Nuclear Instruments and Methods 208 (1983) 659-663 659 North-Holland Publishing Company X-RAY FLUORESCENCE ANALYSIS WITH SYNCHROTRON RADIATION A. KN()CHEL, W. PETERSEN and G. TOLKIEHN Universitiit Hamburg, Institut fiir A norganische und A ngewandte Chemie, Martin - Luther - King - Platz 6, D-2000 Hamburg 13, Fed. Rep. Germany The method of trace element analysis by X-ray fluorescence detection has been improved to an especially efficient multielement method for the ng to pg range in matrices containing light elements by the use of synchrotron radiation for excitation. It was necessary to determine the intensity and polarisation of the synchrotron radiation quantitatively. Inclusion of the vertical electron beam diameter and the divergence into the calculation, and definition of an effective vertical beam diameter by fitting the calculated polarisation spectrum leads to quantitative agreement between experimental and calculated absolute intensity spectra of scattered and fluorescent radiation of well-defined samples. This means that absolute mass determinations are in principle possible. The physical limits of detection calculated with these data agree very well with the experimental results. The limits of detection for special elements can be optimised by using different absorbers in the primary beam. They range from 0.05 to 0.2 p,g for organic matrix. This implies an absolute physical detection limit of 0.1 to 0.4 pg for a diameter of the primary beam of 0.5 mm and a sample of 1 mg/cm 2. 1. Introduction Out of the large number of analytical principles for trace element analysis, there are only a few which can be used for a multielement-method. One of these is X-ray fluorescence analysis (XRFA). This uses the in- tensity of the characteristic X-ray emission lines for trace element detection. With the XRFA all elements with atomic numbers greater than about 15 can, in principle, be detected with comparable accuracy. The major drawback of conventional XRFA is its relatively low sensitivity. The physical limits of detec- tion are determined by the following effects: - scattering background due to the excitation process, - fluorescence lines of major components, - lines in the exciting radiation spectrum. The most general and important point is the first. The second point depends on the sample-type. It does not, for instance, apply for most of the samples from a wide range of environmental and biological materials, as here the major components are light elements. The third point of course only applies for X-ray tube excitation. The outstanding features of synchrotron radiation (SR), which allow significant improvements of the XRF-analysis are: - high flux density, - high degree of linear polarisation, - small divergence, - white spectrum without lines, - calculability of the source. Due to these features it should be possible to develop an absolute method of XRFA with greatly improved limits of detection. 2. Characterisation of the synchrotron X-radiation source The intensity of the X-rays scattered from a thin sample irradiated by a parallel X-ray beam of some degree of linear polarisation P has two minima in the direction of the polarisation. The intensity in these minima compared to the intensity perpendicular to an "unpolarised" parallel beam is 1 - P. The X-ray fluorescence radiation is emitted with a spherical distribution. The synchrotron-X-radiation of a storage ring has a rather sharp maximum of the degree of linear polarisation in the electron orbit plane. It can be higher than P = 0.9. If detection is done only in this position and direction, an improvement of the signal to scattering background ratio of 1/(1 - P), which may be more than a factor of ten, is the result. The effect of the polarisation of the exciting radia- tion is demonstrated by two spectra of a gaseous N 2 sample containing a small amount of Xe. The spectra were taken with detection in the polarisation direction and vertical to it with the same detector and aperture system. The sample was vertically positioned in the center of the synchrotron radiation beam. The spectra were normalised to equal beam current integrated over live-time of the detection system (fig. 1). The limits of detection are improved by the same factor 1/(1 -P), if there are no fluorescence lines of 0167-5087/83/0000-0000/$03.00 © 1983 North-Holland viii. ABSORPTION/FLUORESCENCE SPECTROSCOPY