Introductory investigations performed in order to make quantitative electron-probe microanalysis of very light elements such as boron, carbon, nitrogen or oxygen possible are described. The practical problems encountered in this kind of work are discussed in detail and, where possible, solutions are proposed. It is shown that with very light elements such as carbon it is no longer permitted to measure x-ray intensities at the position of the maximum of the emission peak as the shape of the carbon K a peak is subject to strong alterations, depending on the type of chemical bond involved. As a consequence, integral measurements have to be performed and it is shown that errors of 3040% can easily be made if this is neglected. The lengthy procedures, connected with integral intensity measurements, can be shortened considerably by the introduction of arealpeak factors, which, by definition, represent the ratio between the correct area (integral) and the peak intensity ratio. The accurate determination of such factors for 13 binary carbides is described. This set of factors will permit future measurements simply on the peak; multiplication of the peak intensity ratio by the appropriate aredpeak factors will then yield the correct integral intensity ratios.

A computer program based on the use of Gaussian expressions for the x‐ray distribution with depth [ϕ(ρz)curves] was tested on its usefulness for quantitative electron probe microanalysis. As the good results originally claimed for a similar program could not be reproduced initially, it was subjected to a detailed analysis. As a result, some modifications in the approach are proposed. Apart from increasing the speed of calculation considerably, the modified expressions provide a better insight into the delicate balance which has to exit between the relevant quantities involved. After a new optimization process the modified program was tested on about 450 published microanalyses. The results show that the ϕ(ρz) approach is indeed very promising as a narrow histogram with an r. m. s. value of 5. 4% could be produced. Finally, some suggestions are made for future improvements.

No abstract

An improved correction model for quantitative electron probe microanalysis, based on modifications of the Gaussian ϕ(ϱ) approach, originally introduced by Packwood and Brown, is presented. The improvements consist of better equations for the input parameters of this model which have been obtained by fitting to experimental ϕ(ϱ) data. The new program has been tested on 627 measurements for medium to heavy elements (Z>11) and on 117 carbon measurements with excellent results: an r.m.s. value of 2.99% in the former case and 4.1% in the latter. Finally the new program has been compared to five other current correction programs which were found to perform less satisfactorily.

In succession to our work on aluminium films a systematic database of thin‐film measurements on palladium films by electron probe microanalysis is presented. This time the measurements were performed at accelerating voltages between 4 and 30 kV, again on films of six different nominal thicknesses, ranging from 100 to 3200 Å, which were deposited simultaneously on 20 different substrates, ranging between Be and Bi. The purpose of this work was to provide further systematic data on which thin‐film programs can be tested. A total of 931 k ratios for the film element Pd and 913 k ratios for the various substrate elements were collected. Tests with our most recent thin‐film program, TFA, based on the double Gaussian PROZA96 procedure, on this database showed even better performance than in the previous case for the aluminium films: a mean value of 0.996 for kcalc/kmeas and a relative root‐mean‐square deviation of 3.266% in the histogram for the film element. Copyright © 2000 John Wiley & Sons, Ltd.