I am trying to perform heating and quenching simulations of iron from various temperatures. I took 2000 atom simulation box and performed following methodology

#=====starting from random velocity at 500K  final temp will be 250K ========
velocity all create 500.0 4928459 rot yes dist gaussian
fix 1 all nve
run 10000
unfix 1
# ==== re-relax with npt =========
fix 2 all npt temp 250.0 250.0 0.1 iso 0.0 0.0 1
run 10000
unfix 2
#=========== constant heating=============
fix 2 all npt temp 250.0 600.0 0.1 iso 0.0 0.0 1
run 500000
unfix 2
#==========re-relax at Tmax =============
fix 2 all nvt temp 600.0 600.0 0.1
run 100000
unfix 2
write_restart restart.heat
#================ Quench to 100K ===========
fix 3 all nvt temp 100 100 0.1
run 100000
unfix 3
write_restart restart.quench

I am not sure about the quenching methodology is correct or not. Kindly comment on both heating and quenching process. Timestep :1fs enter image description here


1 Answer 1


In my calculations, what I opt to do is somewhat different. However, I can't comment on whether you method is correct or not. But based on the results which I have obtained, I think that your method could be questionable. My procedure is as follows:

  1. Generate initial temperature and relax sufficiently in the NPT ensemble followed by NVE to make sure that the effects of the random distribution of velocities prescribed initially are lost.
  2. Heat the system in the NPT ensemble at the desired rate by around 50-100K.
  3. Relax the system at constant temperature in the NPT ensemble followed by NVE to ensure the stability of the system (add NVT as well if necessary).
  4. Repeat steps 2 and 3 until the desired final temperature is achieved.
  5. Cool the system from the current (maximum) temperature to the initial temperature using the NPT ensemble in a single run at a desired rate.
  6. The rate of cooling is best determined from literature. However if no published data is available, try several rates and validate with data for amorphous phase (assuming that this is what you are after) from experiments.

You must log in to answer this question.

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