A neural network potential energy surface for the F + CH 4 reaction including multiple channels based on coupled cluster theory.

We report here a new global and full dimensional potential energy surface (PES) for the F + CH4 reaction. This PES was constructed by using neural networks (NN) fitting to about 99 000 ab initio energies computed at the UCCSD(T)-F12a/aug-cc-pVTZ level of theory, and the correction terms considering the influence of a larger basis set as well as spin-orbit couplings were further implemented with a hierarchial scheme. This PES, covering both the abstraction and substitution channels, has an overall fitting error of 8.24 meV in total, and 4.87 meV for energies within 2.5 eV using a segmented NN fitting method, and is more accurate than the previous PESs.