An efficient and flexible approach for local distortion: distortion distribution analysis enabled by fragmentation.

Distortion can play crucial roles in influencing structures and properties, as well as enhancing reactivity or selectivity in many chemical and biological systems. The distortion/interaction or activation-strain model is a popular and powerful method for deciphering the origins of activation energies, in which distortion and interaction energies dictate an activation energy. However, decomposition of local distortion energy at the atomic scale remains less clear and straightforward. Knowing such information should deepen our understanding of reaction processes and improve reaction design. Herein, an efficient, general and flexible fragmentation-based approach was proposed to evaluate local distortion energies for various chemical and biological molecules, which can be obtained computationally and/or experimentally. Moreover, our distortion analysis is readily applicable to multiple structures from molecular dynamics (or the minimum energy path) as well as can be evaluated by different computational chemistry methods. Our systematic analysis shows that our approach not only aids computational and experimental chemists in visualizing (relative) distortion distributions within molecules (distortion map) and identifies the key distorted pieces, but also offers deeper understanding and insights into structures, reaction mechanisms and dynamics in various chemical and biological systems. Furthermore, our analysis offers indices of local distortion energy, which can potentially serve as a new descriptor for multi-linear regression (MLR) or machine learning (ML) modelling.