Data

We make the data behind our published work openly available. Simulation papers from the group are accompanied by permanently archived, citable deposits on Zenodo containing the initial configurations, force field parameters, and analysis output needed to reproduce our results. Each deposit carries a DOI and is released under a CC BY 4.0 license.

Nearly all of these datasets were generated with OpenMD, our group’s open source molecular dynamics engine.

What’s in an OpenMD Archive

The datasets above share a common set of file types. Trajectory (.dump) and status (.stat) files are generally omitted because of their size — both can be regenerated from the included configuration files.

• .omd — combined metadata and configuration file used to start a simulation
• .frc — force field parameter file
• .eor — end-of-run final configuration, in the same format as .omd
• .report — thermodynamic averages computed after a run
• .rnemd — spatial temperature, density, and momentum profiles from RNEMD runs
• .pack — Packmol input for building initial systems
• .agr — Grace graphing package data, used for the published figures
• .xyz — Cartesian coordinates for visualization

Individual deposits document their own analysis-specific extensions and directory layouts. Simulation replicas are generally numbered in the filenames.

Simulation Datasets

Trace Gas Interactions with Ice Surfaces

Molecular dynamics simulations of acetone and methanol layers adsorbed on the basal facet of proton-ordered Ice-I_h, run at 223 K and 265 K alongside bare-ice controls. The archive includes quasi-liquid layer density profiles, tetrahedrality measurements, hydrogen bond statistics, mapped O–H stretching frequencies derived from local electric fields, and sum frequency generation signals computed under the time averaging approximation. Packmol input for building the adsorbate layers and all plotting data are also included. 117 MB.

Harless, B. M. and Gezelter, J. D. (2026). Supporting Data for “Elucidating Trace Gas Interactions with Ice Surfaces using Molecular Dynamics.” Zenodo. DOI: 10.5281/zenodo.18664169
Associated paper: J. Phys. Chem. C 130(28), 9932–9941 (2026). DOI: 10.1021/acs.jpcc.6c01908

Separation of Flexible Enantiomers Using Shear Flow

Simulations of two conformationally flexible chiral drugs — bicalutamide and montelukast — as single molecules in a range of solvents and as racemic solutions under reverse non-equilibrium molecular dynamics shear. The archive contains gas-phase geometry optimizations, GAFF2 parameters adapted for OpenMD, computed hydrodynamic resistance and diffusion tensors, scalar pitch and moments of pitch for all sampled geometries, mean displacements separating the two chiralities, orientational correlation functions, and radial distribution functions. Three shear rates are included for the montelukast free acid systems. 108 MB.

Pham, M. N., Cherek, L., and Gezelter, J. D. (2026). Supporting Data for “Separation of Flexible Enantiomers Using Shear Flow.” Zenodo. DOI: 10.5281/zenodo.19206241
Associated paper: J. Chem. Inf. Model. 66(15), 9458–9467 (2026). DOI: 10.1021/acs.jcim.6c01064

Thermal Transport at Metal Interfaces

Reverse non-equilibrium molecular dynamics simulations of interfacial thermal conductance across gold surfaces functionalized with the cationic surfactants CTAB and MTAB. Three low-index facets — Au(100), Au(110), and Au(111) — are covered at high and low ligand coverage, along with 10 Å radius gold nanoparticles. Runs performed with metal polarizability enabled are marked in the filenames. Analysis output includes axial and radial density profiles, charge densities, and Legendre polynomial orientational correlations. 2.3 GB.

Shavalier, S. A. and Gezelter, J. D. (2024). Supporting Data for “Thermal Transport Through CTAB- and MTAB-Functionalized Gold Interfaces using Molecular Dynamics Simulations.” Zenodo. DOI: 10.5281/zenodo.14232345
Associated paper: J. Chem. Inf. Model. 65(2), 811–824 (2025). DOI: 10.1021/acs.jcim.4c02195

OpenMD

Archived releases of the group’s molecular dynamics engine. OpenMD handles point-particle atoms, atoms with orientational degrees of freedom such as point multipoles and coarse-grained assemblies, and atoms with fluctuating charges. It runs in parallel under MPI and ships with the trajectory analysis tools used to generate most of the derived data in the datasets above. Development happens on GitHub; each release is archived. BSD 3-Clause licensed.

Drisko, C. R., Bhattarai, H., Fennell, C. J., Stocker, K. M., Vardeman, C. F. II, and Gezelter, J. D. OpenMD: A parallel molecular dynamics engine for complex systems and interfaces. Zenodo. DOI: 10.5281/zenodo.13953129 (all versions)

Associated paper: J. Open Source Software 9(102), 7004 (2024). DOI: 10.21105/joss.07004

Ice Structures, Free Energies and Phase Diagrams

Ice-i and Ice-i´ Crystals

The Ice-i and Ice-i´ crystal structures are quite similar to one another, the primary difference being an elongation along z of the entire crystal, together with distortions in the tetramer subunits shown below.

The crystals provided here are proton ordered. Proton disordered crystals are also possible — in fact, the crystallizations we observed were to proton disordered forms — but the anti-ferroelectric proton ordered forms are more stable in our simulations, so those are what we distribute. We have a small program that follows the procedure of Rahman and Stillinger [J. Chem. Phys. 57, 4009–4017 (1972)] to generate proton disordered structures from ice xyz files; email Dan Gezelter if you would like some disordered crystals.

Ice-i, Ice-i´, and ice I_h contain 1024 molecules; ice I_c contains 1000 and ice B contains 1728. Ice B is a simulation-only structure discovered by Báez and Clancy; ice I_h and ice I_c are the two real low-density forms, included for completeness. Files are available in both xyz and pdb format. The comment (second) line of each xyz file carries box dimensions for use with periodic boundary conditions; the same dimensions appear in the REMARK section of each pdb file.

If you are comparing energies, note that we did not use SHAKE or RATTLE in our simulations — OpenMD (and its predecessor OOPSE) treats water molecules as rigid bodies and integrates the orientational degrees of freedom directly, which makes bonding constraints unnecessary. We examined the crystal energies with both CHARMM and AMBER and obtained different values, primarily because of their use of SHAKE and secondarily because of differences in cutoff methods. The rigid body methods in TINKER compare well with those used in OpenMD.

Ice-i and Ice-i´ Tetramers

For those interested in the subunits that make up these crystals, the following clusters were excised from typical Ice-i and Ice-i´ crystals.

Tetramer 1 composes the major tetrameric rods in a proton ordered variant of the crystal. In Ice-i it is a relatively undistorted square; in Ice-i´ the square distorts into a rhombus. The distortion lets the O–H···O angle of the hydrogen bonds approach 180°.

Tetramer 2 shows that tetramer 1 is not the only route to proton ordering. This alternative is a hydrogen bond ring system in which the molecular dipoles are not directly opposed across the diagonal. Its O–H···O angles are roughly 164°, against the 166–169° found in tetramer 1, and the hydrogen bond donations run clockwise around the ring — the counter-clockwise arrangement is energetically equivalent. Under an orientational spin-relaxation approach to proton ordering, these tetramers appeared about as often as tetramer 1. Our free energy calculations used a crystal built from tetramer 1, but a crystal based on tetramer 2 would be worth investigating. We have not yet looked for a distorted Ice-i´ version of it.

Use measure angles to compare the O–H···O geometries directly: click three atoms in turn — donor oxygen, its hydrogen, then the acceptor oxygen — and Jmol will report the angle.

Questions about any of these structures? Send email.

Free Energies

Results for the 9 Å cutoff simulations. Free energies are in kcal/mol; the lowest free energy structure for each model is shown in bold.

Water Model	Ih	Ic	B	Ice-i	Ice-i’	Tm (*Ts)	Tb
TIP3P	-11.41(2)	-11.23(3)	-11.82(3)	-12.30(3)	–	269(4)	357(2)
TIP4P	-11.84(3)	-12.04(2)	-12.08(3)	–	-12.33(3)	266(5)	354(2)
TIP5P	-11.85(3)	-11.86(2)	-11.96(2)	–	-12.29(2)	271(4)	337(2)
SPC/E	-12.87(2)	-13.05(2)	-13.26(3)	–	-13.55(2)	296(3)	396(2)
SSD/E	-11.27(2)	-11.19(4)	-12.09(2)	-12.54(2)	–	*355(2)	–
SSD/RF	-11.51(2)	-11.47(2)	-12.08(3)	-12.29(2)	–	278(4)	349(2)

In all of these simulations, Ice-i (or Ice-i´) has the lowest free energy for all of the water models studied. The melting temperatures — with the obvious exception of SSD/E — show reasonably good agreement with the experimental value of 273 K, considerably better than the values observed for TIP4P from ice I_h (from 214 to 238 K). An important distinction is that these melting transitions are calculated from the most stable phase, Ice-i or Ice-i´. Calculating T_m from ice I_h instead places TIP4P’s melting transition near 210 K, in line with estimates from other groups.

This work is described in: Fennell, C. J. and Gezelter, J. D., “Computational free energy studies of a new ice polymorph which exhibits greater stability than Ice I_h,” J. Chem. Theory Comput. 1, 662–667 (2005). DOI: 10.1021/ct050005s