A program for quick and rigorous XANES analysis for Windows 3 Mar 2006

Improved convergence for the fitting procedure. Now the fitting is much faster!

• Author: K.V.Klementiev
• Address: HASYLAB / DESY
                 Notkestrasse 85,
                 D-22607 Hamburg, Germany
• E-mail Address: nospam@desy.de (BUT replace "nospam" by "Konstantin.Klementiev" ; this is done to short-cut scanners for spam mail)
  Since May 2006 I have been working in ALBA/CELLS in Barcelona. If someone tried to reach me by the old address at DESY, please forgive my silence: the address is not active any longer. You can ask your questions here.
If you suppose that some analysing steps should be changed or a new procedure is worth to be added, please contact me.
• Abstract:

• Energy calibration (using the reference curve or can be set manually).
• Deconvolution of absorption coefficient with monochromator resolution curve and/or core-hole lifetime deconvolution.
• Transformation to a new equidistant grid
• Pre-edge background subtraction (a polynomial with chosen powers)
• Correction by a user-defined function (for example, to account for the fluorescence self-absorption). The function may refer to the theoretical tables of absorption coefficients incorporated into the program.
• Normalization to unity
• Base line subtraction (useful for analysis of pre-edge peaks). The base line is a spline drawn through manually adjusted knots or a smoothing spline drawn through chosen points on μ.
• Fitting by a user-defined formula, any parameter of which is either a fitting value or another loaded spectrum (experimental or calculated one) whose energy shift can be varied. The model curve can be convoluted with an appropriate broadening function. All fit parameters can be fixed and/or constrained. Advanced error analysis is developed and applied: optionally, the fitting uncertainties can be found with neglected pair correlations, with maximum pair correlations, or via Bayesian analysis (with all pair correlations). Contour maps for χ² function can be drawn for each pair of parameters. Statistical χ²- and F- tests can be performed at user request.
• Principal component analysis or target transformation. If a spectrum is one of the spectra forming the vector space dimensionality of which is sought for, this spectrum is reconstructed with a chosen number of principal components. If a spectrum does not belong to this space, the target transformation of this spectrum is performed.
  
  Any time, all curves and their changes under processing are visual. You can see also I₀ and I₁ currents and derivatives of μ (with possible smoothing).
  
  For a pair of spectra, their difference can be taken (for XMCD). Also, for all spectra loaded, the average spectrum and the rms deviation spectrum can be calculated.
  
  Data file structure is tunable and flexible, no previous conversion needed to load multiple column files from different EXAFS stations. User should specify the file header and how to interpret and use the columns of data (or functions of these columns). Once described, the data formats are then recognized automatically. Several data files can be loaded and processed simultaneously (the best way for this is to load one spectrum, chose appropriate settings and to load all other spectra; the settings for these will be cloned). Or just drag and drop your files onto the open program or the icon on your desktop.

• Screenshots:
  
  A series of CuZn-based catalysts under reduction treatment
  
  
  Example of the pre-edge peak processing (this example is among example XANDa Project files which you can open in the program)
  
  
  Example of fitting procedure (this example is among example XANDa Project files which you can open in the program)
  This is a Ce L3 XANES (spectrum #2) fitted by a weighted sum of monovalent 3+ and 4+ contributions. The "3+" cpectrum was calculated by FEFF (spectrum #1, black). The "4+" contribution is represented by the same calculated spectrum shifted by a fitting value to higher energies. Thus the experimental spectrum is fitted by the formula 'a+b*c', where 'a' and 'b' are the shifted spectrum #1, 'c' is a scalar. The average valence is given by (3*1+4*c)/(1+c)
  
  
  Example of Principal Component Analysis (this example is among example XANDa Project files which you can open in the program)
  Here are 6 spectra. How many of them are independent?
  After unselecting the 5th and the 6th component the test spectra #3 and #4 deviate from the experimental ones considerably. Thus, among the 6 spectra there are 5 independent ones (5 principal components).
  


• Documentation:    None. Detailed manual is planned to prepare. Meanwhile, you may send me questions via e-mail.
• Examples: provided in a separate zip-file (see the download page) as XANDa Project files (*.xpj). These files can be associated with XANDa (then you can double-click on them to open), or can be opened from the XANDa 'Load project' dialog.
• Supported Platforms:    Windows NT/2000/XP, Windows 95/98/ME
• References:
Fitting errors based on Bayesian analysis; statistical tests : 
K.V.Klementev, Statistical evaluations in fitting problems, J. Sychrotron Rad., 8, 270-2 (2001) (Proc. of Int. Conf. XAFS XI), and
K.V.Klementev, XAFS analysis. II. Statistical evaluations in the fitting problems, e-arXiv: physics/0003087
Bayesian deconvolution (instrumental or life-time): 
K.V.Klementev, Deconvolution problems in x-ray absorption fine structure spectroscopy, J. Phys. D: Appl. Phys. 34, 2241-7 (2001)
Principal component analysis: 
S.R.Wasserman, The analysis of mixtures: application of principal component analysis to XAS spectra, J. Phys. IV, 7, C2-203-5 (1997) (Proc. of Int. Conf. XAFS IX)



• Citation:   You are requested to cite XANES dactyloscope as: "K. V. Klementiev, XANES dactyloscope for Windows, freeware: www.desy.de/~klmn/xanda.html"