Applicability of XANES spectroscopy and machine learning methods for the determination of local atomic structure of Cu-MOR zeolites

The research is devoted to the development of methods of the determination of the local structure of copper centers in Cu-MOR using a combination of machine learning and X-ray absorption spectroscopy techniques. Cu-zeolites are promising catalysts for processes of environmentally friendly production of methanol from natural methane gas, the catalytic activity of which is mostly determined by the local environment of copper atoms in the zeolite. The irregular distribution of copper centers in the zeolite framework increases the complexity of the problem, since it makes difficult to interpret the experimental Cu $K$-XANES spectra. Machine learning algorithms trained on the synthetic data obtained in the FDMNES software package allowed us to determine the location of copper centers in a particular zeolite ring with an accuracy of 0.97 according to the F1 metric.