Naturally diffused sintering aid for highly conductive bilayer electrolytes in solid oxide cells

A strategy of naturally diffused sintering aid enables the fabrication of defect-free bilayer electrolyte in solid oxide cells.

Abstract

Solid oxide cells (SOCs) are promising sustainable and efficient electrochemical energy conversion devices. The application of a bilayer electrolyte comprising wide electrolytic oxide and highly conductive oxide is essential to lower the operating temperatures while maintaining high performance. However, a structurally and chemically ideal bilayer has been unattainable through cost-effective conventional ceramic processes. Here, we describe a strategy of naturally diffused sintering aid allowing the fabrication of defect-free doped-zirconia/doped-ceria bilayer electrolyte with full density and reduced interdiffusion layer at lower sintering temperature owing to the supply of small but appropriate amount of sintering aid from doped zirconia to doped ceria that makes the thermal shrinkages of both layers perfectly congruent. The resulting SOCs exhibit a minimal ohmic loss of 0.09 ohm cm² and remarkable performances in both fuel cell (power density exceeding 1.3 W cm⁻²) and electrolysis (current density of −1.27 A cm⁻² at 1.3 V) operations at 700°C.

INTRODUCTION

The distinguished features of solid oxide cells (SOCs) over competing technologies are the high conversion efficiency attained from the high-temperature operation (≥800°C) and bifunctionality, which enables electricity generation in accordance with the power demand and fuel production using idle power (1–3). Despite such benefits, the high manufacturing costs of SOCs have hindered their commercialization (4). For cost reduction, the development of high-performance SOCs operating at reduced temperatures (≤750°C) has been widely pursued (5) because a lower operating temperature enables the use of less expensive components in the SOC stack, and a higher performance can reduce the number of cells required for the designated output of the stack.

Particularly in recent years, bilayer electrolytes have been explored as a potent configuration to improve the electrochemical performance of SOCs even at lower temperatures owing to their synergistic effects (6); e.g., the yttria-stabilized zirconia (YSZ)/gadolinium-doped ceria (GDC) bilayer electrolyte exhibits not only chemical stability with highly active oxygen electrodes and higher ionic conductivity relative to the YSZ single-layer electrolyte owing to GDC but also secure electrolytic property in wide range of oxygen partial pressures owing to YSZ. In recognition of such benefits, bilayer electrolytes are adopted in most SOCs for now. However, to realize the synergistic property of a bilayer electrolyte, both layers should be of full density with a physically and chemically defect-free interfacial structure, i.e., tight adhesion without residual pores or resistive phases at the interface between them (7). Unfortunately, whereas a high sintering temperature above 1400°C is typically required to achieve full density of both YSZ and GDC electrolytes, such a high temperature induces the formation of resistive phases by interdiffusion between YSZ and GDC (8, 9). To suppress this, a two-step approach is conventionally used; the fuel electrode/YSZ assembly is sintered, and subsequently, GDC is deposited on YSZ and sintered at a lower temperature below 1250°C. However, the low sintering temperature and constrained condition of GDC on the rigid YSZ substrate result in considerable porosity in GDC (fig. S1). In such cases, since GDC cannot perfectly play a role of electrolyte, it is typically called a buffer layer only to limit the chemical reaction between the oxygen electrode and YSZ. Alternatively, various thin-film deposition techniques have been used to fabricate the dense and extremely thin electrolytes at relatively low temperatures (10, 11); however, their application toward industrial acceptance is still questionable considering the initial equipment cost and scale-up capability.

The use of sintering aids to promote the intrinsic sinterability of electrolytes, thereby achieving full density below 1250°C, is simple but the most effective approach (12–14). However, the direct addition of a sintering aid to GDC poses substantially potential risks such as the generation of several processing flaws and deterioration of electrical properties. Since even a small amount of sintering aid tremendously influences the sinterability of GDC (15), e.g., only 0.7 mole percent (mol %) Fe addition decreases the sintering temperature by approximately 200°C (the details will be discussed below), the direct addition of a sintering aid is highly likely to induce much more rapid shrinkage of GDC than YSZ, thereby generating various defects such as vertical and interfacial pores, microstructural anisotropy, and delaminated interfaces during cosintering of bilayer electrolyte. Moreover, sintering aids are known to accelerate the chemical reaction with adjacent layers, which may lead to the formation of resistive phases (13, 16), and incur an additional increase in the electrical resistivity owing to their precipitation at grain boundaries (17) or incorporation into the lattice (18). Therefore, only a handful of studies report SOCs fabricated through cosintering process with the use of sintering aid, and even the ionic conductance of bilayer electrolyte is not satisfactory (12, 19).

Here, we present a novel strategy of a naturally diffused sintering aid for the fabrication of a defect-free bilayer electrolyte in SOCs via cost-effective ceramic process. Both the layers of YSZ with a small amount of sintering aid and GDC without a sintering aid can be simultaneously densified to almost full density at sufficiently low temperature (1250°C) owing to an internal supply of the sintering aid from YSZ to GDC during the cosintering process. The amount of sintering aid supplied to GDC is not only appropriate for closely matching the sintering behavior of GDC to that of YSZ but also small enough to minimize the influence on electrical properties. The SOCs fabricated through this strategy exhibit exceptionally low ohmic loss, remarkable electrochemical performances, and durable long-term operations in both fuel cell (FC) and electrolysis modes.

RESULTS

Concept of a naturally diffused sintering aid in a bilayer electrolyte

To assess the intrinsic sinterability of YSZ and GDC influenced by sintering aids, we analyzed the in situ linear shrinkages of YSZs and GDCs with and without sintering aids (Fig. 1A). Considering that a linear shrinkage above 17% is typically required to achieve a high relative density (≥95%) when the initial powder packing density is 50 to 55%, YSZ and GDC cannot be sufficiently densified at 1250°C unless a sintering aid is used (fig. S2). For GDC, additives even less than 1 mol % (0.7 mol % Fe or 0.3 mol % Cu) notably enhance the sinterability of GDC, thereby resulting in a large deviation from the shrinkage behavior of YSZ (Fig. 1A and fig. S3). This implies that matching the shrinkage behaviors by controlling the amount of sintering aid in GDC is technically impractical, and such mismatch subsequently induces generation of severe processing defects; we prepared two SOCs with the configuration of NiO-YSZ fuel electrode support/thin (2 mol % Fe-added YSZ/0.7 mol % Fe- or 0.3 mol % Cu-added GDC) bilayer, in which the GDC materials with a sintering aid exhibit higher sinterability than 2 mol % Fe-added YSZ, and then sintered them at 1250°C. While the bilayer electrolyte was fully densified in both SOCs, critical vertical cracks were observed (fig. S4). Since the densification of GDC with sintering aids is almost complete at approximately 1100°C, the following densification of YSZ above 1100°C imposes compressive stress on GDC, thereby producing vertical cracks through both layers, which possibly originate from the GDC surface and subsequently propagate to YSZ layer.

Fig. 1. Thermal sintering behavior of the bilayer electrolyte.

(A) In situ linear shrinkage of YSZ (black continuous), 2 mol % Fe-added YSZ (black dotted), GDC (red continuous), 0.7 mol % Fe-added GDC (red dashed dotted), and 2 mol % Fe-added GDC (red dotted) as a function of temperature. The vertical gray dotted line is a visual guidance corresponding to the temperature of 1250°C. (B) Schematic illustration of the Fe sintering aid naturally diffusing from underlying 2 mol % Fe-added YSZ (white) to GDC (ivory) during the cosintering process of the bilayer electrolyte. The upper three schematics show cross-sectional microstructures of the bilayer electrolyte before (left), during (middle), and after (right) thermal sintering. Red spheres indicate the sintering aid. (C to F) STEM images (top) and EELS maps (bottom) of 2 mol % Fe-added GDC after heat treatment at 900°C (C) and 1000°C (D) and 2 mol % Fe-added YSZ after heat treatment at 900°C (E) and 1000°C (F). (G) In situ linear shrinkage rates of 2 mol % Fe-added YSZ (black), GDC (red), and the 2 mol % Fe-added YSZ/GDC bilayer along the out-of-plane direction (blue). (H) Schematic illustration of the fabrication of fuel-electrode support/(YSZ/GDC) bilayer electrolyte through conventional ceramic processes. (I and J) Comparison of the cross-sectional microstructures of cermet fuel electrode/(YSZ/GDC) electrolyte without a sintering aid (I) and cermet fuel electrode/(2 mol % Fe-added YSZ/GDC) electrolyte (J) after sintering at 1250°C.

Figure 1B shows schematics of the use of a naturally diffused sintering aid as a novel strategy enabling the supply of a small but appropriate amount of sintering aid from the underlying YSZ to GDC layer during the cosintering process. Notably, we chose Fe as the sintering aid because it promotes the sinterability of both YSZ and GDC with relatively minimal influence on their electrical properties (15, 20). Fe can be dissolved into YSZ above 1000°C (21) by substituting Zr, which has a similar ionic radius (Fe³⁺, 0.78 Å; Zr⁴⁺, 0.84 Å), while it is nearly insoluble in GDC owing to the large difference in ionic radii (Ce⁴⁺, 0.97 Å) (14). Therefore, different mechanisms have been suggested to dominate the sintering behaviors of Fe-added YSZ and Fe-added GDC; for YSZ, the Fe substituent promotes the diffusion of the Zr⁴⁺ ion (22), while that for GDC is Fe₂O₃, which is probably in the form of a thin amorphous film reduces the interparticle friction, leading to facile particle rearrangement, the so-called “viscous flow mechanism” (23). To verify the suggested sintering mechanism, we performed scanning transmission electron microscopy coupled with electron energy loss spectroscopy (STEM-EELS) mapping for 2 mol % Fe-added YSZ (hereafter referred to as “Fe-YSZ”) and 2 mol % Fe-added GDC (hereafter referred to as “Fe-GDC”) in transient states in the sintering process; each powder compact was heat treated at 900° and 1000°C and then quenched and ion milled (fig. S5). As expected from the shrinkage behavior of Fe-GDC (Fig. 1A) and the suggested viscous flow mechanism, GDC was not fully covered with Fe at 900°C (initial stage of sintering; Fig. 1C) but was fully covered at 1000°C, and some of Fe was segregated at grain boundaries (intermediate stage; Fig. 1D). Unexpectedly, YSZ was almost covered with Fe even at 900°C (Fig. 1E), implying that the viscous flow mechanism partially contributes to the accelerated sintering of Fe-YSZ. At 1000°C, Fe started being incorporated into the YSZ bulk (Fig. 1F). Since the transition metal oxide in the form of a thin amorphous film around the particle can readily diffuse along the surface, as observed by many investigators (23), this result suggests the possibility that in the Fe-YSZ/GDC bilayer electrolyte, Fe in YSZ can diffuse to GDC at 900° to 1000°C and then accelerate the densification of GDC via the viscous flow mechanism.

To verify this, we assessed the shrinkage behavior of the Fe-YSZ/GDC bilayer and compared the results with those of Fe-YSZ and GDC single components (Fig. 1G). For the Fe-YSZ/GDC bilayer, a single plateau between two plateaus corresponding to the Fe-YSZ and GDC single components was observed at ~1200°C. This result implies that the plateaus at ~1150°C in Fe-YSZ and ~1250°C in GDC become closer to 1200°C owing to the loss of Fe in Fe-YSZ and uptake of Fe into GDC, respectively. We exemplified this strategy in SOC fabrication through conventional cost-effective ceramic processes, tape casting, roll calendering, and cosintering at 1250°C (Fig. 1H). While the SOC without sintering aid addition to either YSZ or GDC showed porous and weak interfacial microstructures in the bilayer electrolyte (Fig. 1I), the SOC comprising the Fe-YSZ/GDC bilayer exhibited very thin (~7 μm), fully densified (>99%, fig. S6), and reliable interfacial structures (Fig. 1J), verifying the concept of a naturally diffused sintering aid. Notably, the improved interfacial structure of fuel electrode/Fe-YSZ (Fig. 1J) compared to fuel electrode/YSZ (Fig. 1I) is possibly because the Fe sintering aid diffuses not only to GDC but also to the NiO-YSZ fuel electrode.

Examination of the Fe-YSZ/GDC bilayer electrolyte

Solid solution phases of YSZ and GDC, e.g., Ce_0.37Zr_0.38Gd_0.18Y_0.07O_1.87, which have more than 10 to 40 times lower ionic conductivity than YSZ (8, 24), are known to form at the interface of YSZ/GDC above 1200°C, and a higher temperature expedites the formation thereof owing to the faster interdiffusion kinetics (9). In addition, the use of sintering aids within YSZ and/or GDC tends to accelerate the formation of such unfavorable phases at the interface owing to the generation of active transient phases during the sintering process, which also accelerate the cation diffusion kinetics (13, 16). Therefore, it is expected that the lower sintering temperature and the use of lower amount of sintering aids will be effective to suppress the formation of interdiffusion layer. For example, while the thicknesses of the interdiffusion layer in the YSZ/GDC bilayer (without a sintering aid) after cosintering at 1300° and 1400°C have been reported to be ~0.5 and ~1.1 μm, respectively (9), the 2 mol % Fe-added YSZ/2 mol % Fe-added GDC bilayer exhibited much thicker interdiffusion layers of ~0.5 and ~2.0 μm after cosintering at 1250° and 1400°C (12), respectively. In this regard, our Fe-YSZ/GDC bilayer (Fig. 1J), cosintered at a relatively low temperature of 1250°C by using a small amount of sintering aid (2 mol % Fe only in YSZ), is expected to have a thinner interdiffusion layer than previous reports. To verify this, we first investigated the crystalline structure and composition using TEM and energy-dispersive spectroscopy (EDS). The cross-sectional TEM image of Fe-YSZ/GDC (Fig. 2A) showed that YSZ and GDC were almost fully densified, as verified by scanning electron microscope (Fig. 1J and fig. S6). The grain size of GDC in the bilayer (0.57 ± 0.06 μm) was identical to that in the 2 mol % Fe-added GDC bulk sample after sintering at 1250°C (0.63 ± 0.06 μm; see fig. S2D) within the error, indicating that the naturally diffused Fe from YSZ to GDC was sufficient to enhance the sinterability of GDC while its absolute amount was small (which will be shown next). The lattice parameters of YSZ and GDC grains, far from the Fe-YSZ/GDC interface (regions “1” and “4” marked by yellow squares in Fig. 2A) and adjacent to the interface (“2” and “3” in Fig. 2A), were calculated using the diffraction patterns (Fig. 2B) that were converted from high-resolution TEM (HR-TEM) images in the selected regions (fig. S7) via fast Fourier transformations (FFTs). The results showed that the lattice parameters in the regions far from interface and adjacent to the interface were identical within the error in both YSZ (1 versus 2) and GDC (3 versus 4) cases, suggesting that the interdiffusion layer was sufficiently thin less than hundreds of nanometers. From EDS analysis of the interfacial region corresponding to “5” in Fig. 2A, the interdiffusion layer was found to have a thickness of ~0.3 μm (Fig. 2C and fig. S8), verifying that the low cosintering temperature and small amount of sintering aid effectively suppress the formation of interfacial solid solution phases.

Fig. 2. Selected characteristics of the bilayer electrolyte.

(A) TEM micrograph of the 2 mol % Fe-added YSZ/GDC bilayer electrolyte after sintering at 1250°C (Fig. 1E). (B) Diffraction patterns in the selected regions [1 to 4 marked by yellow squares in (A)] obtained via FFTs. The given numbers are the lattice parameters calculated from the corresponding diffraction patterns. (C) EDS maps for Gd, Ce, Y, and Zr at the Fe-YSZ/GDC interface to examine the interdiffusion (ID) layer. (D to I) Fe composition across the GBs in Fe-YSZ [(D) to (F), region 6 in (A)] and GDC [(G) to (I), region 7 in (A)] measured by APT. TEM micrographs of needle-shaped samples before APT analysis (D and G), APT 3D reconstruction of atomic Fe species (E and H), and concentration contour maps for Fe in which the red and purple colors correspond to higher and lower concentrations, respectively (F to I).

In addition to the results on the microstructural (Figs. 1J and 2A) and shrinkage (Fig. 1G) features that phenomenologically verify the naturally diffused Fe from YSZ to GDC, we performed atom probe tomography (APT) to directly observe and quantify Fe element in YSZ (“6” in Fig. 2A) and GDC (“7” in Fig. 2A) over regions in which a grain boundary (GB) is traversed. Since APT, in principle, enables us to resolve the chemical identities of individual atoms in three dimensions with near-atom resolution (25), the distribution and concentration of Fe, particularly in GDC, can be analyzed beyond the elemental sensitivity limit of EDS. Figure 2 (D to I) summarizes the APT results for YSZ and GDC samples with the shape of a sharp needle. The two-dimensional contour maps (Fig. 2, F and I) extracted from three-dimensional atom probe data (Fig. 2, E and H) showed the Fe distributions: deep accumulation at the GB in GDC [GB, 0.93 atomic % (at %); bulk, 0.46 at %] and light accumulation at the GB in YSZ (GB, 2.46 at %; bulk, 1.90 at %), which are reasonable in accordance with the solubility limits of Fe in YSZ and GDC at 1250°C (≥5 at % and negligible, respectively; fig. S9 and movies S1 and S2) (26). The results not only directly confirm that Fe naturally diffused into GDC in a small amount (average concentration, 0.47 at %) but also suggest that the GDC in the bilayer could retain its intrinsic electrical properties owing to the small concentration of Fe sintering aid and lower cosintering temperature, because the solubility of Fe in GDC decreases with temperature and the resistance of GB, wherein most Fe is segregated, is relatively negligible at the operating temperatures of SOCs (17). Significantly, microstructural uniformity in the bilayer electrolyte was achieved over the entire area of the SOC (fig. S10), implying that the approach via a naturally diffused sintering aid can readily achieve uniform distribution of the sintering aid even for a small amount thereof and in turn suppress the processing defects during the sintering process.

Electrochemical evaluation of SOC

We then completed the fabrication of SOCs by incorporating PrBa_0.5Sr_0.5Co_1.5Fe_0.5O_5+δ (PBSCF) as an oxygen electrode, which exhibits high electronic conductivity and good surface redox reaction activity, superior to those of La_0.6Sr_0.4Co_0.8Fe_0.2O_3−δ (LSCF) (27, 28) and comparable to those of La_0.5Sr_0.5CoO_3−δ (29, 30) and Ba_0.5Sr_0.5Co_0.8Fe_0.2O_3−δ (31, 32) conventional oxygen electrode materials (fig. S11). To achieve secure interfacial adhesion between GDC and PBSCF, which have a large difference in thermal expansion behavior [GDC (33), ~12 × 10⁻⁶ K⁻¹; PBSCF (34), 21.3 × 10⁻⁶ K⁻¹], a GDC-PBSCF composite with a weight ratio of 50:50 was inserted between them. As a result, good adhesion at all interfaces was achieved while retaining sufficiently porous structures for the efficient gas transport and electrochemical electrode reactions (fig. S12).

To illustrate the functional capability of the bilayer fabricated with a naturally diffused sintering aid, the electrochemical performance and long-term durability of SOCs operating in FC, electrolysis cell (EC), and reversible cell (RC) modes were evaluated. Notably, all the electrochemical tests were performed using metal interconnectors (Inconel 600 alloy), a glass-ceramic sealant, and metal foams (Ni foam at the fuel electrode side and Cu-Mn foam at the oxygen electrode side), not ceramic housings and/or noble metal current collectors, to demonstrate the environments of the actual SOC stack (fig. S13). We characterized the FC with 3% H₂O/H₂, EC with 50% H₂O/H₂, and RC with 50% H₂O/H₂ supplied to the fuel electrode, while dry air was supplied to the oxygen electrode in common. In FC mode, the high open-circuit potentials of 1.078 and 1.097 V at 800° and 700°C, respectively, which are close to the theoretical values (1.103 and 1.120 V), ensured the fully densified bilayer electrolyte, indicating perfect gas tightness and a high transference number of oxygen ions close to unity. The current-potential characteristics (Fig. 3A) represented remarkable performance; e.g., the power density was 1.27 W cm⁻² at 0.7 V and 700°C, which is superior to the values reported for cells using the PBSCF-GDC composite oxygen electrode on the YSZ electrolyte (0.72 W cm⁻² at 0.7 V and 700°C) (35) and LSCF-GDC/LSCF oxygen electrode on Fe-YSZ/GDC bilayer electrolyte (0.85 W cm⁻²) (36). We then examined the durability of the FC at a constant current density of 0.3 A cm⁻² (Fig. 3B) and found that the cell was stable for a prolonged period of 310 hours at 700°C. In EC mode, at a thermoneutral voltage for steam electrolysis (1.3 V; Fig. 3C), current densities of −2.75 and −1.27 A cm⁻² corresponding to H₂ productions of 1.26 and 0.58 liter hour⁻¹ cm⁻² at 800° and 700°C, respectively, were observed. These values are higher than those recently reported for the EC with the identical configuration of Ni-YSZ/(YSZ/GDC bilayer)/PBSCF (37). When operated in EC mode at −1.1 A cm⁻² for over 350 hours and subsequent RC mode at −1.1/+1.1 A cm⁻² over the course of 10 measurement cycles (Fig. 3D), no apparent degradation was observed after 100 hours of prolonged operation. Considering the secure prolonged operations and cyclic durability, the absence of secondary phase formation (e.g., SrZrO₃ resistive phase at YSZ/GDC interface) and of microstructural changes (fig. S14), elemental changes (e.g., Fe segregation at interface or surface of internal pores, thicker interdiffusion layer; figs. S15 and S16) is expected. The primary reason for the observed high performances in FC and EC modes is considered to be the low area-specific ohmic resistance (R_ohm), obtained from the high-frequency intercepts of the impedance spectra (Fig. 4 and fig. S17), representing the resistance of the bilayer electrolyte. To the best of our knowledge, this study achieved the lowest R_ohm in temperature range of interest among all the SOCs comprising doped-zirconia/doped-ceria bilayers (Fig. 4 and table S1) (12, 37–47). The fully and congruently densified YSZ/GDC bilayer with the relatively thin interdiffusion layer and the minimum amount of residuals in GDC, enabled by the naturally diffused sintering aid, results in an electrically highly conductive and thermochemically secure electrolyte in SOCs. Moreover, there is still room for improvement of the electrochemical performances by optimizing the oxygen electrode materials and microstructures. Significantly, since the strategy is exemplified in the fabrication of SOCs through conventional ceramic processes, tape casting, and roll calendering, which are continuous, readily scalable, and cost-effective (48) (fig. S18), the SOCs in this study suggest a realistic way to overcome the challenges that SOCs currently face.

Fig. 3. Electrochemical properties of SOCs comprising the bilayer electrolyte.

(A) Potential and power density curves as functions of the current density under 3% H₂O/H₂ (fuel electrode) and dry air (oxygen electrode) in FC mode. (B) Temporal evolution of the voltage at a constant current density of 0.3 A cm² and 700°C in FC mode. (C) Potential curves as functions of the current density under 50% H₂O/H₂ (fuel electrode) and dry air (oxygen electrode) in EC mode. (D) Temporal evolution of the voltage at a constant current density of −1.1 A cm² in EC mode, followed by continuous cyclic operation between FC (1.1 A cm²) and EC (−1.1 A cm²) modes at 700°C.

Fig. 4. Comparison of the R_ohm of the developed SOC with those of previously reported cells comprising doped-zirconia/doped-ceria bilayer electrolyte.

Detailed information on each cell is listed in table S1.

DISCUSSION

We demonstrate that the Fe sintering aid that naturally diffuses from the underlying YSZ to GDC during the cosintering process enables the fabrication of a fully densified and highly conductive YSZ/GDC bilayer electrolyte in SOCs at a relatively low cosintering temperature via cost-effective conventional ceramic processes. The SOCs comprising this bilayer electrolyte show exceptionally low R_ohm and, in turn, exhibit high performance and secure operation for prolonged periods in both FC and EC operations. The strategy of a naturally diffused sintering aid is technically simple but effective in extending the processing window of multilayer fabrications, thus being readily adoptable into the industry of high-temperature electrochemical devices, in which large-scale production with a high process yield is essentially required.

MATERIALS AND METHODS

Dilatometry analysis

The linear shrinkages against temperature (≤1400°C) for several 8 mol % YSZ, 10 mol % GDC, and YSZ/GDC bilayer green body pellets—including YSZ (Tosoh, Japan), GDC (Rhodia Solvay, Belgium), 0.7 mol % Fe-, 2.0 mol % Fe-, 0.3 mol % Cu-, and 0.5 mol % Cu-added GDC, 2.0 mol % Fe-added YSZ (custom-ordered, Fcelltech, Korea), and 2 mol % Fe-added YSZ/GDC with volume % of 50:50—were analyzed using a dilatometer (DIL 402C, Netzsch, Germany) at a rate of 5°C min⁻¹ in air [200 standard cubic centimeter per minute (sccm)]. For the single-layer pellets, each powder was uniaxially pressed at 45 MPa for 2 min to obtain pelletized samples with a diameter of 10 mm and a thickness of 3.70 ± 0.15 mm. For the bilayer pellet, GDC powder was first uniaxially pressed under a pressure of 20 MPa, and then, 2 mol % Fe-added YSZ powder was added onto the slightly pressed GDC and pressed again under a pressure of 45 MPa.

STEM-EELS analysis

Atomic-resolution STEM-EELS mapping is used to examine the distribution of Fe in 2.0 mol % Fe-added YSZ and 2.0 mol % Fe-added GDC (custom-ordered, Fcelltech, Korea) at early stage of sintering, respectively. The 2 mol % Fe-added YSZ and 2 mol % Fe-added GDC powders were pelletized and heated in dilatometer instrument (DIL 402C, Netzsch, Germany). All pellets were heated to 900° and 1000°C at a rate of 5°C min⁻¹, respectively; then, right after the temperature was reached to the set value, the pellets were rapidly quenched to room temperature at a rate of 10°C min⁻¹. TEM-EELS samples were prepared by focused ion beam–scanning electron microscope (FIB-SEM, Helius NanoLab 600). Microstructure analysis was performed using a Cs-corrected STEM (FEI, TitanTM 80-300) equipped with a fast charge-coupled device camera (Gatan, Oneview 1095) and an EEL spectrometer (Gatan, Quantum 966). The STEM was operated at 300 kV for all samples. Under STEM mode, the probe convergence and EELS semi-collection angles were 21 and 28 mrad, respectively. The EELS spectra dispersion was 0.25 eV per channel and the pixel dwell time was 0.2 s. The EELS analysis was performed using Digital Micrograph program (ver. 3.21).

Tape preparation

For the fuel electrode support layer (FSL) tape, powders of NiO (Sumitomo, Japan), YSZ (Tosoh, Japan), and poly(methyl methacrylate) (PMMA; Sunjin Beauty Science, Korea) pore-forming agent with a ratio of 52.7:41.5:5.8 in weight were ball-milled for 24 hours using zirconia balls (diameter, 5 mm) in solvents [toluene and ethanol, 58:42 in weight % (wt%)] with a dispersant (KD-1, 2.3 wt %) to mix them. The proper amounts of commercial binder (PVB79, 9.3 wt %) and plasticizer [dibutyl phthalate (DBP), 9.3 wt %] were added to the slurry, and then, the slurry was ball milled for 24 hours with zirconia balls to form a compositionally homogeneous slurry and for an additional 24 hours without zirconia balls to extract residual pores. The tapes were fabricated using the prepared slurry through a tape caster (STC-14C, Hansung, Korea). The slurry was coated onto Mylar film (Alphasanup, Korea) at a rate of ~0.7 m/min. The slurries and tapes of the fuel electrode functional layer (FFL), which had an identical composition with that of FSL slurry but without the PMMA pore former, and of the electrolytes [YSZ and GDC with or without sintering aids; for the electrolytes with sintering aids, the custom-ordered powders (Fcelltech, Korea) were used] were prepared through identical procedures. The thicknesses of the FSL, FFL, and electrolyte tapes were approximately 130, 24, and 5 μm, respectively.

Fuel electrode/bilayer electrolyte preparation

Six sheets of FSL and one sheet of FFL tapes were stacked, in which Mylar films were attached onto the top and bottom of the stacked body. The stacked tape was laminated through roll calendering under conditions in which the temperatures of both the upper and lower rolls were set at 75°C, the rotational speed of the rolls (∅ 120 mm × L 200 mm) was maintained at 0.27 rpm, and the gap between rolls was modified to 99% relative to the total thickness of the laminated tape, i.e., 1% less than the initial thickness. After that, YSZ tape was transferred onto the FFL and then laminated using the condition for the lamination of the fuel electrode, followed by identical lamination of GDC tape onto YSZ. After lamination, the Mylar films at the top and bottom of the stacked tapes were replaced by polyimide films, and the laminated body [fuel electrode/(YSZ/GDC) bilayer electrolyte] was passed through the rolls with a gap between rolls of 92% relative to the initial thickness. Here, for the YSZ and GDC tapes, YSZ without a sintering aid, 2.0 mol % Fe added, GDC without a sintering aid, and 2.0 mol % Fe- and 0.3 mol % Cu-added GDC powders were used. The laminated cell was cut into a size of 8 cm by 8 cm and cosintered at 1250°C for 4 hours. To improve the flatness of the large-area cell after cosintering, an ironing step at 1200°C was further carried out. The microstructure of sintered cells was examined using SEM (Hitachi Regulus 8230).

Examination of the bilayer electrolyte

The crystallographic structure and chemical composition of the 2.0 mol % Fe-added YSZ (Fe-YSZ)/GDC bilayer in cosintered cell were investigated by TEM (FEI, Talos F200X). A sample of the bilayer electrolyte including the interface between Fe-YSZ and GDC for TEM analysis was prepared by FIB-SEM (Helius NanoLab 600). HR-TEM images of each grain in Fe-YSZ and GDC at several positions were taken with an electron-accelerating voltage of 200 keV. The lattice parameters of each grain were calculated using a FFT function in Gatan Digital micrograph software (Gatan Inc., USA). Compositional analysis of the bilayer electrolyte was performed using EDS (Super-X system, Bruker) equipped with a TEM system.

The distribution and concentration of the Fe sintering aid in the Fe-YSZ/GDC bilayer in the cosintered cell was analyzed using atom probe tomography (APT). APT needle-shaped specimens were prepared using a dual-beam FIB (Nova NanoLab 600, FEI, USA) with the site-specific “lift-out” method. The GBs in the specimens were observed using TEM (Talos F200X, FEI, USA) at an accelerating voltage of 200 kV (Schottky X-FEG gun) equipped with a Super-X EDS system comprising four windowless silicon drift detectors in STEM mode with a probe current of ~0.7 nA. The APT specimens were measured in a local electrode atom probe (LEAP 4000 X HR, CAMECA, USA) by applying 10-ps, 50-pJ ultraviolet (wavelength = 355 nm) laser pulses with a pulse repetition rate of 100 kHz. The detection rate was 3 ions per 100 pulses on average. The base temperature was 50 K, and the ion flight path was 350 mm. The APT data were processed using the commercial software package IVAS 3.6.14 (CAMECA, France). Correlative analysis with TEM and APT was carried out to observe the GBs and obtain the elemental distribution information near the GBs.

Examination of Fe solubility in electrolytes

YSZ and GDC powders were mixed with 1, 2, 3, 4, and 5 mol % Fe nitrate [Fe(NO₃)₃·9H₂O; Sigma-Aldrich, USA] in solvents (mixed ethanol and distilled water) for 24 hours. After complete removal of the solvent through a drying process at 200°C, the powders were calcined at 500°C to form iron oxide. The obtained powders were pelletized using a uniaxial press with a pressure of 45 MPa for 2 min and then heat treated at 1250°C for 4 hours through an identical schedule as the cosintering process for the fuel electrode/bilayer electrolyte cell. The heat-treated pellets were hand ground again into powder with a mortar and pestle. The lattice parameters of each powder were analyzed using an x-ray diffractometer (CuKα radiation; step rate, 0.02° s⁻¹; D Max 2500/PC, Rigaku).

SOC fabrication and electrochemical characterization

The sintered substrate in the configuration of Ni-YSZ fuel electrode/(Fe-YSZ/GDC) bilayer was cut into a size of 2 cm by 2 cm. An oxygen electrode functional layer (OFL), which consisted of PBSCF (Kceracell, Korea) and GDC powders in a mixing ratio of 50:50 in wt %, was screen printed on the GDC electrolyte with an area of 1 cm by 1 cm and then sintered at 1050°C for 3 hours. In addition, an oxygen electrode current-collecting layer (OCL) comprising only PBSCF was screen printed on the sintered OFL layer and then sintered at a lower temperature of 950°C for 2 hours. The OFL and OCL slurries for the screen printing were prepared using commercial PBSCF and GDC powders, a solvent (α-terpineol, Kanto Cehmical), a dispersant (Hypermer KD6, Coroda Advanced Materials), a binder (BH-3, Sekisui Chemical Co.), and a plasticizer (DBP, Junsei) through a planetary milling process (Pulverisette 5, Fritsch) for 48 hours.

The electrochemical performances of a single cell in FC and EC modes were evaluated in a lab-made station. The cell was placed between Inconel 600 alloy metal interconnectors and compression sealed using a glass-ceramic sealant. Cu-Mn foam (Alantum, Korea) and Ni foam (Alantum, Korea) were used as the oxygen electrode and fuel electrode current collectors, respectively. The open-circuit voltage (OCV), electrochemical impedance, and current-voltage characteristics in a temperature range from 650° to 800°C were analyzed using a frequency response analyzer and a potentiostat (Solartron 1260/1287 and Solatron EnergyLab XM). Electrochemical impedance spectroscopy was performed with an amplitude of 10 mV at OCV over a range of 10⁵ to 0.1 Hz. During the evaluation of the FC mode, humidified hydrogen (3% H₂O in H₂) with flow rate of 200 sccm and dry air with flow rate of 200 sccm were supplied to the fuel electrode and oxygen electrode, respectively. In the EC mode, 50% H₂O in H₂, achieved by forcing the H₂ through a heated water vessel at 83.5°C, and dry air were supplied. Long-term FC/EC modes and cyclic operations were performed at a constant current density of +0.3 A cm⁻² for 300 hours (in FC mode), −1.1 A cm⁻² for 350 hours (in EC mode), and +1.1 A cm⁻² (FC)/−1.1 A cm⁻² (EC) over the course of 10 measurement cycles (cycling mode).

Electrical and kinetic properties of the oxygen electrode

A PBSCF bulk sample for electrical conductivity relaxation (ECR) analysis to evaluate the surface reaction redox kinetics was prepared through the tape casting process described above for the fuel electrodes and electrolytes. The green tape was sintered at 1150°C for 10 hours, resulting in full density above 99%. The sintered body was cut into a rectangular shape with dimensions of 10.4 mm by 5.4 mm by 0.36 mm. The thickness was only 6.7% of the sample width, justifying the analysis of relaxation profiles according to the one-dimensional model, i.e., the overall redox reaction is dominated by the reaction in the thickness direction. Gas and temperature controls were provided by an in-house constructed ECR reactor with a sample chamber approximately 10 cm³ in volume, which is sufficiently small to ensure fast gas exchange. The total electrical conductivity was measured by the DC four-probe method using a current source (6220, Keithley, USA) and a multimeter (2000, Keithley, USA) at 400° to 700°C, and the relaxation behavior at each temperature was characterized under an abrupt oxygen partial pressure change from low pO₂ (~0.17 atm) to high pO₂ (~0.21 atm), achieved by flowing mixtures of dry air and Ar with a total flow rate of 200 sccm. The absence of solid-state diffusion contributions (from all experiments) was established by analyzing the relaxation profiles according to Eq. 1, and, in turn, the chemical surface reaction rate constant, k_chem, was extracted

$$\frac{\mathrm{\sigma }(t)-{\mathrm{\sigma }}_{\mathrm{i}}}{{\mathrm{\sigma }}_{\mathrm{f}}-{\mathrm{\sigma }}_{\mathrm{i}}}=1-\text{exp}(-\frac{k_{\text{chem}}}{a}t)$$ (1)

where σ_i and σ_f are the initial and final conductivity values, respectively, and the sample thickness is 2a (with exchange occurring on both surfaces of the thin sample).

Postanalysis of the microstructure and composition

Cross sections of SOCs after long-term operation in FC mode and EC with subsequent RC modes at 700°C were examined using a SEM (Hitachi Regulus 8230) and an electron probe microanalyzer (JEOL JXA-8500F) to evaluate the microstructure and check the chemical reaction between the oxygen electrode and bilayer electrolyte.

Supplementary Materials

This PDF file includes:

Table S1

Figs. S1 to S18

Legends for movies S1 and S2

Other Supplementary Material for this manuscript includes the following:

Movies S1 and S2