Vitor, Although I agree that you will have to perform a counterpoise correction on the energy, that will only lower the binding energy by about 0.1-0.2 eV. So, something else is wrong here.
I took another look at the original Grimme paper and found that you're using too high of a Grimme potential parameter. Supposing C6 for Argon is 4.61 (Table 1) from the paper J. Comput. Chem. Vol 27, 1787-1799 (2006), the geometric average of C6 for two Argon atoms (eq. 13) is, unsuprisingly, 4.61 and that is the value you should use in your MM.Potentials Block, rather than 47.85. There is a parameter, MM.Grimme.S6, missing in your input file. Experience suggests that the default value for this parameter is usually not good enough to capture the physics of the vdW interaction. As Herbert said, plotting Ar-Ar distance versus the binding energy will clearly show you the trend (you will have to perform a ghost atom calculation for each distance, look elsewhere on the list for help with that, I believe I answered a similar question some time ago). You should see that the PBE functional does not bind Ar-Ar together, with no minimum in Binding E vs distance, and the DFT-D+PBE scheme will probably bind it slightly. Increasing your basis set size is going to play a big role in how your calculation comes out. If you don't want to use the simplex method to optimize your basis, try out TZDP or a similar, larger basis. You must declare your own basis set if you wish to include diffuse orbitals (s-like orbitals of higher principal quantum number). Cheers! On Thu, Jun 27, 2013 at 8:25 PM, Vitor Damiao <[email protected]> wrote: > Hi Abraham, > > follows the file .fdf to see if you can help me. > > argon dimer calculations showed acceptable results with the vasp and g03 > programs. > > thanks, > Vitor. > > > 2013/6/27 Abraham Hmiel <[email protected]> > >> Hello Vitor, >> >> GGA functionals typically do a poor job of capturing the physics of >> systems with strong van der Waals character like the Ar dimer. Typically >> LDA performs better than GGA because of fortunate cancellation of error, >> but is still far from correct. >> >> Try installing a recent version of SIESTA's development trunk, available >> from the website. You will be able to use the vdW-DF (DRSLL), vdW-DF2 >> (LMKLL), and opt-b88 (KBM) functionals which may significantly improve your >> Ar-Ar bond distance and binding energy at the expense of additional time >> spent calculating the xc energy. There is a decent pool of computational >> results in the literature to draw from here, as well. >> >> Furthermore, these self-consistent vdW functionals will give you the >> tools to engineer a better optimized basis for Ar. Using an optimized basis >> rather than the automatically-generated ones can be useful and it probably >> couldn't hurt to add additional zetas, polarization orbitals and diffuse >> orbitals to your basis. Without your .fdf file though, it's difficult to >> give you a more exact perscription, but using some form of vdW-DF could be >> a decent start, I feel. >> >> Best of luck, >> >> >> On Thu, Jun 27, 2013 at 2:28 PM, Vitor Damiao <[email protected]>wrote: >> >>> Hi all, >>> >>> Have anybody calculated the argon dimer with SIESTA program? It appears >>> that the binding energy (BE) is overestimated. Please, see a DFT-PBE >>> calculation using a DZP basis set for Ar---Ar: >>> BE (DFT) = 1.3 kcal/mol >>> BE (DFT+D) = 2.4 kcal/mol >>> BE (Expt.) = 0.28 kcal/mol >>> >>> I have tried using different pseudopotentials schemes, but the binding >>> energies appears to be too high. I would be grateful if someone could >>> explain me this trouble. >>> >>> Thanks in advance >>> Vitor >>> >> >> >> >> -- >> *Abraham Hmiel* >> Katherine Belz Groves Fellow in Nanoscience >> Xue Group, College of Nanoscale Science and Engineering at SUNY Albany >> http://abehmiel.net/about >> >> > -- *Abraham Hmiel* Katherine Belz Groves Fellow in Nanoscience Xue Group, College of Nanoscale Science and Engineering at SUNY Albany http://abehmiel.net/about
