AI-Powered 3D Image Reconstruction for Analyzing Lithium Ion Battery Microstructures at the Nanoscale

Stephen T Kelly,Yulia Trenikhina,William Harris
DOI: https://doi.org/10.1149/ma2024-012360mtgabs
2024-08-09
ECS Meeting Abstracts
Abstract:A clean energy future will rely heavily on lithium-ion batteries to shift away from carbon-based energy sources. These devices are closed off from the outside world and operate under highly non-ambient conditions (e.g., liquid electrolytes, air-free environments). Their performance is largely driven by the 3D microstructures of the internal components and their evolutions. For example, breached layers in batteries due to lithium dendrite growth can lead to catastrophic failures and safety concerns. As such, imaging the internal microstructures and processes is important for observing and measuring device construction details and understanding how defects are distributed and change performance during operation. In particular, 3D imaging allows for analysis of features and parameters that can only be fully understood in 3D, like porosity and tortuosity, and can also provide realistic microstructural input for 3D modelling and simulation approaches. Within this scope, X-ray microscopy provides a unique method to image electrochemical devices because of the non-destructive nature of the technique. Modern x-ray microscopes enable 3D imaging even at the nanoscale, bringing capabilities once reserved for the synchrotron into the laboratory. Advances in tomographic image reconstruction have allowed researchers to increasingly maximize the impact of x-ray microscopy data through higher image quality and enabling faster data acquisition schemes. Recently, artificial intelligence (AI) has been incorporated into these algorithms, dramatically increasing the performance and capability of the technique. Here, we apply multiple novel reconstruction technologies that leverage AI to dramatically improve the performance of laboratory-based nanoscale x-ray microscopes, showing examples of improvements in imaging lithium-ion battery electrode materials. We show how AI-powered tomographic reconstruction can reveal new levels of detail in lithium-ion battery materials and enable up to 10 times faster data collection across the micrometer and nanometer length-scales. These capabilities shift the paradigm in volume analysis for electrochemical devices and give researchers unprecedented insight into the detailed microstructures that drive device performance, enabling new levels of understanding and helping to power the drive to a clean energy future.
What problem does this paper attempt to address?