Article pubs.acs.org/ac
Laboratory Scale X‑ray Fluorescence Tomography: Instrument Characterization and Application in Earth and Environmental Science Brecht Laforce,*,† Bram Vermeulen,† Jan Garrevoet,‡ Bart Vekemans,† Luc Van Hoorebeke,§ Colin Janssen,∥ and Laszlo Vincze† †
X-ray Microspectroscopy and Imaging Group (XMI), Ghent University, Krijgslaan 281 S12, B-9000 Ghent, Belgium PETRA III, Deutsches Elektronen-Synchrotron (DESY), Notkestrasse 85, D-22607 Hamburg, Germany § UGCT-Department of Physics and Astronomy, Ghent University, Proeftuinstraat 86, B-9000 Ghent, Belgium ∥ Environmental Toxicology Unit, Ghent University, Jozef Plateaustraat 22, B-9000 Ghent, Belgium ‡
ABSTRACT: A new laboratory scale X-ray fluorescence (XRF) imaging instrument, based on an X-ray microfocus tube equipped with a monocapillary optic, has been developed to perform XRF computed tomography experiments with both higher spatial resolution (20 μm) and a better energy resolution (130 eV @Mn−Kα) than has been achieved up-to-now. This instrument opens a new range of possible applications for XRF-CT. Next to the analytical characterization of the setup by using well-defined model/reference samples, demonstrating its capabilities for tomographic imaging, the XRF-CT microprobe has been used to image the interior of an ecotoxicological model organism, Americamysis bahia. This had been exposed to elevated metal (Cu and Ni) concentrations. The technique allowed the visualization of the accumulation sites of copper, clearly indicating the affected organs, i.e. either the gastric system or the hepatopancreas. As another illustrative application, the scanner has been employed to investigate goethite spherules from the Cretaceous-Paleogene boundary, revealing the internal elemental distribution of these valuable distal ejecta layer particles.
T
be obtained, using dedicated reconstruction algorithms.6 During the past decade XRF-CT has developed into a quasiroutine technique based on the use of high spatial resolution synchrotron radiation sources.7−11 Recent developments in laboratory X-ray sources, X-ray optics, and detector technologies have made the construction of a laboratory scale XRF tomography instrument feasible. Wegrzynek et al. developed an XRF-CT instrument as part of a multitube laboratory microXRF setup, employing a Mo and a Cr tube combined with a Si(Li) detector.12 This instrument used a collimated X-ray beam to achieve a spatial resolution of a few tens of micrometers. A commercial laboratory scale instrument combining absorption computed tomography (ACT) and XRF-CT has also been developed.13,14 This installation, the SkyScan 2140 (Bruker, USA), uses conical X-ray beams to illuminate the sample. The ACT and XRF-CT images are recorded simultaneously, using an X-ray sensitive CCD camera in transmission mode for the ACT and a second, energy dispersive X-ray camera for the XRF-CT. The achieved spatial resolution is 70 μm, while the energy resolution is 180 eV at 6.4 keV.
hree-dimensional elemental analysis of microscopic samples is of great interest in several scientific fields, such as geology, archeology, and ecotoxicology. Using specialized X-ray fluorescence (XRF) methodologies, 3D analyses can be performed in an elegant, nondestructive fashion with trace level detection limits for a wide range of elements. There exist two main routes to 3D XRF analysis. The first is the confocal XRF methodology where focusing optics with coinciding focal points are used on both the X-ray source and detector so that only a spatially confined microvolume is analyzed during the measurement. Scanning the sample through the confocal volume yields 3D information in a straightforward way. This methodology has been successfully employed both using laboratory and high spatial resolution synchrotron radiation sources.1−5 XRF computed tomography (XRF-CT) is another X-ray based analysis technique yielding 3D elemental information on the microscopic scale. Apart from the need for an extra rotation stage, the setup of an XRF-CT instrument is less demanding compared to the confocal XRF methodology, since only the Xray source needs to be equipped with optics to generate a highly collimated beam. The primary tomographic data obtained by such a setup are represented by so-called XRF sinograms, which are collected by performing line scans of the sample at different rotational angles. From the sinograms, virtual slices of elemental distributions through the sample can © XXXX American Chemical Society
Received: January 12, 2016 Accepted: February 18, 2016
A
DOI: 10.1021/acs.analchem.6b00137 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
providing the beam dimensions. The beam size at the Fe−Kα line was found to be 27 μm (H) × 28 μm (V) at 2 mm working distance, having a 8 mrad divergence. The beamsize at the Mo− Kα line (17.4 keV) was found to be 22 μm. The analytical capabilities of the setup were estimated by determining the achievable detection limits (DL) using a pressed pellet sample of NIST SRM 1577c bovine liver. A sample area of 250 by 250 μm2 was scanned using 10 μm step size with an acquisition time of 60 s live time per point, resulting in a total measurement time of 26 × 26 × 10s = 40560 s (11.3 h). The sum spectrum resulting from this scan was analyzed in terms of net peak areas for the detectable elements and associated background using AXIL, a software package using an iterative least-squares fitting algorithm.16,17 The element dependent detection limits of the instrument were then determined using following relation15
Recently, we have implemented an XRF tomography extension of our existing, in-house developed micro-XRF spectrometer15 based on a new Mo X-ray tube equipped with a monocapillary optic and a silicon drift detector (SDD). This new instrument means an important step forward for lab scale XRF-CT in terms of both spatial resolution (20 μm) and energy resolution (130 eV @ Mn−Kα, 5.9 keV), which opens a new range of possible applications due to the possibility of smaller sample regions to be investigated with a better discrimination between XRF lines than previously feasible. Several samples have been examined to demonstrate the analytical capabilities of the new XRF-CT setup, including a set of test samples based on glass capillaries filled with salt solutions, an ecotoxicological model organism (Americamysis bahia), and goethite spherules from the Cretaceous-Paleogene boundary.
■
MATERIALS AND METHODS Experimental Setup. A photograph of the XRF-CT setup is shown in Figure 1. The scanner uses a 50 W Mo (max.
DLi =
3 Ib Ii
Ci
where DLi is the minimum detection limit of element i in units of concentration defined by Ci, Ib is the background intensity, Ii is the net peak intensity of element i, and Ci is the reference concentration of the element in the selected standard reference material. To obtain information on the performance of the instrument representing a practical situation in terms of measuring time per point, the DL values obtained via this equation were extrapolated to a measurement time of 20 s, which is a typical value used during the tomographic experiments. Analysis of the NIST SRM 1577c scanning experiment revealed relative DLs ranging from 500 ppm for sulfur to 2 ppm for iron; however, using a more realistic acquisition time settings of 20 s the DL for sulfur is about 20000 ppm, while for iron an DL of 115 ppm is found. The XRF spectrum obtained during the DL determination is shown in Figure 2, while the DLs for the total acquisition time and the extrapolated time of 20 s are presented in Figure 3. Test Samples. In order to evaluate the three-dimensional imaging capabilities of the new XRF-CT microprobe, a series of test samples were made using thin walled borosilicate capillaries with an outer diameter of 1 mm and 110 μm wall thickness
Figure 1. Photograph of the experimental setup used during the XRFCT experiments with indications of the main components.
voltage 50 kV) anode tube as an X-ray source, equipped with a monocapillary optic (Xbeam, X-ray Optical Systems Inc., USA), concentrating the generated X-rays into a highly collimated pencil beam. The sample is mounted on a high precision motorized XYZθ sample stage with a movement accuracy of 1 μm. A silicon drift detector having a specified area of a 60 mm2 and a crystal thickness of 450 μm (SiriusSD, e2v, UK) is used to detect the fluorescent X-ray photons. The detector is equipped with a collimator with a 50 mm2 circular window. A digital microscope (Dinolite, Taiwan) facilitates optical sample observation. The detector can be moved relative to the sample, in order to maximize the detection solid angle thus optimizing the signal acquisition efficiency. Source and detector are positioned under an excitation-detection geometry which defines a 90° angle of detection with respect to the optical axis of the incident beam. The X-ray beam size produced by the monocapillary source was determined by line scans over a stainless steel wire of 10 μm in diameter (AISI 302, Goodfellow Cambridge Ltd., UK). To obtain the dimensions of the micro beam, a Gaussian function was fitted to these line scans. The full width at halfmaximum (fwhm) of these fits was calculated and corrected for the diameter of the stainless steel wire and scanning geometry,
Figure 2. XRF sum spectrum of the NIST SRM 1577c standard, 26 × 26 points mapping with 60 s acquisition time per point (40560 s or 11.3 h total acquisition time). The X-ray tube was operated at 40 kV and 0.7 mA. B
DOI: 10.1021/acs.analchem.6b00137 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
divided in several batches that were exposed to varying Cu and Ni concentrations. Two A. bahia samples have been analyzed using XRF-CT; one belonging to the reference population (no artificial Cu and Ni exposure) and one which had been exposed to a medium with 135 μg/L Cu and 84 μg/L Ni. The organisms were held in groups of 20 in a 4 L tank. Exposure to the elevated metal concentrations started from the moment of hatching. After an exposure period of 28 days, the organisms were freeze-dried to be preserved and fixated and were then mounted on a graphite pin in order to be analyzed with the XRF microprobe. Goethite Spherules (Cretaceous-Paleogene Boundary). Upon impact of a (large) extraterrestrial body on Earth, layers consisting of small particles of crushed, melted, and vaporized rock are dispersed over a large area of the surface of our planet.22,23 These layers, among others consisting of spherule particles (ejecta), represent a valuable source of information on the Earth’s impact history. Imaging the elemental distributions in these particles reveals internal structures and phases, providing information that can shed new light on the origin, formation, and history of these spherules. Microbeam XRF methodologies are ideally suited for this kind of investigations.8,22,24 In cooperation with the Earth System Science group (VUB), we investigated some examples from a specific type of ejecta particles: goethite microkrystite spherules from the Cretaceous-Paleogene boundary, collected in Italy.25 Several particles were mounted on plastic pins using a very small amount of epoxy resin based glue. The large number of available spherules allowed for a preliminary selection of the most appropriate samples for the XRF-CT experiments. The particles were inspected using an optical microscope prior to the mounting procedure in order to select intact spherules for further processing. A second inspection took place after the particles were attached to the polymer pin, to ascertain that no damage occurred due to the sample mounting procedure. XRF Tomography. As a first step in the analysis of each sample, a conventional XRF mapping is made to obtain a detailed 2D elemental distribution (projection map) of the sample. Next, a sample height of interest is selected at which the XRF-CT experiment is performed, resulting in the elemental distributions corresponding to the virtual slice/ sample cross-section. XRF line scans are performed for multiple rotational angles, in the rotation angular range of 0−360° with equiangular steps, leading to a so-called sinogram for every detectable element present in the irradiated slice of the sample. The term sinogram is derived from the sinusoidal curves which are the result of this type of measurement, as can be seen in the left part of Figure 5. Several reconstruction algorithms exist to retrieve the internal structure/elemental distributions of the sample from these sinograms.6,26,27 During the experiments described in this paper, an in-house available filtered back projection (FBP) procedure was used. Figure 5 demonstrates the data analysis of an XRF-CT experiment. First, the software AXIL is used to fit the measured spectra, and an in-house developed software package (“microxrf2”) is employed to retrieve the sinograms of every element present in the sample. Next, these sinograms are reconstructed via filtered back projection in order to obtain the elemental distribution of these elements in the virtual slice through the sample.
Figure 3. Relative DLs of the laboratory XRF-tomography instrument for NIST SRM 1577c for a total measuring (live) time of 40560 s (measured data) and extrapolation to a measurement time of 20 s.
(Sutter). These capillaries were filled with an aqueous salt solution of K2Cr2O7, RbCl, or iron alum (8000 μg/g of Cr, Rb, and Fe, respectively), respectively, using capillary forces and sealed at top and bottom with wax. The capillaries were glued together in a triangular arrangement to create a threedimensional test object as can be seen in Figure 4.
Figure 4. Photo of the tomography test samples constructed from three capillaries filled with a salt solution as mounted in the instrument with the Cr/Fe/Rb RGB image of XRF tomography virtual slice throught the test sample (step size 20 μm, 180 projections, 15 s per point, 40 kV, 0.7 mA).
Ecotoxicological Model Organism: Americamysis bahia. The Americamysis bahia is a shrimp-like crustacean, which originates from the East coast of the United States. It is widely used as an aquatic (eco-) toxicological model organism.18−21 The organisms used during these experiments were cultured at the laboratory of the UGent Environmental Toxicology Unit (GhEnToxLab). The population was subC
DOI: 10.1021/acs.analchem.6b00137 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
Ecotoxicological Sample: Americamysis bahia. Two different organisms have been analyzed using XRF-CT, one of which has been exposed to elevated concentrations of both Cu and Ni and an unexposed reference organism. Figure 6 shows a microscope image, captured by the digital optical microscope of the XRF microprobe, and an overlay
Figure 5. Sinogram and reconstructed image (via FBP) for the Ca signal of a virtual slice through an Americamysis bahia sample (20 μm steps, 200 projections, 14 s per point).
■
RESULTS AND DISCUSSION Analysis of the 3-Capillary Test Sample. The initial test experiments of the new tomographic setup were performed using a test object consisting of three attached capillaries filled with a salt solution. The well-known geometry and elemental concentrations in the salt solutions make this a very useful model sample to determine the capabilities of the instrument in terms of XRF-CT experiments. The sample and an XRF tomography virtual slice thereof are represented in Figure 4. Due to the low fluorescence yield for low Z elements, absorption effects in the walls of the glass capillaries, and selfabsorption in the salt solutions, only the relatively high atomic number constituents (Cr, Fe, and Rb) could be analyzed. The information depth for Cr could be derived from the experimental data, determined to be about 450 μm. For Fe it is 850 μm and for Rb the information depth is considerably larger than the diameter of the capillary (>1000 μm). The experimental information depth per element is obtained by determining the number of pixels between the rim of the capillary and the point where the elemental intensity drops to the background intensity. Using the software xraylib,28,29 the theoretical information depth for each of these elements can be determined from the absorption correction factor. The absorption correction factor Tabs can be calculated using the equation
Figure 6. Left and middle: microscope image with Cu/Ca/Br overlay of an XRF mapping on an Americamysis bahia exposed to an elevated Cu concentration (step size 15 μm, 14 s per point, 40 kV, 0.7 mA). Right: Microscope image of reference organism. The dotted lines indicate the position of the XRF-CT experiments.
image corresponding to the red(Cu)/green(Ca)/blue(Br) RGB representation of the conventional 2D XRF mapping of the exposed shrimp. The organism is rich in calcium, an important element of the exoskeleton, while a large copper rich region and a bromine hotspot are also discernible. Since copper is one of the elements of interest the organism has been exposed to, the virtual slice of the tomographic experiment is selected corresponding to this region, as indicated by a dotted line on the figure. The nickel signal was too low for significant detection. The results of this XRF-CT experiment are represented in Figure 7a. The exoskeleton of the organism is clearly visible in
Tabs = e−μρ·ρ·d
where the attenuation coefficient μ0 takes into account the matrix composition. When for a certain path length only 1% of the original intensity of the characteristic X-rays of an element is transmitted (i.e., Tabs = 0.01), the information for that specific element can no longer be significantly discerned from the background. This path length is hence defined to be the information depth for that element. For Cr, an information depth of 430 μm is calculated, for iron it is 990 μm, while for Rb a value of 8920 μm is found. These theoretical values agree very well with the experimentally determined information depths. It could also be calculated that for elements with atomic numbers below 23 (Ti and lighter elements) no signal passes through the capillary wall. Since the theoretical calculation of the information depth has been experimentally validated, we can safely use this procedure to calculate the information depth for other samples, taking into account the effect of the widely varying matrices, ranging from biological tissue to glass or rock.
Figure 7. 2D XRF mappings of two different Americamysis bahia organisms a. Cu/Ca/K RGB image of organism exposed to elevated Cu concentration; b. Cu/Ca/K RGB image of reference organism (step size 20 μm, 180 projections, 15 s per point, 40 kV, 0.7 mA).
the calcium signal. Potassium is present to some extent in the internal structures of the Acamysis bahia. The copper signal is very evident, being localized in a single, lobed structure near the back of the organism. To correlate this copper signal with the normal situation, an XRF-CT experiment was performed on a reference organism at approximately the same position of which the result is shown in Figure 7b. Clearly, no region with an elevated copper content is present in this virtual slice. It can be concluded that the copper D
DOI: 10.1021/acs.analchem.6b00137 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry rich structure found in the exposed organism is due to the uptake of this element from the exposure medium. When investigating the location of the internal organs according to the anatomy of the Americamysis bahia, the copper appears to be accumulated in either the gastric system or the hepatopancreas. Since the latter organ has a lobed structure in these organisms, it is the most probable candidate for the affected tissue. Elemental Distributions within Goethite Spherules. To gain data on the internal structure and elemental distributions within the goethite microkrystite spherules, prior to performing the XRF-CT analysis, a detailed conventional XRF mapping has been performed. The result for Ca is presented in Figure 8, together with an optical image of the
Figure 9. a. Ni/Fe/Ti RGB image of a virtual slice through the goethite spherule. b. Ca/Fe image of the same virtual slice (step size 15 μm, 150 projections, 15 s per point, 40 kV, 0.7 mA).
presence of this enriched rim can be of high relevance to the study of ejecta spherule layers of the Cretaceous-Paleogene boundary, since it indicates a phase separation, shedding light on the formation and transformation history of the particles. The right (b) of Figure 9 shows a Ca/Fe mapping of the particle. This image can be used in correlation with Figure 8 to determine the 3D position of the calcium rich phase in the spherule. Clearly, the Ca signal is most intense in a relatively thin region on the left-hand side of the particle, while a second zone with elevated Ca content can be noticed on the opposite side. Ca is present in a phase at the outer rim of the spherule. However, none of the common phases linked to ejecta spherules from the Cretaceous-Paleogene boundary (goethite, K-feldspar, glauconite, and magnesioferrite spinels) are characterized by such a high Ca content. It is therefore hypothesized that this is a secondary phase incorporated into the particle via phase-transfer or debris sticking to the surface of the original goethite particle.
■
CONCLUSIONS A new XRF microtomography setup was demonstrated based on a laboratory scale instrument, developed at Ghent University. The instrument has a better performance than other laboratory scale XRF-tomography setups reported in the literature, with a higher spatial (20 μm) and energy (130 eV) resolution. This XRF-CT setup opens the possibility to analyze the internal elemental distributions via virtual cross sections in a broad range of samples at the laboratory on a routine basis, with typical measurement times ranging from 8−24 h. Applying a test sample to analyze the capabilities of the instrument revealed both the prowess of the scanner at XRFCT and its limitations. The relatively thick wall of the borosilicate tubing blocked the characteristic X-rays of all elements with an atomic number lower than 23 (i.e., below V). For chromium and iron, self-absorption effects caused an incomplete image of the interior of the test object. The measured information depth for each element was in good accordance with the theoretically calculated values. In the field of bioimaging, illustrative measurements on two ecotoxicological samples demonstrated measurable copper accumulation in a well-defined structure within the organism as a result of exposure to an elevated concentration of this metal. Knowledge on which tissues and organs of the model organism are affected by metal consumption is of high importance when assessing the ecotoxicological impact of metal pollution. In the field of Earth and planetary science, an application example was shown corresponding to the analysis of a goethite spherule stemming from the Cretaceous-Paleogene boundary.
Figure 8. Optical image of the goethite spherule (top) and Ca XRF image (step size 10 μm, 20 s per point, 40 kV, 0.7 mA). The dotted line indicates the position of the XRF-CT experiment.
mounted spherule. The mapping clearly reveals the presence of two Ca-rich regions. In order to investigate the exact nature of these regions, it is important to know their precise location in the spherule. An XRF-CT measurement at the height of the larger of these two Ca-rich clusters was used to obtain this information, yielding the virtual slice at the position indicated by the dotted line on Figure 8. Figure 9 represents the virtual slice through the spherule at the position indicated in Figure 8. Information on the distribution of four elements is given, namely on Fe, Ni, Ti, and Ca. Since goethite is an iron hydroxide phase, the main element composing the bulk of the sample is Fe. The left (a) of Figure 9 is an RGB image representing the relative distribution of Fe, corresponding to the body of the particle, and Ni and Ti are only identified at the outer rim of the microkrystite. While examining ejecta spherules stemming from a Late Archaean impact vapor cloud, Goderis et al. also found these elements, among others, in elevated concentrations at the rims of the spherules.22 According to these authors, passive enrichment is responsible for the migration of these elements from the bulk to the outer regions of the particles during phase transformation. In the case of the goethite spherules, the Ni and Ti content of the initial olivine particle have migrated passively to the rim during the conversion to a goethite phase. The E
DOI: 10.1021/acs.analchem.6b00137 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
(18) Verslycke, T.; Ghekiere, A.; Raimondo, S.; Janssen, C. Ecotoxicology 2007, 16, 205−219. (19) Verslycke, T.; Vangheluwe, M.; Heijerick, D.; De Schamphelaere, K.; Van Sprang, P.; Janssen, C. R. Aquat. Toxicol. 2003, 64, 307−315. (20) Echols, B. S.; Smith, A. J.; Gardinali, P. R.; Rand, G. M. Chemosphere 2015, 120, 131−137. (21) DeLorenzo, M. E.; Key, P. B.; Chung, K. W.; Sapozhnikova, Y.; Fulton, M. H. Environ. Toxicol. 2014, 29, 1099−1106. (22) Goderis, S.; Simonson, B. M.; McDonald, I.; Hassler, S. W.; Izmer, A.; Belza, J.; Terryn, H.; Vanhaecke, F.; Claeys, P. Earth Planet. Sci. Lett. 2013, 376, 87−98. (23) Glass, B. P.; Simonson, B. M. Elements 2012, 8, 43−48. (24) Laforce, B.; Schmitz, S.; Vekemans, B.; Rudloff, J.; Garrevoet, J.; Tucoulou, R.; Brenker, F. E.; Martinez-Criado, G.; Vincze, L. Anal. Chem. 2014, 86, 12369−12374. (25) Goderis, S.; Tagle, R.; Belza, J.; Smit, J.; Montanari, A.; Vanhaecke, F.; Erzinger, J.; Claeys, P. Geochim. Cosmochim. Acta 2013, 120, 417−446. (26) Golosio, B.; Simionovici, A.; Somogyi, A.; Lemelle, L.; Chukalina, M.; Brunetti, A. J. Appl. Phys. 2003, 94, 145−156. (27) Schroer, C. G. Appl. Phys. Lett. 2001, 79, 1912−1914. (28) Schoonjans, T.; Brunetti, A.; Golosio, B.; Sanchez del Rio, M.; Solé, V. A.; Ferrero, C.; Vincze, L. Spectrochim. Acta, Part B 2011, 66, 776−784. (29) Schoonjans, T.; Brunetti, A.; Golosio, B.; Sanchez del Rio, M.; Solé, V. A.; Ferrero, C.; Vincze, L. Advances in Computational Methods for X-Ray Optics II 2011, 8141.
Based on the analyzed virtual slice, the outer rim of the spherule was found to be enriched in Ni and Ti, most probably through a process of passive enrichment during the transformation of the initial olivine phase to the current goethite phase. This information adds to our knowledge on the transformation history of the particles and hence is of high relevance in the study of the formation/alteration processes of these goethite spherules.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS Components of the setup were purchased within the framework of a Hercules-II project (AUG11-024). B.L. and J.G. acknowledge the support by the IWT (agentschap voor Innovatie door Wetenschap en Technologie, Flanders, Belgium). Our special thanks go to David Deruytter for preparing the Americamysis bahia samples and Steven Goderis for providing the microkrystite spherules.
■
REFERENCES
(1) Vincze, L.; Vekemans, B.; Brenker, F. E.; Falkenberg, G.; Rickers, K.; Somogyi, A.; Kersten, M.; Adams, F. Anal. Chem. 2004, 76, 6786− 6791. (2) Patterson, B. M.; Campbell, J.; Havrilla, G. J. X-Ray Spectrom. 2010, 39, 184−190. (3) Cordes, N. L.; Havrilla, G. J.; Usov, I. O.; Obrey, K. A.; Patterson, B. M. Spectrochim. Acta, Part B 2014, 101, 320−329. (4) Kanngießer, B.; Malzer, W.; Reiche, I. Nucl. Instrum. Methods Phys. Res., Sect. B 2003, 211, 259−264. (5) Šmit, Ž .; Janssens, K.; Proost, K.; Langus, I. Nucl. Instrum. Methods Phys. Res., Sect. B 2004, 219−220, 35−40. (6) Kak, A. C.; Slaney, M. Principles of computerized tomographic imaging; Society for Industrial and Applied Mathematics: Philadelphia, PA, 2001. (7) De Samber, B.; Vanblaere, S.; Evens, R.; De Schamphelaere, K.; Wellenreuther, G.; Ridoutt, F.; Silversmit, G.; Schoonjans, T.; Vekemans, B.; Masschaele, B.; Van Hoorebeke, L.; Rickers, K.; Falkenberg, G.; Szaloki, I.; Janssen, C.; Vincze, L. Powder Diffr. 2010, 25, 169−174. (8) Silversmit, G.; Vekemans, B.; Brenker, F. E.; Schmitz, S.; Burghammer, M.; Riekel, C.; Vincze, L. Anal. Chem. 2009, 81, 6107− 6112. (9) Terzano, R.; Spagnuolo, M.; Medici, L.; Vekemans, B.; Vincze, L.; Janssens, K.; Ruggiero, P. Environ. Sci. Technol. 2005, 39, 6280−6287. (10) Bleuet, P.; Gergaud, P.; Lemelle, L.; Bleuet, P.; Tucoulou, R.; Cloetens, P.; Susini, J.; Delette, G.; Simionovici, A. TrAC, Trends Anal. Chem. 2010, 29, 518−527. (11) de Jonge, M. D.; Vogt, S. Curr. Opin. Struct. Biol. 2010, 20, 606− 614. (12) Wȩgrzynek, D. X-Ray Spectrom. 2001, 30, 413−418. (13) Sasov, A.; Liu, X.; Rushmer, D. Developments in X-Ray Tomography VI; 2008; p 70780R-70780R-70789. (14) Bruyndonckx, P.; Sasov, A.; Liu, X. AIP Conf. Proc. 2011, 1365, 61−64. (15) Garrevoet, J.; Vekemans, B.; Bauters, S.; Demey, A.; Vincze, L. Anal. Chem. 2015, 87, 6544−6552. (16) Vekemans, B.; Janssens, K.; Vincze, L.; Adams, F.; Vanespen, P. X-Ray Spectrom. 1994, 23, 278−285. (17) Vekemans, B.; Janssens, K.; Vincze, L.; Adams, F.; Vanespen, P. Spectrochim. Acta, Part B 1995, 50, 149−169. F
DOI: 10.1021/acs.analchem.6b00137 Anal. Chem. XXXX, XXX, XXX−XXX
