I am trying to simulate water by using AIMD simulation with VASP.
The LDA is used and about 70 water molecules are placed in a box. The following parameters are my setting in INCAR. However, the results seem abnormal. A few water molecules prefer to dissociate into H and OH after 5ps. Does anyone know how to solve this problem? Thanks in advance!!!
PREC = low
ENCUT = 300.000
IBRION = 0
NSW = 6000
POTIM = 0.5
SMASS = 2
TEBEG = 300
TEEND = 300
ISIF = 2
ISYM = 0
ALGO = V
ISMEAR = -1
SIGMA = 0.0258
LREAL= Auto
NELECT=1392 #1393
NELMIN=6
Most likely there is a problem with the energies of the system. Have you double checked your molecule's topologies to make sure they're correct?
YEs，Adam, I have checked all the positions. I feels LDA is not proper for water, is that so?
Several questions from the top of my head: How does the temperature of your system behave during the course of the simulation? Have you started from an optimized geometry? Is the dissociation homo- or heterolytic? Can you check this result against the very same setup, but with a different XC functional?
LDA performs quite poorly, e.g. it has virtually no long-range dispersion, so it will give bad results wherever intermolecular interactions are involved. However, it should not be poor enough to cause water to dissociate at low temperature, unless there is something quite unexpected happening to your system.
Thank you for your reply, and this problem is surely caused by LDA after check from our results.
Then this might also be a convergence issue. I haven't specifically worked with VASP, but sometimes - depending on your settings - your SCF might not converge in the given number of iterations, and then the simulation might proceed with an unoptimized wavefunction, giving rise to weird forces and unphysical behavior of the system. Is that the case? Or can you give more details on what exactly is happening to your energies, temperatures etc. at the time the dissociation occurs.
A crude workaround would be to restrain O-H bond lengths from above, i.e., with a wall, so that they do not dissociate at least during a short equilibration period. However, if the setup is OK, 5 ps should be enough to equilibrate such a small system (of course, this will depend e.g. on your thermostat settings).
Hi, Miłosz,
The energy terms given in VASP is like this,
T= 309. E= -.21321262E+04 F= -.22241507E+04 E0= -.22241507E+04 EK= 0.13907E+02 SP= 0.78E+02 SK= 0.53E-01
For my systems, the E is keeping decreasing but the F term is converged after 2-3 ps. I think the latter is the criteria in AIMD. The temperature is also easily converged after 2 ps. As to the SCF problem, I paste the last output, and prefer to think the wavefunction is also converged. What's your opinion?
- Badges
- Science topic