• De Geronimo, Gianluigi
• Paterson, David
• Kuczewski, Tony
• Borg, Stacey
• Hough, Rob
• Cleverley, James
• Etschmann, Barbara
• Lintern, Mel
• De Jonge, Martin
• Howard, Daryl
• Davey, Peter
• Siddons, Pete
• Dunn, Paul
• Moorhead, Gareth
• Mcnulty, Ian
• Jensen, Murray
• Eyberger, Catherine
• Lai, Barry

Abstract

Motivated by the need for megapixel high definition trace element imaging of natural material, faster acquisition and improved counting statistics, using a scanned sample and a microfocused X-ray probe, a large energy-dispersive detector array with dedicated processor called Maia has been developed by CSIRO and BNL for SXRF elemental imaging. Following a year using a 96 detector prototype for a wide range of applications at the X-ray Fluorescence Microscopy (XFM) beamline at the Australian Synchrotron, construction of the Maia annular 384 detector array [1,2] has been completed and tests are underway. This paper provides an update on the Maia concept, reports on developments in methods for real-time processing of event data, methods for spectral deconvolution and quantitative imaging using a large detector array with extended solid-angle, a software approach to cope with large data-sets and illustrates Maia performance using applications involving rare fine-scale features in complex geological samples.Maia combines a 96 or 384 silicon detector array and a high speed, FPGA based, pipelined, parallel processor that enables real-time processing of each detected X-ray event tagged by detector number and XY position in the image scan. Using an embedded spectral deconvolution algorithm based on the Dynamic Analysis (DA) method [3], the processor accumulates deconvoluted element images in real-time. The real-time processing pipeline includes: (i) event correction for linearization, (ii) gain trimming to match spectra from individual detector channels, (iii) pile-up rejection, (iv) DA image accumulation, (v) monitor spectra accumulation, (vi) intelligent sub-sampling to conserve disk space, and (vii) logging of raw and accumulated data to disk. The detector systems may drive the XY sample stage directly, or read back stage encoder or laser interferometer values to sample position and focusing zone-plate location. Real-time event processing enables transit times per pixel as short as 50 µs and scanning up to ~10000 lines to collect high definition SXRF images up to ~100M pixels (or 3D fluorescence tomography [4] and XANES imaging data-sets [5]). The result at XFM is spatial detail spanning 4 orders of magnitude from ~1 µm to ~10 mm that can detect rare µm scale features (e.g. sub-µm gold grains at depth in a section) and place them within a broad and detailed spatial context.References: [1] D.P. Siddons et al., AIP Conference Proceedings 705, 953 (2004). [2] R. Kirkham et al., Proc. of SRI 2009, AIP Conference Proceedings, in press. [3] C.G. Ryan, Int. J. of Imaging Systems and Technology 11, 219 (2000). [4] M.D. de Jonge et al., these Proceedings. [5] B.E. Etschmann et al., American Mineralogist, in press.

Topics

• impedance spectroscopy
• grain
• tomography
• gold
• Silicon
• trace element
• fluorescence microscopy