g98: TD-DFT geometry optimizations ?



Dear CCL members,
 I have a Gaussian 98 related question:
 According to the g98 users manual (p. 155, TD keyword) it should be
 possible to do Time Dependent DFT geometry optimizations of the N-th
 excited state using the Root=N option.
 However, I can´t find a proper input. The easiest input line
  #P TD=(Root=1) B3LYP/STO-3G Test DENSITY OPT=Z-Matrix
 leads to an error in link 1002:
  (Enter /vol/chemie/libexec/g98.A9/g98/l1002.exe)
  Minotr:  Closed-shell wavefunction.
  Cannot mix post-SCF and DFT.
  Error termination via Lnk1e in /vol/chemie/libexec/g98.A9/g98/l1002.exe.
 I played around with several other options, tried to do the job in a two
 step procedure, like it is described for the CIS method in
 Forsman/Frisch´s "Exploring Chemistry with Electronic Structure
 Methods",
 all without success. Is it possible to do these calculations?  Anybody out
 there with experience in this topic? The ultimate goal would be to perform
 a normal mode frequency calculation for excited states within the TD-DFT
 formalism.
 Regards, :-)ohannes.
 ----------------------------------------------------------------------
 | Johannes Weber                     | Email:                        |
 | Universitaet zu Koeln              | Johannes.Weber -8 at 8- uni-koeln.de   |
 | Institut fuer Physikalische Chemie |                               |
 | Lehrstuhl Prof. Dr. G. Hohlneicher | tel: 0049-(0)221-470-4812     |
 | Luxemburger Str. 116               | fax: 0049-(0)221-470-5144     |
 | 50939 Koeln                        |                               |
 ----------------------------------------------------------------------