summary on Ni basis set



 Dear CCL members
 here are the answers (11) to my enquiry on a good gaussians Ni basis
 set.
 I thank the authors of the answers for their help.
 Angelo
 _______________________________________________________________________
 Angelo Vargas
 Laboratory of Technical Chemistry
 Department of Chemical Engineering and Industrial Chemistry
 Swiss Federal Institute of Technology (ETHZ)
 ETH Zentrum, Universitätsstr. 6    Telefon:  0041/1/632 31 54
 CH-8092 Zürich - Switzerland       Fax:      0041/1/632 11 63
 E-mail:   vargas-0at0-tech.chem.ethz.ch
 http://mercury.ethz.ch/members/vargas/vargas.html
 ________________________________________________________________________
 1)
 Hi Angelo,
 I am quite happy with the use of the Schaefer, Horn and Ahlrichs (SHA)
 basis
 sets for 3d metals. You can look them up in J. Chem. Phys. 97, 2571
 (1992).
 You should also try http://www.emsl.pnl.gov:2080/forms/basisform.html
 >From there I downloaded the pVDZ set for Ni (for use in Gaussian
 94/98):
  NI  0
  S   6  1.00
      71074.80321100        0.00142604
      10672.02094100        0.01092824
       2428.13890070        0.05421263
        685.53595148        0.18874769
        223.10072863        0.38324617
         76.84201404        0.29550637
  S   3  1.00
        148.71122016       -0.11014443
         17.45915499        0.64521427
          7.16252807        0.44797838
  S   3  1.00
         12.55613713       -0.22645403
          2.07357405        0.72320959
          0.85382641        0.44868026
  S   1  1.00
          0.10536766        1.00000000
  S   1  1.00
          0.03813409        1.00000000
  P   5  1.00
        916.73608662        0.00934396
        216.06139913        0.06973737
         68.38391482        0.27073495
         24.59384395        0.53078302
          9.13929602        0.34410229
  P   3  1.00
          4.71933717        0.34076082
          1.81618492        0.56580170
          0.67840751        0.23616717
  D   4  1.00
         47.09383211        0.02898232
         13.14646397        0.15494996
          4.41705489        0.37633115
          1.47715651        0.47365096
  D   1  1.00
          0.43735922        0.31247838
  P   1  1.00
          0.14658800        1.00000000
  ****
 Have a nice day
 Micha
 ooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo
  Dr. Michael Hartmann
  Dipl.-Chem. (univ.)
  Research School of Chemistry                   Ph. 61-2-6249-3771
  Australian National University                 Fax.61-2-6249-0750
  Canberra ACT. 0200                   http://rsc.anu.edu.au/~micha
  Australia
 ooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo
 2)
 Angelo,
 Try Bauschlicher's ANO basis set. You'll find it at the PNNL site.
 Regards,
 John Kerkines
 3)
 Dear Angelo,
 I' m sure you tried many of the things I'm suggeting but maybe it's
 still
 helpful. One thing I found helpful with TM complexes and gaussian is to
 start from a small basis set Hartree-Fock calculation and slowly upgrade
 to
 the larger target bases (Guess=Read). HF converges somewhat better than
 DFT
 because the HOMO/LUMO gap is larger. Another thing is to start from a
 related closed shell system because these converge better or from a more
 positively charged ion which also tend to converge better. In some cases
 level shifting proved to be very helpful to obtain convergence (about
 0.1
 Hartree used to be enough). A very big convergence improvement was to
 use
 the DGauss DZVP basis set which can be obtained via the EMSL basis set
 library under
    http://www.emsl.pnl.gov:2080/forms/basisform.html
 This, in my experience is a pretty well designed and efficient basis
 set.
 Other very nice transition metal basis sets are the VDZ and VTZ bases
 >from
 Ahlrichs group that can also be obtained from EMSL. In the end it will
 certainly depend on the specific application that you have in mind which
 one turns out to be the best basis set. Ni appears to be the worst case
 for
 almost any theoretical method and to obtain truly converged results
 requires unrealisitically large basis set. There is a paper by Siegbahn
 and
 I think Taylor on the subject in Theoret. Chim. Acta a few years ago.
 Good luck + best regards
 Frank
 4)
 In our group, lots of research has been done on Ni-complexes with great
 success. The calculations were done with ANO basis sets in MOLCAS. See
 our
 homepage (http://hydra.chem.rug.nl/) for references to publications
 about
 these studies.
 Best regards,
 Marcel Swart.
 =========================================
 drs. Marcel Swart
 Theoretical Chemistry (MSC)
 Molecular Dynamics (GBB)
 Rijksuniversiteit Groningen
 Chemistry Department
 Nijenborgh 4
 9747 AG Groningen
 The Netherlands
 tel : +31 - (0)50 - 3634377
 fax : +31 - (0)50 - 3634441
 E-mail : m.swart-0at0-chem.rug.nl
 WWW: http://hydra.chem.rug.nl/~swart/
 =========================================
 5)
 hi angelo
 for heavy elements relativistic effect can be important. so try
 using basis sets with effective core potentials. in g98 the
 los alamos (key word: lanl2dz) and the stuttgarter basis
 sets (key: sdd) are installed.
 if this does not work try to include diffuse functions (6-31++g**).
 because of there long range effects they may describe bondigs
 in complexes in a better way.
 hope this will help.
 sincerely
 michael braunschweig
 department of chemistry
 university of dortmund
 germany
 6)
 Dear Mr. vargas,
 You can employ a large valence basis set of Hay and Wadt with  effective
 core
 potentials (ECP) for Nickel. It is quite good for nickel complexes.
 Reference.
 Hay P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 299
 7)
 Hello,
 I had a bit of a fight recently with the vanadium basis set in G98 as
 well
 (6-311+G).
 My personal suggestions are:
 If you don't need all-electron basis sets, at least for the geometry,
 then either the LANL2DZ ECP + valence basis set ok, or the CEP-31G
 (Stevens-Basch-Krauss compact ECP).
 For all electron, there's always the Huzinaga basis sets, "Handbook of
 Gaussian Basis Sets", though you have to type them in by hand.
                                                         -fred
 "No science has ever made                 Frederick P. Arnold, Jr.
  more rapid progress in a                 A&HPRC, U. of Chicago
  shorter time than Chemistry."            5640 S. Ellis Ave
         -Martin Heinrich Kloproth, 1791   Chicago, IL 60637
 8)
 Angelo,
         there are also some 6-311G** type basis sets for Ni in G98,
 you might want to try them, the following is from the G94 Manual.....
  6-311G: Specifies the 6-311G basis for first-row atoms and the
 MacLean-Chandler (12s,9p)
 (621111,52111) basis sets for second-row atoms [170-171] (note that the
 basis sets for P, S, and Cl
 are those called "negative ion" basis sets by MacLean and Chandler;
 these were deemed to give
 better results for neutral molecules as well), the Wachters-Hay
 [172-173] all electron basis set for
 the first transition row, using the scaling factors of Raghavachari and
 Trucks [174], and the
 6-311G basis set of McGrath, Curtiss and coworkers for most of the rest
 of the third row (note
 that K and Ca are not currently defined) [290,291,292].
 Note that Raghavachari and Trucks recommend both scaling and including
 diffuse functions when
 using the Wachters-Hay basis set for first transition row elements. You
 will need to use the
 6-311+G keyword form to include the diffuse functions recommended in
 their paper (see
 reference [174]). MC-311G is a synonym for 6-311G.
 ....so you might want to stick some diffuse functions in there too.
 noj
 --------------------------------------------------------------------------
 Dr. N.O.J. Malcolm
 e-mail:malcolm-0at0-mail.chem.tamu.edu
 Department of Chemistry
 Texas A&M University
 College Station
 TX 77845
 U.S.A
 --------------------------------------------------------------------------
 9)
 Often, the reason metal compounds don't converge is because of a poor
 initial guess. You can usually tell if this is the case, since there is
 a
 large change in energy on the first SCF iteration (if you use #P on the
 route card).
 A good work around is to precondition your job to give it a better
 chance
 of converging. I am actually writing a short paper about this which will
 eventually appear on www.gaussian.com.
 The way to precondition the job is this:
 First, run the SCF using only an STO-3G basis set. You should name your
 checkpoint file (using a %chk card) so it is saved after this job
 completes.
 Then run it 6-31G, using the same named checkpoint file with GUESS=READ
 and GEOM=CHECK (or GEOM=ALLCHECK). Then your converged STO-3G run will
 be
 the initial guess for the 6-31G run.
 Finally, run the job with 6-31G**, using the same checkpoint file and
 GUESS=READ and GEOM=CHECK again. The converged 6-31G run will be the
 initial guess for this job.  This final step should converge without a
 problem.
 If this doesn't work, send along your input file, and I'll take a look
 at
 it.
 --
 Joseph Ochterski, Ph.D
 Senior Customer Service Scientist
 help-0at0-gaussian.com
 10)
 Dear Angelo,
         The basis sets from Schafer, Horn and Ahlrichs, J.Chem. Phys.
 1992,97,2571, are very good for the transition metals. They have several
 sets at different levels of approximation. They can also be obtained
 >from the \
 emsl website, with the following address.
 http://www.emsl.pnl.gov:2080/forms/basisform.html
         I doubt whether changing the basis set will solve your
 convergence
 problems. With the transition metal complexes getting convergence can
 sometimes be a time consuming task. Sometimes starting with a smaller
 basis, which may give convergence, and then using the converged
 functions to
 begin calculations with a larger basis is a way to solve the problem,
 but
 it does not always work.
                 Regards,   Graham Chandler
 11)
 eetings,
 The problem may not be with the basis set. It may be that a poor initial
 guess has been made with the extended Huckel routine. (Look at the value
 of S**2--is it close to the desired value?)
 I have been doing calculations on d3 chromium complexes and found it
 expedient to use an initial single point calculation with SCF(QC
 conver=2) to improve the guess of the wavefunction. A second calculation
 uses SCF(Conver=5 Vshift=400) to optimize the geometry.  This has worked
 for me so far, with the cep-31g basis set and ub3lyp.
 good luck,
 Curt Hoganson, Ph.D.
 Univ. Delaware