Bilayer hexagonal Boron Nitride(hBN) simulation : problems with electrostatic energy

LAMMPS LAMMPS Mailing List Mirror
1

Hi all,
I have simulated a bilayer hBN with intra-layer potential described by tersoff potential and inter-layer potential are described by LJ(6-12) and coulomb term. (attached the file)

The elong and ecoul constitute the total electrostatic energy. However, electrostatic energy is changing drastically if I change real space cut-off or even, tolerance in k-space. However, the total electrostatic energy shouldn’t change as, the summation is done in real (ecoul) and momentum space (elong). Also, elong is way larger than ecoul.

For instance, with 15 Ang. cutoff in real space for coulomb interaction of a ~11011050 supercell of bi-layer hBN I get elong= -25160.913 eV and ecoul =-35.969 eV . If I change real space cutoff to 20 Ang for same system, I get elong = -18785.12 and ecoul = -35.9678 eV.

I have carefully read previous questions and answers but haven’t been able to solve it. Is there anything wrong with the input??

Also, please suggest me any reference that can help me to solve it.
Thank you for your time
Indrajit

hBN.in (2.41 KB)

2

Hi,
I realized I should also add the tersoff potential. In this mail I attached the file for that.
Sorry for missing that.
Thanks

hBN.tersoff (1.5 KB)

3

Hi all,
I have simulated a bilayer hBN with intra-layer potential described by
tersoff potential and inter-layer potential are described by LJ(6-12) and
coulomb term. (attached the file)

The elong and ecoul constitute the total electrostatic energy. However,
electrostatic energy is changing drastically if I change real space cut-off
or even, tolerance in k-space. However, the total electrostatic energy
shouldn't change as, the summation is done in real (ecoul) and momentum
space (elong). Also, elong is way larger than ecoul.

For instance, with 15 Ang. cutoff in real space for coulomb interaction of a
~110*110*50 supercell of bi-layer hBN I get elong= -25160.913 eV and ecoul
=-35.969 eV . If I change real space cutoff to 20 Ang for same system, I get
elong = -18785.12 and ecoul = -35.9678 eV.

I have carefully read previous questions and answers but haven't been able
to solve it. Is there anything wrong with the input??

you have an inconsistent model. while you can partition the real-space
interactions with pair_style hybrid (or hybrid/overlay), you cannot do
the same for kspace.
the long-range solver *will* include the long-range coulomb
contributions for pairs that are not included in real-space, hence the
inconsistency when summing up ecoul and elong for different real-space
cutoffs.

axel.

4

Dear Axel,
Thank you for your reply.
Can you suggest me anything if I want to do the following: (within lammps)

  1. L-J force between B of layer1 and N of layer2 or the other way around.
    2)L-J force between N of layer2 and N of layer1
  2. coulomb interaction between layer 1 and layer 2 all atoms but not within same layer.
  3. intra-layer interaction is still described by tersoff potential

I have gone through one of your earlier response to a similar question. However, if you suggest some idea, about building it within lammps I can work on it.
Thanks
Indrajit

5

Dear Axel,
Thank you for your reply.
Can you suggest me anything if I want to do the following: (within lammps)

1) L-J force between B of layer1 and N of layer2 or the other way around.
2)L-J force between N of layer2 and N of layer1
3) coulomb interaction between layer 1 and layer 2 all atoms but not within
same layer.
4) intra-layer interaction is still described by tersoff potential

I have gone through one of your earlier response to a similar question.

can you point out which answer with which recommendations. i have sent
so many e-mails to lammps-users over many years, i don't recall all of
them.

However, if you suggest some idea, about building it within lammps I can
work on it.

well, the obvious conclusion is that you cannot use long-range
electrostatics for the reasons i already have given. the next best
thing would be using a real-space-only method with a compensated
damping scheme, e.g. coul/wolf or coul/dsf with a sufficiently chosen
cutoff.

of course, that would have to be followed by thorough testing of this approach.

axel.

6

Hi,
Thanks for your response. I will try with real space only.

I was referring to “”“Re: [lammps-users] Hybrid REAX/C and LJ Forcefield”""

"

so in my personal opinion the best choice here is to do the entire system in ReaxFF. i don’t see a way to incorporate long-range electrostatics consistently, and having a coulomb interaction between graphene (with ReaxFF) and water (with LJ + Coulomb) would also be inconsistent (water would polarize graphene, but not the other way around, and your water potential would not be parameterized to account for the charge fluctuations in ReaxFF). so the only consistent hybrid model would be using plain pair_style hybrid (not hybrid/overlay).

C-C with ReaxFF
C-O and C-H with lj/cut
O-O, H-H, O-H with lj/cut/coul/cut

but that doesn’t look like it will have any significant advantage over using ReaxFF for everything​.

in summary, pair_style hybrid and hybrid/overlay offer many ways to shoot yourself in the foot and a lot of care and consideration to get it right.

axel"

However, using elong and ecoul like the way I did is a bad idea, as you have cleared it.
Thanks again
Indrajit

Sent with Mailtrack