CCL:RES: Help With Gaussian 98 Hindered Rotor Calculations

Dear Jim,
 Welcome to the club.
 I too have had the same problem.
 I am trying to do the same with propylbenzene radicals using HF and b3lyp
 methods. In both cases where the radical center is on the end carbon or
 middle carbon of the propyl chain G98 fails with the same message that you
 are getting.
 About a month ago I wrote to Douglas Fox at Gaussian
 The answer was this:-
    Dr. Hovell,
    The option FREQ=ReadInfo is not implemented.  This functionality
 is implemented as FREQ=ReadHindered or FREQ=ReadHind for short.
 The input is slightly different than the manual suggests,
 VMAX 25.0
 7 8 3 1 1
 would set the barrier cutoff, between atoms 7 and 8, for a threefold
 barrier with a symmetry number of 1 and only 1 well would be considered.
 Repeat if further barriers need to be considered.
    I am doubtful that even getting this right will fix an error regarding
 the number of degrees of freedom.  If problems persist please send
 the input and the last 100 lines of the output file.
  Douglas J. Fox
   Technical Support
   Gaussian, Inc.
   help{at}gaussian.com
 So I sent him a copy of my problem, I am still waiting for a reply. As far
 as I know this is an old problem that has never been solved.
 When I asked a similar question to the CCL group a while back. I received a
 number of replies. Here there are, I have not yet tried all of their ideas
 yet.
 I have been running into similar problems with freq=hindered, in particular
 when there is a ring present.
 I usually follow this procedure.
 first look at the connectivity list (*.out) and check if there are no
 connection between next nearest neighbors. They have to be removed
 (geom=modred)
 sometimes it is necessary to add some angles and dihedrals.
 what helps for problems with ring structures is to "break" the ring
 structure using the connectivity list (remove a connection, angle,
 dihedral). Though this has no influence on the calculated properties (I
 guess so??) it does sometimes solve the problem with hindered rotors.
 Just as a note: the hindered rotor method is not all that accurate-in some
 cases we found large deviations between a fully correct approach (using a
 calculated rotor PES and solving the corresponding 1D schrodinger equation
 for internal rotation) and the hindered rotor approach of gaussian.
 For the more accurate approach you should look into the recent papers of
 Sumathy, Green et al. and Van Speybroeck, et al.
 Some codes are available on the net, but you might want to contact the
 authors of the previous papers to obtain a version of their code.
 Kind regards,
 Mark Saeys
 Laboratory for Petrochemical Engineering
 Ghent University
 Krijgslaan 281 (S5)
 B9000 Gent
 Belgium
 tel: .32.(0)9.264.56.78
 fax: .32.(0)9.264.49.99
 http://lptnt01.rug.ac.be/chemical.engineering
 	---------------------------------------
 The only other things I can think of might be to use ub3lyp/6-31+G(d),
 include nosymm keyword and maybe reduce the step size with IOp(1/8=n),
 default is 30, use 5 maybe. I had some problems with some b3lyp optimsation
 earlier this year. I got around the problem by performing scans with
 HF/3-21G(d) and then capturing the coordinates and writing XYZ files for the
 two minima and the highest point on the scan. Then I wrote a QST3 file with
 b3lyp keyword and the calculation then behaved!
 Have you tried a opt=stable calculation to check that you have a sensible
 wavefunction? You could maybe constrain some geometries, as in fix the
 aromatic ring ( fix dihedrals to zero around the carbons in the aromatic
 ring ) to be flat and then run the optimisation again.
 These are just ideas and you might think I am mad.
 James Robinson [prsjjr{at}bath.ac.uk]
 	 ------------------------------------------
  I don't know if this would work, but maybe adding nosymm would help?
 	--Shobe, Dave [dshobe{at}sud-chemieinc.com]
 	 -----------------------------------
 try ub3lyp/6-31+G(d) single point to see if that convergences.
 	James Robinson [prsjjr{at}bath.ac.uk]
 	----------------------------------
 Hello,
 I don't know just what causes your problem, but I have heard that diffuse
 (+)
 functions can sometimes be a pain. Plus functions are probably not needed
 for
 propylbenzene (it is not an anion), and also, for DFT (unlike ab initio)
 going
 beyond 6-31G* tends to give little or no improvement:
 J Phys Chem A, 1997, 101, 208
 J Phys Chem 1994, 98, 11623
 elewars [elewars{at}trentu.ca]
 ---------------------------------
 The experimentally determined sixfold "barrier" for the 'hindered'
 rotation
 in toluene is 5 cm^-1 (0.014 kcal/mol!!!) in the s0 state - not really a
 barrier at all! It is 25 cm^-1 in the S1 electronically excited state.
 [J. Mol. Spect. 97 (1983) 186]
 I doubt that it is possible to calculate a barrier accurately which ist as
 low as these numbers are. But perhaps you can try just to calculate the
 barrier height from two single points at the min and max value, take the
 internal methyl rotor rotational constant from the vibrational analysis and
 calculate the contribution to the partition function from the internal
 rotor quantum mechanical problem.
 Regards
 Wolfgang Roth [Wolfgang.Roth1{at}epost.de]
 -------------------------------------
 Thanks for forcing me to produce this short summary. Even if it is
 inconclusive at the moment. If you find a solution to this problem please
 let me know.
 Ian Hovell - Ph.D.
 NUCLEO DE MODELAGEM MOLECULAR-NMM
 Centro de Tecnologia Mineral - CETEM
 Ministerio da Cijncia e da Tecnologia- MCT
 Avenida Ipj, No 900 - Cidade Universitaria
 Ilha do Fundco Rio de Janeiro RJ Brasil
 CEP 21941-590
 tel 00 55 (xx) 3865 7344 ou 3865 - 7216
 Fax 00 55 (xx) 22602837 ou 2290-4286
 e-mail hovell{at}cetem.gov.br