Rare-earth elements are widely used in all kinds of gadgets nowadays. But similarity of their chemistry and natural occurrence force us to develop new effective methods of their separation. Experimental search of new effective and selective ligands requires significant material and human resources, so computational methods are commonly used for estimating potential ligand efficiency. And a question of complexation of rare-earth elements is still the most important in terms of ligands design. Here we would like to introduce new hybrid approach for in silico ligand design based on combination of accurate quantum chemistry (QC) and novel quantitative structure-property relationship (QSPR) methods. QC methods require a lot of computational resources and usage of supercomputer equipment, while QSPR ones can be used on personal computer. On the other hand, lack of experimental data, doesn’t allow us to train accurate model for QSPR. Appropriate combination of the approaches allow us to avoid their weak sides. Proposed method is based on experimental data as well as QC calculations and can be easily scaled to other chemical complexation problems.

Deep learning approach to computational chemistry of lanthanides

255th National Meeting (2018)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.