Accurate Thermochemistry and Kinetics of Molecules with Coupled Motions

 

Detailed reaction mechanisms are essential for modeling of energy and chemical engineering applications. This holds for fuel design applications as ingnition and combustion processes, but also for networks in polymerization. Often, the parameters of the fundamental chemical mechanisms are not experimentally accessible; nevertheless, they can be computed quantum-mechanically. Therefore one needs to model the electrons in the molecules on the one hand and the nuclei on the other.

  Copyright: © LTT

Nuclei move often in a complex coupled manner (cf. Figure of methanol dimer [1]). While methods describing the electrons are very accurate, modeling of the nuclear degrees of freedom turn out to be the bottleneck for the overall accuracy. Widely used models for nuclear motion (like the rigid-rotor harmonic-oscillator approximation) fail to describe coupled motions. Goal of this project is therefore a highly accurate computation of such coupled (and anharmonic) motions at the quantum-mechanical level. By using internal coordinates and some properties of the Jacobian [2], we develop a software that not only computes small systems with benchmark accuracy but also contains approximations for larger systems of technical interest, like biofuels [3] or solvents [4].

One such software is our Configuration Integral Monte Carlo Integration (CIMCI) algorithm [5]. Rather than following the traditional approach of calculating an approximate model for nuclear motion from very few, high-accuracy energy calculations, CIMCI performs an exact numerical integration with a coarse-but-fast electronic model, like a force field, tight-binding method, or machine learning potential. Fully coupled handling of all nuclear degrees of freedom in this way produces substantial accuracy gains for some systems, even when force field-based CIMCI results are compared against coupled cluster-based results from other methods.

 

Another software developed by our group, TAMkinTools, focuses on molecular torsions and is an extension of the python package TAMkin. Using our code, the energy curve of torsions can be modeled more accurately and numerically stable, what we proved for the solvents mentioned above [6]. Currently, we investigate the use of machine-learning methods for further improvements of the potentials.

 

Relevante Publications of our group

[1] Muhammad Umer and Kai Leonhard, Ab Initio Calculations of Thermochemical Properties of Methanol Clusters, The Journal of Physical Chemistry A, 2013, volume 117, issue 7, pages 1569-1582.

[2] Wassja A. Kopp and Kai Leonhard, General formulation of rovibrational kinetic energy operators and matrix elements in internal bond-angle coordinates using factorized Jacobians, The Journal of Chemical Physics, 2016, volume 145, issue 23, page 234102.

[3] Leif C. Kröger, Malte Döntgen, Dzmitry Firaha, Wassja A. Kopp and Kai Leonhard, Ab initio kinetics predictions for H-atom abstraction from diethoxymethane by hydrogen, methyl, and ethyl radicals and the subsequent unimolecular reactions, Proceedings of the Combustion Institute, 2019, volume 37, issue 1, pages 275-282.

[4] Hannes C. Gottschalk et al., The furan microsolvation blind challenge for quantum chemical methods: First steps, The Journal of Chemical Physics, 2018, volume 148, issue 1, page 014301.

[5] Gabriel Rath, Wassja A. Kopp and Kai Leonhard, Coupled Anharmonic Thermochemistry from Stratified Monte Carlo Integration, Journal of Chemical Information and Modeling, 2021, volume 61, issue 12, pages 5853-5870.

[6] Wassja A. Kopp, Matthias L. Mödden, Narasimhan Viswanathan, Gabriel Rath, and Kai Leonhard, Improved modeling of anharmonicity for furan microsolvation, Physical Chemistry Chemical Physics, 2023, volume 25, issue 16, pages 11316-11323.