Dear All,
My name is Evgeny, and I am undergraduate student working on my summer project on LAMMPS. I am trying to simulate ice formation using the lammps ice_crystal example from moltemplate. In my simulation, water initially has an organized crystal structure, then I heat it up using fix nvt command. For some reason, temperature in my simulation goes from 0 up to 2300 degrees (which seems very high), and then it cools down back to 0. My two main questions are:
-
Is it possible to properly simulate water crystallization in LAMMPS?
-
Is there any way to precisely control the temperature of the system, because temperature parameters of run nvt command seem not to work properly?
Attached are the simulation files: data file system.data, input file run.nvt, and output trajectory traj_nvt.lammpstrj
I will appreciate any kind of help
Thank you,
Evgeny
run.nvt (1.62 KB)
system.data (13.8 KB)
Hello Evgeny
Please read about how to do basic MD simulations.
Arun
Dear All,
My name is Evgeny, and I am undergraduate student working on my summer
project on LAMMPS. I am trying to simulate ice formation using the lammps
ice_crystal example from moltemplate. In my simulation, water initially has
an organized crystal structure, then I heat it up using fix nvt command. For
some reason, temperature in my simulation goes from 0 up to 2300 degrees
(which seems very high), and then it cools down back to 0. My two main
questions are:
1. Is it possible to properly simulate water crystallization in LAMMPS?
yes, but it can be very tricky due to crystallization being an
activated process. it is very easy to get "supercooled" water (ever
seen the experiment, where people put a seltzer bottle into the
freezer and remains liquid until it is opened and gas bubbles
appear?). there is a substantial amount of published literature on
this subject.
2. Is there any way to precisely control the temperature of the system,
because temperature parameters of run nvt command seem not to work properly?
please talk to the person supervising your project. what you are
asking about has little to do with using LAMMPS, but everything with
understanding the details of molecular dynamics simulations and how
thermodynamic properties behave differently on the microscopic level
versus macroscopic experiments. this requires some insight into
statistical mechanics and more, but also an understanding of the whole
simulation process with equilibration and how to setup simulations
effectively.
the lammps-users mailing list is primarily for questions about
specific commands.
so to answer your specific question about the fix nvt command: please
compare your parameters with the recommendations and explanations
given in the LAMMPS manual for the fix nvt command. you are using a
time constant of 10000 (femtoseconds), which is quite large and
corresponds to an extremely weak coupling of the thermostat to your
system so that the target temperature under normal circumstances (i.e.
if otherwise energy is well conserved and the system already in
equilibrium) will be reached only after 10 picoseconds. please try
what the documentation recommends.
axel