New Frontiers in Electrochemical Device Characterization Enabled by AI-driven X-Ray Microscope Reconstruction Technologies

A clean energy future will rely heavily on electrochemical energy storage and conversion devices like batteries and fuel cells to shift away from carbon-based energy sources. These electrochemical devices are generally closed off from the outside world and operate under highly non-ambient conditions (e.g., liquid electrolytes, air-free or controlled gas environments, elevated temperatures, liquid / gas flows). The performance of these devices is largely driven by the 3D microstructures of the internal components and their evolutions. For example, the weaves of the gas diffusion layer in a polymer electrolyte fuel cell control the gas flow to the catalyst and electrolyte layers; 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 in these devices 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.<br/><br/>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 also enable high resolution imaging inside relatively large objects – critical for in situ imaging of closed-form electrochemical devices like batteries and fuel cells. 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 X-ray microscopes, showing examples across multiple fields of electrochemical device research.<br/><br/>We show how AI-powered tomographic reconstruction can reveal new levels of detail in lithium-ion batteries and polymer-electrolyte fuel cells, and enable up to 10 times faster data collection at both the micrometer and nanometer length-scales. We additionally show how these technologies can be used to generate massive 3D images that allow researchers to evaluate, measure, and simulate the performance of these devices with the needed micrometer resolution across truly meaningful length scales of many millimeters of device volume. We additionally show how these technologies can be used to generate massive 3D images up to 100 times faster than conventional approaches allowing researchers to evaluate, measure, and simulate the performance of these devices with unprecedented precision and representivity. 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.