Dear Mansur,

If you have ~150 imaginary frequencies (negative eigenvalues of the Hessian), this probably means that your structure is not a stationary point on your potential energy surface. Why this happens, we can only guess.

Perhaps you used a different level of theory (basis set, functional, ...) for your optimization than for your frequency run. Or you need to optimize your 100-atom molecule more tightly...

You can probably share your input / output with the community of users / developers of FHI-aims if you are still in doubt?

Hope this helps,

Fedor