Dear Haoliang, I think when you compare total energies from Siesta and from Atom, you introduce an unknown error, because they are calculated in quite different way (mesh, basis issues appear in Siesta but not in Atom). It is better to compare comparable values, i.e. total energy of dimer against that of two distant atoms. Define a large enough unit cell to accommodate two separated atoms and, keeping the cell fixed, move the atoms together and trace the energy; hopefully you'll find a minimum at some distance. In doing this, do not forget to make the calculation magnetic and to set the magnetic moments on isolated atoms.
Best regards Andrei Postnikov > Dear all, > I was trying to calculate the binding energy of a SiC molecule.The formula > I use is : E(SiC)-E(Si)-E(C), and then divide it by the number of total > atoms to get the binding energy per atom. The experimental value is > 2.34eV, and what I got is over 3.44eV which is way off the experimental > value. I used the ATOM program to calculate the atomic energy of Si and C > which might not be a good idea. I got the energy of Si atom and C atom > from all-electron and pseudopotential calculation separately. The energy > of Si is -7.5 Ry and the energy of C is -10.69 Ry using pseudopotential > calculation. Then I calculated the total energy of a SiC molecule with > structural relaxation which is not satisfying. The total energy is > -254.267753eV which is equal to -18.696 Ry approximately. That gives me > the binding energy per atom 3.44eV. > I used the pseudopotential of Si from ATOM program of Siesta and the > pseudopotential from Javier Junquera (Flavour of the pseudopotential: > Troullier-Martins. Exchange and correlation functional: LDA > (Ceperley-Alder). Relativistic: no. Core corrections: no. Valence > reference configuration: 2s2 2p2 3d0 4f0 Cutoff radius: 2s 1.25 bohr 2p > 1.25 bohr 3d 1.25 bohr 4f 1.25 bohr ). > This is my input file: > SystemName SiC molecule > SystemLabel SiC > NumberOfAtoms 2 > NumberOfSpecies 2 > MeshCutoff 50 Ry > %block ChemicalSpeciesLabel > 1 14 Si # Species index, atomic number, species label > 2 6 C > %endblock ChemicalSpeciesLabel > AtomicCoordinatesFormat Ang > %block AtomicCoordinatesAndAtomicSpecies > 0.000 0.000 0.000 1 > 1.870 0.000 0.000 2 > %endblock AtomicCoordinatesAndAtomicSpecies > %block Ps.lmax > C 1 > %endblock Ps.lmax > > SpinPolarized T > DM.MixingWeight 0.01 > MeshCutoff 150 Ry > MD.TypeOfRun Broyden > MD.VariableCell F > MD.NumCGsteps 100 > > PAO.BasisSize DZ > XC.functional GGA > XC.authors PBE > MaxSCFIterations 400 > SCFMustConverge F > > save-rho T > save-delta-rho T > save-total-potential T > save-neutral-atom-potential T > save-hs T > LongOutput T > > I also attached the output file of Si and C atom. So what I'm wondering is > why the binding energy is not right. Is it because of the pseudopotential > I use or the generation method of pseudopotential? Or something else like > basis size, DM.Mixing weight etc? And how to get a successful optimization > of geometry? Thanks > > Yours, > Haoliang >
