vasp-rep
TRANSCRIPT
Introduction to surface calculations
黃聖峰 中正大學物理系2003 年 3 月 22 日
POTCARKPOINTSPOSCARINCAR
Required input files for VASP
pseudopotentail fileBrillouin zone samplingstructural datasteering parameters
Choosing POTCAR file
LDAGGAPAW_LDAPAW_GGAPAW_PBE(VASP4.5)
Check following line in POTCAR LEXCH= CA or 91 GGA= LPAW= T
PAW Ni_pv 16.0000000000000000 parameters from PSCTR are: VRHFIN =Ni: LEXCH = 91 EATOM = 3503.4980 eV, 257.4998 Ry
TITEL = PAW Ni_pv LULTRA = F use ultrasoft PP ? IUNSCR = 1 unscreen: 0-lin 1-nonlin 2-no RPACOR = 1.500 partial core radius POMASS = 58.690; ZVAL = 16.000 mass and valenz RCORE = 2.000 outmost cutoff radius RWIGS = 2.000; RWIGS = 1.058 wigner-seitz radius (au A) ENMAX = 367.921; ENMIN = 275.941 eV RCLOC = 1.301 cutoff for local pot LCOR = T correct aug charges LPAW = T paw PP EAUG = 703.443 DEXC = .000 RMAX = 2.404 core radius for proj-oper RAUG = 1.300 factor for augmentation sphere RDEP = 2.103 core radius for depl-charge QCUT = -5.200; QGAM = 10.400 optimization parameters
GGA tag in INCAR: (default is determined by POTCAR) PB Perdew-Becke PW Perdew-Wang 86 LM Langreth-Mehl-Hu 91 Perdew-Wang 91 PE Perdew-Burke-Ernzerhof (PBE) RP revised PBE
If doing spin polarized PW91 calculations :ISPIN = 2MAGMON = (initial magnetic moment for each atom)GGA = 91VOSKOWN = 1S.H. Vosko, L. Wilk and M. Nusair, Can. J. Phys. 58, 1200(1980)
“typical” values (never trust them!) Metals (9x9x9)/atom Semiconductors (4x4x4)/atom
k-points for bulk 0Monkhorst11 11 11 0 0 0
commentautomatic generation (= 0)Monkhorst or Gammapoint (centered)mesh parametershift
Setting KPOINTS file
For molecules or atoms (large supercells) use 1 x 1 x 1 (Γ)
For surfaces (one long direction) use 2-D Brillouin-zone, nk1 x nk2 x 1
IALGO = 38(blocked Davidson algorithm)IALGO = 48(RMM-DIIS)IALGO = 38 for 5 initial steps than 48 after ions are moved: 38 for 1st step than 48
ALGO-tag determine how the wavefunctions are optimized
◎VASP4.5 does not support IALGO = 8 (preconditioned conjugated gradient) for copyright reasons◎RMM-DIIS is 1.5 to 2 times faster, but Davidson is more stable.◎Eigenstates can be missed using RMM-DIIS for large system.◎If the Davidson algorithm is used for the first steps, there is praticallynodanger of missing eigenstates.
NormalVery_FastFast
Searching the optimal lattice parameter
(1) Automatic batch jobINCAR: ISTART = 0 ICHARG = 2
initial charge-density from ovelapping atoms
#! /bin/bashfor a in 2.3 2.4 2.5 2.6 2.7 2.8 2.9 3.0 3.1 3.2 3.3 3.4docat >POSCAR <<!bcc Fe $a -0.5 0.5 0.5 0.5 -0.5 0.5 0.5 0.5 -0.5 1direct 0.0 0.0 0.0!vaspE=`tail -1 OSZICAR`echo $a $E >> SUMMARYdone
Sample unix bash script for volume scan
2.3 1 F= -.33734901E+01 E0= -.33734901E+01 d E =0.000000E+00 mag =0.111346E+002.4 1 F= -.54258528E+01 E0= -.54258528E+01 d E =0.000000E+00 mag =0.713446E+002.5 1 F= -.67487749E+01 E0= -.67487749E+01 d E =0.000000E+00 mag =0.136680E+012.6 1 F= -.75334410E+01 E0= -.75334410E+01 d E =0.000000E+00 mag =0.169207E+012.7 1 F= -.79469252E+01 E0= -.79469252E+01 d E =0.000000E+00 mag =0.193771E+012.8 1 F= -.81030668E+01 E0= -.81030668E+01 d E =0.000000E+00 mag =0.211534E+012.9 1 F= -.80790747E+01 E0= -.80790747E+01 d E =0.000000E+00 mag =0.226119E+013.0 1 F= -.79505898E+01 E0= -.79505898E+01 d E =0.000000E+00 mag =0.253218E+013.1 1 F= -.77562691E+01 E0= -.77562691E+01 d E =0.000000E+00 mag =0.264343E+013.2 1 F= -.75101095E+01 E0= -.75101095E+01 d E =0.000000E+00 mag =0.274595E+013.3 1 F= -.72343924E+01 E0= -.72343924E+01 d E =0.000000E+00 mag =0.283323E+013.4 1 F= -.69416562E+01 E0= -.69416562E+01 d E =0.000000E+00 mag =0.290679E+01
Result of volume scan
(2) Relaxing the structureINCAR: ISIF = 3 IBRION = 2 NSW = 40 PREC = high (VASP4.4) ENMAX = 1.3*default value in POTCAR EDIFF = 10E-5 (smaller than default value)
Ex. diamond Si from volume scan a = 5.488 Å from structure relaxation a = 5.465 Å (difference is due to the Pulay stress)
BCC Fe from volume scan a = 2.849 Å from structure relaxation a = 2.827 Å from FLAPW (WIEN97, volume scan) a = 2.832 Å
Building surfaces(1)asymmetric setup (2)symmetric setup
Fixed layers(bulk)
coordinatesare optimized
unit cell
vacuum
FCC (100) fcc – (100) surface – 5 layers3.55 .500000 .500000 .000000-.500000 .500000 .000000 .000000 .000000 5.000000Selective DynamicsCartesian .00000 .00000 .00000 F F F .00000 .50000 .50000 F F F .00000 .00000 1.00000 F F F .00000 .50000 1.50000 T T T .00000 .00000 2.00000 T T TorDirect .00000 .00000 .00000 F F F .50000 .50000 .16667 F F F .00000 .00000 .33333 F F F .50000 .50000 .50000 T T T .00000 .00000 .66667 T T T
FCC (111) fcc – (111) surface – 5 layers3.55 .707106 .000000 .000000-.353553 .612372 .000000 .000000 .000000 5.1961524Selective DynamicsCartesian .00000 .00000 .00000 F F F .00000 .40825 .57735 F F F .00000 .20412 1.15470 F F F .00000 .00000 1.73205 T T T .00000 .40825 2.30940 T T TorDirect .00000 .00000 .00000 F F F .33333 .66667 .11111 F F F .66667 .33333 .22222 F F F .00000 .00000 .33333 T T T .33333 .66667 .44444 T T T
Ex. FCC (111) c(2x4)
Ex. BCC (100) p(2x2)
Adsorbing atoms / molecules Surface reconstruction
Surface energy
)(21
bulkatomsurf ENE
Geometry
getting relaxed structure from CONTCAR
Relaxation of surface layers : %100
idea
ideai
d
dd
Heat of formation of overlayers of A on substrate B
2
)2( )()()(2)()(
AbulkBslabAnBslabAn
nEEEH
(Should use the same energy cutoff for each calculation)
WIEN97F (ev) E (ev) F (ev) E (ev) E(Ry)
w-bulk -12.86 -12.85 -12.77 -12.77 -32332.27w5-100 -59.18 -59.13 -58.95 -58.89 -161660.99w5-110 -60.66 -60.62 -60.66 -60.62 -161661.17w7-111 -81.96 -81.89 -81.66 -81.59 -226325.32
WIEN97surf-F (ev) surf-E (ev) surf-F (ev) surf-E (ev) surf-E (ev)
w5-100 2.56 2.57 2.46 2.48 2.54w5-110 1.82 1.83 1.61 1.62 1.34w7-111 4.03 4.04 3.88 3.90 3.98
relax(%) WIEN97w5-100 -10.92
1.67w5-110 -2.63
1.39w7-111 -13.13
-14.676.55
VASP-PP VASP-PAW
VASP-PP VASP-PAW
VASP-PP VASP-PAW
-11.062.14 1.87
-10.42
6.74-15.496.68
-2.401.20-13.18
-16.02
-2.121.17-15.01
MethodHeat of
Formation (ev)
Co/7W(111) 1.23 -0.03 -0.50 -37.9 -11.1 5.4 0.32Co/7W(111) 1.68 1.04 0.01 0.01 -8.0 -47.9 13.7 -3.5 2.93Co/7W(111) 1.66 1.86 1.49 -0.01 0.57 -29.5 -12.4 -14.7 -7.4 -0.2 0.1
Co/7W(111) 1.13 -0.04 -0.57 -39.7 -12.3 5.7 0.3Co/11W(111) 0.68 0.03 -0.51 -42.1 -11.3 3.7 -6.0 6.7 0.02Co/7W(111) 1.70 1.09 -0.01 0.00 -12.8 -48.6 15.8 -2.3 2.0
2Co/11W(111) 1.70 0.93 -0.02 0.07 -12.3 -50.0 17.0 -3.5 -3.6 3.9 0.03Co/7W(111) 1.66 1.87 1.64 -0.01 0.45 -31.1 -19.4 -3.7 -7.4 0.6 2.1
3Co/11W(111) 1.62 1.83 1.58 0.03 0.54 -32.2 -18.4 -4.9 -6.8 -0.1 1.8 -0.7 0.0
Co/7W(111) 0.86 -0.02 -0.49 -42.8 -11.2 6.6 -0.5Co/11W(111) 0.01 0.00 -0.46 -44.0 -10.5 3.7 -5.6 4.8 1.92Co/7W(111) 1.55 0.84 0.00 0.06 -9.5 -55.3 18.4 -2.0 2.1
2Co/11W(111) 1.56 0.63 0.00 0.15 -9.8 -54.2 18.3 -3.3 -4.5 6.0 -1.23Co/7W(111) 1.58 1.75 1.39 0.03 0.59 -26.1 -15.1 -12.8 -4.3 -0.3 2.6
3Co/11W(111) 1.54 1.69 1.18 0.01 0.62 -25.8 -11.9 -17.7 -2.4 0.2 1.5 1.1 -0.8
k-points: d=
Moment Relaxation (percentage of d)
WIEN97-FLAPW
VASP-PP
0.915(A) for VASP-PP0.919(A) for VASP-PAW
VASP-PAW
12x12x1 0.919(A) for WIEN97
Method Heat of Formation(eV)
Co/5W(110) 1.40 -0.03 -0.10 -15.2 -2.2 1.9
2Co/5W(110) 2.00 1.69 0.02 0.28 -25.2 -8.0 2.0 2.8
Co/5W(110) 1.27 -0.02 -0.19 -18.9 1.2 0.6
Co/7W(110) 1.25 -0.02 -0.19 -19.0 1.1 0.8 0.8
2Co/5W(110) 1.90 1.42 -0.02 0.16 -28.4 -12.4 0.7 1.1
2Co/7W(110) 2.00 1.41 -0.02 0.16 -28.4 -12.4 0.8 1.2 1.0
Co/5W(110) 0.70 -0.03 -0.17 -21.4 1.5 0.7
Co/7W(110) 0.76 -0.02 -0.16 -21.3 1.2 0.8 0.6
2Co/5W(110) 1.92 1.30 -0.02 0.23 -30.3 -14.0 0.4 0.9
2Co/7W(110) 1.92 1.33 -0.02 0.23 -30.1 -13.2 0.8 1.6 1.4
k-points: d=
2.252(A) for VASP-PAW
WIEN97-FLAPW
VASP-PP
VASP-PAW
12x12x1 for VASP 2.250(A) for WIEN97
Moment Relaxation (percentage of d )
16x16x1 for WIEN97 2.242(A) for VASP-PP
Local Density of states
INCAR: RWIGS = γ Å (works only for NPAR = 1 or serial version) LORBIT = 11 (only for PAW) ISMEAR = -5 (use tetrahedron for DOS calculations) NPAR = 1Output file : DOSCAR (energy, s-dos, p-dos, d-dos for each atom) PROCAR (dos for each band and k-point)
LORBIT-tag
Band structure (after selfconsistent run)
1. Setup a k-point list along specific axis in KPOINTS2. Set ICHARG = 11, read charg density from CHGCAR3. Set FFT grid parameter manually to same value, to make sure that CHGCAR file is read properly (NGX, NGY, NGZ, NGXF, NGYF, NGZF)4. Analyze and plot data in EIGENVAL after job done
k-points for band structure 161C .00000 .00000 .00000 1.00000 .02500 .00000 .00000 1.00000 .05000 .00000 .00000 1.00000…
Ex. KPOINTS file for band structure
Si-bulk Band Structure
WIEN2k VASP
using 2-D Brillouin-zone for surfacesto get projected bandstructure
k-points for surface band structure 13R .00000 .00000 .00000 1.00000 .12500 .00000 .00000 1.00000 .25000 .00000 .00000 1.00000 .37500 .00000 .00000 1.00000 .50000 .00000 .00000 1.00000 .50000 .12500 .00000 1.00000 .50000 .25000 .00000 1.00000 .50000 .37500 .00000 1.00000 .50000 .50000 .00000 1.00000 .37500 .37500 .00000 1.00000 .25000 .25000 .00000 1.00000 .12500 .12500 .00000 1.00000 .00000 .00000 .00000 1.00000
Ex. K-points along line Γ-X-M-Γ
Work function
1. Search “E-fermi” in OUTCAR to get fermi-level2. Analyze and plot data in LOCPOT
INCAR: LVTOT = .TRUE.
Output file : LOCPOT (same format as CHGCAR) WRITE(IU,FORM) (((V(NX,NY,NZ),NX=1,NGX),NY=1,NGY),NZ=1,NGZ)
LOCPOT only contain electrostatic part of potential,if exchange correlation potential is to be included,change one line in main.F :
! comment out the following line to add exchange correlation! INFO%LEXCHG=-1
Cou
lomb
poten
tial (eV) C
oulom
b p
otential (eV
)C
oulom
b p
otential (eV
)
Cou
lomb
poten
tial (eV)
W(100) W(110)
W(111) W(211)
Fermi energy
Z-axis(Å) Z-axis(Å)
Z-axis(Å) Z-axis(Å)
Φ
Surface adsorption
Ni – (111)3.53 .707106 .000000 .000000-.353553 .612372 .000000 .000000 .000000 5.19615245 1 1Selective DynamicsDirect .00000 .00000 .00000 F F F .33333 .66667 .11111 F F F .66667 .33333 .22222 F F F .00000 .00000 .33333 T T T .33333 .66667 .44444 T T T .33333 .66667 .54029 T T T .33333 .66667 .60299 T T T
Ex. CO/Ni(111)
POSCAR: INCAR:EXMAX = 400ISMEAR = -5LORBIT = 1
NPAR = 1RWIGS = 1.40 1.29 1.11
LVTOT = .TRUE.IDIPOL = 3LDIPOL = .TRUE.(dipol corrections)
COcleantotalads EEEE
Adsorption energy
(use the same energy cutoff)
Dipol correction
cleanE = -25.730 eV (270eV)= -25.741 eV (400eV)
COcleantotalads EEEE = -40.830 + 25.741 + 14.833= -0.256 eV
Result of CO/Ni(111)
F
CO clean
= 1.66 eV
Vacuum-potential at 8.15 / 6.76 eV
= 6.49 eV = 5.10 eV
But in Ni(111) calculation
clean = 5.24 eV
due to too small vacuum in CO/Ni(111)
Other new feature and advantages
VASP4.5
◎non-collinear magnetic structure and spin orbital coupling◎PREC=Low/Medium/Normal/Accurate/High◎single precision WAVCAR (smaller)◎new MPI communication layer and FFT routines
VASP4.6
◎support L(S)DA+U calculations◎report orbital moment◎new output file “vasprun.xml” used in new vasp utility “p4v” (python for vasp)