Is it possible to use multiple files to define eam potential in LAMMPS?

Question

I have been trying to model ZrNiSn Half Heusler using the eam potential in LAMMPS MD package. For this I use a single eam potential file with the descriptions of all the species available in it. Also, I want to see the effect of the addition of Ti atoms on the thermal conductivity. However, the eam potential file which I have supports only Zr, Ni and Sn.

Even though I can find the eam potential for Ti on the internet, I am not sure about the ability of LAMMPS to model the Ti interactions with other species.

Can someone confirm whether it is possible to use such a separate file to model the Ti interactions. Would LAMMPS be supplied with the required data on the interactions of Ti with other species if this is done?

Answer 1

In general, It is not recommended to mix two inter-atomic potential, Let me go through the consequences, You want to mix $\ce{Ti}$ EAM potential to $\ce{Zr-Ni-Sn}$. You can do simply in lammps using pair style/overlays For example

pair_style hybrid lj/cut/coul/cut 10.0 eam lj/cut 5.0
pair_coeff 1*2 1*2 eam niu3
pair_coeff 3 3 lj/cut/coul/cut 1.0 1.0
pair_coeff 1*2 3 lj/cut 0.5 1.2

But Do you know interaction parameters for $\ce{Ti}$ with $\ce{Zr}$,$\ce{Ni}$ and $\ce{Sn}$. In broader sense, for EAM potential, you need three thing

1. Embedded function

2. Density function

3. Pair-wise interaction

Even If embedded function/density of $\ce{Ti}$ is same as $\ce{Zr}$, $\ce{Ni}$, $\ce{Sn}$, you must know pairwise parameters for $\ce{Zr-Ti}$, $\ce{Ni-Ti}$ and $\ce{Sn-Ti}$ for EAM/alloy. Things will be difficult if you want EAM/fs where density function for each cross-term is needed. Don't forget, all parameters should also be fitted for some properties as well. It is also worthy to note that fitting X number of parameters of material doesn't guarantee to give qualitative results for different Y number of parameters.