Accurate potential energy surfaces with a DFT+U(R) approach

Area of Science:

• Computational Chemistry
• Materials Science
• Quantum Mechanics

Background:

• Density Functional Theory (DFT) with Hubbard U (DFT+U) corrects self-interaction errors for localized electrons.
• Standard DFT+U often uses an averaged Hubbard U, limiting accuracy across different molecular geometries.
• Accurate modeling of localized d and f electrons is crucial for many materials and molecules.

Purpose of the Study:

• To introduce a position-dependent DFT+U (DFT+U(R)) method that incorporates geometry variations.
• To overcome the limitation of using an averaged Hubbard U in energy comparisons.
• To improve the accuracy of DFT+U for systems with significant U variations with geometry.

Main Methods:

• Developed a DFT+U approach where the Hubbard U value is self-consistently calculated and varies with geometry (DFT+U(R)).
• Applied the linear-response method to determine the position-dependent U(R).
• Validated DFT+U(R) against accurate quantum chemistry methods and experimental data for diatomic molecules.

Main Results:

• DFT+U(R) demonstrated a two- to four-fold improvement over standard DFT+U for diatomic molecules.
• Reduced errors in binding energies, vibrational frequencies, and equilibrium bond lengths.
• Showed the method's relevance for systems like transition-metal oxides where U varies significantly with geometry.

Conclusions:

• The position-dependent DFT+U(R) offers a significant advancement over standard DFT+U for localized electrons.
• Provides a criterion to assess the necessity of standard DFT+U versus DFT+U(R).
• The DFT+U(R) approach is applicable to complex systems and can be integrated into standard computational workflows.