Simulating a nanowire in vacuum (free surface effects)

Hi all,

Once again, I am finding very peculiar dissipation of atoms into a vacuum. In this case, the temperature never increases above 50K, which certainly should not result in vaporisation of the atoms. However, the nanowire completely disintegrates and the atoms re-disperse towards the corners of the simulation box. Oh, and the pressure never increases beyond 1200 (which is admittedly quite high). Why?

Is it impossible to simulate a nanowire that is this small, without dispersion of the atomic structure?

My code:

clear
units metal
dimension 3
boundary p p p
atom_style atomic

lattice fcc 4.03

#---------- Define Geometry

region whole block 0 10 0 10 0 10
region wire cylinder y 5 5 4 0 10
create_box 1 whole
create_atoms 1 region wire

pair_style eam/alloy
pair_coeff * * Al99.eam.alloy Al

compute csym all centro/atom fcc
compute pe all pe/atom

dump 1 all cfg 1000 dump.Compress_*.cfg id type xs ys zs c_csym c_pe fx fy fz

#-------- Equilibrate
thermo 500
thermo_style custom temp press pxx pyy pzz lx ly lz pe etotal
velocity all create 50 12345 rot yes mom no
fix 1 all npt temp 50 50 1 iso 0 0 10 drag 1.0
run 20000

echo both

Nathaniel,

I ran your script with Al_zhou_eam.alloy. All that's happening is the
nanowire is drifting through the periodic boundary. The drift is
happening because you chose not to zero the linear momentum when you
assigned initial velocities:

velocity all create 50 12345 rot yes mom no

There are two other problems with the simulation:

1. The barostat is applied to the x and z directions, even though the
structure is only continuous in y. Never try to couple a barostat to
vacuum.
2. The NPT dynamics start from an unrelaxed structure. Better to relax
the structure first using minimization or some other method.

Aidan

Thanks Aidan,

All good points.

However, the question remainsâ€¦ If you want to impose a uniaxial load (i.e. fix deform), you really need to have equilibrated the pressure. If you canâ€™t use a barostat for a cylinder (nanowire), how else can you achieve this?

Cheers!
Nathaniel

I re-ran the simulation, specifying rotational (not linear) momentum and the system did not disperse!

I may see what happens when I apply fix deform.

Thanks