Our Portfolio: Characterization Projects

CMST Index | Characterization | DDFA

X-Ray Fluorescence Metal Analyzer

Information
Resources

 

Tech ID: 2001
Project Overview
 The X-Ray Fluorescence Metal Analyzer, the Alloy Expert (ALEX), is a portable, battery-operated, x-ray fluorescence metal analyzer. The ALEX uses a proprietary algorithm developed at the Naval Research Laboratory; with push-button operation, it determines common alloy elements in the field with accuracy reaching laboratory results. The instrument was initially designed to analyze the percent of alloy content in scrap metal. Other applications range from inventory control to incoming inspection to counterfeit metal identification.
Technology Description
The analysis of an XRF spectrum has two parts--decomposition of the spectrum into intensities for the various elements, and subsequent calculation of elemental composition of the unknown via calibration of the intensities and correction for matrix effects. The algorithm for the spectra decomposition was completed and tested. The algorithm views a spectrum as a superposition of one or more pure element spectra whose shapes are known plus backgrounds with known profiles. We apply a least squares mathematical technique to a set of data sampled from this spectrum to extract (approximate) elemental intensities. We treat the background the same as the component spectra. The partial derivatives with respect to the coefficients of the known pure element spectra and backgrounds are summed over the sampled unknown spectrum and set to zero, giving a linear least squares solution. This can be reformulated as a matrix equation for simplicity and solved using Gauss-Jordan elimination.

One of the advantages of this algorithm is that we can use variable-sized sampling intervals, i.e., large intervals in areas where the spectral profiles are not changing. This can be used to reduce the amount of data processing and increase the insensitivity to distortion of lines, etc. The algorithm was tested with generated data including noise and easily reproduces the original spectral composition within better than 1%, even with distorted peak widths and considerable noise. The algorithm is sensitive to the energy calibration, so an improved version that takes advantage of recalibrations is being written. The output of the algorithm is fractional intensities of pure elements, taking into account both alpha and beta lines and normalized to pure element intensities. These results are already calibrated and give a good first approximation to the unknown sample composition, uncorrected for matrix effects (absorption or enhancement).

Return to Top