Evaluation of DFT-D3 dispersion corrections for various structural benchmark sets

Area of Science:

• Computational Chemistry
• Quantum Chemistry
• Materials Science

Background:

• Accurate molecular structures are crucial for predicting material properties.
• Dispersion corrections in density functional theory (DFT) are essential for describing non-covalent interactions.
• Existing methods like DFT-D3(BJ) offer good accuracy but can be computationally intensive.

Purpose of the Study:

• To evaluate the performance of a new DFT-D3 dispersion correction, D3(CSO).
• To compare D3(CSO) with the established D3(BJ) correction for geometry optimizations.
• To present an optimized algorithm for DFT-D3 gradient calculations.

Main Methods:

• Utilized benchmark datasets for bond lengths, rotational constants, and supramolecular complex center of mass distances.
• Performed geometry optimizations using both DFT-D3(CSO) and DFT-D3(BJ) corrections.
• Developed and implemented an optimized algorithm for DFT-D3 gradient computation.

Main Results:

• Both D3(CSO) and D3(BJ) dispersion corrections yield accurate molecular structures.
• No systematic structural differences were observed between D3(CSO) and D3(BJ).
• The optimized gradient algorithm reduces the computational scaling from O(N^3) to O(N^2).

Conclusions:

• The new D3(CSO) dispersion correction is a viable alternative to D3(BJ) for DFT geometry optimizations.
• The developed gradient algorithm significantly enhances computational efficiency for DFT-D3 calculations.
• These advancements contribute to more accurate and efficient computational chemistry methods.