Pt-Ni phase diagram (phb: the end of two-phase equilibrium)

Alloy Theoretic Automated Toolkit (ATAT)
1

Hello,

I am new to ATAT. When generating a Pt-Ni phase diagram, I ran into a problem that the composition-temperature phase boundaries of single-phase (Pt3Ni and PtNi) are incomplete.
Here is the description of the problem.
(1) I used my own cluster expansions (lat.in, clusters.out, eci.out, and gs_str.out)
Inside gs_str.out, pure Pt, ordered Pt3Ni, ordered PtNi, ordered PtNi3, and pure Ni are listed, which correspond to gs=0, gs=1, gs=2, gs=3, and gs=4, respectively. The disordered phase (Pt-Ni solid solution) is gs=-1, according to the manual of ATAT.
(2) To build the phase diagram in the Pt-rich region, I tried to compute the two-phase equilibria.
For example, for the two-phase equilibrium between phase 1 (Pt3Ni) and phase 2 (PtNi),
(2-i) I ran the following phb command:

phb -gs1=1 -gs2=2 -dT=10 -dmu=0.001 -er=20 -ltep=1e-3 -o=phb-gs1=1-gs2=2-test1.out -k=8.617e-5 -dx=1e-3

This command automatically finished at T=540 and mu=-0.0385916, which meant that the "end" of two-phase equilibrium has been reached, and a third phase appears (disordered phase -1).

(2-ii), in order to compute two new two-phase equilibria (phase 1 and phase -1, & phase 2 and phase -1), I separately ran the following two commands:

phb -T=540 -mu=-0.0385916 -gs1=1 -gs2=-1 -dT=10 -dmu=0.001 -er=20 -ltep=1e-3 -o=phb-gs1=1-gs2=-1-test1-sT=540.out -k=8.617e-5 -dx=1e-3
phb -T=540 -mu=-0.0385916 -gs1=-1 -gs2=2 -dT=10 -dmu=0.001 -er=20 -ltep=1e-3 -o=phb-gs1=-1-gs2=2-test1-sT=540.out -k=8.617e-5 -dx=1e-3

Unfortunately, the two commands in step of (2-ii) did complete the phase diagram. The problems I have are:
(A) the phb output of the calculations for the equilibrium between the ordered phase 1 (Pt3Ni) and phase disordered -1 (solid solution) stopped at Pt%=57% & T=560 K, and resembled the results for the equilibrium between the phase 2 (PtNi) and phase -1, which leads to a problem that the left boundary for the region of single-phase Pt3Ni is missing.
(B) Moreover, the phb command to compute the equilibrium between ordered phase 2 (PtNi) and disordered phase -1 finished at Pt%=58% and T=550K so that I was NOT able to identify the transition temperature for PtNi phase to transit to the disordered solid solution phase, which is usually at Pt%=50%.
(C) the similar problem happened for the composition-temperature phase boundaries between phase 0 (pure Pt) and phase 1 (Pt3Ni), & between phase 2 (PtNi) and phase 3 (PtNi3).

Could you please provide some insightful advice for the calculated incomplete Pt-Ni phase diagram?

Thanks.

Liang

2

Cannot say much but here are some things to try:

  • your cell is rather small with er=20. Try a larger value (40 or more)
  • You can run emc2 in the region that looks troublesome to see how the configurations generated by MC look like (you can visualize mcsnapshot.out)
  • I am not sure why you set dmu, in phb, this is not really necessary.
3

"- your cell is rather small with er=20. Try a larger value (40 or more)"
I tried er=40, 50 and 60. There is no improvement for generated phase diagrams

"- You can run emc2 in the region that looks troublesome to see how the configurations generated by MC look like (you can visualize mcsnapshot.out)"
I tried to generate the Gibbs free energy surfaces (e.g. Figure 5.4 a in the ATAT manual). I am confused which column should be used as "Gibbs free energy"?
The third column is "E-mux". Should I directly use the 3rd column or use 3rd_column + 2nd_column4th_column=E-mux + mux=E as the "Gibbs free energy"?
Thanks

"- I am not sure why you set dmu, in phb, this is not really necessary."
You are right. No need to set dmu for phb function.

4

"Should I directly use the 3rd column or use 3rd_column + 2nd_column4th_column=E-mux + mu*x=E as the "Gibbs free energy"? "
Depends on what you want to do.
If you want canonical quantities, you can just use the -g2c Option.

emc can help you to understand which configurations ATAT produces during equilibrum - explore the problematic regions and look at the structures.

5

Hi mbaeker,

Thanks for your reply. I solved this problem by decreasing the dT.

One quick question: the meaning of er.
If we set er=20, then the # of supercell will be 202020=8000 of unit cells.
Here, the unit cell is the primitive cell defined in lat.in, am I right?
For the Pt-Ni system I work on, the # of atom in the fcc primitive cell is one. Then, the total atoms within the simulation cell will be 8000 * 1 =8000. Am I right?

Thanks.

6

Not exactly. -er is specified in distance units (e.g. angstrom, if that’s what your lat.in uses).
As explained in https://dx.doi.org/10.1088/0965-0393/10/5/304, -er gives the radius of a sphere and the code find the smallest system that can fit that sphere.
If the unit cell is not cubic, the code will give different multiple of unit cells along different directions.