Charge-Density-Wave Phase Transitions in Monolayer 1<i>T</i>-TaS₂ - Supplemental Materials

Valentina Nesterova, Tribhuwan Pandey, Tom Berlijn, Fariborz Kargar, Lucas Lindsay, Konstantin Klyukin

The files support the work "<b>Charge-Density-Wave Phase Transitions in Monolayer 1</b><b><i>T</i></b><b>-TaS</b>_{<strong>2</strong>}<b> from Universal Machine Learning Molecular Dynamics</b>". In this study, phase transitions in monolayer 1<i>T</i>-TaS₂ were investigated using Molecular Dynamics (MD) simulations based on universal Machine Learning Interatomic Potentials (uMLIPs).This repository contains input and output files from UMA calculations, video files of MD trajectories, and scripts used for data processing.<b>MD Trajectories </b>md_60.traj, md_300.traj, md_500.trajMolecular Dynamics (MD) trajectories obtained using UMA s-1p1 interatomic potentials at 60, 300, and 500 K, generated via sequential heating. Due to the file size, other temperatures are not included in this archive. The reader may contact the corresponding author if needed. Length of each trajectory is 50 ps, timestep 1 fs, every fifth step recorded (5 fs sampling interval).md_60.mp4, md_300.mp4, md_500.mp4Visualization of the trajectories listed above. Each frame of the video shows positions of Ta atoms averaged across 2 ps (400 recorded steps). Star-of-David motifs were identified using the same logic as in the plot_sod.py script.<br><b>Scripts</b>UMA_MD.pyRuns the molecular dynamics simulations using the UMA s-1p1 potential. Performs sequential heating MD in the NVT Langevin ensemble: each temperature step is initialized from the final positions/velocities of the previous step, equilibrated, and run for the target duration. Trajectory frames are saved every 5 fs.parse_uma.pyReads an ASE MD trajectory (from UMA_MD.py) and converts the sampled frames into the input format required by TDEP (positions, forces, and metadata files) for extracting temperature-dependent effective force constants.plot_sod.pyPost-processing/visualization script. Reads an MD trajectory, averages Ta atom positions over a chosen frame window, classifies each Ta atom as "primitive" or "SoD" based on Ta–Ta bond-length criteria (tolerance set from the Ta RMSD over the trajectory), and plots a spatial map of the structure with SoD bonds highlighted.<br><b>TDEP Input Files</b>infile.ucposcarTDEP unit-cell structure file (POSCAR format) - the reference small primitive (NM) cell used to define the force-constant expansion.infile.ssposcarTDEP supercell structure file (POSCAR format) - the simulation supercell matching the MD trajectory, used together with infile.ucposcar to map forces/displacements onto the lattice during TDEP fitting.