Mr. Rashid,

What you have understood under ''large'' molecule?

Thank you for your reply. What I meant is it is not a small molecule it a perylene Diimide(PDI) derivative @Bojidarka B. Ivanova

If I take a look at the Perylene diimide structure: that's not a large molecule by today's standards, so unless your substitution is gigantic you can start directly.

However, if you have a really big substituent you might consider a PM7 optimization using MOPAC (http://openmopac.net/). Just install the program; the input can be generated automatically with many structure editors, e.g. Avogadro (you just need to switch the method from PM6 to PM7 manually). The program is free, all they want is a citation if you publish.

Then you can proceed optimizing your geometry; if it's really large, start with a GGA functional, e.g. BP86, and a split valence basis set. In Turbomole I used def2-SVP or def2-SV(P) but you can take anything on that level (e.g. a Pople basis set). After that you can augment your calculation to the functional and basis set that you want.

You can try B3LYP/6-311+G** basis set

@Sivakumar Sivagnanam I have tried this base set, half way in the process I get this error #2070 and link dead.

How you wrote the input file? Through Guassview or Chemultra? @ Rebwar M.Rashid

I used gaussview to draw the molecule @Sivakumar Sivagnanam

Maybe MOPAC or xtb is a great tool for the preliminary optimization of large molecules

Could you please show me your input?

Prior to start your final optimization, first optimize the molecule with Hartree Fock (HF) method and 6-311G basis set for better input of final optimization. It will take very short time to complete. Then, from output construct input file for final optimization in the same method you already tried. Hope #2070 type error will not arrive.

Draw your structure using gaussView, clean it or open pdb format file of your structure and optimize the molecule at ground state, Hartree Fock and default spin and 6-311G basis set. Gaussian will run the command.

In contrast to my preposters I recommend you not to use Hartree-Fock for the first optimization run, but instead a GGA functional like BP86. The reason why that makes more sense is simply because HF has a worse scaling behaviour than GGA functionals (so it's slower without delivering better quality) because GGA doesn't calculate exchange integrals.

As J. Weippert suggested that you may try to enter your input file while using Hartree-Fock to run optimization at GGA or Meta-GGA method.