Dear Adf users,

                We, in our group, study the different conformations of transition metal
clusters.
                We found some problems when the vib. frequencies are calculated for such
systems. When the geometry of a four-atoms cluster is optimized and
restricted to be in the T_d group, an electronic state T is obtained. This
fact seems to anticipate Jahn-Teller effects, that is, distortion along
some imaginary frequencies.
                When the vib. analysis is made all the freqs. obtained are REAL!! We
inspected the freq. calculation output and found that during the scf cycle
on the first geometry, the geometry coming from the optimization, the
program puts 1/3 of the unpaired electron in each t orbital giving an A
state, instead of the original T state. Thus, the freqs. obtained do NOT
correspond to the optimized geometry!!!
                We know that freq. calculaltions are performed without symmetry in ADF
due to absence of analytic Hessians, but our question is, how reliable is a
freq. run performed on systems having degenerate electronic states?
                We'd like very much hearing your opinion and to know if someone else
found similar problems. If you consider that a given technical detail is
needed to elaborate your answer, dont hesitate in sending a mail.
                Regards,

                                                                                                Reinaldo
Received on 1999-02-17 16:24:32