Developing Solid Composite Polymer Electrolytes and Unveiling Layered Oxide Cathodes through Machine Learning for Sodium-Ion Batteries
Date
2024-05-21
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Solid-state sodium-ion batteries (ss-SIBs) are becoming a viable substitute for traditional lithium-ion batteries, offering a sustainable and cost-efficient option for future energy storage needs. The primary advantage is the abundant availability and lower cost of sodium compared to lithium. The advancement of ss-SIBs depends on achieving superior electrochemical, mechanical, interfacial, and thermal stability in solid electrolytes and optimal cathode selection. Solid polymer electrolytes (SPEs) are particularly promising due to their flexibility and potential for enhancement, despite challenges such as low ionic conductivity and high interfacial resistance. This thesis introduces a solid composite polymer electrolyte (SCPE) aimed at enhancing sodium-ion conductivity in ss-SIBs. The SCPE film was produced via a simple solution casting method, incorporating poly(vinylidene fluoride) (PVDF), poly(vinyl pyrrolidone) (PVP), succinonitrile as binders, NaPF6 salt, and NZSP NASICON as a ceramic electrolyte additive. Characterization techniques, including electrochemical impedance spectroscopy (EIS), scanning electron microscopy (SEM), and X-ray diffraction (XRD), were used to analyze the microstructure and electrochemical properties of the SCPEs. The resulting SCPEs exhibited a total conductivity of 6.8 × 10-4 S/cm at 23°C and 2.5 × 10-3 S/cm at 71°C. SEM analysis revealed uniform dispersion of the ceramic electrolyte within the SPE matrix, attributed to the polar nature of the host polymer, which reduces crystallinity and enhances sodium-ion conductivity. A symmetric half-cell assembly with a sodium electrode demonstrated excellent performance in sodium plating and stripping at a current density of 7 mA cm-2 at 23°C. Further, the thesis explores a machine-driven approach to predict critical factors affecting ss-SIB cathode performance using a dataset of about 350 data points of transition metal layered oxide cathode materials. Machine learning techniques were employed to develop three interconnected models aimed at predicting the P2/O3 ratio, initial discharge capacity, and discharge capacity after 50 cycles. The model for predicting the P2/O3 ratio achieved an R2 value of 83%, indicating high accuracy. The subsequent models, using Gaussian Process (GP) and Multilayer Perceptron Regressor (MLPR), achieved around 80% accuracy in predicting initial discharge capacity and 85% accuracy in discharge capacity after 50 cycles.
Description
Keywords
Solid-state batteries, Machine-driven studies, Solid composite electrolyte
Citation
Salari, H. (2024). Developing solid composite polymer electrolytes and unveiling layered oxide cathodes through machine learning for sodium-ion batteries (Master's thesis, University of Calgary, Calgary, Canada). Retrieved from https://prism.ucalgary.ca.