Solvation dynamics linear response

III. Solvation dynamics within the linear response approximation 213... 

T. Fonseca and B. M. Ladanyi, Breakdown of linear response for solvation dynamics in methanol, J. Phys. Chem., 95 (1991) 2116-19. 

K. Ando and S. Kato, Dielectric relaxation dynamics of water and methanol solutions associated with the ionization of /V,/V-dimcltiylanilinc theoretical analyses, J. Chem. Phys., 95 (1991) 5966-82 D. K. Phelps, M. J. Weaver and B. M. Ladanyi, Solvent dynamic effects in electron transfer molecular dynamics simulations of reactions in methanol, Chem. Phys., 176 (1993) 575-88 M. S. Skaf and B. M. Ladanyi, Molecular dynamics simulation of solvation dynamics in methanol-water mixtures, J. Phys. Chem., 100 (1996) 18258-68 D. Aheme, V. Tran and B. J. Schwartz, Nonlinear, nonpolar solvation dynamics in water the roles of elec-trostriction and solvent translation in the breakdown of linear response, J. Phys. Chem. B, 104 (2000) 5382-94. 

Figure 43. Solvation dynamics from MD simulations for isomer 2. (a) The linear-response calculated time-resolved Stokes shifts for indole-protein, indole-water, and their sum. (b) Direct nonequilibrium simulations of the time-resolved Stokes shifts for indole-water, indole-protein, and their sum. Note the lack of slow component in indole-water relaxation in both (a) and (b), which is opposite to isomer 1 in Fig. 42. Also shown is the indole-water (within 5 A of indole) with coupled long-time negative solvation, (c) Relaxation between indole-lys79 and indole-glu4. The interaction energy changes from these two residues nearly cancel each other, (d) The distance changes between the indole and two charged residues, but both residues move away from the indole ring.

The surrogate Hamiltonian is expressed in terms of renormalized solute-solvent interactions, a feature that leads to a simple and natural linear response description of the solvent dynamics in the vicinity of the solute. In addition to the measurable solvation time correlation function (tcf), we can also calculate observables needed to elucidate the detailed mechanism of solvation response, such as the evolution of the solvent polarization charge density around the solute. 

While there is no unique criterion for choosing 4 E, the selection must lead to an accurate theory of solvation dynamics without invoking two-time many-point correlation functions. We have found that this goal can be achieved with a new theory for the nonequilibrium distribution function in which the renormalized solute-solvent interactions enter linearly. In this theory and are chosen such that the renormalized linear response theory accurately describes the essential solute-solvent static correlations that rule the equilibrium solvation both at t = 0 (when solvent is in equilibrium with the initial charge distribution of the solute) and at 1 = oc (when the solvent has reached equilibrium with the new solute charge distribution). ... 

Figure 3 presents a comparison of the non-equilibrium solvent response functions, Eq (1), for both the photoexcitation ("up") and non-adiabatic ("down") transitions (cf. Fig. 2). The two traces are markedly different the inertial component for the downwards transition is faster and accounts for a much larger total percentage of the total solvation response than that following photoexcitation. The solvent molecular motions underlying the upwards dynamics have been explored in detail in previous work, where it was also determined that the solvent response falls within the linear regime. Unfortunately, the relatively small amount of time the electron spends in the excited state prevents the calculation of the equilibrium excited state solvent response function due to poor statistics, leaving the matter of linear response for the downwards S(t) unresolved. Whether the radiationless transition obeys linear response or not, it is clear that the upward and downwards solvation response behave very differently, due in part to the very different equilibrium solvation structures of the ground and excited state species. Interestingly, the downwards S(t), with its much larger inertial component, resembles the aqueous solvation response computed in other simulation studies, and bears a striking similarity to that recently determined in experimental work based on a combination of depolarized Raman and optical Kerr effect data. ...

The nonlinear Smoluchowski-Vlasov equation is calculated to investigate nonlinear effects on solvation dynamics. While a linear response has been assumed for free energy in equilibrium solvent, the equation includes dynamical nonlinear terms. The solvent density function is expanded in terms of spherical harmonics for orientation of solvent molecules, and then only terms for =0 and 1, and m=0 are taken. The calculated results agree qualitatively with that obtained by many molecular dynamics simulations. In the long-term region, solvent relaxation for a change from a neutral solute to a charged one is slower than that obtained by the linearized equation. Further, in the model, the nonlinear terms lessen effects of acceleration by the translational diffusion on solvent relaxation. 

It should be emphasized that this description of solvation as a purely electrostatic process is greatly over-simplified. Short-range interactions exist as well, and the physical exclusion of the solvent from the space occupied by the solute must have its own dynamics. Still, for solvation of ions and dipolar molecules in polar solvents electrostatic solvent-solute and solvent-solvent interactions dominate, and disregarding short-range effects turns out to be a reasonable approximation. Of main concern should be the use of continuum electrostatics to describe a local molecular process and the fact that the tool chosen is a linear response theory. We will come to these points later. 

The continuum dielectric theory used above is a linear response theory, as expressed by the linear relation between the perturbation T> and the response , Eq. (15.1b). Thus, our treatment of solvation dynamics was done within a linear response framework. Linear response theory of solvation dynamics may be cast in a general form that does not depend on the model used for the dielectric environment and can therefore be applied also in molecular (as opposed to continuum) level theories. Here we derive this general formalism. For simplicity we disregard the fast electronic response of the solvent and focus on the observed nuclear dielectric relaxation. 

The nonequilibrium solvation function iS (Z), which is directly observable (e.g. by monitoring dynamic line shifts as in Fig. 15.2), is seen to be equal in the linear response approximation to the time correlation function, C(Z), of equilibrium fluctuations in the solvent response potential at the position of the solute ion. This provides a route for generalizing the continuum dielectric response theory of Section 15.2 and also a convenient numerical tool that we discuss further in the next section. 

One example of the use of linear response theory has been that of Hwang et al. in their studies of an reaction in solution. > o In their work, based on the empirical valence bond (EVB) method discussed earlier, they defined their reaction coordinate Q as the electrostatic contribution to the energy gap between the two valence bond states that are coupled together to create the potential energy surface on which the reaction occurs. Thus, the solvent coordinate is zero at the point where both valence states are solvated equivalently (i.e., at the transition state). Hwang et al. studied the time dependence of this coordinate through both molecular dynamics simulations and through a linear response treatment ... 

This deviation from linearity shows itself also in the solvation dynamics. Figure 4.3.7 shows the linear response functions and the non-equilibrium solvation function, C(t) and S(t), respectively, computed as before, for the di-ether H(CH20CH2)2CH3 solvent. Details of this simulations are given in Ref. 1 lb. If linear response was a valid approximation all the lines in Figure 4.3.7 The two lines for C(t) that correspond to q=0 and q=l, and the two lines for S(t) for the processes q=0—K =l and the process q=l—X =0, would coalesce. The marked differences between these lines shows that linear response theory fails forfliis system. 

Such numerical simulations have played an important role in the development of our understanding of solvation dynamics. For example, they have provided the first indication that simple dielectric continuum models based on Debye and Debey-like dielectric relaxation theories are inadequate on the fast timescales that are experimentally accessible today. It is important to keep in mind that this failure of simple theories is not a failure of linear response theory. Once revised to describe reliably response on short time and length scales, e.g. by using the full k and (O dependent dielectric response function e(k,o , and sufficiently taking into account the solvent structure about the solute, linear response theory accounts for most observations of solvation dynamics in simple polar solvents. 

Numerical simulations have also been instrumental in elucidating the differences between simple and complex solvents in the way they dynamically respond to a newly created charge distribution. The importance of translational motions that change the composition or structure near the solute, the consequent early failure of linear response theory in such systems, and the possible involvement of solvent intramolecular motions in the solvation process were discovered in this way. 

---