Using Noncovalent Interactions to Test the Precision of Projector-Augmented Wave Data Sets

Area of Science:

• Computational chemistry
• Materials science
• Quantum mechanics

Background:

• The projector-augmented wave (PAW) method accelerates electronic structure calculations by approximating core electrons.
• Understanding PAW approximations is crucial for accurate computational results.
• Current tests using atoms or solids are insufficient to diagnose PAW error origins.

Purpose of the Study:

• To demonstrate the utility of molecular dimer interaction energies for assessing PAW data sets.
• To identify the origins of errors in PAW calculations.
• To develop and evaluate a method for correcting PAW-derived interaction energies.

Main Methods:

• Utilized molecular dimers from S22 and S66 test sets and additional dimers.
• Analyzed the components of interaction energy errors, including short-range and long-range electrostatic contributions.
• Proposed and tested a correction scheme for long-range electrostatic errors.

Main Results:

• Molecular dimer interaction energies effectively identify problematic PAW data sets and pinpoint error sources.
• PAW errors in interaction energy comprise a short-range exponential decay and a long-range electrostatic component due to charge density inaccuracies.
• The proposed correction scheme reduced interaction energy errors by over 50% on average for hydrogen-bonded dimers.

Conclusions:

• Molecular dimer interaction energies provide deeper insights into PAW method inaccuracies than atomic or solid-state tests.
• The identified error components offer a pathway for improving PAW data sets.
• The developed correction scheme offers a practical and effective way to enhance the accuracy of PAW calculations for molecular systems.