Convergence and Properties of Intrinsic Bond Orbitals in Solids

Area of Science:

• Condensed Matter Physics
• Computational Materials Science
• Quantum Chemistry

Background:

• Localized Wannier orbitals are crucial for understanding electronic structure in solids.
• Constructing these orbitals in large supercells is computationally intensive.
• Existing methods face challenges with different bonding types and material systems.

Purpose of the Study:

• To investigate the construction and spatial properties of localized Wannier orbitals in large supercells.
• To develop and assess efficient algorithms for Wannier orbital localization.
• To analyze the relationship between orbital localization and material properties across diverse bonding types.

Main Methods:

• Utilized the Pipek-Mezey (PM) functional with intrinsic atomic orbitals (IAOs) to form intrinsic bond orbitals (IBOs).
• Employed plane waves as the basis set for calculations in large supercells.
• Evaluated the performance and scaling of various optimization algorithms (BFGS, CG, SA, DIIS) for orbital localization.

Main Results:

• Observed a correlation between orbital spreads and geometric properties, independent of bonding type and band gap.
• Found comparable sparsity patterns in the Hartree-Fock exchange matrix for covalent, polar covalent, and ionic bonds.
• Demonstrated that metal oxides require significantly more computational effort for Wannier orbital construction.

Conclusions:

• Intrinsic bond orbitals provide a consistent framework for analyzing localized electronic structure across different materials.
• Efficient algorithms for Wannier orbital localization are crucial for large-scale material simulations.
• The computational cost of constructing localized Wannier orbitals varies significantly with material type, with metal oxides posing a particular challenge.