Dear Bitupon,

If you have only one negative frequency it indicates that the structure is a transition state and if you have more than one negative frequency it indicates that the structure is imaginary. In reactants, intermediates and products all frequencies should be positive.

In order to solve your problem you should start from a reasonably good initial geometry such as the X-ray structure geometry. I suggest to first optimize the x-ray structure geometry using UFF or any other MM method, then DFT with low basis set such as B3LYP/6-31G(d,p). When you run the DFT calculations do not make any restriction to bonds or/and angles since if you make a constriction you might get negative frequency.

Hoping this will be helpful,

Rafik

Dear Bitupon,

it is possible that you are overly constraining the geometry by symmetry. This is not an entirely unusual problem. If you can afford it you may have some success if you removed the symmetry constraints by slightly distorting the starting geometry ( there is no way of explicitly declaring the point group symmetry in Gaussian ) and re-running the geometry optimisation. If you started by calculating an initial hessian you may also improve you chances. 

Once the geometry has optimised, inspect the geometry to see if an alternative geometry is preferred.

It wouldn't be surprising if metallocene type molecules prefer a more relaxed symmetry in the gas phase.

One final suggestion, have a look at the displacement coordinates of the imaginary frequency to identify the direction of steepest descent. This might give you an idea of an improved starting geometry.

All the best

Fred

Thank you so much Rafik Karaman and Frederick Bennett for your valuable suggestions. I shall try accordingly.

If the problem persists, please post your input file. It would be easier to diagnose issues if we could see what you're telling the program to do. 

If you're really doing an unconstrained optimization and getting a negative frequency, you can use the direction of motion indicated by the imaginary vibration to locate one or more true ground state structures. Simply take the "optimized" geometry with the imaginary frequency and displace the atom(s) in the directions indicated by the vibration. Use these two geometries (one for each direction of the vibration) as your initial guess for a new optimization. It/they should relax to a true ground state structure. 

To solve your problem , take time to do the following:

You visualize your molecule with GaussView to see how each movement atom is imaginary frequency. On the displayed molecule you try to change the geometry eliminating imaginary frequency one by one after the calculation step by step.

Dear Bitupon,

please send me your .log file with the imaginary frequency you have, I can follow the imaginary frequency mode for you and see which symmetry your structure wants to adopt along that mode.