AMBER Archive (2000)

Subject: Small Molecule Paramaterisation

From: Simon Cross (pcxsc_at_nottingham.ac.uk)
Date: Fri Dec 01 2000 - 04:46:28 CST

Hi, I'm trying to parameterise a diphenylphosphonate molecule PhP(O*2)OPh
for use in an MD simulation in Amber5. The phosphonate is an anion and
therefore has an axis of symmetry. I've run Gaussian98 with 3-21G* basis
set (due to lack of computer time) to generate a fully optimised structure
and used these bond lengths and angles in my parameter file. For the
dihedrals I've fixed the chosen bond (3 in total) at 0, 15, 30..180
degrees and optimised the rest of the molecule which has given me what a
believe to be fairly accurate torsional information. For the charges I ran
G98 with 6-31G* basis set (to be consistent with amber forcefield) and
used RESP to fit the charges.

I am now having problems trying to get amber to reproduce these
data. Setting the new dihedral force constants to zero (to see nonbonding
interactions) I've run a one step minimisation on each of the 0,15,30..180
degree dihedral structures and plotted the energies. For the P-Ph bond
there is a flat line (as expected) except at 2 torsion angles, 30 and
150. These correspond to steric clashes between the hydrogen atoms on the
phenyl ring, and the 2 phosphonate oxygens. I would also expect
this. However, the problem seems to be this:

1) There is no gradual increase - all other points are approx. zero and

2) The magnitude of the energy is about 135 kcal/mol!

I then set the newly paramaterised angle force constants to zero as there
seemed to be some contribution in the amber output file, and this just
reproduced the same 'curves' but at a slightly lesser energy, therefore
all relative values equal to the first 'curve'.

The 135 kcal/mol energy is specified as being in the VderW term in the
amber output files, but there is no clash as far as Gaussian98 is
concerned.

I was wondering whether the charges on the two phosphonate oxygens (about
-0.8 each, P is about +1, OS is about -0.5; total = -1) being relatively
large are causing these problems.

I should say that I have introduced a new atom type for the P, called PN
and borrowed any parameters I could not get from Gaussian from the Parm98
phosphate P.

If anyone can help me I would be grateful, as it seems that this
forcefield does not seem to reproduce the theoretical data, which is
worrying! I will gladly summarise responses.

Cheers,

-----------------------------------------

Simon Cross
School of Chemistry
University of Nottingham
tel. 0115 9514193
Email: pcxsc_at_nottingham.ac.uk