Re: CCL:summary AM1 vs PM3
To echo Gabor Csonka's comments on PM3's problem of unrealistic estimation
of H-H interaction, here are some additional comments on the problem and
fix-up solution.
Original PM3 Bug:
----------------
In original PM3 model, the non-hydrogen interaction are known to be
problematic which is too attractive and gives rise to unusual stability
for crowded arrangements.
PM3 Expt
=== ===
methyl cyclohexane: (axial) <-> (equatorial) 1.1 1.8
t-butyl cyclohexane: (axial) <-> (equatorial) 1.1 5.4
Modification:
------------
This can be corrected by introducing a repulsive term between non-bonded
hydrogens into the PM3 energy expression, which allows it rapidly approaches
to zero with increasing H--H distance and becomes insignificant for
distances greater than about 2 angstrom.
PM3 PM3 Expt
(modified)
=== === ===
methyl cyclohexane: (axial) <-> (equatorial) 1.1 1.3 1.8
t-butyl cyclohexane: (axial) <-> (equatorial) 1.1 4.8 5.4
This result is little if any change for systems which are not crowded,
but significant change for molecules in which hydorgens are crowded t
together.
This modification is within Spartan semiempirical module. The correction
has been introduced as the default for PM3(tm) (the PM3 method dealing
with d-transition metals) and as an option for PM3.
--Wayne
--
+---------------------------------------------------------------------+
| Wayne Huang, Ph.D. | 18401 Von Karman, Suite 370 |
| Computational Chemist | Irvine, California 92715 |
| Wavefunction, Inc. | (714)955-2120 Fax: (714)955-2118 |
| huang "-at-" wavefun.com | World Wide Web: http://wavefun.com |
+---------------------------------------------------------------------+