Re: G98 ONIOM and TM

[Steven Feldgus <sfeldgus -8 at 8- students.wisc.edu>]

We've been doing a lot of ONIOM(B3LYP:HF:UFF) jobs on a transition metal
 complex, and from our experience, we've found that the Gaussian UFF
 routines are not tremendously robust. I sent Gaussian several emails in
 late 1998 telling them of what we found, but never got any response. We
 needed to add several UFF atom types into the UFF code (in the utilam.f and
 utilnz.f files) in order for UFF to handle square planar rhodium complexes
 and coordinated olefins. The potential for linear bonds, like -CN, was also
 wrong (there was a maximum at 180). More recently, we've found that the
 ordering of the first few atoms can make a HUGE difference in whether UFF
 has a prayer of handling the molecule. Virtually everything in our molecule
 is within three bonds of rhodium, but we zeroed out what torsions we could.
 Having proper connectivity was critical for getting our systems to work.
 Steve Feldgus
 Dept. of Chemistry
 University of Wisconsin-Madison