Questions and problems?

Fluorine parameters

Two people have responded to this, Craig Gough in a series of messages to Bill Ross and Oyvind Edvardsen in a response to the AMBER Mail Reflector.

From gough@cgl.ucsf.EDU Wed Dec  2 14:29:11 1992
Date: Wed, 2 Dec 92 14:29:09 -0800
From: gough@cgl.ucsf.EDU (Craig Gough)
To: gough@cgl.ucsf.EDU, ross@cgl.ucsf.EDU
Subject: Re: flourine params
Yes I have some fluorine parameters that are probably worth putting in the force field. I derived an R* of 1.75 Angstroms, and an epsilon of 0.061 kcal/mole for the fluorine atom, by tweaking the parameters until a simulation of pure tetrafluoromethane produced the correct molar volume and enthalpy of vaporization (with 5% of experimental values).

The bonding parameters (both stretching and bending) were taken from MM2 and used without modification (except that K sub R differs by a factor of 2; i.e., I think that MM2 defines the stretching potential as 1/2 K (R - R0)**2, and AMBER defines it as K (R - R0)**2. (Same for the bond bending potential).

These parameters are (for AMBER):

Bond                 K (kcal/mol-Angstrom**2           R0(Angstroms)
CT-F                     367                            1.38
of CF4

CT-F                     367                            1.332
of CHF3
I have found free energy calculations in solution to be somewhat dependent on the exact CT-F bond length used, if the number of fluorines is changing, and would recommend using the correct experimental bond length for the system under consideration, if known, or at least the experimental bond length for a system with the same number of fluorines attached to a give carbon. Thus, although I once used 1.38 Angstroms as a default bond length for all CT-F bonds, I would recommend, for systems with one, two or three fluorines attached to a given carbon, the following bond lengths:
one fluorine to a given carbon:   R0 = 1.38 A
two fluories "  "   "     "       R0 = 1.35 A
three "      "   "  "     "       R0 = 1.332 A

Four "       "   "   "    "       R0 = 1.323 A
(The latter disagrees with the bond length used in the cf4 calculations, but the exact bondlength did not seem to be important there).

The bending parameters (from MM2):

Angle          K0(kca;al/mol-rad**2)           theta sub 0  (degrees)
F-CT-HC             35.0                      109.5
F-CT-F              77.0                      109.1 
The reference to be cited for the non-bonded parameters is: Craig A. Gough, Stephen, E. DeBolt, and Peter A. Kollman; Journal of Computational Chemistry, Volume 13, No.8, Page 963-970 (1992).

This article describes in detail the parameterization process.

From gough@cgl.ucsf.EDU Wed Dec  2 14:36:10 1992
Oh, I forgot: The torsional parameters I have used for fluorine were not arrived at in a very rigorous fashion. I simply used the analogous parameters already in the force field for oxygen; i.e., an O-C-C-F torsion was assigned the same parameters as an O-C-C-O torsion, for example. The logic for this is that both fluorine and oxygen are electronegative, and will be approximately equal in their charge- accepting ability. This affects the torsional energies due to electron-delocalization, according to the theory of Brunck and Weinhold (I forget the reference right now).

Also, the references for the experimental bond lengths are:

Harmony, M.D., et al., Journal of Physical and Chemical Reference Data, Volume 8, page 619 (1979)
Baird, N.C., Canadian Journal of Chemistry, Volume 61, page 1567 (1983).

From gough@cgl.ucsf.EDU Wed Dec  2 14:55:24 1992
Here is the reference discussing torsions; the reasoning in the ref led me to use the fluorine torsional parameters that I did (i.e., the same parameters as for oxygen).
Brunck, T.K.; Weinhold, F.; Journal of the American Chemical Society, Volume 101, page 1700 (1979).
From: (Oyvind Edvardsen Computer Graphics Lab. IMB Univ. Tromsoe)
Fluorine parameters may be found in:
Gregory & Gerig (1989) J.Comput.Chem. 10:711-717.
Gough et al (1992) J.Comput.Chem. 13:963-970.
Veenstra et al (1992) J.Comput.Chem. 13:971-978.