Similar to:

I would like to ask: What are the different types of solvation models?


1 Answer 1



Name: COSMO = Conductor-like Screening Model.


  • Implicit
  • Dielectric continuum model

First appearance:


Like other continuum solvent models, the solvent is treated as a continuum with a permitivity of $\varepsilon$. The energy of the electrostatic interaction between the solute and the solvent is determined based on the charge distribution of the solute molecule (which is determined by whatever electronic structure method you wish, independent of whether or not you're using the COSMO model), and by the effective charges $q$ on the solvent, which are given based on the charges $q^*$ on the surface segments as follows:

\begin{align} \tag{1} q &= \frac{\varepsilon -1}{\varepsilon + 1/2}q^* ~~~, &\textrm{for neutral solvent molecules}, \\ q&= \frac{\varepsilon -1}{\varepsilon }q^* ~~~, &\textrm{for ionic solvent molecules}. \tag{2} \end{align}


  • OpenMOLCAS (open source)
  • Gaussian (apparently uses Eq. 2 even for neutral solvent molecules, erroneously)
  • NWChem (open source)
  • MOPAC (open source until MOPAC 7.1 in 2006, expensive for MOPAC 2007, 2009, etc, then again open source since 2022)
  • Q-Chem
  • Many others

Computational cost:

Due to the simplicity of Eqn. 1 and 2, it is not very computationally costly, and can be used for large and/or irregularly formed molecular structures, whereas the PCM (polarizable continuum model) attempts to use "exact" dielectric boundary conditions instead of the approximations given in Eqs. 1-2, and therefore can only be used for small solvent molecules with spherical, ellipsoidal, or quasi-spherical shapes.


  • A variant of the PCM model was compared to COSMO in this paper and the difference in the two models, was found to be small compared to the deviation from experimental solvation data, so a conclusion that can perhaps be drawn from that paper was that you might as well use the COSMO model (as long as your situation is similar enough to the situation in that study).

  • Like all continuum solvent models, the effects of things like hydrogen bonding and reorientation are neglected, and this is considered to be a bigger issue than the choice of which continuum solvent model is used.


Different implementations may use a slightly different formula (see the part in parentheses after Gaussian was mentioned earlier in this answer, for example). Each implementation might also have its own unique way of treating the geometry of the solvent molecules (e.g. radii of the effective spheres, cavity construction, surface segments, etc.), so unfortunately the numerical result from a COSMO calculation done with one software might not be the same as with another software (this might be a problem for most solvent models though).

Related models / extensions:


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .