** Reference to paper:

https://doi.org/10.1021/acs.jpclett.9b00560

** DOI:

10.1021/acs.jpclett.9b00560

** Title:

Accurate Probabilities for Highly Activated Reaction of Polyatomic Molecules on Surfaces Using a High-Dimensional Neural Network Potential: CHD3 + Cu(111)

** Authors:

Gerrits, Nick; Shakouri, Khosrow; Behler, Joerg; Kroes, Geert-Jan

** Contact e-mail:

n.gerrits@lic.leidenuniv.nl
g.j.kroes@chem.leidenuniv.nl

** Abstract:

An accurate description of reactive scattering of molecules on metal surfaces often requires the modeling of energy transfer between the molecule and the surface phonons.
Although ab initio molecular dynamics (AIMD) can describe this energy transfer, AIMD is at present untractable for reactions with reaction probabilities smaller than 1%.
Here, we show that it is possible to use a neural network potential to describe a polyatomic molecule reacting on a mobile metal surface with considerably reduced computational effort compared to AIMD.
The highly activated reaction of CHD3 on Cu(111) is used as a test case for this method.
It is observed that the reaction probability is influenced considerably by dynamical effects such as the bobsled effect and surface recoil.
A special dynamical effect for CHD3 + Cu(111) is that a higher vibrational efficacy is obtained for two quanta in the CH stretch mode than for a single quantum.

** Description per file:

The program used for DFT is VASP v5.3.5 with a special modification in order to use a SRP functional and with VTST (http://theory.cm.utexas.edu/vtsttools/index.html).
The RuNNer code is used for the training of the HD-NNP, which is available by emailing Joerg Behler (joerg.behler@uni-goettingen.de).
Molecular dynamics are performed with LAMMPS (16 March, 2018 version), as has been described in the article.

Most files are missing as they are already provided in 2018_jcp_CHD3_cu_alloys.

All figures are made with matplotlib 1.5.1

** Folder Figure01:
Several figures were added together and annoted in GIMP.
angles.png - made in Blender 1.79
elbow_2d.pdf - elbow where nothing is relaxed and only Z and r are varied
elbow_15d.pdf - elbow where all degrees of freedom other than Z and r are relaxed. This version is drawn slightly differently (the underlying data is the same), but can't find the other one right now and it is not important. For making a elbow plot, see 2018_jcp_CHD3_cu_alloys.
coupling.pdf - electronic and mechanical coupling
plotcoupling.py - generates coupling.pdf. Note that a mistake in the original script caused a wrong surface, resulting in a different coupling. Here we corrected this. Other figures and calculations are not affected.
elbow.pdf - combined figure

** Folder Figure02:
errorfraction.pdf
errorfraction_plot.py - generates the plot
error_test.dat - contains the error of the HD-NNP vs DFT of the test set
error_train.dat - contains the error of the HD-NNP vs DFT of the training set

** Folder Figure03:
reactionprobabilitymulti.pdf
plotstickingmulti.py - generates the plot

** Folder Figure04:
elbow_zc_traj.pdf
97v2_014829.dat 143v1_014889.dat 181v0_112348.dat - data files for the trajectories
draw_zc_elbow.py - generates the plot
energy_NN_constraint.dat - energies of the elbow

** Folder FigureS01:
correlation.pdf

** Folder FigureS02:
beamparameters.pdf
plotbeamparameter.py - generates the plot

** Folder FigureS03:
beam_velocity.pdf
see FigureS02/plotbeamparameter.py for script

** Folder FigureS04:
zc_incidence.pdf - data is taken from the PostAnalysis.dat of all reacted trajectories

** Folder FigureS05:
See FigureS06/energy_NN_constraint.dat for data and 2018_jcp_CHD3_cu_alloys for making the plot.

** Folder FigureS06:
elbow_ni.pdf
draw_elbow_3.py - generates figure
energy_ni.dat energy_ni_mep.dat energy_NN_constraint.dat energy_nn_mep.dat - data files
phi_scan.py - MEP finder

** Folder FigureS07:
mep_ni.pdf
draw_mep.py - generates plot
energy_ni_mep.dat energy_nn_mep.dat - data files

Other files:
check-CHD3-NN.py - analysis the trajectory and stores the relevant info in PostAnalysis.dat