Directional multiobjective optimization of metal complexes at the billion-system scale.

Area of Science:

• Computational chemistry
• Materials science
• Ligand design

Background:

• Discovering transition metal complexes (TMCs) with optimal properties necessitates extensive ligand libraries and advanced optimization algorithms.
• Existing methods often struggle with the vast chemical space and complex multiobjective optimization challenges.

Purpose of the Study:

• To introduce the tmQMg-L library, a large collection of diverse and synthesizable ligands with assigned properties.
• To develop and validate the Pareto-Lighthouse multiobjective genetic algorithm (PL-MOGA) for efficient TMC discovery.
• To demonstrate the algorithm's ability to optimize multiple properties simultaneously without predefined limits.

Main Methods:

• Generation of the tmQMg-L library comprising 30,000 ligands with characterized charges and coordination modes.
• Creation of 1.37 million palladium TMCs using the tmQMg-L library.
• Development and application of the PL-MOGA, employing whole-ligand mutation and crossover operations.
• Benchmarking the PL-MOGA against established optimization strategies.

Main Results:

• The tmQMg-L library facilitated the generation of a vast dataset of palladium TMCs.
• The PL-MOGA successfully maximized polarizability and the highest occupied molecular orbital-lowest unoccupied molecular orbital (HOMO-LUMO) gap of TMCs.
• The algorithm navigated complex chemical spaces, yielding thousands of diverse TMCs in an interpretable manner.
• PL-MOGA demonstrated efficient optimization without prior knowledge of objective limits.

Conclusions:

• The tmQMg-L library and PL-MOGA provide a powerful, integrated platform for accelerating the discovery of novel transition metal complexes.
• The PL-MOGA's whole-ligand operations offer a more interpretable and efficient approach to exploring large chemical spaces.
• This methodology enables fine control over property optimization, advancing the design of functional materials.