Extract LJ from Airebo

Hello All,

I was wondering if it was possible to extract LJ energy alone while using pair_style AIREBO potential.

Thanks,
Rohit.

Hello All,

I was wondering if it was possible to extract LJ energy alone while using
pair_style AIREBO potential.

no. at least not without modifying the sources.

axel.

Thanks Axel. I guess i have to stick with hybrid potential of REBO and LJ/CUT.

Rohit.

Hi Rohit,
Just one comment, hybrid REBO + LJ/Cut does not equal AIREBO. AIREBO has a switching function that turns on/off the LJ part, which is something hybrid rebo + lj/cut cannot do. Also, REBO potential already includes a LJ term, which is splined to the main potential.
Best,
Ray

Hello Ray,

Thanks for your time :slight_smile:

I do know that AIREBO has a switching function for the LJ part and that is the reason i want to use AIREBO

to simulate large number of CNTs, so that i need not define LJ/CUT interaction separately for every combination of non bonded interactions in between the tubes.
My main intention was to use REBO+LJ/CUT but seeing the switching function of LJ in AIREBO i tried to use
it, but my simulations need extraction of LJ energy separately, hence the REBO + LJ/CUT.

You said that REBO potential already includes LJ term, did you mean the AIREBO? As far as i know REBO
is generally used only in covalent bond systems.

Rohit.

Hi Rohit,
You are welcome. A complete REBO potential has a short-range and a “long-range” part, the rebo and the LJ, respectively. Short range rebo are cut off mostly after 2.0 A, while the “long” range LJ is ~2.0*sigma. These two contributions make up the REBO potential.
Around some distance, which I am not exactly sure what the value is, the repulsive wall of LJ is splined to the rebo part forming a double well potential. For most applications, this splined wall is beyond the equilibrium bond distances so the wall is not accessed, and this works great for most applications. One example when this does not work well is compressing graphite: upon compression atoms in different layers overcome this repulsive wall and forms diamond with a significantly smaller activation barrier.

For your CNT applications, I don’t think you need to hybrid a LJ/cut to your REBO.
Best,
Ray

Hello Ray,

First of all let me be clear when i say REBO i mean the second generation REBO potential energy expression proposed by Brenner et al. in their 2002 paper and when i mean AIREBO i mean the potential model proposed by
Stuart et al. I did not see any long-range part i.e. the LJ part in the Brenner paper.

I originally wanted to simulate the bonded environment with REBO and non-bonded with LJ. Seeing that AIREBO can be used as REBO with the LJ and Torsion part switched off i am using it for my bonded environment and lj/cut for non bonded.

Pardon me for repeating but when you mean “complete REBO potential” do you mean AIREBO model by Stuart?
Because for the AIREBO model it makes sense that there is a long-range part whose cut-off distance
can be controlled.

The point you made about the double potential well is very interesting and gives me a warning to be on look out for such non physical results.

My simulations consists of many number of CNTs. I think LJ interactions among the tubes need to be modeled with LJ/CUT if using AIREBO with LJ and torsion off for bonded interactions.

I use a hybrid of REBO and LJ/CUT only because it gives me pairwise interaction energies when i want otherwise i would have used AIREBO model.

Sorry if i confused you with my wordings.

Thanks,
Rohit.

Hi Rohit,
My terminologies of REBO and AIREBO are the same as yours. After taking a close look at the pair_rebo.cpp and pair_airebo.cpp I agree with you now. I have a copy of Donald Brenner’s original 2nd generation REBO code and I was forming my arguments based on that code, which does include LJ with splined repulsive wall. LAMMPS implementation of REBO, as well as the 2002 2nd REBO paper, does not include this splined LJ. I agree that for your applications hybrid rebo + lj/cut is a good choice.

Best,
Ray

Hello Ray,

Thanks a lot!! For a minute i was terrified i might have to redo all my simulations.
The idea of a splined wall does seem more appealing though.

Again thanks for your time and suggestions.

Regards,
Rohit.

Hi Rohit,
You are welcome and thanks for the helpful discussions, too. I guess the splined LJ was in that code for some reason :slight_smile:

Best,
Ray