Projectability disentanglement for accurate and automated electronic-structure Hamiltonians

Area of Science:

• Solid-state physics and materials science.
• Computational condensed matter physics.

Background:

• Maximally-localized Wannier functions (MLWFs) are crucial for characterizing material electronic structures.
• Constructing MLWFs for isolated or entangled bands often demands expertise and iterative refinement, hindering large-scale computations.

Purpose of the Study:

• To introduce an automated method for generating accurate and compact maximally-localized Wannier functions (MLWFs).
• To provide a robust tight-binding representation of electronic states in crystalline solids.

Main Methods:

• Developed projectability-disentangled Wannier functions (PDWFs) by introducing a projectability measure.
• Utilized this measure to guide the disentanglement process for Bloch states onto atomic orbitals.
• Applied the method to spanned occupied and empty electronic bands.

Main Results:

• Successfully constructed MLWFs for both occupied and unoccupied electronic bands.
• Demonstrated the accuracy of PDWFs on a diverse set of 200 materials.
• Validated the reliability through the generation of 21,737 Wannier Hamiltonians.

Conclusions:

• PDWFs offer an automated and reliable approach to obtaining tight-binding models from first-principles calculations.
• This method overcomes previous limitations, facilitating high-throughput electronic structure analysis.