1 Introduction
Water electrolysis for hydrogen production is considered one of the key pathways to reduce dependence on traditional energy sources [
1,
2]. In this context, a variety of electrochemical devices have been explored [
3]. Among these, SOEC are environmentally friendly and highly efficient high-temperature conversion technologies that do not require expensive catalysts [
4–
7].
In the field of renewable energy, the integration potential of this technology with solar [
8] and wind [
9] energy is becoming increasingly prominent, offering significant socio-economic benefits [
10,
11]. Individual electrolysis cells are typically connected in series or parallel to form an electrolysis stack for large-scale hydrogen production [
12]. Although there have been reports of electrolysis stack achieving ultra-long-term operation [
13,
14], the long-term stability of electrolysis stack performance remains a key research focus.
SOEC stacks can experience irreversible performance degradation during extended operation. As the site of electrochemical reactions, the cells directly influence the overall performance of the electrolysis stack [
15,
16]. Under typical SOEC operating conditions, the oxygen evolution reaction at the interface between electrolyte and air-electrode can lead to interfacial delamination [
17]. In 2024, Yang et al. [
18] conducted electrolysis of N
2/H
2O to produce ammonia feedstock at a current density of 700 mA/cm
2 for 800 h. Strontium (Sr) segregation on the air-electrode surface contributed to performance degradation. Additionally, in studies exceeding 1000 h, more instances of cell degradation were reported [
19–
23], including Ni depletion in the fuel electrode, mechanical failure of the electrolyte causing cracks and voids, interfacial reactions between the electrolyte and electrode materials forming SrZrO
3, and Cr poisoning of the electrode material due to Cr diffusion from the interconnect. These degradation mechanisms pose significant risks to long-term electrolysis performance, making it crucial to enhance cell stability.
The metal interconnect is used to link multiple electrolysis cell units. Both the resistance of the interconnect and the interfacial resistance between the interconnect and the cell are closely tied to overall stack performance [
24,
25]. Although the large amount of oxygen produced at the air-electrode during SOECs operation accelerates interconnect oxidation, studies have indicated that the resistance of the interconnect can remain below 46.61 mΩ·cm
2 after 2000 h of operation, which meets performance requirements [
26]. Previous study has shown the degradation of SOFC performance due to interface issues can reach up to 58.2%, depending on the contact method between the interconnect and the cell [
27]. This indicates that the current-collecting interface of the electrolysis stack significantly impacts its output performance. Moreover, a recent study found that diffusion of interfacial elements between the interconnect and the cell is one of the contributing factors to stack degradation. In 2023, Wu et al. [
28] investigated the degradation behavior of a flat-tube stack operated for 900 h within 64 pulsed cycles during H
2O/CO
2 co-electrolysis. The formation of SrCrO
4 at the interconnect-air-electrode interface contributed to performance decline.
In summary, the performance of the electrolysis stack is constrained by both the electrolysis cells and the current collector interface. However, the quantitative contributions of these two components to the degradation, as well as the underlying degradation mechanisms, remain unclear.
Therefore, this work adopted a segmented research approach to differentiate the contributions of individual stack components [
19,
29]. Quantitative calculations were performed to determine the specific contributions of the cells and the interface to the overall degradation of the electrolysis stack. Moreover, the degradation mechanisms of the electrolysis cells during operation were investigated, providing insights for improving the long-term application of SOEC stacks in water electrolysis.
2 Experiment
The flat-tube SOEC used in this experiment was manufactured by Zhejiang H
2-Bank Technology Company, with an effective active area of 60 cm
2. The cell structure has been described in previous studies [
18,
30,
31]. The metallic interconnects were made of SUS441 stainless steel, with flow channels on the air-electrode side formed by an etching process.
A single cell and an interconnect together formed a stack repeat unit, as illustrated in Fig. 1(a). The cathode (air-electrode) and fuel-electrode sides of the cell were coated with current collection layers: “LSC(La0.6Sr0.4CoO3‒δ) + Ag” on the air-electrode, and “NiO” on the fuel electrode. The interconnect side in contact with the cell’s air-electrode served as the air flow channel, while the fuel side featured a smooth planar surface.
Figure 1(b) shows the stack structure, consisting of three repeat units designated as Cell-1, Cell-2, and Cell-3. The corresponding metal interconnects were labeled IC0, IC1, IC2, and IC3. The interfaces between each interconnect and the cell electrode were designated as IC-I0, IC-I1, IC-I2, and IC-I3, respectively. IC-I0 represented the interface between IC0 and the fuel electrode of Cell-1, while IC-I1 denoted the interface between IC0 and the air-electrode of Cell-1. Similarly, IC-I2 and IC-I3 referred to the interfaces between IC2, IC3 with the air-electrode of Cell-2 and Cell-3, respectively. Cell-3 and IC-3 were positioned near the fuel inlet, while Cell-1 and IC-I0 were located near the fuel outlet.
During the assembly process, a platinum wire with a diameter of 0.1 mm was inserted at each layer interface. Voltage wires were arranged as shown in Fig. 1(b) to enable real-time monitoring of the voltage data across each cell and the fuel-electrode interface (IC-I0) and air-electrode interfaces (IC-I1, IC-I2, and IC-I3). After assembly, the stack was placed in a sintering furnace, heated to 800 °C, and subjected to a compressive load of 2000 N to ensure proper sealing and collection efficiency.
Before starting the electrochemical performance tests, the sintered stack was slowly cooled to 750 °C and held that temperature. The testing setup is shown in Fig. 1(c). Mass flow meters were used to control the flow rates of H2, N2, and air, while a pump (LC-3060B, Beijing Chin-Fine Technology Co., Ltd., China) in the steam generator regulated the water input.
Prior to testing, a reduction step was conducted. nitrogen was first purged through the fuel-electrode side, followed by the introduction of 0.6 L/min H2 into the same side. Simultaneously, 2.0 SLM air was supplied to the air-electrode side. All gases were fed into the stack without preheating. After 4 h of reduction, the open-circuit voltage (OCV) of the stack stabilized, indicating the completion of the reduction process. Performance testing then proceeded.
The electrochemical performance of the stack was evaluated in both discharge and charge modes. In discharge mode, 2 L/min H2 was supplied to the fuel-electrode, and 6 L/min air to the air-electrode. In the charge mode, the fuel side received a mixture of 0.7 L/min H2 (40 vol.%) and 0.844 g/min water vapor (60 vol.%), while 3 L/min air was supplied to the air-electrode side. Long-term electrolysis testing was conducted at 750 °C under a current density of 500 mA/cm2.
The electrochemical performance of was evaluated using a test platform (500W-SOC, H2-Bank). During the test, a soap bubble flow meter (Beijing Ke’an Labor Protection New Technology Co., Ltd., China) was employed to periodically calibrate the flow rate of the exhaust gas from the fuel side after drying. Electrochemical impedance spectroscopy (EIS) measurements were conducted for the stack, individual cells, and interconnects under open-circuit voltage conditions, with 2 L/min H2 and 6 L/min air. The measurements were performed using an electrochemical workstation (VMP3B-20, Bio-Logic, France), over a frequency range of 20 mHz to 40 kHz, with a voltage amplitude of 20 mV.
Moreover, a scanning electron microscope (SEM, S4800, Hitachi, Japan; FEI Quanta FEG250, USA) was used to characterize the microstructure of the stack components, and phase analysis was conducted using X-ray diffraction (XRD, D8 Advance, Germany).
3 Results and discussion
3.1 Instantaneous electrochemical performance of the stack
The electrochemical performance of the stack was evaluated in discharge mode. The instantaneous I‒V performance curves are shown on the right side of Fig. 2(a)‒2(d). The results demonstrate that the initial maximum discharge power of the stack was 104.7 W at 1.8 V and 969 mA/cm2. The corresponding power outputs of the three individual cells were 44.31, 44.54, and 44.07 W, respectively, indicating consistent discharge performances of the three cells. The power loss at the interfaces between the cells and interconnects was 28.22 W, accounting for 21% of the total power output from the three cells.
After long-term electrolysis, the maximum discharge power of the stack decreased to 80.89 W at 1.8 V and 749 mA/cm2. The maximum power outputs of the individual cells declined to 31, 31.00, and 29.06 W, respectively. At this point, power loss at the interfaces was reduced to 10.17 W, representing 11% of the total cell power output.
The I‒V curves in electrolysis mode are shown on the left side of Figs. 2(a)‒2(d). Initially, the stack exhibited a maximum current density of 750 mA/cm2 at 4.2 V, with corresponding voltages of the three cells of 1.3, 1.28, and 1.23 V. The voltage loss at the interfaces was 0.39 V, accounting for 10% of the total voltage across the three cells.
After the long-term constant current electrolysis, the maximum current density decreased to 608 mA/cm2 at 4.2 V. The corresponding cell voltages increased slightly to 1.31, 1.35, and 1.3 V, while the interface voltage loss decreased to 0.24 V, accounting for 6% of the total voltage across the cells.
Combined analysis of the discharge and electrolysis performances reveals a 22.5% reduction in instantaneous discharge performance and an 18.9% decrease in instantaneous electrolysis performance. Notably, the interface between the cell and interconnect showed signs of improvement following long-term electrolysis, suggesting better contact and reduced resistance over time.
3.2 Long-term stability test of the stack
Figures 3(a) and 3(b) show the voltage‒time (V‒t) curves of the stack and individual cells, at 500 mA/cm2, respectively, during 900 h of water electrolysis at a constant current density of 500 mA/cm². Based on curve fitting, the degradation rate of the electrolysis stack was calculated to be 0.93% per 100 h. The degradation rates of the individual cells were 0.93%/100 h, 1.5%/100 h, and 0.84%/100 h, respectively. Notably, the overall degradation rate of the stack was lower than the average degradation rate of the individual cell, indicating that interface optimization occurred during operation.
Table 1 compares the degradation rates of SOEC stacks for water electrolysis reported in recent studies, ranging from 0.43%/100 h to 2.14%/100 h [
19,
20,
32–
34]. This comparison indicates that the stack in this study exhibits good long-term stability relative to similar systems.
The degradation process can be divided into three distinct stages. During the first 300 h, electrolysis remained stable, with only a slight decrease in stack voltage. From 300 and 600 h, both the stack and individual cells experienced accelerated degradation, leading to a significant increase in voltage. In the final phase, voltage stabilized again. Notably, at 540 h, the stack experienced a 10-h interruption in hydrogen supply, resulting in a sharp voltage increase in Cell-3, which was located near the fuel gas inlet.
The voltage data for the stack and its components before and after the constant current electrolysis test are summarized in Table 2. The degradation contributions of the individual cells accounted for 37.09%, 55.87%, and 32.86% of the total stack performance degradation, respectively. Interestingly, the change in interfacial performance contributed a negative value of −25.82%, indicating that interface conditions improved over time, potentially mitigating overall stack degradation.
Figure 3(c) shows the variation in OCV values of the three cells during the long-term electrolysis process. The OCV values decreased to varying degrees: Cell-1 dropped from 1.044 to 1.015 V, Cell-2 from 1.078 to 1.077 V, and Cell-3 from 1.064 to 0.996 V.
Figure 3(d) records the actual hydrogen production and water vapor conversion ratio during the electrolysis. The stack maintained nearly stable performance throughout the test. The actual values were slightly lower than theoretical predictions due to issues with packaging and piping. Over the 900-h period, the stack produced an average of 0.599 L/min of gas, with a Faradaic efficiency exceeding 95%.
The theoretical hydrogen production
q and water vapor conversion rate
n for water electrolysis are calculated based on the number of electron transfers, as detailed in Ref. [
18,
30]:
where I represents the working current (A); t is the time in seconds, 60 s; c is the number of cells, 3; e is the number of electrons transferred in the electrochemical process (2); F is the Faraday constant (96485 C/mol); V is the molar volume of gas under standard conditions (22.4 L/mol), and L is the water vapor flow rate (1.05 L/min).
Based on this, the theoretical hydrogen production q of the stack is 0.627 L/min, and the calculated steam conversion rate n is 59.71%.
3.3 Cells contribution to stack performance
EIS was used to analyze the resistance changes in the stack and cells. Figures 4(a)‒4(d) display the Nyquist plots of the impedance spectra for the flat-tube electrolyzer stack and individual cells [
34–
36]. Figures 4(e) and 4(f) illustrate the resistance variations of the stack and cells over time.
The initial ohmic resistance of the stack was 14.92 mΩ, which increased to 16.93 mΩ after 900 h. The initial polarization resistance was 25.16 mΩ, gradually increasing to 42.25 mΩ at 600 h before decreasing to 34.49 mΩ. The initial ohmic resistances of the cells were 2.45, 2.37, and 2.93 mΩ, which increased post-electrolysis to 3.13, 3.87, and 4.03 mΩ, respectively.
The initial polarization resistances of the cells were 7.98, 10.91, and 8.9 mΩ, which changed to 9.52, 18.37, and 7.7 mΩ after electrolysis, aligning with the observed degradation rate trends of the cells.
The flow pattern of the stack structure leads to significant variations in polarization resistance. Cell-2, positioned in the middle of the stack and farther from both the inlet and outlet (as shown in Fig. 1(b)), experiences high diffusion resistance due to restricted fuel gas and air. As a result, Cell-2 operates at a higher impedance, leading to greater degradation compared to the other cells.
The ohmic resistance of Cell-1, Cell-2, and Cell-3 increased by 27.7%, 63.3%, and 37.5%, respectively, after the 900-h electrolysis. Meanwhile, the polarization resistance changed by 19.3%, 68.4%, and −13.5%. The increases in ohmic resistance for Cell-1, Cell-2, and Cell-3 accounted for 33.83%, 74.63%, and 54.72% of the total stack resistance, respectively, with the cumulative total exceeding 100%. The changes in polarization resistance for the cells contributed 16.5%, 79.95%, and −12.86% to the overall stack, with Cell-2 experiencing the largest increase and Cell-3 showing a decrease in polarization resistance.
Distribution of relaxation times (DRT) was used to analyze the EIS data. Figure 5 shows the DRT images of the test cell at different time intervals. The processes corresponding to different peaks are as follows [
21,
36]:
P1 (100–101): Gas diffusion process within the fuel electrode;
P2 (101–102): Oxygen exchange and oxygen diffusion process at the cathode;
P3, P4 (102–104): Charge transfer reaction in the fuel electrode;
P5 (> 104): Ionic conduction process in the electrolyte.
It can be observed that the peaks at P1, P3, and P4 of the test cell undergo significant changes. These changes align with the variations in the cell’s ohmic resistance and polarization resistance, indicating that the degradation of the cell primarily occurs at the fuel electrode. A more in-depth investigation would require disassembling the cells for microscopic characterization.
After disassembling the stack, the central sections of both the test cell and the reference cell were extracted and embedded using low-pressure vacuum impregnation with epoxy resin. The samples were then ground with sandpaper, polished, and characterized using scanning electron microscope (SEM). The secondary electron (SE) and backscattered electron (BSE) images are shown in Fig. 6. In the Ni/YSZ active layer of the tested cell, the loss and agglomeration of Ni particles were observed. The agglomeration of Ni particles reduces the electron percolation path and increases the cell’s ohmic resistance. The most significant Ni loss was observed in Cell-2, which correlates with its high degradation rate, as further confirmed by the BSE image.
Moreover, BSE imaging revealed changes in the porosity of the active electrode layer within 15 μm of the electrolyte-electrode. Cell-2 exhibited a significant increase in active layer porosity, leading to a pronounced reduction in three-phase boundary (TPB) length. This reduction in charge transfer pathways resulted in an increase in polarization resistance [
37,
38]. Furthermore, in Cell-1 and Cell-3, particularly in regions farther from the electrolyte (10–15 μm from the electrolyte-electrode), Cell-3 demonstrated a more favorable porosity, facilitating better fuel gas transport to the active layer. Consequently, Cell-3 exhibited the lowest diffusion resistance [
39].
The variation in fuel electrode porosity is likely induced by Ni redistribution, though the underlying mechanism remains unclear. The most plausible explanation, as reported by Mogensen [
40], suggests that at high temperatures, Ni particles volatilize in a humidified fuel gas atmosphere in gaseous Ni species, ultimately leading to Ni migration and re-agglomeration.
3.4 Interfaces contribution to stack performance
Figure 7(a) presents the I‒V curves for each interface before and after electrolysis. The results indicate a significant decrease in the voltage slope at each interface, suggesting an improvement in interface contact. Figure 7(b) illustrates the V‒t curves of each interface during water electrolysis. The results show that the voltage at the IC-I0 fuel-electrode interface decreased from 0.015 to 0.01 V, while the voltage at the IC1-I1 air-electrode interface dropped from 0.072 to 0.066 V. The IC-I2 voltage decreased from 0.042 to 0.034 V, and the IC-I3 voltage dropped from 0.094 to 0.053 V, with IC-I3 showing the largest decrease about 0.039 V. Since the current collection area of the fuel electrode is approximately twice that of the air-electrode, the voltage at the IC-I0 fuel-electrode interface is the lowest.
As shown in Fig. 7(c), the initial resistances for IC-I0, IC-I1, IC-I2, and IC-I3 were 0.51, 2.24, 1.25, and 2.89 mΩ, respectively, accounting for 3.42%, 15.01%, 8.38%, and 19.37% of the stack’s total ohmic resistance. The final resistance for IC-I0, IC-I1, IC-I2, and IC-I3 were 0.35, 2.13, 1.12, and 1.73 mΩ, respectively, accounting for 2.07%, 12.58%, 6.62%, and 17.07% of the stack. The interface resistance decreased during the 0‒200 h and 500‒600 h periods. According to Refs. [
41–
44], during the early stages of electrolysis, the actual conductive contact area between the interconnect and electrode materials is significantly smaller than the apparent area. As the oxide layer forms (0–200 h), the conductive area increases, reducing interfacial resistance. However, between 500 and 600 h, the temporary absence of hydrogen and fluctuations in fuel gas composition lead to temperature variations within the stack. These variations, combined with oxide layer growth and thermal stresses, can cause spallation of the interconnect oxide layer. Despite this, the stack operates under a constant pressure of 2000 N, which helps mitigate interfacial resistance over time.
Meanwhile, the interconnect in contact with the air-electrode was analyzed using XRD after disassembly, and the results are shown in Fig. 7(d). After testing, oxides such as Cr2O3 and (Mn,Cr)3O4 appeared on the interconnect surface, supporting the optimization of interface contact. Figure 7(e) shows the interface morphology at the air-electrode after disassembly. The indentation on Cell-2 is the most pronounced, while the indentation on Cell-1 is shallower compared to Cell-3, consistent with the aforementioned resistance trends.
Figure 8 shows the voltage ratio of the cells and interfaces relative to the total voltage of the electrolysis stack during the electrolysis process. The results indicate that the cell voltage exhibited a consistent increasing trend. The voltage ratio of each cell to the total voltage of the electrolysis stack fluctuated between 30% and 33%. The voltage across total interfaces accounted for 4%‒6% of the electrolysis stack’s total voltage, showing a decreasing trend. This suggests that the improvement of the interconnect-electrode interface is the primary reason for the slow degradation of the stack.
4 Conclusions
In this paper, the quantitative contributions of stack’s components to degradation were investigated at 750 °C under a constant current of 500 mA/cm2 for 900 h electrolysis test. The voltage of the electrolyzer stack increased by 0.213 V, corresponding to a degradation rate of 0.93%/100 h. The voltage increases for Cell-1, Cell-2, and Cell-3 were 0.079, 0.119, and 0.07 V, respectively. The interfacial voltage decreased by 0.055 V, accounting for 25.82% of the total stack voltage. Degradation of the cell mainly occurs in the early stage, after which the fluid distribution within the stack leads to rapid degradation of Cell-2. Further optimization can enhance fluid uniformity and reduce this rapid degradation.
The ohmic resistance and polarization resistance of the cell increased. DRT analysis indicates that the degradation originates at the fuel electrode, with microscopic analysis revealing TPB loss between Ni-YSZ grains and changes in the porosity within the Ni-YSZ layer. The ohmic resistance of the stack interfaces decreased from 6.89 to 5.33 mΩ, attributed to the contact optimization at the interconnect.
This study was conducted under fixed external conditions, reflecting the ideal state for SOEC operation. However, SOEC applications often involve more complex environments. Factors such as thermal cycling, fluctuating current densities, and variable external pressure may significantly affect the stack components. Future work will explore these factors to better understand their impact on component degradation and interfacial stability, thereby guiding the design of improved SOEC stacks.
PDF
(7850KB)
