Accuracy of Partial Core Corrections Using Fourier Transforms in Pseudopotential-Density Functional Theory

Area of Science:

• Computational physics
• Materials science
• Quantum chemistry

Background:

• Accurate electronic structure calculations rely on precise exchange-correlation functionals and pseudopotentials.
• Partial core corrections are crucial for improving the accuracy and transferability of these methods.
• A common method involves Fourier transforms to compute partial core charge density.

Purpose of the Study:

• To investigate numerical errors in exchange-correlation potentials arising from Fourier transforms of partial core charge density.
• To identify the origin of these errors and their impact on electronic structure calculations.
• To propose an alternative method for accurate computation.

Main Methods:

• Analysis of partial core charge density calculations using Fourier transforms.
• Investigation of errors in the vacuum region for low-dimensional materials and surfaces modeled with supercells.
• Comparison with direct real-space calculation of partial core charge density.

Main Results:

• Widely used Fourier-space methods can introduce sizable numerical errors in exchange-correlation potentials.
• These errors, stemming from slow decay in reciprocal space, can reach ~1 eV for Kohn-Sham energies of unoccupied states.
• Direct real-space calculation effectively avoids these Fourier transform-induced numerical errors.

Conclusions:

• Fourier transforms in partial core charge density calculations can compromise accuracy, especially in vacuum regions.
• The direct real-space approach offers a more reliable alternative for accurate electronic structure and pseudopotential calculations.
• This finding is critical for reliable modeling of advanced materials and surfaces.