The thermal decomposition of CL-20 (C6H6N12O12)#

This tutorial is proposed to develop a Deep Potential for the thermal decomposition of CL-20, perform MD simulations, and then use the ReacNetGenerator to analyze reactions.

This tutorial introduces:


Liqun Cao, Jinzhe Zeng, Bo Wang, Tong Zhu, and John Z H Zhang. Ab initio neural network MD simulation of thermal decomposition of a high energy material CL-20/TNT. Phys. Chem. Chem. Phys., 24:11801–11811, 2022. doi:10.1039/D2CP00710J.

DP-GEN workflow#

In this study, a workflow was developed for making reference datasets with DP-GEN program. The details of each module in the workflow are given in Cao et al. [CZW+22].

You can install DP-GEN program with the tutorial in DP-GEN.[ZWC+20] In this example, the input files for DP-GEN program are provided here.

During the DP-GEN workflow, each iteration includes three steps: QM calculation, training of the NN PES, and reactive molecular dynamic simulations.

QM calculation#

CP2K package was used for DFT calculation, which can be obtained at CP2k.[KuhneIDelBen+20] PBE-D3 method was employed in this example, because it is generally used for DFT calculation of energetic materials.

Training of the NN PES#

The DeepPot-SE (Deep Potential-Smooth Edition) model was used to train the NN potential by the DeePMD-kit program. You can install DeePMD-kit program with the tutorial in DeePMD-kit.[WZHE18]

Reactive molecular dynamic simulations#

After training, the well-behaved NN PESs can be used for running molecular dynamics simulation by LAMMPS interfaced with DeePMD-kit program.[TAB+22]

Analysis by ReacNetGenerator#

DP-GEN software will produce a series of reactive simulation trajectory. Here we take the final trajectory. The trajectory of a single frame can be integrated into the entire trajectory, and the format of trajectory here is dump. The reaction mechanism can be obtained by the following command:

reacnetgenerator -i cl20.lammpstrj -a H C N O --type dump --nohmm

The results can be viewed at this page.