Equlibration not achieved for a composite structure

LAMMPS LAMMPS Beginners
1

Hello all,
I want to create a composite structure using amorphous silica and crystalline silicon using a long-range user-defined potential and Stillinger-Weber potential, respectively.
I initially made silica glass, and the simulation box is augmented to insert crystalline silicon atoms in the box, using periodic boundary conditions. The structure looks pretty good until I start a minimization run. During minimization, I find that the pressure shoots up unrealistically high, so does the potential energy. This may be because the atoms are close enough at the interface. Having that said, deletion of the overlapping atoms did not help either. Before I start to run any MD algorithm, I find that the crystallinity of the silicon atoms is lost. In this regard, I have the following concerns.

  1. While the constituents of silica have charged particles that will later interact with neutral silicon atoms, how does that interaction work? I mean, is it okay to run the entire simulation using the atom_style charge?
  2. Are such chaos prevalent while using the hybrid style? How would you think that I can improve the system to proceed further? As you will be able to make out now, my main aim is to generate an equilibrated composite structure with x-silicon and a-silica structures using PBC.
    Any guidance would be highly appreciated.
    Thank you so much for your attention and participation.
    Kiran
2

  • re 1: the long-range coulomb will use whatever charges are assigned to the atoms. if atoms have charge zero charge they don’t contribute. the stillinger-weber potential will completely ignore any charge property assigned to atoms.

  • re 2: what you describe sounds like you have incorrect or unsuitable force field parameters and atom type assignments or unsuitable simulation settings or both.
    using a hybrid pair style may make it easy to set up a system with mixed potential functions, but it is hard work and can be very difficult - if not impossible - to make such simulations produce sufficiently accurate results.

3

Dear Dr. Axel,
Thank you for the response.
As of now, I have assigned the following interactions. Let’s say O-1, Si-2 (charged ones of SiO2), and, Si-3 (Crystalline).
1-1, 1-2, 1-3, 2-1, 2-2 and 2-3: long-range with appropriate parameters.
3-3 Stillinger-Weber.
Does this assignment make sense?

4

I need to see the actual input.

5

Please take a look.
#Type 1----> Oxygen (SHIK-1)
#Type 2----> Silicon (SHIK-1)
#Type 3----> Silicon (SW potential)

mass 1 15.9994
mass 2 28.0855
mass 3 28.0855 #c-Si

#SHIK-1 Potential-----rcut 8.0 and 10.0
pair_style hybrid shik/wolf 8.0 10.0 sw
pair_coeff * * sw Si.sw NULL NULL Si
pair_coeff 1 1 shik/wolf 1120.528996 2.892741833 26.13207696 16800.0 0.2 0.2 #O-O
pair_coeff 1 2 shik/wolf 23107.84764 5.097856735 139.6947857 66.0 0.2 0.2 #O-Si
pair_coeff 2 2 shik/wolf 2797.979165 4.407320018 0.0 3423204.0 0.2 0.2 #Si-Si
pair_coeff 1 3 shik/wolf 23107.84764 5.097856735 139.6947857 66.0 0.2 0.2 #O-Si(3)
pair_coeff 2 3 shik/wolf 2797.979165 4.407320018 0.0 3423204.0 0.2 0.2 #Si(2)-Si(3)

neighbor 3.0 bin
neigh_modify check yes every 1 delay 10

6

Pair style shik/wolf is not part of LAMMPS so I cannot comment on the validity on it.

You are using the same parameters for Si in SiO2 and Si in Silicone. That cannot be correct.

7

Thank you for the suggestions. I shall try to modify the system based on your inputs.