A Hessian-Based Interpolation Method for Locating Transition Structures

Area of Science:

• Computational chemistry
• Chemical kinetics

Background:

• Accurate identification of transition states is crucial for understanding reaction mechanisms.
• Current methods for locating saddle points can be computationally expensive, especially for flexible molecules.

Purpose of the Study:

• To develop an efficient Hessian-based interpolation scheme for estimating first-order saddle points.
• To enable automated screening of numerous transition structures and global reaction networks.

Main Methods:

• Constructed two molecular-specific, force-field-inspired energy functions using reactant and product structures and Hessians.
• Calculated saddle point estimates as minima on the intersection of two diabatic energy surfaces.
• Developed an approximate Hessian construction at the intersection minimum for initiating saddle point searches.

Main Results:

• The Hessian-based interpolation (HINT) procedure is computationally efficient.
• The method allows for automated screening of thousands of potential transition structures.
• HINT successfully located 90% of transition structures in an automated workflow for a benchmark set of 262 reactions.

Conclusions:

• The proposed Hessian-based interpolation scheme provides an efficient and automated approach for saddle point estimation.
• This method facilitates the creation of global reaction networks and the screening of transition structures.
• HINT offers a powerful tool for advancing computational chemistry and reaction mechanism studies.