ABSTRACT The three‐dimensional x‐ray absorption near edge structure (3D XANES) imaging technique can analyze the 3D morphology and chemistry at the nanoscale. High‐quality conventional 3D XANES imaging requires sufficient full‐view projections across every near‐edge energy point, prolonging the imaging time. This limits the method's application in dynamic process monitoring (e.g., in situ electrochemical reactions and phase transitions). This study presents a novel workflow to accelerate 3D XANES imaging based on sparse sampling and deep learning. Sparse‐view sampling strategies can significantly reduce acquisition time, while the STC‐UNet deep learning algorithm can correct sampling artifacts and preserve precision in quantitative elemental chemical analysis. The proposed approach was applied to LiNi 0.8 Co 0.1 Mn 0.1 O 2 (NCM811) battery cathode particles to analyze 3D chemical state mapping of nickel. The results demonstrate a tenfold reduction in data acquisition time compared to conventional protocols while maintaining equivalent image quality and quantitative analytical accuracy, thereby enabling potential for operando 3D chemical imaging.
Zhang et al. (Sun,) studied this question.