Large Total Energy Drift in NVT MD Simulation

Message

anoop · #1 Post by **anoop** » Sun Sep 15, 2013 1:47 pm

Hi,

I am doing NVT simulations. 32 molecules of water + Zinc atom, There is charge (+2) in the simulation box (L=9.96 Angstroms). Calculations are done at Gamma Point only. Please find my INCAR below. The energy drift is continuous and its about 0.2 eV/ps. Drift looks quite large.I am using the SCF optimized structure as my initial input. Rest of the parameters i.e Temperature, P.E & K.E with respect to time looks ok.

Any suggestions in this regard would be highly appreciated.

SYSTEM = Water with Zn
!
!---Startparameter_for_this_run---
PREC = Normal !Normal,Medium,High,Low,Accurate
ISPIN = 1 !Spin polarized calculation: 2-yes, 1-no
!
!---Electronic_Relaxation1---
ENCUT = 450
NELM = 100 !Maximum number of self consistent steps for the electronic loop
NELMIN = 4 !Minimumal number of self consistent steps for the electronic loop
NELMDL = -4 !nb_ELM_before_charge_update
EDIFF = 1.0E-05 !stopping-criterion_for_ELM [eV]
ALGO = Very FAST ! Normal
MAXMIX = 50
LREAL = A
LVDW= .TRUE.
!
!---Ionic relaxations---
NSW = 8000 !Maximum number of steps
IBRION = 0 !ionic relax:0-MD,1-DIIS,2-CG
LCORR = .TRUE.
ISIF = 2 !change_int_param:2-ion,3-volume,4-shape
!
!
NPAR = 4
IWAVPR = 11
NELECT = 266
MDALGO = 2
!---DOS_related_values---
ISMEAR = -1 !-5:tetrahedron,-1:fermi,0:Gaussian,N:Methfessel-Paxton(Metal)
SIGMA = 0.1 ! //0.1(with ISMEAR = 1) !smearing_width_ISMEAR
!
!---Write_flags---
LWAVE = F !write_WAVECAR
LCHARG = F !write_CHGCAR
LVTOT = F !write_LOCPOT,local_potential
LELF = F !write_electronic_localiz_function(ELF)
NWRITE = 0 !print the band energy only one ionic step
!---MD setting---
TEBEG = 350
TEEND = 350
!---Canonic (nose) MD
SMASS = 3
NBLOCK = 1
POTIM = 1.5

Request for your suggestions and comments on the INCAR.
Many Thanks in advance!

Thanking You,

Best Regards,
Anoop

<span class='smallblacktext'>[ Edited ]</span>

anoop · #2 Post by **anoop** » Tue Sep 17, 2013 11:50 am

Dear all,

Any suggestions in this regard would be highly appreciated.
I tried with SMASS=0 also, It also shows quite similar drift.

Regards,
Anoop

#3 Post by **admin** » Wed Sep 18, 2013 11:35 am

The drift indicates that the system is not equilibrated properly. A liquid requires much longer simulation times not just 12 ps like you are doing. You can also increase the relative mass (via tuning SMASS) to slow down atomic motions.

anoop · #4 Post by **anoop** » Thu Sep 19, 2013 11:38 am

many thanks for your suggestion.