Misidentification of Major Constituents by AutomaticQualitative Energy Dispersive X-ray Microanalysis:A Problem that Threatens the Credibilityof the Analytical CommunityDale E. Newbury*Surface and Microanalysis Science Division, National Institute of Standards and Technology,Gaithersburg, MD 20899-8370, USAAbstract: Automatic qualitative analysis for peak identification is a standard feature of virtually all moderncomputer-aided analysis software for energy dispersive X-ray spectrometry with electron excitation. Testing ofrecently installed systems from four different manufacturers has revealed the occasional occurrence of misiden-tification of peaks of major constituents whose concentrations exceeded 0.1 mass fraction ~10 wt%!. Testmaterials where peak identification failures were observed included ZnS, KBr, FeS2, tantalum-niobium alloy,NIST Standard Reference Material 482 ~copper–gold alloy!,Bi2Te3, uranium–rhodium alloys, platinum–chromium alloy, GaAs, and GaP. These misidentifications of major constituents were exacerbated when theincident beam energy was 10 keV or lower, which restricted or excluded the excitation of the high photonenergy K- and L-shell X-rays where multiple peaks, for example, Ka~K-L2,3!–Kb~K-M2,3!;La~L3-M4,5!–Lb~L2-M4!–Lg~L2-N4!, are well resolved and amenable to identification with high confidence. Thesemisidentifications are so severe as to properly qualify as blunders that present a serious challenge to thecredibility of this critical analytical technique. Systematic testing of a peak identification system with a suite ofdiverse materials can reveal the specific elements and X-ray peaks where failures are likely to occur.Key words: analytical electron microscopy, automated analysis, energy dispersive X-ray spectrometry, peakidentification, qualitative analysis, scanning electron microscopy, transmission electron microscopy, X-raymicroanalysisINTR ODUCTIONEnergy dispersive X-ray spectrometry ~EDS! performed inthe scanning electron microscope ~SEM! and the transmis-sion electron microscope ~TEM!/analytical electron micro-scope ~AEM! has assumed a dominant role in the arsenal ofmicrostructural characterization tools used across a widerange of the physical and biological sciences, engineeringand technology, and forensic applications ~Williams & Carter,1996; Goldstein et al., 2003!. Historically, manufacturers ofEDS systems have been among the most aggressive develop-ers of computer-based analysis systems to aid the EDS userin achieving meaningful results with a high degree of effi-ciency. Software systems provide active control over spect ralacquisition to provide proper corrections for paralyzabledeadtime, to minimize the effects of pulse coincidence, tomonitor the amplifier gain that defines the energy axiscalibration, and to maintain a quality measurement environ-ment. Software tools are generally provided to perform peakidentification ~qualitative analysis! and to convert measuredcharacteristic X-ray intensities into equivalent concentra-tions ~quantitative analysis! either by standardization andmatrix correction or by the “standardless” method ~Gold-stein et al., 2003!. It is now becoming common practice forthe software system to directly prepare a report, so that asthe analyst selects locations for analysis on the specimen,the software automatically makes peak assignments to spe-cific elements and interprets the measured peak intensitiesas concentrations that are presented with associated mea-surement statistics. The final, automatically prepared reportof results is often divorced from the basic measurement, theEDS spectr um, which can only be recovered from the origi-nal archived data on the EDS host computer. This increas-ing tendency toward a comprehensive, fully automated EDSanalytical procedure is perhaps inevitable given the extraor-dinary advances in computing power now available. How-ever, one serious co nsequence of having such comprehensi vesoftware is the effect of divorcing the analyst from thedetails of the EDS measurement and interpretation process.For novice analysts, this trend results in a lack of challengeReceived October 20, 2004; accepted March 1, 2005.*E-mail: [email protected]. Microanal. 11, 545–561, 2005DOI: 10.1017/S1431927605050531MicroscopyANDMicroanalysis© MICROSCOPY SOCIETY OF AMERICA 2005to their knowledge and an increasing dependence on thesoftware. Because the novice analyst is no longer required tomanually “solve” a spectrum to identify the peaks, thecritical knowledge base needed for manual qualitative analy-sis is not being acquired, which greatly reduces the likeli-hood that a novice or even a more “experienced” analystwho has trained in such an environment will recognize anincorrect peak identification made by automatic qualitativeanalysis. Moreover, there are users for whom SEM-EDS orAEM-EDS is just one of several or even many analyticaltools that they must use in the course of their work. Suchindividuals often come to depend on the accuracy of acomprehensive computer-aided analysis system that pro-duces finished reports that can be handed directly to aclient.W hat if such an automatic analytical system is occasion-ally utterly wrong in its identification of a major constituent~arbitrarily, concentration . 0.1 mass fraction or 10 wt%!,even if the analytical system is being properly operated?Such a mistake when identifying a major constituent wouldbe so egregious it would completely destroy the value of theanalysis. It is a procedural failure so severe as to constitute ablunder ~Bevington & Robinson, 1992; Taylor, 1997!.Itssubsequent impact on the utilization of the analytical re-sults could be catastrophic, whether it is in the context of aresearch result that supports a new interpretation of natureor evidence presented in a court case. Do such blunders inthe identification of major constituents actually occur inmodern EDS analytical systems? To examine this possibility,several commercial EDS systems of recent manufacturewere tested with a series of known binary materials whereall elements were present as major constituents and theelectron excitation was sufficient that the peaks used foridentification were well above the continuum background.MATERIALS AND METHODSA s eries of specimens was subjected to the automatic quali-tative analysis function of recent EDS analysis systems, eachless than 2 years of age since installation, offered by fourmanufacturers. For obvious
View Full Document
Unlocking...