Two-level stochastic search of low-energy conformers for molecular spectroscopy: implementation and validation of MM and QM models

摘要

The search for stationary points in the molecular potential energy surfaces (PES) is a problem of increasing relevance in different fields of molecular sciences especially for large, flexible systems characterized by several large-amplitude internal motions leading to shallow minima with comparable energies and separated by small barriers. After structural biology and medicinal chemistry, also high-resolution molecular spectroscopy, which is the focus of our research activity, is nowadays shifting its attention to this kind of molecular systems. In such circumstances, accurate geometrical structures and relative stabilities of all these minima are a mandatory prerequisite for the vis-a-vis comparison between computed and experimental spectra. This task raises, in turn, the problem of the best compromise between accuracy and feasibility. In our opinion, a promising route is offered by a two-level stochastic search in which a relatively inexpensive MM or QM method is used in the initial search, followed by single point energy evaluation at a higher QM level of a relatively large number of low-energy structures in order to select a final short-list of candidates, whose geometries are fully optimized at the higher QM level. Finally, the relative stabilities and properties of the final short-list of energy minima can be computed by a state-of-the-art QM approach. This strategy defines a general two-level search/three-level evaluation approach, which can be finely tuned in terms of the accuracy of the sought results. Setup of the procedure, interface with a general purpose electronic structure code and validation of the most effective low-level methods for some representative molecular systems (three already well characterized and one new) ended up with a general, robust and user-friendly tool, which can be easily used and extended also by non-specialists to aid experimental spectroscopic studies.