My Community

If I want use the vdW-DF functional proposed by Dion, using
(LUSE_VDW = .TRUE.)
then, Which POTCAR I have to use?

You can use either PBE or LDA potentials, the results are often quite insensitive to this. For Dion's vdW-DF based on revPBE, the PBE POTCARs might be more suitable, but that's a guess.

Usually we find only tiny differences, for example, the interlayer binding in graphite is pretty much the same when both LDA and PBE and the hard or normal POTCARs are compared. (See, e.g., discussion in
http://www.chem.ucl.ac.uk/ice/docs/Graz ... l_JPCM.pdf )

Since the calculation is done on the valence density, the errors will be reduced when you use POTCARs with more electrons or when you use hard POTCARs. Then it shouldn't matter much if your POTCAR is LDA or PBE. For example, for transition metals, the _sv POTCARs give closer results to the all electron calculations (as discussed in Klimes et al PRB 83 195131 (2011)). But maybe you don't want to use them because of the increased cost, then you can run one typical calculation and compare the _sv with normal POTCAR (or hard with normal depending on what atoms you have in your system) and see what the differences are (say, in adsorption energy). There should be some (even PBE will give a slightly different number when used with LDA POTCAR) but they should be acceptable for many purposes.

Thanks a lot, I'm using PBE POTCAR, but I'm obtaining too small adsorption energy values, I will try POTCAR with more electrons.