Young modulus of polyethylene

Hello lammps users,

I am simulating a linear melt polyethylene structure . The structure contains 23 chains of 180 carbon atoms each.

I have modeled the system with the TraPPE-UA potential for united atoms and equilibrated it using the method of “phantom chains simulation” at 300 K and 1 atm.
My current work includes finding its young modulus. For that purpose i tried going step by step enforcing a standard deformation using specific uniaxial stress with the command : fix 1 all npt temp 300 300 1 x -145.2 -145.2 1000 y 1.0 1.0 1000 z 1.0 1.0 1000 drag 1. For some reason the averaged pressure over time does not converge to the specified values, so i cant calculate the correct averaged strain for the stress i have enforced…
These are the results of the pressure after averaging every 100000 steps (step=1femtosecond) for 700000 steps using the command fix 4 all ave/time 1 30000 100000 :

pxx pyy pzz (atm)

-45.9968 -48.1742 -49.0495
-49.4132 -47.4046 -47.7341
-48.6964 -47.5332 -45.7315
-47.2617 -49.2964 -46.8012
-47.4547 -47.5195 -47.2357
-48.0259 -49.923 -45.793

-48.3982 -46.9983 -47.4239

Any suggestions concerning this matter are highly appreciated…

After the failure of this method i tried calculating the young modulus through a constant engineering strain…
I decoupled the direction x from the npt equations of the motion and i used the fixed deform command:

fix 1 all npt temp 300 300 1 z 1 1 1000 y 1 1 1000 drag 2

fix 2 all deform 1 x erate 0.00001 units box remap x

After visualizing the deformation, it seems to be working in the correct manner, however the output pressure seems to be fluctuating a lot (something reasonable for solids after reading many posts in the mail list) so i dont really have an accurate pressure value at each step of deformation…

I was wondering if there is a way to get an averaged value of the pressure at each strain value without using the npt command as i applied above or if there is a standard and more accurate way of calculating the young modulus of polyethylene…

• I am using periodic boundary conditions, units are real and the box is cubic and orthogonal…

Thank you very much in advance

Dedes Grigorios
Nacional Technical University of Athens

Fix ave/time can be used to time average just about any
value LAMMPS produces. E.g. you can use it to get
an average pressure over a time window that corresponds
to one of your strain values.

Steve

Thank you very much Steve for your suggestions,

I think this is the best way to go.However, there is one more issue regarding the method you proposed that troubles me.
I try to use an engineering strain rate as small as possible (for example 10^-8) in order to get a good averaged value of the pressure around a specific strain value
(for example an average for 200000 steps). When i use a very small value of erate after some steps the y and z direction (which are barrostated)
seem to increase towards infinity…When i increase the strain value (to 10^-6 for example) the deformation seems to be happening in the correct manner

(the x direction which i have decoupled from barostat increases according to strain rate and the other too gradually decrease over time)…
Could you provide me an explanation regarding this matter?

Στις 7:27 μ.μ. Τρίτη, 29 Μαΐου 2018, ο/η Steve Plimpton [email protected] έγραψε:

Fix ave/time can be used to time average just about any
value LAMMPS produces. E.g. you can use it to get
an average pressure over a time window that corresponds
to one of your strain values.

Steve

I don’t know. Assuming your 3 box lengths are all correct,
the pressure will simply go where it wants to go. If you
can do something reasonable at a high strain rate, it’s
unlikely there should be a problem with a lower strain rate.

Steve