- Email: [email protected]
Progress and application of the Tohoku microbeam system
NIM B Beam Interactions with Materials & Atoms
Nuclear Instruments and Methods in Physics Research B 260 (2007) 55–64 www.elsevier.com/locate/nimb
Progress and application of the Tohoku microbeam system S. Matsuyama a,*, K. Ishii a, H. Yamazaki a, Y. Kikuchi a, K. Inomata a, Y. Watanabe a, A. Ishizaki a, R. Oyama a, Y. Kawamura a, T. Yamaguchi a, G. Momose a, M. Nagakura a, M. Takahashi a, T. Kamiya b a
Department of Quantum Science and Energy Engineering, Tohoku University, Aramaki-aza-Aoba 01, Aoba-ku, Sendai 980-8579, Japan b Takasaki Advanced Radiation Research Institute, Japan Atomic Energy Agency, Watanuki 1233, Takasaki 370-1292, Japan Available online 15 February 2007
Abstract A microbeam system was constructed at the Tohoku University Dynamitron laboratory and optimization of the system was performed. Parasitic field contamination from tungsten carbide slit chips and an annular Si surface barrier detector limited the beam spot size to 2 · 2 lm2. By replacing these components, the parasitic field contamination of the system was reduced and the performance of the microbeam system was remarkably improved. A measured beam spot size of 0.4 · 0.4 lm2 at a beam current of several tens of pA has been accomplished. The results obtained using the grid shadow method show that a beam spot size of less than 0.4 lm is obtainable in the low-current regime. For easy tuning and operation of the microbeam system, a human–machine interface (HMI) was developed based on the user participatory design concept. The HMI has been accepted favorably by users and is used in routine operations. The analysis system has been re-developed and is applicable to simultaneous in-air/in-vacuum PIXE, RBS, SEM, and STIM analyses and 3D lCT. The system is now applied for studies in biological, environmental, and other fields. 2007 Elsevier B.V. All rights reserved. PACS: 07.78.+s; 41.75. i; 41.85.Lc Keywords: Microbeam; Parasitic field contamination; Grid shadow method; HMI; PIXE; RBS; STIM; 3D l-CT
1. Introduction High-energy ion microbeams are powerful analytical tools that combine various ion beam analysis techniques such as PIXE, RBS, STIM, and SEM [1–3]. They are also attractive as a direct lithographic technique. We have developed a microbeam analysis system for biological applications with sub-micrometer resolution. The primary purpose was to develop a 3D l-CT, in which a microbeam is used as a monoenergetic point X-ray source. The second was to develop a microbeam analysis system for biological samples. The microbeam line was installed in July 2002; the system produced a beam spot size of 2 · 2 lm2 with a beam current of 10 pA [4]. Both the beam size and beam current were insufficient for ion beam analyses. Optimization and *
Corresponding author. Tel.: +81 22 217 7933. E-mail address: [email protected] (S. Matsuyama).
0168-583X/$ - see front matter 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.nimb.2007.01.277
modification of the system were performed and the analysis system was re-developed. After optimization and modification of the system, the beam performance was remarkably improved and a multimodal analysis system was developed. To give the system higher usability both for novices and experienced users, a human–machine interface (HMI) was also developed. This paper presents a description of recent progress of the microbeam system and its recent applications. 2. Microbeam system The microbeam system was designed to achieve submicrometer beam sizes and was developed in collaboration with Tokin Machinery Corp. [4]. The system is connected to the 4.5 MV Dynamitron accelerator at Tohoku University. The Dynamitron accelerator is equipped with a high-current duo-plasmatron ion source. For microbeam
56
S. Matsuyama et al. / Nucl. Instr. and Meth. in Phys. Res. B 260 (2007) 55–64
formation, the brightness of 10 pA/lm2 mrad2 from the accelerator (3 MeV proton) is high enough, but its energy stability is insufficient. A high-resolution energy analysis system is installed upstream of the microbeam line to confine the energy variation so as to achieve sub-micrometer beam sizes. The microbeam line comprises a quadrupole doublet and three slit systems: micro-slit (MS), divergencedefining slit (DS), and baffle slit (BS). These components are mounted on a heavy rigid support with vibration isolation. The micro-slit defines the object size and comprises two wedge-shaped slits, which are made from 4-mm-diameter tungsten carbide cylinders whose spacing is increasing linearly from 0–150 lm. The divergence-defining slit and the baffle slit are four-pole type. The divergence-defining slit defines beam divergence into the quadrupole doublet and is placed 3 m downstream of the micro-slit; the baffle slit reduces scattered beams. The slit openings of the micro-slit and the divergence-defining slit are adjusted remotely using stepping motors, but the baffle slit requires no frequent adjustment and is instead controlled manually. As for the quadrupole lens, the bore radius is 5 mm with yoke dimensions of 60 mm long · 220 mm outside diameter. An iron piece of the magnet with poles and a yoke was cut as one body using a computer (numerically) controlled machine to reduce mechanical irregularities that might cause sextupole contamination [5]. Tolerance of the pole dimensions is smaller than 2 lm. The quadrupole lenses are set on stages that can precisely and separately adjust translation, tilt, and rotation. The lens and the stages are mounted on another stage which can be used to adjust the working distance without changing alignment. We set the working distance and the objective distance at 26 cm and 6 m, respectively. The demagnification factors are 9.2 and 35.4 for horizontal and vertical directions, respectively. The focused microbeam is scanned across a target two-dimensionally using an electrostatic beam scanner. The scanner is located downstream of the quadrupole
doublet and the maximum scanning area is larger than 1 · 1 mm2 for 3 MeV proton beams. 3. Analysis system improvements The target chamber is a rectangular box and is applicable to either in-vacuum or in-air analysis without changing the main body. For multimodal analysis, two X-ray detectors and three charged particle detectors can be mounted simultaneously [6–8]. For 3D l-CT, an X-ray CCD camera (C8800X; Hamamatsu photonics), a rotating sample stage and a target to produce X-rays are also mounted on this chamber [9–11]. In the present study, further improvements were introduced to the target chamber, as shown in Fig. 1. Two X-ray detectors are attached to both sides of the chamber and are not shown. For in-vacuum analysis, hydrogen is analyzed using offaxis STIM. In a previous study, a Si-PIN photodiode was used for both off-axis STIM and direct STIM measurements [8]. Superior energy resolution of the Si-PIN photodiode is efficient for direct STIM measurement [12]. However, the small size of the detector occasionally restricted detection efficiency in hydrogen analyses. For that reason, an ion-implanted Si detector with a larger sensitive area of 50 mm2 was installed. Detection efficiency of the system is now three times higher than that of the previous system. The scattering angle is 28, which is sufficient to separate proton peaks that are scattered from hydrogen and from other elements, even in the thickness of 50 lm of organic films. For the in-air system, on/off-axis STIM for simultaneous density mapping with PIXE and RBS is newly available. It will be useful for damage monitoring in biological cell analysis and for correction of X-ray self-absorption in samples. A thin scattering foil (C or Al) is placed 20 mm downstream of the sample. Scattered protons are detected by a Si-PIN photodiode which is set at 28 with respect
Fig. 1. Side cut view of the target chamber.
S. Matsuyama et al. / Nucl. Instr. and Meth. in Phys. Res. B 260 (2007) 55–64
to the beam axis. The detector is well collimated to reduce energy broadening by kinematics. A compact secondary electron detector is newly installed for secondary electron imaging (see Fig. 1). In the previous system, secondary electrons were detected by a plastic scintillator with a photomultiplier. The detector was not compact, and was therefore not able to cooperate with other detector systems. A compact ceramic channel electron multiplier (CEM, MD-502; Amptek Inc.) is very compact (1.9 cm diameter · 3.8 cm long) and is not obstructive to other detector systems. 4. Microbeam system performance After the installation, performance of the system was tested by a micro-PIXE analysis for a Cu mesh (1000 lines/in.). The lower limit of the beam spot size was 2 lm, which is much larger than the design value [4]. The main reason for the limitation is parasitic field contamination from microbeam line components. Especially, the tungsten carbide slit chips cause strong field contamination. X-ray maps from Cu mesh (1000 lines/in.) measured with the baffle slit open and with the baffle slit closed are shown in Fig. 2. Although the mesh image was clearly apparent when the baffle slit was opened, the beam was defocused and was shifted when the baffle slit was closed. This is an effect of parasitic field existing at the baffle slit. The baffle slit has no an electric driving mechanism which might cause electric or magnetic fields. Two of the slit chips used in the baffle slit were magnetized strongly and attracted paper clips. The field contamination was therefore ascribed to the tungsten carbide slit chips. These slit chips were degaussed, but field strength of 1 lT remained, as measured using a magnetometer. The degaussed chips were reattached and identical measurements were carried out. Although the beam is not defocused by closing the baffle slit, a beam shift of a few micrometers was apparent. The field strength of 1 lT affects microbeam formation. This magnetism is caused by cobalt element that is used as a binder in the slit chip. Although the tungsten carbide itself is not magnetized, cobalt shows strong magnetism and is magnetized easily [13]. The slit chip contains about 10% cobalt. The magne-
Fig. 2. Elemental distribution maps from Cu mesh (1000 lines/in.) measured with the baffle slit open and with the baffle slit closed.
57
tization arises from manufacturing processes and should be eliminated. Finally, the slit chips of the divergence-defining slit and the baffle slit, whose parasitic field strongly affects microbeam formation, were replaced by the other chips whose elemental concentration of cobalt is less than 10% that of previous chips. After replacement of the chips, defocusing and beam shifting do not occur in any slit position. Effects of parasitic field contamination was measured using the grid shadow method [5]. In the grid shadow method, the beam is focused on the image plane, where a fine mesh grid is placed, by a single quadrupole lens and casts shadow pattern on a scintillator downstream. The shadow pattern is influenced strongly by lens aberrations and parasitic field contaminations of the microbeam system. In the present study, a Au mesh (2000 lines/in.) was placed on the image plane and a shadow pattern was obtained on a ZnS(Ag) scintillator located 53 mm downstream of the grid. Beam divergence must be maximized to increase measurement sensitivity of field contamination. The maximum beam divergence of our microbeam system is ±0.5 mrad, which is restricted by the beam pipe (6 mm inner diameter) through the quadrupoles. Because the divergence-defining slit and the baffle are four-pole slits, actual beam divergence is reduced to ±0.35 mrad. In this study, we set the divergence to ±0.35 mrad, which is slightly smaller than that from the accelerator. The sensitivity of the grid shadow method is also increased by reducing the angle between the divergence axis of quadrupole and the grid bar. Smaller angles produce fewer grid bar shadows and higher sensitivity, especially for higher order field contamination. The angle at which a shadow pattern shows a shadow of 4–10 grid bars is the optimum angle to investigate sextupole field contamination which degrades microbeam quality significantly [14]. In this study, we set the angle as 1.75. Fig. 3 shows theoretical grid shadow patterns with 0–0.2% sextupole field contamination superimposed onto the quadrupole field calculated by the beam optics computer codes, PRAM and OXTRACE [3], with identical beam conditions. The theoretical shadow pattern shows that the minimum detectable limit corresponds to approximately 0.05% for sextupole field contamination. Fig. 4 shows the measured grid shadow patterns for the two quadrupole lenses with a 3 MeV proton beam. The beam is screened by the beam scanner in the divergence plane. Therefore, the number of grid bars is fewer than the theoretical pattern. Although measured shadow patterns in the horizontal focus show slight deformation resulting from sextupole field contamination, shadow patterns in the vertical focus are not deformed. Both lenses show a similar trend. In comparison with the calculated results, sextupole field contamination in the horizontal focusing is estimated as less than 0.1%. Contamination of the skew sextupole field is less than 0.1%. Therefore, it is difficult to observe deformation in the measured pattern, which implies that sextupole field contamination in the horizontal plane (skew sextupole field contamination in
58
S. Matsuyama et al. / Nucl. Instr. and Meth. in Phys. Res. B 260 (2007) 55–64
Fig. 3. Theoretical grid shadow patterns calculated by PRAM and OXTRACE.
Fig. 4. Measured grid shadow patterns of the two quadrupole lenses for vertical (X) and horizontal (Y) focusing.
the vertical plane) remains. This contamination source might be the lens itself or another part of the microbeam
line. This source can be checked by rotating the lens by 90. If the sextupole contamination remains in the horizontal plane, then the contamination originates outside the lens. If the sextupole contamination changes to the vertical plane, then the contamination source is the lens itself. Unfortunately, our lens system cannot rotate by 90, so we were unable to identify the contamination source. One possibility is that the cancellation of excitation current of the quadrupole is not perfect. This lens was designed so that the excitation current from pole to pole is canceled by the return current. However, no cancellation is made in the upper part of the lens where terminals exist. Because the coils consist of only 22 turns, such effects might cause the problem. To reduce the effect of the contamination field, the 1st quadrupole uses horizontal focusing and the 2nd lens uses vertical focusing (divergence–convergence); the scanner is located downstream. Then, the beam spot size was measured by scanning the beam across mesh samples (Ni and Au mesh, 2000 lines/ in.) and measuring X-rays. Horizontal and vertical line
Fig. 5. Horizontal and vertical beam spot sizes versus object sizes.
S. Matsuyama et al. / Nucl. Instr. and Meth. in Phys. Res. B 260 (2007) 55–64
profiles were fitted by symmetric double Gaussian convolution and beam spot sizes were obtained. The line profiles were well reproduced using the symmetric double Gaussian convolution, which implies that the beam profile can be assumed to be of Gaussian shape. Fig. 5 shows horizontal and vertical beam spot sizes versus object sizes. These measurements were carried out in the same beam divergence of 0.2 mrad which corresponded to our normal divergence in the high-current regime for ion beam analyses. Beam spot sizes diminished concomitant with the object slit sizes and were better than those that were predicted from the calculations. A beam spot size of 0.4 · 0.4 lm2 was obtained at an object size of 30 · 10 lm2 with a beam current of several tens of pA. Beam size estimations using the deconvolution method are strongly dependent on the mesh quality and are sometimes overestimated in the range less than 1 lm because we assume the mesh edges to be very sharp [6]. The cross sections of mesh grid bars are trapezoidal or round to some degree. Therefore, the beam spot size will be better than 0.4 · 0.4 lm2. Fig. 6 shows secondary electron images of Ni mesh (1000 lines/in.) measured using an electron microscope and the microbeam system with object size of 25 · 10 lm2 and beam divergence of 0.2 mrad. The secondary electron image measured using the microbeam system clearly represents the steps of the bars and round shape of the corner, which corresponds to the image obtained using the electron microscope. This fact shows that resolution of better than 0.4 lm is obtained and is consistent with that obtained using deconvolution method. These measurements were carried out with the divergence of 0.2 mrad for the high-current regime. The result of the grid shadow implies that a better beam spot size will be obtained in a low-current regime where the object size and beam divergence are one-tenth lower. Even in the same beam condition mentioned above, the beam spot size of 1 lm is a lower limit in the actual experimental setup, where X-ray detectors, RBS detector, and other detectors are set in a chamber. The grid shadow method revealed the source of this problem. The grid shadow patterns in this situation are shown in Fig. 7. Strong sextupole or higher order field contamination are observed. By removing the annular Si surface barrier detector (TC-
Fig. 6. Secondary electron images of Ni mesh (1000 lines/in.) measured using an electron microscope and the microbeam system.
59
Fig. 7. Measured grid shadow patterns with a parasitic field from the annular detector.
019-150-100; Ortec) positioned 40 mm upstream of the sample position, field contamination was eliminated. The grid shadow pattern in Fig. 7 was measured without applying bias voltage to the annular detector. The holder of the annular detector is made of plastic, which adds no field. Therefore, the parasitic field was ascribed to the magnetic field of the annular Si surface barrier detector itself. In fact, the annular detector imparted a field strength of 10 lT, as measured using a magnetometer. Elements that show strong magnetism are used in the detector body. The RBS detector was replaced by another annular ionimplanted Si detector (ANPD 300-19-100RM; Canberra). The detector gave a weaker magnetic field, which was also measured using a magnetometer. Following change, the field contamination of the system was reduced to a normal level and performance of the microbeam system was improved greatly, even in the actual analysis. Components that must be set near the beam path should have their magnetism eliminated; the grid shadow method is very effective to clarify the source of parasitic field. 5. Development of a control system The microbeam system is controlled via a programmable controller (FA-M3; Yokogawa Electric Corp.) and a personal-computer-based measurement instrument (WE7000; Yokogawa Electric Corp.) using LabVIEW based software. The power supplies of the analyzing magnet and quadrupole lenses are stabilized within 10 5 and are driven by 16-bit digital-to-analog converters (DACs) in WE7000. The control parameters for operation are stored in the memory space of the programmable controller. Another personal computer used for the user interface (HMI-PC) can only refer to these parameters. During the start-up process, control software retrieves the parameters for a programmable controller and can send commands to change the parameters. This framework was adopted to deal with the situation of a possible computer hang-up and to enable the use of different HMI-PCs.
60
S. Matsuyama et al. / Nucl. Instr. and Meth. in Phys. Res. B 260 (2007) 55–64
Fig. 8. Main and Sub-window of microbeam control software.
The main control window is arranged to neatly fit a small notebook computer display. The control window for each device can be opened by clicking the icon. Fig. 8 shows an example of the main and sub-windows of the HMI-PC. This graphical feature enables novice users to become accustomed to using the system with minimum training effort. A display palette is provided to adjust the two parameters simultaneously when control of more than two related parameters is necessary (e.g. control of the quadrupole doublet) (Fig. 8). The display contents such as the beam currents and images from CCD cameras would automatically switch to the currently focused location according to the change of Faraday cups and viewers, which can reduce the operational load. This control software has been designed based on user requirements and requests. Because the control software was constructed on a LabVIEW system, which is characterized by superior flexibility in alternation, it is rather easy to make modifications or to customize it for better usability. However, it is difficult to maintain the software when many users try to improve the software merely to suit their own preferences. Because the appropriate arrangement of the display might change according to experimental conditions, the interface contents should not be customizable based solely on a specific user’s opinion. In the present study, the user interface was elaborated based on users’ opinions [15] and information of their respective operational histories [16] for actual microbeam analysis operation. In analysis operations, the following procedure is required. First, direct STIM was carried out defining the analysis area. Subsequently, simultaneous PIXE/RBS/off-axis STIM measurements were executed, after which direct STIM was employed once more for the same position. In this procedure, an operator must change
many parameters on different sub-windows depending on the task. The points of the interface design elaboration for our microbeam system can be summarized as follows: I. Combined use of subjective and objective informationAs stated previously, if the subjective opinions of specific users are adopted, the interface might become better for a specific user, but not for the other users. Therefore, objective information (operational history) was used to reveal the possible problems and specific characteristics of operation for different experimental conditions. II. Iterative elaboration processThe elaboration process should be repeated several times until users are satisfied. It is rather difficult to determine final design specifications based on a limited number of evaluations. In the present study, the elaboration process for interface design continued for more than one year after the system was put to actual experimental use. Fig. 9 shows the sub-window developed for actual microbeam analyses. The important finding in our design elaboration process is that the users’ opinions are not always correct and users are aware of the potential problems only after it is revealed through analysis of the operational history. This fact strongly implies the importance of the combined use of subjective and objective information for evaluation of the user interface. The resultant interface design was put to use for daily operation. Two facts indicate the effectiveness of our design methodology: the operational history shows considerable reduction of the number of window focus operations, thus implying that the control window contents were appropri-
S. Matsuyama et al. / Nucl. Instr. and Meth. in Phys. Res. B 260 (2007) 55–64
61
ately re-arranged for experimental purposes; in addition, questionnaire results indicated users’ acceptance, with improved subjective evaluation of users after the elaboration process. Those results imply that our pragmatic approach for interface elaboration is effective for realizing better usability. 6. Recent applications of the microbeam system Microbeam analysis system at Tohoku University is routinely applied for studies in various fields. Typical applications are described below. 6.1. Aerosol study
Fig. 9. Sub-window developed for actual microbeam analyses.
Elemental analysis of atmospheric aerosols is useful for source identification and for the study of their formation mechanisms. These studies were carried out by analyzing bulk samples thereby averaging over many single particles. Therefore, analysis of single aerosol particles is superior to bulk analysis. Our system can quantify single aerosol particles containing from hydrogen to heavy metals and reveal chemical composition of these particles [8]. The microbeam system was applied to analysis of yellow sand dust particles. These particles were impacted on a thin polycarbonate
Fig. 10. Elemental maps of yellow sand dust particles. Scanning area is 100 · 100 lm2.
62
S. Matsuyama et al. / Nucl. Instr. and Meth. in Phys. Res. B 260 (2007) 55–64
film developed especially for this purpose. Yellow sand dust particles from the Asian continent sometimes cause turbid conditions in Japan, especially in spring. Typical elemental images of yellow sand dust particles are shown in Fig. 10. The elemental distribution of Al is similar to that of Si and O, originating mainly from soil dust of alumino-silicate particles. Distribution of hydrogen is visible and resembles that of carbon. These particles are sometimes deformed by mixing with anthropogenic aerosols. The system is useful to study these deformation processes. 6.2. Cell analysis In-air simultaneous PIXE, RBS, and a consecutive STIM analyses are powerful tools for analyses of single cells and tissue sections in biological and biomedical research. We have developed an in-air micro-PIXE system in collaboration with the Japan Atomic Energy Agency (JAEA), Takasaki [17,18] and have applied it to various cells [19–21]. Our in-air analysis system is also adjusted specifically for this purpose [7]. Although in-air analysis reduces damage during beam irradiation [22], morphological changes occasionally occur by beam irradiation, and cell thickness or density is reduced to 60% of the initial value [7]. In this case, the cell density should be monitored simultaneously with analysis. In addition, correction of Xray self-absorption should be done according to the irradiation dose. On/off-axis STIM facilitates performance of simultaneous density imaging in PIXE and RBS analyses. Fig. 11 shows elemental maps of cells and density maps measured using the new in-air system. Distributions of P, S, Cl, and K elements are clarified and correspond to the
Fig. 12. Optical image of pattern written using microbeams.
density. Changes in cell density during beam irradiation can be monitored and will be effective for cell analyses. 6.3. Proton beam writing Proton beam writing was realized by scanning the microbeam with a corresponding scanning pattern translated from a bitmap file. Pattern-to-pattern spacing was accomplished by moving the beam at the fastest beam scanner speed. Details of the system will be described elsewhere. A typical pattern written on a Mylar film is shown in Fig. 12. Although this pattern was written before removing the field contamination from the annular detector, the shape of Japan’s islands is clearly visible. We are planning to use this technique for efficient analyses through combination with a STIM image.
Fig. 11. Elemental maps and density distribution of cells. Scanning area is 100 · 100 lm2.
S. Matsuyama et al. / Nucl. Instr. and Meth. in Phys. Res. B 260 (2007) 55–64
63
Fig. 13. Cross sectional view of an ant’s head.
6.4. 3D l-CT The 3D l-CT technique comprises a point X-ray source, a rotating sample stage and a high-speed X-ray CCD camera. Details of 3D l-CT and applications are described in detail in [9–11]. The microbeam is used as a monoenergetic l-X-ray source by bombarding pure metals. The 2D transmission data are obtained by rotating the sample encapsulated in a polycarbonate tube (1000 lm inner diameter). The 3D images were reconstructed from the obtained transmission data. 3D images of a living ant’s head were obtained using this system with 6 lm spatial resolution. Typical CT images of the ant’s head measured using Ti K X-rays are shown in Fig. 13. The 3D l-CT uses monoenergetic low-energy X-rays. Therefore, the contrast of the image for small insects is superior to that of other CT systems. Furthermore, our system can readily change the X-ray energy to obtain a better contrast depending on the elements contained in the insects. This system can measure small insects of 2 mm in-vivo and will be useful for biological studies. 7. Conclusion A microbeam system was constructed at the Tohoku University Dynamitron laboratory and optimization of the system was undertaken. Minimum beam spot size of the microbeam was limited at 2 · 2 lm2 by parasitic field contamination. This parasitic field contamination was ascribed to the tungsten carbide slit chips and the annular Si surface barrier detector in the case of RBS analysis. The parasitic field contamination of the system was greatly reduced by replacing these components, as confirmed by the grid shadow method, and the microbeam system performance was improved. The minimum beam spot size of 0.4 · 0.4 lm2 at a beam current of several tens of pA is obtained, which is the optimum expected from the accelerator performance. The results obtained using the grid
shadow method indicate that a beam spot size of less than 0.4 lm is obtainable in the low-current regime. For easy tuning and operation of the microbeam system, a human–machine interface (HMI) for the control system was developed based on the user participatory design concept. The operational log of a skilled operator was used to realize the HMI, which is both user-friendly and has higher usability both for novices and experienced users. The new HMI has been accepted favorably by users and is used now in routine operations. While improving the microbeam system, simultaneous in-air/in-vacuum PIXE, RBS, SEM, and STIM analyses, in addition to 3D l-CT, were developed and are now being applied to biological, environmental, and other fields. Acknowledgements This study was partly supported by Grants-in-Aid for Scientific Research, (S) No. 13852017, (B) No. 18360450, (C) No. 16560731, and a Grant-in-Aid for Scientific Research in Priority Areas under Grant No. 14048213 from the Ministry of Education, Culture, Sports, Science and Technology, Japan. The authors would like to thank Prof. D.N. Jamieson of the University of Melbourne, for his valuable suggestions and advice in reducing field contamination and introducing the PRAM and OXTRACE computer codes. The authors would like to acknowledge the assistance of Mr. R. Sakamoto and M. Fujisawa for maintenance and operation of the Dynamitron accelerator. The authors would like to thank Mr. K. Komatsu, T. Nagaya and C. Akama for their assistance in constructing the microbeam and target system. References [1] F. Watt, G.W. Grime, Principal and Applications of High-energy Ion Microbeams, Adam Hilger, Bristol, 1987. [2] S.A.E. Johansson, J.L. Campbell, K.G. Malmqvist, Particle-Induced X-ray Emission Spectrometry (PIXE), John Wiley & Sons, NY, 1995.
64
S. Matsuyama et al. / Nucl. Instr. and Meth. in Phys. Res. B 260 (2007) 55–64
[3] M.B.H. Breese, D.N. Jamieson, P.J.C. King, Materials Analysis using Nuclear Microprobe, John Wiley & Sons, NY, 1995. [4] S. Matsuyama, K. Ishii, H. Yamazaki, R. Sakamoto, M. Fujisawa, Ts. Amartaivan, Y. Ohishi, M. Rodoriguez, A. Suzuki, T. Kamiya, M. Oikawa, K. Arakawa, N. Matsumoto, Nucl. Instr. and Meth. B 210 (2003) 59. [5] D.N. Jamieson, G.J.F. Legge, Nucl. Instr. and Meth. B 29 (1987) 544. [6] S. Matsuyama, K. Ishii, H. Yamazaki, Y. Barbotteau, Ts. Amartaivan, D. Izukawa, K. Hotta, K. Mizuma, S. Abe, Y. Oishi, M. Rodriguez, A. Suzuki, R. Sakamoto, M. Fujisawa, T. Kamiya, M. Oikawa, K. Arakawa, H. Imaseki, N. Matsumoto, Int. J. PIXE 14 (1–2) (2004) 1. [7] S. Matsuyama, K. Ishii, S. Abe, H. Ohtsu, H. Yamazaki, Y. Kikuchi, Ts. Amart aivan, K. Inomata, Y. Watanabe, A. Ishizaki, Y. Barbotteau, A. Suzuki, T. Yamaguchi, G. Momose, H. Imaseki, Int. J. PIXE 15 (1–2) (2005) 41. [8] S. Matsuyama, K. Ishii, H. Yamazaki, Y. Kikuchi, Ts. Amartaivan, S. Abe, K. Inomata, Y. Watanabe, A. Ishizaki, R. Oyama, Y. Kawamura, A. Suzuki, G. Momose, T. Yamaguchi, H. Imaseki, Int. J. PIXE 15 (3–4) (2005) 257. [9] K. Ishii, S. Matsuyama, H. Yamazaki, Y. Watanabe, T. Yamaguchi, G. Momose, Ts. Amartaivan, A. Suzuki, Y. Kikuchi, W. Galster, Int. J. PIXE 15 (3–4) (2005) 111. [10] K. Ishii, S. Matsuyama, H. Yamazaki, Y. Watanabe, Y. Kawamura, T. Yamaguchi, G. Momose, Y. Kikuchi, A. Terakawa, W. Galster, Nucl. Instr. and Meth. B 249 (2006) 726. [11] K. Ishii, S. Matsuyama, Y. Watanabe, Y. Kawamura, T. Yamaguchi, R. Oyama, G. Momose, A. Ishizaki, H. Yamazaki, Y. Kichuchi, Nucl. Instr. and Meth. A 571 (2007) 64.
[12] G. Deves, S. Matsuyama, Y. Barbotteau, K. Ishii, R. Ortega, Rev. Sci. Instr. 77 (2006) 056102. [13] T.W. Penrice, Int. J. Refract. Met. H 15 (1997) 113. [14] D.N. Jamieson, J. Zhu, Y. Mao, R. Lu, Z. Wang, J. Zhu, Nucl. Instr. and Meth. B 104 (1995) 86. [15] A. Dix, J. Finlay, R. Beal, People and Computers VII, in: Proceedings of HCI 92, Cambridge University Press, 1992, p. 429. [16] J. Nielsen, Usability Engineering (1993). [17] S. Matsuyama, K. Ishii, A. Sugimoto, T. Satoh, K. Gotoh, H. Yamazaki, S. Iwasaki, K. Murozono, J. Inoue, T. Hamano, S. Yokota, T. Sakai, T. Kamiya, R. Tanaka, Int. J. PIXE 8 (2–3) (1998) 203. [18] T. Sakai, T. Hamano, T. Hirao, T. Kamiya, K. Murozono, J. Inoue, S. Matsuyama, S. Iwasaki, K. Ishii, Nucl. Instr. and Meth. B 136–138 (1998) 390. [19] K. Ishii, A. Sugimoto, A. Tanaka, T. Satoh, S. Matsuyama, H. Yamazaki, C. Akama, Ts. Amartaivan, H. Endoh, Y. Oishi, H. Yuki, S. Sugihara, M. Satoh, T. Kamiya, T. Sakai, K. Arakawa, M. Saidoh, S. Oikawa, Nucl. Instr. and Meth. B 181 (2001) 448. [20] A. Sugimoto, K. Ishii, S. Matsuyama, T. Satoh, K. Gotoh, H. Yamazaki, C. Akama, Int. J. PIXE 9 (3–4) (1999) 151. [21] A. Tanaka, K. Ishii, Y. Komori, S. Matsuyama, T. Satoh, K. Gotoh, H. Yamazaki, C. Akama, Int. J. PIXE 12 (3–4) (2002) 79. [22] H. Komori, K. Mizuma, K. Ishii, H. Yamazaki, S. Matsuyama, Ts. Amartaivan, Y. Ohishi, M. Rodoriguez, A. Suzuki, M. Satoh, T. Kamiya, T. Sakai, T. Satoh, M. Oikawa, K. Arakawa, Int. J. PIXE 124 (3–4) (2004) 75.
Nuclear Instruments and Methods in Physics Research B 260 (2007) 55–64 www.elsevier.com/locate/nimb
Progress and application of the Tohoku microbeam system S. Matsuyama a,*, K. Ishii a, H. Yamazaki a, Y. Kikuchi a, K. Inomata a, Y. Watanabe a, A. Ishizaki a, R. Oyama a, Y. Kawamura a, T. Yamaguchi a, G. Momose a, M. Nagakura a, M. Takahashi a, T. Kamiya b a
Department of Quantum Science and Energy Engineering, Tohoku University, Aramaki-aza-Aoba 01, Aoba-ku, Sendai 980-8579, Japan b Takasaki Advanced Radiation Research Institute, Japan Atomic Energy Agency, Watanuki 1233, Takasaki 370-1292, Japan Available online 15 February 2007
Abstract A microbeam system was constructed at the Tohoku University Dynamitron laboratory and optimization of the system was performed. Parasitic field contamination from tungsten carbide slit chips and an annular Si surface barrier detector limited the beam spot size to 2 · 2 lm2. By replacing these components, the parasitic field contamination of the system was reduced and the performance of the microbeam system was remarkably improved. A measured beam spot size of 0.4 · 0.4 lm2 at a beam current of several tens of pA has been accomplished. The results obtained using the grid shadow method show that a beam spot size of less than 0.4 lm is obtainable in the low-current regime. For easy tuning and operation of the microbeam system, a human–machine interface (HMI) was developed based on the user participatory design concept. The HMI has been accepted favorably by users and is used in routine operations. The analysis system has been re-developed and is applicable to simultaneous in-air/in-vacuum PIXE, RBS, SEM, and STIM analyses and 3D lCT. The system is now applied for studies in biological, environmental, and other fields. 2007 Elsevier B.V. All rights reserved. PACS: 07.78.+s; 41.75. i; 41.85.Lc Keywords: Microbeam; Parasitic field contamination; Grid shadow method; HMI; PIXE; RBS; STIM; 3D l-CT
1. Introduction High-energy ion microbeams are powerful analytical tools that combine various ion beam analysis techniques such as PIXE, RBS, STIM, and SEM [1–3]. They are also attractive as a direct lithographic technique. We have developed a microbeam analysis system for biological applications with sub-micrometer resolution. The primary purpose was to develop a 3D l-CT, in which a microbeam is used as a monoenergetic point X-ray source. The second was to develop a microbeam analysis system for biological samples. The microbeam line was installed in July 2002; the system produced a beam spot size of 2 · 2 lm2 with a beam current of 10 pA [4]. Both the beam size and beam current were insufficient for ion beam analyses. Optimization and *
Corresponding author. Tel.: +81 22 217 7933. E-mail address: [email protected] (S. Matsuyama).
0168-583X/$ - see front matter 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.nimb.2007.01.277
modification of the system were performed and the analysis system was re-developed. After optimization and modification of the system, the beam performance was remarkably improved and a multimodal analysis system was developed. To give the system higher usability both for novices and experienced users, a human–machine interface (HMI) was also developed. This paper presents a description of recent progress of the microbeam system and its recent applications. 2. Microbeam system The microbeam system was designed to achieve submicrometer beam sizes and was developed in collaboration with Tokin Machinery Corp. [4]. The system is connected to the 4.5 MV Dynamitron accelerator at Tohoku University. The Dynamitron accelerator is equipped with a high-current duo-plasmatron ion source. For microbeam
56
S. Matsuyama et al. / Nucl. Instr. and Meth. in Phys. Res. B 260 (2007) 55–64
formation, the brightness of 10 pA/lm2 mrad2 from the accelerator (3 MeV proton) is high enough, but its energy stability is insufficient. A high-resolution energy analysis system is installed upstream of the microbeam line to confine the energy variation so as to achieve sub-micrometer beam sizes. The microbeam line comprises a quadrupole doublet and three slit systems: micro-slit (MS), divergencedefining slit (DS), and baffle slit (BS). These components are mounted on a heavy rigid support with vibration isolation. The micro-slit defines the object size and comprises two wedge-shaped slits, which are made from 4-mm-diameter tungsten carbide cylinders whose spacing is increasing linearly from 0–150 lm. The divergence-defining slit and the baffle slit are four-pole type. The divergence-defining slit defines beam divergence into the quadrupole doublet and is placed 3 m downstream of the micro-slit; the baffle slit reduces scattered beams. The slit openings of the micro-slit and the divergence-defining slit are adjusted remotely using stepping motors, but the baffle slit requires no frequent adjustment and is instead controlled manually. As for the quadrupole lens, the bore radius is 5 mm with yoke dimensions of 60 mm long · 220 mm outside diameter. An iron piece of the magnet with poles and a yoke was cut as one body using a computer (numerically) controlled machine to reduce mechanical irregularities that might cause sextupole contamination [5]. Tolerance of the pole dimensions is smaller than 2 lm. The quadrupole lenses are set on stages that can precisely and separately adjust translation, tilt, and rotation. The lens and the stages are mounted on another stage which can be used to adjust the working distance without changing alignment. We set the working distance and the objective distance at 26 cm and 6 m, respectively. The demagnification factors are 9.2 and 35.4 for horizontal and vertical directions, respectively. The focused microbeam is scanned across a target two-dimensionally using an electrostatic beam scanner. The scanner is located downstream of the quadrupole
doublet and the maximum scanning area is larger than 1 · 1 mm2 for 3 MeV proton beams. 3. Analysis system improvements The target chamber is a rectangular box and is applicable to either in-vacuum or in-air analysis without changing the main body. For multimodal analysis, two X-ray detectors and three charged particle detectors can be mounted simultaneously [6–8]. For 3D l-CT, an X-ray CCD camera (C8800X; Hamamatsu photonics), a rotating sample stage and a target to produce X-rays are also mounted on this chamber [9–11]. In the present study, further improvements were introduced to the target chamber, as shown in Fig. 1. Two X-ray detectors are attached to both sides of the chamber and are not shown. For in-vacuum analysis, hydrogen is analyzed using offaxis STIM. In a previous study, a Si-PIN photodiode was used for both off-axis STIM and direct STIM measurements [8]. Superior energy resolution of the Si-PIN photodiode is efficient for direct STIM measurement [12]. However, the small size of the detector occasionally restricted detection efficiency in hydrogen analyses. For that reason, an ion-implanted Si detector with a larger sensitive area of 50 mm2 was installed. Detection efficiency of the system is now three times higher than that of the previous system. The scattering angle is 28, which is sufficient to separate proton peaks that are scattered from hydrogen and from other elements, even in the thickness of 50 lm of organic films. For the in-air system, on/off-axis STIM for simultaneous density mapping with PIXE and RBS is newly available. It will be useful for damage monitoring in biological cell analysis and for correction of X-ray self-absorption in samples. A thin scattering foil (C or Al) is placed 20 mm downstream of the sample. Scattered protons are detected by a Si-PIN photodiode which is set at 28 with respect
Fig. 1. Side cut view of the target chamber.
S. Matsuyama et al. / Nucl. Instr. and Meth. in Phys. Res. B 260 (2007) 55–64
to the beam axis. The detector is well collimated to reduce energy broadening by kinematics. A compact secondary electron detector is newly installed for secondary electron imaging (see Fig. 1). In the previous system, secondary electrons were detected by a plastic scintillator with a photomultiplier. The detector was not compact, and was therefore not able to cooperate with other detector systems. A compact ceramic channel electron multiplier (CEM, MD-502; Amptek Inc.) is very compact (1.9 cm diameter · 3.8 cm long) and is not obstructive to other detector systems. 4. Microbeam system performance After the installation, performance of the system was tested by a micro-PIXE analysis for a Cu mesh (1000 lines/in.). The lower limit of the beam spot size was 2 lm, which is much larger than the design value [4]. The main reason for the limitation is parasitic field contamination from microbeam line components. Especially, the tungsten carbide slit chips cause strong field contamination. X-ray maps from Cu mesh (1000 lines/in.) measured with the baffle slit open and with the baffle slit closed are shown in Fig. 2. Although the mesh image was clearly apparent when the baffle slit was opened, the beam was defocused and was shifted when the baffle slit was closed. This is an effect of parasitic field existing at the baffle slit. The baffle slit has no an electric driving mechanism which might cause electric or magnetic fields. Two of the slit chips used in the baffle slit were magnetized strongly and attracted paper clips. The field contamination was therefore ascribed to the tungsten carbide slit chips. These slit chips were degaussed, but field strength of 1 lT remained, as measured using a magnetometer. The degaussed chips were reattached and identical measurements were carried out. Although the beam is not defocused by closing the baffle slit, a beam shift of a few micrometers was apparent. The field strength of 1 lT affects microbeam formation. This magnetism is caused by cobalt element that is used as a binder in the slit chip. Although the tungsten carbide itself is not magnetized, cobalt shows strong magnetism and is magnetized easily [13]. The slit chip contains about 10% cobalt. The magne-
Fig. 2. Elemental distribution maps from Cu mesh (1000 lines/in.) measured with the baffle slit open and with the baffle slit closed.
57
tization arises from manufacturing processes and should be eliminated. Finally, the slit chips of the divergence-defining slit and the baffle slit, whose parasitic field strongly affects microbeam formation, were replaced by the other chips whose elemental concentration of cobalt is less than 10% that of previous chips. After replacement of the chips, defocusing and beam shifting do not occur in any slit position. Effects of parasitic field contamination was measured using the grid shadow method [5]. In the grid shadow method, the beam is focused on the image plane, where a fine mesh grid is placed, by a single quadrupole lens and casts shadow pattern on a scintillator downstream. The shadow pattern is influenced strongly by lens aberrations and parasitic field contaminations of the microbeam system. In the present study, a Au mesh (2000 lines/in.) was placed on the image plane and a shadow pattern was obtained on a ZnS(Ag) scintillator located 53 mm downstream of the grid. Beam divergence must be maximized to increase measurement sensitivity of field contamination. The maximum beam divergence of our microbeam system is ±0.5 mrad, which is restricted by the beam pipe (6 mm inner diameter) through the quadrupoles. Because the divergence-defining slit and the baffle are four-pole slits, actual beam divergence is reduced to ±0.35 mrad. In this study, we set the divergence to ±0.35 mrad, which is slightly smaller than that from the accelerator. The sensitivity of the grid shadow method is also increased by reducing the angle between the divergence axis of quadrupole and the grid bar. Smaller angles produce fewer grid bar shadows and higher sensitivity, especially for higher order field contamination. The angle at which a shadow pattern shows a shadow of 4–10 grid bars is the optimum angle to investigate sextupole field contamination which degrades microbeam quality significantly [14]. In this study, we set the angle as 1.75. Fig. 3 shows theoretical grid shadow patterns with 0–0.2% sextupole field contamination superimposed onto the quadrupole field calculated by the beam optics computer codes, PRAM and OXTRACE [3], with identical beam conditions. The theoretical shadow pattern shows that the minimum detectable limit corresponds to approximately 0.05% for sextupole field contamination. Fig. 4 shows the measured grid shadow patterns for the two quadrupole lenses with a 3 MeV proton beam. The beam is screened by the beam scanner in the divergence plane. Therefore, the number of grid bars is fewer than the theoretical pattern. Although measured shadow patterns in the horizontal focus show slight deformation resulting from sextupole field contamination, shadow patterns in the vertical focus are not deformed. Both lenses show a similar trend. In comparison with the calculated results, sextupole field contamination in the horizontal focusing is estimated as less than 0.1%. Contamination of the skew sextupole field is less than 0.1%. Therefore, it is difficult to observe deformation in the measured pattern, which implies that sextupole field contamination in the horizontal plane (skew sextupole field contamination in
58
S. Matsuyama et al. / Nucl. Instr. and Meth. in Phys. Res. B 260 (2007) 55–64
Fig. 3. Theoretical grid shadow patterns calculated by PRAM and OXTRACE.
Fig. 4. Measured grid shadow patterns of the two quadrupole lenses for vertical (X) and horizontal (Y) focusing.
the vertical plane) remains. This contamination source might be the lens itself or another part of the microbeam
line. This source can be checked by rotating the lens by 90. If the sextupole contamination remains in the horizontal plane, then the contamination originates outside the lens. If the sextupole contamination changes to the vertical plane, then the contamination source is the lens itself. Unfortunately, our lens system cannot rotate by 90, so we were unable to identify the contamination source. One possibility is that the cancellation of excitation current of the quadrupole is not perfect. This lens was designed so that the excitation current from pole to pole is canceled by the return current. However, no cancellation is made in the upper part of the lens where terminals exist. Because the coils consist of only 22 turns, such effects might cause the problem. To reduce the effect of the contamination field, the 1st quadrupole uses horizontal focusing and the 2nd lens uses vertical focusing (divergence–convergence); the scanner is located downstream. Then, the beam spot size was measured by scanning the beam across mesh samples (Ni and Au mesh, 2000 lines/ in.) and measuring X-rays. Horizontal and vertical line
Fig. 5. Horizontal and vertical beam spot sizes versus object sizes.
S. Matsuyama et al. / Nucl. Instr. and Meth. in Phys. Res. B 260 (2007) 55–64
profiles were fitted by symmetric double Gaussian convolution and beam spot sizes were obtained. The line profiles were well reproduced using the symmetric double Gaussian convolution, which implies that the beam profile can be assumed to be of Gaussian shape. Fig. 5 shows horizontal and vertical beam spot sizes versus object sizes. These measurements were carried out in the same beam divergence of 0.2 mrad which corresponded to our normal divergence in the high-current regime for ion beam analyses. Beam spot sizes diminished concomitant with the object slit sizes and were better than those that were predicted from the calculations. A beam spot size of 0.4 · 0.4 lm2 was obtained at an object size of 30 · 10 lm2 with a beam current of several tens of pA. Beam size estimations using the deconvolution method are strongly dependent on the mesh quality and are sometimes overestimated in the range less than 1 lm because we assume the mesh edges to be very sharp [6]. The cross sections of mesh grid bars are trapezoidal or round to some degree. Therefore, the beam spot size will be better than 0.4 · 0.4 lm2. Fig. 6 shows secondary electron images of Ni mesh (1000 lines/in.) measured using an electron microscope and the microbeam system with object size of 25 · 10 lm2 and beam divergence of 0.2 mrad. The secondary electron image measured using the microbeam system clearly represents the steps of the bars and round shape of the corner, which corresponds to the image obtained using the electron microscope. This fact shows that resolution of better than 0.4 lm is obtained and is consistent with that obtained using deconvolution method. These measurements were carried out with the divergence of 0.2 mrad for the high-current regime. The result of the grid shadow implies that a better beam spot size will be obtained in a low-current regime where the object size and beam divergence are one-tenth lower. Even in the same beam condition mentioned above, the beam spot size of 1 lm is a lower limit in the actual experimental setup, where X-ray detectors, RBS detector, and other detectors are set in a chamber. The grid shadow method revealed the source of this problem. The grid shadow patterns in this situation are shown in Fig. 7. Strong sextupole or higher order field contamination are observed. By removing the annular Si surface barrier detector (TC-
Fig. 6. Secondary electron images of Ni mesh (1000 lines/in.) measured using an electron microscope and the microbeam system.
59
Fig. 7. Measured grid shadow patterns with a parasitic field from the annular detector.
019-150-100; Ortec) positioned 40 mm upstream of the sample position, field contamination was eliminated. The grid shadow pattern in Fig. 7 was measured without applying bias voltage to the annular detector. The holder of the annular detector is made of plastic, which adds no field. Therefore, the parasitic field was ascribed to the magnetic field of the annular Si surface barrier detector itself. In fact, the annular detector imparted a field strength of 10 lT, as measured using a magnetometer. Elements that show strong magnetism are used in the detector body. The RBS detector was replaced by another annular ionimplanted Si detector (ANPD 300-19-100RM; Canberra). The detector gave a weaker magnetic field, which was also measured using a magnetometer. Following change, the field contamination of the system was reduced to a normal level and performance of the microbeam system was improved greatly, even in the actual analysis. Components that must be set near the beam path should have their magnetism eliminated; the grid shadow method is very effective to clarify the source of parasitic field. 5. Development of a control system The microbeam system is controlled via a programmable controller (FA-M3; Yokogawa Electric Corp.) and a personal-computer-based measurement instrument (WE7000; Yokogawa Electric Corp.) using LabVIEW based software. The power supplies of the analyzing magnet and quadrupole lenses are stabilized within 10 5 and are driven by 16-bit digital-to-analog converters (DACs) in WE7000. The control parameters for operation are stored in the memory space of the programmable controller. Another personal computer used for the user interface (HMI-PC) can only refer to these parameters. During the start-up process, control software retrieves the parameters for a programmable controller and can send commands to change the parameters. This framework was adopted to deal with the situation of a possible computer hang-up and to enable the use of different HMI-PCs.
60
S. Matsuyama et al. / Nucl. Instr. and Meth. in Phys. Res. B 260 (2007) 55–64
Fig. 8. Main and Sub-window of microbeam control software.
The main control window is arranged to neatly fit a small notebook computer display. The control window for each device can be opened by clicking the icon. Fig. 8 shows an example of the main and sub-windows of the HMI-PC. This graphical feature enables novice users to become accustomed to using the system with minimum training effort. A display palette is provided to adjust the two parameters simultaneously when control of more than two related parameters is necessary (e.g. control of the quadrupole doublet) (Fig. 8). The display contents such as the beam currents and images from CCD cameras would automatically switch to the currently focused location according to the change of Faraday cups and viewers, which can reduce the operational load. This control software has been designed based on user requirements and requests. Because the control software was constructed on a LabVIEW system, which is characterized by superior flexibility in alternation, it is rather easy to make modifications or to customize it for better usability. However, it is difficult to maintain the software when many users try to improve the software merely to suit their own preferences. Because the appropriate arrangement of the display might change according to experimental conditions, the interface contents should not be customizable based solely on a specific user’s opinion. In the present study, the user interface was elaborated based on users’ opinions [15] and information of their respective operational histories [16] for actual microbeam analysis operation. In analysis operations, the following procedure is required. First, direct STIM was carried out defining the analysis area. Subsequently, simultaneous PIXE/RBS/off-axis STIM measurements were executed, after which direct STIM was employed once more for the same position. In this procedure, an operator must change
many parameters on different sub-windows depending on the task. The points of the interface design elaboration for our microbeam system can be summarized as follows: I. Combined use of subjective and objective informationAs stated previously, if the subjective opinions of specific users are adopted, the interface might become better for a specific user, but not for the other users. Therefore, objective information (operational history) was used to reveal the possible problems and specific characteristics of operation for different experimental conditions. II. Iterative elaboration processThe elaboration process should be repeated several times until users are satisfied. It is rather difficult to determine final design specifications based on a limited number of evaluations. In the present study, the elaboration process for interface design continued for more than one year after the system was put to actual experimental use. Fig. 9 shows the sub-window developed for actual microbeam analyses. The important finding in our design elaboration process is that the users’ opinions are not always correct and users are aware of the potential problems only after it is revealed through analysis of the operational history. This fact strongly implies the importance of the combined use of subjective and objective information for evaluation of the user interface. The resultant interface design was put to use for daily operation. Two facts indicate the effectiveness of our design methodology: the operational history shows considerable reduction of the number of window focus operations, thus implying that the control window contents were appropri-
S. Matsuyama et al. / Nucl. Instr. and Meth. in Phys. Res. B 260 (2007) 55–64
61
ately re-arranged for experimental purposes; in addition, questionnaire results indicated users’ acceptance, with improved subjective evaluation of users after the elaboration process. Those results imply that our pragmatic approach for interface elaboration is effective for realizing better usability. 6. Recent applications of the microbeam system Microbeam analysis system at Tohoku University is routinely applied for studies in various fields. Typical applications are described below. 6.1. Aerosol study
Fig. 9. Sub-window developed for actual microbeam analyses.
Elemental analysis of atmospheric aerosols is useful for source identification and for the study of their formation mechanisms. These studies were carried out by analyzing bulk samples thereby averaging over many single particles. Therefore, analysis of single aerosol particles is superior to bulk analysis. Our system can quantify single aerosol particles containing from hydrogen to heavy metals and reveal chemical composition of these particles [8]. The microbeam system was applied to analysis of yellow sand dust particles. These particles were impacted on a thin polycarbonate
Fig. 10. Elemental maps of yellow sand dust particles. Scanning area is 100 · 100 lm2.
62
S. Matsuyama et al. / Nucl. Instr. and Meth. in Phys. Res. B 260 (2007) 55–64
film developed especially for this purpose. Yellow sand dust particles from the Asian continent sometimes cause turbid conditions in Japan, especially in spring. Typical elemental images of yellow sand dust particles are shown in Fig. 10. The elemental distribution of Al is similar to that of Si and O, originating mainly from soil dust of alumino-silicate particles. Distribution of hydrogen is visible and resembles that of carbon. These particles are sometimes deformed by mixing with anthropogenic aerosols. The system is useful to study these deformation processes. 6.2. Cell analysis In-air simultaneous PIXE, RBS, and a consecutive STIM analyses are powerful tools for analyses of single cells and tissue sections in biological and biomedical research. We have developed an in-air micro-PIXE system in collaboration with the Japan Atomic Energy Agency (JAEA), Takasaki [17,18] and have applied it to various cells [19–21]. Our in-air analysis system is also adjusted specifically for this purpose [7]. Although in-air analysis reduces damage during beam irradiation [22], morphological changes occasionally occur by beam irradiation, and cell thickness or density is reduced to 60% of the initial value [7]. In this case, the cell density should be monitored simultaneously with analysis. In addition, correction of Xray self-absorption should be done according to the irradiation dose. On/off-axis STIM facilitates performance of simultaneous density imaging in PIXE and RBS analyses. Fig. 11 shows elemental maps of cells and density maps measured using the new in-air system. Distributions of P, S, Cl, and K elements are clarified and correspond to the
Fig. 12. Optical image of pattern written using microbeams.
density. Changes in cell density during beam irradiation can be monitored and will be effective for cell analyses. 6.3. Proton beam writing Proton beam writing was realized by scanning the microbeam with a corresponding scanning pattern translated from a bitmap file. Pattern-to-pattern spacing was accomplished by moving the beam at the fastest beam scanner speed. Details of the system will be described elsewhere. A typical pattern written on a Mylar film is shown in Fig. 12. Although this pattern was written before removing the field contamination from the annular detector, the shape of Japan’s islands is clearly visible. We are planning to use this technique for efficient analyses through combination with a STIM image.
Fig. 11. Elemental maps and density distribution of cells. Scanning area is 100 · 100 lm2.
S. Matsuyama et al. / Nucl. Instr. and Meth. in Phys. Res. B 260 (2007) 55–64
63
Fig. 13. Cross sectional view of an ant’s head.
6.4. 3D l-CT The 3D l-CT technique comprises a point X-ray source, a rotating sample stage and a high-speed X-ray CCD camera. Details of 3D l-CT and applications are described in detail in [9–11]. The microbeam is used as a monoenergetic l-X-ray source by bombarding pure metals. The 2D transmission data are obtained by rotating the sample encapsulated in a polycarbonate tube (1000 lm inner diameter). The 3D images were reconstructed from the obtained transmission data. 3D images of a living ant’s head were obtained using this system with 6 lm spatial resolution. Typical CT images of the ant’s head measured using Ti K X-rays are shown in Fig. 13. The 3D l-CT uses monoenergetic low-energy X-rays. Therefore, the contrast of the image for small insects is superior to that of other CT systems. Furthermore, our system can readily change the X-ray energy to obtain a better contrast depending on the elements contained in the insects. This system can measure small insects of 2 mm in-vivo and will be useful for biological studies. 7. Conclusion A microbeam system was constructed at the Tohoku University Dynamitron laboratory and optimization of the system was undertaken. Minimum beam spot size of the microbeam was limited at 2 · 2 lm2 by parasitic field contamination. This parasitic field contamination was ascribed to the tungsten carbide slit chips and the annular Si surface barrier detector in the case of RBS analysis. The parasitic field contamination of the system was greatly reduced by replacing these components, as confirmed by the grid shadow method, and the microbeam system performance was improved. The minimum beam spot size of 0.4 · 0.4 lm2 at a beam current of several tens of pA is obtained, which is the optimum expected from the accelerator performance. The results obtained using the grid
shadow method indicate that a beam spot size of less than 0.4 lm is obtainable in the low-current regime. For easy tuning and operation of the microbeam system, a human–machine interface (HMI) for the control system was developed based on the user participatory design concept. The operational log of a skilled operator was used to realize the HMI, which is both user-friendly and has higher usability both for novices and experienced users. The new HMI has been accepted favorably by users and is used now in routine operations. While improving the microbeam system, simultaneous in-air/in-vacuum PIXE, RBS, SEM, and STIM analyses, in addition to 3D l-CT, were developed and are now being applied to biological, environmental, and other fields. Acknowledgements This study was partly supported by Grants-in-Aid for Scientific Research, (S) No. 13852017, (B) No. 18360450, (C) No. 16560731, and a Grant-in-Aid for Scientific Research in Priority Areas under Grant No. 14048213 from the Ministry of Education, Culture, Sports, Science and Technology, Japan. The authors would like to thank Prof. D.N. Jamieson of the University of Melbourne, for his valuable suggestions and advice in reducing field contamination and introducing the PRAM and OXTRACE computer codes. The authors would like to acknowledge the assistance of Mr. R. Sakamoto and M. Fujisawa for maintenance and operation of the Dynamitron accelerator. The authors would like to thank Mr. K. Komatsu, T. Nagaya and C. Akama for their assistance in constructing the microbeam and target system. References [1] F. Watt, G.W. Grime, Principal and Applications of High-energy Ion Microbeams, Adam Hilger, Bristol, 1987. [2] S.A.E. Johansson, J.L. Campbell, K.G. Malmqvist, Particle-Induced X-ray Emission Spectrometry (PIXE), John Wiley & Sons, NY, 1995.
64
S. Matsuyama et al. / Nucl. Instr. and Meth. in Phys. Res. B 260 (2007) 55–64
[3] M.B.H. Breese, D.N. Jamieson, P.J.C. King, Materials Analysis using Nuclear Microprobe, John Wiley & Sons, NY, 1995. [4] S. Matsuyama, K. Ishii, H. Yamazaki, R. Sakamoto, M. Fujisawa, Ts. Amartaivan, Y. Ohishi, M. Rodoriguez, A. Suzuki, T. Kamiya, M. Oikawa, K. Arakawa, N. Matsumoto, Nucl. Instr. and Meth. B 210 (2003) 59. [5] D.N. Jamieson, G.J.F. Legge, Nucl. Instr. and Meth. B 29 (1987) 544. [6] S. Matsuyama, K. Ishii, H. Yamazaki, Y. Barbotteau, Ts. Amartaivan, D. Izukawa, K. Hotta, K. Mizuma, S. Abe, Y. Oishi, M. Rodriguez, A. Suzuki, R. Sakamoto, M. Fujisawa, T. Kamiya, M. Oikawa, K. Arakawa, H. Imaseki, N. Matsumoto, Int. J. PIXE 14 (1–2) (2004) 1. [7] S. Matsuyama, K. Ishii, S. Abe, H. Ohtsu, H. Yamazaki, Y. Kikuchi, Ts. Amart aivan, K. Inomata, Y. Watanabe, A. Ishizaki, Y. Barbotteau, A. Suzuki, T. Yamaguchi, G. Momose, H. Imaseki, Int. J. PIXE 15 (1–2) (2005) 41. [8] S. Matsuyama, K. Ishii, H. Yamazaki, Y. Kikuchi, Ts. Amartaivan, S. Abe, K. Inomata, Y. Watanabe, A. Ishizaki, R. Oyama, Y. Kawamura, A. Suzuki, G. Momose, T. Yamaguchi, H. Imaseki, Int. J. PIXE 15 (3–4) (2005) 257. [9] K. Ishii, S. Matsuyama, H. Yamazaki, Y. Watanabe, T. Yamaguchi, G. Momose, Ts. Amartaivan, A. Suzuki, Y. Kikuchi, W. Galster, Int. J. PIXE 15 (3–4) (2005) 111. [10] K. Ishii, S. Matsuyama, H. Yamazaki, Y. Watanabe, Y. Kawamura, T. Yamaguchi, G. Momose, Y. Kikuchi, A. Terakawa, W. Galster, Nucl. Instr. and Meth. B 249 (2006) 726. [11] K. Ishii, S. Matsuyama, Y. Watanabe, Y. Kawamura, T. Yamaguchi, R. Oyama, G. Momose, A. Ishizaki, H. Yamazaki, Y. Kichuchi, Nucl. Instr. and Meth. A 571 (2007) 64.
[12] G. Deves, S. Matsuyama, Y. Barbotteau, K. Ishii, R. Ortega, Rev. Sci. Instr. 77 (2006) 056102. [13] T.W. Penrice, Int. J. Refract. Met. H 15 (1997) 113. [14] D.N. Jamieson, J. Zhu, Y. Mao, R. Lu, Z. Wang, J. Zhu, Nucl. Instr. and Meth. B 104 (1995) 86. [15] A. Dix, J. Finlay, R. Beal, People and Computers VII, in: Proceedings of HCI 92, Cambridge University Press, 1992, p. 429. [16] J. Nielsen, Usability Engineering (1993). [17] S. Matsuyama, K. Ishii, A. Sugimoto, T. Satoh, K. Gotoh, H. Yamazaki, S. Iwasaki, K. Murozono, J. Inoue, T. Hamano, S. Yokota, T. Sakai, T. Kamiya, R. Tanaka, Int. J. PIXE 8 (2–3) (1998) 203. [18] T. Sakai, T. Hamano, T. Hirao, T. Kamiya, K. Murozono, J. Inoue, S. Matsuyama, S. Iwasaki, K. Ishii, Nucl. Instr. and Meth. B 136–138 (1998) 390. [19] K. Ishii, A. Sugimoto, A. Tanaka, T. Satoh, S. Matsuyama, H. Yamazaki, C. Akama, Ts. Amartaivan, H. Endoh, Y. Oishi, H. Yuki, S. Sugihara, M. Satoh, T. Kamiya, T. Sakai, K. Arakawa, M. Saidoh, S. Oikawa, Nucl. Instr. and Meth. B 181 (2001) 448. [20] A. Sugimoto, K. Ishii, S. Matsuyama, T. Satoh, K. Gotoh, H. Yamazaki, C. Akama, Int. J. PIXE 9 (3–4) (1999) 151. [21] A. Tanaka, K. Ishii, Y. Komori, S. Matsuyama, T. Satoh, K. Gotoh, H. Yamazaki, C. Akama, Int. J. PIXE 12 (3–4) (2002) 79. [22] H. Komori, K. Mizuma, K. Ishii, H. Yamazaki, S. Matsuyama, Ts. Amartaivan, Y. Ohishi, M. Rodoriguez, A. Suzuki, M. Satoh, T. Kamiya, T. Sakai, T. Satoh, M. Oikawa, K. Arakawa, Int. J. PIXE 124 (3–4) (2004) 75.