Features of GUPIXWIN

(updated June 2019 prior to release of GUPIX-3)

Spectral Data are accepted in Guelph's ASCII format. .

The Ion Beam may be:

  • Protons (up to 5.5MeV)
  • Deuterons (up to 6MeV)
  • Helium-4 (up to 12MeV)
  • Helium-3 (up to 12MeV)
  • High energy protons (up to 100MeV - for cyclotron work)

X-ray Detector may be Si(Li), SDD or HPGe.

Non-Linear Least Squares Fit of Marquardt type, modified to prevent premature elimination of weak peaks. The five parameters of a quadratic channel-versus-energy and a linear (peak width)2-versus-energy system calibration are variables, in addition to the height of each element's principal peak. The quadratic calibration is for extreme cases only.

Peak Description can be Gaussian or Voigtian with low energy tailing: in Voigtian case, Gaussian resolution is convoluted with natural Lorentzian line profile; tailing details can be modified for specific detector. Pile-up by Johansson "pile-up element" (including both double and triple pile-up) or by auto-convolution which describes also peak+continuum pileup. Escape peaks are included; in the Si(Li) case the escape peak is asymmetric, reflecting the Si KL satellites.

Background is removed by top-hat digital filtering ; no analytical model needed; the user can choose among filter options that optimize different aspects. 

X-ray Absorbers (Filters) are specified via atomic number thickness and hole fraction (funny filter). 

Specimens may be thin, intermediate, thick, or layered.

Matrix Corrections: There are 2 alternatives.  The Fixed-Matrix approach is for analysis of minor and trace elements in non-thin specimens where the major element concentrations are known a priori (i.e. the matrix is pre-defined). In cases where the matrix is not initially defined i.e. no concentrations ae known a priori, the Iterative-Matrix approach is used..

Elements may be designated as:

  • on the specimen surface
  • in specimen
  • in defined layers of a multi-layer specimen
  • in detector window or filter

and matrix corrections will be appropriate in each case. In analysis of minerals, concentrations may be specified as elements or oxides in input/output. 

Atomic Physics Database includes:

  • Dirac-Fock K X-ray intensities including quadrupole lines and radiative Auger satellites; DF L X-ray intensities; DHS M X-ray intensities.
  • Zeigler stopping powers: TRIM website 1997
  • For protons, a choice of either:
    - Parameterized ECPSSR/DHS proton ionization cross-sections of Chen and Crasemann for K, L and M subshells: Bambynek K fluorescence yields; theoretical DHS fluorescence and CK yields for the L and M subshells. 
    OR 
    - Parameterized proton reference ionization cross-sections: Paul for K-shell: variant of Orlic for L subshells: coupled to Krause CK and fluorescence yields
  • For helium-3, helium-4 and deuterons, parameterized reference K-shell cross-sections from Paul; parameterized MECPSSR/hydrogenic cross-sections for L and M subshells; 
  • XCOM 1987 attenuation coefficients in tabular form. A code is supplied with which user can alter selected attenuation coefficients. 
  • Scofield PE cross-sections for secondary fluorescence.
  • Subprogram GUCSA provides access to data base and computes cross-sections, stopping powers, attenuation coefficients, filter attenuation, etc. 

Principal X-ray Line defaults to K-alpha, K-alpha-1 or L-alpha or L-alpha-1, M-alpha or M-alpha-1, but operator may designate K-beta, L-beta, L-gamma etc, if this appears more favourable as regards detection limits.

X-ray Yields can be computed for thin and thick targets, including full secondary fluorescence for thick targets (GUYLS).

Spectrum Fitting provides tables of peak areas, two error recipes, detection limits, target depth for defined fractional X-ray yield and corresponding ion beam energy. Plots of measured and fitted spectra, residues, data minus fit, etc.

Thin or Thick Target Calibration by the Guelph H-value method. The instrumental constant H is essentially the solid angle of the system, but it also includes any normalization of the measured equivalent charge to real charge. H should be determined through measurements with known samples, pure element targets or standard reference materials. Slight differences in measured concentrations can arise from imperfections in the database or detector description and H can be corrected accordingly. H can even exhibit a dependence upon X-ray energy, and this is catered for by GUPIXWIN. 

Concentrations of Trace Elements are provided in ppm wt. or ng/cm2 if H is specified and matrix elements are defined (Z, CONCENTRATION). Limits of detection (LODs) are also provided for elements observed, and also approx. LODs given for elements not present. Secondary fluorescence by the X-rays of matrix elements is automatically accounted for. Peak areas are augmented for pile-up losses before conversion to concentrations. 

Matrix (Major, Minor) Element Concentrations available by iterative solution that has secondary fluorescence by matrix elements automatically corrected for. One light element that is known to be present but is "invisible" in the spectrum, e.g., S or O, may be included in the matrix using the fact that elements sum to 100%. Similarly, invisible complexes such as CO2 in carbonate or SiO4 in zircon, may be handled.

  • In the case of a multi-layer specimen, the matrix element concentrations in each layer plus the layer thickness are iterated, the constraint of 100% total concentration being used to determine layer thickness. There are two options: iteration may be done until either the layer concentration reaches 100% or until the substrate concentration reaches 100%. 

K-alpha - K-beta Decoupling Option may be used in fit for elements subject to secondary fluorescence or for characteristic lines from critical filter/absorber or for PIXE lines heavily attenuated by absorber.

Weighting Schemes: Various uncertainties, e.g., absorber transmission, K-beta/K-alpha ratio, may be used to augment the conventional chi squared weighting .

Output options include both graphical options and csv files. One csv file option provides concentrations, percent errors, detection limits and decisions as to presence/absence of each element. A second option provides peak areas, percent errors, area limits of detection, and presence/absence decisions.

Batch Fitting can be used to handle large numbers of similar spectra in batch mode. It is available for PIXE using one detector and also for two-detector PIXE where one detector records the trace element spectrum and the other records the major element spectrum

Tests: GUPIX has been tested successfully on various IAEA standards, NIST standard alloys, and synthetic mineral standards, and in numerous comparisons between micro-PIXE, EPMA, and other methods. Nevertheless, it is the user's responsibility to establish GUPIX's accuracy in her or his particular context. The University of Guelph and the Guelph PIXE Group accept no responsibility for the accuracy of analytical results that are generated using GUPIX.

Documentation: An extensive manual accompanies GUPIXWIN. Purchasers should familiarize themselves with this documentation prior to running GUPIX.