Using Machine Learning and Infrared Spectroscopy to Quantify Species Concentrations in Battery Electrolytes

Abstract

In this research, artificial intelligence, deep learning, and chemometric tools were coupled with operando spectroscopy of battery electrolytes to measure species concentrations and elucidate molecular interactions. FTIR spectra from an electrolyte composed of LiPF6 in ethylene carbonate (EC) and ethyl methyl carbonate (EMC) were analyzed with principal component analysis (PCA) and a convolutional neural network (CNN) to discern solvation behavior and quantify component concentrations during cell operation. PCA pinpointed exact band locations of solvation shifting behavior in the IR spectra and improved understanding of the relationship between spectral peak changes, lithium concentrations, and solvation behavior. The CNN was trained with spectral datasets of electrolytes with known lithium and solvent concentrations and made predictions with extraordinarily high accuracy. Additionally, the CNN interpreted FTIR spectral datasets from a graphite half-cell with EC/EMC/LiPF6 electrolyte and accurately determined the lithium concentration in the bulk electrolyte. The CNN also observed lithium depletion events in the graphite anode during battery cycling. These depletion events were previously investigated with traditional spectroscopic techniques but with large errors in absolute concentration. This research breaks new ground on using advanced computational tools for in situ and operando spectroscopic analysis of battery electrolytes to investigate complex molecular-level phenomena important for improving electrolyte transport and stability.