Trouble in G94 with MP2 Freq (Summary)

 Hi all,
 Thanks to Customer Service of Gaussian (Doug Fox) and Stefan Fau for their
 replies to my question. In brief, G94 does not support analytic second
 derivatives for ECP basis sets, although G98 does.
 The answers and original question follow.
 (From Cust. Service Doug)
   Dr. Garcia,
    G94 does not support analytic second derivatives for ECP basis sets.
 We have added that in G98 but you can use FREQ with MP2 in G94 and to these
 with finite differences by specifying FREQ or FREQ=HPModes.  In general
 MP2 by itself with memory and MAXDisk specified will make good choices on
 other MP2 options.
 (From Stefan Fau)
 Dear Pablo,
 >from the name of your chk-file and your choice of basis set I assume
 that your molecule contains Pt. If this is true, you have a problem.
 Pt should be treated with an ECP to include linear relativistic effects,
 but G94 is not able to calculate analytic 2nd derivatives if ECPs are
 present. This feature is implemented in G98 only.
 (Original Question)
 This is the first time I tried to calculate the IR spectrum with MP2, and
 I have the following problem. I read in the manual that analytical
 frequencies need MP2=SEMIDIRECT. So I used the following input:
   (rest omitted)
 And the computation stopped with:
  Analytic MP2 frequencies require a (semi-)direct transformation.
  Error termination via Lnk1e in /usr/local/g94/l1.exe.
  Job cpu time:  0 days  0 hours  0 minutes  0.4 seconds.
 I would appreciate any help. Could it be that analytical frequencies are
 not available with core potentials (LANL2DZ is an only-valence basis set,
 with core replaced by an ECP).
 Thank you very much
 Pablo Vitoria Garcia
 Dpto. Quimica Inorganica, Facultad de Ciencias
 Universidad del Pais Vasco (UPV/EHU)
 Aptdo. 644
 E-48080 Spain
 Phone: +34 94 6012000 Ext. 5529
 Fax: +34 94 4648500