I agree with Maciej in that standard DFT methods are generally found useless, so are semi-empirical and molecular mechanics. If what you want is accuracy (at a reasonable computational cost), then you have to go to DFT methods including dispersive interactions such as DFT-D (Stefan Grimme). In Gaussian09 you can use B97D and wB97XD.

However, I would like to draw your attention to the fact that the existence of pi-pi interactions have been exaggerated throughout the literature. They are not so ubiquitous as some people pretend. Two bencene molecules are not attracted by pi-pi interactions: what type of attraction would you consider between two aromatic electronic clouds?: they should rather repel each other!. As a demonstration, try to find in any database the structure of solid benzene and you will find (perhaps to your surprise) that there is not a planar stacking, but rather the opposite.

Pi-pi interactions require alternating electron-rich and electron-deficient aromatics. I strongly suggest the reading of this excellent review:

Chem. Sci., 2012, 3, 2191

Rethinking the term ‘‘pi-stacking’’

Chelsea R. Martinez and Brent L. Iverson*

DOI: 10.1039/c2sc20045g

If you are considering protein-ligand interactions, you hale LigPlot or LigPlot+. This software can identify non-bonded interactions and hydrogen-bonds between ligands and protein residues.

You you only want to check pi-pi interactions between ligands, it will be a little more difficult, but i think LigPlot can also do that (interactions between ligands).

Using Merck Molecular Force Field MMFF (that has parametrized such pi-pi interactions) will allow you to obtain a first model (for any intermolecular or intramolecular), then you can optimize that MMFF geometry with a density functional method and a good enough basis set (with polarizable and diffuse functions). I usually use Spartan for MMFF and Gaussian gor DFT.

Thanks for the answers. There was a mistake in the title. I meant 'intramolecular, not intermolecular', which probably changes a lot... I have edited the title. Vincente Marti-Centelles: Is the MMFF (or similar method) implemented in Gaussian03 or HyperChem? These two are available at my lab.

Hello Maciej, I'm not familiar with HyperChem, but in

Gaussian 09 it is availible http://www.gaussian.com/g_tech/g_ur/k_mm.htm

Any way, if you want to minimize a structure (not to make an automatic conformational search) you can optimize the structure in Gaussian using the semiempirical level of theory PM6 and then a DFT with the approrpriate basis set.

Within DFT this problem typically solved with employment of LDA functional that implemented in all DFT codes (I use SIESTA). It work proper due to "error cancellation". GGA+vdW (van der Waals) is also good and correct, but implemented only in a few codes.

I have to say I am a bit confused. I have looked through literature and I found that: standard DFT methods are generally found useless, so are semi-empirical and molecular mechanics.

DF-SCS-LMP2 is advised as a good method, however it is not implemented in neither G03 nor HyChem. MP2 is also said to be a proper choice, however it seems out of my computational capabilities.

In that case tho reduce computation

time you can use semiempirical PM6

available in gaussian 09

DFT should be a resonable solution if you use a new feneration of XC functional (e.g. wB97XD or M06)

I agree with Maciej in that standard DFT methods are generally found useless, so are semi-empirical and molecular mechanics. If what you want is accuracy (at a reasonable computational cost), then you have to go to DFT methods including dispersive interactions such as DFT-D (Stefan Grimme). In Gaussian09 you can use B97D and wB97XD.

However, I would like to draw your attention to the fact that the existence of pi-pi interactions have been exaggerated throughout the literature. They are not so ubiquitous as some people pretend. Two bencene molecules are not attracted by pi-pi interactions: what type of attraction would you consider between two aromatic electronic clouds?: they should rather repel each other!. As a demonstration, try to find in any database the structure of solid benzene and you will find (perhaps to your surprise) that there is not a planar stacking, but rather the opposite.

Pi-pi interactions require alternating electron-rich and electron-deficient aromatics. I strongly suggest the reading of this excellent review:

Chem. Sci., 2012, 3, 2191

Rethinking the term ‘‘pi-stacking’’

Chelsea R. Martinez and Brent L. Iverson*

DOI: 10.1039/c2sc20045g

@ German Sastre: It is true that so called "pi-pi interactions" are not so obvious as one may think. Then , what about most of med-chem docking software that treat pi-pi stacking as a reasonably strong interaction? Thanks for the review, I'll try to go through it.

Did you mean the so called "T-Stacking"? I found a nice example of application of MP2 methods to investigations of the reaction mechanism (this is what I am trying to do for another reaction) where this kind of interaction between two phenyl moieties has been found superior than "facial stacking". It was describen in J. Org. Chem. 2010, 75, 885–896.

In my case the diastereoselectivity of the condensation reaction almost always prefers the A isomers. However, when I switched the aliphatic substrate to the one with the phenyl ring, the diastereoselectivity was no longer observed! There is an another phenyl-containing reactant in the same reaction. On the basis of the set of experiments (wet, dirty chemistry :)) I could not find any other reasonable explanation for the failure of diastereoselectivity in this one special case than pi-pi kind of interaction between the two substrates... But it is difficult to explain in just few words.

@ Maciej Dawidowski: just an answer to your very good question:

"what about most of med-chem docking software that treat pi-pi stacking as a reasonably strong interaction? "

And my answer: these interaction are (usually) just pure "van der Waals" (dispersive) interactions. More details: http://en.wikipedia.org/wiki/Van_der_Waals_force