Query regarding use of Counterpoise method for DFT calculations

[krishnad]

Dear all,
  I am using Turbomole 6.3 and I would like to determine the interaction energy between two groups of atoms in a complex. I used dispersion corrected dft (DFT-D3) with def2-TZVP basis set. I proceeding calculating the interaction energy in two ways. First, I performed SP calculations for the complex, group 1 and group 2. Interaction energy was calculated as follows -  interaction energy = (energy of complex) - (energy of group 1) - (energy of group 2). I got a value of -161 kcal/mol which was expected. I also wanted to correct BSSE and employed Counterpoise method by following the Turbomole tutorial. I changed the charges of the group 1 and group 2 atoms to "0" in two separate calculations through the atomic attributes section of define. The interaction energy I obtained after BSSE correction is 229 kcal/mol! That is -161 vs 229 kcal/mol! Could someone kindly suggest why this might be happening?

[amavr78]

Hi krishnad,
I had faced the same problems.
There seems to be a bug when the BSSE is calculated with DFT-D3 methods.
The error is appearing in the 'ghost' calculations with the DFT-D3.

One way to avoid it, is to calculate the BSSE without the $disp3 keyword , and then add the counterpoise correction to your total DFT-D3 energy of the dimer.

You should also include the deformation energy of the monomers into the formula of your binding energy. So, your final binding energy should include  7 terms, as is calculated from the jobbsse script. Check turbomole manual and also these notes useful notes: http://vergil.chemistry.gatech.edu/notes/cp.pdf

Hope that you can solve your problem.

Regards,
Andreas