CCL: translational entropy in solvent

 Sent to CCL by: "Raphael  Ribeiro" []
 Dear Pierre,
 It is quite obvious that the translational entropy changes while going from the
 ideal gas phase to a solvated phase. What is also obvious in my opinion is that
 the translational/rotational/vibrational components of free energy (and also
 entropy) calculated using implicit solvents are not going to reproduce the
 quantities from the real system in a good way. I've said earlier that one of the
 biggest failures of the implicit solvent models is that they do not represent in
 a true way the phase space of the system. The density of states calculated using
 the implicit model for solvation does not represent the reality at all, and that
 is why it is not a good idea to calculate components of the free energy with
 PCM, COSMO, etc.
 Also, in most of the implicit models there are empirical parameters (in the
 non-electrostatic contributions) which are not physical. Implicit solvation
 models were created to give nice solvation free energies (as a whole) and that
 is why they are in most of the times not suitable to reproduce other physical
 properties in a good way.
 For your association/dissociation problem I recommend you the following article:
 In this article the authors create a model for calculating the loss of
 translational entropy in associations accurately. They use molecular dynamics.
 The last thing to say is that the free energy of solvation generated by implicit
 solvent models already include all of the components of the free energy (free
 energy of solvation  = free energy of the solvated phase system - free energy of
 the gas phase system) and as Andreas Klamt already remarked, the addition of
 translational, rotational and vibrational free energy contributions in solutions
 will lead to double counting of effects already implicitly taken into account in
 solvation models and are not recommended at all.
 Raphael Ribeiro