The goal of this work is to understand the relationships between electrode microstructure and mass transport resistances. One can use this information to predict cell performance from fundamental principles. This work includes new types of particle-scale 3D models for correlating and predicting the effects of electrode microstructure on both ionic and electronic transport. The 3D models imitate the sub-micrometer-scale arrangement of active material particles, carbon, binder, and pores and use FIB/SEM images as a basis for parameterization. The 3D models are based respectively on the statistical mechanics techniques of molecular dynamics and Monte Carlo. The approach closely related to molecular dynamics, named the dynamic particle packing (DPP) model, uses aggregates of spheres to recreate electrode microstructures. The other approach, named the stochastic grid (SG) model, is closely related to Monte Carlo techniques in which a small set of fundamental interdomain and bulk energy parameters are used to generate structures.In order to predict electrode microstructures we correlated the fundamental interdomain and bulk energy parameters for the SG model to electrode mass composition and porosity. We used the revised computer program, known as predict SG, to estimate structures of which there are no experimental measurements of electrode structure. From these predicted electrode structures we obtained electronic and ionic transport properties. This allowed us to estimate the trade-offs between ionic and electronic transport for different porosities and carbon fractions. We found from experimental measurements of electrode structure that carbon and binder formed distinct agglomerates. From the 3D models we determined at commercial fractions of carbon and binder that the conductivity of these carbon agglomerates plays a large role in determining both the electronic and ionic pathways. So in order to better understand the role that these carbon/binder agglomerates play, we explored and developed several experimental methods to find the electronic and ionic conductivity of both simulated carbon domains and complete electrode films. The goal was not only to elucidate the role carbon agglomerates play, but also to develop a non-destructive method of determining overall film properties. Although we found that a non-destructive method is extremely challenging due to probe contact resistances, we did find success in determining carbon domain properties using a delamination method.

The goal of this work is to develop a model to predict the microstructure of Li-ion batteries, specifically focusing on the cathode component of the batteries. This kind of model has the potential to assist researchers and battery manufacturers who are trying to optimize the capacity, cycle life, and safety of batteries. Two dynamic particle packing (DPP) microstructure models were developed in this work. The first is the DPP1 model, which simulates the final or dried electrode structure by moving spherical particles under periodic boundaries using Newton's laws of motion. The experience derived from developing DPP1 model was beneficial in making the final model, called DPP2. DPP2 is an improved version of DPP1 that includes solvent effects and is used to simulate the slurry-coating, drying, and calendering processes. Two type of properties were used to validate the DPP1 and DPP2 models in this work, although not every property was used with the DPP1 model. First are the structural properties, which include volume fraction, and electronic and ionic conductivities. Experimental structural properties were determined by analyzing 2D cross sectional images of the battery cathodes. These images were taken through focused ion beam (FIB) planarization and scanning electron microscopy (SEM). The second category are the mechanical properties, which include film elasticity and slurry viscosity. These properties were measured through experiments executed by our group. The DPP2 model was divided into two submodels : active-free and active-composite. The 2D cross sectional images of the simulated structure of the models have a similar particle arrangements as the experimental structures. The submodels show reasonable agreement with the experimental values for liquid and solid mass density, shrink ratio, and elasticity. For the viscosity, both models show shear-thinning behavior, which is a characteristic of slurries. The volume fractions of the simulated structures of the active-free and active-composite models have better agreement with the experimental values, which is also reflected in the 2D cross sectional images of the structure.