Hi Artem, On Thu, Oct 14, 2010 at 7:40 PM, Artem Baskin <[email protected]> wrote: > Thank you Marcos, > I'll definitely try what you suggested, but I have 2 more questions: > > 1)To calculate the energy (binding energy) between nanoribbon and a > molecule (above the ribbon) I used the value of TOTAL energy in block > > siesta: Final energy (eV): > siesta: Kinetic = 7168.537907 > siesta: Hartree = 83393.002729 > siesta: Ext. field = 0.000000 > siesta: Exch.-corr. = -3071.032309 > siesta: Ion-electron = -173190.442014 > > siesta: Ion-ion = 75708.722958 > siesta: Ekinion = 0.000000 > siesta: Total = -9991.210729 > of output file. Am I right doing like this (providing that all > thermodynamic parameters are by default)?
No. You have to use BSSE corrections for that. Check the list archives for this subject. I will send you a review on DFT with localized orbitals by Roberto Dovesi in which this explained in reasonable detail. > > > 2) And finally I have the general question about the meaning of relaxation > procedure and its connection to the convergence. > Trying to calculate the BS for armchair nanoribbon without relaxation part > I got absolutely wrong BS, an spurious dipole moment and other weird > > things. > As I understood, the problem is that my initial geometry was not > appropriate and needed to be corrected. Moreover, I could not get the > convergence within 100 SCF steps (there were oscillations of dDmax level). Problems with SCF convergence have been adressed a billion times before here. Lower your mixing weight as much as you can, increase your electronic energy to the point of convergence, then lower your electronic energy a bit and re-start from the previously converged DM or simply check if your structure is correct using xcrysden, molden or some other visualization program. > > But it seems to be strange. If I want to calculate the electronic energy > of a system with the particular geometry (like in Gaussian) I expect to > have that energy but not the energy corresponding to optimized geometry. > > So far I do not know how to calculate the BS, electronic energy and so on > of non-relaxed geometry. > First of all: what does BS mean? Second: I have sent you huge emails on the subject through the list some time ago, having spent loads of time and energy writing these emails. I don't know exactly what you didn't understand in them, neither have you made any substantial effort to let me know EXACTLY what you didn't understand from my explanations. In this way, communication gets VERY difficult and I get very upset because even if it's not the case, it seems that you are not taking what I wrote into consideration. Either you tell me exactly what part of my emails you didn't understand and we will try to work them out, or I suggest that you ask your supervisor, or a more experienced postdoc in your research group, to explain you this subject. This is NOT a question specific to Siesta. Rather, this is a question on general DFT concepts. Thrid: if you want the total energy of a non-relaxed geometry, all you have to do is to perform a full SCF calculation until you get to convergence. You will then have the total energy of this specific geometry. Marcos
