Dear All!
In a ¡°stacked¡± system, e.g. a copper block and an added (thin) block of Ni (stacked in z-direction) I would like to use periodic boundaries in x-y. However due to lattice mismatch and different expansion coefficients I wonder if it is possible to use an npt ensemble in this configuration and how the system would react if one of the layers expands ¡°beyond¡± the others (pbc-box) boundary?
Your advice is very much appreciated!
Yours,
Chris
Dear All!
In a “stacked” system, e.g. a copper block and an added (thin) block of Ni
(stacked in z-direction) I would like to use periodic boundaries in x-y.
However due to lattice mismatch and different expansion coefficients I
wonder if it is possible to use an npt ensemble in this configuration and
how the system would react if one of the layers expands “beyond” the others
(pbc-box) boundary?
yes, you can use fix npt with coupling only x and y (it is not going to be
an npt ensemble though, because of the non-periodicity in z-direction).
to minimize the impact of the lattice mismatch, you may have to choose a
suitable supercell. however, if i am not mistaken, the lattice constants of
Ni and Cu differ only by a few percent, and then it is usually sufficient
to set the simulation box according to the lattice to that of the "bulk"
medium and create the ad-layer(s) just on top of that, possibly with the
same lattice parameters and then do a proper relaxation. on surfaces,
you'll always have some reconstruction/relaxation anyway.
as for the different expansion coefficients, they will introduce some
strain, but as far as the npt integrator is concerned, it will operate on
the pressure on the box walls and that would be a mix of the respective
components. unless you are doing studies at varying temperatures, there may
not be much benefit to using fix npt, in most cases it would be sufficient
to determined the bulk equilibrium lattice constant for the chosen (Cu)
potential and at the desired temperature and then use a fixed volume with
those parameters instead. this will save you a lot of grief and
complications and the system can still relax/reconstruct in the direction
of the free surface. in a system with a thin coating of one material on a
bulk of a second, the overall dimension are determined by the bulk
component anyway.
this last matter really is not much of a LAMMPS topic but a question of
simulation methodology in finite size and slab conditions and more of a
discussion topic with your adviser or experienced colleagues and a reason
to study the available literature. there should be plenty published
studies, reviews and even text books, as this is not exactly a new kind of
simulation. i remember attending workshops discussing such matters when i
started as a graduate student 20 years ago, when surface science was a
"hip" topic (although i never really got to do much surface science myself).
axel.