Explicit solvent, modelling

C. Explicit Solvent Models and the Importance of Balancing the External Interactions... 

It is well known that a solvent can canse dramatic changes in rates and even mechanisms of chemical reactions. Modem theoretical chemistry makes it possible to incorporate solvent effects into calcnlations of the potential energy surface in the framework of the continnnm and explicit solvent models. In the former, a solvent is represented by a homogeneous medium with a bulk dielectric constant. The second model reflects specific molecule-solvent interactions. Finally, calculations of the potential energy surface in the presence or absence of solvents can be performed at various theory levels that have been considered in detail by Zieger and Autschbach [10]. 

In between the implicit and explicit solvent models, there are mixed models, such as the solvation shell approximation.67-69 This model describes explicitly only the first solvation shell molecules and treats as implicit the solvent region beyond the first solvation shell. Such treatment both provides the information about the solvent structure near the solute and allows for faster computation. 

In the next section, QMSTAT is thoroughly presented and explained. As will be seen, we will return to the three aspects of explicit solvent models presented above. We continue with some previously published representative results obtained with QMSTAT. The results are mainly meant to illustrate features of the model other perspectives on the result are omitted and the reader is referred to the original... 

Although many satisfactory VCD studies based on the gas phase simulations have been reported, it may be necessary to account for solvent effects in order to achieve conclusive AC assignments. Currently, there are two approaches to take solvent effects into account. One of them is the implicit solvent model, which treats a solvent as a continuum dielectric environment and does not consider the explicit intermolecular interactions between chiral solute and solvent molecules. The two most used computational methods for the implicit solvent model are the polarizable continuum model (PCM) [93-95] and the conductor-like screening model (COSMO) [96, 97]. In this treatment, geometry optimizations and harmonic frequency calculations are repeated with the inclusion of PCM or COSMO for all the conformers found. Changes in the conformational structures, the relative energies of conformers, and the harmonic frequencies, as well as in the VA and VCD intensities have been reported with the inclusion of the implicit solvent model. The second approach is called the explicit solvent model, which takes the explicit intermolecular interactions into account. The applications of these two approaches, in particular the latter one will be further discussed in Sect. 4.2. 

Sun et al. found significant differences between the implicit and the explicit solvent models. However, lacking further information they were not able to conclude which of the models gave more accurate results. 

Shelley, J.C., and Patey, G.N. Phase behavior of ionic solutions Comparison of the primitive and explicit solvent models. J. Chem. Phys., 1999, 110, p. 1633-7. 

Along with deciding whether to use implicit or implicit-explicit solvent models, a specific level of theory and basis set must be used for the calculation of the change in free energy of solvation. Similar to the gas-phase free energy, there are a variety of methods and it can be difficult to determine what combination is the most accurate. Further discussion can be found in Section 4. 

It should be further mentioned that the choice of the ensemble also depends on whether or not the solvent is explicitly modeled. In particular, care has to be exerted concerning the number of intensive variables of the ensemble. This can be understood considering the Gibbs phase rule for interfaces [93] f = 2 + c - p, with / being the number of independent intensive variables needed to describe the interface, p the number of different phases in the interface, and c denoting the number of different components. A lipid bilayer comprising one sort of lipid embedded into an implicit solvent corresponds to a one-component system c = 1, in one-phase state so that p = 1 hence / = 2. On the other hand, a model with explicit solvent yields c = 2, p = 1, and / = 3. Thus, implicit solvent models can be simulated within nXT ensemble while for explicit-solvent models an additional intensive quantity has to be controlled, e.g., the nPzzXT ensemble is appropriate. 

Lower computational costs tor many molecular systems, and better scaling on parallel machines. The effective cost reduction may be particularly signiticant it one takes into account the improved sampling in contrast to explicit solvent models, solvent viscosity that slows down conformational transitions can be turned off completely within implicit representations. 

---