Introduction to the analysis of EXAFS data

Transcription

1 Introduction to the analysis of EXAFS data Carlo Meneghini

2 Grado2013_XAFS_TUTORIAL PDF slides: Complete directory XAFS_tutorial_Grado2013 Data Examples Manuals Software Grado_2013_XAFS_Tutorial.pdf a-ge A-GeO2 Cu Cu 10K Fe_align W LCA IFEFFIT VESTA Notepad++

3 Characteristics of a XAS spectrum µt Post edge atomic background Jump Near Edge Region (XANES) Edge Energy Pre-edge background

4 XAFS ANALYSIS: from the experimental data to results Data collection Preliminary data treatment Structural model(s) revision Extraction of XAFS structural signal: χ(k) revision Structural refinement Check the results revision END

5 XANES/EXAFS Fluo/TEY/transmission. Data collection Choose properly the experimental set-up & sample preparation Proposal submission + proposal evaluation + beamtime scheduling = 6 to 12 months! It is difficult to obtain new beamtime for the same proposal in case of failure!!!! Check the data quality during the experiment Quality is better than quantity Optimize your beamtime! Always measure reference data XAFS rarely gives absolute quantitative details Choose properly data collection strategy - Check the proposal submission deadlines, - discuss your experiment with local contacts, - check your data during the experiment ELETTRA: ESRF:

6 Data collection Errors in XAFS data analysis are dominated by systematic errors: - spurious structures - background subtraction - wrong models - inaccurate theory a Experimental set-up & sample preparation Transmission: massive concentrated samples inhomogeneities, holes, not parallel surfaces, etc... Fluorescence: thin concentrated thick diluted samples Self absorption, detector linearity, Bragg reflections TEY: thick concentrated Surface sensitivity, sample charge, Bragg reflections d Measure reference samples 1-energy calibration 2-data analysis procedures calibration The power of XAFS technique consists in highlighting the smallest structural differences c IXS Standards and Criteria Subcommittee Reports Data quality b Collection strategy S/N ratio - step scan / quick EXAFS glitches - Fixed Energy or K steps discontinuities - single scan/repeated scans - temperature points -...

7 Evaluate S/N ratio α(e) J α(e) High degree polynomial E (ev) E (ev) σ = 1.1e α(e) S/N ~ J/σ E (ev) Check regularly during the experiment

8 Check systematically during the experiment α(e) 1.2 Glitches E (ev) edge shift Discontinuities I 2 Ref.Sample I 1 Sample I 0 I ref =ln I 1 /I 2 I exp =ln I o /I 1

9 Preliminary data treatment a Select properly the best spectra and useful data regions don't use blue one! Absorption signal Transmission Fluorescence & TEY µx<<1 Check carefully the data quality during the experiment Experimental set-up generally provide macros for automatic data collection over several days: supervise your macros! b c Preliminary data treatment - Self absorption corrections -deglitching -data interpolation - data alignement Sum together the best spectra Weighting by S/N

10 Extract χ(k) structural signal Transmission Absorption signal Fluorescence & TEY revise preliminary tratments pre-edge subtraction α= α O + α pre α O = (µ O +µ)t α o calculation µ o = free atom absorption µ = embedded atom absorption pre = all but the absorber XAFS structural signal structural signal χ(k) Fourier Transform Fourier Filtering Structural refinement

11 a Background subtraction pre = α(e) - pre pre = ae+b b calculate µ o and χ(k) c Fourier Transform

12 Structural signal Define E o Calculate k k Note: Multiple excitations my affect the background with (small) discontinuities Define µ o Calculate χ(k) E o Check FT µ o = represents the bare atom absorption. It is calculated empirically as a smooth curve across the data. Different XAFS data analysis software apply different (generally equivalent) methods A. Di Cicco et Al. Phys. Rev. B 62, (2000) Large FT contributions at low (unphysical) distances may signify "wrong µ o " Requirements for µ o : 1) Smooth enough to not remove true structural features 2) Structured enough to remove background structures

13 k 3 χ(k) Fourier Transform k 3 χ(k) Shape of FT widely changes as a function of: FT window k w weight data range kχ(k) kχ(k) FT shows more intuitively the main structural features in the real space: the FT modulus represent a pseudoradial distribution function modified by the effct of amplitude, phase and mean free path parameters: FT peaks point out interatomic correlations Peak positions are not the true distances due to the photoelectron phase shift

14 k 3 χ(k) Fourier Filtering Fourier filtering allows isolating contributions of selected regions of the FT FT Background contribution Back Fourier transform (BFT)

15 data refinement Theoretical χ(k) Choose a model Exp. χ(k) Select the relevant structural contributions Experimental χ(k) Refine the structural parameters: N, R, σ 2 exp fit Y Y add new contributions? n Change the model? n Revise your data extraction? Y END

16 Choose a structural model How to find a model structure How to visualize the structure How to calculate distances and geometries Barns.ill.fr Full inorganic structure database (ICSD) (registration fees) Protein Data Base (Free) Search on the web!

17 experimental data Theoretical formula How a data refinement program works Amplitude and phase functions from atomic cluster models ex fit p

18 XAFS data analysis software XAFS organizations e f x i p t GNXAS home-page Amorphous & disordered systems XAFS data analysis FEFF home-page

19 advantages: Freeware, multiplatform (Win, Mac, Linux) easy to handle, semi-automatic procedures Install Ifeffit.exe Ifeffit.exe drawbacks: easy to handle, semi-automatic procedures 1. athena GUI for Data Processing with Ifeffit 2. artemis GUI for XAFS Fitting with Ifeffit 3. hephaestus GUI for general x-ray properties of the elements 4. sixpack GUI for XAFS Processing and Fitting with Ifeffit 5. feff6l Stand-alone program for ab inito EXAFS calculations 6. atoms Stand-alone, command line crystallography->feff.inp 7. autobk Stand-alone background removal program 8. feffit Stand-alone FEFF fitting program 9. ifeffit command-line version of Ifeffit

20 Inside the Ifeffit directory Estract χ(k) data fitting Manuals and tutorials Info on the proposed example Configuration files: NOT FOR NEWBIES Database of crystallographic structures as *.inp files for Atoms

21 Data collection Preliminary data treatment revision Extraction of XAFS structural signal: χ(k) revision Structural refinement Check the results Structural model(s) revision END

22 Preliminary data treatment & Extraction of XAFS structural signal: χ(k) List of files (group) opened into Athena Plot highlighted files Plot (several) marked files Options for plot Warning: check if the full windows is visible on your PC, smal screens (i.e. e-pc) may cut the lower part Warning: check the graphic windows and never close it leaving Athena/Artemis open sometime the PC crashes

23 Example 1: Cu 10 K Open cu_foil_10k.dat into Athena Exercise 1: Look at the results obtained using the automatic extraction playing with plot options

24 Select Eo Takes the value from the cursor on the graph window Exercise 2: Look at the effects of changing background subtraction parameters - Change Eo - Change R bkg -... R cut off for automatic background calculation Rbkg = 0.2 Rbkg = 1 Rbkg = 2 Cut off structural information!!!!

25 Exercise 3: in Real space look at the effects of changing extraction and FT parameters K- space R- space FT of: k kw χ (k)

26 Exercise 4: in q (Back Fourier) space look at the effects of changing BF parameters and regions

27 Exercise 5: Data treatment Shifts the spectrum to have selected point (reference at) at the calibration energy (calibrate to) Experimental data Shifted data

28 Save: cu_foil_10k.dat.chi χ(k) is required for ARTEMIS Note: Arthemis requires equispaced k data!!! (i.e. δk = 0.05) You cannot use χ data if not equispaced

29 Working with several files Go into the Fe_Align directory Open all Fe files into Athena Fe 300 K, scan Fe 60 K, 2scan 1 Fe 60 K, scan 2 Fe 60 K, scan 3 Fe 150 K, scan 1 Fe 150 K, scan 2 Fe 300 K, scan 1 Temperature effect on EXAFS and FT data

30 Exercise 6: align Eo Open Fe files into Athena Eo is shifted among the files!!!

31 Exercise 7: Merge several files Fe 60 are now aligned

32 Full Exercise: 1. Read Fe Files 2. Align Fe files on order to have the same Eo 3. Merge files having the same T 4. Save χ(k) of merged files, to be used for the structural analysis

33 Structural model & Data refinement Cu-fcc metal SPG: fcc, f m 3 m (# 220) a = 3.61 A Cu Sh R N R Cu (A) IV I: a/sqrt(2) II: a I III V III:a. sqrt(1.5) IV: a. sqrt(2) V: a. sqrt(2.5) II... You can visualize the structure using VESTA program icsd_43493_copper.cif

34 Structural model & Data refinement Start the ARTEMIS program The ARTEMIS panel Start the refinement Warning: check the full windows is visible on your PC, smaller screens may cut the lower part Depends on what is highlighted in the Data & Paths panel Controls for plotting Warning: check the graphic windows and never close it leaving Athena/Artemis open sometime the PC crashes

35 Structural model & Data refinement Theory new atoms page - Import Cu.inp file from the Cu_10k directory Or - Blank Atom page (you must type the crystallographic info) Warning: check the full windows is visible on your PC, smaller screens may cut the lower part

36 Define the parameters describing the structure 1 3 Sh R N R Cu (A) I: a/sqrt(2) II: a III: a. sqrt(1.5) IV: a. sqrt(2) V: a. sqrt(2.5) Check the distances and multiplicities are those you expect

37 1 2 3 Theoretical XAFS and FT for Cu fcc

38 Open Cu_foil_10k.inp Compare model and experimental data Using CRTL key you can select groups for plot

39 remove all paths or restart ARTEMIS 1 st shell analysis Right click add 1st shell path to the list select and Right click

40 guess: optimized in the refinement def: math expressions updated during the fit S o 2 Path label Energy shift Distance correction set: numbers or expressions evaluated once at the fit beginning and not updated Debye Waller factor The Data screen allow defining the fitting strategy and parameters

41 ν = N ind -N var The refinement is performed in Real space, on the Real and Imaginary parts of the FT Absolute misfit between experimental data and theory

42 Structural results right click

43 Effects of Fitting parameters Kwt = 1 Kwt = 3

44 Add new contributions IV I III V Visualize the structure and understand the neighbour shells! II SS Take care about multiple scattering contributions Single Scattering MS 3-legs MS 3-legs MS 4-legs MS + Focusing

45 Understand the MS contributions!!! SS paths MS paths Focusing effect may raises some MS contributions well above the SS one

46 In principle the same for each shell In principle different for each shell Warning: The used path must be inside the fit range

47

48

49

50

51 it's your work now! a. Fe bcc b. GeO c....