Dear amber users,
To my mind, on such systems, one problem when fitting charges from an
electrostatic potential is the fitting itself and not only the shape of the
electrostatic potential.
The problem with CCl4 may come from the fact that the central C atom is
"hidden" from the electrostatic potential surface due to the large size of
Cl atoms. The determination of the point charge on the carbon atom is
therefore quite difficult to obtain from the ESP fitting and may highly vary
depending on 1) the fitting method (RESP vs native Gaussian or native
jaguar) 2) the shape of the electrostatic potential (less crucial to my
mind) and 3) the basis set used (of course!).
But the more important task is to represent the electrostatic interaction of
the molecule with other molecules, not to have a unique charge distribution,
i.e. to verify that the point charges nicely represent the electrostatic
interaction on the ESP grid, staying homogenous with the other charges used
(amber calculated, TIP3P, etc.)
Philippe Guilbaud
Received on Thu Jul 05 2001 - 02:09:07 PDT
