X-ray microscopy provides a unique method to image samples non-destructively in 3D across a wide range of materials and life sciences. 3D imaging allows for analysis of features and parameters that can only be fully understood in 3D like porosity and tortuosity in porous materials and additively manufactured parts or network connection maps in neuroscience and can also provide realistic microstructural input for 3D modelling and simulation approaches.

Modern X-ray microscopes also enable high resolution imaging inside relatively large objects – critical for in situ imaging within special environments, or for enclosed 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 materials research including lithium-ion batteries and additive manufacturing. We show how AI-powered tomographic reconstruction can reveal new levels of detail in functioning lithium-ion batteries 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 in super-resolution applications combining high- and low-resolution image volumes to recover the high-resolution information across the expanded field of view of the low-resolution image volume, enabling quantitative porosity mapping in additively manufactured structures. Methods such as this can provide detailed information for computational modelling at relevant, statistically representative length scales across meaningful fields of view that capture device- or part-level properties. These capabilities shift the paradigm in volume analysis within materials research and give researchers unprecedented insight into the detailed 3D microstructures that drive processes across these research topics.