Dear Colleagues,

I'm afraid this is a bit of a "newbie" question, but here goes. Are
there any standard or useful tricks for optimizing the geometry of
molecules that contain several relatively uncoupled methyl or NH2
groups? I am trying to study molecules like B(NH2)3, and having a
terrible time in getting geometries that have all positive frequencies.
I have tried making the convergence criteria very tight or only somewhat
tight; I've used small basis sets (DZ with frozen core) to big basis
sets (TZ2P, all electron); I've used internal and cartesian coordinates,
etc etc. ADF seems to have no problem "converging", but then when I run
a frequency calculation on the converged geometry, I inevitably get some
negative frequencies. Based on the normal modes of the negative
frequencies, the trouble always comes from the NH2 torsion modes (or CH3
torsions etc) which is no surprise given how low energy these are. I
also did a linear transit calculation for B(NH2)3, for variable NH2
dihedral angles, and found a very very shallow minimum where I believe
it should be based on experimental data (namely, just off from planar),
but again at this geometry I get negative frequencies.

Any and all help would be greatly appreciated.

Thanks,
Joe Zwanziger

Josef W. Zwanziger tel: (902) 494-1960
Professor of Chemistry and fax: (902) 494-1867
  Canada Research Chair in net: jzwanzig_at_dal.ca
  NMR Studies of Materials web: http://jwz.chem.dal.ca
Dept. of Chemistry
Dalhousie University
Halifax, NS
B3H 4J3 CANADA
Received on 2004-02-24 20:31:54