ReaLigands: A Ligand Library Cultivated from Experiment and Intended for Molecular Computational Catalyst Design

Area of Science:

• Computational Chemistry
• Materials Science
• Catalysis

Background:

• Computational catalyst design necessitates identifying optimal metal-ligand combinations for desired reactivity and selectivity.
• A significant barrier exists in evaluating the synthesizability of computationally proposed ligands for experimental validation.

Purpose of the Study:

• To address the challenge of ligand synthesizability in computational catalyst design.
• To introduce the ReaLigands library, a comprehensive database of experimentally validated ligands.

Main Methods:

• Developed the ReaLigands library by extracting >30,000 ligands from crystal structures using a modified depth-first search algorithm.
• Assigned charges to ligands using a machine learning model trained on quantum-chemical features.
• Employed principal component analysis (PCA) and uniform manifold approximation and projection (UMAP) for ligand analysis and categorization.

Main Results:

• The ReaLigands library contains diverse monodentate, bidentate, tridentate, and larger ligands, sorted by connectivity and denticity.
• PCA and UMAP analyses revealed the structural diversity and interconnections within tridentate ligand categories.
• Demonstrated library utility by rapidly generating reaction barriers for 750 bidentate ligands in a Rh-hydride ethylene migratory insertion reaction.

Conclusions:

• The ReaLigands library significantly facilitates the translation of computational catalyst designs into experimental practice.
• This resource accelerates the discovery and optimization of novel catalysts by providing a curated set of synthesizable ligands.