Stabilization of Ni-YSZ Anodes in Solid Oxide Fuel Cells using an ALD-Grown Aluminum Titanate Interlayer

The effect of an Al2TiO5 (ALT) interlayer between Ni and YSZ on enhancing the thermal stability of Ni-YSZ solid oxide fuel cell was examined. Atomic layer deposition (ALD) was used to provide precise control of the structure and thickness of the ALT interlayer. The study’s findings demonstrate that a 2 nm thick ALT interlayer deposited by ALD does not adversely affect the cell’s ohmic resistance and effectively prevents Ni sintering and the loss of active area during high-temperature heat treatments. ALT layers thicker than 2 nm, although they enhanced Ni stability, were found to impede oxygen ion transport in the electrode and significantly increase the ohmic resistance of the cell, leading to a decline in performance.

2][3] While Ni-YSZ composite electrodes exhibit good performance and are used in commercial systems, they undergo some degradation in performance over time at typical SOFC operating conditions.This degradation is generally associated with sintering and other morphological changes in the Ni [4][5][6][7][8][9][10] For example, Simwonies et al. observed that the average Ni particle size in the anode increased when they treated a cell at 1273 K for 4000 h. 5 Similarly, Zekri et al. reported that Ni agglomeration occurred in an SOFC anode operating at 1123 K. 4 Iwata also found that after only 20 h of operation at 1200 K the performance of a Ni-YSZ anode decreased significantly.They attributed this to a reduction in the Ni surface area due to sintering of the Ni particles, consistent with previous reports. 66]11 Enhancing the thermal stability of Ni-YSZ anodes is, therefore, a critical requirement to further enhance longterm SOFC performance, especially when operating at temperatures >973 K.3][14][15][16][17] Yokokawa et al. thought that Ni particles could be transported by evaporation/ condensation and diffusion mechanisms and finally aggregated into large Ni particles. 9Simwonies et al. pointed out that over time, spherical Ni particles tend to form and grain coarsening is partially thermodynamically driven by the accompanying decrease in surface free energy. 5The bonding interactions between Ni and underlying YSZ surface play a role in this process.One approach to enhance the structural stability of the Ni that has been explored in several studies is to alter the chemistry at the Ni-YSZ interface by including a thin oxide interlayer that bonds strongly to both the Ni and the YSZ, also referred to as "anchoring."Specific examples of such Ni-anchoring layers include using TiO 2 , 18,19 AlN, 20 and Al 2 TiO 5 (ALT). 21,222][23][24] Law et al. were the first to explore this approach 21 using anodes in which the Ni was added by wet infiltration to a pre-sintered YSZ anode scaffold.
During synthesis of the porous YSZ scaffold they used a mixture of 95 wt% YSZ and 5 wt% ALT powders.After sintering the scaffold at 1723 K they added Ni using multiple cycles of infiltration of a Ni nitrate solution.They then heat treated the cell at 1673 K for 1 h to induce a reaction between Ni and the ALT in the scaffold.Based on XRD data they concluded that the ALT in contact with YSZ reacted to form an interfacial layer of ZrTiO 2 , and that the ALT in contact with Ni reacted to form an interfacial layer of NiAl 2 O 4 which anchored the Ni to the electrode scaffold.While they didn't report polarization curves or EIS data for their cells, current versus time data obtained in the Law et al. study showed slower deactivation for an ALT-modified cell relative to one in which the anode scaffold contained only YSZ.][23] While these studies provide some validation for the approach, the structure of the anodes used in these studies were likely sub-optimal.To optimize both cell performance and anode stability it would be advantageous to selectively place a thin layer of ALT only between the YSZ and the Ni.This was not done in the previous studies, where the ALT was added randomly during synthesis of the YSZ scaffold or during infiltration of the Ni. 21,22In the work reported here, we have attempted to overcome this drawback by using the conformal vapor deposition technique of Atomic Layer Deposition (ALD) to selectively add ALT as a thin layer between the Ni and YSZ, with excellent thickness control.
Our general approach, shown schematically in Fig. 1, was to use ALD to deposit a thin ALT layer on a preformed YSZ anode scaffold, heat treat it at high temperature to induce reaction with the YSZ, followed by infiltrating Ni and heat treating so it would react with the ALT to form NiAl 2 O 4 .The goal was to produce a final anode structure like that shown schematically in Fig. 2. In addition to producing a conformal film of ALT on the surface of the YSZ electrode scaffold, the use of ALD to deposit the anchoring interlayer allowed for precise control of its thickness.The results of this study show that very thin ALT interlayers can dramatically enhance the thermal stability of Ni-YSZ anodes while causing little to no decrease in overall cell performance.

Experimental
The SOFCs used in this study were constructed from porousdense-porous YSZ (8 mol% yttria) scaffolds which were prepared using tape casting as described in detail in previous work. 25,26To produce the scaffold, YSZ tape cast layers were laminated with the z E-mail: vohs@seas.upenn.eduECS Advances, 2024 3 024502 two outer green tapes containing graphite as a pore former.The laminated structure was then calcined at 1673 K (with a ramping rate of 3 K min −1 ) for 4 h producing a dense, 80 μm thick YSZ electrolyte layer (1.5 cm diameter) that was sandwiched between 30 μm thick, 65%-porous layers (0.67 cm diameter).The active components, La 0.8 Sr 0.2 FeO 3−δ (La:Sr:Fe = 4:1:5) for the cathode, and Ni for the anode, were added using multiple cycles of wet infiltration into the porous YSZ electrode scaffolds using metal salt solutions.Each infiltration cycle consisted of adsorbing the aqueous precursor solution containing the desired ratio of metal cations into the porous YSZ electrode layer, followed by drying and then calcining in air at 773 K for 10 min to form the corresponding metal oxide.
For the base case SOFC that did not include the ALT layer, LSF was added to the cathode using 15 infiltration cycles of an aqueous precursor solution containing La(NO 3 ) 3 •6H 2 O (Alfa Aesar, 99.9%), Sr(NO 3 ) 2 (Alfa Aesar, 99%), and Fe(NO 3 ) 3 •9H 2 O (Alfa Aesar, 98.5%)) with a 4:1 la 3+ :Sr 2+ ratio.After this infiltration process, the cell was calcinated at 1123 K for 2 h to induce formation of the LSF perovskite phase.This procedure produced an LSF-YSZ composite cathode that contained 50 wt% LSF.Ni was then added to the anode side of the cell using 15 infiltration cycles of a 4.5 M aqueous Ni (NO 3 ) 2 (Alfa Aesar, 98%) solution.This produced a NiO layer in the anode that was reduced to Ni metal when the cell was heated in H 2 during testing.This cell will be referred to as the pristine cell throughout the remainder of the paper.
For the remaining cells, prior to adding the active components to the electrodes, ALD was used to deposit a conformal ALT layer on the surface of the porous anode.The stoichiometry of ALT was obtained by introducing ALD precursors in super-cycles with equal sub-cycles of the titanium and aluminum precursors.The ALD depositions were caried out with a chamber temperature of 150 °C, resulting in a growth rate of 1.2 nm and 0.88 nm per cycle of Al 2 O 3 and TiO 2 , respectively.The growth rates were determined gravimetrically by growing a film on a YSZ powder and assuming the ALT had its bulk density.The ALT layers were designed such that the TiO 2 layer reacts with YSZ surface to form ZrTiO 4 and the Al 2 O 3 layer reacts with the infiltrated Ni to form NiAl 2 O 4 .Two ALT layer thicknesses were investigated in this study: 2.3 nm and 5 nm which will be referred to as the 2-ALT cell, and 5-ALT cell, respectively.A cell with a 10 nm thick ALT layer was also synthesized and used for X-ray diffraction analysis of the ALT film.The XRD pattern obtained from this cell (Fig. S1) contained a peak at 26.8°2θ which can be assigned to the (110) plane of ALT.
After the conformal ALT layer was deposited, the cell scaffold was calcined at 1673 K (with a ramping rate of 3 K min −1 ) for 1 h.This step was necessary to induce reaction of the TiO 2 in the ALT with the YSZ surface to form an anchoring ZrTiO 4 interlayer.Approximately half of the Ni was then added to the anode side of the cell using 8 infiltration cycles of a 4.5 M Ni (NO 3 ) 3 solution.The cell was then calcined at 1373 K (with a ramping rate of 3 K min −1 ) for 1 h to induce reaction of the Ni with the ALT to form a NiAl 2 O 4 interlayer that served to anchor the Ni to the ALT.LSF was then added to the cathode side of the cell using the same procedure as that for the base case cell.To counterbalance the possible agglomeration of the Ni during the 1373 K calcination step and to ensure a percolated Ni layer in the anode, after adding the LSF an additional 7 Ni infiltration cycles were performed bringing the total number of Ni infiltration cycles to 15, which is the same as that use for the non-ALT modified cell.Figure 3 summarizes the steps that were used to synthesize the ALT modified cells.Several cells were also synthesized using similar procedures as that described here except the hightemperature annealing steps (1673 K and 1373 K) that were used to induce reaction of the ALT with the YSZ and Ni were omitted.
Structural information for the anodes was obtained using XRD and an X-ray energy dispersive analysis (EDS) equipped scanning electron microscope (SEM).The X-ray patterns were collected using a Rigaku MiniFlex diffractometer with a Cu-Ka source (l = 0.15416 nm) and EDS maps were obtained using an Oxford Instruments EDS System on a Hitachi S-4800 Cold-Field-Emission SEM using a 20 kV electron beam.
Ag paste and Ag wires, which have minimal catalytic activity for hydrogen oxidation, were used to make the external electrical  contacts on each electrode.To facilitate testing, each cell was sealed onto the end of an alumina tube using a ceramic adhesive (Aremco Ceramabond 552).Cell performance using H 2 as the fuel was assessed by measuring polarization curves and electrochemical impedance spectra (EIS) at 973 K using a Gamry Instruments impedance spectrometer/potentiostat. For these tests the reactant consisted of humidified (3% H 2 O) H 2 , which was supplied to the anode at a rate of 10 ml min −1 .The LSF-YSZ cathode was exposed to air.

Results
Figure 4 displays Nyquist plots of the EIS data and polarization curves for the pristine, 2-ALT and 5-ALT cells operating at 973 K with humidified H 2 .These data were collected immediately after heating the as-synthesized cells up to the operating conditions with the cathode exposed to air and the anode exposed to humidified H 2 .The Nyquist plot for the pristine cell (Fig. 4a) shows that the area specific ohmic resistance, R Ω , which corresponds to the first intercept with the real Z axis is 0.47 Ω•cm 2 which is close to that expected for the 80 μm thick YSZ electrolyte layer.The area specific polarization resistance for the electrodes, R p , which is given by the difference in the two intercepts of the impedance arc with the real Z axis is 0.45 Ω•cm 2 .In our previous studies we have shown that the R p of an infiltrated LSF electrode similar to that used here is ∼0.1 Ω•cm 2 27 at 973 K; thus, the R p of the infiltrated Ni anode is ∼0.35Ω•cm 2 .The area specific resistances (ASR) values for this and the other cells used in this study are summarized in Table I.
The impedance spectrum for the 2-ALT cell is similar to that for the pristine cell with the only significant difference being a slight increase in R Ω to 0.52 Ω•cm 2 .As noted in the experimental section, the synthesis of this cell included high temperature annealing steps to induce reaction of the ALT layer with both the YSZ and the infiltrated Ni.The fact that the addition of the 2.3 nm ALT layer only caused a 0.05 Ω•cm 2 increase in R Ω demonstrates that this layer did not significantly hinder oxygen ion transport into the electrodes.The R p value for this cell was 0.35 Ω•cm 2 .The similarity in the EIS spectra for the pristine and 2-ALT cells is also reflected in the polarization curves for these cells at 973 K which are nearly identical, as shown in Fig. 4b.
In contrast to the 2.3 nm ALT layer, the 5 nm ALT interlayer adversely affected the performance of the 5-ALT cell.The R Ω and the R p values for the 5-ALT cell were 0.64 Ω•cm 2 and 0.80 Ω•cm 2 , respectively, which are 0.17 Ω•cm 2 and 0.35 Ω•cm 2 higher than the corresponding values for the pristine cell.The increase in the overall ASR (R total ) of the cell is also apparent in the polarization curves where the 5-ALT cell produced significantly less current than the pristine and 2-ALT cells (Fig. 4b).This result clearly shows that the 5 nm ALT acts as a blocking layer and significantly hinders oxygen ion transport across the electrode-electrolyte interfaces.
In order to investigate if the ALT layer was effective in anchoring the Ni and preventing its agglomeration during long-term operation, each cell was heated to 1073 K at a ramping rate of 5 K min −1 , held for 1 h, and then allowed to cool back to the 973 K operating temperature.The performance of each cell was then re-evaluated at this temperature.The EIS and polarization curves obtained after this treatment are displayed in Fig. 5 and the R Ω and R p values are summarized in Table I.The Nyquist plot for the pristine cell (Fig. 5a) shows that while the 1073 K heat treatment had only a small effect on R Ω , causing it to increase by 0.11 Ω•cm 2 compared to that for the cell that had been heated to only 973 K, it caused R p to nearly double to 0.99 Ω•cm 2 .Note that the LSF cathode had been previously heated to 1123 K and in a previous study we have shown that heating a similar LSF/YSZ cathode to 1373 K caused only 0 0.2 Ω•cm 2 increase in its ASR; 28 thus, the large increase in R p that occurred after heating the pristine cell to 1073 K can be mostly attributed to agglomeration of the Ni in the anode resulting in a decrease in the active TPBs of the electrode.The impedance spectrum for the 2-ALT cell, however, shows that it was much more stable and exhibited a significantly smaller increase in the overall cell impedance after heating to 1073 K.For this cell, heat treating at 1073 K caused R Ω to increase by only a 0.03 Ω•cm 2 to 0.55 Ω•cm 2 and R p to increase by 0.26 Ω•cm 2 to 0.61 Ω•cm 2 , relative to the 2-ALT cell that was only heated to 973 K.The R total for the pristine and 2-ALT cells after heating to 1073 K were 1.57Ω•cm 2 and 1.16 Ω•cm 2 , respectively which is reflected in the polarization curves in Fig. 5b.The results for the 2-ALT cell heated to 1073 K are quite interesting and indicate that the ALT interlayer substantially stabilized the Ni layer in the anode and helped prevent Ni agglomeration.
The results for the 5-ALT cell were much less promising and for this cell the 1073 K heat treatment resulted in significant increases in both R Ω (0.84 Ω•cm 2 ) and R p (1.15 Ω•cm 2 ) producing a cell with a R total of 1.99 Ω•cm 2 which is 0.42 Ω•cm 2 higher than that for the 1073 K-treated pristine cell.As noted above, the poor performance of this cell can be attributed to the 5 nm ALT layer being thick enough to hinder oxygen ion transport in the electrodes, compared to 2.3 nm thickness.The increase in R p after heating to 1073 K can again be attributed to partial agglomeration of the Ni in the anode.The origin of the increase in R Ω is less clear but may be due to a decrease in the active area in the electrode.
To further assess the stabilizing effect of the ALT interlayer, the cells were heat treated to 1173 K (ramping rate of 5 K min −1 , 1 h dwell time) and re-tested at 973 K.The EIS spectra and polarization curves for these cells are displayed in Fig. 6 and the R Ω and R p values are summarized in Table I.The data for the pristine cell obtained after this treatment further demonstrates the low thermal stability of the Ni in the anode of this unmodified cell as indicated by the large values of both R Ω (0.89 Ω•cm 2 ) and R p (2.87 Ω•cm 2 ) compared to those for the cell only heated to 973 K.The R total of the pristine cell heated to 1173 K was 3.76 Ω•cm 2 compared to only 0.92 Ω•cm 2 prior to the heat treatments.This again can be attributed to agglomeration of the Ni in anode which reduces the concentration of active TPB sites.
While the higher temperature heat treatment induced some degradation in the performance of the 2-ALT cell, the extent was remarkably less than that of the pristine cell.For the 2-ALT cell the 1173 K heat treatment caused R Ω to increase by only 0.24 Ω•cm 2 to 0.76 Ω•cm 2 and R p to increase by 0.85 Ω•cm 2 to 1.20 Ω•cm 2 .The R total for the 1173 K treated 2-ALT cell was 1.96 Ω•cm 2 compared to 3.76 Ω•cm 2 for the 1173 K treated pristine cell.This result again clearly demonstrates the effectiveness of the ALT interlayer in preventing agglomeration of the Ni in the anode.The results for the 1173 K treated 5-ALT cell are equally interesting.Despite the thicker ALT layer hindering oxygen ion transport into the electrodes, this cell had a R total of only 2.52 Ω•cm 2 and significantly outperforms the heat-treated pristine cell.As would be expected, the polarization curves (Fig. 6b) show the same trends as those described above for the EIS data.Table I.A summary of the area specific resistances (Ω*cm 2 ) for each cell at 973 K.

K
After heating to 1073 K After heating to 1173 K ECS Advances, 2024 3 024502 The performance of a cell with a 2.3 nm thick ALT layer that was not subjected to the high-temperature annealing steps required to induce reaction at the ALT-YSZ and Ni-ALT interfaces was also measured.For operation at 973 K, this cell had an R total of 0.95 Ω•cm 2 which is only 0.08 Ω•cm 2 greater than that of the 2-ALT cell.The stabilizing effect of the ALT layer was not observed for this cell, however, and after treating it at 1173 K for one hour its R total increased to 3.47 Ω•cm 2 which is only slightly less than that of the pristine cell and 1.51 Ω•cm 2 greater than that of the 2-ALT cell after the 1173 K treatment.This result demonstrates that high temperature annealing steps during cell synthesis are necessary to activate the ALT interlayer so that it can anchor the Ni particles.
The enhanced thermal stability of Ni with the activated ALT coating was further confirmed by the cross-section SEM images with EDS mapping of the nickel in the fuel electrode shown in Fig. 7.These images were obtained from cells that were broken in two to allow imaging of the region near the Ni-YSZ interface. of the A key condition for SOFC operation is a percolating Ni network.As can be observed in the Ni EDS map of the pristine, uncoated cell before any heat treatment (leftmost image), the Ni is present throughout the electrode as a porous interconnected layer.However, after the 1173 K treatment (middle image), the Ni has migrated and agglomerated into smaller, disconnected patches.This migration reduces the electrical performance of the cell, as Ni acts as the electrical conductor.The Ni EDS map for the 2-ALT cell (rightmost image) is consistent with the improved EIS and polarization performance observed for this cell and shows that a thin interconnected Ni coating is still present throughout the electrode.Note that while some areas of agglomeration can be observed, this is to be expected as only half of the Ni added to the electrode was properly activated and the other half was added after the heating to activate the anchor.However, critically most of the Ni network remains intact.These results demonstrate that the addition of the thin 2.3 nm ALT coating when properly activated significantly enhances the thermal stability of Ni in a cell with an infiltrated Ni-YSZ electrode.

Discussion
The results obtained in this study provide further demonstration that the addition of an ALT layer between the Ni and the YSZ in a composite Ni-YSZ SOFC anode can significantly increase its thermal stability by preventing agglomeration of the Ni thereby maintaining the concentration of TPB sites.As shown by the R Ω and R p values reported in Table I, the effect of the ALT interlayer can be quite significant.For example, for the pristine cell without the ALT interlayer, heat treating to 1073 K caused the R p measured at 973 K to increase by 0.54 Ω•cm 2 , while that for the 2-ALT cell with the 2.3 nm ALT interlayer it increased by only 0.26 Ω•cm 2 .The effectiveness of the interlayer was even more apparent after heat treating at 1173 K where the corresponding increases in R p relative to those for the cells only heated to 973 K were 2.42 Ω•cm 2 for the pristine cell and only 0.85 Ω•cm 2 the 2-ALT cell.
It is noteworthy that the beneficial effect of the ALD ALT layer was not observed for a cell that was not heat treated at temperatures above 1300 K after deposition of the ALT and prior to addition of all the Ni.This observation is consistent with the XRD results reported  by Law et al. 21and Driscoll et al. 22 which show that such temperatures are required to induce surface reactions of the YSZ with the Ni and with the ALT to form ZrTiO 4 and NiAl 2 O 4 interfacial phases, respectively, and that these compounds play a pivotal role in anchoring the Ni against agglomeration.
Since ALT, as well as ZrTiO 4 and NiAl 2 O 4 , are not likely to be good oxygen ion conductors, one must also consider how the ALT interlayer affects R Ω of the cell and its overall performance.The use of ALD to deposit the ALT layer allowed us to investigate this effect.As shown in Fig. 4 and Table I, the 2.3 nm thick ALT interlayer had only a small effect on R Ω and the pristine and 2-ALT cells heated to 973 K exhibited almost identical performance.The effect of 2.3 nm ALT interlayer on R Ω also did not appear to be very significant for the 1073 K and 1173 K heat treated cells.The 5 nm ALT interlayer, however, clearly adversely affected R Ω and for operation at 973 K before and after heat-treating to 1073 K, R Ω for the 5-ALT cell was significantly larger than that of the pristine and 2-ALT cells.
While the effectiveness of the ALT interlayer in enhancing the stability of Ni in Ni-YSZ composites SOFC anodes has now been firmly established in this and previous studies, the results obtained here show that this layer must be kept relatively thin, less than several nanometers, for it to not significantly increase the R Ω of the cell.The results obtained here show that the ability of ALD to produce a continuous film of uniform thickness in the nanometer regime makes it an ideal technique to both deposit and optimize the structure of stabilizing ALT interlayers that are added to composite Ni-YSZ anodes.

Conclusions
The results obtained in this study provide further demonstration that an ALT interlayer can significantly enhance the thermal stability of the Ni layer in an infiltrated Ni-YSZ electrode.High temperature annealing steps (>1300 K) were found to be required to induce this beneficial effect which is consistent with previous studies that have shown that stabilization of the Ni results from the formation of NiAl 2 O 4 and ZrTiO 2 phases at the Ni-ALT and ALT-YSZ interfaces, respectively, which anchor the Ni to the YSZ.The results of this study also demonstrate that the ALT stabilizing layer needs to be less than approximately 3 nm in thickness to avoid significantly increasing the ohmic losses in the cell and that this can be achieved using ALD to deposit conformal films of ALT on a porous YSZ electrode scaffold.

Figure 1 .
Figure 1.Schematic showing the steps used to produce the anchoring interlayer.

Figure 3 .Figure 4 .
Figure 3. Steps used during the synthesis of cells with an ALT interlayer.

Figure 5 .
Figure 5. (a) Nyquist plots for the indicated cells that had been heat-treated at 1073 K running on humidified H 2 fuel at 0.7 V and 973 K and (b) the polarization curve for each cell.

Figure 6 .
Figure 6.Nyquist plots for the indicated cells that had been heat-treated at 1173 K running on humidified H 2 fuel at 0.7 V and 973 K and (b) the polarization curve for each cell.

Figure 7 .
Figure 7. Cross section SEM image EDS mapping of Ni (green) in (Left) a pristine, uncoated cell before heat treatment, (Middle) a pristine, uncoated cell after heat treatment, and (Right) 2-ALT cell processed to activate the anchoring mechanism after heat treatment.