how to calculate stabilization due to vdW when using vdW-DF?

Message

Ghorbanpour · #1 Post by **Ghorbanpour** » Tue Jan 21, 2014 7:59 am

Hi all,

I'm doing some NEB jobs, and I'd like to assess the role of vdW interactions in stabilizing transition state compared to initial and final states. However, the vdW correction values reported by VASP when using vdW-DF is positive (also when I use LUSE_VDW = .FALSE., the total energy is more negative than when vdW-DF is turned on), which doesn't make sense to me since vdW forces must stabilize the entire structure and lower the total energy. Could someone clarify this to me please?

stephan · #2 Post by **stephan** » Tue Jan 21, 2014 11:50 am

While I agree that it is somewhat surprising, I do confirm your observation.
Please remember that we are talking about total energies, which do not necessarily mean much. Second: the vdW kernel term includes the gradient correction to the correlation energy which might be positive. Third, there are "repulsive" regions in the kernel, i.e., at short distance the point-wise interaction seems repulsive (at least that's what I understand).
In any case, this is a general feature of the vdW-DFs: vdW-DF, vdW-DF2, VV09, VV10 all give a positive contribution to the total energy in the (molecular) examples I checked with VASP and Q-Chem.
In summary: just look at energy differences and keep in mind that the physical content is not as "clean" as you might wish...