**Imaginary or Negative Frequencies** **Quote:**A stable structure should have no negative/imaginary frequencies. A transition state should have one negative/imaginary frequency. If your optimizer converged to a geometry with too many negative/imaginary frequencies, you will need to shift the geometry by hand, and re-run the optimization. Typically this happens with very floppy motions such as internal rotations and sometimes it can be difficult to avoid.

There are 3N cartesian degrees of freedom (N= the number of atoms in the molecule) and hence MCLR will output 3N frequencies.

**Edited: **There are 3 translational and 3 rotational degrees of freedom, these will have imaginary frequencies. 3N-6 normal modes remain. If the normal modes have no imaginary frequencies the geometry corresponds to a stationary point.

I don't think this is true. The following is:

There are 3 frequenties that are zero. These are due to translations. When they are not zero they should be very close to 0 (plus or minus). If not, the structure is not at its minimum.

In case there is a large negative frequency the geometry is optimised to a transition state. You are at a top of the Hessian and not in a minimum.

When optimising structures the vibrational frequencies can be calculated to ensure there are no saddle points. An indication of a saddle point is an imaginary frequency that does not correspond to a translation.

Sources:

Hints on Quantum Chemistry

http://ocw.mit.edu/NR/rdonlyres/Chemistry/5-68JSpring2003/309378CA-314C-4F31-AF2
E-DFA3C4993489/0/qchem.pdf Geometry optimizations and Hessians

http://www.teokem.lu.se/molcas/tutor/node61.html