Molecular dynamics dataset: stacking fault tetrahedra under a supersonic solitary wave in L12 Pt3Al

Description

Molecular dynamics data for the interaction of a supersonic solitary wave with stacking fault tetrahedra in the ordered L12 intermetallic Pt3Al. Stacking fault tetrahedra of 28, 91 and 136 vacancies were built by the Silcox-Hirsch route in a [001] waveguide of 24 x 24 x 500 unit cells (1.152e6 lattice sites, lattice parameter 3.92 A) and exposed to a solitary wave train generated by nonlinear supratransmission, with a piston source of amplitude 0.18, 0.30, 0.40 or 0.50 A and period 0.146 ps (6.85 THz). Interactions were described by the embedded-atom potential of Zhou, Johnson and Wadley [Phys. Rev. B 69, 144113 (2004)]. The set contains 12 defect-bearing runs at 0 K, 4 pristine reference runs, and a seed-matched pair at 300 K. The deposit provides the LAMMPS input scripts and the potential file, the detector time series and the energies of two Lagrangian analysis spheres (r = 20 and 35 A) sampled every 0.25 ps, the space-time maps of the kinetic energy, the relaxed and 600 K-annealed tetrahedra of the verification cell, the final atomic configurations in a +-45 A slab about the defect on which the dislocation analysis was performed, the LAMMPS logs, the complete DXA and common-neighbour analysis of all 19 defect-bearing states and of the five pristine controls, the reduced trajectories from which the displacement of the tetrahedron relative to the surrounding material was measured, and the Python scripts that regenerate every table and figure of the associated article. Reference data for the planar Frank loops used for comparison, obtained in the same cell with the same potential and the same wave source, are deposited separately at https://doi.org/10.17632/d2kk8f7rk3 (vacancy loops) and https://doi.org/10.17632/3xcwrd7k3g (interstitial loops).

Steps to reproduce

1. Requirements. LAMMPS compiled with the MANYBODY package (tested with the 29 August 2024 Windows build with MS-MPI). Python 3 with numpy, pandas and matplotlib, and the OVITO Python module (pip install ovito); on a headless Linux machine the OpenGL runtime libraries libopengl0, libegl1 and libglx0 must also be present. 2. Run the series. Put the contents of input/ in the working directory and execute run_sft_series.bat. It first builds and relaxes the three tetrahedra in the small verification cell (check_sft.lmp, sft_make.mod.lmp) and then performs the 18 production runs defined in in.pt3al_wave_sft.lmp: three tetrahedron sizes at four source amplitudes, four pristine references at the same amplitudes, and a seed-matched defect-bearing and pristine pair at 300 K. One production run of 1.152e6 atoms over 35 000 steps takes a few hours on a desktop machine. The outputs reproduce data/detector_series/, data/ke_profiles/, data/prepared_states/, data/final_states/ and data/logs/. 3. Analyse. Run the scripts of analysis/ in this order: analyze_sft.py - reduction of the detector series to normalised transmission and net energy capture; writes data/summary_su11_en.csv run_dxa_check.py - dislocation extraction and common-neighbour analysis of the relaxed and 600 K-annealed states of the verification cell run_dxa.py - the same analysis for the final state of every run; writes tables/table_dxa.csv count_disorder.py - atoms of neither fcc nor hcp coordination, split into slab faces, defect neighbourhood and bulk; writes tables/table_disorder.csv measure_drift.py - displacement of the tetrahedron relative to the surrounding material; writes tables/table_drift.csv and the reduced series of data/sphere_reduced/ make_fig1.py, make_figs.py, make_ga.py - all figures of the article and the graphical abstract The scripts resolve their input folders automatically when run from analysis/ inside this deposit, and accept an input directory and an output file as command-line arguments otherwise. Running them on the deposited data reproduces every table and figure of the associated article. 4. Note on two inputs that are not deposited. LAMMPS restart files (about 1.7 GB) are omitted: every run is fully determined by the input script and, at 300 K, by the velocity seed recorded in the file name. The trajectories of the r = 20 A analysis sphere (about 340 MB) are also omitted; the per-frame reduction that measure_drift.py derives from them is shipped in data/sphere_reduced/, so the displacement result can be checked without them.

Institutions

• Altai State University

  Altai Krai, Barnaul

Categories

Related Links

Licence