Hello LAMMPS Community,
I would like to use LAMMPS to simulate the physical vapor deposition of
Palladium onto a graphene-coated Copper (111) surface. Due to the hybrid
nature of this system, I need to use the pair_style hybrid/overlay to treat
each interaction with an appropriate interatomic potential.
i disagree with this assessment. first of call, you should see, if you
can find a single potential type that can represent all of these
materials at the same time. if not, i would think that pair style
hybrid and not hybrid/overlay would be the right pair style to go
with. please note, that combining different potential styles via the
hybrid pair style has some serious restrictions on the _physics_ of
systems that can be meaningfully computed this way. in general, the
parts that are treated with different potentials should be well
separated and can only be coupled with pair-wise additive potentials.
manybody potentials can only be split by atom types, not pairs of
types, as they are not pairwise additive. when you have a mixture of
particles (e.g. an alloy), you should use a particular potential type
and parameter set that has been parameterized *and* balanced for this
mixture. going against these guidelines carries a high risk of
computing inconsistent and unbalanced interactions leading to rather
meaningless results. please also keep in mind that many potentials and
parameterizations are not very transferable. if you want a consistent
treatment of a complex system, you almost always are restricted to use
a quantum chemical or at least semi-empirical description.
I have been unable to find a potential file containing the parameters for an
alloyed Pd Cu interaction in the literature, however, I saw in the
documentation for pair_style eam it says that:
" Different single-element files can be assigned to different atom types to
model an alloy system. The mixing to create alloy potentials for type pairs
with I != J is done automatically the same way that the serial DYNAMO code
originally did it; you do not need to specify coefficients for these type
pairs.”
With this in mind, I used the following commands to set the pair
coefficients for the simulation.
#-----------Potentials-----------------------------------------------------
pair_style hybrid/overlay airebo 2.5 1 1 tersoff lj/cut 10.0 eam
pair_coeff * * airebo Potentials/CH.airebo C NULL NULL
pair_coeff * * tersoff Potentials/CCu.tersoff C Cu NULL
please note that this line will compute the carbon-carbon interactions
*twice* unless you modify the potential files, but in that case you
have other problems, since both AIREBO and Tersoff are manybody
potential styles.
pair_coeff 1 3 lj/cut .341 3.003
pair_coeff 2 2 eam Potentials/Cu_u3.eam
pair_coeff 3 3 eam Potentials/Pd_u3.eam
Type 1: Carbon
Type 2: Copper
Type 3: Palladium
This results in an error stating that not all pair coefficients are set.
Which makes sense because I never explicitly set the interaction between
copper and palladium (2 and 3).
My question is as follows:
Is it possible to get LAMMPS to automatically mix the potentials from two
pure eam files using the hybrid/overlay pair_style? If so, how can I do
this?
in addition to what steve said, the restrictions on what can be mixed
automatically for hybrid/overlay are significant. with overlay it is
rarely uniquely defined that you want a particular pair of
interactions to be derived from two other pairs. most of the time, it
could be quite intentional to *not* have them automatically mixed
(unlike for pair style hybrid). if i was in your situation, i would
see if there is a suitable parameterization for all elements and
similar systems with, e.g. MEAM or ReaxFF. the combination of elements
you have is not very exotic and thus there is a good chance that
somebody has already looked into creating a parameter set for that.
that would reduce your headaches massively.
axel.
axel.