Predictive in silico models for competing reaction mechanisms

by Johannes Kircher
Doctoral thesis
Date of Examination:2024-09-10
Date of issue:2025-09-08
Advisor:Prof. Dr. Ricardo Mata
Referee:Prof. Dr. Jonny Proppe
Referee:Prof. Dr. Manuel Alcarazo
Persistent Address: http://dx.doi.org/10.53846/goediss-11500

 

 

Files in this item

Name:Dissertation_Johannes_Kircher.pdf
Size:8.65Mb
Format:PDF

The following license files are associated with this item:

Abstract

English

Reactivity is one of the most fundamental concepts in chemistry, determining if a chemical transformation is taking place, its elementary steps and how fast it is. Due to the exponential dependence of the rate constant on the Gibbs free energy barrier, high accuracy values for these are needed to accurately predict the reactivity and reaction mechanisms. This work aims to shed light on the computational procedures for obtaining transition state structures as well as Gibbs free energy barriers. With this objective in mind, the Julia package STREAMS has been created as it incorporates functions to simplify the optimization and evaluation of transition states and reactants. It can be used to create scripts for automatic workflows. To gain insights into the accuracy of quantum chemically predicted rate constants, benchmarking with a reliable dataset was necessary. Mayr's database of reactivity parameters provides experimental information on reaction kinetics. Utilizing uncertainty quantification on the reactivity parameters, benchmarking under uncertainty could be realized. The evaluation of the different energy contributions to the Gibbs free energy were analyzed. Furthermore, the impact of conformational sampling of transition states has been evaluated. To obtain minimum energy transition states minimum energy paths are needed. One method for the optimization of pathways is the nudged elastic band method. The computational cost and its success are reliant on the quality of the starting pathway that is optimized to the final minimum energy pathway. As several interpolation methods for obtaining starting paths exist their relative performance regarding a dataset of reactions has been evaluated. Two cooperations with the group of Alcarazo are presented in which more complex systems have been studied computationally to explain experimental observations.
Keywords: Reactivity; Reaction mechanisms; Reaction modelling; Transition states; Transition state conformers
 
Statistik