Freeze-and-Release Direct Optimization Method for Variational Calculations of Excited Electronic States.

Area of Science:

• Computational Chemistry
• Quantum Chemistry
• Theoretical Chemistry

Background:

• Variational optimization of excited electronic states is challenging due to saddle points on energy landscapes.
• Charge-transfer excitations, involving significant electron density rearrangement, exacerbate optimization difficulties.

Purpose of the Study:

• To present a simple and effective strategy for variational orbital optimization of excited electronic states.
• To address the challenges in optimizing excited states, particularly those involving charge-transfer excitations.

Main Methods:

• A novel 'freeze-and-release' strategy for orbital optimization is introduced.
• The method combines energy minimization with frozen orbitals, followed by unconstrained saddle-point optimization.
• Direct optimization algorithms are employed, maintaining computational scaling similar to ground-state calculations.

Main Results:

• The freeze-and-release approach successfully avoids variational collapse to spurious solutions, outperforming conventional methods like the maximum overlap method.
• Calculations on intramolecular and intermolecular charge-transfer excited states demonstrate the method's robustness.
• For intermolecular charge transfer, the method accurately captures energy dependencies without needing long-range exact exchange.

Conclusions:

• The presented freeze-and-release direct optimization strategy offers a reliable method for calculating excited electronic states, especially charge-transfer states.
• This approach overcomes limitations of existing methods in time-dependent density functional theory, particularly for intermolecular charge transfer scenarios.