Unexpected zero interfacial force between epoxy and CNt

Hello dear all,

I am trying to measure the interfacial shear strength between a CNT and an epoxy resin. I have a system well equilibrated over 20ns in NPT anisobaric ensemble with 1 fs timestep. The diagram of the total energy of the system shows that the system has reached its equilibrium point

I am using DreidingX6 force field. The epoxy resin is crosslinked with a crosslinking degree of 86%
The CNT and the resin are NOT connected to each other and they just interact with VdW interactions.
My system looks like this. The boundary conditions are periodic in every direction.


I saw in the literature that the interfacial shear strength for pristine CNT with epoxy varies between 10-30 GPa
I am trying to use fix smd command to make the pullout. I constrain the motion of the polymer by using a spring.
I am performing a pullout in the z direction and the position of the CNT center of mass of the looks like that:

But the average force I am measuring for ths fix smd output command is fluctuating over zero like this

The command seems to work fine from a visualisation point of view

I cannot understang what I am doing wrong. I will also attach the input files and the data files for the simulation.

6_6_CNT_smd.data (5.7 MB)
6_6_CNT_smd_100.in (5.7 KB)

Thank you in advance for your time and help!

If there is only vdW between CNT and resin, it is no surprise that there is practically no retention force when pulling the CNT. The vdW potential is very soft so there is effectively a flat potential along the CNT surface.

Hello Axel,
Well in the literature the values from MD simulations with pristine CNT and epoxy interface (like my case) are 10-30 GPa. Does the fact that all boundaries are periodic affect the result? Also the temperature is 300K

This is not my area of research, so I don’t know any details and thus I can just give “common sense physics” kind of answers.

For specific details, especially about how to reproduce published results, you need to discuss with the people that have done these kinds of simulations how they have obtained those values (if not properly documented in their publications).

Some general comments:

You are measuring a frictional force, but then trying to compare it to a shear modulus. I can see the physical connection between those quantities, but I don’t know how they are connected mathematically – I hope you do.

It is not obvious to me from your graph that the frictional force is zero on average – it could have a small non-zero average. This is a very general problem with non-equilibrium molecular dynamics where, because of the very small length and time scale of simulations, realistic stimuli will produce responses with very low signal-to-noise ratio.

Also, the CNT is sliding out at a speed of 10m/s relative to the epoxy. Is it realistic to expect frictional forces to “catch” at that speed? Do you observe a non-uniform stress profile across the epoxy?


I am aware og the mathematical conection between frictional force and shear modulus. Most of the lirerature uses the interfacial force and then divides it with the area if the interface to obtain this quantity.

You are correct about your observation but in fact there is also a zero average.

Also, the CNT is sliding out at a speed of 10m/s relative to the epoxy. Is it realistic to expect frictional forces to “catch” at that speed? Do you observe a non-uniform stress profile across the epoxy?

This is a typical literature value, I will try also smaller ones but i dont think that trere will be such a different result.

Thanks for your comments!

Cool, I learned something new today :slight_smile:

Have you tried replicating a known study from literature? If you have a particular paper you’re working from, post it here and I can take a (very quick) look at it.

This is the reference. In their case they have pbc, but I want to have a periodic geometry

As a first step you should ask them for their LAMMPS input files, which will make it much easier to double-check many things (such as whether force field parameters were correctly set).

Furthermore, it is not obvious to me that their method is physically correct. As they pull the CNT out of the epoxy, the receding end creates a void which will pull back on the CNT. It is entirely possible that their “interface” is in fact the receding end’s cross section rather than the surface area along the CNT circumference. That would explain neatly why your periodic setup does not replicate their results (since a periodic setup would not contain that interface).

As a reviewer I would have recommended against publication until they had addressed this point, either by reference to other literature or by checking with additional analysis (for example a compute force/force between the CNT and the “end slab” of the polymer, which is easily computed in post, or even showing the raw force trace in the SI, which should presumably decrease against trajectory time, since the interfacial contact length l should decrease as the CNT is pulled out of the epoxy).

I agree