Interpolate potential between two type A atoms according to the # of their nearest type B atoms

Hi All,

I’m trying to simulate Zr metal/oxide interface. For Zr-Zr atoms interaction in metal, we have EAM potential, for Zr-Zr ions interaction in ZrO2, we have Buckingham potential. (Both EAM and BK have analytical form).

But for Zr-Zr interaction at the metal/oxide interface, we cannot find a good potential so I hope to do a linear interpolation of EAM and Buckingham, according to the # of nearest Oxygen anions. For example, to calculate the interaction energy between a pari of Zr-Zr, firstly I should calculate how many oxygen atoms are around (e.g. within 3 Angstrom) each Zr atom/ion, then I will use the # of oxygen neighbors to adjust the coefficient of EMA and BK to do the interpolation.

Is it possible to do such things in LAMMPS? If so, could anyone having similar experiences give any advice/suggestion? If not, does anyone know other possible tools?

Thanks so much for the help!
Xing

Hi All,

I'm trying to simulate Zr metal/oxide interface. For Zr-Zr atoms interaction
in metal, we have EAM potential, for Zr-Zr ions interaction in ZrO2, we have
Buckingham potential. (Both EAM and BK have analytical form).

But for Zr-Zr interaction at the metal/oxide interface, we cannot find a
good potential so I hope to do a linear interpolation of EAM and Buckingham,
according to the # of nearest Oxygen anions. For example, to calculate the
interaction energy between a pari of Zr-Zr, firstly I should calculate how
many oxygen atoms are around (e.g. within 3 Angstrom) each Zr atom/ion, then
I will use the # of oxygen neighbors to adjust the coefficient of EMA and BK
to do the interpolation.

Is it possible to do such things in LAMMPS? If so, could anyone having
similar experiences give any advice/suggestion? If not, does anyone know
other possible tools?

this is a very bad idea. you completely neglect the chemical nature of
the mixed compounds and what went into the respective
parameterization. for example neither model is polarizable, but if you
put ions next to a metal it will be significantly polarized and the
ZrO2 potential in term includes the polarization (and
self-polarization) in an averaged manner.

what you need to do is either running a proper quantum mechanical
model (e.g. using DFT) or some approximations to it. the model
implemented in LAMMPS that comes closest is ReaxFF, but you'll have to
check if there is a suitable set of parameters for Zr/ZrO2. other than
that, you're probably confined to using codes like siesta or quantum
espresso or alike.

axel.