Nick Brady is the first CEEC PhD graduate

August 13, 2019

Investigation of Lithium Ion Battery Electrodes: Using Mathematical Models Augmented with Data Science to Understand Surface Layer Formation, Mass Transport, Electrochemical Kinetics, and Chemical Phase Change

Abstract:

This thesis first uses physical scale models to investigate solid-state phenomena - surface layer formation, solid-state diffusion of lithium, electrochemical reactions at the solid-electrolyte interface, as well as homogeneous chemical phase change reactions. Evidence is provided that surface layer formation on the magnetite, Fe3O4, electrode can accurately be described mathematically as a nucleation and growth process. To emulate the electrochemical results of the LiV3O8 electrode, a novel method is developed to capture the phase change process; this method describes phase change as a nucleation and growth process. The physical parameters of the LiV3O8 electrode: the solid-state diffusion coefficient, phase change saturation concentration, phase reaction rate constant, and exchange current density, are all quantified and the agreement with experimental results is compelling. Electrochemical evidence, corroborated by results from density functional theory, indicate that delithiation is a more facile process than lithiation in the LiV3O8 electrode.

Further investigation of the LiV3O8 electrode is undertaken by coupling the crystal scale model to electrode scale phenomena. Characterization of the LiV3O8 electrode by operando EDXRD experiments provides a unique and independent set of observations that validate the previously estimated physical constants for the phase change saturation concentration and phase change reaction rate constant; they are both found to be consistent with their previous estimates. Finally, it is observed that anodic physical phenomena are important during delithiation of the cathode because the kinetics at the anode become mass-transfer limited.

Finally, it is illustrated that coupling physical models to data science and algorithmic computing is an effective method to accelerate model development and quantitatively guide the design of experiments.