##############################################################################################################

Developments and Further Applications of Ephemeral Data Derived Potentials

P. T. Salzbrenner, S. H. Joo, L. J. Conway, P. I. C. Cooke, B. Zhu, M. P. Matraszek, W. C. Witt, C. J. Pickard

https://arxiv.org/abs/2306.06475

##############################################################################################################

# Contents

## Carbon

- Carbon Ephemeral Data-Derived Potentials (EDDPs) used in the manuscript

	- C-EDDP-Figure-4a:
		The EDDP which was used to generate the potential energy curves in figure 4(a).

	- C-EDDP-Figure-4b:
		The EDDP which was used to generate the potential energy curves in figure 4(b). Also referred to as EDDP I in section V.A of the manuscript.

	- C-EDDP-Qamar-data:
		The EDDP trained on the dataset curated by Qamar *et al.* [1]. Also referred to as EDDP II in section V.A of the manuscript.

- Carbon geometry optimisation and bulk modulus calculation:
	Carried out by fitting the Birch-Murnaghan EOS to total energy calculations at a variety of lattice parameters. Total energies derived from:

	- C-DFT-geom-opt:
		Density Functional Theory (DFT)

	- C-EDDP-I-geom-opt:
		EDDP I (The EDDP used to generate figure 4(b))

	- C-EDDP-II-geom-opt:
		EDDP II (The EDDP trained on the dataset curated by Qamar *et al.* [1])

- Input and output of Carbon potential energy surface calculations:
	Generated using total energy calculations for different structures at a range of interatomic separations. These data constitute different components of figure 4:

	- C-PES-DFT:
		DFT points in both figures 4(a) and (b).

	- C-PES-EDDP-Figure-4a:
		EDDP curves in figure 4(a).

	- C-PES-EDDP-Figure-4b:
		EDDP curves in figure 4(b).

## Lead

- Lead EDDPs used in the manuscript

	- Pb-EDDP-no-SOC:
		EDDP trained on DFT data not including spin-orbit coupling (SOC)

	- Pb-EDDP-SOC:
		EDDP trained on DFT data including spin-orbit coupling (SOC)

- Pb-DFT-enthalpies:
	Input and output of the DFT pressure-enthalpy calculations presented in figure 5. The output data are in `good_castep` and a relative enthalpy plot can be generated using the `AIRSS` [3] `cryan` tool with `ca -e`.

- Lead phonon calculations

	- Pb-DFT-phonons-no-SOC:
		Output data of a DFT phonon calculations of Fm-3m lead without SOC, carried out using the finite difference Caesar code [2]. Contains dispersion curve (phonon_dispersion_curve_meV.dat) and vibrational energies (interpolated_free_energy.dat).

	- Pb-DFT-phonons-no-SOC:
		Output data of a DFT phonon calculations of Fm-3m lead with SOC, carried out using the finite difference Caesar code [2]. Contains dispersion curve (phonon_dispersion_curve_meV.dat) and vibrational energies (interpolated_free_energy.dat).

	- Pb-EDDP-phonons-no-SOC:
		Input data for a `wobble` phonon calculation using the EDDP without SOC. Results in the manuscript can be reproduced with `wobble -disp -therm -natom 1000 -unit meV Pb`.

	- Pb-EDDP-phonons-no-SOC:
		Input data for a `wobble` phonon calculation using the EDDP with SOC. Results in the manuscript can be reproduced with `wobble -disp -therm -natom 1000 -unit meV Pb`.

- Pb-coex-example:
	Input files including run command for a coexistence melting point calculation using `ramble`. The example uses the potential including SOC and is for Fm-3m lead at 0 GPa, and can easily be adapted to reproduce the other pressure-points also.

- Lead enthalpy and free energy calculations using the EDDPs: the `good_castep` directory contains the output of the enthalpy calculations, as well as the ensemble deviation data. The latter can also be reproduced using `res2dev` in the EDDP code suite. the `good_castep_*_thermal` directories contain the free energy calculations done using wobble and the input files (melt_curve.dat, phase_diagram_data.dat, phase_transition_points.dat) used to generated the phase diagrams.

	- Pb-EDDP-enthalpy-phase-diagram-no-SOC:
		Calculations done using the potential without SOC and used to generate figure 7(a).

	- Pb-EDDP-enthalpy-phase-diagram-SOC:
		Calculations done using the potential without SOC and used to generate figure 7(b).

## Scandium hydride

- Scandium hydride EDDPs used in the manuscript

	- ScH-EDDP-searching:
		The EDDP used for carrying out the structure searches in section V.C.1 of the manuscript.

	- ScH-EDDP-MD:
		The EDDP used for carrying out the molecular dynamics simulations in section V.C.2 of the manuscript.

- Scandium hydride structure searches
	
	Convex hulls can be generated using the `AIRSS` [3] `cryan` tool with `ca -m`.

	- ScH-structures-broad-search:
		The local minimum structures returned by the AIRSS search of the structure space described in the third panel of figure 9(a).

	- ScH-structures-narrow-search:
		The local minimum structures returned by the AIRSS search of the structure space described in the second panel of figure 9(a).

- ScH-superionicity-example:
	Input files including run command for an NPT MD simulation using `ramble` and testing for superionicity. The example is at 1000K and 350 GPa, and can easily be adapted to reproduce the other temperature-points also.

## Zinc cyanide

- Zinc cyanide EDDPs used in the manuscript

	- ZnCN-EDDP-TS:
		The EDDP trained with the Tkatchenko-Scheffler [4] long-range interaction in the DFT data.

	- ZnCN-EDDP-MBD:
		The EDDP trained with the many-body dispersion [5] long-range interaction in the DFT data.

- ZnCN-search-TS:
	The local minimum structures returned by the AIRSS search of the structure space described in the manuscript. `mof.packed.res` contains all structures found. `good castep` contains known structures, which were added manually and whose energies were calculated for comparison with the *ab initio* search.

- Thermal expansion molecular dynamics simulations

	- ZnCN-MD-TS:
		NPT molecular dynamics simulations at 0 GPa to determine the thermal expansion of several Zn(CN)2 MOFs, using the EDDP trained with the Tkatchenko-Scheffler [4] long-range interaction in the DFT data. Directories for the different runs are labelled in the schema {structure}-{temperature}, and contain the input `.cell` and the output `.track` files.

	- ZnCN-MD-MBD:
		NPT molecular dynamics simulations at 0 GPa to determine the thermal expansion of several Zn(CN)2 MOFs, using the EDDP trained with the Tkatchenko-Scheffler [4] long-range interaction in the DFT data. Directories for the different runs are labelled in the schema {structure}-{temperature}, and contain the input `.cell` and the output `.track` files.

# A note on `ramble` `.track` output files

The columns, from left to right, are Time [natural units] - Volume [Ang<sup>3</sup>] - Current Temperature [K] - Target Temperature [K] - Current Pressure [GPa] - Internal energy [eV/atom] - Target Pressure [GPa] - Total energy [eV/atom]. The zinc cyanide thermal expansion calculations were carried out with an old version of `ramble`. The current release prints time in ps instead of natural units.

## References

[1]: M. Qamar, M. Mrovec, Y. Lysogorskiy, A. Bochkarev, and R. Drautz, “Atomic cluster expansion for quantum-accurate large-scale simulations of carbon,” arXiv:2210.09161 [cond-mat.mtrl-sci] (2022).

[2]: J. H. Lloyd-Williams and B. Monserrat, "Lattice dynamics and electron-phonon coupling calculations using nondiagonal supercells," Phys. Rev. B 92, 184301 (2015).

[3]: C. J. Pickard and R. J. Needs, “Ab initio random structure searching,” Journal of Physics: Condensed Matter 23, 053201 (2011).

[4]: A. Tkatchenko and M. Scheffler, “Accurate Molecular Van Der Waals Interactions from Ground-State Electron Density and Free-Atom Reference Data,” Phys. Rev. Lett. 102, 073005 (2009).

[5]: A. Tkatchenko, R. A. DiStasio, R. Car, and M. Scheffler, “Accurate and Efficient Method for Many-Body van der Waals Interactions,” Phys. Rev. Lett. 108, 236402 (2012).