Help with Electron Denisty in EAM funcfl format

LAMMPS LAMMPS Mailing List Mirror
1

Dear LAMMPS users,

My name is Joey Vella, and I’m a graduate student in the department of Chemical and Biological Engineering at Princeton University. One of my future goals is to generate an EAM potential in the standard funcfl format for lithium. Before I do that I wanted to make sure I understood the lay out of the file. I did this by reading the LAAMPS documentation on the website, and examining/plotting values in the EAM files built-in to the package (mostly the Cu_u3.eam file in bench/POTENTIALS). The embedding function and effective charge tables seem straight forward to me and I was able to verify the values in the table by examining the relevant equations and graphs in the reference given in the file (S.M. Foiles et. al Physical Review B.,33, 7983, (1983)). However, when examining the electron density I’ve had some issues. First of all, I don’t know the units of electron density (I suspected Angstroms^(-3) but I’m not positive), the documentation on the website doesn’t seem to know either. I tried to figure out the units by integrating the values for electron density over a spherical volume and seeing if I recovered the number of valance electrons for Copper (11), but that didn’t work either.

My main issue is I don’t know the units for electron density in the funcfl format, or if the electron density is presented in a different format than simply rho as a function of r.

I checked the previous questions, but I couldn’t find a straight forward answer.

If someone could give me an answer or at least point me in the correct direction that would be great.

Thanks,
Joey

PhD Student

Department of Chemical and Biological Engineering

Princeton University

2

First of all, I don’t know the units of electron density (I suspected Angstroms^(-3) but I’m not positive), the documentation on the > website doesn’t seem to know either.

The format of EAM funcfl (and setfl) files pre-dates LAMMPS.
You could likely work out the units for electron density
by working thru the eqs in the pair_eam.cpp file where it is used.
There are a couple odd conversion factors if I recall.

If the papers don’t clarify, you can also send Stephen Foiles
an email - foiles at sandia.gov
If you work out a definitive answer, let me know and we’ll update
the doc page.

Steve

3

AFAIK the density in EAM is any unit of your choosing, as long as the embedding energy function uses consistent units. So if your transfer function is number density (Angstrom^(-3)) as a function of atomic distance, your embedding energy function is an energy as a function of a number density. But in principle you can call that whatever you like, *unless* you want that quantity to actually mean something - which for many (especially empirical) EAM potentials it does not.

In fact, the combined scaling of the transfer and embedding function is a well-known gauge degree of freedom of EAM potentials, see e.g. Ercolessi et al., Phil. Mag. A, 58 (1), 213–226, 1988. They choose the arbitrary unit of the transfer function so that for an atom of the bulk the total density is equal to the coordination number.

Peter