2
$\begingroup$

In my AIMD simulations with the NVT ensemble at T= 300K (Nosé–Hoover thermostat)for a Pd (111) surface (2*2, 6 layers) and vacuum sizes of 14 Å and 18 Å, All other INCAR tags are same(only NGZ for 14Å (123) and 18Å (143)), I've observed two distinct graph time step versus temperature. I'm seeking clarification on the accuracy of these graphs. If accurate, can it be inferred that increasing the vacuum size leads to a reduction in Temperature fluctuation? vacuum 14Å pd(111)surface time step vs temperature graph at 14Å vacuum vacuum 18Å pd(111)surface time step vs temperature graph at 18Åvacuum CONTCAR file for 14 Angstrom CONTCAR file for 18 Angstrom

$\endgroup$

1 Answer 1

2
$\begingroup$

"Ensemble" is a pretty strong word for a system with fewer atoms than the cards in a poker deck of cards. The Nose-Hoover thermostat attempts to redistribute energy between a system's degrees of freedom to achieve a thermal distribution -- but it fails for small systems with strongly harmonic modes.

My intuition would say that the Nose-Hoover thermostat fails to thermalize your first system because of the vibrational mode associated with the whole system's self-interaction across the vacuum gap. That mode is clearly resonant and harmonic enough to maintain its harmonic oscillations -- whereas in the second system that mode is less resonant and is thus readily thermalized.

The phenomenon is magnified in both systems by how small your unit cell is, which leaves your phonon spectrum much closer to discrete than continuous, and thus hampers thermalization.

$\endgroup$
5
  • $\begingroup$ Thank you @Shern Ren Tee , So Are you saying that the second one is more accurate because it has less periodic interactions? I am asking again because I have never seen this type of graph for equilibrating Surface Temperature Through NVT Ensemble Ab Initio Molecular Dynamics. $\endgroup$ Feb 1 at 10:10
  • $\begingroup$ To be precise, the spurious forces from the slab's self-interaction across the z-periodic gap are smaller in the second system than in the first. It is impossible for me to know whether the spurious forces in the second system are small enough for you to accurately calculate what you want to calculate. $\endgroup$ Feb 2 at 4:35
  • $\begingroup$ I am adding two CONTCAR file of both 14 and 18 angstrom vacuum in my qustion. now can you adding something in your answer. $\endgroup$ Feb 2 at 6:25
  • $\begingroup$ I answered purely in terms of statistical mechanics. I don't routinely do DFT calculations. But even if I did, it is your responsibility to make sure your calculations converge and are physically meaningful. If a reviewer has problems with your results two months from now, you will not be able to reply "well a stranger on the Internet said it was okay!" $\endgroup$ Feb 2 at 10:20
  • $\begingroup$ yeah, you are absolutely right. Thank you so much @Shern Ren Tee $\endgroup$ Feb 2 at 11:01

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .