mat-dft-electronic-transport

mat-dft-electronic-transport

Goal

To determine high-fidelity electronic transport properties (e.g., carrier mobility $\mu$, Seebeck coefficient $S$, and electrical conductivity $\sigma$) across various doping concentrations and temperatures using the AMSET (Ab initio Scattering and Transport) package integrated directly into an atomate2 VASP computational flow.

Background

Machine Learning Interatomic Potentials (MLIPs) only predict energies, forces, and stresses; they lack proper electronic wavefunction representations. True electronic transport capabilities require coupling dense Density Functional Theory (DFT) band structures with detailed scattering matrix calculations (acoustic deformation potential scattering, polar optical phonon scattering, etc.). This skill leverages VaspAmsetMaker to seamlessly chain these calculations natively.

Instructions

1. Construct the AMSET Workflow

Use the provided script to generate the sequence (DAG) of VASP computations targeting electronic transport. This automated DAG coordinates structure relaxation, uniform band structure extraction, evaluation of the elastic tensor, and calculations of deformation potentials.

# Env: atomate2-agent
python .agents/skills/mat-dft-electronic-transport/scripts/generate_inputs.py --output amset_flow.json

2. Job Execution (via jobflow/Fireworks)

Because this workflow contains numerous sequential and parallel VASP evaluations (e.g., generating strained supercells for deformation potentials), it should be passed to your job management framework rather than run individually. The default script simply serializes the theoretical DAG to JSON.

If operating on a compute-capable node with vasp_std available, it can be tested locally using:

import jobflow
# Assuming `flow` is the defined VaspAmsetMaker output
jobflow.run_locally(flow, create_folders=True)

3. Extract Transport Results

Once completed, the final node wraps the AMSET runner. Resulting transport parameters (Mobility, Conductivity) will be dumped into a structured .json and amset.log inside the final Job's folder. Parse these properties natively using standard amset.plot utilities.