Modelling ligand exchange in metal complexes with machine learning potentials

Gers Tusha,Veronika Juraskova,Hanwen Zhang,Lars Schäfer,Fernanda Duarte

DOI: https://doi.org/10.26434/chemrxiv-2024-6lvk7

2024-06-28

Abstract:Metal ions are irreplaceable in many areas of chemistry, including (bio)catalysis, self-assembly and charge transfer processes. Yet, modelling their structural and dynamic properties in diverse chemical environments remains challenging for both force fields and ab initio methods. Here, we introduce a strategy to train machine learning potentials (MLPs) using MACE, an equivariant message-passing neural network, for metal-ligand complexes in explicit solvents. We explore the structure and ligand exchange dynamics of Mg2+ in water and Pd2+ in acetonitrile as two illustrative model systems. The trained potentials accurately reproduce equilibrium structures of the complexes in solution, including different coordination numbers and geometries. Furthermore, the MLPs can model structural changes between metal ions and ligands in the first coordination shell, and reproduce the free energy barriers for the corresponding ligand exchange. The strategy presented here provides a computationally efficient approach to model metal ions in solution, paving the way for modelling larger and more diverse metal complexes relevant to biomolecules and supramolecular assemblies.

Chemistry

What problem does this paper attempt to address?

The main objective of this paper is to develop a strategy to model the ligand exchange process in metal complexes using machine learning potentials (MLPs). Specifically, the authors employed an equivariant message passing neural network—MACE (Message Passing Neural Network)—to train MLPs for specific metal ions (magnesium ion Mg\(^2+\) and palladium ion Pd\(^2+\)) in different solvent environments. For Mg\(^2+\), the study focused on its behavior in aqueous solution. By constructing a series of training datasets of varying sizes and compositions and employing an active learning (AL) strategy, the authors trained an MLP capable of accurately simulating the structural and dynamic properties of Mg\(^2+\) complexes in solution. This MLP not only reproduces the equilibrium structures of the complexes, including different coordination numbers and geometries, but also simulates the structural changes between the metal ion and ligands within the first coordination shell, and can reproduce the corresponding ligand exchange free energy barriers. For Pd\(^2+\), the research focused on its behavior in acetonitrile (MeCN) solvent. Similarly, by constructing training datasets containing different Pd\(^2+\) complex structures and using the AL strategy, the authors trained an MLP capable of accurately simulating the structural and kinetic properties of Pd\(^2+\) complexes in solution. This MLP accurately reproduces the equilibrium structures of Pd\(^2+\) complexes in solution and can simulate the structural changes between the metal ion and ligands within the first coordination shell, as well as the corresponding ligand exchange free energy barriers. In summary, this study aims to demonstrate that MLPs trained using MACE can efficiently and accurately simulate the ligand exchange process of metal ions in solution, paving the way for modeling larger and more diverse metal complexes in biomolecular and supramolecular assemblies.

Modelling ligand exchange in metal complexes with machine learning potentials

Predicting binding motifs of complex adsorbates using machine learning with a physics-inspired graph representation

Deciphering Cryptic Behavior in Bimetallic Transition Metal Complexes with Machine Learning

Towards a Generalizable Machine-Learned Potential for Metal-Organic Frameworks

Many-body Expansion Based Machine Learning Models for Octahedral Transition Metal Complexes

Machine Learning Potential for Electrochemical Interfaces with Hybrid Representation of Dielectric Response

Developing General Reactive Element-Based Machine Learning Potentials as the Main Computational Engine for Heterogeneous Catalysis

Machine Learning Interatomic Potentials for Catalysis

Prediction of stability constants of metal–ligand complexes by machine learning for the design of ligands with optimal metal ion selectivity

Exploiting Ligand Additivity for Transferable Machine Learning of Multireference Character Across Known Transition Metal Complex Ligands

Graph neural networks for predicting metal–ligand coordination of transition metal complexes

A Hessian-Based Assessment of Atomic Forces for Training Machine Learning Interatomic Potentials

metallicious: Automated force-field parametrization of covalently bound metals for supramolecular structures

Ligand Many-Body Expansion as a General Approach for Accelerating Transition Metal Complex Discovery

Modeling electronic response properties with an explicit-electron machine learning potential

Computational modelling of supramolecular metallopeptide assemblies

Quantum-Accurate Machine Learning Potentials for Metal-Organic Frameworks using Temperature Driven Active Learning

Generative Design of Functional Metal Complexes Utilizing the Internal Knowledge of Large Language Models

Construction of Machine Learning Interatomic Potentials for Metals

Parametrization of Non-Bonded Force Field Terms for Metal-Organic Frameworks Using Machine Learning Approach

Constructing and evaluating machine-learned interatomic potentials for Li-based disordered rocksalts