Key aspects of sustainable and high-performance K-ion batteries: A machine learning approach
Journal of Power Sources, cilt.657, 2025 (SCI-Expanded, Scopus)
- Yayın Türü: Makale / Tam Makale
- Cilt numarası: 657
- Basım Tarihi: 2025
- Doi Numarası: 10.1016/j.jpowsour.2025.238215
- Dergi Adı: Journal of Power Sources
- Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Chemical Abstracts Core, Chimica, Compendex, INSPEC
- Anahtar Kelimeler: Potassium-ion battery, Machine learning, Sustainability, XGBoost, Association rule mining, SHAP
- Boğaziçi Üniversitesi Adresli: Evet
Özet
As sustainable and affordable energy storage systems, potassium-ion (K-ion) batteries have received noteworthy attention due to the natural abundance of K, their promise for high performance with low cost. Yet, the complexity and the diversity of the materials and cell design factors make the development of these batteries complicated. Here, we conduct a detailed machine-learning analysis of research directions for sustainable and high-performance K-ion batteries. First, the author keywords are analyzed using 11,243 publications; one of the most dominant trends in publications is the development of carbon-based anodes. Then, extreme gradient boosting (XGBoost) and random forest (RF) prediction, SHAP analysis, and association rule mining are applied to a dataset constructed from 124 experimental papers, reporting 577 experiments. XGBoost/RF models can successfully predict the second discharge capacity and cycle life of the K-ion battery while SHAP and association rule mining analyses determine critical materials and design factors. The anode group, crystal structure, synthesis methods, and loading have the most prominent impact on the performance of anode half-cells, whereas cathode loading and synthesis methods are critical for the cathode studies. Structured electrodes exhibit better performance in terms of capacity and cycling compared to bare forms of both anode and cathode.