Revealing the Markers of Metal Anode Stability Using Substrate Physics and Statistical Learning
Speaker Details:
Solomon Oyakhire
Assistant Professor of Chemical and Biomolecular Engineering
Georgia Tech
Abstract
Next-generation batteries, such as lithium metal and zinc metal batteries, offer significant promise but face significant challenges. Although these metal anodes possess higher charge storage capacity than their non-metallic counterparts, the metals deposited while a battery charges react with the electrolyte. This reaction results in rapid capacity fade of metal anode batteries and has been the subject of scientific investigation for decades.
While there has been significant progress in engineering these batteries, primarily through chemical modification of the electrolyte (the medium that shuttles ions across the battery) and physico-chemical modification of the interfacial layers that isolate the reactive metals from reducible battery components, how these metals react and what truly quells their reactions remain poorly understood. This gap in understanding arises because chemically modifying a battery component to improve its stability changes several reaction mechanisms simultaneously, thereby obscuring the mechanisms that govern stability.
To reveal the principles that dictate battery stability, my group uses two approaches to disentangle the mechanisms of electrodeposition and passivation.
First, we leverage concepts in solid-state physics and nanoscience to build substrates that control electron transfer and thereby allow us to isolate the principles governing battery stability. In my talk, I will discuss some of these principles, ranging from metal-substrate interfacial energy and metal growth driven by radial diffusion in lithium metal batteries, to substrate topography and asperity distribution in zinc metal batteries. With other known drivers of stability held fixed, we show that these principles influence metal anode morphology and, as a result, battery stability.
Second, we leverage concepts in statistical learning to represent electrolytes as high-dimensional combinations of the mechanisms that collectively influence battery stability, thereby allowing us to isolate electrolyte-based drivers of battery stability. In my talk, I will present our work showing for the first time that machine learning can be used to identify electrolyte-based markers of stability in lithium metal batteries. Using our framework, we have found that solvation free energy is a strong driver of battery stability, that galvanic corrosion explains differences in stability among high-performance electrolytes, and, more recently, that explicitly adding molecular-scale solvation features improves the generalizability of our predictions to unseen solvents.
I will end my talk by discussing some grand challenges that remain in batteries and energy science and how my group is working towards addressing them.
Speaker Biography:
Solomon Oyakhire is an Assistant Professor of Chemical and Biomolecular Engineering at Georgia Tech. He earned his BSc in Chemical Engineering from the University of Lagos, Nigeria, graduating with the highest honors and recognized as the department's best graduating student. He completed his PhD in Chemical Engineering at Stanford University working on interfacial electrochemistry with Stacey Bent and Yi Cui as a Knight-Hennessy scholar. During his postdoctoral training at the University of California Berkeley, he developed artificial intelligence algorithms for high dimensional datasets with Kristin Persson and Gerbrand Ceder as a Schmidt Science Fellow.