Anal. Chem. 1999, 71, 2497-2501
Grazing Exit Electron Probe Microanalysis for Surface and Particle Analysis Kouichi Tsuji,*,† Kazuaki Wagatsuma,‡ Rik Nullens,† and Rene´ E. Van Grieken†
Micro- and Trace Analysis Center, Department of Chemistry, University of Antwerp (UIA), B-2610 Antwerpen, Belgium, and Institute for Materials Research, Tohoku University, Katahira-2-1-1, Aoba, Sendai, 980-8577 Japan
We developed a new method of grazing exit electron probe microanalysis (GE-EPMA) and applied it to analyze both Si surfaces and Mg-salt particles. In conventional EPMA, X-rays are detected at an exit (takeoff) angle of approximately 45°. Therefore, when particles collected on a sample carrier are analyzed by EPMA, the X-rays from both the particles and the carrier are detected, although we need only the X-rays emitted from the particle itself. In contrast to this, the X-rays are detected at grazing exit angles in GE-EPMA. The X-rays emitted from deep inside of the sample are not detected under grazing exit conditions, and only X-rays emitted from the surface and the particle are measured. It was found that surface-sensitive analysis of a Si wafer was possible with low background at grazing exit angles. The intensity ratio of O Kr to Si Kr increased near zero degrees, indicating that the Si wafer is covered with a native Si oxide. Moreover, Mg Kr X-rays from a Mg-salt particle, which was deposited on the Si wafer, were detected with a small Si Kr intensity at grazing exit angles of less than 0.5°. By decreasing the exit angle to less than zero, only the top of the particle was observed; therefore, GE-EPMA measurement would make it possible to investigate the surface layer of one particle. Electron probe microanalysis (EPMA) is a common method for elemental analysis of small regions.1,2 Hitherto, many applications of EPMA in the fields of material science, biological chemistry, and environmental chemistry have been developed. Microanalysis is also possible by using other probes, such as laser and ion and micro X-ray beams; however, the minimum probe diameters of these are in the order of micrometers, while an electron beam can be easily focused to the diameter size of less than 1 µm and in nanometer order. In addition to microanalysis, surface analysis is also required in many cases. The smaller the size of the semiconductor device, the larger the importance of the surface. However, as is well-known, electron-induced X-rays are emitted not only from the surface but also from inside of the †
University of Antwerp. Tohoku University. * Corresponding author. On leave from Tohoku University. E-mail: tsuji@ imr.tohoku.ac.jp. (1) Small, J. A. Electron-Induced X-ray Emission. In Handbook of X-ray Spectrometry; Van Grieken, R. E., Markowicz, A. A., Eds.; Marcel Dekker: New York, 1993. (2) Reed, S. J. B. Electron Microprobe Analysis, 2nd ed.; Cambridge University Press: Great Britain, 1993.
Figure 1. Schematic drawing of the conventional EPMA (a) and grazing-exit EPMA (b). A triangular attachment was used to measure X-rays at grazing exit angles. The Si-wafer samples were placed on the triangular ramp for investigation.
solid sample. This is because electrons are elastically or inelastically scattered in the solid sample, and then the scattered electrons excite X-rays again in deeper positions, and in addition, the emitted X-rays excite other atoms. The depth resolution depends on the electron energy, the X-ray energy, and the density of the sample.1 In the case of the electron energy of 20 kV and the Cu target, Cu LR X-rays are emitted from depths of approximately 1-2 µm.1 In conventional EPMA, the exit angle (or takeoff angle) is normally fixed at approximately 45°, as shown in Figure 1a. Therefore, X-rays emitted from both the surface and inside of the substrate are detected. To achieve surface analysis by EPMA, only the X-rays emitted from near the surface should be detected. In the field of X-ray fluorescence (XRF) analysis, the X-ray totalreflection phenomenon has been used to achieve surface-sensitive analysis on the flat substrate. This method is known as total reflection XRF (TXRF).3,4 When the X-ray beam impinges at the grazing angle less than the critical angle for total reflection on the flat substrate, such as a Si wafer, the penetration depth is restricted to approximately 3 nm.4,5 In addition, the primary X-rays are totally reflected, yielding low background intensity.3 Therefore, TXRF is a surface-sensitive method and is applied for trace analysis of contamination on Si wafers.4 Moreover, as a method related to TXRF, grazing exit (or emission) X-ray fluorescence (GEXRF) has been proposed.6-8 In GEXRF, the X-ray total-reflection phenomenon is utilized in the process of the X-ray detection; that is, fluorescent X-rays are measured at grazing exit angles. Similar
‡
10.1021/ac990075p CCC: $18.00 Published on Web 05/21/1999
© 1999 American Chemical Society
(3) Yoneda, Y.; Horiuchi, T. Rev. Sci. Instrum. 1971, 42, 1069. (4) Klockenka¨mper, R. Total-reflection X-ray Fluorescence Analysis; John Wiley: New York, 1997. (5) Parratt, L. G. Phys. Rev. 1954, 95, 359. (6) Sasaki, Y.; Hirokawa, K. Appl. Phys. A 1991, 50, 397. (7) Noma, T.; Miyata, H.; Ino, S. Jpn. J. Appl. Phys., Part 2 1992, 31, L900. (8) de Bokx, P. K.; Urbach, H. P. Rev. Sci. Instrum. 1995, 66, 15.
Analytical Chemistry, Vol. 71, No. 13, July 1, 1999 2497
to TXRF, surface-sensitive analysis is carried out by GEXRF.6-9 Although the excitation method in EPMA (by electrons) is different from that in GEXRF (by X-rays), the technique of detecting X-rays at the grazing exit angle can also be applied to EPMA. Actually, Ino et al.10 have described total-reflection-angle X-ray spectroscopy (TRAXS) in reflection high-energy electron diffraction (RHEED) experiments. The electron beam was impinged upon the sample surface at the grazing incident angle, and the X-rays were detected at the grazing takeoff angle. Therefore, the RHEED-TRAXS method is sensitive to the surface of the sample. Usui et al.11 also carried out TRAXS measurement in scanning electron microscopy (SEM), where the electron beam also impinged at a grazing angle of several degrees. RHEEDTRAXS and SEM-TRAXS make it possible to analyze the surface structure and observe the sample surface, respectively; however, they are not suitable for a microanalysis because the area irradiated by electrons under grazing-incidence conditions is considerably large. We performed EPMA measurements under grazing exit conditions to achieve micro- and surface analysis, as shown in Figure 1b. The grazing exit EPMA (GE-EPMA) method was applied for the surface analysis of a Si wafer and for particle analysis on a Si substrate. In the case of particle analysis, the only information we need is the X-rays from the particle. The X-rays from the substrate merely form the background and frequently influence the measurement of X-rays from the particle. The X-rays from the substrate would decrease under the grazing-exit condition, resulting in a low background. EXPERIMENTAL SECTION The experiments were carried out by EPMA (Superprobe-733, JEOL, Tokyo, Japan) with both a wavelength-dispersive X-ray detector (WDX) and an ultrathin-window energy-dispersive X-ray detector (EDX, Link Pentafet model 5373, Oxford Instruments, England). Since the conventional solid-state detector (SSD) has a beryllium window about 8 µm in thickness, the sensitivity for detection of soft X-rays is reduced. In contrast, it is possible to detect light elements, such as carbon and oxygen, with the ultrathin-window SSD. Light elements are the main components of particles in the atmosphere; therefore, the windowless or ultrathin-type SSD is suitable for particle analysis. The sensitive area of the SSD used was 30 mm2, corresponding to approximately 6 mm in diameter. This indicates that the solid angle for the X-ray detection is approximately 3.4° at the distance of 100 mm from the sample. For the purpose of the strict GE-EPMA experiment, the small solid angle is desirable. It was possible to change the distance between the sample and the SSD from 30 to 100 mm. The longer this distance, the better the angular resolution. Therefore, the solid angle for X-ray detection would be improved by taking away the X-ray detector from the sample. The angular dependence of X-rays was detected at the distance of 100 mm in this work. In the EPMA used, the X-rays are normally measured at the exit angle of approximately 45°, as shown in Figure 1a. The exit (9) Tsuji, K.; Sato, S.; Hirokawa, K. J. Appl. Phys. 1994, 76, 7860. (10) Hasegawa, S.; Ino, S.; Yamamoto, Y.; Daimon, H. Jpn. J. Appl. Phys., Part 2 1985, 24, L387. (11) Usui, T.; Aoki, Y.; Kamei, M.; Takahashi, H.; Morishita, T.; Tanaka, S. Jpn. J. Appl. Phys. 1991, 30, L2032.
2498 Analytical Chemistry, Vol. 71, No. 13, July 1, 1999
angle can be changed from 35 to 90° by tilting a sample holder. Thus, a custom-made brass triangular attachment having an inclination of 45° was put on this sample holder, as shown in Figure 1b. This attachment made it possible to measure the X-rays at the grazing exit angle of around zero. It was difficult to exactly determine zero degrees by measuring the X-ray intensity, because the SSD had a large sensitive area. In this work, the angle of zero degrees was roughly determined as the angle where Si KR could not be measured. The EPMA apparatus was operated at an accelerated voltage of 10 kV and at a beam current of 1 nA under the vacuum of 7.5 × 10-6 Torr in the EDX measurements. The incident angle of the electron beam was approximately 45°. X-ray spectra were measured for 60 s with the SSD for each exit angle, which was changed from -2.5 to 5.5° with a step angle of 0.25°. The takeoff angle was changed from high angle to low angle in one direction. This was necessary in order to obtain more precise adjustment of the exit angle. In the WDX measurements, X-rays were monochromatized with a TAP through a slit of 0.5 mm and then detected with a flow-proportional. After a 2θ scan at a speed of 0.6 Å/min, the area intensity of the characteristic X-ray peak was evaluated. The beam current was adjusted to 20 nA in order to enhance the signal intensity in the WDX measurements. The samples used were flat Si substrates and Mg-salt particles. Si substrates were cut from a Si wafer to the size of 5 × 10 mm2 and then cleaned in acetone solution by using an ultrasonic cleaning machine. The Mg-standard HNO3 solution (5 ppm), in a 10 µL aliquot, was dropped on the Si substrate using a micropipet and a disposable tip and dried in air. At first, the solution spread to the whole substrate, and then it gradually condensed as it was dried. Finally, one Mg-salt (Mg(NO3)2) particle with a diameter of approximately 100 µm was obtained. The Si-wafer samples were placed on the triangular ramp for investigation. RESULTS AND DISCUSSION X-ray Intensity at Grazing Exit Angles. Before the experimental results are shown, we describe theoretical treatment of the X-ray intensity under grazing exit conditions. It is possible to calculate the intensity of X-rays that are emitted from the substrate surface at grazing exit angles. The calculation procedure has been described elsewhere.7,12 The Si KR intensity from a Si wafer (density, 2.2 g/cm3) is shown as a function of exit angle in Figure 2. The X-ray intensity at grazing exit angles was calculated with the assumption that Si KR X-rays impinged on the Si wafer at grazing incident angles. This assumption is reasonable because microscopic reversibility (or reciprocity theorem) applies in the X-ray region.13 The X-ray intensity drastically increases at an exit angle of 1°, and then it slowly increases as the exit angle increases, as shown in Figure 2. The critical angle (θc, in degrees) of total reflection can be approximately calculated according to the following equation: θc ) 0.094λF1/2. Here λ (Å) is the wavelength of observed X-rays and F (g/cm3) is the density of the substrate.6 In the case of Si KR (7.126 Å) on a Si substrate, the critical angle is estimated to be 0.99°, which corresponds to the angle where the Si KR X-rays suddenly increase in Figure 2. Figure 2 indicates that Si KR X-rays are detected much less in grazing exit angles (12) Tsuji, K.; Hirokawa, K. J. Appl. Phys. 1994, 75, 7189. (13) Becker, R. S.; Golovchenko, J. A.; Patel, J. R. Phys. Rev. Lett. 1983, 50, 153.
Figure 4. Illustration of a particle on flat substrate. The variable h is the height of the particle and d is the distance between the particle and the edge of the Si wafer. Electron-induced X-rays were detected by WDX and EDX-SSD. The surface layer of the particle was measured at an exit angle less than zero degrees.
Figure 2. Si KR intensity calculated for the Si wafer as a function of exit angle.
Figure 3. Gross intensities of Si KR and Mg KR as functions of exit angle. Si KR and Mg KR X-rays were detected with WDX-EPMA for Si substrate and Mg-salt particle, respectively.
less than 1°; therefore, it is expected that GE-EPMA would give us an X-ray spectrum with low background intensity. We can also estimate the observing depth of GE-EPMA in the same way as was done for GE-XRF or TXRF. It is approximately 3 nm for the Si wafer at the grazing exit angles below θc, indicating that GEEPMA is a surface-sensitive method. Furthermore, when we measure particles collected on the substrate, X-rays emitted from the substrate should be avoided or reduced. This would be achieved by measuring X-rays at grazing exit angles. The grazing exit condition would be especially suitable for the analysis of particles on a flat surface. GE-EPMA with WDX. The control of the exit angle is important for analyzing the data in GE-EPMA. Since the WDX system of the EPMA apparatus used originally had an entrance slit 0.5 mm wide, it is possible to limit the exit angle to a small solid angle. The combination of the GE-EPMA method and the WDX system is superior, in principle, because any special attachments are not necessary to perform the GE-EPMA. Figure 3 shows the angular dependence of X-ray intensities of Si KR and Mg KR, which were measured at the different positions of the Si wafer and the Mg-salt particle, respectively. Si KR steeply increased at first and then slowly increased as the exit angle
increased. This feature is similar to that of the calculated curve shown in Figure 2. The critical angle (θc) of total reflection of Si KR on the Si wafer is calculated to be approximately 0.99°, as described in the previous section. In the experimental curve shown in Figure 3, Si KR steeply increased at 1°, which roughly agrees with the theoretical critical angle. This suggests that a phenomenon similar to X-ray total reflection occurs on the Si wafer in the X-ray detection process in GE-EPMA. In contrast to this, a clear angular dependence was not obtained for Mg KR intensity, as shown in Figure 3. In the TXRF analysis of thin films on the flat substrate, a characteristic angular dependent curve is measured, which is caused by an interference effect between incident X-rays and reflected X-rays.14 However, in the case of a large particle, a clear angular dependence is not observed, because interference effects cannot occur.15 Moreover, the Mg KR X-rays were still detected in the exit angles less than zero. This result would be explained by using an illustration shown in Figure 4. Since the Mg-salt particle was large (∼100 µm) and considerably high, it was possible to detect the X-rays emitted from the particle at the exit angles below an angle of zero. In the measurement using the WDX, the electron beam current was adjusted to 20 nA to enhance the detected X-ray intensity. However, this current was so strong that the particle was damaged during measurement. Another drawback of the WDX-type GEEPMA method is that the measurement is time-consuming. Surface Analysis of Si by GE-EPMA with EDX. A Si surface was measured by GE-EPMA. The distance between the sample and the EDX-SSD was fixed at 100 mm. Figure 5a shows the Si KR gross and net intensities as a function of exit angle. The Si KR intensity increased with increasing exit angle. The background intensity at the Si KR energy was calculated by subtracting the Si KR net intensity from the Si KR gross intensity. Finally, the ratio of the net Si KR intensity to the background intensity was calculated, as shown in Figure 5b. The ratio has a peak at approximately 0.5°, where the X-rays emitted near surface of the Si wafer are very sensitively detected with low background. It is well-known that the X-rays emitted from deep inside of the Si wafer cannot be detected at grazing exit angles below the critical angle for total reflection.10 The result shown in Figure 5b indicates that a surface-sensitive EPMA with small background can also be possible under the grazing-exit condition. (14) de Boer, D. K. G. Phys. Rev. B: Condens. Matter 1991, 44, 498. (15) Klockenka¨mper, R.; von Bohlen, A. Spectrochim. Acta, Part B 1989, 44, 461.
Analytical Chemistry, Vol. 71, No. 13, July 1, 1999
2499
Figure 5. Typical exit-angle-dependent curves measured for a Si wafer with EDX. (a) Si KR gross and net intensities and (b) intensity ratio of net Si KR intensity to background intensity, which was estimated by subtracting net intensity from gross intensity. (c) C KR and O KR gross intensities and (d) intensity ratios of C KR/Si KR and O KR/Si KR.
Figure 5c shows the X-ray intensities of O KR and C KR as a function of exit angle. As is well-known, the Si wafer has a native oxide layer on the surface; in addition, contamination of carbon exists on the Si surface. The ratios of O KR and C KR intensities to Si KR intensity are shown in Figure 5d. It is found that both ratios increased at an exit angle near zero, indicating that carbon and oxygen exist on the surface of the Si wafer. Mg-Salt Particle Analysis. The same Mg-salt particle measured with WDX was measured by EDX. The net intensities of Si KR, Mg KR, and O KR are shown as functions of exit angle in Figure 6a. The electron beam impinged on the Mg-salt particle; however, the Si KR X-rays from the substrate were also measured. This is because the electron beam passed through the Mg-salt particle and then bombarded the Si substrate. Compared with the experimental result shown in Figure 5a, the Si KR intensity fluctuated with respect to the exit angles. This is probably because the position of the Mg-salt particle, which had a heterogeneous thickness, was adjusted for each exit angle. This procedure was necessary because the sample position shifted upon tilting of the sample holder. The Si KR intensity decreased as the exit angle decreased and was not observed below zero degrees. In contrast to this, Mg KR and O KR were still detected at angles just below zero. This result can also be explained by using Figure 4. Since 2500 Analytical Chemistry, Vol. 71, No. 13, July 1, 1999
the Mg-salt particle has a large thickness, it is possible to observe the top of the particle from the position of the SSD below zero degrees. The intensity ratio of Mg KR to Si KR is shown with solid squares in Figure 6b. It is clear that X-rays from the particle are measured with low Si KR intensity and low background in the exit angle range below zero degrees. In Figure 4, the distance (d) between the particle position and the edge of the Si substrate was approximately 2 mm. The X-rays of Mg KR and O KR were undetectable in the angles below - 2.5 degrees. Therefore, a simple calculation, h ) dtan(2.5), gives us the height (h) of the particle as approximately 90 µm. If both the distance d and exit angle were accurately measured, the height h would be more accurately estimated. The intensity ratio of O KR to Mg KR is also shown with empty circles in Figure 6b. The O KR X-rays originate from both the Mg-salt particle and the Si wafer, which has a native oxide layer. Since Si KR X-rays were observed only in the angles above zero degrees, O KR X-rays emitted from the Si-oxide layer would also be detected in these angles. Therefore, the O KR intensity above zero degrees includes the X-rays from the particle and the Si native oxide. However, in the angle range below zero degrees, O KR X-rays should be emitted only from the particle, because the X-rays from Si were not observed below zero degrees. In this angular
O KR to Mg KR slightly increased. This might indicate that the outside of the Mg-salt particle was covered with an oxide layer. The GE-EPMA measurement in the angles below zero would make it possible to investigate the surface layer of one large particle. CONCLUSION We proposed the GE-EPMA method in this paper. This method was applied to surface analysis and particle analysis. The X-rays emitted from inside of the sample were not detected under grazing exit conditions and only X-rays emitted from the surface (∼3 nm theoretically) were measured. Therefore, GE-EPMA is suitable for surface analysis. For the same reason, GE-EPMA could also become a powerful method to analyze the particle, deposited on a flat substrate. The surface layer of large particles is also measured by GE-EPMA in the exit angle less than zero degrees. The WDX-type EPMA is, in principle, directly suitable for detecting the X-rays at grazing exit angles, because the solid angle for X-ray detection is limited by the slit of the WDX-EPMA. However, it was difficult to adjust the sample surface in parallel to the slit of the WDX under grazing exit conditions in this work. As a next step, we will optimize the WDX-type GE-EPMA and undertake to measure GE-EPMA using an aperture or a slit in front of the EDX-SSD. The sample holder was manually tilted, with a 0.25° step, in this work. To analyze the surface layer of particles in detail, more precise control of exit angle would be necessary.
Figure 6. (a) Net intensities of Si KR, Mg KR, and O KR measured with EDX for Mg-salt particle deposited on the Si wafer. (b) Intensity ratios of Mg KR/Si KR and O KR/Mg KR.
range, elemental analysis of the particle would be possible with low background. At the exit angle of -2.5°, the intensity ratio of
ACKNOWLEDGMENT One of the authors (K. Tsuji) was financially supported by Japan Society for the Promotion of Science (JSPS). The authors would like to thank Dr. Z. Spolnik, Dr. J. Osa´n, and Prof. J. Zhang for useful discussions.
Received for review January 27, 1999. Accepted March 29, 1999. AC990075P
Analytical Chemistry, Vol. 71, No. 13, July 1, 1999
2501
