Hybrid Density Functional Valence Bond Method with Multistate Treatment.

Area of Science:

• Quantum Chemistry
• Computational Chemistry
• Theoretical Chemistry

Background:

• The hybrid density functional valence bond (VB) method, λ-DFVB(U), offers accuracy comparable to CASPT2 for various chemical properties.
• However, λ-DFVB(U) neglects electronic state interactions, leading to incorrect potential energy surface (PES) topologies near conical intersections.

Purpose of the Study:

• To develop a novel hybrid density functional VB method, named λ-DFVB(MS), that incorporates multistate treatments.
• To address the limitations of λ-DFVB(U) by accurately describing the topology of PESs in regions of conical intersection.

Main Methods:

• The proposed λ-DFVB(MS) method constructs an effective Hamiltonian matrix using diabatic states derived from a VB-based compression approach.
• Diagonalization of this effective Hamiltonian matrix allows for the inclusion of interactions between electronic states.
• The VBSCF wave function with selected VB structures can serve as a reference within the λ-DFVB(MS) framework.

Main Results:

• λ-DFVB(MS) successfully reproduces the correct topology of PESs near conical intersection regions in test calculations.
• The method demonstrates the capability to accurately capture the complex interactions between electronic states.

Conclusions:

• λ-DFVB(MS) represents a significant advancement in computational chemistry for accurately modeling PESs, particularly in regions with conical intersections.
• This method provides a more reliable tool for studying chemical reactions and excited states where electronic state interactions are critical.