disagreement in two different ways of adding vacuum

LAMMPS LAMMPS Mailing List Mirror
1

Dear all,

I tried to add vacuum to the bulk with two ways but for some reason I’m getting different energies in each case (for the same geometry) :

Case 0. took bulk, 0 4 0 4 0 4 and the same size of simulation box.

Case 1. created simulation box, 0 4 0 4 -1 5. And created atoms on the smaller region, so that the vacuum was left in +z and -z directions, i.e. create_atoms 0 4 0 4 0 4.

In this case more atoms were created than in bulk case in the same region, and energy also differed from that of bulk. I don’t understand why number of atoms increased. It’s not even the same if I filled the entire box.

Case 2. created bulk 0 4 0 4 0 4 and the simulation box of the same size, then tried to change box using change_box command without remapping atoms:
change_box all z final -1 5

Here again, I got different value for energy than that from bulk, but the number of atoms remained the same.

In summary, I get:

#bulk(pbc, no vacuum): N=256, E=-74.554717/atom
#case 1: N=288, E=-73.342496/atom
#case 2: N=256, E=-73.190968/atom

I would explain energy differences from bulk value with surface effect but I haven’t relaxed the system, just set up geometries. And why the two per atom energies in Case1 and Case2 differ?

Third option could be filling the box and deleting atoms so that the vacuum is left. I have bulk and just want to add vacuum without disturbance. I’m confused which way is the correct one since I get different energies for any option.

Thanks for any suggestion. Files attached.

ffield.reax.Al (21.7 KB)

in.vactest (739 Bytes)

al_control (511 Bytes)

2

Dear all,

I tried to add vacuum to the bulk with two ways but for some reason I'm
getting different energies in each case (for the same geometry) :

no. it is not exactly the same geometry.
the different energies are correct.

Case 0. took bulk, 0 4 0 4 0 4 and the same size of simulation box.

Case 1. created simulation box, 0 4 0 4 -1 5. And created atoms on the
smaller region, so that the vacuum was left in +z and -z directions, i.e.
create_atoms 0 4 0 4 0 4.

In this case more atoms were created than in bulk case in the same region,
and energy also differed from that of bulk. I don't understand why number of
atoms increased. It's not even the same if I filled the entire box.

the different number of atoms is perfectly ok.
you defined your lattice so that your atom positions
would be *exactly* on the box faces and corners.

when lammps creates atoms from a region, it creates
them to fill the defined volume but *only* if they are
inside the simulation cell. due to using periodic boundary
conditions, atoms that are exactly on box corners
and faces are allowed only once since the other
two (or eight) positions would be on top of each other.

so if you enlarge the box in z before creating the atoms,
atoms on *both* z faces are now inside the box while
only half of them were inside in case 0.

Case 2. created bulk 0 4 0 4 0 4 and the simulation box of the same size,
then tried to change box using change_box command without remapping atoms:
    change_box all z final -1 5

Here again, I got different value for energy than that from bulk, but the
number of atoms remained the same.

of course. you pushed atoms apart that were immediate
neighbors in case 0, so they are interacting less and since
the interactions are attractive consequently the energy goes up.
remember, that you have periodic boundary conditions in all directions.

and since you create the atoms *before* enlarging the box,
you get the same number of atoms as in case 0.

if you turn of periodic boundaries in z you should
get the same number of atoms and the same energies
in all three cases.

In summary, I get:

#bulk(pbc, no vacuum): N=256, E=-74.554717/atom
#case 1: N=288, E=-73.342496/atom
#case 2: N=256, E=-73.190968/atom

I would explain energy differences from bulk value with surface effect but I
haven't relaxed the system, just set up geometries. And why the two per atom
energies in Case1 and Case2 differ?

Third option could be filling the box and deleting atoms so that the vacuum
is left. I have bulk and just want to add vacuum without disturbance. I'm
confused which way is the correct one since I get different energies for any
option.

Thanks for any suggestion. Files attached.

i think you discounted the impact of periodicity in z.

axel.

3

Hello Axel,

Thanks for detailed discussion. Yes, I didn’t take into account all the effects of pbc, especially in case 1.Then it’s true, bulk with vacuum will never have the same energy as pure bulk, no matter how pbc is set or whether relaxed or not.

Now ppf (or pps or ppm) did give equal energies for case 0 and case 2 (equal to the value of case2) though didn’t affect case1.

4

Thanks for detailed discussion. Yes, I didn't take into account all the
effects of pbc, especially in case 1.
Then it's true, bulk with vacuum will never have the same energy as pure
bulk, no matter how pbc is set or whether relaxed or not.

Structures with vacuum shall never have the same energy as pure bulk,
due to decreased number of bonds at the surfaces, and the one with
vacuum shall always be less stable.

Before relaxation the energy difference between the two is the
un-relaxed surface energy, while after relaxation is the relaxed
surface energy.

To properly obtain surface energies you will need a vacuum that is
large enough so that surfaces do not interact with their periodic
images, that is, if "p" is used. One lattice with "p" may be too
small.

Best,
Ray