Dear friends from the Ifeffit list, specially Sebastiano and Bruce, Thank you very much for your advices. Here I am giving you a better explanation of the physical problem we are facing, and then I´ll give the approach we will try to follow now. We are working on a 50-50 Cr-Fe maghemite, which structure is known by high resolution XRD and Rietveld method. The synthesis of such a maghemite is difficult and some magnetic properties are being measured. We know, also, that pure Fe maghemite structure at T=10 K has ordered vacancies. We are trying to verify by XAFS if the local environment of Cr or Fe absorbers reflects ordered vacancies. With the aid of crystallographic codes we modeled the various vacancy configurations, including the one corresponding to the ordered Fe maghemite structure. By the Rietveld analysis and by the XANES Cr spectrum we know that Cr occupies only octahedral sites. We have tried to fit with Artemis the experimental Cr and Fe EXAFS spectra, comparing them with one model of the ordered vacancies structure. The results are not the best, and then we “suspect” that the model of ordered vacancies is not the good one. As I have written before, our idea was to obtain “average vacancies” FEFF files and to work with the fully occupied magnetite structure, but with some kind of “gray” atoms. These gray atoms would be the average of vacancies and Cr or Fe atoms in octahedral sites. After your comments, it seems too difficult to obtain the corresponding FEFF functions. As a way for testing the disordered model, we are now working on the much easier problem of fitting the magnetite structure to our EXAFS spectra and to obtain the corresponding interatomic distances and amplitudes. We hope their interpretation will be also easier than before. Thanks a lot again for your attention. Best regards Maria Elena 2012/5/31

Dear friends: Thank you very much for your kind answers. Believe me, now I am in a hurry, and I hope this evening I'll answer with more details. Regards Maria Elena Enviado desde mi oficina móvil BlackBerry® de Telcel

-----Original Message----- From: Bruce Ravel Sender: ifeffit-bounces@millenia.cars.aps.anl.gov Date: Thu, 31 May 2012 09:33:29 To: XAFS Analysis using Ifeffit Reply-To: XAFS Analysis using Ifeffit Subject: Re: [Ifeffit] how to obtain average vacancy files?

On Tuesday, May 29, 2012 10:31:43 PM María Elena Montero Cabrera wrote:

...

Dear friends, I am trying to fit with Artemis, not very successfully, a maghemite structure. The main difficulty seems to be the vacancies in the structure. My idea, following a recently given Bruce’s advice to a person in this list, is to obtain “average vacancies” FEFF files and to work with the fully occupied magnetite structure. I wanted to obtain each file for a given vacancy configuration in the first coordination sphere and then to multiply the obtained functions by the weight of each configuration and summ them all. I had the hope of doing this by using the feff00nn.dat files that I have used when working with SixPack, some years ago. Surprisingly, I couldn’t find the Artemis’ corresponding feff00nn.dat files at any folder. What can I do?

There are two questions here. I'll answer the second one first.

It would be reasonable to say that the whole point of Artemis is to free the user from having to manage the feffNNNN.dat files. They get stashed in an out-of-the-way location so that the user can focus on creating a fitting model rather than keep track of many (dozens, hundreds,...) of little text files.

You are not very clear about what, exactly, your confusion is in terms of the vacancy thing. So I can only make a guess.

In general, vacancies are hard to model in Feff. In any case, it is may be imporssible to do a good job of modeling a vacancy with a /single/ Feff calculation. The reason is the same as the reason as for dopants -- Feff requires that every point in the cluster be occupied by only one kind of thing -- i.e. an atom or nothing. In general, you cannot know how to distribute the vacancy in the small cluster represented in the feff.inp file.

The word "vacancy" makes much more sense in a crystallographic context where the structure is averaged over 10^23 atoms. In an EXAFS context, you have to figure out how to model both of the main effects of missing atom:

* Reduction in coordination

* Change in the distances to other neighboring atoms due to relaxation around the missing atom.

I think that, if you spend some time reading papers by the folks who answer questions around here, you will find that the prefered approach to solving a problem involving vacancies is to run Feff on the "native" structure, then to parameterize the fit in such a way that models the coordinaton reduction and the changes in bond lengths.

That is, I see dealing with a problem like yours as a modeling problem in Artemis, not as a problem requiring a clever use of Feff.

For starters, read some of Scott Calvin's papers on mixed ferrites to understand how he approached the problem of mixed occupancy.

B

--

Bruce Ravel ------------------------------------ bravel@bnl.gov

National Institute of Standards and Technology Synchrotron Methods Group at NSLS --- Beamlines U7A, X24A, X23A2 Building 535A Upton NY, 11973

Homepage: http://xafs.org/BruceRavel Software: https://github.com/bruceravel

_______________________________________________ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit

-- María Elena Dra. María Elena Montero Cabrera Centro de Investigación en Materiales Avanzados (CIMAV) Miguel de Cervantes 120, Compl. Ind. Chihuahua Chihuahua CP 31109, Chih. México Tel (614) 4391123