When DMol3 perform Energy or Geometry Optimization calculations with using the COSMO solvent simulation, the Sigma profile is automatically generated. But when I performed the calculations for simple molecules, like water or formaldehyde, the obtained sigma profiles seems to be strongly "sqeezed" vertically, i.e. the argument (induced solvent surface charge density) values at the local maxima are much less than found in literature. E.g. the "right" maximum for H2O should appear for the argument value above 1 e/nm^2 (i.e. 0.01 e/Ang^2, in the units used in MS6.0), indicating the strong hydrogen bonding, while the MS6.0 shows value of 0.004 e/Ang^2. The result only weekly depends on functional and basis set used.
Below, the sigma-profile of water molecule shown by MS6.0.
Pawel