Chapter 3.35. X-ray fluorescence detection for EXAFS

The largest fluorescence signal is obtained by the collection of all X-ray energies. This signal is unfortunately contaminated with elastic scattering, Compton scattering and fluorescence from atoms that are of no direct interest. For concentrated samples this may not be a problem, but for dilute samples where the desired fluorescence signal is small this is a problem. The solution is to use some sort of energy discrimination to detect only the X-ray photons of interest.

There are three main approaches to this: X-ray filters, X-ray analyzers and discriminating/counting detectors.

The first two methods restrict the energies of the X-rays reaching the detector. The last method relies on energy discrimination in the detector and associated instrumentation.

X-ray bandpass filters are of two general flavours: high-resolution, usually using Bragg diffraction for a narrow energy bandpass, and low-pass X-ray filters for rejecting elastically scattered X-rays.

The low-pass filter is typically a lower Z (typically Z − 1) thin film on a low-Z substrate such as Kapton.The quasi-elastic scattered X-rays have an energy close to the probe X-ray energy, which is higher than the desired fluorescence energy. If the absorbing atom has an absorption edge at an energy greater than the fluorescence energy and lower than the elastic scattered energy, then the elastic background will be reduced relative to the fluorescence signal. For example, the Kα transition for iron is at 6.40 keV, while the iron absorption edge is at 7.11 keV. The absorption edge of manganese at 6.54 keV falls nicely between them, and is suitable as a filter for the iron edge (Fig. 1).

Figure 1 The effect of a `Z − 1' filter on a measured iron fluorescence spectrum. A filter of manganese placed between the sample and the detector will absorb most of the scatter peak, while transmitting most of the Fe Kα emission. For samples dominated by the scatter peak, such a filter can dramatically improve the signal-to-noise level. (Figure courtesy of M. Newville.)

After traversing the energy filter, the X-rays are then detected by an ion chamber or some other detector, which need not have energy discrimination itself. This technique is enhanced by geometrically suppressing refluorescence from the filter by the use of Soller slits (Stern & Heald, 1979). Besides relative simplicity, this allows the use of detectors that are capable of extremely high X-ray fluxes. To optimize path length through the fill gas while allowing a reasonable path length for ions and electrons to the electrodes, these chambers are usually designed with multiple subchambers that the X-ray beam transits (Fig. 2).

Figure 2 A schematic of the geometry of X-ray fluorescence detection. The incident X-rays excite atoms in the sample, and fluorescence and scattered radiation are emitted over a large solid angle. The `Z − 1' filter selectively absorbs higher energy X-rays such as elastically scattered photons. Soller slits are used to suppress refluorescence from the filter. (Figure courtesy of M. Newville.)

Ion chambers are typically used in an integrated current mode, with the small direct current being fed to a current-to-voltage amplifier, with the resulting output voltage recorded by a computer as a function of monochromator position.

Pulse-counting detectors, on the other hand, use the detector itself for energy discrimination. These detectors, along with their limitations, are described more completely in this volume by Kappen et al. (2024).

Another approach to using crystal analyzers is a bent-Laue optic, which has a larger acceptance angle than the flat optical elements because of crystal distortion in the strain of the bent crystal. Although these devices are somewhat tunable for energy, they are generally suited to a particular range of energies (see, for example, Zhong et al., 1999).

High energy resolution (⋘10 eV) can be used for a number of specialized purposes, such as inelastic X-ray scattering or suppression of core-hole lifetime broadening. The instrumentation is described more completely by Glatzel & Ghiringhelli (2024). The suppression of lifetime broadening has been described by Hämäläinen et al. (1999).