[lammps-users] Need your help for alloy EAM potential in Lammps!!

Hi Steve and Guys:

Thank you very much for help!!

Yes, the commands should be:
pair_style eam
pair_coeff 1 1 Zr.eam
pair_coeff 2 2 Ni.eam


------Note that the Zr.eam and Ni.eam files above must be in funcfl format,
------so you would have to convert your fs and alloy files to that.

So I tried to convert Zr_2.eam.fs and Ni99.eam.alloy to funcfl format. It is easy to get
3 parts(Embedding function, pair interaction and electron density—3 parts) from old
files and inserted them into new funcfl files . But I run simulations which got error to
fail or gave “nan” results for both transfered Zr and Ni single funcfl files.

A DYNAMO single-element funcfl file is formatted as follows:
line 1: comment (ignored)
line 2: atomic number, mass, lattice constant, lattice type (e.g. FCC)
line 3: Nrho, drho, Nr, dr, cutoff
embedding function F(rho) (Nrho values)
effective charge function Z® (Nr values) (!!! Not pair interaction like
setfl and FS format !!!)
density function rho® (Nr values)

A DYNAMO multi-element setfl file is formatted as follows:

lines 1,2,3 = comments (ignored)
line 4: Nelements Element1 Element2 … ElementN
line 5: Nrho, drho, Nr, dr, cutoff
line 1 = atomic number, mass, lattice constant, lattice type (e.g. FCC)
embedding function F(rho) (Nrho values)
density function rho® (Nr values)
Following the Nelements sections, Nr values for each pair potential phi® array are
listed for all i,j element pairs in the same format as other arrays.

Same thing happens to FS EAM format.

Now the problem is that in funcfl format the second parts are not pair interaction but
effective charge instead. Maybe this is the possible reason for error.

So How should I do if we only have setfl and FS format for single element?

Thank you very much!!!

best regards


---------------------------Last email from Steve Plimpton

Hi - neither of your command sequences is correct. The 2nd pair_style
command will override the first.

If you want LAMMPS to mix the cross-interaction, then you
need to use pair_style eam like this:

pair_style eam
pair_coeff 1 1 Zr.eam
pair_coeff 2 2 Ni.eam

It will mix for the Zr-Ni interaction the same way DYNAMO did - not
sure what that means vis-a-vis the Johnson paper, but I can point you
to where this happens in the code.

Note that the Zr.eam and Ni.eam files above must be in funcfl format,
so you would have to convert your fs and alloy files to that.

Alternatively you can put both the Zr and Ni into one file (in alloy setfl
or FS format) and do this:

pair_style eam/alloy
pair_coeff * * NiZr.eam.alloy Zr Ni

But this will require you to do the mixing yourself b/c the
cross term is listed in the setfl file.

Please post future Qs to the LAMMPS mail list, so others can

Hope that helps,

[Hide Quoted Text]

Dear Steve:

I am Rongguang Xu, a Ph.D student in Prof Michael Falk’s group in Department of
Materials Science & Engineering at Johns Hopkins University.

I met a problem when using Lammps to simulate alloy system.

From the manual, Lammps can mix the single-element potentials to produce alloy

[Hide Quoted Text]

potentials, which sounds great to me. I tried many times to write down commands for this
and always failed. it was said that “incorrect pair coefficient”.

I have Zr_2.eam.fs for Zr in F-S format and Ni99.eam.alloy for single element Ni in
DYNAMO Setfl format. Following the manual. the most likely command seems like (Zr in
type 1 and Ni in type 2 defined before)

pair_style eam/fs
pair_coeff 1 1 Zr_2.eam.fs
pair_style eam/alloy
pair_coeff 2 2 Ni99.eam.alloy

We do not need to specify pair_coeff 1 2 because Lammps would mix them automatically.
But it failed.

I tried many different combinations of above commands and finally when I find that

pair_style eam/fs
pair_coeff * * Zr_2.eam.fs Zr Zr
pair_style eam/alloy
pair_coeff * * Ni99.eam.alloy Ni Ni

the simulation can be run without error. I am not sure it is correct or not.

Could you tell me the correct commands for producing alloy EAM potential based on 2
single EAM potential.

Furthermore, How does Lammps mix two single element EAM potential. Does it mix geometric
and arithmetic or apply invariance in R. A. Johnson’s alloy EAM model.

Thank you very very much for your offering any suggestions!

Best Regards


Rongguang, nice to know you here. I used to work in Dr. Ma’s group at JHU. As far as I know, the mix up of pure metals to form alloy does not work. It is not as simple as you thought. People have to do some fitting process to get the correct parameters (table files). Our group has devopled the Cu-Zr-Al alloy system and have validated the potential carefully, which is especially good for metallic glasses. You may want to find the following publications of our group useful. I can send you the potential file with lammps format if you wish.

Cheng YQ, Sheng HW, Ma E, “Atomic Level Structure in Multicomponent Bulk Metallic Glass”, Phys Rev Lett 2009;102:245501.

Cheng YQ, Cao AJ, Ma E, “Correlation between the elastic modulus and the intrinsic plastic behavior of metallic glasses: The roles of atomic configuration and alloy composition”, Acta Mater 2009;57:3253.

Cheng YQ, Cao AJ, Sheng HW, Ma E, “Local order influences initiation of plastic flow in metallic glass: Effects of alloy composition and sample cooling history”, Acta Mater 2008;56:5263.

L. Zhang, Y.Q. Cheng, A.J. Cao, J. Xu, E. Ma, “Bulk metallic glasses with large plasticity: Composition design from the structural perspective”, Acta Mater 2009;56:1154.

Cao AJ, Cheng YQ, Ma E, “Structural processes that initiate shear localization in metallic glass”, Acta Mater 2009 (in press).