The electrolyte cells of YSZ (8 mol% Y₂O₃ stabilized ZrO₂) were prepared by cutting into 20 mm diameter using water-jet from commercial dense YSZ square sheet (Kceracell) with a thickness of 250 μm. Both surfaces of the electrolyte cells were cleaned by ethyl alcohol. Each anode material maintained a consistent weight ratio of NiO (Kceracell, BET 8.8 m²/g) to GDC (Kceracell, 20 mol% Gd-doped ceria, BET 12.5 m²/g) of 6:4 wt%, whereas the Bi₂O₃ (Sigma Aldrich, 99.999%) loading was prepared as varied from 0–10 mol% as a sintering aid. The Bi₂O₃ loadings considered were at 0, 2.5, 5, 7.5, and 10 mol% relative to the GDC contained in the anode. The anode pastes were prepared by mixing the anode powder with binder paste (Kceracell) in a 7:3 weight ratio using a Resodyn Acustic Mixer (LabRAM II) at 50 G for 5 min. Both commercial slurry products of the GYBC (Gd_0.135Y_0.015Bi_0.02Ce_0.83O_1.915, BET 15 m²/g) and LSCF-GDC (La_0.6Sr_0.4Co_0.2Fe_0.8O_3-δ-Gd_0.1Ce_0.9O_1.95, BET 7.5 m²/g) were purchased from Kceracell Co., Ltd, and used to buffer layer and cathode, respectively. GYBC, which needs a low sintering temperature of 1100°C, was considered for the buffer layer to minimize the sintering impact on the anode during the sintering process of the cathode side. Three pastes of anode, cathode, and buffer layers were prepared to obtain a uniform quality by mixing at 2200 rpm for 3 min, followed by defoaming at 2000 rpm for 2 min using Thinky Mixer (ARE-310).

All layers of SOFC cells are prepared by screen printing using each mask with 40, 5, and 40 μm depth and 8, 10, and 8 mm diameter for an anode, buffer layer, and cathode, respectively. The anode layer was coated on a clean surface of YSZ and then sintered at 1200°C for 2 h with a 2°C ramp rate of heating/cooling. The buffer layer was coated on YSZ, on the opposite side of the anode, following a dry process at 60°C for half an hour in a convection oven. The cathode layer was coated on top of the dried buffer layer and then co-sintered at 1100°C for 2 h with a 3°C ramp rate of heating/cooling. Finally, the current collectors were attached to each electrode using a homemade NiO sheet for an anode and an LSC (LaSr_0.6Co_0.4O₃-δ) sheet for a cathode.

A single cell with LSCF-GDC | GYBC | YSZ | NiO-GDC-(Bi₂O₃) was electrochemically tested using the current-voltage (I-V) and constant bias methods using an electrochemical workstation. All cells were sealed onto single-cell measurement equipment (Nano Ionics) using sealants (Schott 311, Aremco 552) and heated to 800°C with appropriate heating steps as described by the sealant manufacturer. Air and H₂ were used as the oxidant and fuel, respectively, during the operation, and both gas flow rates were 100 mL/min. The operating stability of fuel cells was carried out at a constant current density of 150 mA/cm² at 800°C. I–V measurement was examined from open-circuit voltage to 0.4 V with 10 mA/cm² intervals. The microstructures and morphologies of the anodes were examined using field-emission scanning electron microscopy (FE-SEM, Hitachi SU-8230).

In this study, the samples were labelled B0, B2.5, B5, B7.5, and B10, corresponding to the Bi₂O₃ loading. For instance, B10 indicates that 10 mol% Bi₂O₃ was added relative to the GDC contained in the anode. Fig. 1a illustrates the relationship between the material weight loading and Bi₂O₃ addition to the anode across all samples. As the Bi₂O₃ loading increased, the weight proportions of NiO and GDC decreased. The anode thickness of all the cells was measured using a micrometer before and after sintering. The anode thickness of the B0 cell decreased from 25 μm to 18 μm after sintering at 1200°C for 2 h. Cells with added Bi₂O₃ had the same initial anode thickness as the B0 cell, but after sintering, they exhibited varied anode thicknesses of 17, 14, 10, and 10 μm for B2.5, B5, B7.5, and B10, respectively, as depicted in Fig. 1b. The use of Bi₂O₃ as a sintering aid produced dense anodes, suggesting that the significantly enhanced sintering activity could be attributed to the presence of Bi₂O₃. Since the resultant Bi₂O₃ has a low melting temperature of ~825°C, a liquid phase is probably formed on the surface of NiO-GDC mixture grains when sintered at 1200°C, which enables the shrinkage and sintering by the effect of liquid phase sintering. The decreasing anode thickness correlates with an increase in Bi₂O₃ loading; however, the thickness remains unchanged when the Bi₂O₃ loading increases from 7.5 mol% to 10 mol%, suggesting that the optimal Bi₂O₃ loading limit to densify is approximately 7.5 mol%. Increasing the Bi₂O₃ loading further could lead to the creation of more pores, which results in reduced density because of the volatile nature of Bi₂O₃ during the high-temperature sintering process [29,30].

Fig. 2 presents cross-sectional SEM images of the NiO-GDC anode after the electrochemical measurements. The interface between the anode and electrolyte of Ni-GDC-(Bi₂O₃) layers can be clearly observed to be attached indicating the successful sintering process of the anode at 1200°C. The sintered Ni-GDC anode nanoparticles exhibited strong interconnectivity alongside the formation of a porous microstructure. The reduction in the anode’s thickness, despite the constant diameter of 8 mm before and after sintering, can be attributed to the presence of Bi₂O₃. Therefore, the densification of the anode is likely due to the Bi₂O₃ as sintering aid. SEM images revealed that the microstructure of the anode containing Bi₂O₃ was denser than that of the anode lacking Bi₂O₃; however, no significant changes were observed. This result indicates that the volatilization of occupied Bi₂O₃ particle inside anode layer generated pore during the sintering process. Furthermore, the low weight percentage of NiO-GDC in the anode due to the addition of Bi₂O₃ likely affects the relationship between porosity and thickness. The micromorphology of the anode without Bi₂O₃ appears to be a framework composed of softly rounded particles arranged in a three-dimensional distribution. However, for the anodes containing Bi₂O₃, the emergence of a sharp surface is contingent on the amount of Bi₂O₃ loaded, suggesting that Bi₂O₃, which melts during the sintering process, adequately coats the NiO-GDC surface and persists even after the reduction of NiO to Ni. This unique shape of particles can increase the pores and the electrode surface to improve the triple-phase boundary. The addition of Bi₂O₃ potentially improved the mechanical strength by forming bonds with the NiO-GDC matrix.

Fig. 3 shows the electrochemical performance of LSCF-GDC | GYBC | YSZ | NiO-GDC-(Bi₂O₃) cells fueled with 3% humidified hydrogen at 800°C. The measured open circuit voltage (OCV) value of the cells was 1.1 V, regardless of the presence of Bi₂O₃. The OCV remained consistent after 240 h of operation under 150 mA/cm² bias, indicating that no significant mechanical damage, such as cracking, occurred in any cell during the measurement period. The initial current-voltage (I–V) curves of the cells overlapped (Fig. 3a), indicating similar electrochemical activities among the cells. The initial peak power density (PPD) values of the cells are also comparable, with values of 0.476, 0.48, 0.491, 0.468, and 0.478 W/cm² for B0, B2.5, B5, B7.5, and B10, respectively, at 800°C. However, a distinct deviation in the PPD was observed in the final I-V measurement after 240 h of operation under 150 mA/cm² bias among the cells, which differed only in the anode, as shown in Fig. 3b. The final PPD of the cells were 0.354, 0.374, 0.404, 0.396, and 0.397 W/cm² for B0, B2.5, B5, B7.5, and B10, respectively, after 240 h of operation. The change in the PPD caused by the anode indicates that it plays a crucial role in electrolyte-supported cells, in contrast to the minimal impedance impact of anode in the anode-supported cells. To examine the impact of Bi₂O₃ on the cell performance, the PPD of all cells, for both the initial and final I–V curves, is re-plotted in Fig. 3c. The initial PPD of the cells was almost constant regardless of the presence of Bi₂O₃ in the anode. These findings indicate that a suitable quantity of Bi₂O₃ does not significantly influence the electrochemical reactions at the triple-phase boundary within the anode, even though molten Bi₂O₃ may cover parts of the Ni-GDC surface. The largest PPD drop of 0.12 W/cm² appears for the B0 cell and the gap can be reduced by increasing Bi₂O₃ loading up to at least 5 mol%. The B5, B7.5, and B10 cells exhibit a comparable PPD degradation rate of approximately 16.7%, which was lower compared to that of the B0 cell at 25.6% and the B2.5 cell at 21.9%.

Fig. 4a shows the changes in PPD over time. The difference in PPD between the cells widened with time, although the initial PPD values were comparable. The cells containing Bi₂O₃ demonstrated improved performance compared to those without Bi₂O₃. Within the range of B2.5–B10 cells, B5 cell exhibited the highest PPD throughout the measurement period. The order of PPD was B5 > B7.5 ≈ B10 > B2.5 > B0. The PPD degradation rate was re-plotted into four regimes to delineate the operating time for each 60 h period, as shown in Fig. 4b. The PPD degradation of all cells occurred critically, with an average rate of 200 %/kh during the initial regime of 0 to 60 h. This rate decreases to approximately 100 %/kh in the second regime and stabilizes at approximately 30 %/kh in the third and fourth regimes. This result indicates that the initial set point, following rapid degradation in the first regime, serves as a more reliable baseline for calculating the degradation rate during prolonged operation. The cells without Bi₂O₃ exhibited the highest PPD degradation rate in the initial regime, whereas the PPD degradation rates in the cells containing Bi₂O₃ were more evenly distributed across both regimes. Consequently, Bi₂O₃ effectively mitigated the anode-induced PPD degradation and enhanced cell longevity.

Fig. 5a shows the long-term stability of the fuel cells equipped with the NiO-GDC-Bi₂O₃ anode layer. The measurement is conducted at a temperature of 800°C and a constant current density of 150 mA/cm². All the cells exhibited a two-phase degradation profile: rapid voltage degradation over 60 h, followed by slower degradation for up to 240 h. These results suggest that a period of approximately 60 h is required for stabilization to ensure long-term operation of the electrolyte-supported cell. Fig. 5b shows the voltage degradation of all cells over 240 h, which was less than 4%. The degradation profile of the cells without Bi₂O₃ exhibited a significantly steeper slope than that of the cells with Bi₂O₃. However, both demonstrate a similar pattern: a substantial voltage drop during the first 60 h, followed by a gradual decline thereafter. In addition, the magnitude of the voltage drop appeared to decrease at higher Bi₂O₃ loadings. The voltage degradation rates of the cells were recalculated for an extended period of 1000 h, as listed in Table 1. Over a significant degradation period from the beginning to 60 h, the voltage degradation rate of cells without Bi₂O₃ was 46 %/kh, while the rate for cells with Bi₂O₃ was reduced by half to approximately 18–24 %/kh. After 60 h, the voltage degradation of B0, B2.5, B5, B7.5, and B10 significantly decreased to 3.2, 6.2, 5.3, 3.8, and 2 %/kh, respectively, over a measurement period of 60 to 240 h. Considering both voltage degradation rate and power density, thus, the heavy loading of Bi₂O₃ may be a viable strategy for enhancing the long-term operation of electrolyte-supported SOFCs.