# File for determining d/dk for LiNbO2 # # LiNbO2 should belongs to space group P63/mmc, group # number 194. # # Unit cell is hexagonal, with # a = 2.90 Angstroms # c = 10.46 Angstroms # # According to paper by Gerd Meyer and Rudolf Hoppe, we have # Li at 2a # Nb at 2d z = 0.25 # O at 4f z = 0.1263 # Relaxation of atomic positions has led to # z = 1/8 for the oxygens. # # We will need three datasets, one to calculate the # self-consistent wave function, and the second to calculate # the actual d/dk values. The third will calculate # the electric field perturbation and the born effective # charges. ndtset 3 ############################################################### # Dataset 1 ############################################################### tolvrs1 1.0e-09 # We want the potential to be really # well converged. iscf1 5 # Do calculation self-consistently. ############################################################### # Dataset 2 ############################################################### rfelfd2 2 # Tell abinit to actually to the d/dk calculation. rfdir2 1 1 1 # Do all three directions (maybe we only have to # do a and c directions? I dunno.) nqpt2 1 # Apparently we need a q point to do this. This qpt2 0 0 0 # corresponds to the perturbation by which the # d/dk values are calculated. getwfk2 -1 # We want to get the wave function from the 1st # dataset. #kptopt2 2 # Most of the examples seem to use only time-reversal # symmetry to reduce the number of k-points, so that's # what I'll do here. iscf2 -3 # Do a non-self consistent calculation. tolwfr2 1.0e-22 # Strict tolerance on convergence of the # wave function, since these results will # be used in later calculations. ############################################################### # Dataset 3 ############################################################### #rfphon3 1 # This will tell abinit to calculate response # to phonons (ie. atomic displacements) #ratpol3 1 8 # This would displace all the atoms rfelfd3 3 # Perform electric field calculation rfdir3 1 1 1 # Set the directions nqpt3 1 # Apparently we need a q point to do this. This qpt3 0 0 0 # corresponds to the perturbation by used in the # d/dk calculation getwfk3 -2 # We want to get the wave function from disk getddk3 -1 # Get the 1st order wave functions from disk #kptopt3 2 # Most of the examples seem to use only time-reversal # symmetry to reduce the number of k-points, so that's # what I'll do here. iscf3 5 # Self consistent calculation. tolvrs3 1.0e-8 # Sorta strict tolerance on the residual potential ############################################################### # Common parameters ############################################################### #Definition of the unit cell acell 5.3808585426 5.3808585426 19.262436687 # Lengths of the primitive vectors (in Bohr) angdeg 90 90 120 # Specify angle between primitive translations #Definition of the atom types ntypat 3 # Three types of atoms, Li, Nb, O znucl 3 41 8 # We have 3-Li, 41-Nb, 8-O #Definition of the atoms natom 8 # I am treating the unit cell as containing # Li2Nb2O4. This will make it much simpler when # going to H(0.5) LiNbO2 typat 1 1 2 2 3 3 3 3 # First two atoms are Li, Second two are Nb, # Last four are O. xred # I will give the atomic coordinates in reduced c # coordinates. 0 0 0 0 0 1/2 1/3 2/3 3/4 2/3 1/3 1/4 1/3 2/3 0.1211522996 2/3 1/3 -0.1211522996 2/3 1/3 0.6211522996 1/3 2/3 0.3788477004 #Definition of the planewave basis set ecut 50.0 # We seem to have convergence here. #Definition of the k-point grid kptopt 1 # Automatic k-pt generation. Time-reversal symmetry # only. ngkpt 14 14 14 # We seem to have convergence here. nshiftk 1 # number of k-point shifts. This is recommended from # the abinit documentation for hexagonal symmetry shiftk 0.0 0.0 0.5 occopt 3 # There was some messages about metallic character # so we will treat this as a metallic substance # for this calculation. tsmear 0.001 # Set the temperature for F-D distribution to 315K #output eigenvalues in ev. enunit 1 #Definition of the SCF procedure nstep 2000 # Maximal number cycles nband 25 # number of bands #diemac 1.0 # Although this is not mandatory, it is worth to # precondition the SCF cycle. The model dielectric #diemix 0.5 # function used as the standard preconditioner # is described in the "dielng" input variable section. # Here, we follow the prescriptions for molecules # in a big box