Defects and Reactions

Overview

This skill group covers calculations involving crystallographic defects and chemical reaction pathways in solid-state materials. Two main approaches are available:

1. MACE (via ASE) -- Fast ML-potential-based calculations. Good for rapid screening of defect formation energies, NEB barriers, and adsorption energies. Seconds to minutes per calculation on typical supercells.
2. Quantum ESPRESSO (QE) -- Full DFT. Required for publication-quality energetics, charged defect calculations, electronic structure at defect sites, and accurate chemical bonding at surfaces.

Both approaches follow workflows inspired by atomate2's defect, NEB, and adsorption flows: create the defect/surface structure, relax, compute relevant energies, and post-process thermodynamic quantities.

Sub-Skills

Method Decision Guide

What do you need?

Defect formation energy (neutral)?
  Quick screening --> ASE + MACE (point-defect/)
  Publication quality --> QE DFT (point-defect/)

Charged defect formation energy / transition levels?
  --> QE DFT required (point-defect/, need electrostatic corrections)

Migration barrier / reaction pathway?
  Quick estimate --> ASE + MACE NEB (neb-transition-state/)
  Accurate barrier --> QE NEB (neb-transition-state/)

Explore unknown transitions / don't know the final state?
  --> ART nouveau (activation-relaxation-technique/)
  Build KMC event catalog --> ART with systematic sampling (activation-relaxation-technique/)

Adsorption energy / surface chemistry?
  Quick screening --> ASE + MACE (surface-adsorption/)
  Publication quality --> QE DFT with slab model (surface-adsorption/)

Non-radiative recombination / luminescence quenching / carrier capture?
  Structural screening (DeltaQ) --> ASE + MACE (configuration-coordinate/)
  Publication quality (ZPL, barrier, Huang-Rhys) --> QE DFT (configuration-coordinate/)

Common Prerequisites

• Structure: Start from a CIF, POSCAR, or Materials Project query. Use pymatgen for structure manipulation (supercells, defect creation, slab generation).
• Pseudopotentials: QE calculations need pseudopotential files (SSSP library recommended).
• Python packages: pymatgen, ASE, mace-torch, numpy, scipy, matplotlib are pre-installed. Install extras with pip install pymatgen-analysis-defects pymatgen-diffusion as needed.
• Supercell sizes: Defect and NEB calculations require supercells large enough to minimize periodic image interactions. Typical minimum: 3x3x3 for cubic, or at least 10 A between periodic images.