Kevin Rossi^1*, Veronika Juraskova^2*, Raphael Wischert³, Laurent Garel⁴, Clemence Corminboeuf^2*, Michele Ceriotti^1*

1 Laboratory of Computational Science and Modeling (COSMO), Institute of Materials, École Polytechnique Fédérale de Lausanne (EPFL), Lausanne, 1015, Switzerland

2 Laboratory for Computational Molecular Design (LCMD), Institute of Chemical Sciences and Engineering, École Polytechnique Fédérale de Lausanne (EPFL), Lausanne, 1015, Switzerland

3 Eco-Efficient Products and Processes Laboratory, Solvay, RIC Shanghai, China

4 Aroma Performance Laboratory, Solvay, RIC Lyon, France

* Corresponding authors emails: kevin.rossi@epfl.ch, veronika.juraskova@epfl.ch, clemence.corminboeuf@epfl.ch, michele.ceriotti@epfl.ch

DOI10.24435/materialscloud:2x-7x [version v2]

Publication date: Mar 26, 2021

How to cite this record

Kevin Rossi, Veronika Juraskova, Raphael Wischert, Laurent Garel, Clemence Corminboeuf, Michele Ceriotti, Simulating solvation and acidity in complex mixtures with first-principles accuracy: the case of CH₃SO₃H and H₂O₂ in phenol, Materials Cloud Archive 2021.50 (2021), https://doi.org/10.24435/materialscloud:2x-7x

Description

Set of inputs to perform the calculations reported in the paper. The i-pi input enables to perform molecular dynamics / metadynamics / REMD / PIMD simulations, with adequate thermostats. The DFTB and LAMMPS input respectively enable to calculate force and energies within the DFTB and Neural Network Forcefield frameworks. The CP2K input files enable to calculate force and energies at PBE and PBE0 level. The latter is used as the reference to train the neural network correction on top of DFTB. Brief description of the work: We present a generally-applicable computational framework for the efficient and accurate characterization of molecular structural patterns and acid properties in explicit solvent using H₂O₂ and CH₃SO₃H in phenol as an example. In order to address the challenges posed by the complexity of the problem, we resort to a set of data-driven methods and enhanced sampling algorithms. The synergistic application of these techniques makes the first-principle estimation of the chemical properties feasible without renouncing to the use of explicit solvation, involving extensive statistical sampling. Ensembles of neural network potentials are trained on a set of configurations carefully selected out of preliminary simulations performed at a low-cost density-functional tight-binding (DFTB) level. Energy and forces of these configurations are then recomputed at the hybrid density functional theory (DFT) level and used to train the neural networks. The stability of the NN model is enhanced by using DFTB energetics as a baseline, but the efficiency of the direct NN (i.e., baseline-free) is exploited via a multiple-time step integrator. The neural network potentials are combined with enhanced sampling techniques, such as replica exchange and metadynamics, and used to characterize the relevant protonated species and dominant non-covalent interactions in the mixture, also considering nuclear quantum effects.

Materials Cloud sections using this data

Files

File name	Size	Description
input.xml MD5md5:012f438cd8a0d09e1ae32cd583c7a09a	3.7 KiB	i-pi input to run basic MD (https://github.com/cosmo-epfl/i-pi)
pbe.cp2k MD5md5:08e632a33d5afb551b9d5f704ceced55	1.5 KiB	example PBE input for DFT calculations (https://www.cp2k.org/)
pbe0.cp2k MD5md5:9a47c035c868b0ccb78729729345949c	3.0 KiB	exemple PBE0 input for DFT calculations (https://www.cp2k.org/)
dftb_in.hsd MD5md5:8e253196041e84221c7cde407c6c5656	1.2 KiB	example DFTB+ input for DFTB calculations (https://www.dftbplus.org/)
input.nn MD5md5:e7acab8f9218b083a66d28a0bcda31a1	22.6 KiB	example input.nn input to train and use a neural network for force and energy predictions (https://github.com/CompPhysVienna/n2p2)
lmp1.in MD5md5:ba8eae7e733c2cd14244540b26535f8e	3.5 KiB	example LAMMPS input for MD calculations via i-pi and using neural network potentials (https://lammps.sandia.gov/)
input.data.direct.gz MD5md5:36254b945cfaf68c226e49644ce5e96f	34.3 MiB	Dataset for learning DFT energies and forces (input.data format as used in N2P2)
input.data.delta.gz MD5md5:d6b5b6ed0368757315593302a3df17ff	35.3 MiB	Dataset for learning DFTB-baselined energies and forces (input.data format as used in N2P2)
direct.tar.gz MD5md5:8475cf5afa4b6543e57b845c96882a14	143.2 KiB	nn weights for direct predictions of forces and energies
delta.tar.gz MD5md5:68cde22ad92072c0452c7941ce792bb0	713.5 KiB	nn weights for DFTB-baselined predictions of forces and energies
README.txt MD5md5:9dc6a1ab9a65080329ee332892c0460f	375 Bytes	README

License

Files and data are licensed under the terms of the following license: Creative Commons Attribution 4.0 International.
Metadata, except for email addresses, are licensed under the Creative Commons Attribution Share-Alike 4.0 International license.

External references

Keywords

machine learning solution chemistry acid homogeneous catalysis catalysis acid artificial intelligence reaction CH3SO3H H2O2 MARVEL