How to read EAM potentials

GULP
1

Hello.

I am trying to load an EAM potential in Gulp to calculate elastic properties in Al.
The potential is available online, here: Al - EAM potentials

However, this potential is given through 3 different files:
-Al.gulp.denfn
-Al.gulp.eamfn
-AlAl.gulp.pair

, and I think that it is not clear how I can read these files within a Gulp input file. They do not have .lib format.

Thanks in advance for your help.
Kind regards,
Jack

2

Hi Jack,
I’m assuming these are Paradyn file formats with numeric EAM. In which you follow the “numeric” input as given below:

@@eam_functional
Topic : eam_functional
Type : option
Format : eam_functional <square_root> <banerjea_smith n> <johnson/glue/foiles/mei-davenport/baskes/vbo/igarashi/spline/numeric>
: if square_root or power :
: atom1 A_1
: atom2 A_2 etc…
: if banerjea_smith :
: atom1 F0_1 F1_1 rho0_1 <3flags>
: atom2 F0_2 F1_2 rho0_2 <3flags> etc…
: if johnson :
: atom1 F0_1 F1_1 rho0_1 alpha beta gamma <6flags>
: atom2 F0_2 F1_2 rho0_2 alpha beta gamma <6flags> etc…
: if glue :
: atom1 rho1 rho2
: c1_4 c1_3 c1_2 c1_1 c1_0
: c2_4 c2_3 c2_2 c2_1 c2_0
: c3_3 c3_2 c3_1 c3_0
: atom2 rho1 rho2
: c1_4 c1_3 c1_2 c1_1 c1_0
: c2_4 c2_3 c2_2 c2_1 c2_0
: c3_3 c3_2 c3_1 c3_0
: if foiles :
: atom1 F0_1 F1_1 F2_1 F3_1 <4flags>
: atom2 F0_2 F1_2 F2_2 F3_2 <4flags> etc…
: if mei-davenport :
: atom1 Ec_1 alpha_1 beta_1 gamma_1 delta_1 phi0_1 s_1_1 s_2_1 s_3_1 <9flags>
: atom2 Ec_2 alpha_2 beta_2 gamma_2 delta_2 phi0_1 s_1_2 s_2_2 s_3_2 <9flags> etc…
: if baskes :
: atom1 Ec_1 A_1 rho0_1 <2flags>
: atom2 Ec_2 A_2 rho0_2 <2flags> etc…
: if vbo :
: atom1 A_1 rn_1 <2flags>
: atom2 A_2 rn_2 <2flags> etc…
: if igarashi :
: atom1 A_1 B_1 <2flags>
: atom2 A_2 B_2 <2flags> etc…
: if spline :
: atom1 A_1 B_1 C_1 D_1 rho0_1 rho_max_1 <4 x flags>
: atom2 A_2 B_2 C_2 D_2 rho0_2 rho_max_2 <4 x flags>
: if numeric :
: atom1 filename

This means you would write:

eam_fn numeric
Al Al.gulp.eamfn

and follow the same process for the other cases.
Of course an easier option is to use one of the EAM or MEAM models that comes with GULP in the libraries meam, meam_2nn, meidavenport, suttonchen etc.

Regards, Julian

3

Dear Julian,

Thanks a lot for your swift response.

It looks like these files have the same format as the ones used in Lammps. Indeed, if these 3 files are put together in a single file, then this would correspond to the Lamps potential file.

Should I use the following?

manubody numeric
Al AlAl.gulp.pair
eam_fn numeric
Al Al.gulp.eamfn
eam_density numeric
Al Al.gulp.denfn

Thanks in advance for your help,

Kind regards,
Jack

4

Dear Jack,

Apologies but my memories were a bit hazy since it’s been a while since I’ve looked at that part of the code. It appears that at present only the numerical format is supported for the functional. While there are options to input spline points for the density and potential they use GULP’s input format and so the files would have to be converted.

The focus of GULP has tended to be on analytic forms (it makes fitting easier) and so it would be better to work with one of these if possible as in the library files. Without looking at the specific model you are trying to use, it’s generally the case that MEAM-2NN improves on EAM for a lot of things.

Regards,
Julian

6

Thanks a lot for your reply.

I will use the meam_2nn library. Is it possible to calculate phonon dispersions with this MEAM potential? I get the following error:

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!! ERROR : Finite difference phonons only allowed for gamma point
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
7

Unfortunately it’s not currently possible to do phonon dispersion with MEAM. This is because the analytic second derivatives are not yet available (it gets complicated due to the use of a manybody screening term within a manybody formalism) and so only the numerical gamma point force constants are available. Sorry. I’ll add it to the list of things to try to implement, but it’s not a trivial thing.

8

Does this happen with EAM potentials too?

9

No - EAM is fine with k points since them manybody screening term is not used.

10

Dear Julian,

Thanks a lot for your swift reply.
The EAM potential will be enough.