# X-ray Detectors in Analytical Electron Microscopes

A technique for experimentally determining the efficiency function of x-ray detectors in electron beam instruments is described. Submicron spheres of pure materials (elements or stoichiometric compounds) produced by a process called electrohydrodynamic atomization (EHD), are used as calibration samples. The main objective of this work is to develop a technique for routine determination of a detector efficiency function, DEF, as a basis for composition determination with x-ray data. This is accomplished by using a single sample containing different elements in spherical form to measure the DEF at discrete energies throughout a detector’s range.

In this work the different terms involved in the expression that relates the detector efficiency to the production rate of x-rays in thin spheres are determined. These include: (a) Derivation of the conventional absorption and fluorescence corrections for thin spheres and comparing them with previous results for thin foils. Both sphere corrections are found to be less severe compared to films. It is also found that for submicron spheres both corrections can be safely neglected. (b) Modification of the absorption and fluorescence corrections for films and spheres to account for detector geometry. The absorption and fluorescence corrections for spheres are found to be insensitive to detector geometry. This advantage adds to the many advantages for using spheres to calibrate the DEF. (c) Experimental determination of the x-ray depth distribution function, $\phi$($\rho$z), of thin spheres of five elements dispersed in the periodic table (namely Al, V, Ni, Pd and W) by using a tracer technique. The tracer-technique results were empirically fit to a general expression for $\phi$($\rho$z) for thin spheres. and (d) Theoretical determination of the x-ray distribution function using a Monte Carlo technique. The Monte Carlo results were found to be sensitive to the parameters used in the stopping power and inner-shell ionization-cross section models.

The technique was tested using an HB501 dedicated STEM interfaced with a lithium-drifted silicon x-ray detector. The results show that the detector efficiency curve constructed is sensitive to the parameters used for the ionization cross-section expression. The ionization cross-section parameters by Mott and Massey were found to produce a DEF curve that agrees well with the theoretical DEF curve. This technique for determining composition should work well if more accurate values for the ionization cross-section are determined.