Automated Transition State Theory Calculations for High-Throughput Kinetics.

A scarcity of known chemical kinetic parameters leads to the use of many reaction rate estimates, which are not always sufficiently accurate, in the construction of detailed kinetic models. To reduce the reliance on these estimates and improve the accuracy of predictive kinetic models, we have developed a high-throughput, fully automated, reaction rate calculation method, AutoTST. The algorithm integrates automated saddle-point geometry search methods and a canonical transition state theory kinetics calculator. The automatically calculated reaction rates compare favorably to existing estimated rates. Comparison against high level theoretical calculations show the new automated method performs better than rate estimates when the estimate is made by a poor analogy. The method will improve by accounting for internal rotor contributions and by improving methods to determine molecular symmetry.

Transition state geometry prediction using molecular group contributions.

Detailed kinetic models to aid the understanding of complex chemical systems require many thousands of reaction rate coefficients, most of which are estimated, some quite approximately and with unknown uncertainties. This motivates the development of high-throughput methods to determine rate coefficients via transition state theory calculations, which requires the automatic prediction of transition state (TS) geometries. We demonstrate a novel approach to predict TS geometries using a group-additive method. Distances between reactive atoms at the TS are estimated using molecular group values, with the 3D geometry of the TS being constructed by distance geometry. The estimate is then optimized using electronic structure theory and validated using intrinsic reaction coordinate calculations, completing the fully automatic algorithm to locate TS geometries. The methods were tested using a diisopropyl ketone combustion model containing 1393 hydrogen abstraction reactions, of which transition states were found for 907 over two iterations of the algorithm. With sufficient training data, molecular group contributions were shown to successfully predict the reaction center distances of transition states with root-mean-squared errors of only 0.04 Å.