Hi Aftab,

It's difficult to know for certain without knowing a bit more information, such as number of atoms, element types, basis sets and functionals. Your issue is likely to do with convergence. For example some transition metals can struggle to converge to a solution, and often require different parameters or methods to main group chemistry.

Cheers,

Kas.

Thanks dear Kasimir Gregory for your reply. here is attached file contining 108 atoms. i have been using M06 functional and 6-31g* basis set, 12 Processors and more than 40 GB Memory 

I would look at what the gradient is doing. The methyl/methoxy groups can rotate between several different local minima during an optimization. I would look at the gradient values and see what they're doing (are they going down, bouncing around/nearby a couple of values without making any progress, etc.). Did you do something like a PM6 semi-empirical or other molecular mechanics optimization first? If not, that might be a good place to start.  I don't know if gaussian has the semi-empirical options or not, but if they don't, mopac is open source and relatively easy to compile and use.

For systems where the potential surface is very flat with many groups that have very shallow local minima (the technical term for such systems is "floppy") you will have a very hard time locating the global minimum. Smaller floppy systems can be optimized with reasonable certainty because you can feel somewhat confidant that you've sampled most/all of the combinations of all of the various degrees of freedom, but in a system like yours, you won't get such an answer. Even if you do, you would likely see it change somewhat if you increased your basis set. 

The point is, do the best that you can by trying all of the standard tricks (optimize at a low level of theory first, fix some parts and optimize the rest before relaxing all of them at the same time etc) and don't wait three months for ANYTHING to converge unless you expect it to take so long. You should really write yourself a script to pull out the values of the gradient at each iteration or, if you have access to it (or the money to buy it) use something like Chemcraft that will read the output file and allow you to plot the energy/gradient vs iteration to see if it's converging or just hopping around between local minima. 

Good luck!

As Adam W. Peltzer suggested, look at the gradient. Sometimes your optimization will get into an infinite loop. If you have similar molecules, you can always start to build your molecules from those. You can try to reduce the number of freedom of your molecule, for example remove methyl groups and replace them with hydrogen, once that is optimized, add the methyl groups and go from there.

If these doesn't work, in this case I would use a Z-matrix to put my molecule together using pre optimized parts (like the aromatic rings that you have there), and when optimizing I would let only the 2x2 dihedral angles between the aromatic groups to rotate (along the bonds 19-23, 20-24 and 29-35, 30-36), and leave the rest fixed, then I would try to go from there by adding more degrees of freedom. These kind of tricks are not needed usually if you start from a pre optimized structure.

Good luck, and don't expect working with large molecules to be easy, because it's not.

Thanks alot for your explanation Sir  Jared hansen, Adam W. Pelzer, Csongor Matyas

First, he can optimize the geometry of your molecule with a small basis STO-3G and then by using 6-31G.