Automated Transition State Searches without Evaluating the Hessian.

Area of Science:

• Computational Chemistry
• Chemical Kinetics
• Reaction Mechanism Elucidation

Background:

• Accurate transition structure (TS) determination is crucial for computational reaction pathway studies.
• Existing methods often require computationally expensive electronic structure calculations and good initial guesses.
• The freezing string method (FSM) offers a more cost-effective approach than the growing string method (GSM).

Purpose of the Study:

• To improve the efficiency of the freezing string method (FSM) for TS searches.
• To reduce the computational cost associated with Hessian calculations in TS determination.
• To present a combined approach of an optimized FSM and an approximate Hessian for faster TS identification.

Main Methods:

• Replacing the conjugate gradient (CG) optimization in FSM with a quasi-Newton line search and BFGS Hessian update (FSM-BFGS).
• Developing an approximate Hessian construction using tangent direction and local curvature at the TS guess.
• Employing the partitioned-rational function optimization algorithm with the approximate Hessian.

Main Results:

• The FSM-BFGS method demonstrates improved efficiency over FSM-CG.
• The approximate Hessian construction performs comparably to an exact Hessian in most test cases.
• The combined techniques significantly reduce the computational cost of finding transition structures.

Conclusions:

• The FSM-BFGS method offers a more efficient approach for TS searches.
• Using an approximate Hessian is a viable strategy to lower computational expenses.
• These advancements provide a faster and more cost-effective route for exploring chemical reaction pathways computationally.