Linear Response Properties of Solvated Systems: A Computational Study

TL;DR

This study evaluates various computational methods to accurately predict the linear polarizabilities of organic molecules in solution, considering solvent effects through multiscale approaches and electron correlation.

Contribution

It introduces a combined multiscale computational framework integrating QM/MM, quantum embedding, and EOM-CC methods for response property calculations in solvated systems.

Findings

Polarizabilities vary with solvent and method used.

Multiscale approaches improve agreement with experimental data.

Electron correlation effects are significant for accurate predictions.

Abstract

We present a computational study of static and dynamic linear polarizabilities in solution. We use different theoretical approaches to describe solvent effects, ranging from quantum mechanics/molecular mechanics (QM/MM) to quantum embedding approaches. In particular, we consider non-polarizable and polarizable QM/MM methods, the latter based on the fluctuating charge (FQ) force field. In addition, we use a quantum embedding method defined in the context of multilevel Hartree-Fock (MLHF), where the system is divided into active and inactive regions, and combine it with a third layer described by means of the FQ model. The multiscale approaches are then used as reference wave functions for equation-of-motion coupled cluster (EOM-CC) response properties, allowing for the account of electron correlation. The developed models are applied to the calculation of linear response properties of two organic moieties -- namely, para-nitroaniline and benzonitrile -- in non-aqueous solvents -- 1,4-dioxane, acetonitrile, and tetrahydrofuran. The computed polarizabilities are then discussed in terms of the physico-chemical solute-solvent interactions described by each method (electrostatic, polarization and Pauli repulsion), and finally compared with the available experimental references.