Overview

A comprehensive series of molecular dynamics simulations studying atomic diffusion on metal surfaces using LAMMPS. This project generates high-quality trajectory data specifically designed for training machine learning models in materials science applications.

Key Features

  • Complete LAMMPS simulation setups for Cu and Pt surface systems
  • Detailed analysis tools for trajectory and energy data
  • ML-ready datasets optimized for materials property prediction
  • Reproducible workflows with full documentation
  • Surface catalysis focus with applications to real-world materials problems

Technical Details

Simulation Systems

  • Cu(111) surface with adatom diffusion studies
  • Pt(111) surface with comparative analysis
  • Temperature-dependent simulations (300K - 800K)
  • Long-timescale trajectories for statistical significance

Analysis Pipeline

  • Trajectory processing and feature extraction
  • Energy landscape analysis
  • Diffusion coefficient calculations
  • ML-ready data formatting

Applications

  • Neural network potentials for accelerated MD simulations
  • Surface catalysis modeling for chemical reaction prediction
  • Materials discovery through automated property screening
  • Activation energy prediction using ML approaches

Results & Impact

The simulations have generated datasets used in several machine learning studies focused on:

  • Predicting diffusion barriers from atomic configurations
  • Learning interatomic potentials for extended simulations
  • Understanding temperature-dependent diffusion mechanisms

This project directly supports the research documented in:

Usage

The simulation setups and analysis scripts are designed to be easily reproducible and extensible to other metal surface systems. All workflows are documented with step-by-step instructions.