[lammps-users] Unstable Si/Ge boundary

Dear LAMMPS users,

I'm trying to relax a Si/Ge superstructure, consisting of a block of Si
atoms and a block of Ge atoms, and I find that the boundary between the
two blocks isn't stable. I start off with a diamond lattice where one
half is populated with Si atoms and the other half with Ge atoms, and
let it evolve at 300K. After less than 100 ps the boundary between the
two regions collapses (the Si and Ge layers at the boundary
intercalate), but this disagrees with literature results which show that
the boundary is stable. Does anyone have any idea what I might be doing
wrong?

I've tried reducing the timestep, slowly heating up the sample and
fiddling with the Stillinger-Weber parameters that describe the Si-Ge
interaction, but it doesn't help. Note that the T=0 stucture and an
alloy are both stable. Below are the input script and the
Stillinger-Weber interaction parameters.

Thanks for any help,

Philip

############### Definition of lattice #############

units real
dimension 3
boundary p p p
atom_style atomic

lattice diamond 5.56
region simbox block 0 4 0 4 0 8
create_box 2 simbox

region Si_layer block 0 4 0 4 0 3.99
region Ge_layer block 0 4 0 4 4 7.99
create_atoms 1 region Si_layer
create_atoms 2 region Ge_layer
mass 1 28.0855
mass 2 72.64

############### Definition of Si potential #########
pair_style sw
pair_coeff * * SiGe.sw Si Ge

velocity all create 300 573196 mom yes rot yes dist gaussian
timestep 0.5

############### Equilibration ######################

fix NVT all nvt 300 300 10 drag 0.2
dump dumpXYZ all xyz 100 SiGe_initial.xyz
log SiGe_initial.log
thermo_style custom step temp press vol etotal pxx pyy pzz
thermo 10

run 50000

undump dumpXYZ
unfix NVT

################ SiGe.sw #############################

# format of a single entry (one or more lines):
# element 1, element 2, element 3,
# epsilon, sigma, a, lambda, gamma, costheta0, A, B, p, q, tol

Si Si Si 50.0 2.0951 1.80 21.0 1.20 -0.333333333333
            7.049556277 0.6022245584 4.0 0.0 0.0
Si Si Ge 0.0 0.0 0.0 26.0 1.20 -0.333333333333
            0.0 0.0 0.0 0.0 0.0
Si Ge Si 0.0 0.0 0.0 26.0 1.20 -0.333333333333
            0.0 0.0 0.0 0.0 0.0
Si Ge Ge 47.2536 2.1381 1.80 26.0 1.20 -0.333333333333
            7.049556277 0.6022245584 4.0 0.0 0.0
Ge Si Si 47.2536 2.1381 1.80 26.0 1.20 -0.333333333333
            7.049556277 0.6022245584 4.0 0.0 0.0
Ge Si Ge 0.0 0.0 0.0 26.0 1.20 -0.333333333333
            0.0 0.0 0.0 0.0 0.0
Ge Ge Si 0.0 0.0 0.0 26.0 1.20 -0.333333333333
            0.0 0.0 0.0 0.0 0.0
Ge Ge Ge 44.5049 2.181 1.80 31.0 1.20 -0.333333333333
            7.049556277 0.6022245584 4.0 0.0 0.0

Hi Philip,

Your setting for 'units' is wrong. It should be 'metal' instead of
'real' if you are using the Stillinger-Weber potential files provided
with LAMMPS. This could possibly be the source of your errors.

Zhun-Yong

Dear Zhun-Yong,

Thanks for the suggestion. However, I've already changed the values of sigma and epsilon in the potential file to real units, and for pure Si my results agree with literature values - I only get a problem with Si/Ge.

Philip

Zhun-Yong,

When you said that the results from the literature are different,
do you refer to simulations with the same potential? If not, may
be the problem is with the use of Ding and Andersen Ge, that
have very high lambda.

There are other Ge parameterizations available that you may try
if you think the problem is with the force field. I reparameterized Ge
to epsilon=45 kcal/mol, sigma= 2.1836 A, and lambda = 20.
See the details in "Vitrification of a monatomic metallic liquid",
Nature 448 p787 (2007)

vale

The question should be addressed to Philip.

2008/4/1 Valeria Molinero <[email protected]>: