Скачать или смотреть The analysis of Ru K-edge XANES with using machine learning approaches

The project aims at the rational design of ruthenium-based gamogenic catalytic systems. This system can be used to hydrogenate sugar alcohols to alkenes, which offers an efficient pathway for an environmentally friendly and sustainable chemical process.
The system of the RuX3 salt (X = Cl, Br) dissolved in the ionic liquid tetrabutylphosphonium bromide (Bu4PBr) are proposed as a catalyst for such type of reactions. The idea of the work is to trace the evolution of the catalytic system using the XAS method supplemented by machine learning and DFT calculations.
In homogeneous catalysis, obtaining quantitative information from XANES data using machine learning methods represents a promising alternative or addition to the classic EXAFS analysis.
Structures of the Ru(m)Cl(x)Br(y)CO(z) type were taken as the initial ones, divided into two large groups according to the number of ruthenium centers: m = 1 and m = 2. In all structures, each ruthenium atom was in an octahedral environment with a total number of ligands x + y + z = 6. In total, more than 10,000 different structures were generated. For each of these structures XANES spectra were calculated. The calculated theoretical spectra formed several databases used for training machine learning algorithms. The work was carried out using the original program codes written in Python, using the PyFitIt library, developed by the authors of this command1. We apply an inverse approach to predict interatomic distance from Ru sites to CO, Br, and Cl ligands from XANES spectra and compare results with analysis EXAFS and DFT calculations.
These methodological advances have been successfully used to establish important structural patterns for ruthenium-based catalytic systems and have enabled us to predict both geometry and ligand surrounding of ruthenium-based catalysts using machine learning.