Anal. Chem. 2003, 75, 355-359
Correspondence
Micro X-ray Fluorescence Imaging without Scans: Toward an Element-Selective Movie Kenji Sakurai* and Hiromi Eba
National Institute for Materials Science (NIMS), 1-2-1 Sengen, Tsukuba, Ibaraki 305-0047, Japan
Micro X-ray fluorescence imaging is a promising method for obtaining positional distribution on specific elements in a nondestructive manner. So far, the technique has usually been performed by a 2D positional scan of a sample against a collimated beam. However, the total measuring time can become quite long, since a number of scanning points are needed in order to obtain a highquality image. The present report discusses a completely different way of performing imaging of elements much more quickly. A combination of grazing-incidence geometry using a rather wide beam and parallel optics for detecting X-rays can produce an X-ray fluorescence image with ∼1 M pixels and with ∼20-µm resolution in 1-2 min or less. The technique has the potential to open up new frontiers in X-ray imaging, particularly in element-selective movie applications. Microscopic imaging that is capable of distinguishing chemical composition is of extreme importance as a scientific tool. X-ray fluorescence has been used in routine analysis for the identification and determination of elements contained in specimens in environmental and life sciences as well as in industrial products.1 There are several methods of exciting X-ray fluorescence: the use of electron and ion beams is particularly significant in the analysis of extremely small areas, but X-ray beams and, in particular, synchrotron radiation are quite suitable for a wide range of applications because of their nondestructive nature and also because of the absence of charging effects even for insulators or organic materials. So far, X-ray fluorescence imaging has been based on scans with a collimated beam, and therefore, a lot of work has been devoted to fabricating nonspherical mirrors for X-ray focusing2-6 that now permit resolution at the 0.1-µm level * Corresponding author. E-mail:
[email protected]. (1) Janssens K.; Adams F.; Rindby A. Microscopic X-ray Fluorescence Analysis; John Wiley & Sons: London, 2000. (2) Bohic S.; Simionovici A.; Snigirev A.; Ortega R.; Deve`s G.; Heymann D.; Schroer C. G. Appl. Phys. Lett. 2001, 78, 3544. (3) Cai Z.; Lai B.; Yun W.; Ilinski P.; Legnini D.; Maser J.; Rodrigues W. A Hard X-ray Scanning Microprobe for Fluorescence Imaging and Microdiffraction at the Advanced Photon Source. X-ray Microscopy: Proceedings of the Sixth International Conference; Meyer-Ilse W., Warwick T., Attwood D., Eds.; AIP: St. Louis, MO, 2000; Vol. 507, pp 472-477. (4) Adams F.; Janssens K.; Snigirev A. J. Anal. At. Spectrom. 1998, 13, 319. (5) Chevallier P.; Dhez P.; Erko A.; Firsov A.; Legrand F.; Populus P. Nucl. Instrum. Methods Phys. Res. B 1996, 113, 122. 10.1021/ac025793h CCC: $25.00 Published on Web 12/05/2002
© 2003 American Chemical Society
when used with third-generation synchrotron sources. However, the technique has remained confined to the category of still imaging. Extending the technique to movie applications is not easy, because it requires scans, leading to long measuring times, especially when pixel numbers increase to enhance the quality of the image. The problems encountered in this respect relate to step-scan methods that use a conventional X-ray detector and counting system. The novel idea presented here brings much more rapid X-ray fluorescence imaging. It is rapid because it works without scans. The principle might be more comprehensible if we make a comparison with a normal optical microscope, which usually does not require any scans. An optical microscope uses the visible light from outside for illuminating objects in order to obtain images corresponding to the difference in both the optical properties (absorption and reflection of specific wavelengths) and physical surface shape (scattering). Focusing optics employing a combination of lenses to make a one-to-one coincidence between the object and the image. Therefore, one could expect that micro X-ray fluorescence imaging could be performed without scans even in the hard X-ray region,7 if good lenses are available. Primary X-rays from the outside excite specific elements in the objects, and the X-ray fluorescence generated in this way mainly reflects the chemical composition. The problem here is the difficulty of preparing good focusing optics for hard X-rays. However, as discussed in the earlier preliminary work,8 instead of focusing optics, a combination of grazing-incidence geometry and parallel optics for the detection can be also used for micro X-ray fluorescence imaging. The present paper presents details of the instrumentation, as well as several typical images that were obtained experimentally. EXPERIMENTAL SECTION Figure 1 schematically shows the present X-ray florescence microscope, which combines grazing-incidence geometry (0-2°) using a rather wide beam (∼1 cm in width) and parallel optics (smaller than 10 mrad) for detecting X-rays by a two-dimensional detector.8,9 The experiments have been performed with mono(6) Iida A.; Hirano K. Nucl. Instrum. Methods Phys. Res. B 1996, 114, 149. (7) Takeuchi A.; Aoki S.; Yamamoto K.; Takano H.; Watanabe N.; Ando M. Rev. Sci. Instrum. 2000, 71, 1979. (8) Sakurai K. Spectrochim. Acta B 1999, 54, 1497. (9) Sakurai K.; Eba H. Japanese Patent 3049313, 2000.
Analytical Chemistry, Vol. 75, No. 2, January 15, 2003 355
Figure 1. Schematic view of the X-ray fluorescence microscope without scans.
Figure 2. Typical X-ray fluorescence imaging with monochromatic synchrotron beam (7.2 keV). (A) Silver deposits. (a) Final and (b) early growing stages, exhibiting different patterns. Exposure time 2 min. (B) Section of rat brain (sample courtesy of Prof. I. Nakai, Science University of Tokyo, Japan). Exposure time 100 s. (C) Rock sample, silicified wood. Exposure time 2 min.
chromatic X-rays (typically 6-12 keV) from a normal bending magnet synchrotron source (BL-4A, Photon Factory, Tsukuba, Japan). The primary X-ray beam impinges on the surface of the object at quite a shallow incidence and generates X-ray florescence. Since X-ray fluorescence is basically divergent to the 4π direction, the use of a collimator is necessary to ascertain the position where it is emitted. A combination of a CCD camera and collimators has been used for X-ray diffraction10 and later X-ray (10) Wroblewski T.; Geier S.; Hessmer R.; Schreck M.; Rauschenbach B. Rev. Sci. Instrum. 1995, 66, 3560.
356
Analytical Chemistry, Vol. 75, No. 2, January 15, 2003
fluorescence imaging.8,11-13 Microchannel plates11,12 as well as capillary-assembled plates8,9 can be used as collimators. Initially, the technique uses total reflection (typical glancing angle smaller than 5°) and is therefore considered as a method to be employed mainly for the imaging of mirror-polished surfaces.8,12 However, primary X-rays impinging at approximately 1-2° can illuminate the sample and, by generating X-ray fluorescence, can thus create (11) Wroblewski T. Synchrotron Radiat. News 1996, 9, 14. (12) Cazaux J. X-ray Spectrom. 1999, 28, 9. (13) Eba H.; Sakurai K. Photon Factory Activity Rep. 2000, 18 126.
Figure 3. Angular dependence of X-ray fluorescence images for the section of sintered ore. (A) Optical microscope image. (B1, B2) X-ray images obtained at different glancing angles, which have been adjusted by monitoring the reflected X-ray intensity using the detector placed downstream of the camera system. The results clearly show different contrasts, indicating the segregation of the elements at a specific z-position (height or depth). Exposure time 100 s. (C) X-ray fluorescence spectra measured as a function of the glancing angle, reflecting the depth profile of the elements.
a contrast corresponding to the elements, not only from the object surface but also from deeper parts of the specimen. The technique can accept quite normal specimens, which do not have perfectly flat and smooth surfaces as well. Therefore, the technique is now used for more general imaging in a wide variety of applications. In the present system, a TC215 CCD (Texas Instruments, 1000 × 1018 pixels, pixel size 12 µm squared) and Hamamatsu C4880 controller (transfer rate 0.25 flames/s, 14 bit) are employed. A capillary-assembled plate (6 µm hole, 1 mm thick) is stored in the camera as a collimator for X-ray fluorescence. Here, the distance between the object surface and the detector is crucial to both detection efficiency and resolution. The resolution is expressed approximately as the product of the distance and the collimation, of which typical values in the present study are 3 mm and 6 mrad, respectively. Optimizing the structure inside a CCD camera is important, because the camera system is not typically designed to store the collimator and also because of the need to shorten the distance to the object as much as possible. The CCD is cooled electronically (-30 °C) and is therefore equipped with a vacuum chamber and window. In our case, we had to rebuild the whole cover of the camera head. RESULTS AND DISCUSSION Figure 2 shows typical examples of X-ray fluorescence imaging obtained with a monochromatic synchrotron beam (7.2 keV).
Figure 2A is an image of Ag L X-rays (integration of LR 2.98 keV, Lβ1 3.15 keV, Lβ2 3.35 keV, Lγ 3.52 keV) from sliver crystalline deposits grown from a small piece of copper foil, on which a silver nitrate solution is dropped. The pattern changes during growth, (a) final and (b) early stages, reflecting the different diffusion velocity. Figure 2B is a section of rat brain. The bright area corresponds to Fe K X-rays (integration of Fe KR 6.40 keV and Fe Kβ 7.06 keV), suggesting the aggregation of iron in specific parts of the tissue. Figure 2C is an example of rock samples, silicified wood. The stripe pattern is Fe K X-rays, indicating precipitation of iron in the interfaces. In Figure 2B and C, when the primary beam energy goes below the iron K absorption edge (7.11 keV), Fe K X-ray fluorescence becomes very dark, and the pattern is not visible. The spatial resolution for those images is only ∼20 µm, but it should be noted that imaging with ∼1 M pixels can be performed in only 1-2 min or sometimes even less. One can see detailed patterns of the precipitation of metallic crystals, aggregation to the specific part of the tissue, and segregation at the mineral interfaces. Although it is possible to perform the experiments even with a laboratory X-ray source,8 the use of tunable monochromatic or quasi-monochromatic synchrotron X-rays is promising for the selective excitation of the elements contained in the specimen. Another advantage would be the availability of more specific imaging, in addition to Analytical Chemistry, Vol. 75, No. 2, January 15, 2003
357
Figure 4. Time dependence of X-ray fluorescence images for ion-exchange resin (Dowex A-1, Dow Chemical Co.) on wet filter paper with a copper sulfate solution. Primary X-ray energy 9.1 keV. Exposure time 1 min. (A) Images from 52 to 135 min. (B) Enlarged images from 115 to 145 min. Scattering X-ray image taken at 8.8 keV (below the copper K edge) is also shown.
information on normal chemical composition, such as chemical states, local structure, and atomic location in the periodic structure unit obtained by scanning the primary X-ray energy, making use of the X-ray absorption fine structure14 and X-ray standing waves. Figure 3 shows images of another realistic sample, a section of sintered ore. The specimen is not perfectly flat and smooth. The primary energy here is set at 7.0 keV, below the iron K absorption edge, to see other minor elements by suppressing Fe K X-rays. We observed different contrasts when changing the glancing angle slightly. Since angular-resolved X-ray fluorescence measurements generally give a depth profile for the concentration,15 the present data relate to the depth or height profile of the elements. In the X-ray fluorescence spectra taken by a Si PIN detector (Amptek, XR-100T), one can see the manganese X-ray fluorescence peak and the angular dependence as well. Calcium is the major element contained in the glass substrate in this case and has a uniform spatial distribution. Therefore, the different spatial contrast would suggest 3-D manganese segregation: the bright area in B2 indicates manganese in the deeper parts (estimated penetration depth is ∼5000 Å), whereas that in B1 is manganese close to the surface (∼100 Å). One should note that the present imaging also provides information on the physical surface shape by scattered X-ray radiation. Sometimes different contrasts are obtained for the same specimen when the angles are changed quite substantially (∼1°), owing to the shadowing (14) Sakurai K.; Iida A.; Takahashi M.; Gohshi Y. Jpn. J. Appl. Phys. 1988, 27, L1768. (15) Iida A.; Sakurai K.; Gohshi Y. Nucl. Instrum., Methods A 1986, 246, 736.
358
Analytical Chemistry, Vol. 75, No. 2, January 15, 2003
effects, which can be corrected by several exposures with the rotation of the specimen. In our experience, combining the above with conventional angular-resolved X-ray fluorescence measurements could help in the interpretation of the images, particularly when elemental distribution and physical shape are correlated. One of the most important advantages of the present nonscanning X-ray fluorescence imaging would be its feasibility in movie applications, because it is possible to repeat the exposure in rather a short time, typically every 40-60 s in the present case. One can shorten the time by employing a more brilliant source, which can provide more photon flux. Initially, our camera system had a vertical rotation axis for the sample to make full use of the linear polarization of synchrotron beams to reduce the scattering background. However, this meant that the sample had to stand up.8 For the sake of convenience for movie experiments, a new design with a horizontal rotation axis, which fits the observation of dynamical changes in the specimens, was subsequently introduced. Figure 4 shows one typical example of a recording movie; successive copper images during ion-exchange reaction. This gives a comprehensive visual description of the exchange starting from the surface of each resin particle and then the ions being absorbed inside. Since it is possible to obtain separate images for specific elements by choosing the primary X-ray energy, one can obtain multielement movies by quick switching of the monochromator. We have studied the growing patterns during electrodeposition for a binary metal system by recording an X-ray fluorescence movie,13 although such research was carried
out only for a simple system composed of only one kind of metallic ion.16,17 This is significant for future developments, because most of themicroscopes currently available are not sensitive to differences in chemical composition. One could conceive of analyzing X-ray energy by using a CCD camera in such multielement applications. Operating a CCD camera in photon-counting mode and frequently reading out the signals is a useful way of X-ray energy-dispersive imaging,18 while normal accumulation is feasible and efficient as long as the processing images obtained above and below the absorption edges are effective. To perform even more rapid movies on the order of 1-30 ms with multielement
images, further developments would be required for both technologies.
(16) Sawada Y.; Dougherty A.; Gollub J. P. Phys. Rev. Lett. 1986, 56, 1260. (17) Sander L. M. Sci. Am. 1987, (Jan), 81-88. (18) Tsunemi H.; Hiraga J.; Yoshita K.; Miyata E.; Ohtani M. Nucl. Instrum. Method A 2000, 439, 592.
Received for review May 22, 2002. Accepted November 15, 2002.
ACKNOWLEDGMENT The authors thank Prof. Atsuo Iida (Photon Factory) for his valuable advice on the present work. This work was performed with the approval of the Photon Factory Program Advisory Committee (Proposals 1999G-085, 2001G-144, 2002S2-003) and was partly supported by the Active Nano-Characterization and Technology Project, Special Coordination Funds of the MEXT, Japanese Government.
AC025793H
Analytical Chemistry, Vol. 75, No. 2, January 15, 2003
359
