Automatic identification of relevant reactions
The discovery of chemical reactions is a workflow consisting of three steps:
- Reactive molecular dynamics based on the NanoReactor or Lattice Deformation
- Network Extraction using ChemTraYzer2: Reactive MD Analysis and geometry optimizations
- Product Ranking
The NanoReactor is a spherical region of space centered at the origin and surrounded by an elastic wall (represented by a harmonic potential), its size changing over time. The idea of the nanoreactor is to have two phases: reaction and relaxation. In the following, each nanoreactor phase is referred to as a reactor. During the relaxation phase, the reactor is large and if the temperature is high enough the reactants are distributed more or less evenly throughout its volume. In the reaction phase, the reactor shrinks and the atoms that now fall outside the reactor’s sphere accelerate towards its center. During this phase, high-energy collisions may lead to reactions. After the reaction phase is over, the relaxation phase begins again and the reactor is expanded. This cycle is repeated indefinitely.
ChemTraYzer2 is postprocessing tool that reads each frame of a reactive MD trajectory, keeps track of all reactions that occur, and summarizes all of this information into several useful quantities: a list of unique reactions, reaction rate constants for these reactions, the net fluxes of each species encountered in the simulation, and other kinetic and population measures.
During the final Ranking of products both a value for kinetic stability (cost) and thermodynamic stability are assigned. Ultimately, the molecules are orderd based in the first place on cost, and in the second place on (thermodynamic) stability.
