Since in your case it is not a random alloy (I just realized that Co does not substitute for Ga, only Ga for Co, and thus this material should have compensating vacancies on Ga sites), you should include the possibility of vacancy in your fitting scheme. For the first neighbors it will be: chi(Co-1NN) = chi(Co-Ga)*y1 where y is the Ga site occupancy (provided that chi(Co-Ga) was obtained for a model CsCl structure with 8 nearest neighbors per atom). chi(Ga-1NN) = chi(Ga-Co)*x + chi(Ga-Ga)*y2 where x + y are less than 1 (again, provided that each contribution was calculated for the ideal CsCl structure where Ga substituted for Co as described below). It also help that your alloy has a 50/50 composition of Ga and Co - it is only in that case y1 is always equal to y2 and you should constrain it as such in the fit. Then, when you do the fit to the 1NN shell, do it simultaneously for both Ga and Co edges, and vary x and y. It is not easy to say in advance what kind of uncertainties you will be getting since you will vary two parameters, not one as it is usually done in the case of random alloys, and if the correlations will be large you may decide to introduce another contraint, e.g., a known value of site occupancy of Ga (i.e., y = 0.2). Anatoly ________________________________ From: ifeffit-bounces@millenia.cars.aps.anl.gov on behalf of Frenkel, Anatoly Sent: Sun 11/8/2009 12:15 PM To: XAFS Analysis using Ifeffit Subject: RE: [Ifeffit] modeling anti-site disorder I would make a feff.inp file corresponding to the Co as a central atom in CoGa in its original CsCl structure. Then, I would substitute all Ga atoms by Co (the two sites are crystallographically equivalent in the bcc lattice, and there is no need to do it separately for Cs and Cl sites). That would give me Co-Co contributions for all single scattering, and Co-Co-Co...-Co contributions for all multiple-scattering paths. Then, I would do the same by making a feff.inp file for Ga as an absorber in its CsCl structure. Then, again, substitute all Co for Ga and create Ga-Ga and Ga-Ga-Ga...--Ga contributions. If you REALLY want to include multiple-scattering contributions of collenear paths (going from the corner through the body center through the opposite corner along the cube diagonal) you would have to also generate mixed multiple-scattering paths by leaving one and/or two shells around Co(Ga) un-substituted in such models. That would give you Co-Ga-Ga, Co-Ga-Co, etc. linkages and their corresponding double and triple scattering collinear paths. In short, it is just to make a number of models and save the resulting feffxxxx.dat paths on your disk for each of them, and make a table for yourself, which one corresponds to what linkage. Then, you can add them all together in your fitting model and vary their mixing fractions. For example, if you want to fit the first shell only, you should fit it as: x*chi(Co-Co) + (1-x) chi(Co-Ga) for Co edge. It is of course best to do it by fitting Co and Ga edge concurrently, within the same run of your fitting program, by imposing multiple constratints on your Co-Ga distances, mixing fractions and debye-waller factors. There are large number of papers describing multiple data set (multiple edge) fitting schemes in alloys which use this very simple method. Anatoly ________________________________ From: ifeffit-bounces@millenia.cars.aps.anl.gov on behalf of Shashwat Sent: Fri 11/6/2009 9:37 PM To: ifeffit@millenia.cars.aps.anl.gov Subject: [Ifeffit] modeling anti-site disorder Dear All, I am studying the local disorder caused in CoGa alloy (B2: CsCl structure) due to ball-milling; it is expected that Ga atoms substitute most of the Co atoms in the Co sublattice after prolonged milling. In order to model this "anti-site" disorder (occupation of Ga atoms on the Co sites) in ATOMS, I am considering to change the occupancies at the corresponding sites as given below: Before ball-milling: site a: occupancy of Ga = 1 site b: occupancy of Co = 1 (standard .cif data) After ball-milling (assuming 80% anti-site concentration): site a: occupancy of Ga = 0.2 (plus compensating vacancies) site b: occupancy of Co = 0.2 site b: occupancy of Ga = 0.8 Could anyone suggest if this approach is reasonable. Many thanks ... Regards, Shashwat