Progress further in this web-site constitutes acceptance of the terms of our copyright and disclaimer.
CSIRO Science in Motion GeoPIXE Features
Recent feature additions to GeoPIXE ™ ...

GeoPIXE enquiries ...

PIXE and SXRF Spectrum Fitting

The user interface provides interactive selection of elements in the fit and tools for X-ray line identification. There is no inbuilt limit on the number of elements or X-ray lines that GeoPIXE can handle. Silicon and germanium detectors, and detector arrays, are catered for.

See C.G. Ryan, D.R. Cousens, S.H. Sie and W.L. Griffin, "Quantitative Analysis of PIXE Spectra in Geoscience Applications", Nucl. Instr. Meth. B49 (1990), 271-276 for details of the fitting algorithm. This method has been developed over the past decade, and tried and proven in interlaboratory and blind industry tests (e.g. Federowich et al., Can. Min. 33(1995) 469-480, show GeoPIXE accuracy of better than 3% in a published blind test).

The PIXE/SXRF yield calculation includes secondary fluorescence and is able to treat complex multilayered targets. A special feature of the calculation is that the yields of ALL X-ray lines are integrated through the layered structure so that the X-ray relative intensities accurately reflect the full target structure. This gives much better spectrum fits for complex samples.

The fit uses the SNIP background treatment (C.G. Ryan, E. Clayton, W.L. Griffin, S.H. Sie and D.R. Cousens, "SNIP, a Statistics Sensitive Background Treatment for the Quantitative Analysis of PIXE Spectra in Geoscience Applications", Nucl. Instr. Meth. B34 (1988), 396-402). Background shape can also accommodate the effects of detector efficiency and filter absorption.


GeoPIXE: spectrum fitting

GeoPIXE: Quantitative imaging

The Dynamic Analysis Method

The PIXE/SXRF spectrum fit generates a Dynamic Analysis (DA) transform matrix. This enables the projection of EVT data (list-mode, or event-by-event data) directly onto quantitative elemental images. This process resolves element overlaps, strongly rejecting artifacts from overlapping elements and detector response effects (escape peaks, tails) and subtracts background. Pileups are well treated in many cases which involve a single dominant element or majors in fixed ratios. (A more general approach is also available in GeoPIXE.)

The results are quantitative images in ppm.charge or ppm.flux units. Concentration variance images are also accumulated (at half resolution to minimize memory usage), so that error estimates and detection limits can be provided for all extracted concentration values and line profile projections, etc.

The DA method is detailed elsewhere (C.G. Ryan, "Quantitative Trace Element Imaging using PIXE and the Nuclear Microprobe", International Journal of Imaging Systems and Technology (Special issue on Quantitative Imaging) 11, (2000) 219-230; C.G. Ryan, B.E. Etschmann, S. Vogt, J. Maser, C.L. Harland, E. van Achterbergh and D. Legnini, (2005), "Nuclear Microprobe – Synchrotron Synergy: Towards Integrated Quantitative Real-time Elemental Imaging using PIXE and SXRF", Nucl. Instr. Meth. B231, 183-188; C.G. Ryan, D.N. Jamieson, C.L. Churms and J.V. Pilcher, "A New Method for On-Line True Elemental Imaging using PIXE and the Proton Microprobe", Nucl. Instr. Meth. B104 (1995), 157-165; C.G. Ryan, "Developments in Dynamic Analysis for Quantitative PIXE True Elemental Imaging", Nucl. Instr. Meth. B181 (2001) 170-179; C.G. Ryan and D.N. Jamieson, Nucl. Instr. Meth. B77 (1993) 203-214).

Extraction of spectra from other ADCs (e.g. non-PIXE/SXRF data) is also supported here. GeoPIXE also supports sorting using energy regions of interest or cuts. Cuts are set-up and named interactively, and can also perform background subtraction during list-mode file processing.

Quantitative Image Exploration

The quantitative images generated using the DA method can be directly interrogated to extract concentration data from within arbitrary "regions", or projected along traverses or line projections. These results include error estimates and detection limits.

"Regions" supported at present include box (sizeable, rotatable, movable rectangle), circle, ellipse, 10-point spline curve, 32-point spline curve, a line projection rectangle (sizeable, rotatable, movable rectangle including shear control) and traverse projection along non-linear spline curves. All can be positioned and moved, or changed in shape or orientation using "handles". If many image windows are opened, or cloned, to display many elements, then these regions will appear in the same place on all images. A subtract mode enables the removal of defects (cracks, pits, scratches, etc.) from the region selection.

Clicking on a button will then extract average concentrations (and errors, MDLs) from within these areas (or projected back to a centre line, in the case of line projection mode). These results are tabulated, and can be exported.


GeoPIXE: profile extraction

Quality Control

Spectra can be extracted from these regions, by sorting the EVT file, by simply pressing a button. Detector array elements can be merged together, mapped onto a common energy calibration. The resulting spectra from these regions appear in a spectrum window, overlaid by the "fit" to each. This “fit” is estimated from the DA deduced concentration data and pure-element spectral signatures (saved along with the DA matrix). This enables an immediate assessment of the quality of the images by looking at the accuracy of the "fit" to the extracted spectra from specific features. This can often show up a rare (concentrated in one spot) element that was missed from the initial inspection of the overall PIXE/SXRF spectrum, for example. Such omissions can be corrected, a new DA matrix generated, and the images re- projected.

GeoPIXE: spatial matrix correction

Spatially Varying Matrix Effects and Pile-up

In general, sample composition varies with position across an image area. This means that the PIXE/SXRF yields (counts/ppm) will also vary. This effect can be corrected in GeoPIXE using a method based combining the yields calculated for end-member components. This method is described elsewhere (C.G. Ryan, "Developments in Dynamic Analysis for Quantitative PIXE True Elemental Imaging", Nucl. Instr. Meth. B181 (2001) 170-179; C.G. Ryan, E. van Achterbergh, C.J. Yeats, T.T. Win and G. Cripps, "Quantitative PIXE trace element imaging of minerals using the new CSIRO-GEMOC Nuclear Microprobe", Nucl. Instr. Meth. B 189 (2002) 400-407). Using this approach even dramatic variation in sample composition can be accounted for to produce quantitative images.

Another effect that arises from the spatial variation in sample composition is differential absorption of X-rays from neighbouring pixels of different composition. PIXE/SXRF yields implicitly treat each pixel independently. However, a pixel with a neighbouring pixel towards the detector of lower mean atomic number, for example, will produce a higher effective yield. This is a geometric effect related to the direction and take-off angle for the detector. This effect can be corrected in the latest versions of GeoPIXE. The effect is described elsewhere (C.G. Ryan, E. van Achterbergh, G. Mark, C.J. Yeats, S.L. Drieberg, B.M. McInnes, T.T. Win, G. Cripps and G.F. Suter, "Quantitative, High Sensitivity, High Resolution, Nuclear Microprobe Imaging of Fluids, Melts and Minerals", Nucl. Instr. Meth. B188 (2002) 18-27).

GeoPIXE also enables images to be corrected for the effects of a spatially varying pile-up spectrum. The nature of pile-up, in terms of dominant sources of pile-up and the relative intensities of the 'pile-up element' lines, can vary widely with position in an image area. This firstly makes it difficult using normal approaches to fit spectra from image areas. GeoPIXE can fit pile-up components in spectra taking account of the spatially varying nature of the pile-up spectrum. It can then use this information to remove pile-up related effects from images.

Element Correlations and Overlays

Another aspect of image data exploration is looking at the distribution of chemical phases in a sample, phase relationships and modal analysis. GeoPIXE facilitates this work with two new window classes. Correlation windows plot two element 2D histograms of the distribution in pixel composition. This shows distinct phases or solid-solution trends (as shown opposite between Fe and Cu; see also "Electron microprobe mapping as a tool in ilmenite characterisation" by Colin MacRae, Nick Wilson and Mark Pownceby, Microscopy and Microanalysis 7 (Suppl 2: Proceedings) p. 710, Microscopy Society of America 2001, LongBeach meeting). Features observed in the correlations can be surrounded by a spline curve, and the corresponding pixels in the element images selected (see green areas opposite; original image is below). This is useful for modal analysis. Alternatively, regions of images can be selected and just those pixels plotted in Correlation. This can distinguish fine complex intergrowths.
GeoPIXE: RGB, 3 element image overlays Spatial correlation between elemental images can also be viewed using Image RGB windows. These form 24-bit colour images using three elements to set the intensity of the Red, Green and Blue colour components. As with other GeoPIXE windows, these windows are linked to the Image Display windows and respond to setting of image intensity scales for each element.

The example opposite shows a composite image of As (Red), Au (Green) and Cu (Blue) spatial distribution in a sulfide ore.


GeoPIXE: element correlations and highlighting

GeoPIXE: image display window

List-mode Formats and Platforms Supported

A new version of the GeoPIXE ™ software package has been released. It adds user source for Object oriented Device modules and a user guide for writing and modifying device-specific device drivers to customize quantitative PIXE and SXRF image. Example code for 23 devices is presently available.

These include EVT files written by Unix and Linux versions of MPsys (MARC, Melbourne University), MDA data-cube files from sector 2 and 20 (PNC-CAT) of the Advanced Photon Source, MCA pixel spectra and HDF formats (GSE-CARS, APS; NSLS), CSIRO Maia large detector array formats, OM-DAQ event files (Oxford Microbeams), MPA3 list-mode files (FAST ComTec), Sparrow KMax Macintosh data aquisition (Lund PIXE group), XSYS data acquisition on VAX (iThemba Labs, South Africa), Sandia BD12 list-mode data (Primecore U48) and Labo list-mode (Tohoku University, Japan).

GeoPIXE ™ is written in the IDL ™ language (Exelis Visual Information Solutions), which provides good graphics, GUI element, image processing and vector and matrix arithmetic support, and a level of platform independence. The current release of GeoPIXE runs using IDL 7.1 - 8.1 runtime, under the Windows (32, 64 bit), Linux (32, 64 bit) and Mac X (unsupported; 32, 64 bit) operating systems.

For more details or information on availability, or to discuss list-mode and spectra formats, you are welcome to send an email.

Return to the main GeoPIXE page.
Go to the CSIRO-GEMOC Nuclear Microprobe web page.
Exelis Visual Information Solutions

For further information contact: Dr. Chris Ryan via email: (Chris.Ryan@csiro.au)
Phone +61-3-9545 2090
Fax orders +61-7-3327 4455
CSIRO Earth Science and Resource Engineering, c/o CMSE, Gate 5, Normanby Road, Clayton VIC 3168, Australia
CSIRO Australia

CSIRO Earth Science and Resource Engineering
CSIRO © 2001-2012


Please direct comments about this web page to Chris Ryan.
Amiga Graphics for these pages have been prepared using Amiga rendering and image processing software.