Re: [Ifeffit] Problem reproducing spectrum for Ti complex
Janine, I don't have a lot of advice, but there are a few more things to try. The RSIGMA keyword completely removes the broadening -- that's clearly no correct, but I think it will give you a sense of how the spectrum is being broadened by the exchange model. For that matter, have you tried one of the other exchange models? You might also consider using the remaining parameters in the POTENTIALS list. I don't quite remember how the default angular comentum maxima are determined, but if might be helpful to include more l states. Perhaps one of the Seattle folk has some better ideas...? B On Thursday 28 August 2008 04:42:31 Janine GRATTAGE wrote: > I have been trying to reproduce the XANES spectrum of a titanium > compound (titanium phthalocyanine dichloride) using FEFF 8.4. The > overall shape is reproduced reasonably well, but I wanted to ask the > mailing list users if they have any ideas as to how to improve the > details such as the pre-edge (to give two distinct peaks) and peak > positions over the absorption edge. > > I have attached a file showing the experimental spectrum, two FEFF input > files and the corresponding XMU output files. These two files have the > same atomic co-ordinates for the molecule, but one has 'standard' > broadening and the Fermi level has been shifted by -1eV (geom1) where > the other has reduced experimental broadening (-3) a shifted Fermi > level, and smaller step size (geom1acc). > > I have tried changing the geometry of the molecule by moving one or both > of the Cl atoms, and by trying a flat molecular geometry rather than the > 'umbrella' shape in the attached files, but the spectra attached are the > best match to experiment. Any help or suggestions on improvements would > be greatly appreciated. -- Bruce Ravel [EMAIL PROTECTED] National Institute of Standards and Technology Synchrotron Methods Group at NSLS --- Beamlines U7A, X24A, X23A2 Building 535A Upton NY, 11973 My homepage:http://xafs.org/BruceRavel EXAFS software: http://cars9.uchicago.edu/~ravel/software/exafs/ ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Comparison of "amp" between Feff8.40 and feff6 - Curved waves or Plane waves?
Hi Ying, I haven't tried your project files, but in general paths are sorted by reff. IFeffit makes no assumptions, however, about how paths have been sorted; by importing paths from different FEFF calculations, for instance, you can end up with different orders without a problem. Different versions of FEFF may sometimes number paths differently, though, primarily because the criteria for throwing out unimportant paths may be different. So it's important when comparing to make sure that you're actually using paths that represent the same scattering, and not just paths that are numbered the same. --Scott Calvin Sarah Lawrence College On Aug 29, 2008, at 1:54 PM, Ying Zou wrote: > Dear all, > > Nice to get so prompt response from Bruce and Scott, thanks for > their effort to make this list vital! > > Looks like "amp" is a factor which can be reduced. Then there is a > question: how would all paths be sorted? by Reff or by "amp"? More > techically related another question is then: can FEFFIT and FEFF > work seamless together on this? Is there any presumption in FEFFIT > (or more explicityly, ARTEMIS/ATHENA) which assumes all the paths > have been sorted by descending "amp" down from 100% with ascending > REFF? > > The background story behind this: using the same structural model > (thanks Bruce!), I tried to fit either use paths from FEFF840 or > from FEFF6. The fitting with FEFF6 paths does a quite good job, > giving a R_factor~1%, reduced chi square~ 90. Yet the one with > FEFF840 paths terminates abnormally because two parameters cannot be > determined. > > Attached please find two ARTEMIS project files. And appreciate for > comments! > > Ying > > ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Comparison of "amp" between Feff8.40 and feff6 - Curved waves or Plane waves?
By looking at the FEFF.inp file it appears that the path in question is a Ti-O-Ti where the angle of this linkage is very close to 180 degree (had it been 180 degree, the total length would be not 3.9317 but either 3.90500 or 4.15600. In a perovskite with Ti on center such angle would be 180 degree but in your input file you have TiO6 octahedron displaced relative to the Pb corners in 100 direction (also known as tetragonal distortion). Still, the angle is close to 180 degree which makes a contribution of forward scattering, triple-scattering paths that go through O in forward direction twice, the strongest - and that should explain why your path amplitude increased over 100 percent. Why the two programs give different amplitude, must be a bug, as Bruce and Scott explained. Anatoly -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Behalf Of Ying Zou Sent: Thursday, August 28, 2008 6:42 PM To: ifeffit Subject: Re: [Ifeffit] Comparison of "amp" between Feff8.40 and feff6 - Curved waves or Plane waves? Dear all, While I am calculating FEFF path through Feff8.40 using the "feff8.inp" as attached, I notice that the value of "amp" shown on "FEFF interpretation page" for some paths goes over 100%. For example, path 14 has a Reff=3.9317 far away from first shell, yet it has an "amp"=169.913. On the contrary, if I do the same in Artemis through embedded Feff6.0(using "feff6.inp"), none of paths has an "amp" over 100%. I am guessing this could be all right because Feff8 is using curved waves Feff6 instead plane waves. Any comments on this would be greatly appreciated! Ying Ying Zou Research Assoc., Dr.Phys. Physics Department, UWM ===feff8.inp=== * This feff.inp file generated by ATOMS, version 2.50 * ATOMS written by and copyright (c) Bruce Ravel, 1992-1999 * -- * -- * -- * -- * -- * -- * -- * -- * -- * -- * -- * -- * -- * -- * * total mu = 5194.4 cm^-1, delta mu = 761.6 cm^-1 * specific gravity = 7.942, cluster contains 25 atoms. * -- * -- * -- * -- * -- * -- * -- * -- * -- * -- * -- * -- * -- * -- * * mcmaster corrections: 0.00093 ang^2 and 0.165E-05 ang^4 * -- * -- * -- * -- * -- * -- * -- * -- * -- * -- * -- * -- * -- * -- * TITLE Perovskite: PbTiO3 EDGE K S02 1.0 * potxsph fms paths genfmt ff2chi CONTROL 1 1 1 1 1 1 PRINT 1 0 0 0 0 3 * r_scf [ l_scf n_scf ca ] SCF 3.88586 0 15 0.1 * ixc [ Vr Vi ] EXCHANGE 0 0 0 EXAFS RPATH 7.77172 * kmax [ delta_k delta_e ] *XANES 4.0 0.07 0.5 * r_fms [ l_fms ] *FMS 3.88586 * *RPATH 0.1 * emin emax resolution *LDOS -2020 0.1 POTENTIALS * ipot z [ label l_scmt l_fms stoichiometry ] 0 22Ti -1 -1 0 18O -1 -1 3 2 82Pb -1 -1 1 3 22Ti -1 -1 1 ATOMS 0.0 0.0 0.00 Ti 0.0 0.0 0.0-1.766301 Oapical 1.76630 1.95250 0.0 0.324171 Oplanar 1.97923 0.0-1.95250 0.324171 Oplanar 1.97923 -1.95250 0.0 0.324171 Oplanar 1.97923 0.0 1.95250 0.324171 Oplanar 1.97923 0.0 0.0 2.389701 Oapical 2.38970 -1.95250-1.95250 1.915922 Pb 3.36084 1.95250-1.95250 1.915922 Pb 3.36084 -1.95250 1.95250 1.915922 Pb 3.36084 1.95250 1.95250 1.915922 Pb 3.36084 1.95250-1.95250-2.240082 Pb 3.55563 -1.95250 1.95250-2.240082 Pb 3.55563 -1.95250-1.95250-2.240082 Pb 3.55563 1.95250 1.95250-2.240082 Pb 3.55563 -3.90500 0.0 0.03 Ti 3.90500 0.0 3.90500 0.03 Ti 3.90500 3.90500 0.0 0.03 Ti 3.90500 0.0-3.90500 0.03 Ti 3.90500 0.0 0.0 4.156003 Ti 4.15600 0.0 0.0-4.156003 Ti 4.15600 3.90500 0.0-1.766301 Oapical 4.28589 0.0-3.90500-1.766301 Oapical 4.28589 -3.90500 0.0-1.766301 Oapical 4.28589 0.0 3.90500-1.766301 Oapical 4.28589 END feff6.inp
Re: [Ifeffit] Comparison of "amp" between Feff8.40 and feff6 - Curved waves or Plane waves?
Ying and Bruce, Some versions of FEFF, at least, do allow amps above 100. They assign 100 to the first path, and then scale from there. The nearest neighbor path in your list is a single oxygen. I didn't run atoms on your input file, but I bet your path 14 is some combination of high degeneracy, focused, and involving the lead. Despite being further, it is more prominent, and thus correctly assigned a higher amp. It may be that this is true for only some versions of FEFF. I've certainly seen it frequently, although I haven't paid attention to whether it's version or even platform dependent. Personally, despite being documented, I would describe the behavior that Bruce describes, where the meaning of amp gets rescaled if a more prominent path is encountered, as a "known bug," while the behavior where amps greater than 100 assigned as more ideal. In any case, this is just cosmetic; it shouldn't affect the feff path itself. --Scott Calvin Sarah Lawrence College On Aug 29, 2008, at 9:32 AM, Bruce Ravel wrote: > On Thursday 28 August 2008 18:42:12 Ying Zou wrote: >> While I am calculating FEFF path through Feff8.40 using the >> "feff8.inp" as >> attached, I notice that the value of "amp" shown on "FEFF >> interpretation >> page" for some paths goes over 100%. For example, path 14 has a >> Reff=3.9317 >> far away from first shell, yet it has an "amp"=169.913. On the >> contrary, if >> I do the same in Artemis through embedded Feff6.0(using >> "feff6.inp"), none >> of paths has an "amp" over 100%. I am guessing this could be all >> right >> because Feff8 is using curved waves Feff6 instead plane waves. Any >> comments >> on this would be greatly appreciated! > > > Ying, > > Two things: > > 1. Both Feff6 and Feff8 are a curved wave calculations > > 2. From the Feff6 document (the definition is unchanged in Feff8, to > the best of my knowledge, although this definition is not included > in the Feff8 document): > > The curved wave importance ratios are the importance of a > particular path relative to the most important path encountered > so far in the calculation. If the first path is the most > important in the problem, all the importance factors will be > expressed as a fraction of that path. However, if the third > path considered is the most important, and the first path is the > next most important, path 1 will have a factor of 100%, path 2 > will be a fraction of the first path's importance, path 3 will > have an importance of 100% (since it is now the most important > path), and subsequent paths will be expressed as a percent of > path 3's importance. > > > So, an amp of 170 for path 14 must be a bug in Feff8. By its own > definition, 100 is the largest possible value. > > I suspect that if you compare that path from the two Feff > calculations, > they will, in fact, be quite similar. > > B ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Comparison of "amp" between Feff8.40 and feff6 - Curved waves or Plane waves?
On Thursday 28 August 2008 18:42:12 Ying Zou wrote: > While I am calculating FEFF path through Feff8.40 using the "feff8.inp" as > attached, I notice that the value of "amp" shown on "FEFF interpretation > page" for some paths goes over 100%. For example, path 14 has a Reff=3.9317 > far away from first shell, yet it has an "amp"=169.913. On the contrary, if > I do the same in Artemis through embedded Feff6.0(using "feff6.inp"), none > of paths has an "amp" over 100%. I am guessing this could be all right > because Feff8 is using curved waves Feff6 instead plane waves. Any comments > on this would be greatly appreciated! Ying, Two things: 1. Both Feff6 and Feff8 are a curved wave calculations 2. From the Feff6 document (the definition is unchanged in Feff8, to the best of my knowledge, although this definition is not included in the Feff8 document): The curved wave importance ratios are the importance of a particular path relative to the most important path encountered so far in the calculation. If the first path is the most important in the problem, all the importance factors will be expressed as a fraction of that path. However, if the third path considered is the most important, and the first path is the next most important, path 1 will have a factor of 100%, path 2 will be a fraction of the first path's importance, path 3 will have an importance of 100% (since it is now the most important path), and subsequent paths will be expressed as a percent of path 3's importance. So, an amp of 170 for path 14 must be a bug in Feff8. By its own definition, 100 is the largest possible value. I suspect that if you compare that path from the two Feff calculations, they will, in fact, be quite similar. B -- Bruce Ravel [EMAIL PROTECTED] National Institute of Standards and Technology Synchrotron Methods Group at NSLS --- Beamlines U7A, X24A, X23A2 Building 535A Upton NY, 11973 My homepage:http://xafs.org/BruceRavel EXAFS software: http://cars9.uchicago.edu/~ravel/software/exafs/ ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit