Re: [Ifeffit] problem with E0 (enot) parameters
Hi Zajac, What happens if you constrain all E0's to be the same? In the fit where they come out large, what are the uncertainties in the E0's? What are their correlations with other fitted parameters? There has been some debate in this list on the past as to how useful it is to allow for different E0's for different paths. It may be that Artemis is shifting the E0's for those paths in lieu of some other correlated parameter. --Scott Calvin Sarah Lawrence College On Jun 19, 2009, at 5:33 AM, Zajac, Dariusz A. wrote: > Hi all, > can anybody help me and send some link about problems with enot in > Artemis? > in google, tutorials I couldn't find any help. Of course are few posts > about delr or ss parameters, but enot is somehow omited (or I can not > find it...) > > the problem is with fiting K4W(CN)8*2H2O at W:L3 edge. > fit and others parameters looks ok, except enot for C and N, where > both > are around 12 and don't want to fit to other values... > thanks > darek ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] problem with E0 (enot) parameters
Hi Darek, You've got a false fit. The E0's aren't the biggest problem; look at the delr for potassium! Your fit is scrambling all the paths in non- physical ways. Your initial description of the system suggests that you have a decent guess as to the structure to start with. What happens when you run a fit with very few free parameters? (Perhaps none, or perhaps just floating an overall S02, E0, and maybe a couple of sigma2's.) Does it look qualitatively right, with peaks where there should be peaks? If so, then you probably need a tighter set of constraints. If not, then the material is not what you think it is. --Scott Calvin Sarah Lawrence College On Jun 19, 2009, at 7:30 AM, Zajac, Dariusz A. wrote: > Hi Calvin, > thanks for the email. I have tried to find something on mailing list > about E0 but it seems that my "searching words" were not correct. I > found only few posts but non of them explained (or followed) the > problem > with large enot. > enot parameters are closed to eachother e.g. enot_C = 12.597(+-0.654) > and enot_N=12.407(+-1.776); correlations e.g. enot_N and delr_N=0.744, > enot_C and delr_C=0.605 enot_C and enot_N=-0.313 > more detail you can find in the attachment or below. I have found that > the value of the first background variable is large (-3437(+-23034)) > when I constrain both enot's parameters I do not see huge changes in > parameters, chi^2 enot, amp, delr and ssh are the same in the range of > uncertainties. > I agree with you that program can shift not the correct parameter, but > why every time I change parameters (I shift parameters from the local > minimum) their come back? > > cheers > darek > > > Independent points = 45.279296875 > Number of variables = 30.0 > Chi-square =219223.767 > Reduced Chi-square = 14347.765383235 > R-factor= 0.022707172 > Measurement uncertainty (k) = 0.000102780 > Measurement uncertainty (R) = 0.000306293 > Number of data sets = 1.0 > > > Guess parameters +/- uncertainties (initial guess): > amp = 0.8496050 +/- 0.0767160(guessed as > 0.849812 (0.095158)) > enot_C =12.5967170 +/- 0.6539780(guessed as > 12.601228 (0.744286)) > delr_C =-0.0055260 +/- 0.0068920(guessed as > -0.005265 (0.007572)) > ss_C= 0.0016740 +/- 0.0008460(guessed as > 0.001676 (0.000991)) > enot_N =12.4065290 +/- 1.7757010(guessed as > 12.895967 (2.018563)) > delr_N = 0.0358140 +/- 0.0193390(guessed as > 0.042607 (0.029913)) > ss_N= 0.0068930 +/- 0.0016280(guessed as > 0.008196 (0.002599)) > enot_K = 0.0064050 +/- 9.7412920(guessed as > -4.066616 (13.931541)) > delr_K = 0.9265430 +/- 0.1815320(guessed as > -0.018996 (0.154984)) > ss_K= 0.0074030 +/- 0.0124260(guessed as > 0.011457 (0.010167)) > enot_O =-8.4988040 +/- 13.7627610(guessed as > 0.906319 (10.059312)) > delr_O =-0.3610320 +/- 0.0975640(guessed as > -0.139554 (0.069016)) > ss_O= 0.0010060 +/- 0.0050660(guessed as > 0.001140 (0.004395)) > > Def parameters (using "FEFF0: Path 1: [C5_1]"): > enot_CN =12.5016230 > delr_CN = 0.0151440 > ss_CN = 0.0042840 > enot_CNC=12.5333210 > delr_CNC= 0.0082540 > ss_CNC = 0.0034140 > enot_NCN=12.4699250 > delr_NCN= 0.0220340 > ss_NCN = 0.0051540 > enot_KN = 6.2064670 > delr_KN = 0.4811790 > ss_KN = 0.0071480 > enot_KC = 6.3015610 > delr_KC = 0.4605080 > ss_KC = 0.0045380 > enot_KNC= 8.3365500 > delr_KNC= 0.3189440 > ss_KNC = 0.0053230 > > Set parameters: > enot_H = -0.920939 (0.00) > delr_H = 0.103335 (0.00) > ss_H= 0.00749823 (0.00) > > Background parameters +/- uncertainties: > bkg01_01= -3437.6705554 +/- 23034.7299514 > bkg01_02=-0.0609379 +/- 1.8005458 > bkg01_03= 0.1948870 +/- 0.1595573 > bkg01_04=-0.0398830 +/- 0.0358870 > bkg01_05=-0.0025572 +/- 0.0147361 > bkg01_06= 0.0067564 +/- 0.0081250 > bkg01_07=-0.0031386 +/- 0.0052991 > bkg01_08=
Re: [Ifeffit] problem with E0 (enot) parameters
Hi Darek, OK, so if the K, H, and O don't affect the fit much for the C and N, and the K, H, and O are returning nonsensical values, then a logical possibility is that the E0's for C and N are correct. If you add 12 eV to the E0 you chose in Athena, where in the spectrum does it fall? Is it still before the white line? If so, it seems to me you don't have a problem. If not, then we have to ponder further. --Scott Calvin Sarah Lawrence College On Jun 19, 2009, at 8:28 AM, Zajac, Dariusz A. wrote: > Hi Scott, > look also at H and O, > but for me and for this fit important are only W-C and W-C-N bondings. > This sample is an reference sample for other cyano-brigded networks. > So > you suggest to focuse on K ions? how can it help with first 2 peaks? K > is at ~5A. > I have analysed in larger R space only to see how the spectrum behave. > contribution from K, O etc. at k highers than 5A is for me too low to > analyse it resonable for such compound. > I have attached in the previus post the last version of results. > Anyway, > enots for C and N do not change if I am enlarging R region (when I am > including next paths, also for K). > about material I am quite sure ;) and crystal structure is from > literature > > in the attachment you will find bmp file of the fit: data, fit, bkg > and > K path. > fitting ranges k(3-15) R(1.7-6) dk 2 dr 0.5, phase correction - > first C > > cheers > darek > ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] ODP: problem with E0 (enot) parameters
My email system didn't even let the message through, so I lost the thread for a bit. On the other hand, I for one DO want to encourage people to attach stuff that's less than 2 Mb when appropriate. It's very convenient to have, say, an Artemis project file to look at in some cases, and I often look at/write responses to these emails when I'm not connected to the web and thus can't follow a link. Is that an OK rule of thumb with the rest of you? Under 2 Mb => OK to attach; over 2 Mb use a link? Or would people prefer a lower threshold? --Scott Calvin Sarah Lawrence College On Jun 20, 2009, at 10:37 AM, Zajac, Dariusz A. wrote: > I appologize, > first and last time... > cheers > darek > > > -Wiadomość oryginalna- > Od: ifeffit-boun...@millenia.cars.aps.anl.gov w imieniu Stefan Mangold > Wysłano: So 2009-06-20 12:44 > Do: XAFS Analysis using Ifeffit > Temat: Re: [Ifeffit] problem with E0 (enot) parameters > > Please do not attach 10 Mb of data on your Mails. Just send an Link to > an Web-space with your e-mail. People can then download the stuff if > needed. > > Best regards > > Stefan ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] edge height proportional to the molar density?
Hi Haiyan, Yes, relative concentrations for different elements within the same sample can be estimated from edge jumps. Be sure to normalize by the difference in absorption coefficient for each edge--you can get those figures from Hephaestus, among other places. This method won't be as accurate as techniques like ICP or XRF, largely because it's hard to normalize XAS spectra consistently between different edges. But you can easily do better than 20% accuracy. --Scott Calvin Sarah Lawrence College On Jul 24, 2009, at 5:16 PM, Haiyan Zhao wrote: I am wondering whether the edge height could be used to estimate molar density or not. If so, is there any reference about such kind of calculation? Thanks! Haiyan ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
[Ifeffit] Kapton in glove box
Hi Todd, I've taken the liberty of posting your question to the Ifeffit mailing list. You're likely to get more accurate and quicker answers to these kinds of questions there. (For the rest of you: Todd is asking about the technique of preparing air-sensitive samples in a glove box, putting them on Kapton tape, sealing them in plastic bags, and transporting them to the beamline, shooting right through the bags.) I'll take my shots, though: It's hard for me to imagine adsorbed oxygen on the Kapton being more significant than the other sources of stray oxygen that can be present in a glove box. After all, the Kapton's in there too. And I don't think it's going to be more significant than the oxygen that diffuses through the plastic bags during transport. The thinner the Kapton tape, the better, as that will minimize the absorption due to the tape. It used to be hard to find 1 mil Kapton tape with adhesive, but now it's easy. Hephaestus will give you the absorption of Kapton, so you can judge how big an effect it will be at the energies at which you'll work. --Scott Calvin Sarah Lawrence College On Aug 4, 2009, at 2:39 PM, Monson, Todd wrote: Scott, Thanks. Is it pretty reasonable to assume that the kapton tape that you put your samples on doesn’t have any adsorbed oxygen that could affect your samples? Do you do anything to clean the kapton? Where are some good places to buy the kapton (and do you need to purchase rather thin kapton tape for doing XAFS)? Thanks again, Todd From: Scott Calvin Sent: August 04, 2009 11:34 AM To: Monson, Todd Cc: Scott Calvin Subject: Re: mossbauer Hi Todd, Regular zip-loc bags work just fine. For heat sealers I've used everything from a heat sealer manufactured for the purpose to a little propane torch--even a cigarette lighter should work. Putting one sealed bag inside another, if the energy you're working at allows it, seems to work quite well. --Scott On Aug 4, 2009, at 1:15 PM, Monson, Todd wrote: Scott, I had another question – what kind of plastic bags and heat sealers do you use for sealing up your air-sensitive XAFS samples? And where could I buy them? Thanks, Todd ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Kapton in glove box
Good point on the cardboard, Richard. I always pre-cut strips of Kapton tape, so that I wasn't bringing in cardboard at all. --Scott Calvin Sarah Lawrence College On Aug 5, 2009, at 8:13 PM, Richard Mayes wrote: Todd, Are you working with oxygen sensitive or moisture sensitive samples (or both)? If it's just moisture sensitive, then you can use regular 2-sided tape from your local office supply and polypropylene film to seal samples in polycarbonate or aluminum holders (or even pellets if you're lucky enough to be able to press pellets that hold their shape). Chemplex Industries is where I have gotten the polypropylene films I have used (and Kapton as well - www.findtape.com also has a good selection of Kapton tape). I used this method with many samples that involved heavily chlorided titanium on silica and had few problems if they're used within 5-7 days after packing in a glove box (the samples with problems resulted from improperly sealed samples). You can get jars (baby food jars work very well to ship individual samples) to store the samples for shipping and if you pack the jars in the glove box, you will have the box atmosphere in the jars, for a little while anyway. A note on oxygen sensitivity (and to an extent moisture sensitivity): you probably already know this, but I'll say it anyway...if cardboard is present in the role of Kapton tape, you may have oxygen/water diffusion from the cardboard for a few days after you take it into the box. Our rule of thumb was to pull vacuum on anything involving cardboard for at least 48 hrs before taking it into the box. All that to say, take your supplies into the box a few days ahead of time to allow your box catalyst to take care of any residual oxygen/water that make their way in. HTH, -Richard ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Kapton in glove box (Todd Monson)
Hi Todd, Kapton is resistant to most solvents, but that's not necessarily true for the adhesive on it! I've had the adhesive completely washed away by samples with a little solvent on them before. So you should test that before preparing your samples. --Scott Calvin Sarah Lawrence College On Aug 6, 2009, at 11:47 AM, Monson, Todd wrote: > Thanks for all your comments regarding kapton tape and measuring air > and moisture sensitive compounds. Many of my samples are indeed air > sensitive and not just moisture sensitive (iron nanoparticles). If > my particles are dispersed in solvent will the kapton be resistant > to that solvent (at least during the time that the solvent is > evaporating from the tape in the glove box)? Darek mentioned kapton > dots - could you tell me where I can purchase these? > > Thanks again for everyone's help, I thought this mailing list was > primarily for software related questions but I am finding it is > useful to get help on any XAFS related questions. > > Todd ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Chi in arthemis
Hi euG, I don't report chi-square at all. I know that the official IXS reports suggest it, but until the question of what to use for the measurement uncertainty is adequately addressed within the community, it seems to me that the number is not very meaningful in an absolute sense. It is meaningful, however, for comparing fits on the same data. (I do report R-factors to give some sense of closeness of fit.) On the other hand, uncertainties in fitted parameters are calculated in a highly defensible way and should be reported. If the uncertainties are large for parameters that are of interest to you, then that is indicative of a problem in the fit and should not be swept under the rug. --Scott Calvin Sarah Lawrence College On Aug 18, 2009, at 4:02 PM, Eugenio Otal wrote: > Hi all, > I see that the reports of the fits a perform in arthemis have Chi > really big. The fit is really good and the Chi should be closer to 1 > than to 100. > The problem, is what should I report as the Chi. Uncertain in delr > or ss have the same problem, are big like the results. > I found the topic in the list but not a good answer to what should I > report. > Thanks, euG > ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Chi in arthemis
Hi euG, That amp is not physically reasonable, unless you're using it as a proxy for the coordination number. The uncertainty on the other variables does not seem high to me for a single-shell fit. Well, 2 eV is a bit high for an uncertainty on E0, but not crazy high. There are some approaches that can be used to try to reduce the uncertainties, but you shouldn't even think about that until you get the amp (S02) straightened out. --Scott Calvin Sarah Lawrence College On Aug 18, 2009, at 6:18 PM, Eugenio Otal wrote: > Hi Scott, > here I copy a part of the report: > > Independent points = 6.222656250 > Number of variables = 4.0 > Chi-square = 247.145092496 > Reduced Chi-square = 111.193574128 > R-factor= 0.017422216 > > Guess parameters +/- uncertainties (initial guess): > amp = 6.7815290 +/- 1.4687660(1.) > enot= 2.2173620 +/- 2.1499920(0.) > delr= 0.0514640 +/- 0.0163900(0.) > ss = 0.0074020 +/- 0.0025220(0.0030) > > I see that the R-factor is pretty good, 1.74%, amp is high cause is > correlated with the coordination number and always have big errors., > delr has the error of the total distance, so it is ok, but ss and > enot have a a really big error, is this normal? > Thanks, euG ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Chi in arthemis
Report the error bars as given; otherwise you're reintroducing the unknown measurement uncertainty factor. Then somewhere in your paper cite ifeffit and make clear that you used that method to determine uncertainties. --Scott Calvin Sarah Lawrence College On Aug 18, 2009, at 7:31 PM, Eugenio Otal wrote: > Hi Scott, > I forget the amp, I left the N =1, that is the mistake, sorry I have > de S02 from other oxide to transfer it. > I found that error bars in Ifeffit are scaled by Chi2 reduced, so > should I transform the error or inform them directly as reported? > Regards, euG > ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Chi in arthemis
Matt, Is this the most recent IXAS report on error reporting standards? http://www.i-x-s.org/OLD/subcommittee_reports/sc/err-rep.pdf It uses a rather expansive definition of epsilon, which explicitly includes "imperfect" ab initio standards such as FEFF calculations. It indicates that statistical methods such as that used by ifeffit for estimating measurement error yields a lower limit for epsilon, and thus an overestimate of chi square. So I think my statement and yours are entirely compatible. As far as what should be reported, I do deviate from the IXAS recommendations by not reporting chi-square. Of course, I tend to work in circumstances where the signal-to-noise ratio is very high, and thus the statistical uncertainties make a very small contribution to the overall measurement error. In such cases I have become convinced that the R-factor alone provides as much meaningful information as the chi-square values, and that in fact the chi-square values can be confusing when listed for fits on different data. For those working with dilute samples, on the other hand, I can see that chi-square might be a meaningful quantity. At any rate, I strongly agree that the decision of which measurements of quality of fit to produce should not be dependent on what "looks good"! That would be bad science. The decision of what figures of merit to present should be made a priori. --Scott Calvin Sarah Lawrence College On Aug 18, 2009, at 10:40 PM, Matt Newville wrote: > Having a "reasonable R-factor" of a few percent misfit and a reduced > chi-square of ~100 means the misfit is much larger than the estimated > uncertainty in the data. This is not at all unusual. It does not > necessarily mean (as Scott implies) that this is because the > uncertainty in data is unreasonably low, but can also mean that there > are systematic problems with the FEFF calculations that do not account > for the data as accurately as it can be measured. For most "real" > data, it is likely that both errors FEFF and a slightly low estimate > for the uncertainty in the data contribute to making reduced > chi-square much larger than 1. > > And, yes, the community-endorsed recommendation is to report either > chi-square or reduced chi-square as well as an R-factor. I think some > referees might find it a little deceptive to report R-factor because > it is "acceptably small" but not reduced chi-square because it is "too > big". ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] limits for second shell
Hi Eugenio, To my eye, I strongly suspect there's real signal there. To confirm, try several different k-ranges and k-weights. If the feature persists, it is likely physical. (It may change size and shape to some extent.) --Scott Calvin Sarah Lawrence College On Aug 20, 2009, at 11:54 PM, Eugenio Otal wrote: > Hi, > I have a sample of a pure Er2O3 (blue line in the attached graph) > and a sample of doped ZnO with erbium that has segregated the same > oxide (red line) by thermal treatmen. > The signal for de segregated oxide gets noisy around k=9 because the > sample is so diluted, but the radial distribution shows second shell > signal, smaller than the pre oxide, but still a signal. My doubt is > about how to know if the second shell is real and if that second > shell can be useful to obtain information. Is there a criteria to > know that? Some limit in k-space? > Thanks, euG > ___ > Ifeffit mailing list > Ifeffit@millenia.cars.aps.anl.gov > http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] limits for second shell
Hi Eugenio, Perform your analysis on the data, using different values for k-max. If you are successfully extracting structural information which is different from the standard, you will find fitted parameters which: --are stable with changes in k-max --take on values that are different from those for the standard, taking into account the error bars (i.e. the values for the standard do not fall within the error bars for the sample for at least one parameter) --are stable with changes in k-weight. --Scott Calvin Sarah Lawrence College On Aug 21, 2009, at 11:35 AM, Eugenio Otal wrote: > Hi, > thanks for the help. > Let me be more clear. I am trying to know the limits to use > information from the second shell, from XRD I found that the phase > is present, but I want to know if I can use the signal to obtain > more information about second neighbors. > I have this sample and some more where I find a second shell signal, > the doubt is about how far can I trust in this information, is there > a limit in k-space to believe in the information of the second shell > and try to fit it? > I attach the R-space graph with the same limits for the FT, there > are still differences, what should I check to be sure the > information I can get from the second shell is trustable and not a > problem of the noisy signal? > Thanks, euG > > ___ > Ifeffit mailing list > Ifeffit@millenia.cars.aps.anl.gov > http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] mu(E) of theoretically calculated spectra
Hi Bhoopesh, There are files that are identified as cementite in Matt Newville's archive: http://cars9.uchicago.edu/~newville/ModelLib/search.html If you manage to get any other data, let me know; I've used the standard at the link above, but I'd like to confirm it. --Scott Calvin Sarah Lawrence College On Sep 24, 2009, at 11:45 AM, bhoopeshm wrote: > Dear All, > I am writing to see if anyone on this mailing list has > ever measured "Cementite" (Fe3C) EXAFS. It is basically an Iron > Carbide, nothing exotic at all. > > I would be very grateful to anyone who might want to share an EXAFS > spectrum of Cementite (Fe3C). > > > Regards, > Bhoopesh ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Cementite EXAFS - thanks
One thing I can say is that the EXAFS data in Matt's archive is reasonably well fit by a cementite structure. Since there's a lot of variation in cementite-like materials, that doesn't tell us exactly what it is, but at least it's in the right family. --Scott Calvin Sarah Lawrence College On Sep 24, 2009, at 5:49 PM, Bhoopesh Mishra wrote: > Hi Matt, > I imagined you might not remember about the data or how was > it prepared, so I decided not to bother you asking about it. But > sounds like you do have an amazing memory. > > Yes, I would be happy to characterize the sample by running > both EXAFS and XRD on it if you still have it. I would of course > share the information with the mailing list, once I characterize it. > I will collect the sample from you sometime soon. Thanks again for > offering. > > > Regards, > Bhoopesh > ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
[Ifeffit] Tenure track position for chemist
Hi all, Since presumably a lot of people on this list know early-career chemists with an interest in a position where teaching is valued and they can also collaborate closely with a XAFS expert (that's me, I guess), I thought I'd pass along information on our opening: Sarah Lawrence College, a coeducational liberal arts college dedicated to individualized education, invites applications from broadly trained chemists for a tenure-track position beginning August 1, 2010. The successful candidate will teach general chemistry, other undergraduate chemistry courses of interest to liberal arts students, and upper-level courses including physical chemistry. An interest in environmental science is desirable. A commitment to working closely with students on an individual basis is essential. Candidates should have a Ph.D. in Chemistry or expect to receive one by August 1, 2010. Application materials must include: cover letter, CV, statement of teaching philosophy and research interests, graduate transcript, descriptions of two courses suitable for liberal arts students, and three letters of recommendation (at least one of which must address the candidate's ability to teach general chemistry). Deadline for applications is November 16, 2009. To apply for the position please go to: https://slc.simplehire.com/applicants/jsp/shared/frameset/Frameset.jsp?time=1254862871561 For information on Sarah Lawrence College, our curriculum, teaching methods, and philosophy of education, please see our Web site at: http://www.slc.edu . SLC is an Equal Opportunity Employer committed to achieving a racially and culturally diverse community. --Scott Calvin Sarah Lawrence College___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Cu oxide fitting result
Hi Bruce, From April 10 to April 19 of this year, there was a thread discussing the basic strategy Abhijeet is using (first under the subject line "High S02" and then "fitting procedure". As I recall, you were busy with something else at the time, and saw there were enough experts answering so that the question would likely be fully addressed. What Abhijeet is doing, I believe, is fitting the first shell first, setting the relevant values, including S02, to those found from the first shell, and then shifting the R-range to fit a new set of paths. He then continues that iterative procedure, shell by shell. A number of people spoke up for that procedure, although concerns were raised as to whether it was the best approach for the particular system being discussed. What Abhijeet was doing before that discussion was: fit the first shell first, set the relevant values to those found from the first shell, and then extend the R-range to fit a new set of paths, continuing this as an iterative procedure. This meant that the fitting range continued to include paths that he was now constraining to their values from a previous fit. My recollection was that was generally agreed on to be bad practice, and the idea of shifting the R-range through narrow bands was offered as an alternative. Having participated in the discussion the first time around, I personally would not choose this procedure for the kind of system Abhijeet is looking at here: a highly crystalline system with strong contributions from paths at a wide range of distances that is expected to be similar to a known structure. But the point may be to practice this procedure for systems where it is more appropriate, in which case it makes sense to try it first on a known system. --Scott Calvin Sarah Lawrence College On Oct 9, 2009, at 11:47 AM, Bruce Ravel wrote: Abhijeet, I took a quick peak at your project and I find it very confusing. In the most recent fit, you are fitting only from 3.5 to 4.4 -- the area under the *third* peak in the data. I don't really know how to comment on this project because you have set a large number of parameters to seemingly arbitrary values. From a numerical perspective, the reason for questions 1 and 2 is because you are attempting to fit only a narrow and of the data and you have set the majority of your parameters. That seems an unlikely strategy to me. I think the best thing you could do would be to fit *all* of your data rather than an arbitrary and small band of it. ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Rbkg value
Hi Chris, One thing to keep in mind is that it is not wrong to eliminate part of your signal; it just means you're losing a little bit of data. This is similar to a common filtering mistake that beginners make when trying to choose the maximum end of their fitting range: they look at the paths they are including, and try to set Rmax high enough to include most of the contribution from the paths they are including. What they should be doing, however, is to look at the paths they are not including, and set Rmax low enough so that the contributions from those paths are tolerably small. Looking at correlations between background parameters and fitted parameters when the "fit background" option is selected also helps provide you with information. I'll admit that the "fit background" button confuses me a bit, though (I always have to spend ten minutes convincing myself again as to what exactly it is doing), so someone else should explain how that can be used to help address your question. --Scott Calvin Sarah Lawrence College On Oct 23, 2009, at 3:51 PM, Chris Patridge wrote: Hello everyone, In removing background, most literature suggests Rbkg value of 1.0 because below this represents mostly noise and low freq components not part of the scattering effect. In using Feff and viewing the individual paths calculated a number of them have paths near 1.5 A, therefore meaning that they have some contributions very close to 1.0 A and below in R space due to phase shift. Has anyone modeled materials which contain these rather short Reff and how did you decide what was scattering and noise? Thank you all, ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] SrSO4 (Celestite) and SrCO3
Hi Peter, I think I have one of SrCO3, but it will take me a couple days to dig it up. --Scott Calvin Sarah Lawrence College On Oct 26, 2009, at 7:10 PM, Peter Nico wrote: Hello All, Would anyone out there have a SrSO4 or SrCO3 standard XANES spectra they would be willing to share? There aren't any in the databases of which I am aware, namely "Lytle, GSECARS, and NSLS X18b." There is a SrCO3 but its really an EXAFS standard (the edge consists of ~3 points.) I am sorry that I can't necessarily promise a publication acknowledgment because I am not sure if the work will ever be published. However, if it is, I'll definitely acknowledge the help. Thanks. --Peter ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Bug in Athena?
Hi Matt, I'm the one who requested the merged reference channel. If the data is ideal, of course only one reference scan is needed. But there are two common ways it can be nonideal that are relevant: 1) The monochromator does not hold calibration; i.e. there is an energy shift between scans 2) The reference channel is very noisy, perhaps because of an inherently thick sample If 1) is a significant problem and 2) is not, then it makes sense to align the scans using the reference, at which point any reference scan will do for determination of the chemical shift of the merged data from the sample. If 2) is a significant problem and 1) is not, then it makes sense to merge the references along with the sample data, because that will make it easier to determine the chemical shift. If both problems are significant, then you've got a headache. --Scott Calvin Sarah Lawrence College On Nov 19, 2009, at 10:26 AM, Carlo Segre wrote: Hi Matt: I agree. It is useful to have the reference channel from the first of the merged data pulled over as reference for the merged data but this actuallly only makes sense if the user first aligned using the reference. carlo On Thu, 19 Nov 2009, Matt Newville wrote: Is there ever a case where a merged reference channel is useful? I thought the only possible use for a reference channel was for comparing individual scans. That is, prior to merging. --Matt On Thu, Nov 19, 2009 at 6:45 AM, Zajac, Dariusz A. wrote: Dear Bruce, Dear All, maybe it is a naïve question but I want to ask and to point this problem... Windows XP. Athena 0.8.059 Sc.Linux. Athena 0.8.060 I have a set of data with refernces (one sample, many scans). I have marked sample's groups and do "merge marked data in mu(E)" then I get merged data together with reference (2 groups: merge - sample, and Ref merge - reference). But... ...if I have marked reference sample's groups and do "merge" then I get 2 groups: merge - which is merged data of reference, and Ref merge - which is marged data of sample. Oposite to that I did in first example! Is any hidden idea, I can not see, why it should be that way? If you don't know about that, can confuse and surprise... cheers darek ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit -- Carlo U. Segre -- Professor of Physics Associate Dean for Graduate Admissions, Graduate College Illinois Institute of Technology Voice: 312.567.3498Fax: 312.567.3494 se...@iit.edu http://www.iit.edu/~segre se...@debian.org___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Bug in Athena?
Hi Bruce, On Nov 19, 2009, at 11:48 AM, Bruce Ravel wrote: Why does Athena make a merge of references? As Matt points out, that is an odd thing to do. I may be confused as to what we're talking about. Why is this an odd thing to do? It seems perfectly normal to me to want the reference scans merged as well as the sample scans, in order to get a clean measure of chemical shift. --Scott Calvin Sarah Lawrence College ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Bug in Athena?
On Nov 19, 2009, at 2:59 PM, Matt Newville wrote: For this case, wouldn't it be better to measure the reference separately to determine the chemical shift, and not rely on the reference channel for this purpose? How often is the reference channel both noisy AND improved by merging? That would imply a transmission measurement that was poor due to low flux. But if this is because the sample is thick as you suggest, the x-rays hitting the reference could be dominated by harmonics, and the reference data may just be bad, not noisy due to counting statistics. It's a good point. But pick your poison. When I am trying to be careful about chemical shift, I don't trust that the mono won't just happen to skip a step between measuring the standard separately and measuring the sample. So I do both. I measure a standard in the sample channel, with a reference in the reference channel. I then leave the reference in the reference channel, and put my sample in. If the sample is a "reasonable" thickness for transmission, but a bit on the high side (say 2.3 absorption lengths), the photon count is down pretty far by the time it gets to the reference. The reference is also often the worst detector and amplifier that a line has, as the good stuff is used for I0, It, and If. So the reference channel may well have a considerable amount of random noise which can be improved by merging. If that's the case, and if my sample appears to be suffering no beam damage (scans when aligned, lie on top of each other), then I align used the sample data. I then merge the sample data and the reference data. By comparing the sample to the reference and the previous scans where I measured the standard to the reference, I can see if there's been any energy shift between scans. As far as harmonics, this procedure should detect them. If the merged reference looks different from sample to sample (including the case where a standard was also in the sample channel), that suggests that there are issues with harmonics. If those issues move the first peak of the first derivative, I know they're going to affect my determination of chemical shift. Also, if I get a nonzero chemical shift from this procedure for the standard, I know there's an issue. If not, they're not a problem. The net result is that I have good confidence that I'm getting accurate chemical shifts, as loss of energy calibration, harmonics, and noise should all become evident by this procedure. I'm not recommending this procedure over others; it's just what I do in some cases. But it doesn't seem like an unreasonable procedure to me. --Scott Calvin Sarah Lawrence College ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
[Ifeffit] Lattice parameters: EXAFS vs. XRD
Merry Christmas, everyone! Yes, I'm pondering EXAFS on Christmas... Here's an issue that I bet has been worked out, and I bet someone on this list knows the result and where it's been published. It's well known that the MSRD ("sigma squared") for EXAFS differs substantially from the "Debye-Waller factor" in XRD, because the first is the variance in the interatomic distance, and the second is the variance in the atomic position relative to a lattice point. But what about the lattice parameter implied by the nearest-neighbor distance in EXAFS as compared to the lattice parameter found by XRD? It is certainly true that in most materials, particularly highly symmetric materials, the nearest-neighbor pair distribution function is not Gaussian, and generally has a long tail on the high-r side. (This is largely because the hard-core repulsion keeps the atoms from getting much closer than their equilibrium positions.) So imagine a set of atoms undergoing thermal vibrations around a set of lattice points. For concreteness, let's consider an fcc material like copper metal. The lattice points themselves are further apart than they would be without vibration, sure, but that's not the question. The question is whether the square root of two multiplied by the average nearest- neighbor distance is still equal to the spacing between lattice points. My hunch is that the answer is no, and that the EXAFS implied value will be slightly larger. While the average structure is still closed- packed, the local structure will not be. And in a local structure that is not closed-packed, the atoms will occasionally find positions quite far from each other, but will never be very close. In a limiting case where melting is approached, it's possible to imagine an atom migrating away from its lattice point altogether, leaving a distorted region around the defect. While XRD would suppress the defect, EXAFS would dutifully average in the slightly longer nearest-neighbor distances associated with it. Just to be clear, I am not talking about limitations in some particular EXAFS model used in curve-fitting. For example, constraining the third cumulant to be zero is known to yield fits with nearest-neighbor parameters that are systematically reduced. In fact, limitations like that mean the question can't be answered just by looking at a set of experimental results: I can make my fitted lattice parameter for copper metal go up or down a little bit by changing details of a fitting model or tinkering with parameters that themselves have some uncertainty associated with them, like the photoelectron's mean free path. (Fortunately, this kind of tinkering will affect standards and samples in similar ways, and thus don't affect my confidence in EXAFS analysis as a tool for investigating quantitatively differences between samples, or between samples and a standard.) My question is about the ACTUAL pair distribution function in a real fcc metal. To the degree it's a question about analysis, it's about XRD: "In an fcc metal should the expectation value of the nearest-neighbor separation, multiplied by the square root of two, equal the lattice spacing as determined by XRD?" --Scott Calvin Sarah Lawrence College ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Lattice parameters: EXAFS vs. XRD
Thanks, Matt--you give a complete and satisfying discussion of this on January 23 on this list. I forgot about that because it came at the tail end of a long discussion as to whether C3 could ever be 0, but I suspect what you said then was rattling around in the back of my head and only settled in last week. --Scott Calvin Sarah Lawrence College On Dec 30, 2009, at 8:47 AM, Matt Newville wrote: Hi Scott, I believe we had a conversation about this last January. XAFS is not sensitive to the crystallographic lattice constants. It measures the spacing between atoms. Because of thermal vibrations and other disorder terms, the average distance between atoms is larger than the distance between the lattice points. --Matt ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Fitting using Experimental standard
Hi Fiona, An experimental standard is a spectrum of a known material (the term is also often used for the known material itself, in addition to its spectrum). If you use Athena to do a linear combination fit, you are most commonly using experimental standards to do it. A theoretical standard is a theoretically simulated spectrum of some structure. Thus, the theoretical standard is not the crystallographic data itself, although that data can serve as a basis for generating a theoretical standard using software such as FEFF. --Scott Calvin Sarah Lawrence College On Jan 4, 2010, at 7:16 PM, Fiona R. Kizewski wrote: Dear all, Can somebody please explain to me what is theoretical standard and what is experimental standard. My understanding of theoretical standard is the crystallographic data. However, it is first time I heard experimental standard. Thanks ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Bruce ayuda, problemas con athena
You'll probably get Russian before I get Latin. --Scott Calvin Sarah Lawrence College On Feb 13, 2010, at 11:38 AM, Frenkel, Anatoly wrote: Can't wait for a question in Russian. Anatoly From: ifeffit-boun...@millenia.cars.aps.anl.gov on behalf of Ravel, Bruce Sent: Sat 2/13/2010 10:52 AM To: XAFS Analysis using Ifeffit Subject: Re: [Ifeffit] Bruce ayuda, problemas con athena On Friday 12 February 2010, 07:10:17 pm, Jaziel Soto wrote: > Me encuentro usando el athena 0.8.061 y el ifeffit 1.2.11c instalado en > windows vista de 32 bits. Tengo puesta la compatibilidad de windows xp > service pack 2, ejecutándolo con privilegios de administrador y después de > haber trabajado algo, cuando quiero grabar los .prj no pasa nada. Entonces > lo que trabajo no tiene sentido por que no lo puedo guardar. > Jaziel says he's using Athena 0.8.061 and Ifeffit 1.2.11c on 32 bit Vista with service pack 2. While running with Admin privileges and after working for a while, he cannot save .prj files. Jaziel, Primer cosa -- entiendo bien que prefieres escribir en tu idioma, pero te ayuda mucho escribir en ingles. Asi que puedes pedir ayuda de toda la communidad. Muchos de ellos saben mucho mas de Windows que yo. Aqui no hay sufficiente informacion. En esta situacion, es util abrir el command window y empezar athena con cle. Asi, si hay mensajes con informacion sobre el problema, pouedes copiarlos en el email. Hay la possibilidad que el problem es el mismo que esto: http://millenia.cars.aps.anl.gov/pipermail/ifeffit/2010-January/009243.html Si intentes grabar el fichero an un sitio con letras con accente, intente un sitio solo con letras sin accente. And then I said: First thing, write in English. Clearly it is esier to write in your own language, but if you write in English anyone on the list will be able to help you. Many of the people here know more about Windows than I do. Your email doesn't have quite enough information. Try opening the command window and starting Athena using the keyboard. If there are useful error messages sent to the screen, you can copy then into your email. There is the chance that that you are having the same problem as here: http://millenia.cars.aps.anl.gov/pipermail/ifeffit/2010-January/009243.html The problem might be that you are trying to save the file to a folder with accented characters. Try saving to a folder without any accented characters. B -- Bruce Ravel bra...@bnl.gov 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 ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Bruce ayuda, problemas con athena
Primo Latinam linguam aptiorem dicere disce. Tum te laete respondebo. --Scott On Feb 13, 2010, at 7:05 PM, Matt Newville wrote: On Sat, Feb 13, 2010 at 10:59 AM, Scott Calvin wrote: You'll probably get Russian before I get Latin. Quam utor postulo notitia quinymo quam Feff ut a vexillum in Artemis? But you're right: Russian did come before Latin ;) --Matt ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Anharmonic correction
Hi Aaron, I find a nearest-neighbor third cumulant is frequently a useful parameter for nanoscale materials. It's not just the anharmonicity of individual bonds, it's also the anharmonicity of the distribution of environments. In other words, in nanoscale materials there are core atoms and surface atoms, atoms one monolayer below the surface, and so on. (Or, in your case, read "interface" for "surface.") The atoms on the surface may very well have interatomic distances a bit different from those further in, and that distribution is often not symmetric, for essentially the same reason that thermal vibrations are not symmetric: it's energetically more favorable to stretch a bond from equilibrium than to compress it by the same amount. To use a third cumulant in Artemis, go to the Paths menu and check "extended path parameters." The path dialogs will then include a blank for "3rd," which is the third cumulant in the literature. It is then used like any other path parameter. (Or, of course, you can access it through IFEFFIT scripts, again using "3rd.") --Scott Calvin Sarah Lawrence College On Mar 26, 2010, at 1:13 PM, Aaron Slowey wrote: Dear XAFS community: I am fitting Hg L3-edge EXAFS of what I think are mercury sulfide nanoparticles. I fit Fourier filtered 1st shell Hg-to-Sulfur pair correlations for 5 spectra and obtain interatomic distances (r) that are 0.2 angstroms shorter than a cubic HgS(s) (i.e., metacinnabar) and Hg-to-S coordination numbers (N) that range from 2.6 to 3.0 (compared to N = 4 for metacinnabar). Delta_E0 values are less than a few eV, so I think the r's are not 'incorrect' as far as these preliminary fits are based on harmonic atomic vibrations/Gaussian pair-distribution functions. What intrigues me most about these data is that the fitted N's are consistent with the average 1st-shell Hg-S coordinations of 1 to 2 nm HgS clusters obtained by isotropically truncating the metacinnabar crystal lattice. In one case, I can also fit first- nearest Hg neighbors in the first shell, and this N is also consistent with a 1 nm HgS cluster. My objective is to scrutinize the tentative conclusion that the mercury sulfides in the samples consist of 1 to 2 nm subunits (within a larger aggregate, as determined by DLS). For instance, while the fitted N's are consistent with nanoclusters, the assumption of a metacinnabar lattice to estimate N of nanoclusters is undermined by the shorter interatomic distances fitted to the data. This got me reading the work of Manceau and Combes from the late 1980s and Frenkel et al. (2001) J. Phys. Chem. B. In Frenkel's paper (p. 12691), they describe that they used a "third cumulant" to account for anharmonic corrections, but I'm not sure how exactly this is implemented. It is something that you request in a feff.inp file, or is it a path parameter for IFEFFIT to include in its calculations? I am using Artemis on Mac OS X 10.6 (thanks to iXAFS 2.1.1 beta!!) to execute FEFF 7 calculations and fit my data. I noticed parameters called "3rd" and "4th" in the path dialog; is "3rd" the same parameter as, for example, the sigma_sub_i_superscript_(3) term in eqn (2) of Frenkel et al. (2001)? Thanks for reading this lengthy note... Aaron ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Anharmonic correction (Aaron Slowey)
I concur with Grant. But I would like to also make a complementary observation. For moderately disordered systems such as nanoscale samples, there's also no a priori reason to assume that the distribution is not skewed. In other words, the question of convergence isn't avoided by just ignoring higher cumulants! In general, a reasonably good process is: 1) Try the fit with the third cumulant forced to 0 (i.e. not fit). 2) Try the fit with the third cumulant guessed. 3a) If the third cumulant refines to 0 to within the reported uncertainty and the other parameters don't move outside their original error bars, then the third cumulant is not needed; go back to fit 1. 3b) If the third cumulant refines to a nonzero value to within the reported uncertainty, then look at the value. If it violates the limit Grant gives, then it's not appropriate, but you need to find another way of dealing with the fit. (Sometimes you may have a splitting between two different path lengths that you haven't modeled, for instance.) If it is within Grant's limit, then proceed with the usual caution you accord to EXAFS fits. For example, evaluate the physical sensibility of parameters, the stability of the fit to small changes in the data ranges, etc.. --Scott Calvin Sarah Lawrence College On Mar 26, 2010, at 4:09 PM, grant bunker wrote: Aaron - There are a couple of things you should watch out for when fitting cumulants. First, you should make sure in the fitting process that the third cumulant C3 doesn't get much more than twice C2^(3/2) (i.e. 2 sigma^3) - values much larger than that are probably unphysical, even if they happen to give you a better fit. Second, the cumulant expansion loses its utility if it doesn't converge quickly enough. It's essentially an expansion in terms of order k*sigma, and if that approaches 1 the higher order cumulants may be large enough that convergence is questionable. If you are lucky and the effective distribution is Gaussian, or most of the variance is due to Gaussian broadening of a skewed distribution, it may converge OK, but that shouldn't be assumed a priori. Grant Bunker http://gbxafs.iit.edu ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Measuring particle size of metal oxide
Hi Bill, Sorry for my slow response; I've been swamped. And ironically, the fact that I'm very interested in your topic and wanted to give a good reply slowed me from getting to it! I'm also going to repost your question and my response to the IFEFFIT mailing list, as I think it's of general interest. I think you're missing a key idea concerning this procedure. Both my method and Anatoly Frenkel's method for finding particle size rely on comparing the effective coordination for different paths. I am skeptical of EXAFS determinations of particle size using only one path, as coordination number is hard to tease out from other effects that suppress amplitude (sample quality, disorder, vacancies...). But the relative coordination number of different shells is more reliable. So you should be using more than one path, each with its own r (the absorber-scatterer distance), in order to refine a single value for R, the crystallite radius. If you're using Artemis, this is simple enough to do, as each path has a reff value (which is r in the formula you give), and you can define a guessed parameter R. (To do this right, you need to make sure that the r for multiple-scattering paths is the distance from the absorber to the furthest scatterer, not the half-path length. That means putting the value in "by hand" for those paths, rather than using reff.) (If you want a really quick and dirty method, collect a reference for a bulk standard and for your sample, and multiply the FT of the bulk spectrum by the formula, adjusting R until you get a good match. The quick and dirty method doesn't handle multiple-scattering correctly, and has the usual problem of the fact that even for direct-scattering paths the peaks of the FT are shifted from actual absorber-scatterer distances. But it is model-free, which can be nice in some cases.) By the way, the formula you've cited is derived for spherical particles. If they're kinda sorta spherical, it will still give a decent approximation and fit. But if the particles are needle-shaped or flat plates, then it doesn't work well. You either have to derive another formula, or look at Anatoly's papers, which have addressed a number of common morphologies. So far, everything I've said applies to metals, and you asked about oxides. While most of this also applies to oxides, it's important to realize that the nearest-neighbor oxide paths are not suppressed at all; i.e. the formula doesn't apply until the first metal-metal path (but does apply to metal-oxide paths further out). This is because the surface of such particles is generally comprised of oxygen atoms, and the first scattering shell is thus fully populated no matter how small the particles are. --Scott Calvin Sarah Lawrence College On Mar 8, 2010, at 1:45 PM, bill.schwa...@yale.edu wrote: Hi Scott, We met a couple of times at the EXAFS last two EXAFS workshops at BNL, for relative beginners like me. I am attempting to determine the particle size of PdO particles supported on Alumina (3% PdO/Al2O3). I am wondering if I can collect EXAFS data and use the formula in your 2003 Journal of Applied Physics paper(*) to estimate PdO particle size: N_nano = [1-3/4(r/R)+1/16(r/R)^3]N_bulk Here is a little more background: PdO particles on metal oxide supports are generally smaller and more dispersed in comparison to Pd metal on the same support. Also, if PdO is reduced and then re-oxidized at varying temperatures, the re- oxidized PdO particle size varies with temperature, with higher oxidation temperatures resulting in smaller PdO particle size. I have prepared a series of samples of 3% PdO/Al2O3, where the PdO has been been reduced and then re-oxideized at various temperatures, and I would like to use EXAFS to determine the coordination number of my various samples. A major difference in my planned experiment compared to what you described in your paper is that you examined nickel metal, while I am looking at a metal oxide. So my current questions are: 1) Is the above formula reasonable for determining bulk particle radius (R) of metal oxides? 2) If yes, then what is r, since scattering distance between Pd - Pd and Pd - O are different? Is it reasonable to average the distances? 3) Can N(nano) be determined by averaging the coordination numbers for the Pd-Pd and Pd-O paths? Any guidance you can provide will be greatly appreciated. Sincerely, Bill (*) Determination of crystallite size in a magnetic nanocomposite using extended x-ray absorption fine structure ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] can sigma square ever be less than zero?
Hi Chris, Not to be picky, but I think we have to consider the semantics of what you're asking very carefully. "Can ss be negative," as in, can the physical quantity which is the variance in the absorber-scatterer distance be negative? No, since variances are the square of a real number. "Can ss be negative," as in, can ifeffit output a negative best-fit value for ss? Yes, as you've seen. "Can ss be negative," as in, can a fit with a negative best-fit value for ss be considered a valid fit? That's really what you're asking, I think, and the answer is that it could be, depending on what you are trying to claim. Since the uncertainty in your case is quite large, it's certainly possible that your fit is consistent with believable values. But it also means that your fit gives you very little idea of what ss should actually be. The twin facts that the uncertainty is large and that the best-fit value is very close to 0 make the fit less convincing. The simplest explanation in your case is that Ifeffit is finding a fit with an S02 and a ss that are both a bit low. Since they tend to correlate highly, that's not uncommon. Have you tried fitting using different k-weights, or, better yet, several k-weights simultaneously? At any rate, I'd say your fit is a promising preliminary fit. As far as a publication-quality fit, it would be nice to get the nearest- neighbor ss pinned down a bit better. --Scott Calvin Sarah Lawrence College On Jun 28, 2010, at 4:10 PM, Chris Patridge wrote: Hello all, I am working on W L3 edge data. W is acting as a substitution dopant in vanadium dioxide at rather low concentration. In a past mailing conversation discussing Feff6 overestimation of E0 for heavier elements it was mentioned that the E0 could be past the rising edge due to the white line from W data. Well using this comment I aligned data using the theory method well explained by Shelly Kelly SnO2 example. Literature suggests W approximates WO2 cubic structure locally instead of the VO2 unit structure. Then fitting the first oxygen coordination shell paths which are well isolated from the other paths, it gives reasonable values for amplitude and enot of 0.77 (0.17) and -1.14 (3.06) respectively. delr is -0.067 (0.028) and then ss comes out to -0.00036 (0.00414). Can ss be negative if the uncertainty brings it above 0? Thank you all, Chris Patridge PhD Candidate Department of Chemistry NSC 405 SUNY Buffalo 315-529-0501 ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Vanadium center fitting issues
Hi Chris, Yes, you do need to account for the inequivalent vanadiums. Each feff calculation needs to be weighted by the fraction of vanadiums of that type (if you're looking at a tabulation of the asymmetric unit in a crystal structure, the number in the position codes like "8f" next to the atoms tell you the relative numbers of each). Use the S02 field to apply this weighting. In one sense, that doubles the number of parameters. But the V-V paths are present in both calculations, so that cuts it down a little. From there, you have the usual options of trying to apply constraints that aren't physically unreasonable, and seeing how the fit responds to them. --Scott Calvin Sarah Lawrence College On Jul 10, 2010, at 11:03 AM, Christopher Allen wrote: Hi, I wanted to get some advice on a vanadium centered material I’ve been trying to fit for quite a while w/ repeated failure. Based on XRD and electrochemistry I’m pretty confident in the material/model, and I have the input file for the material. The first shell out to 2.2 angstroms in R space is a distorted VO6 w/ bond lengths ranging from 1.6 to 2.2 angstroms, similar to the V2O5 which has been discussed on the mailing list recently (July 7th). The last two peaks from 2.2 to 3.3 angstroms presumably include 4 V-P and 1 more V-O single scattering paths. (the best fit I could get on the 4V-P/ 1V-O used only the first of these two peaks.) With regards to the first shell fitting, is it appropriate to be grouping the V-O bonds in terms of 1 short, 4 medium, and 1 long bond distance to cut down on variables? I noticed in the V2O5 previously discussed, that he used one ss and delr term for all 6 V- O paths, but I guess that just means making some assumptions that any variations in ss and delr are isotropic throughout. Is that correct? I’ve had real issues w/ those peaks representing V-P/V-O from 2-3.3 angstroms so I wonder if it’s unreasonable to try and extract this information based on the k space or if possible thats not what I have there. Could it be that I need to account for the two inequivalent V centers? (In that case, wouldn’t the number of variables be doubled?) Thanks for any comments, Chris -- Chris Allen Northeastern University Center for Renewable Energy Technology 317 Egan Research Center 360 Huntington Ave. Boston, MA 02115 617-373-5630 < livopo4 .prj>___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Case of "nonequivalent multiple atomic sites of absorbing atoms"
Hi Rana, Currently, S02 is usually described as being due to the relaxation of the other electrons in an atom when a core electron is removed, resulting in incomplete overlap of initial and final states. This appears to be a fairly good description, as careful experiments show good agreement with theoretical calculations based on this idea. Note, however, that there could be some other contributors to S02. A photon could, in addition to exciting the core electron at the edge, also excite a valence electron. There's a small thread on the transferability of S02 here: http://www.mail-archive.com/ifeffit@millenia.cars.aps.anl.gov/msg01626.html E0 is a tricky concept, in my opinion: it is the energy origin in the EXAFS equation. Perhaps a theorist can give me a pithy physical interpretation of what happens at that energy, but I don't know there needs to be anything; at k near 0, the path expansion is not convergent, so I'm not sure we should expect anything special to happen exactly at 0. In other words, it's not exactly the Fermi level or any other special energy. E0 is dependent on oxidation state; it can shift by an electron volt or two when oxidation states vary. Note that oxidation state is a simplistic measure of what's happening with the electron distribution in a material. Suppose fluorine is substituted for iodine in some material. Formally, the oxidation state of the atom they are bonded to is not changed by the substitution. But in reality, the electron distribution is different, and a small E0 shift would not be surprising. I think the bottom line, then, is this: S02 is completely transferable for the same element at multiple absorbing sites. Delta E0 is transferable with some caution for the same element at multiple absorbing sites if the oxidation state is the same. One other note: there's no rule that when trying constraints, you have to start unconstrained and add constraints to see the effect on the fit. With complicated systems like yours, it often pays to start with unrealistically simple constrains (not only E0's and S02's the same, but also sigma2's), and see if you're on the right track. Then look at the effect of relaxing constraints. --Scott Calvin Sarah Lawrence College On Jul 28, 2010, at 4:00 AM, Jatinkumar Rana wrote: Dear Users, Since long, i was trying to understand the physical meaning of term "Delta E0" and "S02" in EXAFS equation. I have little bit of idea about both of them. for example, S02 is element specific and it is transferable between samples (if we consider same absorbing atom). However, I am not able to realize their importance in terms of their "physical meaning" as far as interaction of photoelectron is concerned. Therefore, it is difficult for me to understand their influence on EXAFS. I am dealing with a case of "nonequivalent multiple atomic sites of absorbing atoms". It is quite obvious that in such kind of case no. of variables are more than no. of independent points and there is a need to constrain the parameters to solve such problems. I have following questions : How do i understand "Delta E0" and "S02" theoretically (in terms of photoelectron interaction) ? Can i constrain "Delta E0" for all absorbing atomic site as same ? (my assumption : all absorbing atoms are at same oxidation level) your comments and suggestions would be highly appreciated... Best regards, Rana ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Large Amplitude Values
Hi Gavin, What are the uncertainties on the high S02 values? Fluorescence is unlikely to be the culprit. While it can affect your ability to normalize properly, you're unlikely to account for a factor of 2 by normalization if the data is relatively decent. And self- absorption tends to suppress S02, not exaggerate it. Why did you switch to fluorescence on just the handful of data sets? That might provide us a clue. --Scott Calvin Sarah Lawrence College On Jul 30, 2010, at 10:47 PM, Gavin Garside wrote: Fellow X-Ray Absorption Enthusiasts, I have recently compiled a model that gives excellent visual fits in R, q, and k space for bond spacing in a BCC structure. This model gives bond spacings that make sense, and are very close to what would be expected from this set. The R factors are very low, and the enot values correspond quite well to the edge. However, our amplitude values are much larger than typically expected. They come in at the range of 1.8 up to 5.0, but only on a few data sets. On all the rest the amplitude values are 0.4 to 1.0. Could this increase in amplitude be attributed to the fact that we ran florescence measurements instead of transmission, and have a weaker signal coming to the detector? What else could be causing this in only one data set? All samples used in this model have the same structure. Thanks in advance to any replies, your help and time is appreciated. ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Large Amplitude Values
Hi Gavin, The problem isn't the value of S02; it's the uncertainty. I've noticed that Ifeffit has a tendency to push up the best fit value of S02 when it's very uncertain, although I'd have to think more about how it determines error bars to confirm that. But it does make sense, in a "Price is Right" kind of way--negative S02's presumably fit horribly, because they turn chi(k) upside down, and so that biases uncertain S02's to move the whole range up. In any case, your focus should be on reducing that uncertainty. 1) One way to do that is to fit with multiple k-weights, assuming you're not doing that already. Check the kw 1, kw 2, and kw 3 boxes all together and run a fit. The reason this works is that the EXFAS equation shows no k-dependence for S02, but a k^2 dependence for sigma^2, which often shows a high correlation with it in fits. Fitting multiple k-weights sometimes helps break that correlation. 2) Along the same lines, if you can squeeze out any additional k-range that may help. 3) If you're fitting coordination numbers, then adding additional scattering shells with some physically defensible scheme for constraining coordination numbers to a small number of parameters can help a lot. 4) Another good technique is to fit multiple samples simultaneously, constraining S02 to be the same for all of them. Or fit the sample and a standard measured in a similar way simultaneously, again constraining S02 to be the same. 5) Along the same lines, you could fit a standard measured in a similar way to determine S02, and then constrain the fit of your sample to take on that value. 4 and 5 are similar, so you may wonder if I have a preference. I'd say that if the samples, beam, detectors, data, and data reduction are all well behaved, then #5 is probably best, and has the benefit of being a technique with a long pedigree. If you're a little suspicious of something in the chain, though (for example, it's difficult to tell if you've been consistent in normalizing your standard and sample, because one has a big white line and the other doesn't), then #4 has the benefit that it distributes the error in the parameters you are fitting between sample and standard. This is good both because your sample has less error than otherwise, and because the values for the standard act as a "canary in a coal mine," warning you by their deviation from known values as to the magnitude of the errors you're looking at. --Scott Calvin Sarah Lawrence College On Jul 31, 2010, at 11:56 AM, Gavin Garside wrote: Scott, Thank you for a quick response. The value I am getting for SO2 in the fit most of interest is 2.95 plus/minus 3.72. So with the error bar I am in range, but I was just suspicious of it before I make any claims about it. All my experiments were done in florescence because we have ordered bulk material. By creating a sample that would work in fluorescence I may have introduced dislocations or imperfections that would have effected the physical properties of interest in this sample. Gavin Garside University of Utah From: Scott Calvin To: XAFS Analysis using Ifeffit Sent: Sat, July 31, 2010 4:52:35 AM Subject: Re: [Ifeffit] Large Amplitude Values Hi Gavin, What are the uncertainties on the high S02 values? Fluorescence is unlikely to be the culprit. While it can affect your ability to normalize properly, you're unlikely to account for a factor of 2 by normalization if the data is relatively decent. And self-absorption tends to suppress S02, not exaggerate it. Why did you switch to fluorescence on just the handful of data sets? That might provide us a clue. --Scott Calvin Sarah Lawrence College On Jul 30, 2010, at 10:47 PM, Gavin Garside wrote: Fellow X-Ray Absorption Enthusiasts, I have recently compiled a model that gives excellent visual fits in R, q, and k space for bond spacing in a BCC structure. This model gives bond spacings that make sense, and are very close to what would be expected from this set. The R factors are very low, and the enot values correspond quite well to the edge. However, our amplitude values are much larger than typically expected. They come in at the range of 1.8 up to 5.0, but only on a few data sets. On all the rest the amplitude values are 0.4 to 1.0. Could this increase in amplitude be attributed to the fact that we ran florescence measurements instead of transmission, and have a weaker signal coming to the detector? What else could be causing this in only one data set? All samples used in this model have the same structure. Thanks in advance to any replies, your help and time is appreciated. ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mai
Re: [Ifeffit] Stoichiometry from EXAFS data
Hi Peter, I've done this as well, and compared to reliable methods (e.g. ICP). I'd be skeptical of 1%. It's generally quite difficult to determine edge steps to that accuracy. Assuming you're using Athena to determine the edge step, find the most extreme pre- and post-edge lines that seem acceptable and note the range of edge steps. That will yield an uncertainty range. If you have strong features at the white line and just past it, I'd be surprised if you can do much better than 10%. If features in that region are small, such as you might have in an intermetallic alloy, then you might get down to the sub-5% range. While I think that determining the edge step is likely the major source of error, you also have to be aware of the usual suspects in XANES analysis, such as the presence of harmonics in transmission or self-absorption in fluorescence. Testing for linearity with tricks like putting sheets of aluminum foil before I0 can help detect some (but not all) of those kinds of issues. --Scott Calvin Sarah Lawrence College On Aug 2, 2010, at 4:03 AM, Peter Zalden wrote: Dear Feff users, lately, we measured a sample containing Sb and Te at EXAFS beamline CEMO, Hasylab and are wondering whether one can determine the stoichiometry from the height of the different K edges´ steps, if one normalizes the values on the edge steps of the elements (cf. http://physics.nist.gov/PhysRefData/XrayMassCoef/tab3.html) . The absorption gases and the specimen were not changed for the different K edges. Of course, I have already tried doing so and from statistical reproducibility and from the resulting values compared to the expected ones I would estimate an error of this method of about 1%. A source of error that I could imagine originates from the different beam position at different energies combined with a slight inhomogenity in the pressed sample powder. Now my question is: Are there any other sources of error that I should take into accout? Is there any reference on this method from a more experienced user that I could cite? Kind regards, Peter ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Stoichiometry from EXAFS data
Hi again, Your tellurium edge in particular has the kind of nice regular structure around the edge that makes edge step determination relatively accurate. Also, you're working at such high energies that harmonics are unlikely to be much of an issue. I'd probably consider your data as good to +/- 3% for stoichiometry. Calling that "sigma" or "two sigma" kind of implies that the error in treating many similar systems in this way would be normally distributed, and that's very unlikely to be true. Unlike with population statistics or counting statistics, you wouldn't occasionally end up way off "by chance." It's almost more like a report of precision: when carefully measuring a length with a ruler marked in mm, it's reasonable to interpolate between the mm marks and report a measurement as good to, perhaps +/- 0.3 mm. If, by eye, I claim 11.3 mm, it might conceivably be 11.0 or 11.6 mm, in part because of the ability to eyeball it, and in part because of problems with lining up marks using rulers. But unlike with Gaussian statistics, where two- or three-sigma events happen now and then, I'd often be off by 0.2 mm yet never off by 0.6 mm. --Scott Calvin Sarah Lawrence College On Aug 2, 2010, at 8:51 AM, Peter Zalden wrote: Hi Scott, thanks a lot for your quick response! I found your suggestion very helpful and tried to change the edge step to both extremes by tuning the fitting range for the pre- and post-edge lines in Athena. Due to the very flat structure in the XANES range (cf. attachement), I could modify the value for the edge step by 3% total, which corresponds to an error of +/-2%. One could possibly discuss if this value represents the one-sigma or maybe the two-sigma interval, but the error is nicely small anyway. In the last campaign, we measured a sample of Sb_2Te_1 in two different annealing conditions and from those different data sets (as concerns the EXAFS range), I determined the stoichiometries: Sb_2.06Te_0.94 and exactly the same for the second sample. Therefore, a sub-5% error seems reasonable to assume for these semi- metallic systems. Concerning the influence of higher harmonics: The beam was usually detuned to 70% intensity of the main reflection so that this should not have a strong influence, since the amount of detuning was not changed for both edges. Best regards, Peter ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] two measurements of the same compound in different beam-lines
Hi Maria, You have several choices. First, note that the merge function in Athena allows you to select options of weight by chi-noise, or weight by importance. If you weight by chi-noise, noisier data will be counted less. If you have some other way of estimating the quality of the data sets, you can enter different numbers in the "importance" field and then weight by importance. If working on two different beamlines, you might consider merging chi(k) data, rather than norm(E), as the backgrounds may be different. And mu(E) probably makes no sense at all. Another option is to obtain separate chi(k)'s, but then fit both spectra simultaneously in Artemis. The Artemis/Ifeffit default behavior in that case will be to use high-R noise for weighting, but you can override that by assigning an epsilon to each data set if you choose. This method has several advantages: it lets you see if one data set is fitting differently from the other; it lets you choose different k-ranges if noise begins dominating one data set at a lower value of k; it lets you use different values of S02 if there are pinhole, harmonic, or self-absorption effects; and it lets you use different values of delE0 if the data sets are hard to align properly. --Scott Calvin Sarah Lawrence College On Aug 12, 2010, at 8:44 PM, María Elena Montero Cabrera wrote: Hi all! I have performed two independent XAFS measurements of Cr K-edge of the same Fe-Cr sample at two different beam-lines at SSRL. I have obtained the Fe-K edge data only once. The quality of data are different in each measurement. However, I cannot average spectra from different Cr-K measurements, and I don't know if I could take somehow advantage from having almost twice the information for the Cr- K adge, or I have to use only the better quality data and discard the other. What do you advice? If I can use both measurements, how can I do the fitting in Artemis? Thank you very much and all take care -- María Elena Dra. María Elena Montero Cabrera Departamento de Medio Ambiente y Energía Centro de Investigación en Materiales Avanzados (CIMAV) Miguel de Cervantes 120, Compl. Ind. Chihuahua Chihuahua CP 31109, Chih. México Tel (614) 4391123 ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
[Ifeffit] Haha
Hi all, For a little comic relief, I just came across this graph: http://www.usablemarkets.com/wp-content/uploads/2010/06/fed-rate-3.jpg I have never seen a graph unrelated to XAFS looks more like (noisy) XAFS data... --Scott Calvin Sarah Lawrence College ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Ifeffit Digest, Vol 90, Issue 21
Hi Dominik, I haven't had time to look at your question in detail, but try the link below, and see if you find it helpful: http://cars9.uchicago.edu/ifeffit/Doped --Scott Calvin Sarah Lawrence College On Aug 30, 2010, at 7:43 AM, Jatinkumar Rana wrote: Dear Dominik, Thank you so much for your reply. However, i am not able to understand the logic behind removal of following atoms from Feff.inp file as mentioned by you **In principle, after running ATOMS, you just need to remove atoms from the generated feff.inp file so that the occupancy is correct. I.e., in this case, you need to remove roughly 1/4 of the Fe7, 1/2 of the Cr1, 1/3 of the P61, 2/3 of the P62, and 1/3 of the O614 atoms. Try finding those which are too close to other atoms. The FEFF output might help, so try deleting preferentially those causing FEFF to fail.** can you please give me the idea behind doing that and is it physically defendable to do such changes ?? Looking forward to your reply.. Best regards, Jatin On 27.08.2010 14:00, ifeffit-requ...@millenia.cars.aps.anl.gov wrote: Send Ifeffit mailing list submissions to ifeffit@millenia.cars.aps.anl.gov To subscribe or unsubscribe via the World Wide Web, visit http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit or, via email, send a message with subject or body 'help' to ifeffit-requ...@millenia.cars.aps.anl.gov You can reach the person managing the list at ifeffit-ow...@millenia.cars.aps.anl.gov When replying, please edit your Subject line so it is more specific than "Re: Contents of Ifeffit digest..." Today's Topics: 1. Re: Help with cif file (Dominik Samuelis) 2. catalysis workshop at the karlsruher synchrotron ANKA (Matthias Bauer) -- Message: 1 Date: Fri, 27 Aug 2010 10:09:05 +0200 From: Dominik Samuelis To: XAFS Analysis using Ifeffit Subject: Re: [Ifeffit] Help with cif file Message-ID:<4c7772a1.2000...@fkf.mpg.de> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Dear Jatin, in your original cif file, you have site occupancies as low as 0.5 (Cr1 site) and 0.358 (P62). Just setting them to 1 will not help, because ATOMS assumes them to be unity anyway. For such a complicated unit cell, the typical recipe with using prototypical structures of course does not help, just because there is not a simple prototype structure for arrojadite. In principle, after running ATOMS, you just need to remove atoms from the generated feff.inp file so that the occupancy is correct. I.e., in this case, you need to remove roughly 1/4 of the Fe7, 1/2 of the Cr1, 1/3 of the P61, 2/3 of the P62, and 1/3 of the O614 atoms. Try finding those which are too close to other atoms. The FEFF output might help, so try deleting preferentially those causing FEFF to fail. Another solution might be loading the structure's cif file into a structure editor such as DIAMOND. There, you can then check the bond distance histograms and delete atoms accordingly. At the end, just export the data in xyz format and use this as the atom positions list in the feff.inp file. Regards, Dominik On 20.08.2010 09:30, Jatinkumar Rana wrote: Dear Dominic, Dear Bruce, I am also facing the same problem as experienced by Kleper. I am working on the EXAFS analysis of Arrojadite mineral. We have refined the strucutre using neutron diffraction to get crystallographic information which can be fed to ATOMS. The original .cif file contains the fractional occupancy so ATOMS give similar error report as mentioned by Kleper. After reading your post, i changed all site occupancy to 1 and then run ATOMS but still it gives me the error report. Can anybody tell me, why ATOMS report error ? I have attached both original .cif file (with fractional occupancy) and modified .cif file (all site occupancy = 1). I have EXAFS spectrum at Fe and Mn K-edge. Looking forward to your answer. Thanks a lot in advance... Best regards, Jatin On 18.08.2010 19:00, ifeffit-requ...@millenia.cars.aps.anl.gov wrote: Send Ifeffit mailing list submissions to ifeffit@millenia.cars.aps.anl.gov To subscribe or unsubscribe via the World Wide Web, visit http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit or, via email, send a message with subject or body 'help' to ifeffit-requ...@millenia.cars.aps.anl.gov You can reach the person managing the list at ifeffit-ow...@millenia.cars.aps.anl.gov When replying, please edit your Subject line so it is more specific than "Re: Contents of Ifeffit digest..." Today's Topics: 1. Re: Help with cif file (Dominik Samuelis) 2. Re: Help with cif file (Bruce Ravel) -- Message: 1 Date: Wed, 18 Aug 2010 08:36:03 +0200 From: Dominik Samuelis To: XAFS Analysis using Ifeffit Subject: Re
[Ifeffit] More than 256 paths on Mac OS 10.5?
Hi all, Do any of you have a version of Ifeffit compiled for Mac OS 10.5 that allows more than 256 paths? --Scott Calvin Sarah Lawrence College ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] sigma^2 values for multiple scattering paths
Although I agree with the main points that Bruce makes, I do want to comment on one piece: On Oct 6, 2010, at 7:03 AM, Bruce Ravel wrote: . In no case can I understand a physical explanation for the the MS sigma^2 being smaller than for the SS. Actually, there is a physical situation where something like that can occur, although it sounds like it's not the one that Han Sen has. Consider an absorbing atom rattling around in a relatively fixed cage or lattice. And then consider a linear (or near-linear) arrangement: S1 -- A -- S2 One multiple scattering path that can sometimes have a sizable contribution is A --> S1 --> S2 --> A. This path will have a sigma^2 that is a bit larger than the single-scattering path S1 --> S2 --> S1, because of the perpendicular component of the motion of A. But it's quite frequently the case that S1 --> S2 --> S1 is not modeled in a fit, because the S edge is not measured. On the other hand, the single scattering paths A --> S1 --> A and A -- > S2 --> A ARE included in the fit. Those two have high sigma^2's, because A is rattling around a lot. Under that circumstance, a multiple-scattering path included in the fit may indeed have a lower sigma^2 than the single-scattering paths included in the fit. The moral, of course, is that it's not hard to think physically about what sigma^2 means for a multiple scattering path. If one appears to have an "unphysically" small sigma2, then the explanation is probably one of the ones given by Bruce or Shelly. One more thought on this. How much does it change your fit, Han Sen, if you set the sigma^2 for the multiple-scattering path to some "reasonable" value. If the scientific information you want from your fit is not sensitive to exactly what sigma^2 the MS path gets, and is not significantly different when given a "reasonable" value than when allowed to find its "best-fit" value, then there's probably no need to resolve the issue. In my experience, this is often the case with low- amplitude MS paths: the fit is improved by their inclusion, but may not be particularly sensitive to the details of their path parameters. --Scott Calvin Sarah Lawrence College ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] More than 256 paths on Mac OS 10.5?
Thanks, Matt! --Scott Calvin Sarah Lawrence College On Oct 6, 2010, at 9:08 AM, Matt Newville wrote: Hi Scott, The attached zip file has dynamic libraries (and static program ifeffit) built with 1024 paths and feff files. It contains the files lib/libifeffit.dylib lib/libifeffit.so bin/ifeffit The zip file should be unzipped under /Applications/iXAFS.app/Contents/Resources/local/ to overwrite the above files. You should be able to open the iXAFS Shell and type cd /Applications/iXAFS.app/Contents/Resources/local/ unzip ~/Downloads/iXAFS_1024paths.zip athena Athena and Artemis will automatically use the new dynamic library. ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] sigma^2 values for multiple scattering paths (Scott Calvin); Re: Ifeffit Digest, Vol 92, Issue 4
On Oct 6, 2010, at 10:41 AM, Han Sen Soo wrote: Hello Scott, Just to make sure I understand what you mean, are you saying that in your 3 atom system, the S1 and S2 atoms have relatively fixed locations but A may have large vibrational amplitudes in the A-S1 and A-S2 directions? So the round-trip 3 atom MS path has a small sigma^2 value since the variation in the A-S1-S2-A path is dictated by the more or less fixed S1 and S2 end-points (with minimal perpendicular contribution), whereas the 2 individual SS paths have large sigma^2 value due to the large A-S vibrations? Yes--you explained it far better than I did. :) --Scott Calvin Sarah Lawrence College ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] schemes for delr and sigma2 for multiple scattering paths
Jatin, IF the uniform expansion model is valid for single scattering paths, then it is for multiple scattering paths as well. For some materials, particularly those with cubic space groups, that's got a good chance of being a useful model. Others tends to distort with changes in temperature, doping, etc., and it may not work as well. But even in those cases, if you've decided how to constrain the delr for single- scattering paths, you'll do reasonably well by using some kind of appropriate average of the delr's for related single-scattering paths. --Scott Calvin Sarah Lawrence College On Oct 7, 2010, at 3:05 AM, Jatinkumar Rana wrote: Dear all, It is reasonable to assign a constant fraction by which unit cell expands at a given temperature of XAFS measuremnt and so the variation in the path lengths for every single scattering paths could be assigned as delr = alpha * Reff. Similarly, one can assign sigma2 value for each single scattering path depending on both type of scatterer and its distance from the absorbing atom. Now coming to multiple scattering paths, Sigma2 for multiple scattering paths can be constrained based on the sigma2 of related single scattering paths and a definite path-geometry-dependent scheme (Triangle, collinear, reversed etc.) could be applied. Is there any such scheme for delr of multiple scattering paths ? or we can simply assume that all paths (single scattering and multiple scattering) undergo uniform expansion by a factor alpha. Thank you so much in advance for your valuable time... With best regards, Jatin Rana ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] sigma^2 values for multiple scattering paths (Scott Calvin/Shelly Kelly/Abhijeet Gaur); Re: Ifeffit Digest, Vol 92, Issue 5
Not at all unusual, Han Sen. If you think about the EXAFS equation, you'll see that sigma^2 and amplitude primarily affect the amplitude of the signal, while distances affect the position of the peak in the Fourier transform (or equivalently, the spacing of peaks in chi(k)). So sigma^2 and amplitude can trade off without affecting distance- based aspects of the fit much. That's why I suggested you try forcing the sigma^2 to a "reasonable" value to see what happened to your fit. Sometimes none of the aspects of the fit you're interested in depend strongly on the sigma^2 of low- amplitude paths--particularly if what you're interest in is distances or information that is in part derived from distances, like phase identification. In those cases, the anomalous sigma^2 can be a "yellow flag" (think about what might be causing it and decide if it's a problem to your scientific case) rather than a "red flag" (drop everything and resolve the problem before proceeding). Also, note from the EXAFS equation that sigma^2 is weighted by k^2, and amplitude is not. If fits using different k-weights result in significantly different values of sigma^2, that can be a clue that the issue is actually one of amplitude, as in your case. At any rate, I'm glad you solved your issue in such a satisfying way! --Scott Calvin Sarah Lawrence College On Oct 7, 2010, at 11:22 AM, Han Sen Soo wrote: Hello Shelly and Scott, Thank you both again for your suggestions. It seems that after making the MS path more linear in my cif file, the FEFF calculation increased the amplitude value of the path and dramatically increased the sigma^2 value in the fit. Strangely, the fit values for the distances remain pretty much the same and the statistical figures of merit have improved, but the sigma^2 values are now much more reasonable (about twice as large, but I have a more triangular than linear model, so you're right Scott, your explanation does not work for my case). I guess the increased amplitude made a difference? Hello Abhijeet, I used a rudimentary geometrical way to get my bond angles. For a 3 atom triangle M-O-A, the effective MS path length (R_MOA) is twice the sum of the individual bond distances. So if you have the R_MOA, R_MO, and R_MA distances from your fits, you can use R_MOA - R_MO - R_MA to get the O-A bond length. And with the 3 sides of the triangle, you can use the geometrical Cosine Rule to get any of the 3 bond angles. This is just geometry so I don't know what the error propagation for this would be. Thanks again everyone! han sen ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] calibration/alignment
Hi Ornel, Alignment is used to compensate for monochromators that do not maintain stable energy calibration between scans. In conventional measurements, you'll generally have several scans that are supposed to be of the same sample under identical conditions, and those scans may need to be aligned with each other. That is not the case for a time series, which is what I think you are saying you have. So in a time series, how do you compensate for any energy drift of the monochromator? If you are recording a simultaneous reference spectrum, you can align the reference spectra to each other. (Athena automatically will shift the sample spectra by the same amount that the reference spectra are shifted.) If you have a time series but don't have a simultaneous reference spectra, it becomes tougher. If you collected a reference spectrum before and after the time series, you could try to interpolate any shift that's seen, although that's dicey; shifts sometimes occur in jumps. But if there's no shift, you're probably OK! If you have a time series and no reference at all, or a reference only before the series, you're out of luck. You're relying then on the assumption of energy stability, which on some beamlines might be OK...but it is best to confirm that by at least measuring a reference before and after. --Scott Calvin Sarah Lawrence College On Oct 14, 2010, at 2:46 AM, ornella smila castro wrote: Hi everyone, I am trying to do some data processing with Athena but I am already stuck at the first step. The thing is: I read the worked example section of the "Athena's user guide" and on the example on the iron foil, it is mentioned to calibrate the data at the right energy (until here evrything is fine) but then it is said to align the data. Can anyone explain to me what does "alignment" exactly means, and what is the aim of "aligning the data". The data that I have collected were through a channel through which a solution flow (~200 microliters/hr) so I am not convinced that alignment makes sense. Many thanks, Ornel ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] calibration/alignment
On Oct 14, 2010, at 7:39 AM, Matt Newville wrote: If you're just getting started, I would say to not worry about energy alignment until it becomes an obvious problem. A cautionary tale (with details made up, since I don't remember them!) from when I was just starting out as to what constitutes an "obvious problem": I collected five transmission EXAFS scans on the same sample. The scans were on top of each other when I looked at the graph, so I merged them...and proceeded to get somewhat screwy fits. The problem? I only looked at the graph across the whole spectra--say, 1500 eV. It turns out there was about a 0.7 eV shift between each scan and the next one, for a total of roughly 3 eV . That was small enough so as to be invisible when looked at on that scale. When I looked at just the XANES, though, the shift did become "obvious." I aligned the spectra and merged them, and suddenly the problems in the fit went away! Since then, I've seen the same thing happen with students to whom I am teaching the technique. On the other hand, there's no magic "blessing" given by the process of alignment. Suppose I have ten scans of very noisy data, and no reference. If I used the auto-align procedure in Athena, it sometimes shifts a scan 0.3 eV one way, sometimes 0.2 eV the other way, with no apparent rhyme or reason. Looking at the graphs, even zoomed in, just shows a bunch of noisy data roughly on top of each other. In that case, there's no reason to believe there are actual shifts between scans, and I would NOT align them prior to merging. Finally, beamline scientists usually have a very good idea whether their line is prone to drifts. Ask them! --Scott Calvin Sarah Lawrence College ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] calibration/aligment...2
Thanks, Ornella, that clarifies what you're doing. My recommendation is to look closely at the 6 reference spectra and see if there appears to be a systematic energy shift between them. For example, each spectrum might be shifted by about 0.3 eV from the spectrum before it. Or the first two spectra might appear aligned, but then the third through sixth are shifted by 1.5 eV. In either of those cases, you should align them. In my second example, I might throw out the second spectrum as an additional precaution (if the shift occurred "all at once," it might have occurred during the scan before which it appears). In either case, it doesn't really matter which scan you choose to align to (and calibrate, if you have a way of doing that). If, on the other hand, the 6 reference spectra appear to basically overplot except for random noise, I would not try to align them further. I would treat the 6 spectra for the electrolysed solution similarly-- align them to each other if there is a systematic energy shift. What you should not do, in my opinion, is to align the electrolysed scans to the reference scans. You actually expect there to be a chemical shift between the two sets of data, and aligning one to the other would remove that! --Scott Calvin Sarah Lawrence College On Oct 14, 2010, at 11:00 AM, ornella smila castro wrote: Hi Matt and Scott, First, thank you so much for replying to my questions. I realised that I should have been a bit more accurate on the type of experiments I am doing. To start with, I am doing electrochemistry combined with EXAFS. I am using an integrated electrolysis/EXAFS cell. Our experiments are as follow: in the case of this experiment in particular, we have a solution of ruthenium-based compound that we flow through a channel through which the beam passes and that we call "reference". we record let's say 6 spectra in a row (we use a flow in order to avoid beam damage on our sample).we didn't do any reference spectra (if you mean running a scan of a Ru foil before starting the actual experiment). then, we make up a new solution but this time we electrolysed the solution (by applying a potential) in order to get the cationic produced species. (the obtained spectra will therefore be a mixture of the neutral and the cationic species). and we record as well a set of 6 spectra. Here is what I did and what I was wondering if it was the right thing to do: I have considered the first spectra of the first set of scans as the reference for calibration/alignment. is it ok? Thus, I am wondering if when I process the second set of data (electrolysed solution), which spectra I should take as a starting point for calibrating/aligning my second series of scans: the one I used for the first serie or the first of the second set of scans? I hope this is clear... Thanks again for your help, Ornella ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
[Ifeffit] Asymmetric error bars in IFeffit
Hi all, I'm puzzling over an issue with my latest analysis, and it seemed like the sort of thing where this mailing list might have some good ideas. First, a little background on the analysis. It is a simultaneous fit to four samples, made of various combinations of three phases. Mossbauer has established which samples include which phases. One of the phases itself has two crystallographically inequivalent absorbing sites. The result is that the fit includes 12 Feff calculations, four data sets, and 1000 paths. Remarkably, everything works quite well, yielding a satisfying and informative fit. Depending on the details, the fit takes about 90 minutes to run. Kudos to Ifeffit and Horae for making such a thing possible! Several of the parameters that the fit finds are "characteristic crystallite radii" for the individual phases. In my published fits, I often include a factor that accounts for the fact that a phase is nanoscale in a crude way: it assumes the phase is present as spheres of uniform radius and applies a suppression factor to the coordination numbers of the paths as a function of that radius and of the absorber- scatterer distance. Even though this model is rarely strictly correct in terms of morphology and size dispersion, it gives a first-order approximation to the effect of the reduced coordination numbers found in nanoscale materials. Some people, notably Anatoly Frenkel, have published models which deal with this effect much more realistically. But those techniques also require more fitted variables and work best with fairly well-behaved samples. I tend to work with "messy" chemical samples of free nanoparticles where the assumption of sphericity isn't terrible, and the size dispersion is difficult to model accurately. At any rate, the project I'm currently working on includes a fitted characteristic radius of the type I've described for each of the phases in each of the samples. And again, it seems to work pretty well, yielding values that are plausible and largely stable. That's the background information. Now for my question: The effect of the characteristic radius on the spectrum is a strongly nonlinear function of that radius. For example, the difference between the EXAFS spectra of 100 nm and 1000 nm single crystals due to the coordination number effect is completely negligible. The difference between 1 nm and 10 nm crystals, however, is huge. So for very small crystallites, IFeffit reports perfectly reasonable error bars: the radius is 0.7 +/- 0.3 nm, for instance. For somewhat larger crystallites, however, it tends to report values like 10 +/- 500 nm. I understand why it does that: it's evaluating how much the parameter would have to change by to have a given impact on the chi square of the fit. And it turns out that once you get to about 10 nm, the size could go arbitrarily higher than that and not change the spectrum much at all. But it couldn't go that much lower without affecting the spectrum. So what IFeffit means is something like "the best fit value is 10 nm, and it is probable that the value is at least 4 nm." But it's operating under the assumption that the dependence of chi-square on the parameter is parabolic, so it comes up with a compromise between a 6 nm error bar on the low side and an infinitely large error bar on the high side. Compromising with infinity, however, rarely yields sensible results. Thus my question is if anyone can think of a way to extract some sense of these asymmetric error bars from IFeffit. Here are possibilities I've considered: --Fit something like the log of the characteristic radius, rather than the radius itself. That creates an asymmetric error bar for the radius, but the asymmetry the new error bar possesses has no relationship to the uncertainty it "should" possess. This seems to me like it's just a way of sweeping the problem under the rug and is potentially misleading. --Rerun the fits setting the variable in question to different values to probe how far up or down it can go and have the same effect on the fit. But since I've got nine of these factors, and each fit takes more than an hour, the computer time required seems prohibitive! --Somehow parameterize the guessed variable so that it does tend to have symmetric error bars, and then calculate the characteristic radius and its error bars from that. But it's not at all clear what that parameterization would be. --Ask the IFeffit mailing list for ideas! Thanks! --Scott Calvin Sarah Lawrence College___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Asymmetric error bars in IFeffit
I don't think it's at all strange, Anatoly, and I think Matthew's solution is the right one--it seems obvious in retrospect that the parameter that Ifeffit should evaluate is 1/R, but apparently it wasn't obvious to me on Friday. :) As for obtaining N instead of R, the beauty of both of our algorithms is that they don't depend on finding N; they depend on finding the ratio of N's for different shells. Finding N accurately is notoriously challenging: you need some way of getting S02, you need to have the normalization right, and you're sunk if there are data quality issues like an inhomogeneous sample, uncorrected self-absorption, or significant beam harmonics. But finding the ratio of N for two or more different shells doesn't depend so strongly on any of those things. Since my method implicitly involves multiple ratios of coordination numbers, it is not so clear how to invert it. In any case, I expect Matthew's solution to work, and will pursue it further on Monday. --Scott Calvin Sarah Lawrence College On Oct 24, 2010, at 5:59 PM, Frenkel, Anatoly wrote: Scott, It is a strange result. Suppose you fit a bulk metal foil and vary the 1nn coordination number. You will not get 12 +/- 1000. You will get about 12 +/- 0.3 depending on the data quality and the k range, and on the amplitude factor you fix constant. Then, suppose you take your formula for a particle radius from your JAP article and propagate this uncertainty to get the radius uncertainty. That would give you a huge error because you are in the flat region of the N(R) function and R does bit affect N. The meaning of your large error bar is, I think, that you are in such a large limit of sizes that they cannot be inverted to get N and thus the errors cannot be propagated to find Delta R. Why don't you try to obtain N instead of R? You will get much smaller error bars and you can find the lower R limit from your N(R) equation (by plugging in N - deltaN you will find R - delta R). The right limit is infinity as you pointed out. Anatoly From: ifeffit-boun...@millenia.cars.aps.anl.gov > To: XAFS Analysis using Ifeffit Sent: Fri Oct 22 16:23:08 2010 Subject: [Ifeffit] Asymmetric error bars in IFeffit Hi all, I'm puzzling over an issue with my latest analysis, and it seemed like the sort of thing where this mailing list might have some good ideas. First, a little background on the analysis. It is a simultaneous fit to four samples, made of various combinations of three phases. Mossbauer has established which samples include which phases. One of the phases itself has two crystallographically inequivalent absorbing sites. The result is that the fit includes 12 Feff calculations, four data sets, and 1000 paths. Remarkably, everything works quite well, yielding a satisfying and informative fit. Depending on the details, the fit takes about 90 minutes to run. Kudos to Ifeffit and Horae for making such a thing possible! Several of the parameters that the fit finds are "characteristic crystallite radii" for the individual phases. In my published fits, I often include a factor that accounts for the fact that a phase is nanoscale in a crude way: it assumes the phase is present as spheres of uniform radius and applies a suppression factor to the coordination numbers of the paths as a function of that radius and of the absorber-scatterer distance. Even though this model is rarely strictly correct in terms of morphology and size dispersion, it gives a first-order approximation to the effect of the reduced coordination numbers found in nanoscale materials. Some people, notably Anatoly Frenkel, have published models which deal with this effect much more realistically. But those techniques also require more fitted variables and work best with fairly well-behaved samples. I tend to work with "messy" chemical samples of free nanoparticles where the assumption of sphericity isn't terrible, and the size dispersion is difficult to model accurately. At any rate, the project I'm currently working on includes a fitted characteristic radius of the type I've described for each of the phases in each of the samples. And again, it seems to work pretty well, yielding values that are plausible and largely stable. That's the background information. Now for my question: The effect of the characteristic radius on the spectrum is a strongly nonlinear function of that radius. For example, the difference between the EXAFS spectra of 100 nm and 1000 nm single crystals due to the coordination number effect is completely negligible. The difference between 1 nm and 10 nm crystals, however, is huge. So for very small crystallites, IFeffit reports perfectly reasonable error bars: the radius is 0.7 +/- 0.3 nm, for instan
Re: [Ifeffit] LCF analysis
On Oct 25, 2010, at 8:25 AM, Wayne W Lukens Jr wrote: A more useful way to look at this is that the probabilities that A, B and C are present are 99.%, 93%, and 77%, respectively. An excellent post, Wayne, but I don't think that last statement is quite right. If the F-test gives a probability of 0.23 for material C, I believe it's saying that there is a 23% that, given the noise level in the data, the fit would indicate that C was present when it was not. That is not the same thing as saying there is a 77% chance of C being present. To see this, imagine very, very noisy data. Including C in the fit might very well improve the fit in the sense of an R-factor--maybe, in fact, there's a 45% chance of a modest improvement with a given set of very noisy data, even if there's no C present. That does not mean that a result like that should lead to the conclusion that C is more likely than not present (55%). --Scott Calvin Sarah Lawrence College___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Asymmetric error bars in IFeffit
Yes; it's a case of trying to distinguish between a few boulders and lots of pebbles; the total volume isn't the issue. What I'm looking at is something like surface/volume ratio, but with "surface" being path-dependent and gradual. For a nearest-neighbor path, only the top monolayer of atoms are on the surface. For a 5 angstrom path, the transition region from "surface" to "core" extends 5 angstroms in. But that more sophisticated definition of "surface" doesn't change the fact that the dominant dependence is 1/R, so that should address the issue. --Scott Calvin Sarah Lawrence College On Oct 25, 2010, at 4:43 AM, Matt Newville wrote: Hi Scott, That's a pretty amazing use case. But I'm not sure I understand the issue exactly right. I would have thought the volume (r**3) was the important physical parameter, and that a 1000nm particle would dominate the spectra over 3nm particles. Or is it that you are trying to distinguish between 1 very large crystal or 100s of smaller crystals? Perhaps the effect you're really trying to account for is the surface/volume ratio? If so, I think using Matthew Marcus's suggestion of using 1/r (with a safety margin) makes the most sense. --Matt ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] something is wrong with Ruthenium-Oxygen bond amplitudes
Hi Maria, What was the physical form of the samples (powder, thin film, etc.) and how were they measured (transmission, fluorescence, ...)? Sometimes this kind of thing can stem from sample/beamline/data effects. --Scott Calvin Sarah Lawrence College On Nov 5, 2010, at 5:50 PM, María Elena Montero Cabrera wrote: Hello friends, Hopping someone could help us. We are having some problems in fitting Ru K-edge in a Ruthenium-cuprate sample on Artemis, with path functions obtained using FEFF 8.4, where we got amplitude values of less than 0.50 for Ru-O first shells. We think this value probably is wrong, although there are some publications where some oxygen deficiency is studied and recorded as true. The ATOMS input data was found to be ok due to the Rietveld analysis results told us these are good. The reason we think we may be wrong with something during the fitting appeared because we tried to fit our reference sample of RuO2 (measured under same conditions as the experiment of Ru-Cu at SSRL) and we came to the same results, even worst, of amplitude lower than 0.40. We are attaching the Artemis files so you can take a look on it and give us some light to continue with our analysis. Thanking in advance, take care ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
[Ifeffit] Origin of terminology "self-absorption"
Hi all, As some of you know, I'm currently working on a textbook on XAFS analysis. Because of that, I'm going to occasionally pose some questions for the list that may seem a bit random. I hope none of you mind me using the list in this way; the questions may seem to come out of left field, but I think they will still be of interest to many. With that said, here's my question for today: What is the origin of the use of "self-absorption" to describe the suppression of fine-structure observed in thick, concentrated samples measured in fluorescence? I understand the physics of the effect itself, my question is the curious wording. Compared to a thin concentrated sample, the effect might better be described as "saturation," while compared to a thick dilute sample, it's actually related to a lack of absorption by other elements. --Scott Calvin Faculty at Sarah Lawrence College Currently on sabbatical at Stanford Synchrotron Radiation Laboratory ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Origin of terminology "self-absorption"
I tried a few searches, but rapidly get lost in other uses of the term. My guess is we borrowed it from some other spectroscopy, much the way we borrowed "Debye-Waller factor" from XRD, and then proceeded to change its meaning. But it would be nice to be able to track that down. --Scott Calvin Faculty at Sarah Lawrence College Currently on sabbatical at Stanford Synchrotron Radiation Laboratory On Nov 16, 2010, at 10:54 AM, Matthew Marcus wrote: It's definitely a misnomer. I use "overabsorption" and encourage others to do so. I suppose to track it down would require going back over the seminal papers on the subject. mam On 11/16/2010 10:19 AM, Scott Calvin wrote: Hi all, As some of you know, I'm currently working on a textbook on XAFS analysis. Because of that, I'm going to occasionally pose some questions for the list that may seem a bit random. I hope none of you mind me using the list in this way; the questions may seem to come out of left field, but I think they will still be of interest to many. With that said, here's my question for today: What is the origin of the use of "self-absorption" to describe the suppression of fine-structure observed in thick, concentrated samples measured in fluorescence? I understand the physics of the effect itself, my question is the curious wording. Compared to a thin concentrated sample, the effect might better be described as "saturation," while compared to a thick dilute sample, it's actually related to a lack of absorption by other elements. --Scott Calvin Faculty at Sarah Lawrence College Currently on sabbatical at Stanford Synchrotron Radiation Laboratory ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Transmission EXAFS sample
Hi Jatin, Matt covered most of what I would say, but I'll add a few comments of my own. I'm not sure how you arrived at the conclusion that you have only a few percent of what you need--you must be assuming a sample area somehow. I have frequently made transmission measurements on samples where I only had a few milligrams available. Generally, I did it by spreading it on a layer of tape as well as I could and then folding the tape over and over again--sometimes to make as many as 16 layers. (Of course, that many layers is not advisable if you're below 6 keV or so, as the absorption of the tape itself would kill the signal). Even if there are lots of pinholes because you can't cover the tape effectively, 16 layers from folding will make them cancel out fairly well. I can then narrow the beam a bit to match the size of my sample. Flux isn't really the issue here, so I don't even need a focussed beamline--I can just narrow the slits. Two other tips: 1) Realize that even with a tiny amount of sample that much of it won't end up on the tape. The process of brushing on tape is designed to separate the small grains from the big ones, with only the small ones ending up on tape. Allow that to happen! 2) You can sometimes get a second piece of tape to have some sample on it by putting it sticky side down on your mortar and peeling it back. A thin layer of dust from the sample will stick to the tape, and give you a little more absorption and a bit more of a uniform distribution. If you stack that with the primary piece of tape and then fold a few times, you may end up in pretty good shape, as long as you're not operating at a low enough energy so that all the layers of tape are a problem.. This procedure doesn't give me the best data I've ever seen, but it's often not bad. --Scott Calvin Sarah Lawrence College On Nov 19, 2010, at 8:13 AM, Matt Newville wrote: Dear Jatin, The idea that the optimum absorption length (mu*t) for transmission experiments is 2.3 assumes that the errors in the measurement are due to counting statistics of the x-rays. For any synchrotron experiment, the number of x-rays in the transmission chamber is high enough that the noise from counting statistics is rarely significant. This means that using a value of 2.3 is really not that important. The more important issues are a) having a uniform sample. b) not having (mu*t) so high that higher-order harmonics dominate the transmission measurement. For transmission measurements, it's difficult to overstate the importance of a uniform sample. For an ideal thickness, I would say that the better rules of thumb than mu*t = 2.3 are to aim for an edge step of 0.1 to 1.0, and a total absorption less than 3.0. If you only have enough material for an edge step as low as 0.02 (as you imply), then measuring in fluorescence or electron emission is probably a better choice. Such a sample won't be severely affected by "self-absorption" (or "over absorption" to use the term this mailing list prefers) in the fluorescence measurement. I would recommend simultaneously measuring transmission and florescence for such a sample. My concern about a very thin sample is uniformity. Specifically, is the grain size really well below mu/0.02 so that a collection of particles can give a uniform thickness? Since you didn't give any details of the system, it's hard to guess. Is it feasible to pack that material into a smaller area so that the thickness is increased and use a smaller x-ray beam? -- Can my sample be only few percentage of the "actual amount" (i.e. calculated based on above fact) required, and still i can perform transmission EXAFS ? How would this affect my data ? (I guess, it will be heavily dominated by noise) I would guess that a sample with mu*t of 0.02 would be dominated by pinholes. -- What if, i have required amount of sample but since material's density is so high that it yields only small volume of powder (for a given weight), that it can not be covered up on multiple layers of Kapton tape to ensure pinhole-free sample ? If you cannot get the grain size small enough to have many overlapping grains in the sample, the sample won't be uniform enough for good transmission data. The techniques of using multiple layers of mixing with a low-Z binder don't solve this problem. These do help to make a uniform collection of overlapping grains, but don't make the grains smaller. I would recommend trying to increase the thickness at the expense of cross-sectional area, and/or measuring in both transmission and fluorescence. Hope that helps, --Matt ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Transmission EXAFS sample
On Nov 21, 2010, at 2:45 AM, Jatinkumar Rana wrote: Hi Scott, Yes I have assumed the sample cross section area to be 1 sq. cm. and then calculated the amount of sample required for that. What i planned is following : I would calculate the amount of sample required for 1sq.cm area, take that amount of sample and make it very fine paste using mortar and pestle, and then apply it uniformly on a piece of kapton tape. Then fold the tape over and over again in such a way that final bunch of tapes will yield to 1 sq.cm. area containing the required amount of sample. Will it be the right approach ?? OR I can take randomly few milligrams of powder (i.e. not strictly as per calculation) and make a several uniform layers of tape ?? With best regards, Jatin -- Hi Jatin, I'm not sure I understand. If you have enough sample for the 1 square centimeter target, then there shouldn't be a problem, right? I was assuming from your initial question that you weren't going to have enough sample to do that. --Scott Calvin Faculty at Sarah Lawrence College Currently on sabbatical at Stanford Synchrotron Radiation Laboratory___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Transmission EXAFS sample
To my mind, when considering sample preparation the important thing is not so much the "right" thickness, as knowing the effects to guard against as the thickness deviates toward the thin or thick side. As transmission samples become thicker, the problem of "unwanted" photons becomes more severe. Those photons may be harmonics, photons scattered into the It detector, or photons from the tails of the resolution curve of the monochromator. As transmission samples become thinner, uniformity becomes more of an issue. If you play with the equations, you'll see that if your sample is a mixture of regions that have a thickness of 1.0 absorption lengths and regions that have a thickness of 2.0 absorption lengths, the spectrum is less distorted than if it is a mixture of 0.5 and 1.0 absorption lengths. So if a sample is on the thick side, it is particularly important to guard against harmonics in the beam and scattered photons. If it is on the thin side, it is particularly important to guard against nonuniformity. To put it another way, problems are synergistic. With a well- conditioned beam, a uniform sample, and linear detectors, the thickness almost doesn't matter (within reason)--at a modern beamline, a total absorption of even 0.05 or 4.0 will work. But as each of those conditions deviates from the ideal, distortions become much more severe. There's an old joke about someone on a diet going in to a fast food joint and asking for a double bacon cheeseburger, a large fries...and a diet Coke. In XAFS measurements, that attitude actually kind of works, because of the synergies I just discussed. Personally, I trust my ability to condition the beam and minimize scattering more than I trust my ability to make a uniform sample, so I lean a little toward the thicker side. --Scott Calvin Faculty at Sarah Lawrence College Currently on sabbatical at Stanford Synchrotron Radiation Laboratory On Nov 22, 2010, at 5:13 AM, Welter, Edmund wrote: Dear Jatin, the optimum mued of 2.x is not just derived by simple photon counting statistics. As Matt pointed out, for transmission measurements at a synchrotron beamline in conventional scanning mode this is seldom a matter. Nevertheless, one should avoid to measure subtle changes of absorption at the extreme ends, that is, transmission near 0 % or 100 %. In optical photometry this is described by the more or less famous "Ringbom plots" which describe the dependency of the accuracy of quantitative analysis by absorption measurements (usually but not necessarily in the UV/Vis) from the total absorption of the sample. This time the number is only near to 42, the optimum transmission is 36.8 % (mue = 1). So, to achieve the highest accuracy in the determination of small Delta c (c = concentration) you should try to measure samples with transmissions near to this value (actually the minimum is broad and transmissions between 0.2 and 0.7 are ok). In our case, we are not interested in the concentration of the absorber, but we are also interested in (very) small changes of the transmission resp. absorption in our samples. Or, using Bouger, Lambert Beer's law, in our case mue (-ln(I1/I0) is a function of the absorption coefficient (mue0). The concentration of the absorber and the thickness (d) of the sample are constant. -ln(I1/I0) = mue0 * c * d But then: If the optimum is a mue between 0.35 and 1.6 why are we all measuring successfully (ok, more or less ;-) using samples having a mue between 2 and 3? ...and 0.35 seems desperately small to me! Maybe sample homogeneity is an issue? Cheers, Edmund Welter ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Transmission EXAFS sample
On Nov 22, 2010, at 7:09 AM, Jatinkumar Rana wrote: Hi Scott, Sorry for mixing up the things. For the case, when i have very limited amount of sample that i can not cover 1sq.cm area, you, Matt and others have given very very clear explanation about possible solutions and the probable effects on data quality. I am really very thankful to all of you for sharing your experience and expertise. My last post was with reference to the case when i have enough powders (i.e., reference oxide compounds). It is just to be ensured that i am doing things 100% exactly in a same way it has to be done. With best regards, Jatin -- Jatinkumar Rana Yes, Jatin, the procedure you described is fine. There is no "right" way to make samples, although there are many wrong ways. --Scott Calvin Faculty at Sarah Lawrence College Currently on sabbatical at Stanford Synchrotron Radiation Laboratory___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
[Ifeffit] Distortion of transmission spectra due to particle size
Hi all, I'm tracking down a piece of EXAFS lore which I think is incorrect. I've seen it said that you cannot compensate for the distortion introduced by large particle sizes by making the sample thicker. Certainly thick samples have their own set of issues (e.g. "thickness effects" from harmonics), but I've seen the claim that the mathematics of the distortions introduced by nonuniformity means that there is a particle-size distortion that is independent of thickness. This claim is sometimes accompanied by an equation giving chi_eff/chi_real as a function of particle size diameter D and various absorption coefficients. I've eventually traced this equation back to a paper by Lu and Stern from 1983, have walked through the derivation, and believe there is a flaw in the logic that has led to the erroneous--and widely quoted-- conclusion that thickness cannot compensate for particle size. The paper, for those who want to follow along, is K. Lu and E. A. Stern, "Size effect of powdered sample on EXAFS amplitude," Nucl. Instrm. and Meth. 212, 475-478 (1983). They calculate the intensity transmitted by a spherical particle, and from there calculate the attenuation in the normalized EXAFS signal for a beam passing through that particle. They then, however, extend this to multiple layers of particles by the following argument: "Finally, the attenuation in N layers is given by (I/I0)^N, where I is the transmitted intensity through one layer. Xeff for N layers is then the same as for a single layer since N will cancel in the final result." This is not the case, is it? It seems to me that their analysis assumes that the spheres in subsequent layers line up with the spheres in previous ones, so that thick spots are always over thick and thin spots over thin. It's little wonder, then, that making the sample thicker does not improve the uniformity according to that analysis. I've done a calculation for the effects of uniformity in a somewhat different way, and found that it is indeed true that multiple layers on particles show less distortion due to nonuniformity that a single layer of particles of the same size, just as one would intuitively imagine, and in contrast to Lu and Stern. Do you agree that the extrapolation to multiple layers in the original Lu and Stern paper is not correct, or have I misled myself somehow? --Scott Calvin Faculty at Sarah Lawrence College Currently on sabbatical at Stanford Synchrotron Radiation Laboratory P.S. None of this should be taken as an endorsement of overly thick samples! Harmonics and the like are a concern regardless of the uniformity issue. ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Distortion of transmission spectra due to particle size
Some follow-up. This, for example, is from an excellent workshop presentation by Rob Scarrow: Errors from large particles are independent of thickness The relative (%) variation in thickness depends on the ratio (particle diameter / avg. thickness), so it is tempting to increase the avg. thickness (i.e. increase μx) as an alternative to reducing the particle diameter. However, simulations of MnO2 spectra for average Δμ0x = 1, 2 or 3 show that the errors in derived pre-edge peak heights and EXAFS amplitude factors are significant when diameter > 0.2 / Δμ0, but that they are not affected by the average sample thickness. (Δμ0 refers to the edge jump) The equation at right is given by Heald (quoting earlier work by Stern and Lu). D is particle diameter, μ1 is for just below the edge, and Δμ =μ(above edge) - μ1. I've seen similar claims elsewhere, although Scarrow's is particularly clear and unambiguous. The equation Scarrow gives is indeed the one from Lu and Stern, and the simulations are based on that equation. That Lu-Stern equation is derived for a monolayer of spheres, and then experimentally tested with multiple layers of tape. I'm still trying to work through the math to see how it works for multiple layers. I'm not convinced that the N divides out as is claimed in the article. As Matt says, it wasn't their main point. There is no question that if the particle size is large compared to an absorption length there will be nonuniformity and thus distortions. But compare a monolayer of particles with a diameter equal to 0.4 absorption lengths with four strips of tape of that kind stacked. Do we really think the distortion due to nonuniformity will be as bad in the latter case as in the first? In practice, I think many transmission samples fall in roughly that regime, so the question isn't just academic. I'll keep trying to work through the math and let you know what I find. --Scott Calvin Faculty at Sarah Lawrence College Currently on sabbatical at Stanford Synchrotron Radiation Laboratory___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Distortion of transmission spectra due to particle size
On Nov 22, 2010, at 2:55 PM, Scott Calvin wrote: But compare a monolayer of particles with a diameter equal to 0.4 absorption lengths with four strips of tape of that kind stacked. Do we really think the distortion due to nonuniformity will be as bad in the latter case as in the first? In practice, I think many transmission samples fall in roughly that regime, so the question isn't just academic. OK, I've got it straight now. The answer is yes, the distortion from nonuniformity is as bad for four strips stacked as for the single strip. This is surprising to me, but the mathematics is fairly clear. Stacking multiple layers of tape rather than using one thin layer improves the signal to noise ratio, but does nothing for uniformity. So there's nothing wrong with the arguments in Lu and Stern, Scarrow, etc.--it's the notion I had that we use multiple layers of tape to improve uniformity that's mistaken. A bit on how the math works out: for Gaussian distributions of thickness, the absorption is attenuated (to first order) by a term directly proportional to the variance in the distribution. The standard deviation in thickness from point to point in a stack of N tapes generally increases as the square root of N (typical statistical behavior). This means that the fractional standard deviation goes down as the square root of N. In casual conversation, we would usually identify a sample with thickness variations of +/-5 % as being "more uniform" than one with thickness variations of +/- 8%, so it's natural to think that a stack of tapes is more uniform than a single one. But since the attenuation is proportional to the variance (i.e. the square of the standard deviation), it actually increases in proportion to N. Since the absorption is also increasing in proportion to N, the attenuation remains the same size relative to the absorption, and the spectrum is as distorted as ever. This result doesn't actually depend on having a Gaussian distribution of thickness. if each layer has 10% pinholes, for instance, at first blush it seems as if two layers should solve most of the problem: the fraction of pinholes drops to 1%. But those pinholes are now compared to a sample which is twice as thick, on average, and thus create nearly as much distortion as before. Add to this that there is now 9% of the sample that is half the thickness of the rest, and the situation hasn't improved any. I've worked through the math, and the cancellation of effects is precise--a two layer sample has the identical nonuniformity distortion to a one layer one. (There is probably a simple and compelling argument as to why this distortion is independent of the number of randomly aligned layers for ANY thickness distribution, but I haven't yet found it.) * * * For me personally, knowing this will cause some changes in the way I prepare samples. First of all, I'm going to move my bias more toward the thin end. My samples are generally pretty concentrated, so signal to noise is not a big issue. If I'm also not improving uniformity by using more layers of tape, there's no reason for me not to keep the total absorption down around 1, rather than around 2. Secondly, I'll approach the notion of eyeballing the assembled stack of tapes for uniformity, whether with the naked eye or a microscope, with more caution--particularly when teaching new students. The idea that a sample which has no evident pinholes is a better sample than one that does is not necessarily true, as the example above with the single layer exhibiting 10% pinholes as compared to the double layer exhibiting 1% demonstrates. Stressing the elimination of visible pinholes will tend to bias students toward thicker samples, but not necessarily better ones. --Scott Calvin Faculty at Sarah Lawrence College Currently on sabbatical at Stanford Synchrotron Radiation Laboratory ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Distortion of transmission spectra due to particle size
Matt, Your second simulation confirms what I said: The standard deviation in thickness from point to point in a stack of N tapes generally increases as the square root of N (typical statistical behavior). Now follow that through, using, for example, Grant Bunker's formula for the distortion caused by a Gaussian distribution: (mu x)eff = mu x_o - (mu sigma)^2/2 where sigma is the standard deviation of the thickness. So if sigma goes as square root of N, and x_o goes as N, the fractional attenuation of the measured absorption stays constant, and the shape of the measured spectrum stays constant. There is thus no reduction in the distortion of the spectrum by measuring additional layers. Your pinholes simulation, on the other hand, is not the scenario I was describing. I agree it is better to have more thin layers rather than fewer thick layers. My question was whether it is better to have many thin layers compared to fewer thin layers. For the "brush sample on tape" method of sample preparation, this is more like the question we face when we prepare a sample. Our choice is not to spread a given amount of sample over more tapes, because we're already spreading as thin as we can. Our choice is whether to use more tapes of the same thickness. We don't have to rerun your simulation to see the effect of using tapes of the same thickness. All that happens is that the average thickness and the standard deviation gets multiplied by the number of layers. So now the results are: For 10% pinholes, the results are: # N_layers | % Pinholes | Ave Thickness | Thickness Std Dev | # 1| 10.0 |0.900 |0.300 | # 5| 10.0 |4.500 |0.675 | #25| 10.0 |22.500 |1.500 | For 5% pinholes: # N_layers | % Pinholes | Ave Thickness | Thickness Std Dev | # 1| 5.0 |0.950 |0.218 | # 5| 5.0 |4.750 |0.485 | #25| 5.0 |23.750 |1.100 | For 1% pinholes: # N_layers | % Pinholes | Ave Thickness | Thickness Std Dev | # 1| 1.0 |0.990 |0.099 | # 5| 1.0 |4.950 |0.225 | #25| 1.0 |24.750 |0.500| As before, the standard deviation increases as square root of N. Using a cumulant expansion (admittedly slightly funky for such a broad distribution) necessarily yields the same result as the Gaussian distribution: the shape of the measured spectrum is independent of the number of layers used! And as it turns out, an exact calculation (i.e. not using a cumulant expansion) also yields the same result of independence. So Lu and Stern got it right. But the idea that we can mitigate pinholes by adding more layers is wrong. --Scott Calvin Faculty at Sarah Lawrence College Currently on sabbatical at Stanford Synchrotron Radiation Laboratory On Nov 24, 2010, at 6:05 AM, Matt Newville wrote: Scott, OK, I've got it straight now. The answer is yes, the distortion from nonuniformity is as bad for four strips stacked as for the single strip. I don't think that's correct. This is surprising to me, but the mathematics is fairly clear. Stacking multiple layers of tape rather than using one thin layer improves the signal to noise ratio, but does nothing for uniformity. So there's nothing wrong with the arguments in Lu and Stern, Scarrow, etc.--it's the notion I had that we use multiple layers of tape to improve uniformity that's mistaken. Stacking multiple layers does improve sample uniformity. Below is a simple simulation of a sample of unity thickness with randomly placed pinholes. First this makes a sample that is 1 layer of N cells, with each cell either having thickness of 1 or 0. Then it makes a sample of the same size and total thickness, but made of 5 independent layers, with each layer having the same fraction of randomly placed pinholes, so that total thickness for each cell could be 1, 0.8, 0.6, 0.4, 0.2, or 0. Then it makes a sample with 25 layers. The simulation below is in python. I do hope the code is straightforward enough so that anyone interested can follow. The way in which pinholes are randomly selected by the code may not be obvious, so I'll say hear that the "numpy.random.shuffle" function is like shuffling a deck of cards, and works on its array argument in-place. For 10% pinholes, the results are: # N_layers | % Pinholes | Ave Thickness | Thickness Std Dev | # 1| 10.0 |0.900 |0.300 | # 5| 10.0 |0.900 |0.135 | #25| 10.0 |0.900 |0.060 | For 5% pinholes: # N_layers | % Pinholes | Ave Thickness | Thickness Std Dev | # 1| 5.0 |0.950 |
Re: [Ifeffit] Distortion of transmission spectra due to particle size
Nov 24, 2010, at 12:04 PM, Matt Newville wrote: Scott, You said: the distortion from nonuniformity is as bad for four strips stacked as for the single strip. As I showed earlier, a four layer sample is more uniform than a one layer sample, whether the total thickness is preserved or the thickness per layer is preserved. For 1% pinholes: # N_layers | % Pinholes | Ave Thickness | Thickness Std Dev | # 1| 1.0 |0.990 |0.099 | # 5| 1.0 |4.950 |0.225 | #25| 1.0 |24.750 |0.500| Yes, the sample with 25 layers has a more uniform thickness. As before, the standard deviation increases as square root of N. Using a cumulant expansion (admittedly slightly funky for such a broad distribution) necessarily yields the same result as the Gaussian distribution: the shape of the measured spectrum is independent of the number of layers used! And as it turns out, an exact calculation (i.e. not using a cumulant expansion) also yields the same result of independence. OK... The shape is the same, but the relative widths change. 24.75 +/- 0.50 is a more uniform distribution than 0.99 +/- .099. Perhaps this is what is confusing you? The thicker sample is more uniform (in the sense of fractional uniformity), but the distortion in the normalized mu(E) due to the nonuniformity is identical. That is exactly what is surprising, and what I initially did not believe. Now I am firmly convinced that it is true. So Lu and Stern got it right. But the idea that we can mitigate pinholes by adding more layers is wrong. Adding more layers does make a sample of more uniform thickness. Perhaps "mitigate pinholes" means something different to you? By "mitigate pinholes" I mean that we can obtain a normalized energy spectrum that is closer to the mu(E) we are trying to measure. But we can't do that by adding more identically distributed layers if the thick and thin spots are randomly stacked--we end up with exactly the same normalized mu(E). (As usual, this analysis neglects signal to noise, harmonics, and the like.) I used a lot of words there to try to be precise. But basically, stacking two layers of tape with the hope that pinholes will tend to be cancelled out will not work. In your original message (in which you set out to "track down" a piece of "incorrect lore") you said that Lu and Stern assumed that layers were stacked "so that thick spots are always over thick and thin spots over thin". They did not assume that. Given that initial misunderstanding, and the fact that you haven't shown any calculations or simulations, it's a bit hard for me to fathom what you think Lu and Stern "got right" or wrong. They got everything right. I was trying to save a different piece of lore that I think is even more widely disseminated--that stacking thin layers of tape reduces the amount of distortion due to nonuniformity as compared to one thin layer of tape. I thought I had shown calculations (the Gaussian case), and Jeremy has shown simulations which confirm the result for the pinhole case. The main point of their work is that it is better to use more layers to get to a given thickness. You seem to have some objection to this, but I cannot figure out what you're trying to say. I agree that your statement is a true one which is also consistent with their paper, but would respectfully differ as to what the main point of the paper is. The abstract reads: Powdered samples samples are commonly used to measure the extended X- ray absorption fine structure (EXAFS) and near-edge structure. It is shown that sizes of particles typically employed produce significant reduction in the EXAFS amplitude for concentrated samples. The distortion is calculated and found to be in good agreement with measurements on FeSi2 samples. To obtain accurate EXAFS amplitudes on powdered samples it is necessary to use particles significantly smaller than 400 mesh when the atom of interest is concentrated. The use of increasing number of layers as the particles are sieved more finely is done to provide a controlled experiment in which differences in signal to noise and thickness effects are not a factor. Suggesting that it is better to use more layers to get a given thickness is an indirect way of getting at the real issue, which is particle size. For a given particle size, multiple layers provide no help whatsoever with nonuniformity-related distortions, but merely allow for the desired signal to noise. Lu and Stern provide the correct emphasis in their paper. It is only some of the subsequent XAFS practitioners, including myself until about 24 hours ago, who placed the stress on the multiple layers per se as addressing the uniformity issue. --Scott Calvin Faculty at Sarah
Re: [Ifeffit] Cadmium K-edge
Hi Alan, (To the list: I discussed Alan's system a bit with him when he came to SSRL, but don't recall the details. That's why in this reply I have a little more knowledge about it than what he posted yesterday.) I suspect the peak at 2.3 A does indeed represent structure of some kind. Your k-space data for sample 1 has very little noise, so I don't think it's spurious in that sense. What strikes me about your data is that the FT's have so little structure beyond the 1.8 A peak. I don't recall exactly the nature of your system: it was cadmium loaded on to the surface of a substrate, but I don't recall what the substrate was. If it's low-Z (for example, silica), then the scattering off of whatever the oxygen is bonded to (in my example, silicon) would show up pretty weakly, and that could be the peak at 2.3 A. It's absence in sample 2 could represent the cadmium ion being less selective about what surface site it bonds to, so that the Cd-Si distance (in my example) varies and the peak washes out. That's all just an example of the kind of thing that could be responsible for what you're seeing. I agree that a good plan is to use a Cd-O structure to at least let you model the first peak. For the second peak, though, you may have to take more inspiration from the structure of the substrate than the structure of Cd-O. It's also intriguing that the white line in your sample is larger than in the CdO standard from the database. Someone with more XANES experience than I might be able to suggest why that is--something about these being bound to the surface? Is my recollection of your system correct? If so, what was the substrate? --Scott Calvin Faculty at Sarah Lawrence College Currently on sabbatical at Stanford Synchrotron Radiation Laboratory On Jan 5, 2011, at 11:23 PM, Alan Du wrote: Hi all, I processed my XAS data and compare them with standards from Farrel Lytle Database. (http://img717.imageshack.us/i/xasw.jpg/). In the RSF, both sample 1 and 2 has a major peak at R = ca. 1.8 A. sample 1 has an additional peak at R = ca. 2.3 A. I wonder if this tiny peak is significant. It seems both samples has peak positions similar to those of CdO standard, which sound logical because the cadmium will bind to the surface oxygen of the material. I looking at fitting sample 1 and 2 using whatever crystallographic info of CdO I can get my hands on. Wonder if the plan make sense. Happy new year, Alan ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
[Ifeffit] Athena crash during LCA
Hi all, Athena has been crashing for me during one particular linear combination analysis, and I'm wondering if any of you have an explanation. A project file which demonstrates the problem is attached. I am using Athena 0.8.054 with ifeffit 1.2.10 on a MacBook Pro using OS 10.5.8. The behavior can be seen by trying to fit the Data group by using a linear combination of Standard 1 and Standard 2 in chi(k). It gets as far as "plotting in k-space from group 'Data'...done!' and then hangs, with the watch icon remaining indefinitely. This doesn't happen in norm(E) or deriv(E) fits, and doesn't happen when Standard 1 is not used. But Standard 1 appears to have uncorrupted chi(k) data when plotted directly. Standard 1 is from one of the XDAC lines at NSLS, as are the standards that work. I've tried saving Standard 1 and reading it in as a new group, but no luck. I've also looked at the data in Standard 1, and I don't see anything out of place, such as a place where the energy backs up. I've tried starting the chi(k) file with zero values, truncating the end values, and changing the background spline to end at the same point the data does. And it always hangs at the same point. Any ideas? I'd like to use this data for a workshop next week! --Scott Calvin Faculty at Sarah Lawrence College Currently on sabbatical at Stanford Synchrotron Radiation Laboratory LCACrash.prj Description: Binary data ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Athena crash during LCA
Jeff--sorry, I guess my grammar got a little convoluted. It DOES work with norm(E) and deriv(E) fitting for me as well. It does NOT work with chi(k) fitting for me. --Scott Calvin Faculty at Sarah Lawrence College Currently on sabbatical at Stanford Synchrotron Radiation Laboratory On Jan 7, 2011, at 7:58 PM, Jeff Terry wrote: Did not try really hard to work with it. Fitting Data as norm(E) from -20.000 to 30.000 Fit included 98 data points and 3 variables R-factor = 0.000263 chi-square = 0.04347 reduced chi-square = 0.0004482 groupweight === 2: Standard 1 1.000(0.000) 3: Standard 2 0.455(0.000) groupe0 shift === 2: Standard 1-0.905( 0.000) 3: Standard 2-0.905( 0.000) On Jan 7, 2011, at 8:07 PM, Scott Calvin wrote: Hi all, Athena has been crashing for me during one particular linear combination analysis, and I'm wondering if any of you have an explanation. A project file which demonstrates the problem is attached. I am using Athena 0.8.054 with ifeffit 1.2.10 on a MacBook Pro using OS 10.5.8. The behavior can be seen by trying to fit the Data group by using a linear combination of Standard 1 and Standard 2 in chi(k). It gets as far as "plotting in k-space from group 'Data'...done!' and then hangs, with the watch icon remaining indefinitely. This doesn't happen in norm(E) or deriv(E) fits, and doesn't happen when Standard 1 is not used. But Standard 1 appears to have uncorrupted chi(k) data when plotted directly. Standard 1 is from one of the XDAC lines at NSLS, as are the standards that work. I've tried saving Standard 1 and reading it in as a new group, but no luck. I've also looked at the data in Standard 1, and I don't see anything out of place, such as a place where the energy backs up. I've tried starting the chi(k) file with zero values, truncating the end values, and changing the background spline to end at the same point the data does. And it always hangs at the same point. Any ideas? I'd like to use this data for a workshop next week! --Scott Calvin Faculty at Sarah Lawrence College Currently on sabbatical at Stanford Synchrotron Radiation Laboratory ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Athena crash during LCA
Hi Matt and Jeff, I just upgraded to 0.8.061, and it works in that version, so apparently whatever the bug was has been fixed. Thanks for checking it out for me! --Scott Calvin Faculty at Sarah Lawrence College Currently on sabbatical at Stanford Synchrotron Radiation Laboratory On Jan 8, 2011, at 7:08 AM, Matt Newville wrote: Hi Scott, Like for Jeff, the chi(k) LCA fit works for me with Athea 0.8.061 on OSX 10.6. If I had to make a guess or recommendation for something to try it would be this: Standard2 extends to pretty far out in K (>20Ang^-1) and Standard2 is pretty glitchy beyond 12Ang^-1 or so. You might try truncating the data and/or removing the obvious glitches before doing the background subtraction and LCA fit. Again, that's a guess. Does the latest iXAFS3 not work for you on 10.5.8? I haven't tried it, but the intention was that it would work on 10.5. --Matt ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
[Ifeffit] R dependence of S02
Hi all, I've been pondering how much R dependence we should expect to see in S02; that is, how much R dependence is shown by intrinsic losses. I've looked at quite a bit of the literature, not to mention old threads from this mailing list. One recent (2002) reference is this one by Campbell et al.: http://prb.aps.org/pdf/PRB/v65/i6/e064107 My understanding is that S02 is intended to account for intrinsic losses; that is, those that are determined when the core hole forms. Extrinsic losses, such as inelastic scattering of the photoelectron and the effect of the core-hole lifetime, are accounted for by a mean free path term. The mathematics of how ifeffit implements this are here: http://cars9.uchicago.edu/~newville/feffit/feffit.ps Intrinsic losses are dependent on k for at least two reasons. One is that the cross-section of the intrinsic effects themselves depends on the energy of the incident x-ray. This is evident if one thinks in terms of shake-up and shake-off events. But another reason is that shake-up and shake-off events rob some energy from the primary photoelectron. At low k, ten or twenty electron volts can alter the phase of the primary photoelectron significantly, and thus shake-up and shake-off events will tend to cancel each other out. But at high k, the energy robbed from the photoelectron is less significant, because the EXAFS oscillations are more spread out in energy. Thus, shake-up and shake-off events, while still occurring, will not suppress the EXAFS amplitude as much at high k. S02 therefore gradually rises through most of the EXAFS region to reach a limiting value of 1 well above the top of the EXAFS region. The latter effect--the fact that removing a specified amount of energy from the primary photoelectron has less of an effect at high k than at low--also implies an R dependence. Low R oscillations are further apart in energy than high R oscillations, and thus over a specified k range low R oscillations should be less affected. In other words, S02 should show a modest decrease with increasing R over typical EXAFS ranges. Some papers, in particular those with John Rehr as an author, confirm that S02 should have an R dependence, but don't discuss the implications much. While ifeffit allows for floating ei, a parameter related to the mean free path, as I understand it that will still give a damping of the amplitude that is exponential in R. It seems to me that the S02 dependence on R, in contrast, is likely to be more gentle. Why is this important? Several authors, including myself, have analyzed crystallite size and/ or morphology by comparing the coordination number of successive scattering shells. This is potentially much more accurate than just finding the first-shell coordination number, because it is independent of any amplitude effects that are independent of R, such as normalization errors and many experimental effects. Errors in the mean free path are a bit more significant, but the exponential dependence of the mean free path gives it a very different shape than effects from size and morphology. But an R-dependence of S02 would be troubling, as the functional form might look a bit more like a size effect. So one question is this: does anyone have an order-of-magnitude estimate of how much R dependence to expect in S02 over the EXAFS range? If over a range of 2 to 6 angstroms S02 changed by even a few percent, that could have a significant effect on the kind of size analyses I mentioned in the preceding paragraph. Of course, another question is if I've completely blown it anywhere in my discussion above; I've just been puzzling this out over the last few days! --Scott Calvin Faculty at Sarah Lawrence College Currently on sabbatical at Stanford Synchrotron Radiation Laboratory ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Fe in glassy ceramics
Hi Andrei, It's not unreasonable to model such a thing. Use a highly constrained model for the spinel, not allowing coordination numbers to float and using very few sigma2 parameters. I've published a model like that for fitting spinels here: http://prb.aps.org/abstract/PRB/v66/i22/e224405 My model is more complicated than yours needs to be, because you don't really have site disorder; all the sites are Fe. Then multiply the amplitudes by a guessed parameter representing the fraction of the iron that is in the spinel phase. The remaining amplitude gets used for the glassy phase, which can possibly be fit by a single iron-oxygen path, which can be cloned from one of the spinel near neighbor paths, but now with most parameters floated. You may very well need to fix S02 based on the value from a standard. Fit over a wide range in the Fourier transform, so that you take advantage of high R peaks to pin down the spinel part. Worth a shot, anyway. The key is being aggressive with assumptions and constraints, and then being honest about how that affects your uncertainties in the end. --Scott Calvin Faculty at Sarah Lawrence College Currently on sabbatical at Stanford Synchrotron Radiation Laboratory On Feb 16, 2011, at 4:53 AM, Andrei Shiryaev wrote: Dear colleagues, Probably similar questions were already asked, but nevertheless I would appreciate some advice how to solve the problem. We are looking at Fe environment in complex glasses, which underwent partial crystallization. A fraction of Fe has precipitated as spinel- like nanocrystals, another fraction remains in glass. Spinel itself is already challenging for XAFS, but our task is to understand the Fe fraction in glass (e.g., Fe-O bond length). We do not know the exact composition of the crystalline phases, thus we can not record a suitable standard. Does anybody have an idea how could we try to separate the Fe in glass from Fe(cryst) contributions? I fully realize that the problem is ill-posed, perhaps, some people had somehow solved this problem. Thanks a lot, Andrei Shiryaev Instituteof Physical Chemistry Moscow Russia ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
[Ifeffit] Looking for reviewers for XAFS textbook chapters
Hi all, I am looking for people interested in reviewing portions of the XAFS textbook I am working on. I am particularly hoping that some people less experienced in XAFS will volunteer, so that I can get feedback on how well it teaches and how well it covers the topics you'd like to know more about. Experts are also welcome! The first chapter I've got ready is the chapter on sample preparation. If you are interested in reviewing this chapter and can get feedback back to me within two weeks, please email me at scal...@slc.edu . --Scott Calvin Sarah Lawrence College ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Looking for reviewers for XAFS textbook chapters
Darn it! I did not mean to send the packet to the whole list. If anyone is interested in reviewing, please send a request to scal...@slc.edu . I apologize for spamming you all with a multi-MB file. --Scott On Mar 8, 2011, at 12:24 PM, Scott Calvin wrote: Great! I'm attaching the review packet. The page numbering is a bit chaotic, because it's actually several documents stitched together: The copyright notice. My publisher is allowing me to distribute review copies, but only on the condition that reviewers not redistribute them. A detailed table of contents. This is more detailed than the final version will be, but gives context to reviewers as to what will be covered elsewhere. An excerpt from the preface explaining the origin of the cartoon characters. (Yes, there are cartoon characters...) An introduction to the cartoon characters. Chapter 3, on sample preparation. References In the chapter, I've placed numbers along the margin to aid in making comments. For example, you might want to comment on page 18, lines 6-7. The numbers won't appear in the final version. If you have any questions about the context of the chapter, the intended audience, or whatever, feel free to ask! --Scott Calvin Sarah Lawrence College On Mar 8, 2011, at 11:27 AM, Prof. Dr. Peter Leinweber wrote: hello, i am interested in reviewing this. i am not experienced but i want to learn more. i look forward to hearing from you, peter leinweber Am 08.03.2011 04:13, schrieb Scott Calvin: Hi all, I am looking for people interested in reviewing portions of the XAFS textbook I am working on. I am particularly hoping that some people less experienced in XAFS will volunteer, so that I can get feedback on how well it teaches and how well it covers the topics you'd like to know more about. Experts are also welcome! The first chapter I've got ready is the chapter on sample preparation. If you are interested in reviewing this chapter and can get feedback back to me within two weeks, please email me at scal...@slc.edu . --Scott Calvin Sarah Lawrence College ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit -- Prof. Dr. Peter Leinweber Soil Science Institute for Land Use University of Rostock D-18051 Rostock/Germany Tel.: +49 381 498 3120 Fax: +49 381 498 3122 ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
[Ifeffit] Looking for reviewers for another chapter
Hi all, I have completed the draft of another chapter of my textbook, and am again interested in reviewers (repeat reviewers are fine!). The chapter I've currently got ready is about the parameters of the EXAFS equation, covering physical meaning, effect on graphs, a bit on common constraint schemes, etc.. This will be chapter 10 in the book, and so it's not suitable for absolute beginners. I am interested, though, in hearing from people who are still learning to fit, as well as experts. If you are interested in reviewing the chapter and can get feedback back to me by 4 May, please contact me directly at scal...@slc.edu . --Scott Calvin Sarah Lawrence College ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
[Ifeffit] Bug in Athena merge preferences?
Hi Bruce (and Ifeffiteers), I'm not positive this is a bug--maybe I'm just misunderstanding how the features are supposed to work. (It wouldn't be the first time!) In Athena 0.8.061 on a Mac running OS 10.5.8, I tried the following (I tried it in more than one project, to make sure it's not an intermittent bug): --Under Edit Preferences/Merge/Merge_Weight, choose "n" and apply to future sessions --Close Athena and reopen --Open a project --Mark two data sets. --Change the "importance" of one of them. --Under the merge menu, choose "weight by importance." --Under the merge menu, "merge marked data in chi(k)" The result for me is that the merged file ignores the importance I've assigned. I've also tried it in mu(E), with the same result. It seems that perhaps the "n" value in the preferences overrides the "weight by importance" option in the menu? If so, that's not the way I expect preferences and menu options to interact. Trying it with applying the "n" only to the current session also has an interesting behavior, although not necessarily the "wrong" one: it seems to just change the radio button in the menu. Also, if I change the radio button in the menu directly, the option under Preferences changes to match. For the current session, I can understand the argument that this is a reasonable behavior. But it is likely to cause confusion if a user changes it in the menu and then tries to use the preferences to check what the future sessions value is set to--if they haven't previously changed the preferences directly during the session, they might naturally assume that the value there is the value for future sessions, but in reality it just represents the most recent menu choice they made, and could be different in a future session. My suggested fix to both issues: the preferences should control what value the radio button takes on when Athena is first opened, and changes to the preferences should cause an immediate change to the status of the radio button. Subsequent changes to the radio button should not change the value under the preferences. And the behavior of the merge function should always reflect the current value of the radio button. --Scott Calvin Sarah Lawrence College___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] EXAFS on multi phase systems
Hi Jatinkumar, It's fairly common to do so. With linear combination analysis or principle component analysis, it's necessarily the case. But it's also done with modelling using FEFF. I personally have published many papers of this type. One early paper of mine that does this is: “X-ray absorption spectroscopy of highly-cycled Li/CPE/FeS2 cells,” E. Strauss, S. Calvin, H. Mehta,* D. Golodnitksy, S. G. Greenbaum, M. L. denBoer, V. Dusheiko, and E. Peled, Solid State Ionics 164, 51 (2003). --Scott Calvin Sarah Lawrence College On May 5, 2011, at 12:26 AM, Jatinkumar Rana wrote: Dear ifeffit users, I was wondering if one could apply EXAFS to multi-phase systems (e.g. two phase systems) where both phases could be crystallographically different but contain same absorbing atom. Can anyone suggest any literature which dealt with such a problem ?? With best regards, Jatinkumar Rana ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] What does FEFF stand for?
Right, Chris. There is a factor in the EXAFS equation, f(k). In different parts of the literature, f(k) sometimes has different meanings, but within the context of FEFF it refers to the effect of the potential of the scattering atom on both the scattering amplitude (the real part) and phase (the imaginary part). Thus, it stands for "f effective." My understanding, although I could be wrong is that the "effective" part came from an improvement of the theory to account for curved-wave effects. In other words, early theories approximated the photoelectron as a plane wave, but of course it spreads out radially from the absorbing atom. That change necessitated tweaking the definitions of the factors, so it became the "effective" f. --Scott Calvin Sarah Lawrence College On May 10, 2011, at 11:53 AM, Christopher Patridge wrote: On 5/10/2011 2:49 PM, Francisco Garcia wrote: Dear all, I wish to ask a somewhat novice question: What does the acronym FEFF stand for? Thank you. ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov<mailto:Ifeffit@millenia.cars.aps.anl.gov > http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit I am pretty sure it stand for the calculated Effective Scattering factor F(eff)ective. buena salud, Chris Patridge ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] What does FEFF stand for?
I looked into this a bit further, Bruce, and I'd tentatively say the curved-wave corrections do turn out to be the source of the "eff": The earliest use of f_eff I can find is from a 1986 Phys. Rev. B article entitled "Spherical-wave effects in photoelectron diffraction," by Sagurton et al. (John Rehr is also in the author list). It says "an approximaton for including SW [spherical wave] corrections suggested recently by Rehr, Albers, and Natoli has been incorporated in some of our calculations...the net limiting result is a calculation procedure in which an effective scattering factor f_eff,j(r,theta_j) which depends on r takes the place of the usual PW [plane wave] f_j(theta_j)." In addition, was FEFF3 a multiple-scattering code? The comments in its header and the 1991 JACS article on it mention only single-scattering. It would make an extraordinary amount of sense that the "eff" would refer to FEFF's ability to handle multiple-scattering paths, but I don't think that's the actual historical origin of the terminology. And as for Anatoly's suggestion, I'll, uh, leave that one be for the moment. --Scott Calvin Sarah Lawrence College On May 10, 2011, at 12:17 PM, Bruce Ravel wrote: On Tuesday, May 10, 2011 03:03:23 pm Scott Calvin wrote: My understanding, although I could be wrong is that the "effective" part came from an improvement of the theory to account for curved- wave effects. In other words, early theories approximated the photoelectron as a plane wave, but of course it spreads out radially from the absorbing atom. That change necessitated tweaking the definitions of the factors, so it became the "effective" f. I think you are mistaken. My memory of the etymology has to do with the formalism dating back to Feff5 for computing MS paths. For a purely single scattering theory, you have an F and a phi (without the subscript eff). That is, you can simply compute the scatting function for the one scatterer and be done with it. Feff's path expansion introduced two clever things to the EXAFS business. One is that it provided a formalism for computing a single function that takes into account the angle-dependent scattering functions of all atoms in an arbitrary-geometry multiple scattering path. This allows one to treat a MS path with the familiar SS EXAFS equation only by replacing F and phi with F_eff and phi_eff. That innovation is central to how Ifeffit works. The second clever thing is that it's really fast. That's not such a big deal today, but back in the mid-90s, when a Feff run could take several minutes, a faster algorithm was very welcome indeed. B ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Question about transform windows and statistical parameters
Hi Brandon, I don't find this terribly surprising. First, a little background which you may or may not know: Reduced chi-square is a statistical parameter which requires a knowledge of the uncertainty of the measurement to compute. In theory, therefore, it "knows" that a "good" fit to noisy data will not be as close in an absolute sense as a "good" fit to high-quality data. The problem, however, is that it's difficult to know what is the proper quantity to use for the uncertainty of the measurement in EXAFS analysis. One could use the standard deviation of subsequent scans, but that is only sensitive to random scan-to-scan error. Something like, say, a monochromator glitch is quite reproducible, and yet most of us would consider it to be part of the measurement error. So the default behavior of Ifeffit is to look at the Fourier transform between 15 and 25 angstroms, and figure that any amplitude there is due to error of some kind, and not signal. It then makes the assumption that the same amount of error is present in the range being fit (i.e. the error is "white"), and from there computes the reduced chi-square. This is in some sense a dubious procedure, but the real problem is that we don't have a good method for estimating the measurement uncertainty, so we have to do something. As long as we are comparing fits to exactly the same data, on the same k-range, with the same k-weight, with the same windows, then changes in reduced chi-square are worth looking at. If all you've done is change a constraint or change the R-range being fit, for instance, a lower reduced chi-square is a good sign (use the Hamilton test if you want to be rigorous about it.) But change the k-range, or the k-weight, or the window, or the data, and Ifeffit's estimate of the uncertainty can change wildly, causing a correspondingly wild change in reduced chi-square. After all, one glitch toward the end of the k-range you are thinking can introduce a lot of high-R amplitude in to the Fourier transform, and different windows would treat it very differently. But single-point glitches often don't have much effect on the results of your fit, precisely because they do affect the high-R part of the Fourier transform much more than low-R part. Ifeffit's default behavior can be overridden, if you so choose. The parameter "epsilon" (available on the Data panel of Artemis) overrides Ifeffit's usual estimate for uncertainty. So in your case, I suggest putting a number--any number--in for epsilon, and then comparing fits using the two windows. Probably you will find that the reduced chi- squares become much more similar to each other. Incidentally, while in this case the default behavior of Ifeffit is merely distracting, there is a circumstance where it can be a more substantial problem: mutliple data-set fits (e.g. on multiple edges of the same sample). If Ifeffit finds uncertainties for the different data sets that are quite different from each other because, for instance, of the presence of a glitch in one, it will in effect weight the data very differently when doing a fit. In multiple-data set fits, therefore, it is often advisable to come up with your own scheme for setting epsilons (perhaps inversely proportional to the edge jump of the set, or something like that), to avoid wonky weightings. --Scott Calvin Sarah Lawrence College On May 11, 2011, at 12:47 PM, Brandon Reese wrote: Hello everybody, I am working on fitting some EXAFS of amorphous materials and have noticed an odd (in my mind) behavior when changing transform windows. I settled on a fit using all three k-weights and the Hanning transform window obtaining statistical parameters of R=0.0018 and chi_R=361. I decided to change the transform window to a Kaiser-Bessel to see what would happen. The refined parameters came out more or less the same, well within the error bars, with the Hanning windows having slightly smaller error bars. But my statistical parameters changed significantly to R=0.0022 and chi_R=89.354. It seems that this large change may be related to why we can't use the chi_R parameter to compare fits over different k- ranges, but I am not sure about that. Have other people seen this? I would guess it means that when looking for trends in different data sets, it is more important to be consistent, rather than which specific window type is used. Thanks, Brandon ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] EXAFS
Hi Francisco, I'll take a shot at some of these questions--I'm sure others will chime in as well. On May 11, 2011, at 2:34 PM, Francisco Garcia wrote: Dear all, I tried computing K-edge chi(k) for a solvated metal ion. For each snapshot, I simply carved out a cluster of radius 6 Ang around the metal. 1. if I set RMAX to say 5.5 in feff.inp, the chi(k) plot looks totally different from the case where the RMAX line is commented out (particularly k>10). I thought RMAX is the largest metal-H or metal-O separation but I think I am wrong RMAX is the longest half-path length to include in chi(k). For direct scattering paths, it is therefore the most distant scattering pair that is allowed to contribute. So your understanding of the meaning is essentially correct. In FEFF7, the documentation says it defaults to 2.2 times the nearest neighbor if it is not included. So commenting out RMAX does not mean to include all paths! That could be responsible for a considerable difference in chi(k). 2. I dont quite understand why the number of paths increases (sometimes by a factor of 3) when I set RMAX as opposed to the case where I comment it out. See above. If your nearest neighbor is at, say, 1.8 angstroms, then commenting out RMAX is equivalent to setting it to 3.96 angstroms. That's much less than 5.5, and explains the smaller number of paths. 3. From time to time I get warnings indicating that a water O-H distance is too short: WARNING: TWO ATOMS VERY CLOSE TOGETHER. CHECK INPUT. atoms 57 50.144002.097001.16300 70.774002.674001.00800 I tried setting FOLP to 0.8 for H but the warnings persist. Don't worry about it. This FEFF warning doesn't really expect hydrogens, and so sometimes gives a warning for atoms that would be close if they were anything bigger, but are OK if one is a hydrogen. Looking at the coordinates it specifies in your example, the two atoms are about 1 angstrom apart--OK if one is a proton. 4. On what criteria are sigma2 and s02 chosen? They are not usually reported in published theoretical papers and I was wondering if they are arbitrary. FEFF is not usually used for EXAFS on its own. Most people use a fitting program, such as Ifeffit, to optimize the values of S02 and sigma2 to provide the best fit. (Although S02 in particular is often fit for a known material, and then that value used.) When used in that way, FEFF should be run with S02 = 1 and sigma2 = 0 to avoid confusion. 5. Finally, is the chi(k) dependent of which version of FEFF is being used. I am asking this because the peak height of my FEFF6 chi(k) is slightly lower than that of published results using FEFF8. Yes, they are slightly different. For EXAFS, they probably aren't different enough to affect the answers to your scientific questions much, though. --Scott Calvin Sarah Lawrence College ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Question about transform windows and statistical parameters
Hi Brandon, Matt and Bruce both gave good, thorough answers to your questions this morning. Nevertheless, I'm going to chime in too, because there are some aspects of this issue I'd like to put emphasis on. On May 11, 2011, at 8:46 PM, Brandon Reese wrote: I tried your suggestion with epsilon and the chi-square values came out to be very similar values with the different windows. Does this mean that reporting reduced chi-square values in a paper that compared several data sets would not be necessary and/or appropriate? Bruce said "no" emphatically, and I say "yes," but I think we've understood the question differently. As Bruce says: Of course, reduced chi-square can only be compared for fitting models which compute epsilon the same way or use the same value for epsilon. That's the key point. I've gotten away from reporting values for reduced chi-square (RCS). That's a personal choice, and is not in accord with the International X-Ray Absorption Society's Error Reporting Recommendation, available here: http://ixs.iit.edu/subcommittee_reports/sc/ I think the difficulty in choosing epsilon is more likely to make a reduced chi-square number confusing than enlightening. But I am moving increasingly toward reporting changes in reduced chi-square between fits on the same data, and applying Hamilton's test to determine if improvements are statistically significant. Would setting a value for epsilon allow comparisons across different k-ranges, different (but similar) data sets, or a combination of the two using the chi-square parameter? Maybe not. After all, the epsilon should be different for different k- ranges, as your signal to noise ratio probably changes as a function of k. Using the same epsilon doesn't reflect that. In playing around with different windows and dk values my fit variables generally stayed within the error bars, but the size of the error bars could change more than a factor 2. Does this mean that it would make sense to find a window/dk that seems to "work" for a given group of data and stay consistent when analyzing that data group? The fact that your variables stay within the error bars is good news. The change in the size of the error bars may be related to a less-than- ideal value for dk you may have used for the Kaiser-Bessel window. But yes, find a window and dk combination that seems to work well and then stay consistent for that analysis. Unless the data is particularly problematic, I'd prefer making a reasoned choice before beginning to fit and then sticking with it; a posteriori choices for that kind of thing make me a little nervous. * * * At the risk of being redundant, four quick examples. Example 1: You change the range of R values in the Fourier transform over which you are fitting a data set. For this example, RCS is a valuable statistic for letting you know whether the fit supports the change in R-range. Example 2: You change the range of k values over which you are fitting your data. For this example, comparing RCS is unlikely to be useful. You are likely trying different k-ranges because you are suspicious about some of the data at the extremes of your range. Including or excluding that data likely implies epsilon should be changed, but by how much? Thus the unreliability of comparing RCS in this case. Example 3: You change constraints on a fit on the same data range. For this example, comparing RCS is very useful. Example 4: You compare fits on the same data range, with the same model, on two different data sets which were collected during the same synchrotron run under similar conditions. For this example, proceed with caution. You may decide to trust Ifeffit's method for estimating epsilon, or you may be able to come up with your own (perhaps basing it on the size of the edge jumps). Hopefully issues like glitches and high-frequency jitter are nearly the same for both samples, which gives you a fighting chance of making reasonable estimates of epsilon. Done with a little care, there may be value in comparing RCS for this kind of case. --Scott Calvin Sarah Lawrence College ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Athena Artemis
Hi George, I've confirmed the behavior you describe on a Mac running OS 10.5.8, Artemis 0.8.014, Ifeffit 1.2.12, and Athena 0.8.061. It does seem to be a bug or a corrupted file, but I can't figure out what it's actually doing. It's not grabbing the wrong data from the Athena project--none of the other data sets look that close. It sort of seems like it imports a version with a different background subtraction (maybe a different background k-weight, for instance) when you import from the larger project. --Scott Calvin Sarah Lawrence College On May 12, 2011, at 11:18 AM, George Sterbinsky wrote: Also, for anyone who just wants to see the difference in the two data sets I've attached a figure. George On Thu, May 12, 2011 at 1:23 PM, George Sterbinsky <mailto:georgesterbin...@u.northwestern.edu>> wrote: Hello, I've noticed an odd behavior in Athena and Artemis and I was hoping someone could explain it to me. I've attached an Artemis file Data3.apj and an Athena file Data4.prj. First open Data3.apj. Then from the Artemis file menu choose open, then open Data4.prj and import the Data4.xmu file. Now plot the data in k-space you will see a slight difference between the two spectra, most noticeably above 15 k, so plotting in k^3 is best to see the difference. Now close Artemis and don't save. Open Data4.prj with athena and choose to import only the file Data4.xmu. Save the project as something else. I saved as Data4C.prj, which I have also attached. Close Athena and don't save. Open Data3.apj again. Then from the Artemis file menu choose open, then open Data4C.prj or whatever you may have named the file and import the Data4.xmu file, which should now be the only data file in the project. Now plot the data in k- space and you will find that the two data sets are now the same. The differences at high k are no longer present. Can anyone explain what is going on here? Thank you, George ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Question about transform windows and statistical parameters
Matt, On May 13, 2011, at 8:39 AM, Matt Newville wrote: After all, the epsilon should be different for different k-ranges, as your signal to noise ratio probably changes as a function of k. Using the same epsilon doesn't reflect that.Without seeing the data in question, this seems like speculation to me. I'm not at all sure why epsilon (the variation in chi(k)) should depend strongly on the k-range. In my experience, it usually does not. The S/N ratio will surely change with k, but that would surely be dominated by the rapid decay in |chi(k)|, rather than a change in epsilon.I'm confused. We Fourier transform k-weighted data. Since Ifeffit uses the high-R amplitude to estimate uncertainty, it seems to me that what matters is signal-to-noise, not just noise in the original unweighted chi(k). Am I wrong in that? I may be misunderstanding how epsilon_r is calculated. And epsilon_r is the relevant epsilon for a fit in R space, right?I think your assumption that epsilon will depend strongly on k may not correct. Do you have evidence for this? I would say that it is not strongly dependent on k, and that reduced chi-square is useful in comparing fits with different k-ranges.I just tried it on the FeC2O4 chi(k) attached to this post. It's a good example of data where it's not immediately clear to me what the "best" value for kmax is, so it would be tempting to use RCS to compare fits over different k-ranges. I used k-weight 3, and Hanning windows with dk = 1. I chose kmin as 2 and stepped kmax by 0.5, recording epsilon_r for each:kmax epsilon_r7 0.034840105 7.5 0.041843848 8 0.082627337 8.5 0.0875503679 0.0860320079.5 0.08599621610 0.088679339 10.5 0.090364699 11 0.092509939 11.5 0.108103081 There's a general trend of increasing epsilon_r with an increase in k. There's also a jump of a factor of 2 between 7.5 and 8. Why? Because there's a glitch there, and the glitch adds high-R structure.To make sure there wasn't something odd about this particular chi(k), I took one of the data sets included with the horae distribution: the file y300.chi in the ybco folder.I followed the same procedure as before, except I stepped by 1 inverse angstrom each time, because of the greater data range.kmax epsilon_r7 0.0128661258 0.0733836959 0.07825577210 0.08001604011 0.09163457212 0.10541947313 0.16434170114 0.19526695715 0.22472759316 0.41113988217 0.480293296If anything, the trend is more clear here.I find it confusing that you expect the noise in the data to depend (strongly, even) on k, but not on R. The general wisdom is the estimate of epsilon from the high-R components is too low, suggesting that the R dependence is significant. Every time I've looked, I come to the conclusion that noise in data is either consistent with "white" or so small as to be difficult to measure. I believe Corwin Booth's recent work supports the conventional wisdom that epsilon decreases with R, but I don't recall it suggesting a significant k dependence.I'm not making any claims as to whether, in general, the noise in the data depends on R. I can speculate about circumstances where low R noise is greater (due, for instance, to temperature fluctuations in cooling water, which are likely to be fairly slow), or where high R noise is greater (an example here would be if whatever system is keeping the beam on the sample vertically as the mono scans is tending to overshoot).But Ifeffit's estimation of epsilon_r demonstrably does not depend on the R-range used for fitting, regardless of the distribution of noise in R. That's a very different thing. Thus, changing the R-range of a fit is completely safe as far as comparing RCS goes.--Scott Calvin Sarah Lawrence College FeC2O4.chi Description: Binary data y300.chi Description: Binary data ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Question about shift in E0
Hi Brandon, I don't find this terribly surprising if your samples are mixtures of a metal and a metal oxide. If you look at the derivative of the XANES spectra of a pure transition metal and its oxides, you'll generally find that the biggest difference is not in the position of the derivative peaks, but in their relative size; for the metal the lower energy derivative peak is much larger than it is for the oxide. A mixture of oxide and metal, therefore, will be primarily evident in XANES through the relative heights of these peaks; sometimes the apparent shift can be very subtle indeed. For example, consider what happens if the first peak in the first derivative is 20 times lower in the oxide than in the metal, and is also 1.5 eV higher in energy. If you have one sample which is 70% metal, and you compare it to a sample which is 40% metal, the tiny contribution of the oxide to the first peak will hardly be visible in either case, and you might see no measurable shift in energy in the XANES for that peak. (You'll see a big change in its amplitude, though.) Likewise, the white line may be dominated by the oxide, and show little shift in the XANES (but once again, a big change in amplitude). An EXAFS fit, however, doesn't suffer from weighting issues in the same way, and may indeed show an E0 shift when the mixture of oxidation states changes. For my description to apply to your system, though, I'd expect you to be seeing substantial changes in the amplitudes of XANES features, even if their position doesn't shift much. Is that the case for your data? --Scott Calvin Sarah Lawrence College On Jun 5, 2011, at 11:56 PM, Brandon Reese wrote: Hi all, I am looking at EXAFS of thin film metal oxides. I am varying both metal content and the oxygen content of the films. I aligned the scans with a metal reference foil collected simultaneously. In Artemis, I have noticed that when changing between films with no extra oxygen versus those with extra oxygen there is a shift in the fitted E0 of ~1.5 eV (after aligning to the foil). I tried setting the E0 in Athena to the peak of the 1st derivative and the peak of the white line with the same result (~7 eV difference). I was a little surprised by the offset because in Athena the E0 values varied by <0.5 eV. I am not sure if the argument could be made that this shift is a result in a changing oxidation state because it doesn't show up in the XANES (at least qualitatively). Are there other experimental effects that could cause a shift like this, or is this likely something real in my material? If anyone want to see a representative group of data, let me know. ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
[Ifeffit] McMaster correction
Hi all, I've been pondering the McMaster correction recently. My understanding is that it is a correction because while chi(k) is defined relative to the embedded-atom background mu_o(E), we almost always extract it from our data by normalizing by the edge step. Since mu_o(E) drops gradually above the edge, the normalization procedure results in oscillations that are too small well above edge, which the McMaster correction then compensates for. It's also my understanding that this correction is the same whether the data is measured in absorption or fluorescence, because in this context mu_o(E) refers only to absorption due to the edge of interest, which is a characteristic of the atom in its local environment and is thus independent of measurement mode. So here's my question: why is existing software structured so that we have to put this factor in by hand? Feff, for instance, could simply define chi(k) consistently with the usual procedure, so that it was normalized by the edge step rather than mu_o(E). A card could be set to turn that off if a user desired. Alternatively, a correction could be done to the experimental data by Athena, or automatically within the fitting procedure by Ifeffit. Of course, having more than one of those options could cause trouble, just as the ability to put sigma2 into a feff calculation and in to Ifeffit sometimes does now. But wouldn't it make sense to have it available (perhaps even the default) at one of those stages? --Scott Calvin Sarah Lawrence College___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] McMaster correction
Thanks, all! Here's what I got out of the discussion: FEFF is calculating the "correct" chi(k), and applying an approximate correction introduces additional sources of error. But the only way to measure chi(k) is to extract it from unnormalized data, and the original definition of chi was an arbitrary, if sensible, one: chi(E) = mu(E)/mu_o(E) - 1. And mu_o(E), while not known with great accuracy, depends only on the element and the edge (perhaps excepting minor contributions from AXAFS). Not applying a correction, whether McMaster or something more accurate (such as the ones Anatoly and John suggested), is equivalent to using the approximation mu_o(E) = mu_o(E_o), which is less accurate than the alternatives. On the other hand, the effect is almost entirely a shift in the absolute (as opposed to relative) value of sigma^2. Considering that, it seems to me that this would be a good option for Athena when calculating chi(k). (I think it would be more problematic to apply when calculating normalized energy-space data, as in that case the correction would depend on instrumental effects and the absorption of other edges in the sample.) So, Bruce, I guess this was first a discussion and then a feature request. :) --Scott Calvin Sarah Lawrence College On Jun 17, 2011, at 4:14 AM, John J. Rehr wrote: Hi Scott et al., Thanks for bringing up this issue. Whether or not McMaster corrections are useful does seem to depend on details of the measurement. But my question is: for the cases where they are useful, can one do better? As the data & theory get better and better, perhaps we should try to extract more accurate cross sections mu(E). For example, is it at all of interest to have embedded atom cross-sections to replace the atomic based Cromer-Liberman cross sections or empirical tables? John On Thu, 16 Jun 2011, Scott Calvin wrote: Hi all, I've been pondering the McMaster correction recently. My understanding is that it is a correction because while chi(k) is defined relative to the embedded-atom background mu_o(E), we almost always extract it from our data by normalizing by the edge step. Since mu_o(E) drops gradually above the edge, the normalization procedure results in oscillations that are too small well above edge, which the McMaster correction then compensates for. It's also my understanding that this correction is the same whether the data is measured in absorption or fluorescence, because in this context mu_o(E) refers only to absorption due to the edge of interest, which is a characteristic of the atom in its local environment and is thus independent of measurement mode. So here's my question: why is existing software structured so that we have to put this factor in by hand? Feff, for instance, could simply define chi(k) consistently with the usual procedure, so that it was normalized by the edge step rather than mu_o(E). A card could be set to turn that off if a user desired. Alternatively, a correction could be done to the experimental data by Athena, or automatically within the fitting procedure by Ifeffit. Of course, having more than one of those options could cause trouble, just as the ability to put sigma2 into a feff calculation and in to Ifeffit sometimes does now. But wouldn't it make sense to have it available (perhaps even the default) at one of those stages? --Scott Calvin Sarah Lawrence College ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] McMaster correction
I apologize if I have abused the list while working on my text. I will find a different channel for raising these questions and requests once the current discussion is complete. --Scott On Jun 16, 2011, at 8:11 PM, Matt Newville wrote: Hope that helps.I have to admit I'm a little uneasy with the frequency of "I've been pondering..." discussion topics alternating with requests to review book chapters, and find myself being more cautious in my response than I would if someone was actually asking a question. On the other hand, I don't think anyone would object if you added a button to Athena that normalized the data in a way that included a correction for the expected decay of mu(E). ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] SS and MS contributions to EXAFS
Hi Francisco, Following up on Matt's comment, a FEFF calculation with default option assigns sigma2 to be 0, and that probably won't result in relative contributions between SS and MS that look much like what happens in the real material. A better strategy would be to read the calculation into Artemis, as Matt suggests, and use a Debye model for the sigma2 for the paths, trying a few different Debye temperatures to get a sense of how it affects the relative contributions. While the Debye model is not an appropriate model for quantitatively fitting most materials, at least it takes a stab at how sigma2 might depend on R, which will help with understanding how important the MS paths are to chi(k). --Scott Calvin Sarah Lawrence College On Jun 17, 2011, at 7:21 PM, Matt Newville wrote: Hi Francisco, On Fri, Jun 17, 2011 at 7:34 PM, Francisco Garcia wrote: Dear users, I would like to quantify the single scattering (SS) and multiple scattering (MS) contributions to the EXAFS spectra over a range of k values. I adopted the following approach and I would like ask experienced users if my approach is sound: (1) Run a regular EXAFS (default NLEG=8); I assume this includes all possible scattering paths. Call the chi(k) data chi.dat (2) Run another EXAFS but this time I set NLEG=2. I assume this is for all SS contributions. Call the chi(k) data chi_ss.dat (3) To obtain the MS contribution, I subtracted chi_ss.dat from chi.dat If my approach is faulty, can you tell me how to remedy it? Thank you. That approach would work under the assumption that the atomic coordinates in the feff.inp file fully described the distribution of atoms in the system. If, on the other hand, the feff.inp had an idealized structure, then the approach would model the MS contributions for that structure. For example, a feff.inp file often assumes no static or thermal disorder in the system. In that case, the MS contributions would have no disorder terms (eg, sigma^2) applied to them. An equivalent approach that may be somewhat simpler would be to use Artemis to run the Feff calculation with NLEG=8 (4 is probably good enough below 6Ang in most structures), read in all the paths, and sum the paths with NLEG>2. --Matt ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] E0 issues
Hi Hana, On Jun 27, 2011, at 2:04 PM, Hana wrote: I guess my question is obvious to most of you, but after some practice and reading, the following is still unclear to me (hopefully some other beginners will benefit): Normally I have calibrated my spectra using a calibration foil – so energy shift was done according to that companion standard. Doing a simple linear combination fit, I have set all spectra to the same E0 value, somewhere on the edge, above the first derivative maxima (was it a good practice?). That is a good practice, in my opinion. If you're going to do a linear combination fit, it doesn't really matter how you choose E0, as long as: 1) The spectra are all aligned on the same energy scale 2) The choice of E0 is in some way consistent By using a calibration foil, you have assured #1, and then by simply making E0 the same for all spectra, you have assured #2. Recently, I found myself actually confused about E0; since my samples inherently have a phase difference (which is the base to my ability to differentiate them), how a certain reference point on the spectrum can be determined? "Reference point" is indeed the correct term. As such, it is somewhat arbitrary, and just needs to be consistent for all spectra being compared. What can be done when I do not have the calibration foil (especially for these heavy elements that do not have specific sharp feature)? Consistency is the only requirement. There are many ways to align consistently. If you've got really noisy reference data for some reason, you could even fit some kind of function to the edge and use that, but for reference data that's not usually necessary. And further, now that I am starting to work on the structural model; how actually IFEFFIT determined the energy shift when there is no specific reference point? If you float E0 when fitting to feff files, the reported value is a shift relative to wherever you initially picked it. Thus if you change the initial choice of E0 by 2 eV upward, the shift ifeffit reports should be 2 eV less. Hope that helps... --Scott Calvin Sarah Lawrence College ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] E0 issues
Hi again, Thanks Scott, I think it is a mistake done by many beginners. To be confident that I understood you well: so basically the reproducibility of E0 between samples while fitting to FEFF files mainly depends on the quality of my calibration. For reproducibility, yes. Actually, if there is no clear feature in my edges and they are different in ‘slopes’, I must have either a calibration foil, or a known standard, measured at the same time; otherwise there is no good way to get comparable data? Some beamlines are more stable in energy than others. Sometimes it suffices to measure a standard occasionally between measuring data. If the calibration does not drift, or drifts in a predictable way, then you're OK. But if the calibration jumps around by an eV or two, as is not uncommon, then your data inherits that uncertainty unless you measure a reference material at the same time. Is that right, or the fitting, considering all spectra’s components will finally lead to a good fit, even with relatively poor calibration (I will definitely be more careful in the future; but asking for a set of samples that were mistakenly measured for me without a standard). This is a different question. If you are fitting to FEFF, then it doesn't matter if the E0's were defined consistently for each sample, or if the calibration drifted. You will most likely float E0 as a free parameter anyway, so the fitting process will adjust for the differences. You do lose the ability to compare the E0's of your different samples, and you lose the ability to constrain them to be the same, but the rest of the fitting process is fine. --Scott Calvin Sarah Lawrence College P.S. Also note that if ifeffit returns an E0 shift of more than about 10 eV, that's a warning sign. Check if that would correspond to an E0 still on or near the rising portion of the edge (a bit past the white line is still OK). If it's not, then the fit is not a good one. If it is, then it's best to choose a new E0 in Athena (or SixPack) that is closer to to where ifeffit wants it; FEFF loses accuracy when the E0 has to be shifted by that much. ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Opening SSRL data in Athena
Actually, Athena can open binary SSRL files--I just successfully opened one of Lia's files in Athena. Here's what I did: Under Settings, choose Plugin registry. Check the SSRLB box. From there, it opened fine. Does trying that work for you, Lia? I am using Athena 0.8.061 on a Mac running OS 10.5.8. --Scott Calvin Sarah Lawrence College On Jul 7, 2011, at 6:20 PM, Wayne W Lukens Jr wrote: Hi Lia, Those are binary files from SSRL, which athena cannot open. There are a few ways that you can deal these files. Easiest is to use Six-Pack to work up the data. I believe it will open SSRL binary files. Six-Pack is a GUI-driven suite of programs analogous to Athena and Artemis. Slightly less easy is to convert your binary data to ascii data, which can be done on the SSRL computer that you used to collect the data. At this point, you will have to ask the beamline scientist to do this for you. You can definitely open the ascii files in Six-Pack. Somewhat more difficult is to use EXAFSPAK to work up the data. These are a suite of programs for EXAFS data analysis that are not particularly easy to learn. They will open the binary files. EXAFSPAK is easy to use once you have learned the commands. Six-Pack: http://home.comcast.net/~sam_webb/sixpack.html EXAFSPAK: http://ssrl.slac.stanford.edu/exafspak.html Sincerely, Wayne ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Question on GDS values: amp
Hi Nic, The delr and ss values may adopt reasonable values with a negative amp, but they're almost certainly not the right values. A negative amp turns the chi(k) upside-down. To make it ft, the other parameters then have to shift the graph over by half an oscillation, yielding values of E0 and delR that are wrong. It is unusual that a first-shell fit to a standard doesn't seem to be giving a qualitatively reasonable result with an amp of 1. What happens if you do a sum of the most important paths without a fit? Does the fit look qualitatively correct over a large range of R? (The amplitudes will of course be substantially off without a fit.)It sounds like something substantial is wrong: either the material isn't actually dadolinium oxide, or the structure you're using for the feff calculation isn't right (is there more than one crystallographic setting?), or there was serious distortion in the measurement (perhaps a thick, concentrated standard was measured in fluorescence without correction). --Scott Calvin Sarah Lawrence College On Aug 1, 2011, at 12:36 AM, nicholas@csiro.au wrote: Dear All, Just wondering what could I do to make the amp guess value as a positive number? If I run it and let it float (i.e. guess), the amp becomes negative in the resulting fits, but the fit has nice delr and ss values (i.e. make sense). If I restrict the amp value to 1, everything else doesn’t fit. I am only fitting the first nearest neighbour in a measured standard; Gadolinium oxide. Regards, Nic ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Question on GDS values: amp
Hi Nic, On Aug 2, 2011, at 6:41 AM, nicholas@csiro.au wrote: Just a sidenote, is the general workflow for fitting XAFS data the following: Fit first shell and get reasonable Enot and Amp and then make set them, then incrementally add more of the scatter paths and adjust the delR for each path correspondingly. Adjusting degeneracy for atom assuming there are slight differences in the atoms spacing. I know the question is very trivial but I don't seem to be able to find a general guideline for knowing when the fitting process is over in these type of analysis. As for your side note, there are several workflows that have success. The most appropriate one depends both on your level of knowledge about the material and your personal preference. I don't particularly favor the one you describe, though, because it's an attempt to fit a few parameters at a time to avoid wrestling with correlations. That doesn't end up actually avoiding correlations; it just locks them in. For instance, if you first guess amp, find a "best fit" value, set it, and then run a fit with something else varied (say, a degeneracy), then you've artificially broken the correlation between amp and degeneracy. But you've done it in a completely arbitrary way...you haven't really explored the space of the two varied jointly. (Of course, if you're only doing a single shell, you can't vary both S02 and N, because they correlate completely. But you certainly can't pretend you can by first varying S02 with N fixed, and then varying N with S02 fixed!) The two most prevalent valid fitting strategies I've seen are: "Bottom up." (That's my name for it. I've also heard it called "shell by shell.") In this strategy, you start with a single shell with few constraints; perhaps you guess N (taking S02 from a standard), E0, delR, and ss. You try different nearest-neighbors and see what works best. As you begin to gain knowledge about the system, you add more distant shells and begin to add reasonable constraints. In a biological system where you've determined nearest neighbors are sulfur, for instance, then your knowledge of the particular system may suggest what ligands are present, which might provide information about second nearest-neighbors. For instance, the number of second nearest-neighbor carbons might be equal to the number of near-neighbor sulfurs, constraining some of the degeneracies. "Top down." Start with a highly-constrained, multi-shell fit. For instance, you might include all important paths out to 5 angstroms, with the only free parameters S02. E0, and a sigma2 parameter or two. If the fit appears qualitatively close, constraints can then be relaxed to more realistically model the particular features of the material (as one example, you can vary degeneracies to allow for vacancies or nanoscale effects). If the fit does not appear qualitatively close initially, it is probably the wrong starting material. (Amplitudes are often far off initially with this approach, and there may be small phase shifts, but the first few big peaks should be roughly in the right place. A minor variation on this approach is to start with a sum of paths rather than fit at all.) Note that both strategies end up in the same place: a modestly constrained multi-shell fit. (And no, that's not always possible-- sometimes first shell information is all you can get.) When do you use each? If your material is a modestly modified version of a known crystal, top down is a good way to go. For instance, you might have a doped zinc oxide. You know the structure of zinc oxide; the question is the effect of the dopant. Why waste effort trying to fit just a nearest-neighbor first, which can be quite difficult, when you think you know the rough structure out to several angstroms? On the other hand, for something like a protein you probably don't initially know much about the structure at all. Then bottom up works better. Many problems, such as some environmental problems, fall in between the two, and either approach might be effective. A strategy that I've sometimes seen used by beginners in workshops, which is not a good idea, is to fit one shell, get the results, set the values of those parameters to the result of the fit, add another shell, and so on. This is a misunderstanding of the bottom up approach! Rather than using information from outer shells to achieve better fits on the inner shells (or vice-versa), you lock in distortions and make it difficult to evaluate statistically-related measures such as error bars. Hopefully that helps! --Scott Calvin Sarah Lawrence College___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Fe K-edge second shell problems
Hi Paul, Looking at your data, I agree that there does seem to be second-shell scattering present in the signal. Unfortunately, that most likely means that your material is not close to one of the "obvious" model compounds. One possibility to keep in mind is that you may have some kind of overlapping paths in that second "shell"--for example, partially Fe-Na and partially Fe-Ca, or an assortment of Fe-Na's at different distances. A clue can perhaps be obtained by noting the relative height of the peak near 2.3 angstroms compared with the large peak you've fit. As k- weight is raised from 0 to 1 to 2 to 3, the peak at 2.3 angstroms does not grow relative to the first peak. That suggests the scattering may be from another low-Z element like oxygen. So I'd tentatively try an Fe-O path around 2.7 angstroms with its own delR, ss, and N guessed. (Meanwhile, set N_1 to 4 to reduce correlations--you've said you expect the first shell to be tetrahedral.) Good luck--sounds like a stubborn one! --Scott Calvin Sarah Lawrence College On Aug 3, 2011, at 9:49 AM, Paul A Bingham wrote: Dear Ifeffit users, I have been struggling with this problem on and off for many months and I cannot resolve it - hopefully someone out there can help I have collected fluorescence Fe K-edge EXAFS of oxide glasses doped with low levels (0.2%) of Fe. The glasses are typified by their major components SiO2, Na2O, CaO and also low levels of Fe2O3 and CeO2 dopants. I'm currently trying to fit the Fe EXAFS. The first shell is relatively easy to fit and I'm reasonably happy with the fit I obtained using a tetrahedral Fe3+ standard, in this case FePO4. The fits are consistent, as I expected to find, with Fe3+ tetrahedrally-coordinated with four oxygens. The problem comes - and here's where I could really use some suggestions - when I try to fit second Fe-x distance. It seems clear to me that a second Fe-x distance (and possibly a third) are present in the data. However, despite expending a great deal of time I am unable to get a fit that appears anywhere near sensible and robust and for which the output parameters are sensible. I suspect the second Fe-x distance (I reckon about 2.8 Angstroms) to be Fe-Na but Fe-Ca, Fe-Si or Fe-O may also be possible. It's also possible that it is Fe-Fe or Fe-Ce. I have tried all of the "obvious" Fe model compounds (aegirine, clinopyroxine, etc) and also many others and I simply cannot get anything approaching a decent fit. The vast majority of distances in model compounds are Fe-O distances around 1.9-2.1A, then there is usually a "gap" until about 3.1A. I have checked my background subtraction and tried out many different options, changes and tweaks that I know or can find suggested but I cannot obtain a fit that is any good. And so I ask my colleagues out there who are more experienced than I with EXAFS - can anyone help with this conundrum? I have attached the Artemis file with the data and simple one-shell fit using FePO4 cif file; and the Athena file FYI. Thanks in advance for your time and I look forward very much to reading any suggestions you may have. Warm Regards Paul Bingham -- Dr. Paul A. Bingham Immobilisation Science Laboratory Dept. of Engineering Materials University of Sheffield Mappin Street Sheffield S1 3JD UK Email: p.a.bing...@sheffield.ac.uk Direct Line: (0114) 2225473 < Bingham_Fe_EXAFS_Glass_Ifeffit_Artemis > ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] asking questions effectively (yes, *you* need to read this email)
One addition to Bruce's appeal: for some subscribers to the list, large attachments are a problem. For instance, some people are still working at dial-up speeds (due to the US' rural digital divide, I was one of those recently), or even have limits on the amount of data they can download in a month. The question, then, is what is a "large" attachment? We've had some discussion of that on the list previously, and never arrived at a hard- and-fast rule. Nonetheless, let me suggest that anything below 1 megabyte is fine--in fact, it should be encouraged so that we can help with the kind of questions Bruce just enumerated. Paul's files, for instance, were 214 KB, or 0.24 MB. I suggest, therefore, that if you have a project file that is large because, for instance, it has many, many fits in its history, please re-save it in a smaller version, and attach that. You should also be careful with screenshots that they are not needlessly large--e.g. saved in a resolution far beyond what is necessary. In the occasional case that the problem or question requires a large file to manifest, such as that described by Nirawat yesterday, some other arrangement needs to be worked out. It's possible, for instance, to use a service such as Dropbox to make the file available without actually attaching it to an email. --Scott Calvin Sarah Lawrence College On Aug 5, 2011, at 3:25 PM, Bruce Ravel wrote: Hi everyone, This has been a particularly troubling week for me here on the Ifeffit mailing list. This week we have seen an unusually large number of poorly asked questions. Not bad questions, mind you, just questions that have been asked in a manner that makes it hard to provide a useful response. On Tuesday, someone had a question about a fit in Artemis, but only posted the project file which demonstrated the problem after being prompted to do so. On Wednesday, someone had an issue about LCF fitting in Athena that is contrary to most people's experience with the program. That person did not bother to provide an example project file or any other supporting information to clarify what happened. On Thursday, another person had an Artemis problem which was described in a short and cryptic email. Only after being prompted 3 times to post an example was someone able to be of help. Also on Thursday, we saw the third example in one week of a problem with Artemis, but no example project file to demonstrate the problem. Today, we see someone with a crystallography problem, but we do not see the actual data that would allow someone to reproduce the problem on their own computer. Happily, on Wednesday Paul Bingham posted a clear question and attached Athena and Artemis project files. He very quickly got two useful answers. You do see the lesson here, don't you? If your problem cannot be reproduced on someone else's computer, it is unlikely that you will get a satisfying answer. Don't wait to be prodded. Supply the project file or crystal data that demonstrates the problem *in your first email*. The so-called experts on this list, including me, really do want to help you with your problems. But we are not mind readers. You have to meet us half way. B ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Athena: problems with LCF
Hi Nina, Thank you for bringing your question to the list. As others have said, the problem lies with not having enough data to normalize effectively, and is compounded by not taking control of the normalization process to make sure its relatively consistent. Standard A, for instance, has a post-edge line that cuts very low through the data, as does D. The 1:1 A to D calculated mixture, though, comes across more evenly. This is a problem even with calculated mixtures, as you chose to import the calculated mixture as a mu(E) file, and not as a norm(E) file. In theory, this should be fixable by fixing the edge step of the calculated spectrum to 1, so that it doesn't try to normalize it again. Something's not quite right when I do that, because the calculated average of A and D sometimes slips below both of the other graphs, and an average shouldn't do that. Nevertheless, it gets us close: an LCF fit now gives us 54.5% A and the rest D. Another approach is to remove the "force weights to 1" checkbox. This is often necessary if normalizations are in doubt. That works quite well here, delivering an A to B ratio of 49 to 53. Summary: --It's best to collect data far enough above the edge so that you establish an unambiguous post-edge trendline, if possible. --Post-edge lines should be examined, and changed if they are inconsistent between spectra being used for LCF. It's better to just eyeball normalization than to use radically different trendlines, for instance. I sometimes play around by eye with trendlines to see what range of normalizations they give, and incorporate that in to my final reported uncertainties. --If normalizations are difficult for a particular set of spectra, it is often better to remove the requirement that weights sum to 1. To the degree that normalizations are off, there will be some error in the values that are found, but at least the fitting routine is able to try to compensate for normalization differences by adjusting the weights. In other words, forcing the sum to be 1 when the normalizations are different forces a bad fit. Allowing them to total to anything allows the algorithm to transfer errors in normalization to errors in weighting. Hope that helps! --Scott Calvin Sarah Lawrence College On Aug 15, 2011, at 5:40 AM, Nina Siebers wrote: Dear All, I acquired Cd L3-edge spectra of some binary and ternary mixtures in varying proportions and for the individual components. The mixtures were created on Cd-mass basis. Then, I tried to fit the reference spectra to the spectra of the mixtures using linear combination fitting of Athena to get their abundance. However, the results were disappointing despite all spectra were carefully energy calibrated and normalized, so I decided to create simple mathematical binary and ternary mixtures by summing up the spectra of the individual reference spectra. After that I did an edge-step normalization in excel and imported the normalized calculated mixtures into Athena. Then, I tried the fitting again to exclude mixing-failures and check sensitivity of LCF with the idealized spectra. Even though the results of the LCF of the mathematical mixtures were better compared to the real mixtures, LCF was also not able to reliable deconvolute these spectra into the individual reference spectra. Does anybody have an explanation for that? It would be nice if somebody could give me information about the mathematical fitting algorithm implemented in Athena. Attached is a data file of three mixtures (two ternary and one binary mixture) including the mathematical mixture created in excel (named calculated at the end). Mixing ratios are named 1to1to1 (meaning 1:1:1 of the components in the same order). For the 1:1:1 ternary mathematical mixture the deconvolution was very good, but the others need improvement. I hope I made my problem clear this time. Thanks a lot! Wishes, Nina ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
[Ifeffit] Looking for reviewers for data reduction chapter
Hi all, Once again, I have a draft of a chapter of my textbook ready for review. This chapter is on data reduction, including normalization, background subtraction, and Fourier transforms. This will be Chapter 4 in the book, so it should be suitable for near-beginners as well as those with more experience. This is also a chapter where I had to make a lot of decisions on nomenclature and presentation, so experts might be interested as well, particularly if you have strong opinions on that kind of thing! It is a bit more than 40 pages long, which makes it the longest so far. If you are interested in reviewing the chapter and can get feedback to me by 29 September, please contact me directly at scal...@slc.edu . As always, repeat reviewers are welcome! --Scott Calvin Sarah Lawrence College ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Negative Sigma^2
It's probably not initial guesses, Robert. Look at the uncertainties, and look at correlations. If you are getting, say, sigma2 = -0.001 +/- 0.004 A^2, then the values aren't even wrong, exactly, as the error bars are consistent with a reasonable value. But sigma2 is not well-determined. If, on the other hand, the error bars are such that the value is unambiguously negative, it's handy to see what correlates strongly with the sigma2 in question, as that can give you some clues. Sometimes, a negative sigma2 is indicative of an incomplete model, rather than one that is flat out wrong. Perhaps, for instance, there are paths within the fitting range that contribute to the signal, but you have not included them in your model. Of course, it could also be that the model is simply wrong. Without more details from you, it is impossible to say what the problem is in your particular case. --Scott Calvin Sarah Lawrence College On Sep 13, 2011, at 1:32 PM, Palomino, Robert wrote: I am trying to fit data I recently collected and occasionally I am getting negative sigma squared values. Could anyone tell me what this is indicative of: am I using the wrong model or are my initial guesses of some parameters way off? Robert ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
Re: [Ifeffit] Spline range?
Wei, You may not have looked at it yet, but the chapter draft I sent to you has several pages devoted to the topic. --Scott Calvin Sarah Lawrence College On Sep 18, 2011, at 10:10 AM, Wei Li wrote: Dear all, I am wondering about what range should be chosen in Athena? Some literature says from 1 to the end; some says from 0.5 to the end. Wei -- Wei Li Postdoc researcher Environmental Soil Chemistry Group Delaware Environmental Institute University of Delaware, Newark,19713 Tel:631-949-0663 http://ag.udel.edu/soilchem/li.html<http://ag.udel.edu/soilchem/> ___ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit