Energy Deposition in SN2 Reaction Products and Kinetic Energy Effects on Reactivity

TSNet: predicting transition state structures with tensor field networks and transfer learning

Transition states are among the most important molecular structures in chemistry, critical to a variety of fields such as reaction kinetics, catalyst design, and the study of protein function. However, transition states are very unstable, typically only existing on the order of femtoseconds. The transient nature of these structures makes them incredibly difficult to study, thus chemists often turn to simulation. Unfortunately, computer simulation of transition states is also challenging, as they are first-order saddle points on highly dimensional mathematical surfaces. Locating these points is resource intensive and unreliable, resulting in methods which can take very long to converge. Machine learning, a relatively novel class of algorithm, has led to radical changes in several fields of computation, including computer vision and natural language processing due to its aptitude for highly accurate function approximation. While machine learning has been widely adopted throughout computational chemistry as a lightweight alternative to costly quantum mechanical calculations, little research has been pursued which utilizes machine learning for transition state structure optimization. In this paper TSNet is presented, a new end-to-end Siamese message-passing neural network based on tensor field networks shown to be capable of predicting transition state geometries. Also presented is a small dataset of S_N2 reactions which includes transition state structures - the first of its kind built specifically for machine learning. Finally, transfer learning, a low data remedial technique, is explored to understand the viability of pretraining TSNet on widely available chemical data may provide better starting points during training, faster convergence, and lower loss values. Aspects of the new dataset and model shall be discussed in detail, along with motivations and general outlook on the future of machine learning-based transition state prediction.

Chemical dynamics simulations of X- + CH3Y → XCH3 + Y- gas-phase S(N)2 nucleophilic substitution reactions. Nonstatistical dynamics and nontraditional reaction mechanisms

Extensive classical chemical dynamics simulations of gas-phase X(-) + CH(3)Y → XCH(3) + Y(-) S(N)2 nucleophilic substitution reactions are reviewed and discussed and compared with experimental measurements and predictions of theoretical models. The primary emphasis is on reactions for which X and Y are halogen atoms. Both reactions with the traditional potential energy surface (PES), which include pre- and postreaction potential energy minima and a central barrier, and reactions with nontraditional PESs are considered. These S(N)2 reactions exhibit important nonstatistical atomic-level dynamics. The X(-) + CH(3)Y → X(-)---CH(3)Y association rate constant is less than the capture model as a result of inefficient energy transfer from X(-)+ CH(3)Y relative translation to CH(3)Y rotation and vibration. There is weak coupling between the low-frequency intermolecular modes of the X(-)---CH(3)Y complex and higher frequency CH(3)Y intramolecular modes, resulting in non-RRKM kinetics for X(-)---CH(3)Y unimolecular decomposition. Recrossings of the [X--CH(3)--Y](-) central barrier is important. As a result of the above dynamics, the relative translational energy and temperature dependencies of the S(N)2 rate constants are not accurately given by statistical theory. The nonstatistical dynamics results in nonstatistical partitioning of the available energy to XCH(3) +Y(-) reaction products. Besides the indirect, complex forming atomic-level mechanism for the S(N)2 reaction, direct mechanisms promoted by X(-) + CH(3)Y relative translational or CH(3)Y vibrational excitation are possible, e.g., the roundabout mechanism.

Post-transition state dynamics for propene ozonolysis: Intramolecular and unimolecular dynamics of molozonide

A direct chemical dynamics simulation, at the B3LYP6-31G(d) level of theory, was used to study the post-transition state intramolecular and unimolecular dynamics for the O3 + propene reaction. Comparisons of B3LYP6-31G(d) with CCSD(T)/cc-pVTZ and other levels of theory show that the former gives accurate structures and energies for the reaction's stationary points. The direct dynamics simulations are initiated at the anti and syn O3 + propene transition states (TSs) and the TS symmetries are preserved in forming the molozonide intermediates. Anti<-->syn molozonide isomerization has a very low barrier of 2-3 kcalmol and its Rice-Ramsperger-Kassel-Marcus (RRKM) lifetime is 0.3 ps. However, the trajectory isomerization is slower and it is unclear whether this anti<-->syn equilibration is complete when the trajectories are terminated at 1.6 ps. The syn (anti) molozonides dissociate to CH3CHO + H2COO and H2CO + syn (anti) CH3CHOO. The kinetics for the latter reactions are in overall good agreement with RRKM theory, but there is a symmetry preserving non-RRKM dynamical constraint for the former. Dissociation of anti molozonide to CH3CHO + H2COO is enhanced and suppressed, respectively, for the trajectory ensembles initiated at the anti and syn O3 + propene TSs. The dissociation of syn molozonide to CH3CHO + H2COO may also be enhanced for trajectories initiated at the syn O3 + propene TS. At the time the trajectories are terminated at 1.6 ps, the ratio of the trajectory and RRKM values of the CH3CHO + H2COO product yield is 1.6 if the symmetries of the initiation and dissociation TSs are the same and 0.6 if their symmetries are different. There are coherences in the intramolecular energy flow, which depend on molozonide's symmetry (i.e., anti or syn). This symmetry related dynamics is not completely understood, but it is clearly related to the non-RRKM dynamics for anti<-->syn isomerization and anti molozonide dissociation to CH3CHO + H2COO. Correlations are found between the stretching motions of molozonide, indicative of nonchaotic and non-RRKM dynamics. The non-RRKM dynamics of molozonide dissociation partitions vibration energy to H2COO that is larger than statistical partitioning. Though the direct dynamics simulations are classical, better agreement is obtained using quantum instead of classical harmonic RRKM theory. This may result from the neglect of anharmonicity in the RRKM calculations, the non-RRKM dynamics of the classical trajectories, or a combination of these two effects. The trajectories suggest that the equilibrium syn/anti molozonide ratio is approximately 1.1-1.2 times larger than that predicted by the harmonic densities of state, indicating an anharmonic correction.