EDS Using an SDD with X-ray Optic
Improvements and Limitations for Light Element Detection
- Fig. 1: Interior view of the SEM chamber with the LoMAX X-ray optic mounted on the EDS detector tube; distance from X-ray emitting sample surface to optic entrance aperture: 4 mm; working distance: 11 mm; take-off angle: 35°
- Fig. 2: Assembled LoMAX: optic, optional electron trap (in this image with magnets removed), ring clamp adaptor and adaptor insert
- Fig. 3: SEM images of a sample stub with a cross scratched across the center. From left to right: Misaligned and aligned pointer
- Fig. 4: Ring clamp adaptor, optic and pointer
- Fig. 5A: X-ray spectra, measured with optic (red) and without optic (blue). (a): CaCO3 (E0: 7 keV); (b) oxidized Be foil (E0: 7 keV); (c) Nylon fiber (E0: 2.7 keV)
- Fig. 5B: X-ray spectra, measured with optic (red) and without optic (blue). (A): CaCO3 (E0: 7 keV); (b) oxidized Be foil (E0: 7 keV); (c) Nylon fiber (E0: 2.7 keV)
- Fig. 5c: X-ray spectra, measured with optic (red) and without optic (blue). (a): CaCO3 (E0: 7 keV); (b) oxidized Be foil (E0: 7 keV); (c) Nylon fiber (E0: 2.7 keV)
Organic and mineral samples, containing light elements like oxygen, nitrogen or beryllium have been analyzed using an attachable X-ray optic with an SDD in SEM. Compared to the measurements without the additional optic, there is a significant increase of the detection sensitivity for X-rays in the energy range below 1 keV. In contrast the spectrum above 1 keV is not affected. The optic can provide similar intensities for low and high energy X-ray lines even at high accelerating voltage.
The silicon drift detector (SDD) technology has reached good energy resolution and exceeds the Si(Li) detector in particular at low X-ray energies. Below 600 eV the hardware-based pile-up rejector of EDS systems does not work and there is no applicable software-based pile-up correction so far. Because the SDDs pulse processing time is about twentyfold faster compared to the Si(Li) detector, the SDD is able to process much higher count rates in the low energy region without producing sum peaks . But the analysis of light elements is still disappointing due to the insufficient production of low energy X-rays and the absorption inside the sample matrix and by the X-ray window. To benefit from the SDDs excellent energy resolution and line separation in the region below 600 eV it is important to significantly increase the low energy detection sensitivity.
To get more signal from the light elements, the beam current might be increased or the detector could be moved closer to the sample to enlarge the covered solid angle. But both would increase also the high energy counts and the bremsstrahlung, thus leading to a rising dead time. Another way to become more sensitive for the low energy region is to reduce the accelerating voltage of the SEM from 15 or 20 kV down to 5 kV or less. But this is only possible if no high energy X-ray lines are to be stimulated.
In the following an X-ray optic is described, which significantly increases the detection sensitivity for low energy X-rays. The spectrum above 1 keV is not affected and the settings of the SEM will not change. The experiments are carried out with the LoMAX EDS optic from Parallax Research Inc., attached to a 10 mm2 e2v SDD equipped with an ultra-thin AP3.3 Moxtek window and a digital pulse processor from XIA LLC.
The optic works by total external reflection, similarly to a shaped mono-capillary .
It is a grazing incident X-ray optic, using grazing angles from about 6 down to about 2 degrees. X-rays diverging into a large solid angle were captured by reducing that angle and thus directed towards the detector. The optic that I use has three reflecting surfaces in the shape of cones of revolution with different opening angles. X-rays, going from the sample into a large angle with respect to the detector axis hit the optic near the small entrance while those emitted into smaller angles hit the optic near the detector. But due to the different grazing angles within the optic they all fall onto the surface of the detector. The inner surface of the optic is nickel, which is polished to about 1 nm rms smoothness. Nickel has very good reflecting properties for the examined energy range whereas the materials of glass capillaries are not as good. The lower the X-ray energy the higher the gain of the nickel-based optic. High energy X-rays pass directly through the optic without reflection, and the optic does not alter the spectrum above 1 keV. The ratio of peak intensities with and without optic shows a unity gain for energies above 1 keV and an approximately linearity of the gain below 750 eV with a maximum gain of about 10 times for the Be-K and the Si-L line. The optic body consists of non-magnetic copper and so does not influence the primary electron beam.
Alignment of the Optic
The optic is mounted by sliding the ring clamp adaptor over the end of the EDS detector tube and tightened by means of an Allen screw (fig. 1, 2). Alignment has to be done carefully so that the optics axis passes through the X-ray emitting point of the sample. A deviation of 200 µm leads to a drop down of the X-ray signal. The precise adjustment is done by a pointer which is screwed into the ring clamp adaptor instead of the optic and is adjusted by means of 4 tiny screws on the front of the split clamp. The pointer is imaged using an SEM magnification of about 200 times to see its position relative to the center and to find out the right working distance. The optic is designed for a distance of 4 mm from the X-ray emitting spot of the sample to the optic entrance aperture. Contacting the sample surface with the pointer, which is 4 mm longer than the optic, results in the exact working distance of the SEM. The adjustment process has to be repeated until the pointer is within 0.1 mm of the center (fig. 3, 4).
Comparative Studies and Results
All measurements with and without the X-ray optic are carried out with the EDX detector at the same position. Probe current, acquisition time and accelerating voltage of compared spectra are the same.
The measurements on a standard material of calcium carbonate (CaCO3) with the optic well aligned result in a unity gain for the X-rays of the Ca-K line which are passed directly through the optic without reflection, and a gain of about 8 to 9 times for the low energy X-rays of oxygen and carbon (fig. 5a).
Maximum gain at the lowest energy detectable with an EDS detector is obtained by the measurement of an oxidized beryllium foil (fig. 5b). The intensity of the beryllium line is strong even at high accelerating voltage of 20 kV. Thus it is possible to detect low and high energy X-ray lines with similar sensitivity. To work with these X-ray optic with high gain at low energies it is important to use an EDS detector and pulse processor with short shaping time. Otherwise the pile-up effects will disturb the spectrum. The minimum detection limit for beryllium is reached, measuring a sample of chrysoberyl (BeAl2O4) which contains about 7 weight percent of Be. The curve of the chrysoberyl spectrum is only slightly higher at the energy of the Be line compared to an EDS spectrum of aluminum oxide (Al2O3). High absorption of the Be photons by the chrysoberyl matrix is the main reason for this limitation.
Detection and quantification of nitrogen is important e.g. for food chemistry, fertilizer industry and soil science. Unfortunately the X-ray transmission of the detector window is particulary weak for the N-K line. Uncoated polyamide fibers (Nylon) have been analyzed. Nylon contains about 12 weight percent nitrogen, which results in a strong N peak in the EDS spectrum, measured with the X-ray optic (fig. 5c). Quantitative analysis using the X-ray optic must be done via comparison to standards. If one attempts to use normal standardless routines, they will predict far too much of the light elements. But experience shows that even without the optic it is necessary to use calibration standards to obtain good quantitative results of energies below 1 keV. Further measurements of nitrogenous samples like iron nitride (Fe4N) and chicken egg white have confirmed the greatly increased detection sensitivity for nitrogen using the X-ray optic.
Conclusions and Perspectives
A main problem for detecting low energy X-ray lines is the contamination of the sample surface during the measurement. The X-ray optic makes the EDS detector very sensitive in the energy range from fluor (677 eV) to beryllium (110 eV), thus probe current and acquisition time can remain low and so contamination is effectively reduced.
The silicon drift detector provides an excellent energy resolution and is able to process more than 100,000 counts per second. Therefore it is a very suitable tool for X-ray mapping. Usually high and low energy X-ray lines have to be mapped together, thus the energy of the primary electron beam has to be high enough to stimulate the heavier elements. A high tension setting of 15 to 20 kV is normal. With the X-ray optic installed the EDS detector is able to detect the high energy X-rays with normal sensitivity but also additionally the low energy lines with up to 10 times the sensitivity as without the optic, and in this way excellent X-ray mappings will be achieved.
Such an optic can also be configured as a ‘low pass filter' to minimize dead time and preferentially increase low energy counts by blocking the direct X-ray path to the detector and only allow reflected X-rays to pass through. Increasing the beam current then will not cause high energy counts to increase but will only increase the desired low energy X-rays.
Additionally, for energies below 200 eV, palladium is a better X-ray reflector than nickel so that an optic can be made that will give another 2.5-fold gain for bor and beryllium compared to the Ni-coated optic .
The described X-ray optic is an appropriate tool to increase low energy counts. The current drawbacks are the sensitivity to exact working distance and the position of the optic very close to the sample surface.
I kindly acknowledge the support by Parallax Research, Inc.
 Gernert, U.: Proc. 14th. Europ. Microscopy Congress Aachen 2008, Vol. 1, pp. 697-698
 McCarthy, J., et al., Microscopy and Micro-Analysis, Vol. 4, pp. 632-641, Cambridge University Press, 1998
 O'Hara, D., et al., Microscopy Today, March 2008, Vol. 16, No. 2, pp. 6-9