PhD Thesis Colloquium: Mr. Dereje Bekele Tekliye (27/02/26)
Thesis title:
Discovery and evaluation of cathode materials for calcium-ion batteries through first-principles and machine learning
Faculty advisor(s):
Prof. Sai Gautam Gopalakrishnan
When?
27th February, 2026 (Friday), 10:00 AM (India Standard Time)
Where
KPA Auditorium, Department of Materials Engineering
Abstract:
Electrochemical energy storage is central to a sustainable energy future, yet progress in lithium-ion batteries is increasingly limited by intrinsic materials constraints. Multivalent systems, particularly calcium ion batteries, promise higher energy densities using earth abundant elements but are impeded by the lack of cathodes that are thermodynamically stable and can reversibly host Ca at high voltage, fast rates, and practical capacities. In this thesis, these challenges are addressed through a multi scale, first principles driven computational framework that integrates high throughput density functional theory, rigorous methodological benchmarking, machine learning accelerated screening, and fundamental studies of thermodynamic and transport phenomena to derive predictive design rules and identify promising calcium cathode chemistries across a broad chemical space.
Exploiting the similarity between the ionic radii of Na+ and Ca2+ as a guiding strategy, we hypothesize that frameworks capable of reversible sodium intercalation may also serve as effective hosts for calcium. This work first investigates the polyanionic NaSICON framework within the CaxM2(ZO4)3 chemical space (M = transition metal; Z = Si, P, S). Our study identifies CaxV2(PO4)3, CaxMn2(SO4)3, and CaxFe2(SO4)3 as promising Ca-cathode candidates, exhibiting favorable thermodynamic stability, intercalation voltages, and Ca2+ migration barriers. To ensure quantitative reliability for fluoride-based systems, we benchmark the SCAN and r2SCAN meta-GGA functionals, optimizing Hubbard U parameters against experimental data. This rigorous validation enables the extension of the Na-Ca design strategy to weberite (CaxM2F7) and perovskite (CaxMF3) frameworks, identifying CaxCr2F7 and CaxMn2F7 as promising Ca-cathode candidates.
Beyond targeted structural families, a machine-learning-driven high-throughput workflow employing machine-learning foundation models and transfer-learning-based property predictors screens over 52,900 compounds from the Materials Project database. Combined with density functional theory-based nudged elastic band validation, this approach identifies 34 previously unexplored promising Ca-cathode frameworks. Finally, to explain the experimentally observed capacity limits and complex phase evolution in post spinel CaV2O4, this work employs cluster expansion and Monte Carlo simulations to construct temperature composition phase diagrams. The resulting thermodynamic landscape reveals a strong driving force for phase separation, while migration barrier analysis confirms that calcium diffusion is severely impeded at intermediate compositions, rationalizing the performance losses and reaction mechanisms of CaV2O4.
Overall, this thesis develops a transferable, data-driven theoretical framework for identifying and understanding promising calcium cathode materials, offering clear guidance for experimental validation and accelerating progress toward next-generation multivalent energy storage systems.