The microstructure variation of materials under extreme conditions have attracted great attention in many fields. The average grain size and grain distribution are the key microstructure characteristics that affect the physical and chemical properties[1], representing the fundamental qualities of polycrystalline materials, therefore are of vital importance in predicting material responses, evaluating the kinetic phase transition process, and having an insight into the physical connotations. As a microstructure characterization technology, small angle X-ray scattering (SAXS) has been widely used in materials science. During the SAXS analysis process and the microstructure inversion, it is usually necessary to establish a sample model first, run simulations of the scattering process and compare them with the experimentally observed pattern, and then repeat the process several times as well as adjust several parameters until this model is closest to the experimental result. Running simulations for each optimization step is time-consuming and requires considerable computing power. With the emergence of the fourth generation X-ray free electron lasers, it is possible to quickly image and obtain in situ response of materials. Large quantities of data also pose a demand for rapid inversion of SAXS.
Here, we have discussed how traditional analytical and Monte Carlo method can be applied to extreme dynamic high-pressure materials. Moreover, a novel SAXS data inversion method based on machine learning was also proposed. We have discussed and evaluated the effectiveness of applying the above methods to extreme dynamic high-pressure phase transition experiments[2]. In addition, we also have analyzed the feasibility of conducting dynamic in-situ SAXS diagnosis on line stations based on high-energy light source and large high-power laser devices.

SAXS, high pressure, shock compressed, in situ diagnostic