In May of 1955 Professor Shinoda introduced the development of the electron probe microanalyzer (EPMA) in a review article published by Oyobutsuri (the Applied Physics journal written in Japanese), which showed a schematic diagram of an EPMA system. That paper, published in the most popular journal in Japan, was very probably the first paper which drew the attention of the people involved in the field of material science, in particular, the field of metallurgy, who were especially interested in EPMA. That paper stimulated the development of EPMA in Japan.

In 1957, the funding provided by the Grant-in-Aid for Developmental Research to the university research groups headed by Profs. Akutagawa, Gokyuu, and Abe at the University of Tokyo was used, in a cooperation with the Hitachi group headed by Drs. Tadano and Watanabe, to develop the EPMA in Japan. Their research results led to the funding of a 2 year Grant-in-Aid for Cooperative Research project (April 1960 to March 1962) called “The Study of X-ray Microanalyzer”, which was chaired by Prof. Sakaki and consisted of 37 members from various universities, national institutes and industries. Prof. Shinoda, who became the Chairman of that project in April of 1962, led that group for another year until March of 1963.

It was just after the start of the Cooperative Research project, in September of 1960, that Professor Castaing visited Japan and gave several talks as a representative of the French Mission Culturelle. Professor Castaing gave a talk at the Hitachi Central Research Laboratory on the 19th of September, another talk at the Nagoya University on the 24th of September, and also a talk at Osaka University on the 28th of September. During his stay in Japan from September 18th to October 4th he talked about his work on EPMA a total of six times by also giving two informal presentations. The visit of Professor Castaing gave a great push forward for the commercial development of EPMA instruments in Japan.

An x-ray spectrometer is actually a scanning monochromator because two conditions must be satisfied, namely: (1) nλ, = 2d sinθ and (2) θI{= θR to within the rocking curve of the crystal 2Δθ. Before Castaing (B.C.) there were four known types of focussing monochromators, namely: Johann, Johansson, Cauchois, and von Hamos. Castaing was fortunate in having Andre Guinier (well known for his work in x-ray crystallography) as his thesis advisor. Thus, he wisely chose to use a spectrometer of the Johansson type. Had he used a flat crystal spectrometer, the result would have been unsatisfactory and the history of microanalysis would have been significantly different. Many of the spectrometers that were used in the early instruments were either of the Johansson type, in which the crystal planes are curved to a radius of 2R and the surface has a radius of R, the focal circle radius. This is sometimes called exact focussing because rays in the plane of the focal circle converge exactly to a point focus. For simpler construction, the Johann crystal which has the surface parallel to the atomic planes (also curved to a radius of 2R) was sometimes used. While both types provide one-dimensional focussing, the Johansson type is preferred because the effective width of the crystal can be greater.

In addition to the type of geometry used, early instruments developed after Castaing (A.C) also had variations in the mechanism of the crystal movement. Many used a relatively simple mechanism with the crystal and detector moving about a fixed center for the focal circle as shown in Fig. 1A. Some later instruments used curved crystals that were fixed in position, and scanning was accomplished either by rotating the crystal as in Fig. 1B or by a mechanism that flexed the crystal and changed the radius of the focal circle as in Fig. 1C. The most satisfactory arrangement and the one subsequently adopted by most manufacturers of commercial microprobes utilized a linear motion of the crystal carriage with a provision for tilting the crystal so that the focal circle always passed though the x-ray source as shown in Fig. 1D. If the detector moves so that it remains on this focal circle, its motion is along a leaf-shaped curve. With this design it is easy to have a readout proportional to wave length.

EPMA... .it all starts with Bragg’s Law. It’s all about crystals. WDS crystals are really the rate-limiting step to the usefulness of the EPMA technique. As such they represent the focal point of as much research as any other component of the instrument. There is an inherent need to improve the performance of crystals to assure the survival of the technique. As electron columns permit smaller and smaller spot sizes and as the need to analyze smaller and smaller things increases the search for improved crystals intensifies.

There are generally two crystal parameters that beg for improvement: 1) Energy Resolution and 2) Intensity (especially for light elements). The resolution of crystals is considered adequate for most applications but it’s always useful to have more resolution to better deal with overlapping lines. It’s also becoming useful to measure valence states by observing relative shifts in peak positions. Intensity is the thing that one never gets enough of. It seems that more and more applications demand more intensity. Trace element analysis, particle analysis, heat sensitive sample analysis, light element analysis all scream for more intensity.

The Sepia years. 1951. Castaing’s thesis(1) : Genesis of a Non-Destructive and truly Quantitative Microanalysis method. 1958. The beginning of commercial microprobes : Early instruments had no computers and were lacking special analyzing crystals, but overall they were well designed. Modern features we are familiar with today, such as light element analysis, field emission gun, Energy Dispersive analysis, analysis of insulated material and scanning analysis, although not widely implemented, were discussed in scientific reviews even then(2) 1963. One of my early personal experience: The daring job of obtaining Castaing’s acceptance for his Cameca-buih EPMA at the University of Paris/Orsay.

From early models to present microprobes Early microprobes were developed, after WWII, in a world driven by metallurgy. They had few WD spectrometers, usually at low take off angle. However, the need for light element analysis and the fast growing use of EPMA in geology have sent the manufacturers back to the drawing board. Why ? The design of modern microprobes was a compromise between light optics, electron optics and higher take off angle X-ray spectrometry. There were three possible designs of X-ray path geometry : X-rays through the(final) lens, X-rays through the gap of the lens, X-rays outside the lens, hence three suppliers arose based on these concepts.

Present and future EPMA improvements. As in the initial era of EPMA newer applications will point the direction in which the electron microprobe of the future should evolve.

Early Development Automation of electron probe analysis began to flourish in the early 1970s spurred on by advances in computer technology and the availability of operating systems and programming languages that the individual researcher could afford to dedicate to a single instrument. By the end of the decade, most researchers and vendors in the microanalysis field had adopted the PDP-11 minicomputer, and languages such as FOCAL, FORTRAN and BASIC that ran on these computers. A good summary of these early efforts was given by Hatfield. The first use of the energy dispersive detector on the electron probe in 1968 added the need to control the acquisition, display and processing of EDS spectra. As a result, the 70’s were also a time when much attention was focussed on development of software for on-line data reduction and analysis. These efforts produced a suite of programs to provide matrix corrections and spectral processing, and automation of WDS data collection. The culmination of these development efforts was first reported in 1977 with the analysis of a lunar whitlockite mineral by simultaneous EDS/WDS measurement. This analysis determined the concentration of 23 elements, 8 by EDS and took a total of 37 minutes for data collection and analysis. In this paper, the authors noted the complementary use of the EDS and WDS (WDS for trace elements and severe peak overlaps, EDS for other elements and rapid qualitative analysis) in their automated instrument, a convention that remains common on the electron probe even today. Toward the end of the decade the analytical accuracy and precision achieved by automated analysis of bulk samples approached the limits of the instrumentation, with the exception of analysis of light element concentrations.

Two Decades of Improvements The explosive growth in digital electronics and microprocessors for data processing and control functions during the 80’s was rapidly applied to electron probe automation. Second and third generation automation systems included direct control of many microscope functions, beam position and imaging conditions. Motor positioning was more precise and far faster. As a result, the data collection and analysis of 23 elements reported in 1977 could be accomplished at least three times faster on a modern instrument.

A central theme of modern materials science has been the exploration of the relationship between the microstructure of a material and its macroscopic properties. Beginning in the late 19th century, the developing field of metallography permitted scientists to view the microstructure of metal alloys. Mechanical polishing followed by selective chemical etching produced differential relief on chemically distinct phases or at grain boundaries. With such specimens, reflection optical microscopy revealed structures with micrometer and even finer dimensions. The microstructural world that was found proved to be highly complex, and most alloys were observed to be chemically differentiated into two or more distinct phases. The answer to many materials science questions required knowledge of the specific composition of such fine scale phases. Castaing’s research was motivated by these considerations, as evidenced by the title of his first paper, “Application of electron probes to metallographic analysis”. The subsequent impact of Castaing’s electron probe microanalyzer (EPMA) has occurred across a broad range of the physical and biological sciences. Materials science has been one of the most active areas and chief beneficiaries, as well as a source of researchers who not only employed electron beam microanalysis to solve their problems but who also contributed innovations that advanced the microprobe field. For example, the abstracts of the First National Conference on Electron Probe Microanalysis contain numerous examples of advanced applications of the electron microprobe to materials science, including analysis of (1) refractory metal coatings (P. Lubllin and W. Sutkowski), (2) diffusion in the Ti-Nb system (D. Nagel and L. Birks), (3) Au-Al alloys (C. Nealey), (4) various steels (H. Nikkei), (5) Al-V-Mo-Ti alloys (R. Olsen), (6) corrosion of Ni-Co alloy (C. Spengler and R. Stickler), and (7) analysis of metal oxides and carbides (T. Ziebold). Integrated over 50 years, the impact of electron probe microanalysis on materials science has been so broad and varied, especially in its continued development and incorporation within scanning electron microscopy and analytical electron microscopy, that a suitably comprehensive review could consume the entire presentation time available for all topics to be covered in this conference! Instead, selected examples of critical applications will be presented to illustrate the impact in materials science.

Castaing, in his remarkable PhD thesis on the electron microprobe, not only detailed the construction of the prototype of what was to be the first commercial electron microprobe and demonstrated its application as a working tool in metallurgy, geochemistry and other materials sciences. In, addition, he also outlined the physical factors involved in the x-ray emission process and developed the general scheme of data correction for quantitative analysis that is still in use today.

In his thesis, Castaing outlined the nature of the processes involved in electron inelastic scattering leading to inner shell ionization and x-ray production (the so-called “stopping power” correction) and the loss of generated x-ray intensity due to elastic scattering of energetic electrons resulting in their emission from the sample (the so-called “backscatter” correction). He showed that the combination of these factors resulted in a correction primarily dependent upon the sample’s average atomic number (the so-called “atomic number” or “Z” correction), which for a thin specimen or one with minimal x-ray absorption and fluorescence would suffice in relating relative intensity to relative concentration. He noted that, in what is known as “Castaing’s first approximation”, the ratio of emitted x-ray intensity of a given line from element “A” in a specimen to that in a standard, KA = IA,spec/IA,std, can be related functionally to the mass concentration of the analyzed element:

Of the different aspects of electron probe microanalysis(EPMA)which were studied by Castaing during his doctorate the work on characteristic x-ray fluorescence was the most definitive. In his thesis, which was completed in 1951, Castaing established the physical and mathematical framework for a correction procedure for fluorescence which is essentially still used in EPMA today. Much of the effort since then has been in refining and improving the accuracy of the correction and extending the scope of the correction to a wider range of specimen types. The Castaing formula was developed for the case of a K x-ray from element A being excited by a K xray from element B (K-K fluorescence) and in 1965 Reed extended the range of the correction by including the K-L, L-L and L-K interactions. In the same paper Reed also introduced the expression from Green and Cosslett for the calculation of K intensities, which was believed to be more accurate than the expression used by Castaing. The original formula included a somewhat unrealistic exponential term to allow for the depth of the production of the primary x-rays and a number of workers have tried replacing this term with a more accurate expression, however, in general this has led to only small changes in the final correction. Reed also simplified the formula in order to make the calculation easier in the days before fast computers; in particular he replaced the jump ratio variable by two constants, one for the K-shell and one for the L-shell. Much later Heinrich showed that this simplification was no longer necessary and that the jump ratio could in fact be calculated directly.

The development of quantitative X-ray microanalysis in the 1950s spurred the need for knowledge of the many parameters which describe the electron interaction, such as the ionization cross-sections, fluorescent yields, the electron stopping power, mass absorption coefficients, and others. Although classical microanalysis, which proceeds by measurements of the unknown specimen against a standard, can eliminate the need to know many of these parameters accurately, much current microanalysis is done on highly inhomogeneous samples for which comparison with a standard is much less useful procedure. The increased use of low beam energies also means that data is now required for L-, and M-lines which previously have been little used. Consequently there is an enhanced need for a reliable and agreed set of data on which to base calculations.

A common misconception is the belief that all of the quantities that are needed must already have been measured by somebody and so it is simply a matter of accessing this data.

Quantitative analysis with the electron microprobe analyzer (EMPA) has yielded more accurate results over time as a result of improvements in ZAF and other correction routines, mass absorption coefficients, synthetic pseudocrystals for ultralight elements, computers, software programs, backscattered electron (BSE) and energy dispersive (EDS) X-ray detectors. Consequently, many geoscientists view EMPA as routine, and details of procedures, standards, and operating conditions are seldom provided in current publications. However, in overseeing a facility with many users, we have learned that acceptable analytical data are sometimes difficult to obtain even with established analytical procedures and a choice of several hundred standards. After novice users have mastered the routines of sample polishing, cleaning, coating, handling and machine focus, their choice of nonoptimal standards often prevents them from obtaining the most accurate results possible. Optimal analysis for geological problems requires choosing appropriate standards, selection of optimal operating conditions, as well as consideration of the possibility of omitted elements, peak and background overlaps, matrix absorption effects, beam damage and elemental migration, reintegration of heterogeneous materials, fluorescence effects, and variations in the oxidation state of iron.

The purpose of this paper is to present a few examples of art objects that have been analyzed with the electron microprobe over the past 40 years at the Boston Museum of Fine Arts. The results of many examinations have led to the fact that the object was a forgery. However, the results of most examinations have proved that the object has the necessary structure and chemistry to be authentic. It should be pointed out, that an object can be proved to be a forgery. But you cannot prove authenticity. The materials used for the object may not fit the time frame for the period, therefore a forgery, or the materials may fit the time period, the structure and style may fit the work of a particular artist or craftsman and still be a copy.

In the early 1960’s a 15th century painting, a Portrait of a Lady, see FIG. 1, was brought to the laboratory at the Boston Museum of Fine Arts for examination.

Castaing first presented his idea to use secondary X-rays excited by a focused electron beam from a polished solid sample for microanalysis in 1949 at the First European Regional Conference on Electron Microscopy in Delft, Netherlands. As part of his dissertation, he then not only built the first electron probe microanalyzer (EPM), but also established many of the theoretical and analytical principles of the technique. This technique offered enormous analytical advantages to earth scientists over other analytical methods available at the time. For example, it allows qualitative and quantitative analysis of individual mineral grains a few microns in diameter; mineral grains can be viewed during analysis, thus ensuring accurate correlation between composition and grain morphology; for most purposes, the method is non-destructive; in situ analysis of minerals in polished thin sections results in retention of textural relationships among coexisting minerals; because of the small excitation volume, the technique is ideally suited for the study of zoned minerals, minute inclusions, exsolution lamellae, etc.; and once suitable standards are prepared and correction procedures are established, a large number of quantitative analyses can be obtained in a comparatively short time.

The matrix of bio-organic samples is composed primarily of carbon, hydrogen, oxygen and nitrogen. Dispersed within this matrix there is a wide spectrum of other elements whose local concentration can be measured in situ by x-ray spectroscopy. The analysis continues to present a challenge because most samples are highly hydrated, beam sensitive and of low density. In light of these restrictions there is only one sensible approach to the use of an analytical procedure which depends on the intereaction of high energy beam electrons with a labile specimen. Samples must be prepared by the least invasive preparative procedure and irradiated with the minimum amount of energy in order to obtain the maximun amount of chemical information. Low temperature sample preparation and low voltage electron beam instruments are central to these processes. By way of example, a brief description will be given of an on-going biological investigation which makes full use of these processes.

Electron probe microanalysis (EPMA) is being utilized to characterize the composition of superconductors in the Bi-Sr-Ca-Cu-O system (BSCCO) that are doped with minor amounts of transition metal s. Specifically, the Fe Group (Fe, Ru, Os), the Co Group (Co, Rh, Ir), the Ni Group (Ni, Pd, Pt), the Zn Group (Zn, Cd, Hg), and the other elements in the Cu Group (Ag, Au) can be used as dopants. These first long period ions in their common valence states have similar ionic radii, so only small lattice distortions are expected when other first period ions are substituted for Cu. Our research involves the systematic study of the effect of transition metal substitutions (for Cu) on the properties of the two high temperature superconductors, Bi2Sr2CaCu2O8+δ (2212) and Bi2Sr2Ca2Cu3O10+δ (2223).

Substitution for Cu with transition metal ions enables one to see what effect the changing d and s character of the ions have on the superconducting properties of these compounds.

Quantitative X-Ray microanalysis is a well established technique for the chemical analysis of homogeneous materials having a flat surface. Precision better than 2% can be routinely obtained if the analysis is performed used the so called ZAF where standards of known composition must be used for all the elements presents in the system under analysis. However, real materials have generally different phases and their composition in a specific phase is not necessary homogeneous. Also, real materials may have surfaces that are not flat and where it is inappropriate to polish them, like a fractured or a corroded surface. It is therefore of paramount importance to develop quantitative schemes for such materials. The first step is to understand the shape of the φ(ρz) curves of heterogeneous materials. This paper present some of these curves that have been obtained with the Monte Carlo program CASINO available for free at www.gme.usherb.ca/casino.

In 1971, Art Chodos and Arden Albee of Caltech contracted Corning Glass Works to produce synthetic glasses containing a number of elements at about 1 weight percent, to be used as electron microprobe trace element standards. Three glasses were produced, 95IRV, 95IRW, and 95IRX (informally known as Corning/Caltech glasses GLV, GLW, and GLX, respectively, in the microanalysis community), which contain the following elements in a Ca-Mg-Al borosilicate glass matrix:

95IRV K Ti Cr Fe Ce Hf

95IRW V Mn Co Cu Cs Ba La Th

95IRX Ni Zn Rb Sr Y Zr Pb U

The source materials for each oxide were weighed-in at 0.80 percent, and each glass was batched in 2 pound lots and drawn into a 0.25 inch cane that was cut into 9 rods of approximately 5 inch length. Initial distribution to end-users was of 1/8 inch slices from the beginning of each cane, as well as sufficient material for chemical analysis to be performed by volunteers. The glass standards were formed in much the same way as the NBS 612, 614, and 616-series trace element glasses, but used a different matrix composition to avoid problems with alkali migration under the electron beam.

A technique has been developed for the interpretation of composition depth profiles from angleresolved x-ray data using a Monte Carlo electron scattering simulation. Conventional methods for the interpretation of angle-resolved depth profiles used in the fields of x-ray photoelectron spectroscopy (XPS) and Auger electron spectroscopy (AES) assume that the outgoing signal is exponentially attenuated along its path. This assumption if not valid for angle-resolved x-ray techniques, as the x-ray signal is dependent on both the paths of the incident electrons and the path of the emitted x-rays. In this case, while the latter can be treated using an exponential attenuation, the path of the incident beam is more complex and corresponds to the well known “pear-shaped” interaction volume. In order to reliably interpret angle-resolved depth profiles in which the angle of the incident beam is varied, it is necessary to be able to obtain the distribution of x-ray emission within the sample for any angle of incidence.

Knowledge of inner-shell ionization cross sections by electron impact is needed for quantitative procedures in electron probe microanalysis (EPMA) and Auger electron spectroscopy (AES) The common practice is to use semi-empirical formulas, based on the asymptotic limit of the Bethe theory, which sometimes are used beyond their domain of validity. Experimental measurements of ionization cross sections are scarce and affected by considerable uncertainties, thus a mere comparison with experimental data does not permit to draw a definite conclusion abou the accuracy of the various formulas. In this communication, we present new measurements o the relative variation of K- and L-shell ionization cross sections deduced from the counting rate of characteristic x-rays emitted by extremely thin films of Cr, Ni, Cu, Te, Au and Bi bombardec by keV electrons.

The studied films were produced by thermal evaporation on backing self-supported 30 nm carbon films.

It has previously been shown that microbial polyphosphate bodies are the main sites of metal sequestration. In our present study, we have exposed Synechococcus leopoliensis, grown to log phase and to overplus phase, to five different heavy metals, Al, Cu, Cd, Pb, and Zn, independently. The cells were analyzed using the STEM mode of a transmission electron microscope in conjunction with a PGT IMIX energy dispersive x-ray spectrometer.

Cells of Synechococcus leopoliensis were grown for two weeks at 21°C on a 12-hour light/dark cycle in modified Fizgerald’s medium The second cell type were the overplus cells, which were grown as previously described. The cells were exposed to 100PPM of each metal for two hours, then were fixed and embedded in EPON according to Luff’s procedure.

For x-ray analysis, cells of interest were first located using the TEM mode. The microscope was then switched to the STEM mode.

Ultra-thin polymer x-ray windows have been developed for energy dispersive spectroscopy (EDS) that enable analysis of the elements lighter in mass than sodium while protecting the detector from light and gases. Windowless detectors produce the ultimate in detector sensitivity, however that sensitivity is lost when ice and other contaminants form on the detector. Polymer windows have had a problem with icing in the past, however modern ultra-thin polymer windows contain metalized layers to prevent the diffusion of water. Nevertheless, over the course of time it has been observed that some detectors with polymer windows that are kept continually cool with liquid nitrogen form a thin layer of ice that attenuates light element sensitivity. The source of this water has been hypothesized to be a gradual leak of water vapor through the polymer x-ray window. This hypothesis has been questioned on the basis that the windows were helium leak tight to 1 x 10−9 mbar L/sec and helium is a smaller molecule than water.