Help with AlSi10Mg alloy hybrid potential (LJ parameters issue)

[Tasmiah_Zaman]

Hello everyone,
I am trying to validate a paper on AlSi10Mg alloy:
“Effect of fabrication process parameters and graphene reinforcement on mechanical behaviour of additively manufactured AlSi10Mg alloy – A molecular dynamics simulation study.”
I am facing issues creating the correct hybrid potential for the alloy as described in the paper.
What I did so far:

1. Built the atomic structure using Python, LAMMPS, and MATLAB
2. After running the simulation, the atoms spread apart instead of forming a stable structure
3. I suspect the issue is with my Lennard-Jones (LJ) parameters
4. I cannot find suitable ε (epsilon) and σ (sigma) values for the cross-interactions:
   Al–Si
   Si–Mg
   Al–Mg

Below is the input script I used (simplified):

LAMMPS input script for AlSi10Mg (incomplete example)

‘’’ units metal #gm/mole, angstrom, ps, ev, bars, K, gm/cc
dimension 3
boundary p p m
atom_style atomic

simulation box and lattice,potential files

lattice fcc 1.0
region whole block 0 60 0 60 0 60 units box
create_box 3 whole
read_data alsimg_sphere.data add append

-------------------- POTENTIAL SETUP --------------------

Atom type mapping:

1 = Al

2 = Si

3 = Mg

Hybrid/overlay: allows EAM, Tersoff, AIREBO, and LJ cross terms

pair_style hybrid/overlay eam/alloy eam/fs tersoff lj/cut 10.0

pair_coeff * * eam/alloy Al99.eam.alloy.txt Al NULL NULL
pair_coeff * * tersoff Si.tersoff NULL Si NULL
pair_coeff * * eam/fs Mg.eam.fs.txt NULL NULL Mg

Cross-interactions using Lennard-Jones

pair_coeff 1 2 lj/cut 0.01954 3.91728 # Al–Si
pair_coeff 1 3 lj/cut 0.01026 3.34978 # Al–Mg
pair_coeff 2 3 lj/cut 0.00915 3.25891 # Si–Mg

reset_timestep 0
timestep 0.0001

fix NVE all nve
thermo 100
dump mydump all custom 100 mydump*.txt id type x y z vx vy vz
run 1000
write_data mydata.data
quit ‘’’

After running, the structure looks like this (screenshot attached).
My questions:
Does anyone know reliable LJ (or alternative) potential parameters for AlSi10Mg?

Am I correct to assume that the spreading issue is due to incorrect LJ parameters?

Would another potential type (e.g., EAM + LJ hybrid) be more suitable for this alloy system?

Any guidance would be highly appreciated!
Thank you,
Tasmiah Zaman

[akohlmey]

Using a hybrid potential is almost never a good idea; hybrid/overlay even less so. Especially for alloys.
Which potential(s) exactly did the authors of the paper use that you are trying to reproduce?

[Tasmiah_Zaman]

The author has used these potentials. I am trying to reproduce these. I tried to find the exact value of LJ potentials using the reference, but I could not.

[akohlmey]

I’ve explained many times why this kind of potential “assembly” is a bad idea and why it is missing important contributions to the interactions. Using LJ to model the cross terms is band aid at best.
You can find my previous explanations in the archives. If there are any useful results in the paper you are referring to, they are obtained by accident, not by design.

Thus a much better choice modeling an alloy would be a potential covering all elements like this: Interatomic Potentials Repository
There are no missing or double counted interaction terms.

[Tasmiah_Zaman]

Okay. Thank you