PPTA (Kevlar) buckling problem during imposing a temperature with ReaxFF

Dear LAMMPS users,

I’m doing research about PPTA (Kevlar) using LAMMPS (August 2019) and ReaxFF potential parameter sets developed by Liu et al. (field.reax.lg) and I want to do a tensile simulation of PPTA as a goal.

But I have a problem in imposing a temperature of PPTA which is necessary before the tensile simulation.

I can finish the minimization well, but when I start giving a temperature to PPTA, the structure of PPTA is messed and PPTA chains buckle showing the zig-zag shape.

I attached the OVITO image of PPTA structure before and after imposing a temperature here https://drive.google.com/open?id=1FrjkLCGyRQ1VsEpExtgHheT6gPEwboDC .

I also attached the input file, control file at the end of this mail.

I did many things to solve this messing structure problem.

For example,

1, I always use a simulation cell of size 3×5×2 PPTA unit cells (1680 atoms) to get a quicker reply, but I changed this to 6×10×4 PPTA unit cells (13440 atoms).

2, I once changed a time constant of pressure control as small as possible.

3, Since I didn’t add partial charge information to coordinate file at first, I added the partial charge information.

4, Since I didn’t use the control file at first, I am using the example control file which I got in LAMMPS.

5, I changed the LAMMPS version from August 2017 to August 2019.

6, I changed the way of imposing a temperature from NVT to NPT.

7, I deleted “velocity” command from input files because I got less structural messing problem.

8, I changed the way to control pressure from iso to aniso because I got less structural messing problem.

All of the above did not dramatically solve the structural messing problem.

I’m thinking that I might have problems in values in the control file.

I am also thinking that the pressure during simulation is high because the values of stress per atom are in order of 10^6 atmospheres.

If you notice any problem in my control file or input file please give me advice.

There are some previous studies about PPTA with LAMMPS and ReaxFF and potential parameter sets developed by Liu et al.

I am always referring them and it seems that they didn’t have the zig-zag shape after imposing a temperature. So, I am thinking that this is a problem.

Thank you very much. I look forward to hearing from you.

Best regards,

Naoki (from Japan)

-----------------------------------------------------------------INPUT FILE----------------------------------------------------------------------------------------------------------------------------------

##-----Setting-------------------------------------------------------------------------------------------------------------------------------------------------------------------------

variable ts equal 0.1 #timestep

variable iT equal 500 #initial temperature

variable T equal 300 #target temperature

##-----Model setting------------------------------------------------------------------------------------------------------------------------------------------------------------------

units real

atom_style charge

boundary p p p

read_data ./model/PPTAmodel_chargesN.txt

replicate 3 5 2

region PPTA block INF INF INF INF INF INF units box

pair_style reax/c control.reax.1st lgvdw yes

pair_coeff * * ffield.reax.lg H C N O

fix 0 all qeq/reax 1 0.0 10.0 1.0e-6 reax/c #Cutoff distance is less than twice the smallest dimendion of the replicated simulation cell

timestep ${ts}

##-----Minimization-------------------------------------------------------------------------------------------------------------------------------------------------------------------

fix f1 all box/relax iso 0.0

minimize 1.0e-10 1.0e-12 10000 10000

unfix f1

reset_timestep 0

##-----Initial temperature heating 0K to 500K-------------------------------------------------------------------------------------------------------------------------------------

fix f2 all npt temp {iT} {iT} 100 aniso 0.0 0.0 1000 tchain 3 pchain 3 # When iso, tchain pchain values are 1. When aniso, 3 (default).

run 20000

unfix f2

reset_timestep 0

##-----Initial temperature cooling 500K to 300K ----------------------------------------------------------------------------------------------------------------------------------

fix f3 all npt temp {T} {T} 100 aniso 0.0 0.0 1000 tchain 3 pchain 3 # When iso, tchain pchain values are 1. When aniso, 3 (default).

run 50000

unfix f3

reset_timestep 0

##-----Temperature control 300K to 300K ------------------------------------------------------------------------------------------------------------------------------------------

fix f4 all npt temp {T} {T} 100 aniso 0.0 0.0 1000 tchain 3 pchain 3 # When iso, tchain pchain values are 1. When aniso, 3 (default).

run 100000

unfix f4

reset_timestep 0

Hello Naoki, were you trying to reproduce results based on the SI from this work?
https://doi.org/10.1016/j.polymer.2017.03.012

I used that model briefly but didn’t observe buckling. However, I didn’t heat up to 500K.

Also, given the complexity of ReaxFF, which is further complicated by using the -lg variant, I recommend making sure your input is consistent. For example, from the reaxff pair_style documentation (https://lammps.sandia.gov/doc/pair_reaxc.html):
Use of this pair style requires that a charge be defined for every atom. See the atom_style and read_data commands for details on how to specify charges.

So this is not something that you should try to"fix" after the fact: you should follow the correct syntax from the start.

My experience with ReaxFF is very limited, but it seems to me that most will start from the default values, and only add a control file for special cases. For -lg, the only one that I can see in the doc is thb = 0.01 as opposed to 0.001 (default). Yet you used the latter in your control file.

It would take very long to find other sources of error in your input: my recommendation is to start with a chemically simple material (how about CO2 or N2 from the Liu paper?) and ensure that your procedure is correct before moving on to a four-element material like PPTA.

Giacomo