Characterizing Performance of Electrocatalyst Nanoparticles Infiltrated into Ni-YSZ Cermet Anodes for Solid Oxide Fuel Cells

The in ﬁ ltration of nanoparticle electrocatalysts into solid oxide fuel cell (SOFC) electrodes has been proven to produce a high density of electrochemically active sites, and reduce charge transfer polarization losses for SOFC electrodes. This is crucial for intermediate temperature operation, as these losses increase greatly at lower temperatures. Nickel-yttria stabilized zirconia (Ni- YSZ) cermets are low-cost, and exhibit excellent stability, but their main disadvantage stems from nickel coarsening and performance loss over their operational lifetimes. In ﬁ ltration of electrocatalyst nanoparticles has been shown to mitigate nickel coarsening and the consequent anode degradation. In this study, the effects of these in ﬁ ltrants are observed in a standard Ni-YSZ electrode. In addition to nickel, mixed ionic and electronic conducting (MIEC) phases were in ﬁ ltrated into Ni-YSZ scaffolds and their performances were characterized using electrochemical impedance spectroscopy (EIS). Cross-sectional microscopy of fractured cells was used to compare electrode microstructure and particle statistics. A model is proposed for how the nanoparticle electrocatalysts improve the anode performance.

High temperature SOFCs operating over long timescales have been observed to experience extreme phase coarsening and instability by end-of-life. Operation at a lower temperature allows broader materials selection as well as slower cell degradation rates. [1][2][3][4][5][6] Previous work has shown that while lower operating temperatures do slow degradation kinetics, the electrocatalytic activity of the cell becomes sluggish as well, leading to larger charge transfer polarization. 7,8 Thus, it is imperative to address the larger charge transfer polarization at lower temperatures to maintain high performance. One direct way to decrease charge transfer polarization in the electrodes is to increase the number of electrochemically active sites in the electrodes. 9 Liquid phase infiltration of nanoparticle electrocatalysts into porous substrates has been shown to be an effective method to improve solid oxide fuel cell (SOFC) electrodes and subsequent cell performance. Some electrocatalysts help decrease feature size within the cermet scaffolds, provide new electrochemical sites, and reconnect isolated grains within the scaffold. Alternatively, other electrocatalysts can improve the SOFC tolerance to contaminants in the reactant gases. [9][10][11][12] Both of these infiltrant functions can reduce polarization resistance in the electrodes, especially at lower operating temperatures. Previous works have also employed metal infiltration into YSZ skeletons, and have also addressed anode connectivity issues. [13][14][15] In the present work, since electrocatalysts have been infiltrated into a pre-existing Ni-YSZ scaffold, connectivity issues are not as relevant as the in the work of Gross et al., 13 McIntosh et al. 14 and Costa-Nunes et al. 15 Ni-Zr 0.84 Y 0.16 O 2 (Ni-YSZ) cermets have long been state-of-theart anode electrodes in SOFCs due to their improved performance and stability at high temperatures. Previous work has shown that infiltration of nickel within Ni-YSZ networks can improve cell performance by increasing the length of triple phase boundaries (TPB) with the introduction of nickel nanoparticles. 12,16,17 Separately, Ce 0.9 Gd 0.1 O 1.95 (GDC) has also been used as an infiltrant due to its mixed ionic and electronic conductivity (MIEC) in reducing environments, showing significant improvements in cell performance for GDC-infiltrated scaffolds. [18][19][20] In this work we demonstrate how repeated cycles and different infiltrant solutions affect Ni-YSZ electrode performance using symmetric cells. Results show that multiple infiltration cycles initially improve cells; however, further increasing infiltration cycles lead to diminishing returns, i.e. the performance improvement either levels off or worsens with additional cycles. Possible reasons for the polarization improvements are also explored.
MIECs have been shown to improve overall cell performance as an infiltrant, but not as an electrode's ionic phase. In other words, imparting MIEC characteristics into the YSZ ionic phase should increase polarization of the cell, since by imparting larger electronic conductivity into YSZ requires decreasing the ionic conductivity of YSZ as well. However, the previous observation has not considered the infiltration of electrocatalysts which would increase not only the TPB density, but also expansion of the TPBs into triple phase zones (TPZ) as seen in analogous work on SOFC cathodes. 21 In this manuscript we present and interpret electrochemical measurements using complex impedance spectroscopy (EIS) on symmetric Ni-YSZ/YSZ/Ni-YSZ cells. The variables examined are the infiltrants, number of infiltration cycles and operating conditions. Additionally, this work presents how the ionic conducting phase of the anode active layer interacts with electrocatalyst infiltration and its effect on cell performance. The use of symmetric cells allows isolation of the anode charge transfer processes from not only oxygen electrode processes found in full cells, but also the anode concentration polarization due to the thickness of the anode support.
Microstructural characterizations of pristine and tested cells, equivalent circuit diagrams, and the distribution of relaxation times (DRT) are compared to elucidate how infiltration improves the polarization processes of the cell.

Experimental
Materials synthesis.-A transition metal-doped YSZ (TiYSZ) ionic phase was formed in solid state solution by calcining 3% TiO 2 doped into YSZ. This was done by mixing the appropriate precursors in ethanol and ball milling the resulting slurry. The slurry was dried and calcined at 700°C in air for 9 h. X-ray diffraction (XRD) of powder samples indicate peak shifts of TiYSZ compared to YSZ without Ti doping, indicating complete incorporation of TiO 2 into z E-mail: sgopalan@bu.edu *Electrochemical Society Student Member. **Electrochemical Society Member. the YSZ lattice. The TiYSZ was used in place of YSZ as the ionic conducting phase in the anode active layer (AAL) in some subsequent experiments.
Cell fabrication.-The anode slurry consisted of a 50-50 weight ratio of NiO-YSZ (J. T. Baker, Tosoh) with V6 (Heraeus) as binder and LP1 (Croda) as dispersant dissolved in alpha-terpineol (Alfa-Aesar). Symmetric cells were fabricated by screen printing the cermet (Ni-YSZ or Ni-TiYSZ) onto commercially purchased 8YSZ electrolyte substrates (Fuel Cell Materials) and sintered at 1400°C for 2 h. Cells were then reduced at 800°C under humidified forming gas for 7 h to reduce the NiO to nickel. The reduction step increases the porosity of the scaffold and allows easier infiltration of nanoparticles. The 3 M infiltration solutions containing nickel or GDC were prepared by dissolving precursors containing Ni(NO 3 ) 2 or (Gd(NO 3 ) 3 ) 0.1 (Ce(NO 3 ) 3 ) 0.9 in 50 ml of ethanol at 90°C, respectively. The 3 M Ni-GDC nitrate solution was similarly prepared by adding the nitrate precursors that results in a 1:1 molar ratio of Ni to GDC into an ethanol solvent. The infiltration procedure involved saturating each electrode with infiltrant solution under vacuum. After each round of infiltration, the cells were heated to 320°C to evaporate the solvent and decompose the nitrate salts into metal oxides. This intermediate heating step was necessary as it reopens the pores for additional infiltration. The processing variables studied include (i) the number of infiltration cycles, (ii) the infiltrated electrocatalyst material and (iii) the ionic conducting phase in the AAL. Group A represented cells comprising Ni-YSZ scaffolds, while Group B represented cells comprising Ni-TiYSZ scaffolds. Groups 1, 2, and 3 represented cells infiltrated with metal nitrate solutions which decomposed to result in infiltrants of Ni, GDC, or Ni-GDC, respectively. For example, group A1 represents cells with a Ni-YSZ anode scaffold that was infiltrated with Ni. A visual representation of the cell parameters is shown below in Table I, including the infiltration cycles tested for each cell group.
Electrochemical testing.-Nickel meshes were affixed with nickel paste onto both electrodes of the symmetric cell as electrical contacts. Nickel wire was used as voltage and current leads for the cell and the entire device was placed in a single chamber setup. All cells were heated to 800°C under humidified forming gas to prevent the re-oxidation of nickel. Forming gas was first passed through a bubbler at 25°C to achieve a total flow rate of 300 ml min −1 (92%N 2 /5%H 2 /3%H 2 O) and fed into the electrochemical test chamber. EIS was performed from 600°C to 800°C under a total gas flow rate of 300 ml min −1 and gas compositions of 3% and 50% H 2 O, and balance hydrogen. The desired chamber gas composition was obtained by changing the bubbler temperature. Flow rates into the test chamber were maintained at 300 ml min −1 . The furnace temperature and water bubbler temperature were both monitored before commencing electrochemical measurements. The cell was equilibrated for 30 min after reaching target operating conditions. EIS spectra were acquired using a Parstat 4000 workstation with a frequency sweep of 10 5 −10 −1 Hz with a 50 mV RMS perturbation. Equivalent circuit models were fit using R-RQ-RQ circuit based on previous work to isolate and identify the main polarization processes. 7,8,18 These processes were correlated with the spectra's DRT to compare process time constants and magnitudes.
Scanning electron microscopy (SEM).-Pristine and tested symmetric cells were fractured and their cross sections were coated with 10 nm of Au and contacted with copper tape. The electrodes were imaged using a Zeiss Supra 55 SEM at 15 kV using an In-lens detector at 6-8 mm working distance to obtain nanoparticle deposition and morphologies from fractured cross-sections. Other cross-sections were also vacuum infiltrated with epoxy and polished to compare overall cell porosity and microstructure.
DRT analysis.-The DRT is obtained from convolving EIS data into the time domain, obtaining peaks which are proportional to the cells' polarization resistances centered around different timescales. More information can be found in work by Leonide et al. 7 The equivalent circuit model is used to sharpen the DRT analysis to further elucidate how the kinetics of the various electrochemical processes in the electrodes change with infiltration and different operating parameters.

Results
Electrochemical characterization of Ni-infiltrated cells.- Figure 1 shows the Nyquist plots, corrected for ohmic resistance, of A1 cells with varying infiltrant cycles under 3% humidity (balance hydrogen) at 800°C and 700°C. From the Nyquist plots, infiltration appears to improve cell performance, as seen from the decreasing polarization resistance; these are the magnitudes of the low frequency intercepts on the Nyquist plots. However, multiple infiltration cycles result in diminishing improvements, eventually stagnating overall cell performance. Additionally, infiltration of nickel appears to impart greater cell improvements at lower temperatures. Figure 2 compares relative polarizations of different A1 cells at different operating conditions. In this work, relative polarization is the ratio of the cell polarization when compared to the group's baseline uninfiltrated cell, under identical test conditions. Although all cells showed improvement under all operating conditions after infiltration, there is an optimal number of infiltration cycles (around 3 or 4 in Ni-infiltrated Ni-YSZ) before additional cycles begin to increase polarization resistance. However, all infiltrated A1 cells showed a more modest level of improvement over the uninfiltrated cell at 800°C and 50% humidity compared to other test conditions, implying that infiltration is most effective at lower temperatures and lower humidity levels in the fuel. This is in line with the general observation that at higher temperatures and humidity conditions, all charge transfer processes are more facile. This is also in accord with observations in prior work 22 and will be discussed later in the paper. Figure 3 shows the relative polarizations of B1 cells under the same operating conditions as above. Infiltration improves overall cell performance, however, the optimal number of nickel infiltration cycles is one for Ni-TiYSZ scaffolds. Under all test conditions, one cycle of infiltration shows significant improvement over the baseline cell without the need for additional infiltrations. Figure 4 compares the A1 and B1 cells' relative polarizations, showing uninfiltrated and optimally infiltrated cells with the uninfiltrated A1 cell as the baseline. The increase in polarization resistance for the Ni-TiYSZ scaffold indicates that the cermet alone performs worse than Ni-YSZ cermets, yet infiltration allows the Ni-TiYSZ scaffolds to perform on par with Ni-YSZ scaffolds, with even fewer cycles of infiltration.
Electrochemical characterization of GDC and Ni-GDC-infiltrated cells.- Table II shows total and relative polarizations of GDC and Ni-GDC infiltrated within the Ni-YSZ and Ni-TiYSZ symmetric    with one cycle of infiltration among all cell groups, with the exception of B3 cells. However, all cells infiltrated with some GDC loading showed an increase in performance by at least one order of magnitude.
Microstructural analysis.- Figure 5 shows SEM fractured crosssections of nickel-infiltrated A1 cells. The bottom right number indicates the number of infiltration cycles. On infiltrated cells the infiltrated nanoparticles appear hemispherical and can be seen on the ionic conducting phase, i.e. YSZ or TiYSZ. Although initially infiltration increases nanoparticle density on the scaffold, it appears that the density decreases with additional infiltration cycles, and instead nanoparticle size increases, indicating coarsening of the infiltrated phase. Previous work in our group has shown that although NiO can deposit on the entire Ni-YSZ scaffold, during the reduction process, any Ni deposited on bulk Ni will form a thin film on said bulk phase due to zero surface energy between Ni and Ni. 17 In other words, all rounded Ni nanoparticles are only found to be deposited on the YSZ surface. Secondary electron SEM images presented in Fig. 5 from our prior work 17 clearly show this. Figure 6 shows the SEM cross-sections of GDC and Ni-GDC infiltrated A2 and A3 cells. As before, the number on the bottom right corner indicates the number of infiltration cycles. The GDC morphology is drastically different from Ni nanoparticles. The GDC phase appears to form a thin yet porous film over both the Ni and YSZ phase of the Ni-YSZ scaffold. The Ni-GDC morphology also appears to form a porous thin film on the scaffold, but with more porosity than the GDC infiltrant. This may be due to the infiltrated nickel coalescing into the bulk metal phase to reduce surface energy, and reducing the overall GDC nanoparticle loading. This hypothesis also would explain why nickel infiltration in experiments described earlier showed an absence of nanoparticles deposited on nickel surfaces.

Discussion
Electronic pathways in the active layer.-The infiltration of nanoparticle electrocatalysts is expected to improve the cell performance through the introduction of phases with higher surface-tovolume ratios, and increasing the overall TPB length thereby increasing the overall density of reaction sites in the electrode. However, SEM cross-sections show that many nickel nanoparticles do not make contact with the bulk Ni in the Ni-YSZ scaffold, yet still result in improved performance as seen by the lower polarization resistance in infiltrated scaffolds.
There are two possible mechanisms that could explain this behavior. The first is that all TPBs where they are formed between isolated Ni particles, YSZ, and the pore phase can be active over a certain distance away from the actual TPB line. Thus the TPB line should really be thought of as a three phase zone (TPZ). Depending on the infiltrated particle density, these TPZs can overlap and enable isolated nickel particles to be electrochemically active. This has been observed in Horita's work on the cathodic side. 21 The second possible mechanism for isolated nickel nanoparticles to contribute to electrochemical reaction involves electronic transport through the predominantly ionic conducting phase in the anode active layer, namely YSZ. Even if the deposited particles were farther away from the metal (Ni) phase in the scaffold, the ionic phase with minority electronic charge carriers can shuttle electrons to the bulk percolated metal (Ni) network. YSZ doped with TiO 2 is known to possess a higher electronic conductivity than YSZ. 23 When the B1-group of cells are infiltrated with Ni nanoparticles, they performed as well as the A1 cells with a lower number of infiltration cycles, thereby providing evidence for the second mechanism.
The latter pathway is applicable to GDC infiltrated cells as well. The GDC electronic conductivity is much greater than that of the cermet's ionic phase, greatly extending the TPZs as the infiltrant morphologies cover both nickel and YSZ grains. Previous work has shown that GDC infiltrated into Ni-YSZ symmetric cells has been shown to significantly improve anode charge transfer. [18][19][20] Whereas in these papers GDC infiltration results in hemispherical particles deposited onto the scaffold, the GDC in this work results in a porous film on the scaffold. Additionally, the infiltration of GDC has been found to reduce the polarization resistance in Ni-YSZ scaffolds by around 95%. It is evident that GDC is a much better electrocatalyst than Ni especially at high humidity levels in the fuel. Presumably its excellent mixed conductivity under SOFC anode conditions also plays a role in obtaining very low polarization resistances.
Effects of humidity on nanoparticle electrocatalysts.-In addition to the 3% humidity condition, a 50% humidity condition was also tested to observe the effect of infiltrated nanoparticles under varying humidity conditions. As shown in Table II, infiltration appears to improve the cell performance by decreasing the polarization resistance under both the humidity conditions tested. However, GDC infiltration appears to greatly decrease the polarization resistance under both humidity conditions under which the cells were tested. The mechanistic reasons for the dramatic decrease in polarization resistance in GDC infiltrated electrodes and the behavior as a function of gas composition are still being explored.
Limits to nanoparticle infiltration.-As discussed above, infiltration has improved cell performance, but multiple infiltration  cycles result in diminishing improvements in performance and eventually increase the cell polarization resistance. A DRT analysis of the EIS spectra showed that two main polarization processes contributed to the charge transfer of the cell. Figure 7 compares the DRT analysis performed on EIS spectra obtained from infiltrating of A1 cells at 700 o C and 50% humidity. The DRT analysis suggests the presence of a high frequency (HF) response at ∼2000 Hz and medium frequency (MF) response at ∼100 Hz. These processes are attributed to charge transfer occurring at electrochemically active sites such as the TPBs, and oxygen vacancy transport, respectively. 20 Infiltration appears to initially decrease the magnitude, and increase the frequency of these processes; this suggests the presence of nanoparticles decreases the polarization resistance and increases the rate of electrochemical reactions. However, the addition of further  infiltration cycles led to smaller improvements in peak processes. In fact, after four infiltration cycles, the peaks revert to higher magnitudes and lower frequencies, suggesting an inversion in cell performance.
The explanation for this behavior is that the first nickel infiltration cycle deposits the first nanoparticles onto the YSZ grains, most of which range from 5-50 nm in size. This infiltration cycle introduces a high nanoparticle density in the electrode scaffold, resulting in a high TPB density. Although additional infiltration cycles do increase the nanoparticle density, some particles are in close enough proximity to coalesce. As Ni(NO 3 ) 2 deposits onto preexisting nanoscale NiO particles rather than the ionic conducting phase, the infiltrant loading increases the Ni nanoparticle sizes and aggregates previously isolated Ni particles. The increasing size of the nanoparticles at higher number of infiltration cycles also decreases TPB density and average particle surface-to-volume ratio. The inversion in polarization resistance can be attributed to the overall increase in average nanoparticle size and the decrease in nanoparticle density.
Effect of GDC loading on Ni-YSZ cells.-As in the case of Niinfiltration, infiltration of GDC nanoparticles improves cell performance by decreasing feature sizes and increasing electrochemically active sites; the electrocatalyst appears to dramatically decrease the cells' polarization resistance (by 95%) as seen in Table I. The DRT analysis of GDC-infiltrated cells in Fig. 8 shows almost complete suppression of the main HF and MF polarization peaks. The remaining peaks are seen around 6000 Hz and 40000 Hz. This is seen not only in GDC-infiltrated cells, but in Ni-GDC-infiltrated cells as well. It is unknown whether these remaining peaks are due to processes that were eclipsed by the main charge transfer processes in Ni-YSZ cermets, or if the peaks are attributed to new processes related to the infiltrated GDC electrocatalyst in the Ni-YSZ scaffold.

Conclusions
Ni-YSZ cermet anodes on electrolyte supported symmetric cells were infiltrated with nanoparticle electrocatalysts to improve the electrocatalytic performance of the anode. Aqueous solutions of metal nitrates were repeatedly infiltrated into the reduced Ni-YSZ Figure 7. DRT results showing improvement of the cell performance of group A1 cells at 700 C, 50% humidity with increasing infiltration cycles of nickel (above) and subsequent inversion with additional infiltration cycles(below). thin electrodes on symmetric cells and said cells were electrochemically tested under EIS. Fractured cross-sections showed hemispherical individual nickel particles deposited on YSZ grains, while GDC-infiltration resulted in a thin, porous, film covering both nickel and YSZ grains. The effects of infiltration were also compared by observing changes in the DRT analysis of the EIS spectra of the cells.
Despite nickel nanoparticles not physically contacting the metal network, infiltrated cells exhibited a decrease in polarization resistance under almost all operation conditions tested. Cells were also exposed to a maximum of 50% humidity, well below the threshold for critical nickel wetting on YSZ surfaces. In other words, nanoparticles contributed to the faradaic reaction in the electrodes due to the close proximity to the metal network, either through the active and overlapping region of TPBs of the infiltrated nanoparticles and bulk metal phase or electronic transport in the ionic phase.
These experimental results also show the addition of GDC suppresses the main charge transfer peaks in the EIS spectra of cells with Ni-YSZ scaffolds, greatly improving the performance of the cells. The remaining polarization peaks appear to be high frequency processes, but it is unknown if these peaks are present due to the GDC or if they only appeared due to the suppression of these peaks in cells with pristine Ni-YSZ scaffolds.