Benchtop XES: Why is it so fast?

In a previous article, I discussed the broad case for benchtop advanced x-ray spectroscopies. While that article focused on x-ray absorption fine structure (XAFS), the present article addresses benchtop x-ray emission spectroscopy (XES). [I've previously discussed XES and its similarities and differences from traditional WD-XRF]. The goal of this article is to explain an unexpected fact about benchtop XES: despite the use of low-powered x-ray sources, benchtop XES has the same energy resolution as obtained at the synchrotron light sources and also has count rates between a few percent and those fully comparable to those for XES performed at the highest intensity monochromatized synchrotron beamlines.

The figure to the right is from Mortensen, et al., 2016. It gives a comparison of Co K_beta and valence-to-core (VTC) XES for two reference cobalt-rich materials. The perfectly smooth red curve was taken at ESRF, with the full might of a world-class 3rd-generation synchrotron behind it. The blue data was taken with the earliest benchtop spectrometer in my group at the University of Washington. This prototype instrument used only a tiny 10-W x-ray tube, essentially the least powerful on the market, but achieved XES count rates of about 1% of those at ESRF. The 'extra' KL_beta feature is due to a multielectron transition that is suppressed in the ESRF data by choice of the incident energy.

While the above result is encouraging, it is important to be realistic. The high brilliance and time-structure of x-ray beams at the synchrotron facilities hold indisputable value that cannot be replicated in the lab. Rather than trying to achieve ultrafine imaging (such as with Fresnel zone plates) or ultrafast soft x-ray or VUV pulses (such as with high-harmonic generation), in our work at the University of Washington and easyXAFS we have a more pedestrian mission.

Our goal is to develop 'analytical' XES as a new routine tool for chemistry and materials science research and also for process and quality control in industry, just as has long been the case for XRD, XPS, and XRF methods.

Below, first, I address the coupling of the source spectrum to the spectrometer for XAFS and for XES, with an emphasis on qualitatively explaining the high efficiency of benchtop XES. Second, it is useful to better understand the preferred x-ray tubes for stimulating the high fluorescence levels needed for rapid benchtop XES, and from that to understand the speed of benchtop XES in a more quantitative way. Third, and finally, I'll give some opinions and speculation on the range of possible applications of benchtop XES. From an R&D perspective, there are clear applications in the catalysis and electritcal energy storage communities, i.e., those who are already heavy users of synchrotron XAFS (and are at least aware of XES). The question is how far benchtop XES could expand beyond that specialist community.

A comparison of efficiencies for benchtop XAFS and XES

When using conventional x-ray tubes the source spectrum will consist of a broad bremsstrahlung spectrum together with occasional fluorescence lines from the anode material itself, see the figure below, where for clarify I've omitted showing the very intense, very narrow fluorescence lines. The lower-energy portion of the bremsstrahlung shows a hard roll-off due to absorption effects in the anode itself and by absorption from the beryllium exit window from the x-ray tube's vacuum space.

In the left panel, the 1-eV wide bandpass of a high-resolution point-focusing monochromator is shown. The inefficiency is obvious: the purpose of the monochromator is to reject almost all of the considerable (often >10^13/sec) photons being emitted by the x-ray tube, and to pass only those in the desired, narrow band. While dispersive spectrometers make use of a broader region of the bremsstrahlung, such 'efficiency' comes at the cost of correspondingly inefficient use of the optic at any one photon energy. The limitations shown in the left panel are intrinsic to diffraction-based optics.

By contrast, the right panel shows the situation for XES. Now all photons above the binding energy E_b of the relevant species can, in principle, stimulate fluorescence. This must lead to a far more efficient use of the tube-source flux. While the details of course vary depending on sample chemistry, as a rough approximation the flux between E_b and 2*E_b is relatively effective at stimulating fluorescence close enough to the surface that the fluorescence can often escape the sample and enter the spectrometer. Higher-energy incident photons penetrate too far into the sample and contribute much less to the 'useful' core-hole generation rate that sets the overall count rate.

This explains why benchtop XAFS is typically limited to concentrated samples in transmission-mode geometry whereas benchtop XES can study rather dilute samples. Given the above schematic explanation of the efficient use of the tube flux by XES, the issue now becomes more quantitative. How much 'useful' core-hole generation rate, i.e., useful fluorescence rate that escapes the sample, can we actually get on the benchtop?

X-ray tubes, fluxes, and core-hole generation rates

To begin, the problem of creating a very, very high flux of polychromatic x-rays onto mm-scale samples has been solved by the decades of commercial refinement of so-called 'XRF tubes'. These are conventional x-ray sources that satisfy the high industrial demand for x-ray based elemental analysis. Whether for lower-resolution hand-held instruments used in field work or high-power wavelength-dispersed spectrometers in advanced analytical chemistry labs, all x-ray fluorescence (XRF) systems whose purpose is to analyze elemental composition start with the goal of shining as many hard x-rays as possible onto the sample.

To emphasize the considerable benefits of XRF-style tubes for stimulating fluorescence, it is useful to compare two designs for XRF-style tubes with that of a traditional 2-kW scale tube commonly used for powder x-ray diffraction (XRD), as in the figure below. The leftmost tube is a crude schematic of an XRF-style tube with a transmission-style anode. The electron beam impacts a thin film that is bonded to the back of the Be exit window, which also serves as a heat sink path. The sample can approach as close as a few mm from the window, and hence also the same few mm from the anode (i.e., the x-ray source spot). Such very low-powered XRF-style tubes are sold by several vendors, with Moxtek being among the best known.

The middle x-ray tube shows a higher-powered variant, such as is made by Varex, where the anode is several mm behind the exit window. This has the advantage of improved cooling that allows for much higher electron beam power. The higher tube power easily offsets the larger anode-to-sample distance, and added benefit comes from the better escape geometry for x-rays leaving the anode. XRF tubes with this general geometry are available from ~50W power up to ~4 kW.

As a final comparison, a schematic for a conventional 2-kW XRD-style tube is shown as the right-most example in the figure. Here, the effective spot size is again mm-size but the exit window is ~40 mm from the anode. The sample distance from the tube window is also larger because of clearances needed for the tube housing and shutter. As the flux on a fixed-size sample decreases as 1/distance^2 from the anode, the higher powered XRD tube is comparatively ineffective for stimulating fluorescence.

Detailed consideration of the tube spectrum and sample geometry can estimate the 'useful' core-hole generation rate, i.e., the rate of absorption that makes fluorescence capable of escaping the sample and entering the spectrometer. Such calculations indicate nearly 10^11/sec 'useful' core-hole generation rate for concentrated 3d transition metal samples for even the small 10 W transmission-style tube, with an increase to ~10^12/sec for higher-powered tubes of the second type, discussed above. These are, of course, the same general order of magnitude as for monochromatic synchrotron beamlines.

The surprising performance of the benchtop XES systems with respect to the synchrotron is therefore solved. The benchtop XES systems based on the University of Washington and easy XAFS design use essentially the same spectrometer layout as is most commonly used at the synchrotron light sources. Since the spectrometer is more-or-less the same, then the energy resolution and efficiencies will be very similar (module the use of multiple analyzers in many synchrotron XES and RIXS endstations). Since count rate simply then scale with the 'useful' core-hole generation rate, the high performance of the benchtop systems follows from the above considerations.

(Old) Research Uses and (New) Analytical Uses of Benchtop XES

Synchrotron XAFS and XES experts have a pretty good idea, already, of possible uses for benchtop XES, especially in chemistry and magnetic materials research. For example, a comprehensive review is given by Glatzel and Bergmann and also by the more theory-oriented article of DeGroot. More recent work by Pollock and DeBeer addresses the exciting realm of valence-level XES for ligand effects and also for special studies of the geometry of transition metal centers. Some subset of previously synchrotron-based XES studies of catalysts and electrical energy storage materials, to name two prominent examples, are going to become far more accessible due to benchtop spectrometers.

A parallel question is the potential range of ‘routine’ analytical applications that can accelerate materials discovery. In-house XRD and XPS are indispensable tools in any major research institution. XES has analogous 'routine' merits as a bulk-sensitive non-vacuum alternative to XPS, in addition to giving an often strong information on ion spin state and unique fingerprinting of coordination and ligand identity in the valence-to-core region. Benchtop XES clearly has a high potential to join the ranks of 'simply necessary' x-ray tools in University shared equipment facilities.

Finally, very different considerations arise in the discussion of possible process and quality control applications of benchtop XES (or XAFS). These mostly fall into two categories.

First, where are existing industrial applications of, e.g., NMR or XPS, that are simply tedious or otherwise technically unsatisfying? For example, the high efficiency of benchtop XES can often beat measurement times for solid-state NMR, with less sample material and fewer constraints on impurity content. Also, the surface insensitivity of XES gives it improved performance compared to XPS for the truly broad class of materials where surface contamination is unavoidable.

Second, we need ask about process or quality control problems that are simply unsolved by existing analytical methods. Will the first 'killer app' for benchtop XES be in regulatory compliance of heavy metal content for environmental purposes, quality control involving dopant coordination in Li-ion battery electrodes, sulfur characterization in soils or fossil fuels, or catalyst process monitoring in refineries? Each is possible, but none of these applications are established. Only time will tell.