【公開日：2025.06.10】【最終更新日：2025.04.15】
課題データ / Project Data
課題番号 / Project Issue Number
24NM0045
利用課題名 / Title
Microstructure of electrochemical energy devices
利用した実施機関 / Support Institute
物質・材料研究機構 / NIMS
機関外・機関内の利用 / External or Internal Use
外部利用/External Use
技術領域 / Technology Area
【横断技術領域 / Cross-Technology Area】（主 / Main）計測・分析/Advanced Characterization（副 / Sub）-
【重要技術領域 / Important Technology Area】（主 / Main）革新的なエネルギー変換を可能とするマテリアル/Materials enabling innovative energy conversion（副 / Sub）-
キーワード / Keywords
Machine learning,電極材料/ Electrode material,電子顕微鏡/ Electronic microscope,燃料電池/ Fuel cell,集束イオンビーム/ Focused ion beam
利用者と利用形態 / User and Support Type
利用者名（課題申請者）/ User Name (Project Applicant)
SCIAZKO Anna
所属名 / Affiliation
東京大学 Institute of Industrial Science
共同利用者氏名 / Names of Collaborators in Other Institutes Than Hub and Spoke Institutes
Naoki Shikazono,Nakazawa Yuichiro
ARIM実施機関支援担当者 / Names of Collaborators in The Hub and Spoke Institutes
利用形態 / Support Type
（主 / Main）機器利用/Equipment Utilization（副 / Sub）-
利用した主な設備 / Equipment Used in This Project
報告書データ / Report
概要（目的・用途・実施内容）/ Abstract (Aim, Use Applications and Contents)
The performance of many energy devices, in particular solid oxide cell (SOC), depends greatly on their microstructure. The three dimensional quantitative evaluation of microstructural parameters can help in clarifying the dependence between microstructure and performance as well as degradation mechanisms. At the same time machine learning can help in prediction and evaluation of the physical mechanisms in electrochemical devices. Here, the advanced measurement techniques are couples with developed machine learning processing methods.
実験 / Experimental
The various designs of Solid Oxide Fuel Cells (SOFC) and Solid Oxide Electrolysis Cells (SOEC) electrodes were fabricated to study the dependence between SOC performance and materials, particle morphology and electrochemical operation strategies. The accelerated electrochemical degradation tests were carried out. In order to investigate microstructural properties, the porous SOC samples were resin infiltrated and polished by the CP polishing. The sequence of cross-sectional images was acquired by a dual beam Focused Ion Beam Scanning Electron Microscopes (FIB-SEM SMF-1000) and reconstructed into 3D model. Additionally, the analysis by the Energy-dispersive X-ray spectroscopy (EDX) mapping was carried out.
結果と考察 / Results and Discussion
The focus of this study was on microstructural observations of solid oxide cell (SOC) electrodes using FIB-SEM, with an emphasis on degradation mechanisms under different operational conditions. The investigations included stability of a pure gadolinium-doped ceria (GDC) electrode during solid oxide fuel cell (SOFC) and electrolysis cell (SOEC) operation and predicting microstructural evolution using machine learning. To achieve high-resolution imaging of the porous SOC electrodes, FIB-SEM was utilized to clearly recognize metal, ceramic, pores and contaminants phases. The results indicated that the primary degradation mechanism in nano-GDC electrodes is the formation of cerium silicates near the current collector layer, leading to pore clogging and restricted gas transport. Silica contamination, originating from raw materials, sealings, and environmental exposure, was identified as a key factor contributing to performance degradation. Additionally, a novel machine learning-based framework was applied to predict microstructural evolution in SOC cermet electrodes. The approach integrated an unsupervised image-to-image translation (UNIT) network with a physically constrained loss function to model the nickel reduction process. The framework demonstrated high predictive accuracy, even for microstructures absent from the training dataset. Furthermore, the conditional UNIT (C-UNIT) network was introduced to extend predictive capabilities by incorporating process conditions and time-dependent microstructural changes, enabling a more comprehensive understanding of degradation processes. These findings contribute to advancing SOC electrode design by addressing key degradation mechanisms and proposing mitigation strategies.
図・表・数式 / Figures, Tables and Equations
その他・特記事項（参考文献・謝辞等） / Remarks(References and Acknowledgements)
This work was partly supported by the New Energy and Industrial Technology Development Organization (NEDO), by Japan Society for the Promotion of Science KAKENHI [grant number 23K13261] and by Advanced Research Infrastructure for Materials and Nanotechnology in Japan (ARIM Japan).
成果発表・成果利用 / Publication and Patents
論文・プロシーディング（DOIのあるもの） / DOI (Publication and Proceedings)
-
Anna Sciazko, Prediction of electrode microstructure evolutions with physically constrained unsupervised image-to-image translation networks, npj Computational Materials, 10, (2024).
DOI: doi:10.1038/s41524-024-01228-3
口頭発表、ポスター発表および、その他の論文 / Oral Presentations etc.
- Sciazko, A., Komatsu, Y., Shimura, T. and Shikazono, N., "Mitigating degradation of GDC electrodes in SOEC operation," 16th European SOFC & SOE Forum, B1602, Lucerne, Switzerland, 2 - 5 July (2024).
- Sciazko, A., Komatsu, Y., Shimura, T. and Shikazono, N., "Predicting dynamic microstructure degradation by physically constrained machine learning,"16th European SOFC & SOE Forum, B0202, Lucerne, Switzerland, 2 - 5 July (2024).
- Sciazko, A., Komatsu, Y., Yamagishi, R., Lyu, Z., Tao, J., Shimura, T., Onishi, J., Nishimura, K., Shikazono, N., SOC Microstructural Studies with Machine Learning: Insights and Applications, 7th Asian SOFC Symposium, IB-07, Sapporo, Japan, October 23-25 (2024).
特許 / Patents
特許出願件数 / Number of Patent Applications：0件
特許登録件数 / Number of Registered Patents：0件