Designing smart electrochemical devices using multifunctional components and data-driven analysis

Abstract

Local energy sources are becoming increasingly important for the future energy landscape. Electrochemical devices are one of the most promising options for addressing these needs that range from powering mobile electronics to stationary grid energy storage. Each application has a unique set of challenges and performance requirements. There is therefore a need for designing specialized cells with these criteria in mind. This work will demonstrate the use of multifunctional components and data-driven cell design of electrodes as routes to more efficient battery design and testing. The first project demonstrates textile based biomechanical energy harvesting and continuous motion sensing utilizing electrochemical generators (ECG), which rely on electrochemical-mechanical coupling to generate electricity at the timescale of human motion. The energy harvesting performance is optimized my maximizing the partial molar volume term, experimentally verifying the mechanical-electrochemical coupling mechanism. A scalable electrodeposition process produces nanoparticle coated fiber-based electrochemical cells that can be stitched into fabrics. This work demonstrates the promising performance and adaptability of the ECG platform for wearable energy generation. Next, a safer battery design is presented using a graphene modified separator within a cell to detect dendritic metal plating during cell operation. The 2D graphene layer adds negligible cell mass and is engineered with homogenous nanopores that facilitate ion transport, meaning that extra cell functionality is acquired without adverse effects on the gravimetric and/or volumetric battery performance. The strategy developed here is a promising route for designing multifunctional components within electrochemical cells through the use of nanoscale materials. The third chapter discusses the application of data-driven approaches to accelerate battery materials development. Namely, sequential Bayesian optimization is used to model the relationship between the properties of a carbon-based substrate for sodium metal plating and the performance of a half-cell constructed using this substrate. We quantify the increased optimization rate compared to brute force methods such as random selection of properties in finding the optimal cell performance. Through this work, we demonstrate sequential Bayesian optimization as an efficient tool for guiding the typically costly and slow process of battery materials optimization.