CCL:G: Opt with Full point group Symmetry in G09

 Sent to CCL by: Hugo Alejandro Jimenez Vazquez [hjimenez * woodward.encb.ipn.mx]
 Hello to All,
 As I am having the same problems as Yi Yang, I would like to comment in
 some of the answers he has received so far:
 > Sent to CCL by: makowskm__chemia.uj.edu.pl
 > Enforcing symmetrization of the density matrix should help. It is done by
 > SCF=DSymm
 I am aware of this option. However it does not work, at least not with
 C60. I have tried both SCF=DSYMM and SCF=SYMM without success.
 > Sent to CCL by: Cory Pye [cpye|,|ap.smu.ca]
 > Hello Mi,
 >
 > I recommend the use of opt=z-matrix and a very carefully constructed
 z-matrix.
 >
 >-Cory Pye
 As Yi Yang comments later (I apologize for the confusion with his name,
 but I am not familiar with Oriental names), the problem with this approach
 is that making a z-matrix with the full symmetry of the molecule involves
 _a lot_ of work, particularly for higher fullerenes with lower symmetry,
 and still there is no guarantee that it will work in the context of the
 current issue. I will try that though, and I will keep you posted.
 > Sent to CCL by: "Mr Mi Yang" [agri_chemist\a/yahoo.com]
 >
 > First of all thanks for replies of Jimenez Vazquez, Cory Pye and others.
 > Actually, I have used Z-matrix made by molden and it works perfect to
 maintain
 > symmetry in G09 only for small molecules like benzene. But for a Fulleren
 like
 > molecule to build Z-matrix manually is almost impossible.
 >
 > Is there any software which can help to build Z-matrix for 3 dimensional
 > fulleren like big molecule by using some dummy atoms so that it can
 > maintain symmetry in G09 opt...?
 >
 > I observed below mentioned problem in most of molecules with DnD or DnH
 > symmetries where G09 try to maintain largest Abelian group instead of full
 > point group.
 >
 > I have not included any diffused function in basis set but will check a
 > bigger set without diffused function.
 >
 > regards,
 >
 > Mr.MiYang
 Mi, you should be aware that the z-matrices created by molden _are not_
 created in order to represent the symmetry of your molecules. The correct
 definition of the point group of a molecule within the z-matrix involves
 careful choosing of the internal coordinates. Gaussian identifies the
 symmetry from the relative positions of the atoms, not from the way the
 z-matrix is built (of course, as Cory suggested, the advantage of defining
 the point group within the z-matrix is that you can choose the opt=zmat
 Gaussian option for the optimization).
 As you suspect, the definition of a z-matrix containing the full symmetry
 of the fullerene would require the use of dummy atoms in order to define
 at least one of the main axes of symmetry. It would also help if you
 mentioned which basis set you are using. I can say that I can carry out
 calculations on C60 (as an example) within the Ih point group with the
 3-21G, 6-31G(d), 6-31G(d,p) basis sets. However, the symmetry of the
 wavefunction is lost with the 6-31+G(d,p) basis set (all within the
 Hartree-Fock approximation). In addition, I have tried the smallest
 Dunning cc basis set (cc-pVDZ) and the optimization runs smoothly.
 However, the inclusion of diffuse functions (AUG-ccpVDZ) creates the same
 problem as before. So this is not an issue particular to Pople's basis
 sets.
 > Sent to CCL by: "Dillen, Jan [jlmd-$-sun.ac.za]"
 <JLMD-$-sun.ac.za>
 > Hi
 > Using a Z-matrix for a molecule containing rings is looking for
 > trouble, unless you use a so-called "Cartesian" Z-matrix. Also,
 Gaussian
 > uses (internal) redundant coordinates to minimize the energy, unless your
 > override that, so all your hard work may be for nothing.
 > What is not clear from your question is whether the C2h symmetry
 > structure is an energy minimum and why you insist on higher symmetry. I
 > suggest to distort the molecule with the eigenvector of the lowest
 > vibrational frequency, and minimise/maximise to the nearest stationary
 > point.
 > Jan
 Jan is right on the first paragraph. Even if we create a z-matrix with the
 full symmetry of the molecule, Gaussian uses redundant internal
 coordinates for the optimization. It is only when the opt=z-mat option is
 used that the optimization is carried out in internal coordinates.
 In my particular case the optimization of C60 at the HF/6-31+G(d,p)
 changes the point group from Ih to Ci, leading to a change in the degrees
 of freedom for the optimization from 2 to 87. As a result of this change
 in symmetry, the optimization time goes from less than an hour to more
 than one day. This is why it is so important, at least for me, to keep the
 maximum symmetry of the molecule. I guess that if it were not for the
 time, this would not matter, because even if the optimization is carried
 out within the Ci point group, the final geometry does have Ih symmetry.
 Best regards,
 --
 ---
 Hugo A. Jimenez Vazquez
 hjimenez|-|woodward.encb.ipn.mx
 Departamento de Quimica Organica
 ENCB-IPN
 Mexico, DF
 MEXICO