and no changes were observed at Peak P2 in case of both the EC operated cells. The slight variations at the middle frequency attributed to the air electrode (Peak P3) were observed also in case of the EC operation. The  polarisation resistance increase was to a similar extent as compared to the FC operation and the overall contribution of the air electrode remains small. Both EC operated cells showed an increase of ∼36-37 mΩcm 2 kh −1 similar to the 60/40 H2/H2O FC operated cell. Thus, in case of both the FC as well as the EC operation, the increase in polarisation contributions occurred to the same extent. Thus, the operation mode did not contribute significantly to the degradation at these fuel utilization ranges (compare Figs. 15b and  16). Although, with a higher inlet steam composition (compared to the 96/4 H2/H2O inlet composition), the polarisation resistances showed up to ∼13-14 mΩcm 2 kh −1 higher degradation rate at the measured fingerprint conditions. The EIS spectra for the OCV aged cells aged in the same fuel composition of 60/40 H 2 /H 2 O and at the temperature of 850°C (Fig. 17a) and 950°C (Fig. 17b) are studied. During the OCV aging, the cell experienced significant changes in the ohmic contributions as shown in Figs. 12a, 12b. In case of the OCV aged cell at 850°C (Fig. 17a), the degradation of the electrodes was more prominent at the middle frequency region around 50 Hz owing to the oxygen kinetics. Only a slight increase was observed in the low frequency region. The change in total polarisation losses was comparable to the in situ aged cells with ∼38 mΩcm 2 kh −1 . The significant increase in the ohmic resistance could be the reason for this increase observed in the air electrode. The OCV variation throughout of the aging was negligible and hence there was not a significant deterioration of gas tightness. The delamination of the electrodes could also result in such observation, however this variation from the in situ operated cells need to be analysed further. The effect of higher temperature (950°C) was studied using the OCV aging approach (Fig. 17b). Higher aging temperature is seen to induce severe changes in the diffusion and electrochemical contributions (∼0.1 Hz) in comparison to the in situ operated cells as well as the OCV aged cells under similar fuel compositions. Variation at the air electrode and ionic rail ∼100 Hz and 1 kHz was observed as well. In addition to the increase in ohmic contributions, the polarisation resistance of the cell increased by ∼96 mΩcm 2 kh-1. This is ∼3 times higher than the other in situ and OCV aged long term aged cells under the same inlet fuel composition of 60/40 H 2 /H 2 O. Thus, loss of the reaction sites at the fuel electrode were accelerated significantly by high temperature and in combination with certain content of steam in the hydrogen fuel. For more detailed understanding of the changes, microstructural changes for the cells aged using the two approaches are to be examined.

Conclusions
Two commercial cell configurations, FSC and ESC were aged under different conditions to understand the effect of isolated  parameters on the cell durability. For this purpose, one set of cells was operated with different hydrogen to steam ratios in either FC or EC mode (in situ operation). Another set of cells was treated at the same fuel gas composition and at different temperatures without applying a load (OCV treatment).
For the FSCs, the most significant changes were observed at the high frequency region assigned to the charge transfer contribution on the fuel electrode under long-term operation. A higher steam content under fuel cell operation led to an increased degradation of the ohmic resistance at a rate of ca. 20 mΩcm 2 kh −1 . However, in the electrolysis and OCV aging, the ohmic contributions remained fairly unchanged. In case of the polarization resistances under 96/4 conditions, a negligible change was observed whereas the cell operated under high steam conditions, with 90% inlet steam composition operated in EC mode showed the highest degradation in the charge transfer at >1 kHz. The OCV aged cell under the same inlet composition increased ∼88% in the charge transfer contribution. This shows that the high steam composition plays a vital role in terms of the degradation observed associated with Ni TPB loss. The OCV aged cell was able to capture more than 80% of the cell degradation treated in EC mode under the same inlet fuel compositions. As a next step, to identify the exact degradation mechanism requires detailed microstructural analysis. The charge transfer resistance of the FC operated cell with 60/40 H2/H2O composition increased by ∼70%.
For ESCs, the cells experienced a severe degradation of the ohmic contributions, in particular during OCV treatment, as well as processes at the low and middle frequency (ca. 0.1-0.5 Hz and ∼50 Hz) assigned to the charge transfer and oxygen kinetics, respectively. Degradation of the charge transfer processes were observed to a significant extent on the OCV treated cell with 60/40 hydrogen/steam, at 950 ⁰C. This indicates that high temperature and high steam content (likely induced further with overpotential at lower temperatures) causes this degradation phenomenon on ESCs. Further microstructural analysis is necessary to understand the occurring mechanisms especially to the large changes in the ohmic resistance in case of the OCV aged ESCs. Apart from the high temperature operation, the in situ operated cells under FC and EC operation under 60/40 H2/H2O experienced a comparable increase in the polarisation resistances, particularly at the low frequency region. The OCV aged cell at 850 C under similar inlet fuel composition showed degradation predominantly at the middle frequency region. Thus a detailed microstructural analysis is desirable to further understand the observed increase in ohmic resistance in case of the OCV aged cells and the associated effect on the polarisation resistances.