Performance of Multilevel Methods for Excited States

Area of Science:

• Computational chemistry
• Quantum chemistry
• Theoretical chemistry

Background:

• Multilevel quantum chemical approaches are crucial for modeling electronic excited states.
• Atom-based system partitioning schemes are commonly used in these methods.
• Accurate modeling of excited states is essential for understanding chemical processes.

Purpose of the Study:

• To evaluate the performance of various atom-based multilevel quantum chemical methods for modeling electronic excited states.
• To compare the accuracy and robustness of different embedding techniques.
• To identify the most reliable and cost-efficient methods for high-accuracy excitation energy calculations.

Main Methods:

• The study employed the second-order algebraic-diagrammatic construction [ADC(2)] as the high-level method.
• Several embedding techniques were assessed: mechanical-embedding (ME) and electronic-embedding (EE) of ONIOM, point charge embedding (PCE), frozen density-embedding (FDE), and projector-based embedding (PbE).
• The XH-27 test set, featuring organic dyes interacting with solvent molecules, was used for the assessment.

Main Results:

• Local domain-based ADC(2) [L-ADC(2)] demonstrated the highest reliability for modeling excited states.
• ONIOM-ME and ONIOM-EE were found to be the least robust methods.
• PbE, FDE, and PCE techniques achieved accuracy comparable to L-ADC(2) but exhibited lower precision.

Conclusions:

• L-ADC(2) is the most reliable approach for modeling electronic excited states among the tested methods.
• Improvements in subsystem selection or the inclusion of charge-transfers are needed for atom-based methods to achieve high accuracy.
• Further development is required for cost-efficient methods to match the precision of L-ADC(2).