- Molecular mechanics is essentially empirical, while methods like PPP, CNDO,
and AM1/PM3 are semiempirical. What are the analogies in PPP etc. to MM
procedures of developing and parameterizing a forcefield? Why are PPP etc.
onlysemiempirical? - What do you think are the advantages and disadvantages of parameterizing
semiempirical methods with data from ab initio calculations rather than from
experiment? Could a SE method parameterized using ab initio calculations
logically be called semiempirical? - There is a kind of contradiction in the Dewar-type methods (AM1, etc.) in that
overlap integrals are calculated and used to help evaluate the Fock matrix
elements, yet the overlap matrix is taken as a unit matrix as far as diagonaliza-
tion of the Fock matrix goes. Discuss. - What would be the advantages and disadvantages of using the general MNDO/
AM1 parameterization procedure, but employing a minimal basis set instead of
a minimal valence basis set? - In SCF semiempirical methods major approximations lie in the calculation of
the Hcorers ,(rs|tu), and (ru|ts) integrals of the Fock matrix elements Frs
(Eq.6.1¼5.82). Suggest an alternative approach to approximating one of
these integrals. - Read the exchange between Dewar on the one hand and Halgren, Kleir and
Lipscomb on the other [ 28 ]. Do you agree that semiempirical methods, even
when they give good results “inevitably obscure the physical bases for success
(however striking) and failure alike, thereby limiting the prospects for learning
why the results are as they are.”? Explain your answer. - It has been said of semiempirical methods: “They will never outlive their
usefulness for correlating properties across a series of molecules...I really
doubt their predictive value for a one-off calculation on a small molecule on the
grounds that whatever one is seeking to predict has probably already been
included in with the parameters.” (A. Hinchliffe, “Ab Initio Determination
of Molecular Properties”, Adam Hilger, Bristol, 1987, p. x). Do you agree
with this? Why or why not? Compare the above quotation with ref. [ 24 ],
pp. 133–136. - For common organic molecules Merck Molecular Force field geometries are
nearly as good as MP(fc)/6-31G geometries (Section3.4). For such molecules
single-point MP(fc)/6-31G calculations (Section5.4.2), which are quite fast,
on the MMFF geometries, should give energy differences comparable to those
from MP(fc)/6-31G//MP(fc)/6-31G calculations. Example: CH 2 ¼CHOH/
CH 3 CHO,DE(MP2 opt, including ZPE)¼71.6 kJ mol"^1 , total time 1,064 s;
DE(MP2 single point on MMFF geometries)¼70.7 kJ mol"^1 , total time¼48 s
(G98 on a Pentium 3). What role does this leave for semiempirical calculations? - Semiempirical methods are untrustworthy for “exotic” molecules of theoretical
interest. Give an example of such a molecule and explain why it can be
considered exotic. Why can’t semiempirical methods be trusted for molecules
like yours? For what other kinds of molecules might these methods fail to give
good results?
444 6 Semiempirical Calculations