Predicting Defect States: A Quick Screening Protocol for Substitutional Point Defect Engineering

Abstract

Point defects in crystalline materials play a central role in determining electronic, optical, and magnetic properties. However, systematic exploration of defect configurations remains computationally expensive because large supercell calculations are required to approximate isolated defects under periodic boundary conditions. We present a unit-cell-based tight-binding protocol that enables rapid pre-screening of substitutional defects. The protocol extracts Wannier tight-binding Hamiltonians from small, fully relaxed unit cells of the host and defect-like systems, replicates the host Hamiltonian to construct a supercell model, and introduces the defect by modifying only the on-site energies at the substitution site while leaving the hopping parameters unchanged. We validate the protocol across three diverse systems: isostructural substitutional defects in transition-metal dichalcogenides (M$_\mathrm{Mo}$ MoS$_2$, M = Ce, Zr, Nb, Tc, and Ru), symmetry-breaking carbon substitutions in hexagonal boron nitride (C$_\mathrm{B}$C$_\mathrm{N}$ h-BN), and nitrogen-vacancy (NV$^-$) centers in diamond. These case studies span two-dimensional and three-dimensional hosts, simple substitutions, and substitution-vacancy complexes. In all cases, the protocol successfully captures the number of in-gap states, their degeneracies, and their shallow or deep character relative to host band edges, despite some quantitative deviations in absolute energy positions. We further identify limitations for vacancies of highly electronegative atoms and for charge-state or spin-polarization effects, both of which involve self-consistent charge redistribution not captured by the protocol.