Volume 107, Number 6, November–December 2002Journal of Research of the National Institute of Standards and Technology[J. Res. Natl. Inst. Stand. Technol. 107, 627–638 (2002)]Sample Preparation for Electron ProbeMicroanalysis—Pushing the LimitsVolume 107 Number 6 November–December 2002Joseph D. Geller andPaul D. EngleGeller MicroAnalytical Laboratory,Topsfield, MA 01983-1216There are two fundamental considerationsin preparing samples for electron probemicroanalysis (EPMA). The first one mayseem obvious, but we often find it isoverlooked. That is, the sample analyzedshould be representative of the populationfrom which it comes. The second is adirect result of the assumptions in thecalculations used to convert x-ray intensityratios, between the sample and standard,to concent rations. Samples originate froma wide range of sources. Dur ing theirjourney to being excited under the electronbeam for the production of x rays thereare many possibilities for sample alteration.Handling can contaminate samples byadding extraneous matter. In preparation,the various abrasives used in sizing thesample by sawing, grinding and polishingcan embed themselves. The mostaccurate composition of a contaminatedsample is, at best, not representative ofthe original sample; it is misleading. Ourlaboratory performs EPMA analysis oncustomer submitted samples and preparesover 250 different calibration standardsincluding pure elements, compounds, alloys,glasses and minerals. This large varietyof samples does not lend itself tomass production techniques, includingautomatic polishing. Our manualpreparation techniques are designedindividually for each sample. The use ofautomated preparation equipment doesnot lend itself to this environment, and isnot included in this manuscript. Thefinal step in quantitative electron probemicroanalysis is the conversion of x-rayintensities ratios, known as the “k -ratios,”to composition (in mass fraction oratomic percent) and/or f ilm thickness. Ofthe many assumptions made in the ZAF(where these letters stand for atomicnumber, absor ption and fluorescence)corrections the localized geometry betweenthe sample and electron beam, or takeoffangle, must be accurately known. Smallangular errors can lead to significanterrors in the final results. The samplepreparation technique then becomes veryimportant, and, under certain conditions,may even be the limiting factor in theanalytical uncertainty budget. This paperconsiders preparing samples to getknown geometries. It will not addressthe analysis of samples with irregular,unprepared surfaces or unknowngeometries.Key words: electron probe microanaly-sis; gr inding; mounting; polishing; quantita-tive analysis; sample preparation; sawing.Accepted: September 20, 2002Available online: http://www.nist.gov/jres1. Sample Collection, Transport, andParticulate MountingEPMA samples originate from a large variety ofsources. They may be powders, corrosion scales on ahost surface, coated substrates or from bulk specimenssuch as metals, ceramics, glasses, and plastic or organicmaterial. In collecting the sample one must be certainthat nothing is added that will later be analyzed andmistaken for the analyte. One area of analysis where thisoften happens is with the analysis of small particles,which have to be collected and transported.The challenge here is to collect the particles andembed them in a media suitable for grinding and polish-ing. The purpose of embedding or mounting is to holdthe particle during preparation so known geometriescan be established between the electron beam and x-raydetector. During analysis small airborne particles, thatmay range in size from sub-micrometer and up, areoften filtered or collected electrostatically. Particlesmay also be suspended in liquid form. How can such627Volume 107, Number 6, November–December 2002Journal of Research of the National Institute of Standards and Technologysmall particles be collected and transported for analysis?Air and liquid filtration may embed these particles intothe depth of three-dimensional filters, preferably singlesurface filters can be used with the particles populatingon a single surface (polycarbonate materials preferreddue to its chemical inertness), but with much reducedconductance resulting in far less volume filtered. Withdepth filtering the filter can be dissolved concentratingthe particles for later surface filtration or embedding.There is a risk, however, that the solvent used will reactwith the analyte by either putting it into solution,corroding the sample, or causing a chemical reactionresulting in a new compound or that impurities in thefilter itself be mistaken for the analyte. Larger particlesmay exist as scale and be simply scraped off a surfacetaking care not to sacrifice the scraping tool in theprocess. For instance, this author has had the unfortu-nate experience of finding tungsten in unknownparticles. With no tungsten used in the process thatgenerated these particles the or igin was later determinedto be a tungsten probe that was used to dislodge theparticle. Unknowingly, a result was produced that hadnothing to do with the sample.Particles may be suspended in oil that were createdwhile an engine or mechanical device was in operation.Concentrating the particles can be done by removingthem from the liquid with a tweezers or fine probe whilebeing observed under a microscope. They can also befiltered, as described above, except care must be takento use solvents that are compatible with the filters andthe particles. Sometimes particles are collected withadhesive tape. While this may be easy for collection, itcan be difficult to completely and cleanly remove them.Adhesive residue on the particles may be very difficultto remove, even with solvents and sonication. If theparticles are embedded with adhesive residue they maymove during grinding and polishing. This results in poorsurface finish or loss of the particle. We prefer using thePost-Itnote type of adhesive paper. While particlebonding may not be as good the adhesive is thinnerand remains on the paper providing less chance forcontamination.Once the particles are collected they are nextprepared for polishing, the purpose of which is toestablish the necessary take-off angle for analysis. Clearembedment media offers the technician the advantage ofbeing able to observe the particles during the prepara-tion process. We have had much success using two-partepoxies (available from the major