Nanoscale X-ray Computed Tomography Based Modeling of Lithium-ion Battery Electrodes

Because of their high energy/power density, long cycle life, and extremely low rate of self-discharge, lithium-ion batteries (LIBs) have dominated portable electronics, smart grid, and electric vehicles (EVs). Although they are the most developed and widely applied energy storage technology, there is still a strong desire to further enhance their energy/power density, cycle life, and safety. While all of these battery requirements are macroscopic and stated at cell/pack scale, they have to be addressed at particle or network of particles scale (mesoscale). At mesoscale, active material particles having different shape and morphologies are bound together with a carbon-doped polymer binder layer. This percolated network of particles serves as the electron conductive path from the reaction sites to the current collector. Even though significant research has been conducted to understand the physical and electrochemical behavior of material at the nanoscale, there have not been comprehensive studies to understand what is happening at the mesoscale. \nMathematical models have emerged as a promising way to shed light on complex physical and electrochemical phenomena happening at this scale. The idea of using mathematical model to study multiphysics behavior of LIBs is not new. Traditional models involved homogeneous spherical particles or computer generated electrode structures as the model geometry to simulate electrode/cell performance. While these models are successful to predict the cell performance, heterogeneous electrode’s structure at mesoscale questions the accuracy of their findings related to battery internal behavior and property distribution. \nThe new advances in the field of 3D imaging including X-ray computed tomography (XCT) and Focused-ion beam/Scanning electron microscopy (FIB-SEM), have enabled the 3D visualization of the electrode’s active particles and structures. In particular, XCT has offered nondestructive imaging and matter penetration capability in short period of time. Although it was commercialized in 70’s, with the recent development of high resolution (down to 20 nm) laboratory and synchrotron radiation tomography has been revolutionized. 3D reconstructed electrodes based on XCT data can provide quantitative structural information such as particle and pore size distribution, porosity, solid/electrolyte interfacial surface area, and transport properties. In addition, XCT reconstructed geometry can be easily adopted as the model geometry for simulation purposes. For this, similar to traditional models, a modeling framework based on conservation of mass/charge and electrochemistry needs to be developed. The model links the electrode performance to the real electrode’s structure geometry and allows for the detailed investigation of multiphysics phenomena. When combined with mechanical stress, such models can also be used for electrode’s failure and degradation studies. The work presented in this dissertation aims to adopt 3D reconstructed structures from nano-XCT as the geometry to study multiphysics behaviour of the LIBs electrodes. In addition, 3D reconstructed structure provides more realistic electrode’s morphological and transport properties. Such properties can benefit the homogeneous models by providing highly accurate input parameters. \nIn the first study, a multiscale platform has been developed to model LIB electrodes based on the reconstructed morphology. This multiscale framework consists of a microscale level where the electrode microstructure architecture is modeled and a macroscale level where discharge/charge is simulated. The coupling between two scales is performed in real time unlike using common surrogate based models for microscale. For microscale geometry 3D microstructure is reconstructed based on the nano-XCT data replacing typical computer generated microstructure. It is shown that this model can predict the experimental performance of LiFePO4 (LFP) cathodes at different discharge rates more accurately than the traditional/homogenous models. The approach employed in this study provides valuable insight into the spatial distribution of lithium within the microstructure of LIB electrodes. \nIn the second study, a new model that keeps all major advantages of the single-particle model of LIB and includes three-dimensional structure of the electrode was developed. Unlike the single spherical particle, this model considers a small volume element of an electrode, called the Representative Volume Element (RVE), which represent the real electrode structure. The advantages of using RVE as the model geometry was demonstrated for a typical LIB electrode consisting of nano-particle LFP active material. The model was employed to predict the voltage curve in a half-cell during galvanostatic operations and validated against experimental data. The simulation results showed that the distribution of lithium inside the electrode microstructure is very different from the results obtained based on the single-particle model. \nIn the third study, synchrotron X-ray computed tomography has been utilized using two different imaging modes, absorption and Zernike phase contrast, to reconstruct the real 3D morphology of nanostructured Li4Ti5O12 (LTO) electrodes. The morphology of the high atomic number active material has been obtained using the absorption contrast mode, whereas the percolated solid network composed of active material and carbon-doped polymer binder domain (CBD) has been obtained using the Zernike phase contrast mode. The 3D absorption contrast image revealed that some LTO nano-particles tend to agglomerate and form secondary micro-sized particles with varying degrees of sphericity. The tortuosity of the pore and solid phases were found to have directional dependence, different from Bruggeman’s tortuosity commonly used in homogeneous models. The electrode’s heterogeneous structure behaviour was also investigated by developing a numerical model to simulate a galvanostatic discharge process using the Zernike phase contrast mode. \nIn the last study, synchrotron X-ray nano-computed tomography has been employed to reconstruct real 3D active particle morphology of a LiMn2O4 (LMO) electrode. For the first time, CBD has been included in the electrode structure as a 108 nm thick uniform layer using image processing technique. With this unique model, stress generated inside four LMO particles with a uniform layer of CBD has been simulated, demonstrating its strong dependence on local morphology (surface concavity and convexity), and the mechanical properties of CBD such as Young’s modulus. Specifically, high levels of stress have been found in vicinity of particle’s center or near surface concave regions, however much lower than the material failure limits even after discharging rate as high as 5C. On the other hand, the stress inside CBD has reached its mechanical limits when discharged at 5C, suggesting that it can potentially lead to failure by plastic deformation. The findings in this study highlight the importance of modeling LIB active particles with CBD and its appropriate compositional design and development to prevent the loss of electrical connectivity of the active particles from the percolated solid network and power losses due to CBD failure. \nThere are still plenty of opportunities to further develop the methods and models applied in this thesis work to better understand the multiscale multiphysics phenomena happening in the electrode of LIBs. For example, in the multiscale model, microscale solid phase charge transfer and electrolyte mass/charge transfer can be included. In this way, heterogeneous distribution of current density in microscale would be achieved. Also, in both multiscale and RVE models, the exact location of CBD can be incorporated in the electrode structure to specify lithium diffusional path inside the group of particles in the solid matrix. Finally, in the fourth study, the vehicle battery driving cycle can be applied instead of galvanostatic operating condition, to mimic the stress generated inside the electrodes in real practical condition. \n

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.