Explicitly correlated renormalized second-order Green's function for accurate ionization potentials of closed-shell...

Area of Science:

• Computational Chemistry
• Quantum Chemistry
• Theoretical Chemistry

Background:

• Accurate calculation of molecular properties is crucial in chemistry.
• Traditional methods for ionization potential (IP) calculations can be computationally expensive.
• The nondiagonal renormalized second-order (NR2) Green's function method is a powerful tool, but its computational cost can be prohibitive.

Purpose of the Study:

• To introduce an energy-dependent explicitly correlated (F12) formalism for the NR2 Green's function method.
• To improve the accuracy and efficiency of IP calculations for closed-shell molecules.
• To reduce the computational cost associated with high-accuracy electronic structure calculations.

Main Methods:

• Development of an explicitly correlated (F12) extension to the NR2 Green's function method.
• Application of the NR2-F12 method to compute ionization potentials (IP) for sets of small and medium-sized molecules.
• Comparison of NR2-F12 results with the standard NR2 method and equation-of-motion ionized coupled-cluster singles and doubles (EOM-IP-CCSD).

Main Results:

• NR2-F12 with an aug-cc-pVTZ basis set achieved significantly lower mean basis set errors in IP compared to NR2 with larger basis sets (e.g., aug-cc-pV5Z, aug-cc-pVQZ).
• For small molecules, NR2-F12 (aug-cc-pVTZ) showed a mean basis set error of 0.028 eV, outperforming NR2 (aug-cc-pV5Z) at 0.044 eV.
• For organic electron acceptor molecules (OAM24), NR2-F12 (aug-cc-pVTZ) yielded a mean basis set error of 0.015 eV, compared to 0.067 eV for NR2 (aug-cc-pVQZ).

Conclusions:

• The NR2-F12 method provides accurate ionization potentials at a reduced computational cost.
• The computational cost of NR2-F12 scales as O(N^6) noniteratively and O(N^5) iteratively with system size.
• NR2-F12 performance is comparable to EOM-IP-CCSD at a small basis set, offering a more efficient alternative for accurate IP calculations.