next up previous contents
Next: Gas phase computations Up: Method Previous: Density Functional and Molecular   Contents

Free energy calculation

The reaction equilibrium constant and the transition state theory estimate of the reaction rate of a chemical reaction are determined by the free-energy profile along the reaction path. The free energy reaction barrier is the reversible work necessary to bring the system from the stable state of reactants (RS) to the transition state (TS). We will characterize these states by some order parameter, a reaction coordinate $\xi $, which is a function of the positions of the nuclei. As the reversible work is independent of the path, the precise choice of the reaction coordinate is not crucial, but a practical and physically appealing choice is one that incorporates the asymmetric stretch vibration along the Cl-C-Cl axis of the transition state complex CH$_3$Cl$_2$. Given a choice for the reaction coordinate, wa can obtain the free energy change $\Delta A(\xi)$ along the reaction path, using the technique of thermodynamic integration (see e.g. ref frenkel_smit):


\begin{displaymath}
\Delta A(\xi) = \int_{\mathrm {\xi_{\mathrm{RS}}}}^{\xi} \le...
...rac{\partial H}{\partial \xi}\right>_{\xi^\prime} d \xi^\prime
\end{displaymath} (47)

Here $H$ is the Hamiltonian of the system of nuclei, defined as the sum of the kinetic and (Born-Oppenheimer) potential energy, and minus the integrand - $\left<\partial H/\partial
\xi\right>$ is usually called the mean force. The brackets denote an ensemble average and the subscript indicates that the integrand is evaluated at the point $\xi({\bf r})=\xi^\prime$ along the reaction coordinate, where $\bf r$ denotes the total of nuclear coordinates.

The S$_\mathrm{N}$2 reaction studied in the present work is an activated process, which implies that the barrier is too high for the reaction to take place spontaneously within the time scale accessible to an MD simulation. The probability to find the system close to the transition state is very small, and the reaction is therefore a rare event. To nevertheless be able to estimate the mean force at the reaction coordinate values of low probability, the method of constraint can be used, where the dynamics of the system is performed with the reaction coordinate fixed at a specified value $\xi^\prime$. The theoretical framework of the study of activated processes in an MD simulation has been established some time ago in ref chandler78 and also ref CaCi89, which provided a microscopic expression for the mean force. An approximate version of this method [72,73], only valid for reaction paths controlled by special classes of constraints, has been successfully used in quite a number of simulations of reactions.[74,75,54] More recently, generally applicable expressions for the mean force have been outlined [17,76], which include the explicit terms to correct for the bias introduced in the ensemble by applying the constraint (see also the appendix).

Figure 3.1: The reaction coordinate (equation 3.2) is a function of the positions of the attacking chloride Cl$^\prime $, the carbon atom C and the leaving chloride Cl. One characteristic of the chosen reaction coordinate is that for a particular value of the coordinate, the carbon is allowed to move only in a plane perpendicular to the line through the chlorine atoms.
The reaction coordinate used in the present study is defined as the normalized projection of the C-Cl bond $R_{\mathrm{CCl}}$ on the Cl$^{\prime}$-Cl bond $R_{\mathrm{Cl^{\prime}Cl}}$ (see figure 3.1):
\begin{displaymath}
\xi = \frac{R_{\mathrm{CCl}} \cos(\angle_{\mathrm{CClCl^{\prime}}})}
{R_{\mathrm{ClCl'}}}.
\end{displaymath} (48)

Here $\angle_{\mathrm{CClCl^{\prime}}}$ is the angle between the C-Cl and Cl$^{\prime}$-Cl bonds. In the following we will refer to Cl$^{\prime}$ and Cl as the attacking and leaving chloride, respectively. Since the reaction is symmetric, the transition state is at a point where both C-Cl distances are equal. At this transition state the value of the reaction coordinate (equation 3.2) is $\xi =0.5$, whereas it approaches zero for separated reactants and one for separated products.


next up previous contents
Next: Gas phase computations Up: Method Previous: Density Functional and Molecular   Contents
Bernd Ensing 2003-06-13