Table of Contents
Fetching ...

Automated Microsolvation for Minimum Energy Path Construction in Solution

Paul L. Türtscher, Markus Reiher

TL;DR

This work introduces Kingfisher, an automated pipeline that identifies active solvent participation in reactions in solution and constructs minimal microsolvation models by integrating QM/MM with continuum solvation. Activation barriers and free energies are assessed using a Microsolvated Free Energy of Activation (MiFEA) framework that adds cavity entropy corrections to a gas-phase-like thermochemical model, enabling rapid, high-throughput comparisons with experiment. The approach is demonstrated on formaldehyde hydrolysis to methanediol, chlorination of phenol, and CO2 hydration, showing that explicit solvent participation generally lowers activation barriers relative to continuum-only models and that cavity entropy corrections improve entropy estimates. This framework supports efficient exploration of reaction networks in solution and can handle solvent mixtures, providing a practical balance between accuracy and computational cost for large-scale mechanistic studies.

Abstract

Describing chemical reactions in solution on a molecular level is a challenging task due to the high mobility of weakly interacting solvent molecules which requires configurational sampling. For instance, polar and protic solvents can interact strongly with solutes and may interfere in reactions. However, to define and identify representative arrangements of solvent molecules modulating a transition state is a non-trivial task. Here, we propose to monitor their active participation in the decaying normal mode at a transition state, which defines active solvent molecules. Moreover, it is desirable to prepare a low-dimensional microsolvation model in a well-defined, fully automated, high-throughput, and easy-to-deploy fashion, which we propose to derive in a stepwise protocol. First, transition state structures are optimized in a sufficiently solvated quantum-classical hybrid model, which are then subjected to a re-definition of a then reduced quantum region. From the reduced model, minimally microsolvated structures are extracted that contain only active solvent molecules. Modeling the remaining solvation effects is deferred to a continuum model. To establish an easy-to-use free-energy model, we combine the standard thermochemical gas-phase model with a correction for the cavity entropy in solution. We assess our microsolvation and free-energy models for methanediol formation from formaldehyde, for the hydration of carbon dioxide (which we consider in a solvent mixture to demonstrate the versatility of our approach), and, finally, for the chlorination of phenol with hypochlorous acid.

Automated Microsolvation for Minimum Energy Path Construction in Solution

TL;DR

This work introduces Kingfisher, an automated pipeline that identifies active solvent participation in reactions in solution and constructs minimal microsolvation models by integrating QM/MM with continuum solvation. Activation barriers and free energies are assessed using a Microsolvated Free Energy of Activation (MiFEA) framework that adds cavity entropy corrections to a gas-phase-like thermochemical model, enabling rapid, high-throughput comparisons with experiment. The approach is demonstrated on formaldehyde hydrolysis to methanediol, chlorination of phenol, and CO2 hydration, showing that explicit solvent participation generally lowers activation barriers relative to continuum-only models and that cavity entropy corrections improve entropy estimates. This framework supports efficient exploration of reaction networks in solution and can handle solvent mixtures, providing a practical balance between accuracy and computational cost for large-scale mechanistic studies.

Abstract

Describing chemical reactions in solution on a molecular level is a challenging task due to the high mobility of weakly interacting solvent molecules which requires configurational sampling. For instance, polar and protic solvents can interact strongly with solutes and may interfere in reactions. However, to define and identify representative arrangements of solvent molecules modulating a transition state is a non-trivial task. Here, we propose to monitor their active participation in the decaying normal mode at a transition state, which defines active solvent molecules. Moreover, it is desirable to prepare a low-dimensional microsolvation model in a well-defined, fully automated, high-throughput, and easy-to-deploy fashion, which we propose to derive in a stepwise protocol. First, transition state structures are optimized in a sufficiently solvated quantum-classical hybrid model, which are then subjected to a re-definition of a then reduced quantum region. From the reduced model, minimally microsolvated structures are extracted that contain only active solvent molecules. Modeling the remaining solvation effects is deferred to a continuum model. To establish an easy-to-use free-energy model, we combine the standard thermochemical gas-phase model with a correction for the cavity entropy in solution. We assess our microsolvation and free-energy models for methanediol formation from formaldehyde, for the hydration of carbon dioxide (which we consider in a solvent mixture to demonstrate the versatility of our approach), and, finally, for the chlorination of phenol with hypochlorous acid.

Paper Structure

This paper contains 17 sections, 25 equations, 17 figures.

Figures (17)

  • Figure 1: A reactive complex assembled of one phenol and one hypochlorous acid molecule. Carbon atoms are depicted in gray, hydrogen atoms in white, oxygen atoms in red, and the chlorine atom in green. Reactive atoms are highlighted in blue, active atoms in gold.
  • Figure 2: Schematic overview of the three steps of the Kingfisher pipeline. Structures in Step 1 are a partially optimized reactive complex and a TS guess. Step 2 and Step 3 contain optimized stationary points along the minimum energy paths. Atoms modeled by a QM model are represented with spheres, water molecules modeled by a MM model are represented as light blue sticks. Carbon atoms are depicted in gray, hydrogen atoms in white, oxygen atoms in red, and the chlorine atom in green.
  • Figure 3: a) Reactive complex b) Atoms of the lQM region. c) Full lQM/MM system before the elementary step search. Atoms within the QM region are represented as spheres, water molecules modeled by a MM model are represented as light blue sticks. Carbon atoms are depicted in gray, hydrogen atoms in white, oxygen atoms in red, and the chlorine atom in green.
  • Figure 4: a) Transition state guess of the full lQM/MM system. b) Transition state guess of the lQM region. c) Extracted mQM region with three water molecules. Atoms within the QM region are represented as spheres, water molecules modeled by a MM model are represented as light blue sticks Carbon atoms are depicted in gray, hydrogen atoms in white, oxygen atoms in red, and the chlorine atom in green.
  • Figure 5: a) Optimized TS of the full mQM/MM system with three water molecules in the mQM region. s) Optimized TS of the sQM region with two water molecules in the sQM region, schematically embedded in the cavity constructed by the continuum solvation model. Atoms within the QM region are represented as spheres, water molecules modeled by a MM model are represented as light blue sticks Carbon atoms are depicted in gray, hydrogen atoms in white, oxygen atoms in red, and the chlorine atom in green.
  • ...and 12 more figures