Creating DLC via melting and cooling

Pressures fluctuate, often by a lot. That's the nature
of MD.



I am very sorry to have caused so much trouble. I have changed my approach according to the literature where I am initializing the system with simple cubic lattice at 5000K. Set the initial density to 2.9 g/cm3, equilibrate the system and cool the system according to the time given in the literature. I am trying out with the system size of 125 atoms according to the literature as well.

I have also added the dihedral contribution for the Airebo potential as well.

However, during the initial equilibration process, the temperature fluctuates wildly even though I am using the nvt fix.
The Tdamp values used is 100 times of the timestep values while the damp value is 0.8. The equilibration is as shown below.

fix fix_1 all nvt temp 5000 5000 0.01 drag 0.8

May I know what could be the problem? Please advice on this matter.

Thank you very much

However, during the initial equilibration process, the temperature
fluctuates wildly even though I am using the nvt fix.

Can you quantify "wildly"? In a small system like yours and starting
from a highly nonequlilibrium structure such as cubic carbon
temperature fluctuations can be expected. How many steps are you

Hi, sorry for taking a long time to reply.

The fluctuation is such that for a system of 125 particles, the temperature can suddenly jump to 9674 from 5800+ and then go up to even 57999140. Soon afterwards it might even drop down to 900+ K even though the fix nvt is at temperature of 5000K.

Its quite hard to quantify it but it can be seen from the attached log file that the temperature is fluctuating a lot instead of keeping near 5000K.

In addition, I tried to look at the system using VMD but somehow the atoms are lost even though the boundary condition is periodic.

Sorry to keep asking questions.


DLC (2.95 KB)

log.lammps (141 KB)

Also, it is interesting to note that the magnitude of the temperature that is fluctuating is dependent on the lattice constant. For a larger lattice constant, that is for a crystal that is less dense, the temperature does not blow up to such a large value nor does it drops down to zero.

In addition, the simulation done for a large system with the same lattice constant of 1.9 seems to cause the temperature to become 0 during the initial equilibration process as well as the cooling process.

To add on the info which I previously forgot to mention, the number of steps run is 8000 for all except during cooling which was done in 5000 steps. The timestep used is 0.0001.

Thank you very much.

Your temperature is bogus because your dynamics is bogus. You should study the mailing list archives since this comes up over and over again. People blame the thermostat, but their set is just broken. MD requires somew basic considerations about stability of numerical integrators and the properties of potentialf as well as the thermodynamic boundary conditions. Doing MD without that is creating entropy and playing the lottery.