VASP Guideline

Basic notes on how to use VASP

General comments regarding VASP:

In order to run VASP, you need 4 input files: INCAR (contains calculation parameters), POSCAR (contains the geometry of the system), POTCAR (contains psuedopotential information), KPOINTS (contains which kpoint you want to use).

INCAR:

1. ISTART = Initial guess of the orbitals

• | 0=new, 1=continue from WAVECAR with const E cut-off, 2=continue from WAVECAR with const basis |

2. ICHARG = Initial guess of charge density

• | 0=from orbitals, 1=read from CHGCAR, 2=from atomic charges, +10=non-scf(const), 11=to get evalue for band plots or DOS read from CHGCAR |

3. LCHARG = Whether to save CHGCAR or not

• | .FALSE. or .TRUE. |

4. SYSTEM = Name of the system ( for example pt4sn3-sio2)

5. ENCUT = Plane Wave cut-off Energy in eV

6. LREAL = Real space projection

• | .FALSE. or .AUTO. |

• Note: For large supercells or hybrid functionals 'Auto' is recommended (faster with a negligible loss in accuracy).

7. ISMEAR = Smearing method

• | 0=gaussian, -1=fermi, -4=tetrahedron, -5=blochl |

8. SIGMA = Broadening of Smearing in eV (width of smearing)

• | recommended: 0.1 |

• Note: For calculating DOS and very accurate total energy calculations (no relaxation in metals) use the tetrahedron method (ISMEAR=-5).

9. NELM = Num of electronic relaxation steps (SCF)

10. NELMIN = Min electronic relaxation steps that should be done between each ionic (geometric) update.

11. EDIFF = Stopping criteria for ELM (electronic minimization, SCF)

• | recommended: 1e-6 |

12. IBRION = Relaxation method

• | 0=MD run, 1=quasi new(RMM-DIIS for geom opt)RECOMMENDED FOR NEB, 2=CG algorithm(for hard relaxation), -1=static, 5-8=freq; 7,8 using DFTPT; 5,6 using finite displacement, 6,8 with sym; 5,7 w/o sym |

• Note: For difficult relaxation problems it is recommended to use the conjugate gradient algorithm (IBRION=2).

• Note: Damped molecular dynamics (IBRION=3) are often useful when starting from very bad initial guesses.

• Note: Close to the local minimum the RMM-DIIS (IBRION=1) is usually the best choice.

13. EDIFFG = Stopping criteria for ionic (geometric) relaxation, IOM (geometry optimizations)

• | Default : EDIFFx10, recommended: -0.01 |

• Note: If EDIFFG is negative it has a different meaning: In this case the relaxation will stop if all forces are small than the |EDIFFG|. This is usually a more convenient setting. EDIFFG does not apply for MD-simulations.

14. NSW = Num of steps for ionic relaxtion(geometry).

15. ALGO = Algorithm used for SCF (electronic minimization)

• | Normal=Davidson, Fast=RMM, Very Fast=RMM-DIIS, All=good for hybrid functionals |

16. ISPIN = Whether the calculation is spin polarized or not

• | 2=spin polarized, 1=no |

17. NUPDOWN = Num of e with spin up - spin down = -1(default,full relaxation)

• | 0-singlet, 1-doublet, etc |

18. ISIF = Used for relaxing the unit cell

• | 2=relax ions only, 3=ions + volume |

19. LORBIT = Density of States calculations

• | use 11 for projected DOS, LORBIT<10, the spd- and site projected wavefunction character of each band is, evaluated, and the local partial DOS is calculated. LORBIT>=10, RWIG is ignored, works only for PAW method |

20. PREC = Precision of the calculation

• | Accurate or Normal |

21. LWAVE = Save Wavecar or not

• | .FALSE. or .TRUE. |

22. POTIM = Trial step, time-step for ion-motion in MD calculations in fs

23. IVDW = VDW correction

• | recommended: 11 |

   

Functionals:

24. GGA = Override the type of density functional specified in the POTCAR

• | default: PBE = PE |

25. LHFCALC = Switch on Hybrid and Hartree-Fock type calculations

• | .FALSE. or .TRUE. |

   

• HF Setting:
• LHFCALC = .TRUE.
• AEXX = 1.0
• ALDAC = 0.0
• AGGAC = 0

   

• B3LYP Functional Setting:
• LHFCALC = .TRUE.
• GGA = B3
• AEXX = 0.2
• AGGAX = 0.72
• AGGAC = 0.81
• ALDAC = 0.19

   

• PBE0 Functional Setting:
• GGA = PE
• LHFCALC = .TRUE.

   

• HSE06 Functional Setting:
• LHFCALC = .TRUE.
• HFSCREEN = 0.2

POSCAR:

Pt2Si cluster isolated from MgO               ! comment line

1.00000000000000               ! universal scaling factor

12.6256980228000000       0.0000000000000000       0.0000000000000000               ! first Bravais lattice vector

0.0000000000000000       12.6256980228000000       0.0000000000000000               ! second Bravais lattice vector

0.0000000000000000       0.0000000000000000       30.0000000000000000               ! third Bravais lattice vector

Si       Pt               ! elements

1       2               num of each element

Selective dynamics               ! for relaxing some coordinates, F=fixed, T=relax

Cartesian               ! direct (fractional) or cartesian (only first letter is significant)

4.210900000000          4.218200000000          12.610000000000          F       F T

6.205100000000          4.215700000000          13.363500000000          T       T       F

2.216900000000          4.239300000000          13.364400000000          F       F       F

• The positions can be given in direct (fractional) or Cartesian coordinates. Note that the lattice vectors are always scaled by the universal scaling factor.

POTCAR:

KPOINTS:

Bader Charge Calculations:

LAECHG = .TRUE. | for Bader charge analysis; the core charge will be written so that it would be used in Bader | In quantum espresso it is recommended to perform the bader analysis only if PAW pseudopotentials are employed. I made some test using QE and ultra-soft pseudopotentials (what we usually use) and VASP with PAW pseudop. The results are very similar. However, for VASP there is a correction to consider also the core electrons. In this case I obtained different results (and they make more sense), specially with atoms such as B or Al. So, I think this is the best way to proceed:

                     To use this correction you need to include in your INCAR file:
                     LAECHG=.TRUE.

                     Once the calculation is done:
                     ./chgsum.pl AECCAR0 AECCAR2
                     ./bader CHGCAR -ref CHGCAR_sum

AIMD Calculations:

• For MD, we recommend the use of PREC=Normal although PREC=Low yields often satisfactory results.

• MAXMIX = 40, Reuse mixer from one MD step to next

• NELMIN = 4, Minimum 4 steps per time step, avoid breaking after 2 steps

• IBRION = 0

• LREAL = A

• NSW = 10000

• NWRITE = 0

• LCHARG = .FALSE.

• LWAVE = .FALSE.

• TEBEG = 2000, Temperature begin

• TEEND = 2000, Temperature end

• SMASS = 0, -1, 0 = NVT-canonical (Nose) MD, -1 = NVE-micro canonical MD with temperature scaling every 50 steps

• NBLOCK = 50

• POTIM = 1.0, timestep in fs

• POTIM = 1.0

• IWAVPR = 12, Determines how orbitals and/or charge densities are extrapolated from one ionic configuration to the next configuration. MD=12, relax=11

• Note: Use ALGO=Very Fast (RMM-DIIS for electrons) for large molecular dynamics runs.For surface or difficult systems, you might need to increase NELMIN = 8.

NEB Calculations:

• IBRION = 1, recommended

• ALGO = fast

• ICHAIN = 0, (int) Indicates which method to run. NEB (ICHAIN=0) is the default.

• IMAGES = 5, (int) Number of NEB images between the fixed endpoints.

• SPRING = -5.0, (float) The spring constant, in eV/Ang^2 between the images; negative value turns on nudging

• LCLIMB = .TRUE., Flag to turn on the climbing image algorithm (CI-NEB)

• LTANGENTOLD = .FALSE., Flag to turn on the old central difference tangent

• LDNEB = .FALSE., Flag to turn on modified double nudging

• LNEBCELL = .FALSE., Flag to turn on SS-NEB. Used with ISIF=3 and IOPT=3.

Band Structure Calculations:

Work Function Calculations:

LVTOT = .TRUE. | to generate a LOCPOT-file |

IDIPOL = 3 | dipole correction in all x,y,z |

LDIPOL = .TRUE.

LVHAR = .TRUE. | so that the LOCPOT contains only the electrostatic potential and not the entire local potential |

Solvation Calculations:

LSOL = .TRUE. LAMBDA_D_K = 3.0 | Debye screening length | EB_K = 78.4 | Relative permittivity of water | TAU = 0 | Effective cavity surface tension |