JCS 第134巻 第10号 2026年10月 1 日発行(毎月1回1日発行) ISSN 1348-6535 CODEN: JCSJEW Japan - October vol.134 Journal of the Ceramic Society of Japan 2026 Journal of the Ceramic Society of Japan 134 [10] 689–696 2026 DOI https://doi.org/10.2109/jcersj2.26043 JCS - Japan FULL PAPER Nanocomposite-derived Ni/YSZ hydrogen electrode active layer for enhancing performance of solid oxide electrolysis cells Yurika Sone1,2, Kazuyoshi Sato1,³, Toshiaki Yamaguchi2,3, Katherine Develos-Bagarinao3 and Haruo Kishimoto3 1 Division of Materials & Environments, Graduate School of Science and Technology, Gunma University, 1–5–1 Tenjin-cho, Kiryu, Gunma 376–8515, Japan 2 National Institute of Advanced Industrial Science and Technology, Energy Process Research Institute, AIST Tsukuba-West, 16–1 Onogawa, Tsukuba, Ibaraki 305–8569, Japan 3 National Institute of Advanced Industrial Science and Technology, Global Zero Emission Research Center, AIST Tsukuba-West, 16–1 Onogawa, Tsukuba, Ibaraki 305–8569, Japan A Ni/YSZ (Yttria-stabilized zirconia; YSZ) hydrogen electrode active layer was employed to improve the performance of solid oxide electrolysis cells (SOECs). The active layer was derived from pre-fabricated NiO/ YSZ nanocomposite particles grown by a co-precipitation method. The hydrogen electrode supported cells with and without the active layer were prepared to investigate the effect of the active layer on performance and durability. The cell with the active layer exhibited a high current density of 1.93 A·cm¹2 at 750 °C and 1.3 V. The performance was approximately 1.5 times higher than that of the cell without the active layer. Electrochemical impedance spectroscopy (EIS) revealed that the higher performance can be attributed not only to a decrease in polarization resistance but also to a reduction in ohmic resistance. Furthermore, the cell with the active layer showed durability comparable to or higher than that of the cell without the active layer, despite being composed of submicron-sized fine grains and pores. Keywords: Solid oxide electrolysis cells, Ni/YSZ cermet, Nanocomposite particles, Hydrogen electrode active layer, Steam electrolysis [Received May 8, 2026; Accepted July 25, 2026] 1. Introduction Solid oxide electrolysis cells (SOECs) are regarded as a promising electrochemical device for efficiently converting electric power into chemical energy.1) SOECs typically operate at high temperatures of 700–800 °C, so the total energy required for hydrogen production can be supplied as not only electricity but also heat. Moreover, the high temperature operation results in reducing the reaction overpotential. For these reasons, the required operating voltage for electrolysis in SOECs is significantly lower than other electrolyzers operating at lower temperatures (e.g., alkaline electrolysis cells, proton-exchange-membrane electrolysis cells) owing to the utilization of thermal energy at elevated temperatures. Consequently, the energy conversion efficiency of SOECs is superior to those of other electrolysis technologies. When cell performance is improved (i.e., reaction overpotential and ohmic resistivity of the cell are reduced), higher current density can be obtained at the same operating voltage. This will lead to reduction of hydrogen ³ Corresponding author: K. Sato; E-mail: kazuyoshi-sato@ gunma-u.ac.jp ©2026 The Ceramic Society of Japan production costs. Therefore, considerable efforts have been devoted to material development and cell design to decrease the internal cell resistance. The hydrogen-electrodesupported cells are one of the most common cell configurations in SOECs. They have thin electrolyte films, resulting in a reduction in the internal ohmic resistance. For reduction of polarization resistance, it is necessary to realize both good gas diffusivity in the hydrogen electrode support and a higher density of reaction sites at the electrode/electrolyte interfaces. To improve mass transportability through the hydrogen electrode support, a large amount of pore formers is typically added to increase its open porosity.2–6) In contrast, increasing the porosity of hydrogen electrode leads to a decrease in the number of reaction sites, known as triple phase boundaries (TPBs) consisting of electrode material, electrolyte material and gas phase. To increase the number of TPBs, an active layer is generally introduced into the hydrogen electrode supported cells.7,8) A Ni-YSZ cermet is commonly used as the hydrogen electrode active layer material in SOECs. Recently, we have reported that the hydrogen electrode, derived from the NiO/YSZ nanocomposite particles synthesized via a liquid-phase method, exhibits superior steam electrolysis performance compared with electrodes 689 This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by-nc/4.0/), which permits use, distribution, and reproduction in any medium for non-commercial purposes, provided the original work is properly cited. Sone et al.: Nanocomposite-derived Ni/YSZ hydrogen electrode active layer for enhancing performance of solid oxide electrolysis JCS-Japan cells prepared from commercial powders. This performance enhancement is attributed to the formation of a high density of TPBs, resulting from the uniform distribution of fine Ni, YSZ and pore structures.9) In this study, a hydrogen electrode active layer containing a high density of reaction sites was fabricated using the NiO/YSZ nanocomposite particles on a hydrogen electrode-supported cell with high porosity. As a result, good gas diffusivity and a high density of electrochemical reaction sites were achieved. The effect of introducing the active layer on the polarization resistance (Rpol) and ohmic resistance (Rohm) was electrochemically evaluated. In addition, the durability was also examined. It was revealed that incorporation of the active layer significantly enhanced the SOEC performance without negatively affecting durability. 2. Experimental Coin-sized hydrogen electrode-supported cells were adopted in this study. The cells consisted of a thin and porous hydrogen electrode support layer designed in order to promote gas diffusion and dense Ce0.9Gd0.1O2¹¤ (GDC) interlayer. This GDC interlayer reduced ohmic resistance, enabling accurate evaluation of the effect of active layer. These cells were prepared under identical fabrication conditions, except for the presence or absence of the active layer. The detailed preparation procedure is described below. Hydrogen-electrode supported SOECs with sizes of ³20 mm diameter were fabricated according to the following sequence. First, green sheet for hydrogen-electrode support layer of approximately 750 ¯m thick was fabricated by tape casting of a slurry containing commercial NiO (Type-F, Nikko Rika Co., Japan) and YSZ (TZ-8Y, Tosoh Co., Japan) powders with mass fraction of 0.656:0.344 (corresponding to a volume fraction of Ni and YSZ of 0.5:0.5 in cermet), polymethyl methacrylate powder Fig. 1. 690 (PMMA, Soken Kagaku Co., Japan) as pore former, and organic substances including dispersant (polyethyleneimine, Junsei Kagaku Co., Japan), binder (polyvinyl butyral, Sekisui Chem., Co., Japan) and plasticizer (di-butyl phthalate, Kanto Chemicals Co., Japan). After drying, the sheet was cut into disks with ³30 mm diameter. A slurry containing the NiO/YSZ nanocomposite particles grown via a co-precipitation approach9) was spin-coated onto the supporting layer disk to form the hydrogen electrode active layer. Then YSZ electrolyte layer was deposited onto the active layer by spin coating. Afterwards, the laminated layers were co-fired at 1350 °C for 2 h in air. A dense GDC interlayer of approximately 2 ¯m thick was fabricated onto the YSZ electrolyte using pulsed laser deposition.8) Finally, the air electrode with ³10 mm diameter was fabricated by screen printing method using commercial La0.6Sr0.4Co0.2Fe0.8O3¹¤ (AGC Seimi Chemical Co., Japan) and GDC (Shinetsu Chemical Co., Japan) powders with a mass fraction of 0.7:0.3, followed by sintering at 1000 °C in air. A cell without the hydrogen electrode active layer was also prepared using identical conditions for systematic comparison. Hereafter, the cells with and without the active layer are denoted as Cell A and Cell B, respectively. The electrochemical performance was evaluated using the pseudo four-terminal method with a ceramic holder (CHINO Co., Japan) as schematically shown in Fig. 1. The cell was set on the holder and then placed into an electric furnace and heated up to 750 °C. The flow rate of synthetic air supplied to the air electrode side was 200 sccm and the gas to the hydrogen electrode side was set to 100 sccm of 3 %H2O-97 %H2 to perform NiO reduction. After the completion of NiO reduction into Ni, the gas composition of the hydrogen electrode side was changed to 100 sccm of 30 %H2O-70 %H2, and this condition was kept during the performance evaluation. Schematic illustration of performance testing setup of hydrogen electrode supported SOEC. Journal of the Ceramic Society of Japan 134 [10] 689–696 2026 The performance evaluation was carried out at 750 °C using an electrochemical interface (1287 Solartron analytical Inc., UK) with a frequency response analyzer (1260A, Solartron analytical Inc., UK). Current densitycell voltage (I-V) characteristics were measured from open-circuit voltage (OCV) up to 1.5 V. The short-term durability of electrolysis performance was evaluated by recording the time-dependent current density change at potentiostatic conditions at 1.3 V. Electrochemical impedance spectroscopy (EIS) was performed at OCV in the frequency range 10¹1–105 Hz with an applied amplitude of 10 mV. The EIS spectra were deconvoluted and analyzed using distribution of relaxation times (DRT) approach.10,11) The microstructures of the post-test Ni/YSZ hydrogen electrodes were observed using SEM (Helios 5 Hydra CX, ThermoFisher). Prior to the observation, the fractured cells were embedded in resin and mechanically polished. 3. Results and discussion Figure 2 shows the cross-sectional SEM images of the as-prepared Cell A and Cell B. Cell A exhibits a uniformly distributed microstructure composed of small NiO Cross-sectional SEM backscattered electron images of the as-prepared (a) Cell A and (b) Cell B. (Dark gray, light gray and black regions correspond to NiO, YSZ and pore respectively.) Fig. 2. JCS-Japan (dark gray) and YSZ (light gray) grains. This structure has the potential to provide a higher density of TPBs, as fine pores are formed through the reduction of NiO. On the other hand, Cell B shows larger NiO and YSZ grains with a nonuniform distribution. Figure 3(a) shows the I-V characteristics of Cell A and Cell B. No significant gas leakage was confirmed for both cells, since the OCV of 0.980 V is close to that of the theoretical value of 0.992 V calculated based on the Nernst equation. Cell A showed a very high current density of 1.94 A·cm¹2 at a cell voltage of 1.3 V, which is approximately 1.5 times higher than that of Cell B (1.31 A·cm¹2). The corresponding hydrogen production rates at 1.3 V were 0.81 and 0.55 L·h¹1·cm¹2 for Cell A and Cell B, respectively, assuming a Faradaic efficiency of unity. The higher performance of Cell A can be attributed to inserting the Ni/YSZ hydrogen electrode active layer, since the only difference between Cell A and Cell B is the presence of this layer. Figure 3(b) shows a comparison of the current densities at 750 °C and 1.3 V for Cell A and Cell B with Fig. 3. (a) Voltage-current density characteristics of Cell A and Cell B, and (b) current density dependence on the steam content in the hydrogen electrode gas for the cells in this study and the literature. 691 Sone et al.: Nanocomposite-derived Ni/YSZ hydrogen electrode active layer for enhancing performance of solid oxide electrolysis JCS-Japan cells Fig. 5. Current density changes of Cell A and Cell B during 150 h operation measured at 750 °C and 1.3 V under 30 %H2O70 %H2. (a) EIS spectra of the cells and (b) DRT data measured at 750 °C and OCV. Fig. 4. literature data published in the last decade.7,8,12–19) The literature data were selected from hydrogen-electrode supported cells with configurations similar to that used in this study, namely, Ni/YSZ hydrogen electrode, YSZ electrolyte, GDC interlayer and mixed ionic electronic conductive perovskite-based air electrode. Cell A with the active layer exhibited the highest performance compared with the literature data including cells with active layer, despite operating under a relatively lower steam partial pressure. Therefore, there is a possibility that Cell A would show a higher current density under higher steam conditions. In contrast, the performance of Cell B is consistent with the trend of the literature data. Figure 4(a) shows the EIS spectra of the cells measured at 750 °C under OCV condition. Rohm represented by the high-frequency intercept of the impedance arc was 0.060 and 0.076 ³·cm2 for Cell A and Cell B, respectively. The reason for this will be discussed in detail after. The impedance spectra were deconvoluted into five peaks (P1–P5) in DRT analysis, as shown in Fig. 4(b), for understanding the impact of the active layer on the reaction processes at the hydrogen electrode. The DRT peaks of P2, P3, P4 and P5 were decreased by insertion of the hydrogen electrode active layer and were determined to be associated with hydrogen electrode reactions. In particular, the DRT peak of P5 at the highest frequency range changed significantly, accounting for close to 70 % of the 692 total variation. The peak intensity of P5 is associated with charge-transfer reaction at the hydrogen electrode,20–22) suggesting that the increased density of TPBs contributed to an enhanced charge-transfer reaction rate. In contrast, the peak intensity of P1, associated with gas-diffusion process at the hydrogen electrode is almost the same for Cell A and Cell B at OCV, and remains unchanged with the insertion of the active layer. Figure 5 shows the short-term durability testing results of the SOECs with potentiostatic condition at 1.3 V and 750 °C. The initial current densities of Cell A and Cell B were 1.97 and 1.33 A·cm¹2, respectively. After the durability testing, the current densities of Cell A and Cell B decreased to 1.84 and 1.19 A cm¹2, respectively. The increases in resistance of Cell A and Cell B per unit time during durability testing were calculated assuming an overpotential of 0.3 V, yielding values of 8 and 13 m³·cm2·h¹1, respectively. Figure 6(a) shows Rpol before and after the short-term durability test. Rpol of Cell A was lower than that of Cell B before and after testing. Moreover, the rates of Rpol increase during the test were almost the same for Cell A and Cell B, at 0.3 m³·cm2·h¹1. The DRT data for Cell A and Cell B are shown in Figs. 6(b) and 6(c), respectively. The DRT peaks corresponding to P2, P3, P4 and P5 increased in both Cell A and Cell B. As discussed earlier, these peaks were associated with hydrogen electrode reactions. The cross-sectional SEM backscattered electron images of the cells after short-term durability testing are shown in Fig. 7 to correlate the microstructure of the hydrogen electrode with the performance and durability. The active layer fabricated from the NiO/YSZ nanocomposite particles is approximately 10 ¯m thick with fine and uniformsized pores as shown in Fig. 7(a). In contrast, the hydrogen electrode of Cell B [Fig. 7(b)] exhibits a highly porous and nonuniform microstructure. Larger pores with sizes of several micrometers originated from the pore former (PMMA), while the finer ones were formed mainly by the reduction of NiO. Journal of the Ceramic Society of Japan 134 [10] 689–696 2026 JCS-Japan (a) Rpol of Cell A and Cell B and DRT data of (b) Cell A and (c) Cell B measured at 750 °C and OCV before and after short-term test. Fig. 6. Figures 8(a) and 8(b) shows magnified SEM images near the electrolyte/hydrogen electrode interfaces of Cell A and Cell B, respectively. In these images, Ni grains appear more protruded compared to the flatter features of the YSZ phase,23) and are highlighted with yellow boundaries. It is clear that the Ni and YSZ grains in the active layer are finer and more uniformly distributed compared to the hydrogen electrode of Cell B. The porosity calculated from the images of analysis area (frame with a red dotted lines) by Image J analysis were 27 and 45 %, respectively. The active layer of Cell A had homogeneous microstructure with smaller Ni and YSZ grain sizes and porosity, which led to increased TPB density in comparison to the inhomogeneous and more porous structure of Cell B. Such a microstructure of the active layer contributes to the reduction of Rpol. Figure 9 shows the average grain size of Ni calculated from the area near the electrolyte in the SEM backscattered electron images before and after durability test. The Ni grain size of Cell A was smaller than that of Cell B. However, Cell A and Cell B showed similar increases in Ni grain size. Therefore, one possible reason why the resistance component (P2, P3, P4 and P5) and Rpol increased to a similar extent regardless of the presence of the active layer was that these structural changes slightly reduced the electrochemically active reaction sites. However, it had no negative effect such as impeding gas diffusion and degradation with the active layer. Figure 10 shows the breakdown of Rohm before and after short-term durability testing. The contributions of YSZ (Rohm,YSZ) and GDC (Rohm,GDC) were calculated based on their thicknesses estimated from Fig. 7 and their ionic conductivities.24) The values of Rohm,YSZ and Rohm,GDC were almost identical for Cell A and Cell B. On the other hand, the region shown in blue represents the difference between the cells, which is attributed to the O2¹ transport resistance through the YSZ network in the hydrogen electrode (Rohm;O2 trans: ). After 150 h, only slight changes in Rohm;O2 trans: were observed for Cell A and Cell B assuming that Rohm,YSZ and Rohm,GDC did not change. Figure 8 is analyzed to explain the difference of Rohm;O2 trans: between Cell A and Cell B. The first is that the actual effective electrode area of Cell B is smaller than that of Cell A although the current density was calculated using the same electrode area (0.785 cm2). The contact area fraction at the active layer/electrolyte interface in Cell A and Cell B, calculated from Fig. 8, was 46.4 and 34.4 %, respectively. The interfacial contact area fraction of Cell B between the active layer and electrolyte was lower than that of Cell A. In other words, Cell B has a smaller number of 693 Sone et al.: Nanocomposite-derived Ni/YSZ hydrogen electrode active layer for enhancing performance of solid oxide electrolysis JCS-Japan cells Fig. 7. Cross-sectional SEM backscattered electron images of (a) Cell A and (b) Cell B after performance testing. (Yellow arrow indicates the interface at the active layer/electrolyte or hydrogen electrode support layer/ electrolyte.) O2¹ current pathways within the hydrogen electrode, i.e., a structure with less parallel circuits of O2¹ pathways, compared to Cell A. In a parallel circuit, the resistance increases as the number of parallel paths decreases. Therefore, Rohm of Cell B was larger than that of Cell A. The second is that the O2¹ migration length of Cell B in the hydrogen electrode is longer than that of Cell A. As shown in Fig. 8(a), the active layer had a higher density of TPBs so the electrochemically effective electrode thickness will be thin. Moreover, the large pores originated from PMMA existed everywhere in the hydrogen electrode of Cell B. Therefore, O2¹ generated from TPB of Cell B traveled a tortuous route due to increased tortuosity of O2¹ transport whereas O2¹ moved a relatively straight path in Cell A. From these reasons, it caused increasing O2¹ transport distance of Cell B and its Rohm became larger than that of Cell A. Thus, the higher Rohm observed for Cell B can be attributed to the lower connectivity of its relatively sparse YSZ network, which limits ionic transport pathways. In contrast, the denser and more continuous YSZ network in Cell A facilitates O2¹ conduction, thereby resulting in a lower Rohm. Furthermore, the slight increase in Rohm after 150 h, suggests that the YSZ networks in the 694 hydrogen electrodes of both Cell A and Cell B remained stable throughout the durability testing. 4. Conclusion Introduction of a Ni/YSZ hydrogen electrode active layer using the pre-fabricated NiO/YSZ nanocomposite particles is a promising approach for remarkable enhancement of steam electrolysis performance of SOECs without sacrificing durability. The higher performance can be attributed to the reduction of both Rohm and Rpol. The nanocomposite active layer which contains many active TPBs due to homogeneous distribution of fine Ni particles and pores can realize a decrease in the polarization resistance without impeding gas diffusion. Rohm of Cell A was also improved compared to Cell B because it has many contact interfaces between the electrolyte and hydrogen activation layers, and the transport distance of O2¹ in Cell A was shorter than that in Cell B. In consequence, the current density of Cell A with the nanocomposite active layer was 1.94 A·cm¹2 at 1.3 V and 750 °C, which is higher than reported values including cells with an active layer. In addition, the cell with active layer maintained high performance without significant degradation after 150 h. JCS-Japan Journal of the Ceramic Society of Japan 134 [10] 689–696 2026 Fig. 8. Cross-sectional SEM backscattered electron images near electrolyte/hydrogen electrode interface of (a) Cell A and (b) Cell B after performance testing. (Dark gray, light gray and black regions correspond to Ni, YSZ and pore respectively.) The average Ni grain size of Cell A and Cell B before and after durability test. Fig. 9. Acknowledgement This study is partially based on results obtained from a project, JPNP21022, commissioned by the New Energy and industrial Technology Development Organization (NEDO). Fig. 10. The breakdown of Rohm in Cell A and Cell B measured at 750 °C and OCV before and after short-term test. References 1) S. E. Wolf, F. E. Winterhalder, V. Vibhu, L. G. J. de Haart, O. Guillon, R.-A. Eichel and N. H. Menzler, J. Mater. Chem. A 11, 17977 (2023). 695 Sone et al.: Nanocomposite-derived Ni/YSZ hydrogen electrode active layer for enhancing performance of solid oxide electrolysis JCS-Japan cells 2) K. J. Yoon, J.-W. Son, J.-H. Lee, B.-K. Kim, H.-J. Je and H.-W. Lee, ECS Transactions 57, 3099 (2013). 3) H. Shimada, T. Yamaguchi, T. Suzuki, H. Sumi, K. Hamamoto and Y. Fujishiro, J. Ceram. Soc. Jpn. 124, 213 (2016). 4) H. Shimada, T. Suzuki, T. Yamaguchi, H. Sumi, K. Hamamoto and Y. Fujisjiro, J. Power Sources 302, 53 (2016). 5) M. A. Laguna-Bercero, A. R. Hanifi, L. Menand, N. K. Sandhu, N. E. Anderson, T. H. Etsell and P. Sarkar, Electrochim. Acta 268, 195 (2018). 6) L. Mingyi, Y. Bo, X. Jingming and C. Jing, Int. J. Hydrogen Energ. 35, 2670 (2010). 7) R. Yang, Y. Tian, Y. Liu, J. Pu and B. Chi, J. Rare Earth. 41, 599 (2023). 8) R. A. Budiman, T. Yamaguchi, T. Ishiyama, K. DevelosBagarinao, K. Yamaji and H. Kishimoto, Mater. Lett. 309, 131419 (2022). 9) Y. Sone and K. Sato, Mater. Lett. 407, 140054 (2026). 10) H. Schichlein, A. C. Müller, M. Voigts, A. Krügel and E. Ivers-Tiffée, J. Appl. Electrochem. 32, 875 (2002). 11) A. Maradesa, B. Py, J. Huang, Y. Lu, P. Iurilli, A. Mrozinski, H. M. Law, Y. Wang, Z. Wang, J. Li, S. Xu, Q. Meyer, J. Liu, C. Brivio, A. Gavrilyuk, K. Kobayashi, A. Bertei, N. J. Williams, C. Zhao, M. Danzer, M. Zic, P. Wu, V. Yrjänä, S. Pereverzyev, Y. Chen, A. Weber, S. V. Kalinin, J. P. Schmidt, Y. Tsur, B. A. Boukamp, Q. Zhang, M. Gaberšček, R. O’Hayre and F. Ciucci, Joule 8, 1958 (2025). 12) H. Shimada, T. Yamaguchi, T. Suzuki, H. Sumi, K. 696 13) 14) 15) 16) 17) 18) 19) 20) 21) 22) 23) 24) Hamamoto and Y. Fujishiro, J. Ceram. Soc. Jpn. 124, 213 (2016). L. Zhou, C. Li, L. Fan, D. Nematov and L. Jia, Sustain. Mater. Techno. 46, e01759 (2025). X. Liu, Y. Li, C. Xiong, B. Chi and J. Pu, Chem. Eng. J. 519, 164428 (2025). J. H. Park, K. J. Kim, C. H. Jung, W. Jeong, H. R. Shin, J.-E. Hong and K. T. Lee, Fuel Cells 20, 712 (2020). H. Shimada, Y. Fujimaki and Y. Fujishiro, Ceram. Int. 46, 19617 (2020). J. Lin, L. Chen, T. Liu, C. Xia, C. Chen and Z. Zhan, J. Power Sources 374, 175 (2018). Q. Ma, S. Dierickx, V. Vibhu, D. Sebold, L. G. J. de Haart, A. Weber, O. Guillon and N. H. Menzler, J. Electrochem. Soc. 167, 084522 (2020). S. B. Karki, L. Le, L. Seymour, R. Springer, T. Liu and O. A. Marina, ACS Appl. Mater. Inter. 17, 58270 (2025). X. Sun, P. V. Hendriksen, M. B. Mogensen and M. Chen, Fuel Cells 19, 740 (2019). S. N. Sampathkumar, P. Aubin, K. Couturier, X. Sun, B. R. Sudireddy, S. Diethelm, M. Pérez-Fortes and J. Van herle, Int. J. Hydrogen Energ. 47, 10175 (2022). P. Caliandro, A. Nakajo, S. Diethelm and J. Van herle, J. Power Sources 436, 226838 (2019). M. B. Mogensen, M. Chen, H. L. Frandsen, C. Graves, A. Hauch, P. V. Hendriksen, T. Jacobsen, S. H. Jensen, T. L. Skafte and X. Sun, Fuel Cells 21, 415 (2021). H. Shi, C. Su, R. Ran, J. Cao and Z. Shao, Prog. Nat. Sci.-Mater. 30, 764 (2020).