AMBER Archive (2009)

Subject: Re: [AMBER] Different optimized structures with HF and B3LYP

From: Karl Kirschner (kkirsch_at_scai.fraunhofer.de)
Date: Fri Oct 02 2009 - 03:07:08 CDT

Hello,

        Yes, it is possible to obtain different conformations when you start from the
same input conformation and optimize at different QM levels. I see this
happen sometimes in my force field parametrization work.

        One way to determine which conformation is more reliable is to increase the
theory level and the basis set, until you obtain convergence of conformation
across several QM levels. However, this only increases the reliability that
you have modeled the experimentally relevent conformation, but it does not
tell you if it is the experimentally correct one. For that, you really have
to compare to the experimental structure if it is available, otherwise you
are predicting the experimental conformation.

        I would guess, in your case, that the basis set is too small to correctly
model the orbital space of the molecule, especially since it is an anion.

        The molecular mechanics results will depend on how the force field was
parametrized.

Cheers,
Karl

P.S. Perhaps this question is more appropriate for the CCL list serve.

On Friday 02 October 2009 09:50, Sanghwa Han wrote:
> Dear Colleagues,
>
> I am trying to obtain a geometry optimized structure of C6H5-O-SO3- (singly
> deprotonated form of phenyl hydrogen sulfate).
>
> When optimized with HF/6-31G*, the dihedral angle of C-C-O-S was close to
> 90 degrees but with B3LYP/6-31G* or MP2/6-31G* it was close to 0 degrees.
>
> In addition, the C-C-O-S torsional profile had a minimum at 90 degrees with
> HF/6-31G* but at 0 degrees with B3LYP/6-31G* or MP2/6-31G*.
>
> Molecular mechanics minimization (for example, that in PRODRG site) yielded
> a structure with 0 degrees.
>
> Is it possible to obtain completely different optimized structures with
> different QM methods?
>
> _______________________________________________
> AMBER mailing list
> AMBER_at_ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber

_______________________________________________
AMBER mailing list
AMBER_at_ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber