Improving the Accuracy of Particle Analysis

Published

July 1, 2000

Author(s)

John A. Small, J R. Michael, Dale E. Newbury

Abstract

Historically the procedures for the quantitative X-ray analysis of particles in the electron probe have been similar to the methods used for bulk electron probe samples.The main difference was that corrections had to be made to the experimental k-ratios or calculated compositions to compensate for particle geometry and often these particle corrections were simply added onto an existing bulk analysis procedure. In general these particle procedures improve significantly the analytical accuracy and reduce the relative uncertainty distributions from several tens of percent for uncorrected data to about 10% for corrected data. [1] Although a significant improvement, arelative uncertainty of 10% is often not sufficient to distinguish between many materials with similar compositions. Recently we have investigated two methods of analyzing particles that may improve the analytical accuracy associated with particle analysis and our capabilities to correctly identifying a particle's composition. The first method we investigated was to lower the excitation potential of the electron beam below the conventional range of 15-25 keV. To study the effects of reducing the accelerating voltage on particle analysis, we analyzed irregularly shaped glass shards ranging in size from about 1 to 24 m at beam energies of 25, 22, 17, 15, and 10 keV. The shards were analyzed in an electron probe employing energy dispersive x-ray spectrometry. Quantitative analysis was performed with a conventional bulk-sample ZAF correction algorithm. The relative errors for all elements were calculated for each particle as a function of beam energy and are plotted in Fig. 1. The error distributions are for quantitative results without any normalization or other corrections for particle effects. The results indicate that the beam energies greater than 20 keV the error distribution is large, ranging from a high of +60% to a low of 50%. In contrast, the error distributions for beam energies less than 20 keV drops dramatically with the smallest values, ranging from only + 13% to -8%, occurring at 10 keV. Based on these results we hope to develop a low-voltage particle analysis procedure that further incorporates particle corrections to provide accuracies approaching those for bulk-polished samples, {{less then or equal to} 5% relative. The second method we investigated to identify particle composition was the phase identification of micrometer- and submicrometer-sized crystalline particles employing electron backscatter diffraction (EBSD). Unlike conventional EBSD samples, which consist of infinitely thick targets with flat surfaces, individual particles present a unique set of problems for phase identification because of their shapes and limited mass. The results of this study indicate that phase identification is possible on submicrometer particles as shown in Fib 2. Phase identificaiton based on EBSD when used in conjunction with an analytical SEM or EPMA provides the analyst with a very powerful and straightforward method for identifying the composition of submicrometer and larger crystalline particles.

Proceedings Title

Microbeam Analysis 2000 | | Proceedings Institute of Physics Conference Series | IOP Publishing Ltd.

Volume

165

Conference Dates

July 9-13, 2000

Conference Title

European Microbeam Analysis Society

Pub Type

Conferences

Keywords

electron backscatter diffraction, electron probe, low voltage analysis, particle analysis, scanning electron micr

Citation

Small, J. , Michael, J. and Newbury, D. (2000), Improving the Accuracy of Particle Analysis, Microbeam Analysis 2000 | | Proceedings Institute of Physics Conference Series | IOP Publishing Ltd. (Accessed November 9, 2024)