Functionalized Metal‐Supported Reversible Protonic Ceramic Cells with Exceptional Performance and Durability

Reversible protonic ceramic cells (RePCCs) are limited by several factors, including high cost, poor stability, and insufficient fuel electrode activity toward fuel oxidization/generation reactions. Herein, a novel Ni−Fe metal‐supported RePCC (MS‐RePCC) to address these issues simultaneously is proposed. Specifically, the Ni−Fe support possesses good mechanical strength and thermal compatibility with cermet‐based electrodes/electrolytes, ensuring a facile cell fabrication and robust durability. Density functional theory calculations suggest that Fe in the Ni−Fe support enhances the fuel electrode functional layer by providing additional and more active sites for the electrocatalytic reactions. The as‐fabricated MS‐RePCC at 700 °C achieves an excellent peak power density (PPD) of 586 mW cm−2 and an electrolysis current of −428 mA cm−2 (at 1.3 V). Furthermore, the cell is exceptionally stable, as evidenced by 930 h of fuel cell operation with ultralow degradation (≈0.78% kh−1), and much better than an analogous anode‐supported cell (≈17.78% kh−1). In addition, the cell is stable for 50 h of reversible fuel cell/electrolyzer cycling, further demonstrating the potential of this MS‐RePCC. This article proposes a simple and new approach to enhance the electrochemical activity and durability of RePCC, thereby accelerating the commercialization of this technology.


Introduction
The development of efficient energy storage/conversion technologies is crucial to support the rapid growth of power generation from renewable resources, including solar and wind. [1] Reversible solid oxide cells (ReSOCs) are excellent candidates for that because they can provide highly efficient, scalable, and fuel-flexible energy generation and storage at the grid level. ReSOCs can either function as fuel cells to generate power or as electrolyzers to produce hydrogen. [2] Conventional oxygen-conducting ReSOCs usually operate at high temperatures (700À900 C) to activate oxygen conduction. [3] Working at these hightemperature triggers several issues, including poor durability and high cost of components, including interconnects and sealings. Compared to ReSOCs, reversible protonic ceramic cells (RePCCs) can operate at lower temperatures (500À700 C) due to the lower activation energy of proton conduction in oxides compared to oxygen conduction. [4] Despite the widespread interest in this technology, RePCCs have drawbacks, such as long-term operational degradation, [4d] which limit their large-scale diffusion and application. The literature [5] suggests that a major issue with commonly used Ni/electrolyte cermet supports is poor durability during cycling. [6] Metalsupported ReSOCs (MS-ReSOCs) [7] are promising alternatives to conventional Ni-based anode-supported SOCs. [8] By replacing expensive ceramic materials with inexpensive metals as the support (Table S1, Supporting Information), MS-ReSOCs are characterized by lower cost, higher mechanical strength, better redox tolerance, and more effective thermal cycling ability. [9] Common MS-ReSOC support materials are Ni-based or Fe-based alloys. [9a] Ni-based materials have shown excellent catalytic activity. [10] However, high thermal expansion coefficient (TEC, %16À18 Â 10 À6 K À1 [11] ) and poor redox tolerance [12] limit the long-term operational stability of Ni-based cells. In comparison to Ni, Fe-based alloys (e.g., SUS430, [13] ZMG232, [14] and Crofer22 APU [15] ) are cheaper (Table S1, Supporting  Information) and have a lower TEC value (10À12 Â 10 À6 K À1 [16] ) [17] but are characterized by lower catalytic activity, compromising cell performance. [18] To overcome the challenges outlined earlier, NiÀFe alloy was used as a support to fabricate a metal-supported RePCC (MS-RePCC). This is in contrast with conventional RePCCs, which are cermet supported ( Figure 1a). Specifically, the fabricated full cell consisted of 1) an alloy of Ni and Fe as the support; 2) NiÀBaZr 0.1 Ce 0.7 Y 0.1 Yb 0.1 O 3Àδ (BZCYYb) as the fuel electrode; and 3) BZCYYb as the electrolyte [19] and Sr 0.9 Ce 0.1 Fe 0.8 Ni 0.2 O 3Àδ (SCFN) as the air electrode [20] (Figure 1b). The alloy showed excellent compatibility and synergy with the Ni/electrolyte functional cermet. [21] Further, the alloy was thermally compatible with the fuel electrode, [8] ensuring robust operational durability and easy cosintering during cell fabrication. [22] Fe was also found to diffuse from NiÀFe to the fuel electrode during cell preparation (Figure 2e), providing additional active sites for hydrogen oxidation reaction (HOR)/hydrogen evolution reaction (HER), [23] while simultaneously overcoming the complexity and high cost of metal impregnation. [24] In turn, the addition of Fe to the electroactive layer enhanced the HER kinetics as suggested by DFT calculations, which show that H Ã adsorbs more weakly on NiÀFe than on Ni. The developed MS-RePCC exhibited a high PPD of 586 mW cm À2 in fuel cell mode and a current density of À428 mA cm À2 at 1.3 V in electrolysis mode at 700 C.
Remarkably, in fuel cell mode, the developed MS-RePCC had an ultralow degradation rate of 0.78 % kh À1 during 930 h of operation. As a comparison, the degradation rate of an analogous anode-supported cell (NiÀBZCYYb|BZCYYb|SCFN) prepared by one of the coauthors was measured to be %17.78% kh À1 . [25] Furthermore, the fabricated MS-RePCC could be operated reversibly for 50 h (25 cycles) at 600 C without noticeable degradation.

Physicochemical Properties
The NiÀFe precursors (NiOÀFe 2 O 3 ) were prepared by ball milling NiO and Fe 2 O 3 in a molar ratio of 1:1 ( Figure S1, Supporting Information). X-ray diffraction (XRD) (Figure 1a) analysis showed that the NiÀFe precursor powders, calcined at 1400 C, formed single-phase NiFe 2 O 4 (NFO, JCPDS 54-0964, Figure 1a). NFO was then reduced in H 2 at 700 C for 10 h to form the NiÀFe alloy (NiÀFeÀH 2 , JCPDS 47-1405, Figure 1a). To identify Ni and Fe oxidation states before and after reduction in H 2 , X-ray photoelectron spectroscopy (XPS) was performed ( Figure 1b,c). Ni and Fe in NFO-1400 C were found as Fe 2þ /Fe 3þ and Ni 2þ /Ni 3þ , while the NiÀFe-containing and H 2reduced samples (NiÀFeÀH 2 ) also had noticeable metallic Fe 0 and Ni 0 peaks. [26] XRD analysis of the synthesized SCFN powder ( Figure S2, Supporting Information) suggests that, in agreement with previous studies, SCFN consisted of two major perovskite phases (T-SCFN and RP-SCFN) and two minor oxide phases (NiO and CeO 2 ). [22,25] The cross section of the reduced MS-RePCC was imaged by scanning electron microscopy (SEM), as shown in Figure 2d. The porous NiÀFe alloy support layer ( Figure S3, Supporting Information) was 650 μm thick and made tight contact with the 90 μm-thick NiÀBZCYYb layer. The thicknesses of the dense BZCYYb electrolyte and porous SCFN cathode layers were 30 and 35 μm, respectively. Energy-dispersive X-ray spectroscopy (EDS) indicates that the cell had a multilayer structure (Figure 2e). It is particularly important to note that Fe diffused into the fuel electrode from the support after cosintering, as highlighted by the yellow dotted line in Figure 2e and the cross-sectional EDS line scans shown in Figure S9, Supporting Information.

Sintering Behavior
To allow the diffusion of reactants and products and achieve high performance, the MS-RePCCs' electrodes and supports need to be porous. Furthermore, all of its components need to be thermally compatible. [3] Figure 3a compares the diameter of NFO, NiOÀBZCYYb, and BZCYYb pellets after 1 h of sintering in the 1200À1500 C range. As the temperature increased, all three samples shrank with similar compactification rates, implying that NFO, BZCYYb, and NiOÀBZCYYb are likely to experience negligible stresses during cosintering. It is worth noting that even after high-temperature sintering, NFO maintained a porosity in the 62À48% range ( Figure S4, Supporting Information). The TEC is another important parameter used to evaluate the thermal compatibility among MS-RePCC components. [27] The TEC of NFO estimated from high-temperature XRD ( Figure S5, Supporting Information) is 11.97 Â 10 À6 K À1 in the 100À800 C range (Figure 3b). This value is close to those of NiOÀBZCYYb (10.42 Â 10 À6 K À1 , close to reported literature [28] ) and BZCYYb (9.82 Â 10 À6 K À1 , close to reported literature [29] ), demonstrating that the support has excellent compatibility with other components. It is worth noting that the TEC value of NFO is also similar to that of other proton ceramic conductors and other reported NiÀFe alloys (i.e., 10.2 Â 10 À6 K À1 of BaZr 0.75 Y 0.25 O 3Àδ , [30] 11.2 Â 10 À6 K À1 of BaCeO 3-δ , [31] 11.6 Â 10 À6 K À1 of BaCe 0.8 Y 0.2 O 3Àδ , [32] and Table S6, Supporting Information).
To characterize the Fe diffusion, two pressed NiOÀBZCYYb pellets were prepared, one of the pellets was used as control and the other was placed above powders of NFO ( Figure S10, Supporting Information). After sintering (at 1400 C for 10 h) and polishing, the two pellets were analyzed using EDS, XRD, and XPS. Fe was detected only in the NFOÀBZCYYb sample, as evidenced by EDS ( Figure S11 , Supporting Information), XPS ( Figure S12, Supporting Information), and XRD ( Figure S13, Supporting Information).

Electrochemical Performance
A cell with a NiÀFe|NiÀBZCYYb|BZCYYb|SCFN configuration was prepared (Figure 2c). Before testing, H 2 was fed into the fuel electrode of the cell for 10 h at 700 C so that NFO could be reduced to NiÀFe alloy and NiOÀBCZYYb to NiÀBCZYYb. During fuel cell testing, the anode was fed H 2 , while the cathode was exposed to ambient air. As shown in Figure 4a, the  MS-RePCC had PPDs of 586, 435, 320, 222, and 139 mW cm À2 at 700, 650, 600, 550, and 500 C, respectively. The performance exceeds that of many previously reported anode-supported PCFCs, as shown in Table S2, Supporting Information. In addition, the low impedance measured by electrochemical impedance spectroscopy (EIS) at the open-circuit voltage (OCV) confirms low cell polarization ( Figure S6, Supporting Information). It is worth noting that the MS-RePCC was exceptionally stable for about 930 h under H 2 /ambient air operation at a constant current density of 300 mA cm À2 at 600 C. The calculated degradation rate of the fuel cell is only %0.78% kh À1 , a value significantly lower than those reported in the literature (Table S3, Supporting Information) and within the range of commercial SOFC (0.2À1.6% kh À1 ). [33] We must stress again that an analogous anode-supported cell (NiÀBZCYYb| BZCYYb|SCFN) previously prepared by one of the coauthors of this article could only be operated for %300 h and had a far higher degradation rate of %17.78% kh À1 . [25] Such a significant difference highlights the excellent stability of the MS-RePCC developed here. The fabricated MS-RePCC was also tested for reversible performance. The IÀV curves were measured in the temperature range of 600À700 C by feeding pure H 2 [25,34] and air with 3 vol% H 2 O to the fuel and air electrodes, respectively (Figure 4c). It can be observed that the current densities at 1.3 V were À428, À331, and À124 mA cm À2 at 700, 650, and 600 C, respectively. Electrolysis performance exceeded that of many previously reported anode-supported cells (Table S4, Supporting Information). Cell reversibility was evaluated by cycling between 1.3 V (electrolysis mode) and 0.9 V (fuel cell mode) at 600 C. The cell operated stably for 50 h (25 cycles) at 600 C (Figure 4d). The MS-RePCC's performance also demonstrates the potential of NiÀFe support in water splitting in RePCCs.

DFT Calculations
We carried out DFT calculations to elucidate the contribution of the NiÀFe support to HER ( Figure 5). In particular, we computed the H Ã adsorption energy on the Ni(111) surface. As shown in the literature, H Ã adsorption on Ni(111) can occur at four different sites (i.e., face-centered cubic (FCC), hexagonal close packed (HCP), bridge, and top) with the fcc being the most favorable [35] ( Figure S7, Supporting Information). For NiÀFe, the (001) termination is the most stable. [36] In that termination, H Ã adsorption can occur at three different sites (i.e., Fe top, Ni top, and bridge) ( Figure S8, Supporting Information). H Ã adsorption on NiÀFe (001) is favored at the bridge site with a free energy (ΔG HÃ ) of À0.11 eV. In contrast, the ΔG HÃ on the Ni (111) is more negative (À0.28 eV), implying a stronger interaction between Ni and H Ã . [37] To unveil the mechanisms leading to the different adsorption energies of H Ã between the substrates, we further calculated the Bader charge, listing the resulting values in Table S5, Supporting Information. The weaker interaction between NiÀFe (001) and H Ã can be attributed to the charge transfer from surface Fe to surface Ni in NiÀFe (001), where a negatively charged Ni enhances the repulsion between the slab and the adsorbed H Ã . The weakened adsorption of H Ã decreases the energy required for desorption on NiÀFe (001), as shown in Figure 5c. Therefore, HER on the NiÀFe surface is expected to be more favorable than on pure Ni, in agreement with the excellent electrochemical performances of the metal-supported cell shown in Figure 4c. www.advancedsciencenews.com www.advenergysustres.com

Conclusion
This work develops a novel NiÀFe support for high-performance and exceptionally durable MS-RePCC in the NiÀFe| NiÀBZCYYb|BZCYYb|SCFN configuration. This NiÀFe alloy support exhibited similar sintering shrinkage characteristics and TECs to BZCYYb and NiO-BZCYYb, implying facile cell fabrication and long-term stability. The fabricated MS-RePCC achieved a high PPD of 586 mW cm À2 at 700 C in fuel cell mode, while exhibiting exceptional operational stability of about 930 h with an ultralow degradation rate of %0.78% kh À1 . In reversible tests, the fabricated MS-RePCC has the current densities of À428 mA cm À2 at 1.3 V at 700 C, while retaining robust reversible cycling of 50 h in 25 fuel cell/electrolysis cycles at 600 C. Furthermore, DFT calculations showed that Fe self-diffusion from the support to the fuel electrode facilitates HER. This study provides a promising direction for the development of future commercial RePCCs with exceptional performance and durability.

Experimental Section
Materials Synthesis and Cell Fabrication: NiO and Fe 2 O 3 (molar ratio of 1:1) with 20 wt% extra corn starch (AR, Sinopharm Chemical Reagent Co., Ltd) were first mixed and ball milled (QM-3SP04) for 24 h and then dried overnight. The BZCYYb electrolyte material was synthesized using a conventional solid-state reaction method. [21] NiOÀBZCYYb was prepared by ball milling 45 wt% of NiO and 55 wt% of BZCYYb. An extra 10 wt% of corn starch was added. SCFN was synthesized by a combined EDTAÀCA complexing method described in the study by Wang et al. [22] Metal-supported half cells in the NiOÀFe 2 O 3 |NiOÀBZCYYb|BZCYYb configuration were prepared by copressing and cosintering and then sintered at 1400 C for 10 h in air to densify the electrolyte. The SCFN slurries were finally sprayed onto the surface of the BZCYYb electrolyte with an active area of 0.4 cm 2 and then sintered in air at 900 C for 5 h. More details are given in the Supporting Information.
Electrochemical Measurements: IÀV data were obtained using a Keysight B2901A source meter. Before the test, H 2 was fed into the support side for 10 h of in situ reduction at 700 C. The MS-RePCC cells were tested between 500 and 700 C (for fuel cell testing) and 600 and 700 C (for reversible cell testing). Potentiostatic EIS was carried out at the OCV using a VSP Biologic workstation. More details are given in the Supporting Information.
Characterizations: The crystal structures were analyzed by XRD (Philips X'Pert) with Cu Kα radiation (λ ¼ 1.5406 Å, 40 kV, 40 mA). The scanning angle 2θ ranged from 20 to 80 with steps of 0.02 . The TEC was calculated from high-temperature XRD carried out in the 100À800 C range. XPS (PHI 5000 VersaProbe) was performed on an Axis Ultra DLD instrument with an achromatic Mg Kα X-ray source. SEM characterization was performed on a JEOL-6390 instrument, and EDS was conducted on a Bruker QUANTAX 70 instrument.
DFT Calculations: All spin-polarized DFT calculations were performed using the Vienna ab initio simulation package [38] (VASP) with the projectoraugmented wave (PAW) method. [38,39] The PerdewÀBurkÀErnzerhof (PBE) functional under the generalized gradient approximation (GGA) scheme was used to handle electron-exchange correlations. [40] The energy cutoff for the planewave basis set was 450 eV. A vacuum layer along the out-plane direction of 15 Å was constructed to limit the interactions between adjacent slabs. The Brillouin zone for Ni(111) and NiÀFe(001) was sampled by a 4 Â 4 Â 1 Monkhorst-Pack k-point mesh. The Broyden method was employed for the structural relaxation until the atomic force maximum was below 0.02 eV Å À1 and the relative energies converged within 10 À5 eV. The catalytic activity for HER was correlated to the hydrogen adsorption Gibbs free energy, ΔG H Ã , as proposed by Noskov et al. [41] ΔG H Ã can be defined as where the hydrogen adsorption energy, ΔE H , is given by and E H Ã and E slab are the total energies of the catalyst with and without a hydrogen atom, respectively. In this study, ΔE ZPE À TΔS H (zero-point energy ΔE ZPE , temperature T, and entropy ΔS H ) was selected as 0.24 eV according to the literature. [42] Supporting Information Supporting Information is available from the Wiley Online Library or from the author.