On Thursday 09 February 2006 09:24, Valmor Mastelaro wrote:

...

Hello Bruce,

could you please help me with the following question:

I truing to calculate the xanes spectra of Pb1-xLaxTiO3 solid solutions. I calculated the PbTiO3 xanes spectra and is quite similar to published one. When I substitute some Pb atoms by La atoms, even for x=15%, I only observed few changes on the calculated spectra compared with the undoped one.

Please, could you verify if I´m using a correct feff imput file (in attach)...

thanks for your help

Regards

Hi Valmor, I hope you don't mind that I am CCing my response to the Ifeffit mailing list. Your question is one that I see often, so I think that others would benefit by seeing this discussion. In a solid solution, one expects the dopant to substitute stochastically into the lattice. That is, in the absence of some kind of structural phase transition that leads to a different crystal symmetry, we expect the dopant to enter randomly into the crystal. In your case, each Pb site in the crystal has a 15% chance of being replaced by a La atom. If we ignore La-La correlations, the probability of a Pb site being replaced by La is unaffected by the occupancy of the adjacent Pb sites. Here are the first several lines of the feff.inp file you sent me, with several lines not related to this discussion deleted: TITLE PLT10 Com lantânio,a=3.885,c=4.139 HOLE 1 1.0 POTENTIALS * ipot z label 0 22 Ti 2 3 0.01 1 8 O 2 3 3.00 2 82 Pb 3 3 0.90 3 22 Ti 2 3 0.99 4 57 La 3 3 0.10 ATOMS .00000 .00000 .00000 0 Ti .00000 .00000 .00000 -1.87041 1 axial 1.87041 -1.95921 .00000 .33980 1 Planar 1.98846 .00000 -1.95921 .33980 1 Planar 1.98846 1.95921 .00000 .33980 1 Planar 1.98846 .00000 1.95921 .33980 1 Planar 1.98846 .00000 .00000 2.15145 1 axial 2.15145 1.95921 -1.95921 1.91754 4 La 3.36956 1.95921 1.95921 1.91754 2 Pb 3.36956 -1.95921 -1.95921 1.91754 2 Pb 3.36956 -1.95921 1.95921 1.91754 2 Pb 3.36956 1.95921 -1.95921 -2.10432 2 Pb 3.47925 -1.95921 -1.95921 -2.10432 2 Pb 3.47925 1.95921 1.95921 -2.10432 2 Pb 3.47925 -1.95921 1.95921 -2.10432 2 Pb 3.47925 .00000 -3.91842 .00000 3 Ti 3.91842 .00000 3.91842 .00000 3 Ti 3.91842 3.91842 .00000 .00000 3 Ti 3.91842 -3.91842 .00000 .00000 3 Ti 3.91842 .00000 .00000 4.02186 3 Ti 4.02186 .00000 .00000 -4.02186 3 Ti 4.02186 This feff.inp file does NOT adequately represent a solid solution. The Pb sites here are NOT occupied by La 15% of the time. In this example, there is one Pb site that is occupied by La 100% of the time and all the rest are never occupied by La. That's NOT a solid solution -- that's a different crystal. Thus, this feff.inp file might be some kind of approximation of a solid solution, but probably not a very good one. Unfortunately, Feff does not allow you to put 15% of an atom at a site. Each site is either all or nothing. That doesn't mean that Feff is not helpful in the case of a solid solution -- it just means that you have to do more work. One way of better approximating a solid solution is to use a random number generator and to run Feff many times. Each time you run feff, pre-process the feff.inp file, using a random number generator to decide for *every* lead site whether it should be replaced with a La atom. If you run enough feff calculations with Pb randomly substituted in this way, the ensemble will be a good representation of a solid solution. You can add up the subsequent calculations until the sum of spectra converges. There is at least one person out there in Ifeffit-land who has done some work on approximating solid solutions in this way. He might be inspired to comment further on this topic. There is an additional issue that you are not considering in the feff.inp file that you sent me. It is quite unlikely that the La atoms rests exactly on the Pb site. Much more likely is that the La atom introduces some kind of local disorder, moving off-site, pushing surrounding atoms away or drawing them closer. Those distortions might be large enough to be significantly effect the local environment about the Ti atom. If so, ignoring them in the feff.inp file is obviously incorrect. One last thing: take a look at the PDF file of a talk I have been giving recently on multiple scattering theory: http://cars9.uchicago.edu/~ravel/talks/pimst.pdf. In section 3, I show a simple sequence of calculations on PbTiO3. One rule of thumb that I try to get across in that part of the talk is that light atoms tend to effect the XANES more than heavy atoms. I don't want to overstate this issue, but it is quite possible that the difference between Pb and La in the XANES of an oxide is pretty subtle. That is, the XANES is probably more sensitive to where the oxygen atoms are than to whether a Pb site is occupied by Pb or La. If that is the case, then, of the two issues I discuss above, the more serious is the possibility that La introduces distortions to the oxygen octahedra. Although your feff.inp file certainly does not do a complete job at modelling a solid solution, the more serious issue is probably that it does not introduce new distortions. And that is my best guess for why your feff.inp file does not seem to replicate the changes you see in your Ti-edge spectra. Hope that helps, B -- Bruce Ravel ---------------------------------------------- bravel@anl.gov Molecular Environmental Science Group, Building 203, Room E-165 MRCAT, Sector 10, Advanced Photon Source, Building 433, Room B007 Argonne National Laboratory phone and voice mail: (1) 630 252 5033 Argonne IL 60439, USA fax: (1) 630 252 9793 My homepage: http://cars9.uchicago.edu/~ravel EXAFS software: http://cars9.uchicago.edu/~ravel/software/exafs/