A computationally efficient double hybrid density functional based on the random phase approximation.

Area of Science:

• Quantum Chemistry
• Computational Chemistry
• Materials Science

Background:

• Density Functional Theory (DFT) is a cornerstone of modern computational chemistry.
• Double Hybrid Density Functionals (DHDFs) offer improved accuracy but can be computationally expensive.
• Accurate prediction of molecular properties is crucial for catalyst design and understanding chemical reactions.

Purpose of the Study:

• Introduce a revised double hybrid density functional, PWRB95.
• Improve computational efficiency and accuracy for a range of chemical systems.
• Evaluate PWRB95 against established methods for thermochemistry, non-covalent interactions, and transition metal catalysis.

Main Methods:

• Developed PWRB95 incorporating Perdew-Wang exchange, Becke95 correlation, Fock exchange, random phase approximation (RPA) for non-local correlation, and VV10 dispersion.
• Adjusted three empirical scaling parameters.
• Tested PWRB95 on the GMTKN30 thermochemical database, transition metal catalysis, and challenging electronic cases.

Main Results:

• PWRB95 shows accuracy comparable to standard DHDFs for thermochemistry.
• PWRB95 demonstrates superior or similar performance for non-covalent interactions and transition metal systems.
• Achieved significant computational savings (approx. 40x) due to reduced basis set dependence, enabling triple-zeta quality calculations.

Conclusions:

• PWRB95 offers a computationally efficient and accurate alternative to existing DHDFs.
• The functional shows promise for studying complex chemical systems, including catalysis and electronically challenging molecules.
• PWRB95 provides an accurate description of bond distances, outperforming other advanced methods.