Computational Discovery of Transition-metal Complexes: From High-throughput Screening to Machine Learning

Abstract

Transition-metal complexes are attractive targets for the design of catalysts and functional materials. The behavior of the metal−organic bond, while very tunable for achieving target properties, is challenging to predict and necessitates searching a wide and complex space to identify needles in haystacks for target applications. This review will focus on the techniques that make high-throughput search of transition-metal chemical space feasible for the discovery of complexes with desirable properties. The review will cover the development, promise, and limitations of "traditional" computational chemistry (i.e., force field, semiempirical, and density functional theory methods) as it pertains to data generation for inorganic molecular discovery. The review will also discuss the opportunities and limitations in leveraging experimental data sources. We will focus on how advances in statistical modeling, artificial intelligence, multiobjective optimization, and automation accelerate discovery of lead compounds and design rules. The overall objective of this review is to showcase how bringing together advances from diverse areas of computational chemistry and computer science have enabled the rapid uncovering of structure−property relationships in transition-metal chemistry. We aim to highlight how unique considerations in motifs of metal−organic bonding (e.g., variable spin and oxidation state, and bonding strength/nature) set them and their discovery apart from more commonly considered organic molecules. We will also highlight how uncertainty and relative data scarcity in transition-metal chemistry motivate specific developments in machine learning representations, model training, and in computational chemistry. Finally, we will conclude with an outlook of areas of opportunity for the accelerated discovery of transition-metal complexes.

Spin-Phonon Relaxation in Magnetic Molecules: Theory, Predictions and Insights

“…After the training stage, ML models can be used to make predictions at very little computational cost, therefore offering a significant speed-up of these simulations. Further adaptation of machine learning schemes to this tasks is needed, but encouraging proof-of-concept applications have recently been presented [56,57,83,85,121,122]. Another interesting strategy involve the use of parametrized Hamiltonians to compute the spin-phonon coupling coefficients.…”

Unlocking the computational design of metal–organic cages

“…There have been recent advances in exploring transition-metal chemical space by developing cheminformatic-inspired approaches. 75,76 Overall, recent literature, including our own work, has shown significant progress in user-friendly, high-throughput approaches for tackling metalcontaining systems.…”

“…When evaluating approaches to modelling metal-containing systems, the diversity in electronic configuration, spin state, oxidation state and bonding/geometry presents issues in finding a ''bestpractice'' or universal solution. 50,75 It is possible to focus on a single ''well behaving'' metal and on exploring the ligand chemical space. However, this approach is fundamentally limiting.…”

Autonomous Reaction Network Exploration in Homogeneous and Heterogeneous Catalysis

“…Many optimization and design strategies for more stable or active catalysts have been developed for specific fields such as biocatalysis [ 261 – 270 ], homogeneous catalysis [ 271 – 281 ], or heterogeneous catalysis [ 282 – 288 ]. In these strategies, the activity of a catalyst is judged on various physical descriptors.…”