Elemental X-ray images obtained by grazing-exit electron probe microanalysis (GE-EPMA) Kouichi Tsuji,*{a,b Rik Nullens,a Kazuaki Wagatsumab and Rene� E. Van Griekena aMicro- and Trace Analysis Center Mitac, University of Antwerp (UIA), B-2610 Antwerpen, Belgium bInstitute for Materials Research, Tohoku University, Katahira 2-1-1, Aoba, Sendai, 980-8577, Japan. E-mail: tsuji@imr.tohoku.ac.jp Received 1st July 1999, Accepted 2nd September 1999 A new method, grazing-exit electron probe microanalysis (GE-EPMA), was studied.Only X-rays emitted from the near-surface layer are measured at grazing-exit angles (e.g. v0.5�), whereas, with conventional EPMA, X-rays emitted from deep positions are also measured. Therefore, X-ray spectra with low background are obtained by GE-EPMA. Here, elemental mapping by GE-EPMA is shown for the Ærst time. It was found that surface-sensitive elemental X-ray images were obtained for a thin Au Ælm deposited on a Si wafer. The problems that occur at boundaries of different heights are discussed.Furthermore, it was difÆcult to recognize elemental distributions of Si, S, Ca, Na and Fe for aerosols deposited on a Si wafer in noisy X-ray images when using conventional EPMA; however, clear X-ray images were obtained under grazing-exit conditions. Introduction Electron probe microanalysis (EPMA) is a powerful technique for elemental analysis of small regions, and it is applied abundantly for microanalysis of metals, semiconductor devices, biochemical samples, aerosol particles, etc.1,2 Elemental mapping, which yields two-dimensional information about elemental distributions in samples, is also obtained by scanning electron beams.In this scanning mode, X-ray spectra are measured frequently with an energy-dispersive X-ray detector (EDX) for each measured point. In many cases, elemental X-ray images of near-surface layers (v5 nm) are needed. However, characteristic and Bremsstrahlung X-rays produced in depth are also detected in conventional EPMA.This is the reason why strong continuous X-rays are observed in EPMA spectra. When aerosols and biochemical samples deposited on sample carriers are analyzed, we need only the information from the deposited samples. However, Xrays emitted from the sample carrier are also measured, as shown in Fig. 1(a). The primary electrons penetrate into the sample carrier and are scattered; therefore, characteristic and Bremsstrahlung X-rays are produced from the broadened regions.These unnecessary X-rays inØuence the quality of Xray images. For example, when an Al foil is used as a sample carrier for the analysis of aerosols, strong characteristic and continuous X-rays emitted from the Al foil will obviously disturb the determination of Al and other trace elements in aerosols.3 We have developed a new method: grazing-exit EPMA (GEEPMA). 4 In conventional EPMA, electron-induced X-rays are measured at large take-off (exit) angles, e.g. 40�. In contrast, Xrays are measured at grazing-exit angles of less than 1� in GEEPMA. Under grazing-exit conditions, a phenomenon similar to X-ray total reØection is observed.5 Only the X-rays emitted from near-surface layers of about 5 nm in thickness are detected at grazing-exit angles of less than 0.5�, as shown in Fig. 1(b). Therefore, the X-ray spectra are obtained with low background,6 and the detection limits are also improved.7 In this paper, elemental X-ray images obtained at grazing-exit angles are shown for the Ærst time.A thin Au Ælm and aerosol samples are measured. Aerosols containing toxic elements are dangerous to humans and aerosols also inØuence the global climate.8 In order to understand the type and origin of aerosols, in addition to elemental composition, the elemental analysis of individual aerosol particles is necessary. For this purpose, individual particle analysis by computer-controlled automated EPMA is a promising method.9,10 Experimental The experimental details have been described elsewhere.3,6,7 The experiments were carried out by EPMA (Superprobe-733, JEOL, Tokyo, Japan) with an ultra-thin window EDX (Link Pentafet Model 5373, Oxford Instruments, UK).The EPMA apparatus was operated at an acceleration voltage of 20 kV and at a beam current of 1 nA under a vacuum of less than 261025 Torr. A slit of 0.5 mm in width was attached parallel to the surface of the sample on the front of the EDX.The sensitive area of the EDX used was 30 mm2, of which only an area of 0.566 mm was uncovered in this set-up. The distance between the sample and the EDX was about 100 mm. A brass triangular attachment having an inclination of 45� was placed on the sample holder. The exit angle was controlled by tilting {On leave from Tohoku University. Fig. 1 A simple illustration of the interaction volume of electron beams for a single particle on a Øat sample carrier. The regions observed by conventional EPMA (a) and GE-EPMA (b) are shaded (a). In conventional EPMA, X-rays from both the particle and the sample carrier are detected.In contrast, only the X-rays emitted from the particle are measured by GE-EPMA. J. Anal. At. Spectrom., 1999, 14, 1711±1713 1711 This Journal is # The Royal Society of Chemistry 1999this sample holder. An incident angle of 90� for electron beams is desirable for microanalysis; however, this set-up was difÆcult because the electron source and the EDX were Æxed.Therefore, an incident angle of the electron beam of approximately 45� was used in this work with the triangular attachment. The samples were a thin Au Ælm and aerosol particles. The thin Au Ælm was deposited to a thickness of about 100 nm on a Si wafer by an evaporation method. Aerosol particles were collected on a Si wafer at the campus of the University of Antwerp (UIA), Belgium. Air was sucked with a rotary vacuum pump into a multiple-oriÆce impactor (Berner-type, Hauke, Austria); Ænally, spots (#1 mm in diameter) of deposited atmospheric particles were obtained on a Si substrate.Results and discussion X-ray images of thin Ælm The thin Au Ælm deposited on the Si wafer was intentionally scratched by a needle. The width of the scratch was about 33 mm. X-ray images of AuMa and Si Ka taken at an exit angle of 6� are shown in Fig. 2(a) and (b), respectively. X-rays of Au Ma were practically not observed at the scratched line, where the Au layer was removed.X-rays of Si Ka were also observed on the region covered with the Au Ælm, because the exit angle was large enough to observe Si Ka emitted from the Si wafer under the thin Au layer. The two boundaries of the scratched line are clearly shown in the Au Ma X-ray image [Fig. 2(a)]. However, the right boundary of this line is not very clear in the Si Ka X-ray image [Fig. 2(b)], and the left boundary is even less sharp, in this case where the X-rays were detected from the right side of the sample.This phenomenon was not observed when the sample of the scratched line was placed parallel to the X-ray detector. Similar results have been reported for grazing-exit X-ray Øuorescence analysis.11 This result can be explained by using the simple model shown in Fig. 3. Since X-ray images were taken at a small exit angle of 6�, SiKa X-rays produced under the Au layer near the left boundary can also be detected through the side of the Au layer.In addition, electron beams can excite Si Ka X-rays through the side of the Au layer, because the incident angle of the electron beam is 45� in this work. The opposite phenomenon occurs near the right boundary: the Si Ka X-rays emitted near the boundary are absorbed by the Au layer. This phenomenon occurs when boundaries of different heights are observed at small take-off angles. Fig. 2(c) and (d) shows elemental X-ray images taken at a grazing angle of about 0.8�.In order to calculate the critical angle for detection of characteristic X-rays in GE-EPMA, it is assumed that the characteristic X-rays impinge upon the surface of the substrate at the grazing-incidence angles,12 because the principles of microscopic reversibility and Lorentz reciprocity are applied.5 is estimated by the following equation: h&1:65=E|ÖZr=AÜ1=2 Ö1Ü where E(keV) is the energy of characteristic X-rays, Z is the atomic number, A is the atomic weight, and r (g cm23) is the density of the substrate.13 For Au Ma (2.1 keV) from a Au layer (19.3 g cm23), h is estimated to be 2.2�.Since the exit angle (0.8�) for the X-ray image of Au Ma [Fig. 2(c)] was less than this critical angle (2.2�), the AuMa intensities decreased in comparison with those in Fig. 2(a). As shown in Fig. 2(d), Si Ka X-rays are practically not observed at regions covered with the Au Ælm.This indicates that surface-sensitive elemental X-ray images can be obtained by GE-EPMA. In the Si Ka Xray image in Fig. 2(d), a similar problem to that in Fig. 2(b) is found. The left boundary of the scratched line is obscure, and Si Ka X-rays emitted from the Si wafer covered with the Au layer near this boundary are also detected. This result can be explained by using the model in Fig. 3 again. In the grazing-exit arrangement, the absorption of Si Ka X-rays in the Au upper layer increases because of the longer path.Therefore, the right boundary of the scratched line in the Si Ka X-ray image is clearer than the left boundary, as shown in Fig. 2(d). This problem at the boundary should require attention when X-ray images are observed at grazing-exit angles. X-ray images of aerosols Fig. 4(a) shows a secondary electron image of aerosols of 2± 10 mm sizes. For this sample, elemental X-ray images of Si Ka, S Ka, CaKa, NaKa and Fe Ka were taken at different exit angles.The X-ray images taken at 8� are shown in Fig. 4(b)± (f). As shown in Fig. 4(b), since aerosols were collected on the Si wafer, Si Ka X-rays emitted from the Si wafer are too strong to allow observation of the Si Ka X-rays from the aerosols. Xray images of other elements are also not clear. White dots appear at the points where the corresponding elements do not exist. This is due to high background intensities caused by Bremsstrahlung X-rays in the Si wafer.The elemental X-ray images taken at a grazing-exit angle of 0.5� are shown in Fig. 4(g)±(k). The X-ray image of Si Ka for aerosols is clearly found, as shown in Fig. 4(g). The critical angle of Si Ka (1.74 keV) from a Si (2.2 g cm23) wafer is approximately 1.0�. Hence, Si Ka X-rays emitted from the surface of the Si wafer are drastically decreased at exit angles of less than 1.0�. This is the reason why clear X-ray images of aerosols were obtained at an exit angle of 0.5�. Similarly, X-ray Fig. 2 Elemental X-ray images of AuMa [(a), (c)] and Si Ka [(b), (d)] for a Au thin layer deposited on a Si wafer.Part of the Au layer was removed by scratching it. The exit angles are 6� [(a), (b)] and 0.8� [(c), (d)]. X-rays were detected from the right side of these pictures. X-ray images were measured at 2006140 pixels for a total measuring time of about 1 h at each exit angle at an area of 95665 mm. Fig. 3 Model of a Au thin Ælm on a Si wafer to explain the Si Ka X-ray image near the boundary.Si Ka X-rays produced under the Au layer near the boundary are detected through the side of the Au layer as shown in the left side, while the Si Ka X-rays emitted from the left of the boundary are absorbed by the Au layer as shown in the right side. 1712 J. Anal. At. Spectrom., 1999, 14, 1711±1713images of S Ka, CaKa, NaKa and Fe Ka are clearly obtained, as shown in Fig. 4(h)±(k). From these elemental X-ray images, SiO2 particles (in the bottom left and in the upper right), a CaSO4 particle (in the bottom right) and Fe-rich particles (in the center) are identiÆed in Fig. 4(a). Conclusions Elemental X-ray images obtained by GE-EPMA were demonstrated. The X-rays emitted from the depth of the carrier cannot be detected under grazing-exit conditions because of refraction effects at the vacuum±sample interface;4,6,7 therefore, X-ray spectra with low background are measured at each measured point. As a result, clear X-ray images are obtained with low noise levels.GE-EPMA is useful especially for the observation of particles deposited on a Øat sample support. Absorption in the particle itself has to be considered for quantitative analysis of large particles.6 The problem of X-ray imaging for near boundaries of different heights was pointed out. Acknowledgement One of the authors (K. Tsuji) was Ænancially supported by Japan Society for the Promotion of Science (JSPS) and by a Grant-in-Aid (11650828) from the Ministry of Education, Science, Sports and Culture.Part of this work was supported by the Belgium OfÆce for ScientiÆc, Technical and Cultural Affairs under contract MN/10/01. The authors thank Dr. Z. Spolnik and Professor J. Zhang for the sample preparation of aerosols, and also thank Dr. J. Injuk for useful suggestions. References 1 S. J. B. Reed, Electron Microprobe Analysis, Cambridge University Press, Cambridge, UK, 1993. 2 P. Duncumb, J. Anal. At.Spectrom., 1999, 14, 357. 3 C. Ro, J. Osa�n and R. Van Grieken, Anal. Chem., 1999, 71, 1521. 4 K. Tsuji, K. Wagatsuma, R. Nullens and R. Van Grieken, Anal. Chem., 1999, 71, 2497. 5 R. S. Becker, J. A. Golovchenko and J. R. Patel, Phys. Rev. Lett., 1983, 50, 153. 6 K. Tsuji, Z. Spolnik, K. Wagatsuma, R. Nullens, J. Zhang and R. Van Grieken, Spectrochim. Acta, Part B, 1999, 54, 1251. 7 K. Tsuji, Z. Spolnik, K. Wagatsuma, R. Nullens and R. Van Grieken, Mikrochim. Acta, in press. 8 Atmospheric Particles, ed. R. M. Harrison and R. E. Van Grieken, Wiley, Chichester, 1998. 9 H. Van Malderen, S. Hoornaert and R. Van Grieken, Environ. Sci. Technol., 1996, 30, 489. 10 W. Jambers and R. Van Grieken, Environ. Sci. Technol., 1997, 31, 1525. 11 T. Noma and A. Iida, Rev. Sci. Instrum., 1994, 65, 837. 12 L. G. Parratt, Phys. Rev., 1954, 95, 359. 13 R. Klockenka»mper, Total-reØection X-Ray Fluorescence Analysis, Wiley, New York, 1997. Paper 9/05301H Fig. 4 Secondary electron image (a) of aerosol particles deposited on a Si wafer, and elemental X-ray images of Si Ka [(b), (g)], S Ka [(c), (h)], Ca Ka [(d), (i)], Na Ka [(e), (j)] and Fe Ka [(f), (k)]. The exit angle was 8� [(b)±(f)] and 0.5� [(g)±(k)]. X-ray images were measured at 3006200 pixels for a total measuring time of about 1h at each exit angle at an area of 65645 mm. J. Anal. At. Spectrom., 1999,