Negative frequency or imaginary frequency suggests that your molecule is not minima on PES (Potential Energy Surface); instead, it's a transition state structure (one negative freq. refers to the first-order saddle point). You need to follow up on the negative frequency to get the equilibrium geometry. There are two methods to do so

1. manually (by changing the respective coordinate w.r.t the negative freq.)

2. use gauss view to do so...

step 1. Open the .out file/.log in gauss view

step 2. right-click on the molecule and select Vibrations in the Result menu

step 3. Now select the respective negative frequency and tick the manual displacement menu

step 4. increase the manual displacement value from 0 to 1

step 5. save this structure and use this as an input for further calculations

these are the steps to follow up the negative frequency in Gauss view.

I hope it may help you.

Regards

G. J

While I confirm what reported by Gaurav Jhaa , I would add that it is important to evaluate the module of the wavenumber of the imaginary frequency. In general, you should be concerned if the module of the frequency is larger than 50-100 cm-1, which point to the optimization of a TS. Negative frequencies in the range 0 – -15 cm-1 can occur because a vibration breaks the molecular symmetry (provided that the molecule belong to a point group different than C1).

I agree with Massimiliano Arca; generally, we avoid negative frequency below 50cm-1 (mainly with C1 symmetry), but sometimes we consider the negative frequency in the case of highly symmetric molecules with high symmetry like D3h, etc.

Adjust the molecular structure.

It is important to evaluate if the negative eigenvalues have been obtained with or without the SCRF. In fact the SCRF introduces (unless you use a simple Onsager approach) some additional forces on the molecule, because the shape of the cavity, that break the molecular symmetry. According to a clarification I asked to the Gaussian Help they confirmed that because the cavity has an truncated octahedral shape it can not fit the symmetry of the molecule giving rise to unphysical negative eigenvalues. In case of C1 pgs this is irrelevant and you do see this effect.

First, you need to open the output file by GaussView. Then you determine the bond holding the negative frequency. Changing the bond distance from 0.1 to 0.5 Angstroms. Save file and optimize again. If it is not good, you should try to change the dihedral angle around the bond from 1 to 10 degrees. Save file and optimize again. I hope it will help you.