Data Files Accompanying: Uptake of N2O5 by Aqueous Aerosol Unveiled Using Chemically Accurate Many-Body Potentials
Authors: Cruzeiro, Vinícius Wilian D.; Galib, Mirza; Limmer, David T.; Götz, Andreas. W.
Journal: Nature Communications

Contact:
Andreas W. Goetz, agoetz@sdsc.edu, San Diego Supercomputer Center, University of California San Diego

Cite as:
Cruzeiro, V. W. D.; Galib, M.; Limmer, D. T.; Götz, A. W. (2022): Data from: Uptake of N2O5 by Aqueous Aerosol Unveiled Using Chemically Accurate Many-Body Potentials. In Center for Aerosol Impacts on Chemistry of the Environment (CAICE) Collection. UC San Diego Library Digital Collections.
DOI: https://doi.org/10.6075/J0FF3SHB

Data folder content:


# Sample input of molecular dynamics simulations

Contains sample AMBER input files for
- Umbrella Sampling simulations of N2O5 in a water slab
- Simulation of N2O5 in bulk water to determine the diffusion coefficient
- Scattering simulations

Energies and forces are set to be computed with the MBX software, that calls our MB-nrg model for N2O5 in water. Please refer to the AMBER manual for further instructions on how to compile AMBER interfaced with MBX.
Note that parameters in the prmtop file are not used to compute energy and forces during the simulations because these are computed by MBX.

The AMBER software and manual can be obtained from https://ambermd.org

The MBX software can be obtained from http://paesanigroup.ucsd.edu/software/mbx.html

The software has been compiled and simulations performed on Linux systems using CentOS versions 7 and 8.3, GNU compiler version 8.4.1, Intel compiler and Intel MKL library version 19.1.1.217 and OpenMPI version 4.1.1.
Software installation can be accomplished within a few hours.
Computational throughput of the molecular dynamics simulations is about 18 ps/day using 4 Intel Xeon E5-2670 CPU cores clocked at 2.6 GHz using OpenMP parallel MBX (setting environment variable OMP_NUM_THREADS=4). About 70 ps/day can be achieved using 16 CPU cores.

## Umbrella sampling

File description:
- mdin.h2o: Amber input file
- mbx.json: MBX configuration file
- disang  : Restraint file for umbrella sampling with Amber
- n2o5_h2o-slab.prmtop: Parameter and topology file
- n2o5_h2o-slab.ncrst : Amber restart file in netcdf format

The initial configuration in the restart file contains the N2O5 approximately placed at the Gibbs Diving Surface. The DISANG file has been set to constrain the N2O5 approximately at the same position.

## Diffusion constant from root-mean square displacement

File description:
- mdin.h2o: Amber input file
- mbx.json: MBX configuration file
- n2o5_bulk.prmtop: Parameter and topology file
- n2o5_bulk.ncrst : Amber restart file in netcdf format

The initial configuration in the restart file contains the N2O5 in bulk water.

## Scattering simulations

File description:
- mdin.h2o: Amber input file
- mbx.json: MBX configuration file
- n2o5_h2o-slab.prmtop: Parameter and topology file
- n2o5_h2o-slab.ncrst : Amber restart file in netcdf format

The initial configuration in the restart file contains the N2O5 placed in vapor with initial velocity pointing towards the water/vapor interface.


# Source data for the figures

The folders Fig1, Fig2 and Fig3 contain the source code used to generate each figure in the manuscript and the source data used.

## Source code to solve reaction diffusion model (Fokker-Planck equation)

Folder Fig3 contains the Python source code to solve the reaction-diffusion equation (FP_prop.py) and produce data for Figure 3 of the manuscript as described in the secion "Reactive uptake through interfacial and bulk hydrolysis" of the manuscript.

FP_prop.py has been tested on MacOS Version 11.6 (Big Sur) with Python 2.7.16. In addition to Python 2, it requires pylab.

The settings for the numerical solution of the reaction-diffusion equation are embedded within the Python source code. The source code is appropriately commented so users can easily edit settings if they wish to change simulation parameters.

By defalt the software will execute 15 million time steps for 10 different hydrolysis rate constants k. It takes about 15 minutes for each k to complete on a 2018 13-inch MacBook Pro laptop computer.

Results (time series of concentration profile and reactive uptake) are printed both to stdout and to files. Running the Python code without modifications will produce the output files that can be found in folder FP. The first line should be ignored. The output contains on each line in order
- time/ps
- hydrolysis concentration
- evaporation concentration
- reactive uptake gamma