Re: CCL:semi-empirical methods

From: polowin;at;hyper.hyper.com (Joel Polowin)
Subject: Re: CCL:semi-empirical methods
Date: Thu, 26 Oct 95 11:58:37 -0400
 > Date: Tue, 24 Oct 1995 10:59:15 +0000 (GMT)
 > From: Paddy Kane <94970459;at;vax1.dcu.ie>
 > 	Calculation of heat of formation of benzoic acid
 > 		delta Hf/(kcal/mol)
 > AM1		-54.1467462
 > PM3		-53.6776001
 > ZINDO/1	-3117.2768454
 > ZINDO/S	-5062.7571910
 >
 > 	Calculation of heat of formation of cyclopentanone
 > 		delta Hf/(kcal/mol)
 > AM1		-52.5334612
 > PM3		-53.1513803
 > ZINDO/1	-2586.4691313
 > ZINDO/S	-4012.5111479
 >
 > 1)  The HyperChem Computational Chemistry manual quotes values of
 > -70.1 kcal/mol and -46.0 kcal/mol for the enthalpy of formation of benzoic
 > acid and cyclopentanone respectively. So how accurate can we expect
 > Hyperchem's semiempirical methods to be?
 As has been stated here before, several times and recently -- we can expect
 HyperChem's calculations to be *exactly* as accurate as the same calculations
 performed by other software.  We use published methods with standard
 parameters; we have not had any reports of our results differing from the
 results of others' calculations that did not turn out to be because of
 differences in the options selected or in the structure being calculated.
 My calculations with HyperChem for benzoic acid gave heats of formation of
 -68.1 kcal/mol for AM1, -66.4 kcal/mol for PM3, and -3147.6 kcal/mol for
 ZINDO/1.  ZINDO/S did not converge.  The AM1 and PM3 results are consistent
 with literature values for those calculations.  (See J.J. Stewart, _J.
 Computer-Aided Mol. Design_ *4*, 1-105 (1990).)  The ZINDO methods do not
 reproduce heats of formation well.
 My results for cyclopentanone are: -55.5 kcal/mol for AM1, -55.3 kcal/mol
 for PM3, -2608.3 for ZINDO/1, and no convergence for ZINDO/S.  The Stewart
 article gives -36.1 kcal/mol as the heat of formation calculated with AM1
 with MOPAC, but our repetition of that calculation with AM1 with MOPAC
 gives -55.3 kcal/mol.  This appears to be an error in the article (and in
 our manual, which quotes results from the article).
 > 2)  Why are the ZINDO/1 and ZINDO/S methods giving me such apparently
 > incorrect values?
 The ZINDO methods weren't parameterized by fitting to experimental heats
 of formation.  ZINDO/S was optimized to give good electronic transition
 (UV/visible) results.
 > 3) Another question (apparently unrelated to the first two questions) is
 > why the following message appeared in my log file for calcium but not for
 > magnesium in the ZINDO/1 calculations on the calixarene complexes:
 > 	WARNING message received from node=0:
 > Atom 221 with atomic number 20 the value of bonding parameter of s orbital
 > is less than or equal to zero, please check your parameter file
 zindo1_1.abp.
 The s and p bonding parameters for Ca are actually negative, unlike those
 parameters for all of the other atoms.  Usually, for those parameters to be
 negative would indicate an error in the parameter set, so HyperChem is
 warning you of a possible problem.  In this case, you should ignore the
 warning and continue the calculation.
 Joel
 ------------
 Joel Polowin, Ph.D.   Manager, Scientific Support
 Email to: polowin;at;hyper.com
 Hypercube Inc, 419 Phillip St, Waterloo, Ont, Canada N2L 3X2 (519)725-4040
 Info requests to: info;at;hyper.com    Support questions to:
 support;at;hyper.com
 Email group: Send "subscribe hyperchem" to
 hyperchem-request;at;hyper.com
 WWW: http://www.hyper.com/