Polarizable continuum model limits

The purpose of this chapter is to present an overview of the computational methods that are utilized to study solvation phenomena in NMR spectroscopy. We limit the review to first-principle (ab initio) calculations, and concentrate on the most widespread solvation model the polarizable continuum model (PCM), which has been largely described in the previous chapter of this book. 

Organometallic systems such as porphyrines have been investigated because of the possibility to fine tune their response by functionalization[105-107]. Systems of increased the dimensionality have been of particular interest [108-111], Concomitant to the large effort to establish useful structure-to-properties relationships, considerable effort has now been put to investigate the environmental effects on TPA[112-114], For example, the solvent effect has been studied for a small linear push-pull chromophore using a self-consistent reaction field (homogeneous solvation) method employing a spherical cavity and an internal force field (IFF) method[l 12] in another study the polarizable continuum model has been employed to calculate the relevant quantities to obtain the TPA cross-section in the limit of a two-state model[113] Woo et al. made a critical study of experimental comparison of TPA cross-sections in different solvents[114]. 

Moving now to QM/continuum approaches, we shall limit our exposition to the so-called apparent surface charges (ASC) version of such approaches, and in particular to the family known with the acronym PCM (polarizable continuum model) [11], In this family of methods, the reaction potential Vcont defined in Eq. (1-2) has a form completely equivalent to the Hel part of the Z/qm/mm operator defined in Eq. (1-4), namely ... 

The most serious limitation remaining after modifying the reaction field method as mentioned above is the neglect of solute polarizability. The reaction field that acts back on the solute will affect its charge distribution as well as the cavity shape as the equipotential surface changes. To solve this problem while still using the polarizable continuum model (PCM) for the solvent, one has to calculate the surface charges on the solute by quantum chemical methods and represent their interaction with the solvent continuum as in classical electrostatics. The Hamiltonian of the system thus is written as the sum of the Hamilton operator for the isolated solute molecule and its interaction with the macroscopic... 

In the IPCM calculations, the molecule is contained inside a cavity within the polarizable continuum, the size of which is determined by a suitable computed isodensity surface. The size of this cavity corresponds to the molecular volume allowing a simple, yet effective evaluation of the molecular activation volume, which is not based on semi-empirical models, but also does not allow a direct comparison with experimental data as the second solvation sphere is almost completely absent. The volume difference between the precursor complex Be(H20)4(H20)]2+ and the transition structure [Be(H20)5]2+, viz., —4.5A3, represents the activation volume of the reaction. This value can be compared with the value of —6.1 A3 calculated for the corresponding water exchange reaction around Li+, for which we concluded the operation of a limiting associative mechanism. In the present case, both the nature of [Be(H20)5]2+ and the activation volume clearly indicate the operation of an associative interchange mechanism (156). 

Much like the RISM method, the LD approach is intermediate between a continuum model and an explicit model. In the limit of an infinite dipole density, the uniform continuum model is recovered, but with a density equivalent to, say, the density of water molecules in liquid water, some character of the explicit solvent is present as well, since the magnitude of the dipoles and their polarizability are chosen to mimic the particular solvent (Papazyan and Warshel 1997). Since the QM/MM interaction in this case is purely electrostatic, other non-bonded interaction terms must be included in order to compute, say, solvation free energies. When the same surface-tension approach as that used in many continuum models is adopted (Section 11.3.2), the resulting solvation free energies are as accurate as those from pure continuum models (Florian and Warshel 1997). Unlike atomistic models, however, the use of a fixed grid does not permit any real information about solvent structure to be obtained, and indeed the fixed grid introduces issues of how best to place the solute into the grid, where to draw the solute boundary, etc. These latter limitations have curtailed the application of the LD model. 

Our conq>utadonal strategy for the study of biochemical systems in solution apphes a solvent model belonging to the family of polarizable continuum noodels (PCM), introduced in 1981 [101] and continuously updated and extended [102-113]. Recently, very refined and effective PCM algorithms have been implemented in widely used ab initio codes, so that molecules in solution can now be studied at any level of the theory (from molecular mechanics to Hartree-Fock, MP2 and configuration interaction) with a very limited computational burden with respect to the corresponding calculations in vacuo. 

The above examples illustrate that continuum models such as the Kirkwood model are reasonably successful in describing the static permittivity, provided one has an independent means of estimating the correlation parameter Unfortunately, these estimates are available for only a few polar solvents, so that gK must be considered an independent parameter. The version of Kirkwood s theory presented here only considers orientational polarization. When distortional polarization, that is, the effect of molecular polarizability, is included, interpretation of experimental results is less clear. Since the approach taken here involves continuum concepts, it is necessarily limited. In the following section, a simple model based on a molecular description of a polar liquid is presented. 

Continuum electrostatic models [72,108-113] are presently most developed and commonly nsed for the evaluation of the solvation energies in proteins however, they carry a nnmber of limitations and uncertainties, which cannot be avoided unless the microscopic interactions of the quantum subsystem and the protein are taken into account [114], For example, it is not clear which dielectric constant of the polarizable water cavities one should use in such calculations even the usually assumed dielectric constant of a dry protein (typically assumed as 4 [99,115,116]) is not that well defined—many studies indicate that the effective dielectric of the protein is much higher [114,117-119]— primarily due to internal water [120], and partially due to protein (nonlinear) charge relaxation. Proteins are also inhomogeneous media. It is understood that only microscopic simulations should eventually provide a correct picture and remove the inherent uncertainty of phenomenological approach [71,114,115,121-132]. Despite the drawbacks, the continuum models provide most computationally efficient approach for the treatment of the protein electrostatics, which make possible large-scale investigation of the enzyme properties, such as CcO. 

Although continuum models have been quite sueeessful in assessing macroscopic optical response, they have intrinsie limitations for probing microscopie optical properties, such as molecular polarizability and photoconductivity. The limitations stem from the faet that eontinuum eleetrodynamies, as applied to metal nanostructures, are intended to deseribe the eolleetive motions of the electrons and are thus not applieable to any physieal phenomenon that occurs at small length scales (typieally a few nanometers for typieal eondensed-phase systems). For small length seales, many-body theories need to be applied to account for the quantum eharaeteristies of individual eleetronie transitions, for example, light absorption by an organie sensitizer and subsequent electron injection to semiconductor layer. 

Figure 3.31 As (due to orientational response of aqueous solvent) versus e, calculated for ET in a large binuclear transition metal complex (D (Ru2+/3+) and A (Co2+/3+) sites bridged by a tetraproline moiety) molecular-level results obtained from a nonlocal polarization response theory (NRFT, solid lines) continuum results are given by dashed lines, referring to numerical solution of the Poisson equation with vdW (cont./vdW) and SAS (cont./SAS) cavities, or as the limit of the NRFT results when the full k-dependent structure factor (5(k)) is replaced by 5(0) 5(k) for bulk water was obtained from a fluid model based on polarizable dipolar spheres (s = 1.8 refers to ambient water (square)). For an alternative model based on TIP3 water (where, nominally, 6 = ), ambient water corresponds to the diamond. (Reprinted from A. A. Milishuk and D. V. Matyushov, Chem Phys., 324, 172. Copyright (2006), with permission from Elsevier).

From Table 3-9 one learns that the continuum-only results are only for water in reasonable agreement with experiment. In contrast, the discrete solvent model leads, even in this very limited version, to shifts that compare well with experiment. Notice also that going from water to MeCN and CCI4, the dispersion is of increasing importance MeCN has an appreciable dipole moment but is also more polarizable than water, especially along the CN triple bond, while for CCI4 the polarizability is the only parameter of importance. 

The obvious limitations of the continuum representation of the solvent necessitated the development of microscopic models of the surroundings. Whereas for liquid phases this task is not trivial at all, for structurally well-characterized environments, like proteins [190, 207] or crystals [208] it is possible to calculate the reaction field from the polarizability distribution [209]. Assuming the existence of strongly bound solvent... 

Solvation effects have been incorporated into the calculations of anionic proton transfer potentials in a number of ways. The simplest is the microsolvation model where a few solvent molecules are included to form a supermolecular system that is directly characterized by quantum mechanical calculations. This has the advantage of high accuracy, but is limited to small systems. Moreover, one must assume that a limited number of solvent molecules can adequately model a tme solution. A more realistic approach is to explicitly describe the inner solvation shell with quantum calculations and then treat the outer solvation sphere and bulk solvent as a continuum (infinite polarizable dielectric medium). In this way, the specific interactions can be treated by high-level calculations, but the effect of the bulk solvent and its dielectric is not neglected. An ej tension of this approach is to characterize the reaction partners by quantum mechanics and then treat the solvent with a molecular mechanics approach (hybrid quantum mechanics/molecular mechanics QM/MM). The low-cost of the molecular mechanics treatment allows the solvent to be involved in molecular dynamics simulations and consequently free energies can be calculated. In more recent work, solvent also has been treated with a frozen or constrained density functional theory approach. ... 

---