Interface adhesion stress calculation - reducing large fluctuation

Hello all,

I am working to calculate the interface adhesion energy between ZnO crystal and graphene layers by seperating these two layers in the z-direction. I have questions on 1) how to equilibrate the ZnO and graphene system with zero initial force by doing the minimize, NPT and NVT relaxation procedures. 2) The ZnO layer (upper) was fixed and the graphene layer (lower) was moved in the Z direction. How to calculate the adhesion force between these two layers?

The input file is copied as below. Please check and give me suggestions or comments for this simulation. Currently, the calculated stress has large fluctuation and it seems that the testing system need be better stabilized/equilibrated before separation and during separation.

Thank you very much.

dimension 3
units metal
boundary p p p
atom_style charge
kspace_style ewald 1.0e-4

create geometry start

read_data data.Zn_G

atom interactions start

group zinc type 2
group oxygen type 1
group graphene type 3
group hydrogen type 4
group lower union graphene hydrogen
group upper union zinc oxygen
mass 2 65.409
mass 1 15.999
mass 3 12.011
mass 4 1.008

pair_style hybrid buck/coul/long 8.5 airebo 3.0 lj/cut 2.5
pair_coeff 2 2 buck/coul/long 0.0 1e-10 0.0
pair_coeff 1 1 buck/coul/long 9547.96 0.21916 32.0
pair_coeff 1 2 buck/coul/long 529.70 0.3581 0.0
pair_coeff * * airebo CH.airebo NULL NULL C H
pair_coeff 1 3 lj/cut 0.004231 2.921
pair_coeff 2 3 lj/cut 0.0194 2.9274
pair_coeff 1 4 none
pair_coeff 2 4 none

atom interactions end

minimize 1.0e-6 1.0e-6 1000 10000

timestep 0.002
thermo 100

fix 1 all npt temp 20 1 20 iso 1.0 1.0 1000.0
thermo_style custom step temp lx ly lz press pxx pyy pzz
run 10000
dump 1 all xyz 100 traj.xyz

unfix 1

fix 2 all nvt temp 1 1 5
thermo_style custom step temp lx ly lz press pxx pyy pzz
run 10000
dump 2 all xyz 100 traj.xyz
unfix 2

fixing the upper layer

fix 3 upper setforce 0.0 0.0 0.0

dump 3 all xyz 100 traj.xyz
unfix 3

moving the lower layer in the Z-direction and calculate adhesion force

timestep 0.002
thermo 100
variable Vz equal 0.0125
variable dz equal -2
fix 4 lower move variable NULL NULL v_dz NULL NULL v_Vz
compute peratom graphene stress/atom
compute sumstress graphene reduce sum c_peratom[1] c_peratom[2] c_peratom[3]
variable sysstr1 equal -c_sumstress[1]/vol
variable sysstr2 equal -c_sumstress[2]/vol
variable sysstr3 equal -c_sumstress[3]/vol
thermo_style custom step temp press v_sysstr1 v_sysstr2 v_sysstr3
run 80000
dump 4 all xyz 100 traj.xyz
dump 5 all xyz 100 traj.xyz1
unfix 4

Just a quick comment. Is your system composed of one graphene layer and one ZnO layer? If yes, does ZnO forms single layers at all? Are you sure your potential for ZnO will render stable ZnO layers? Additionally, are you sure such combination of interactions can handled the hybrid systems and you are not just mix-&-mingling potentials from different sources?

Carlos

The system has one layer of ZnO and one layer graphene. The potentials for each separate layer was calculated and validated. Do you have some suggestions on this hybrid system for force field or interaction definition?

pair_style hybrid buck/coul/long 8.5 airebo 3.0 lj/cut 2.5
pair_coeff 2 2 buck/coul/long 0.0 1e-10 0.0
pair_coeff 1 1 buck/coul/long 9547.96 0.21916 32.0
pair_coeff 1 2 buck/coul/long 529.70 0.3581 0.0
pair_coeff * * airebo CH.airebo NULL NULL C H
pair_coeff 1 3 lj/cut 0.004231 2.921
pair_coeff 2 3 lj/cut 0.0194 2.9274
pair_coeff 1 4 none
pair_coeff 2 4 none

You can give COMB3 a try (pair_style comb3) which descibes ZnO with C
in one parameter file.

Ray

There you have Ray giving you a hint. Yet, I would be double careful with your system and the specific application you are after. You are forcing your metallic oxide to be basically two-dimensional (which surface cut?) and I wonder how stable this config is even at the ab-initio level. Maybe your test already included some mode analysis for mechanical stability and not just some force/energy matching between your parametrized potential and the first-principles calculations. But note that my comments are just intuition-guided as I have no previous expertise working with such hybrid systems.

Carlos