Issuu on Google+


JOURNAL of AUTOMATION, MOBILE ROBOTICS & INTELLIGENT SYSTEMS

Editor-in-Chief Janusz Kacprzyk

Executive Editor: Anna Ładan aladan@piap.pl

(Systems Research Institute, Polish Academy of Sciences; PIAP, Poland)

Associate Editors: Mariusz Andrzejczak (PIAP, Poland) Katarzyna Rzeplińska-Rykała (PIAP, Poland)

Co-Editors: Dimitar Filev (Research & Advanced Engineering, Ford Motor Company, USA)

Kaoru Hirota

Webmaster: Tomasz Kobyliński tkobylinski@piap.pl

(Interdisciplinary Graduate School of Science and Engineering, Tokyo Institute of Technology, Japan)

Witold Pedrycz

Proofreading: Urszula Wiączek

(ECERF, University of Alberta, Canada)

Roman Szewczyk (PIAP, Warsaw University of Technology, Poland)

Editorial Office: Industrial Research Institute for Automation and Measurements PIAP Al. Jerozolimskie 202, 02-486 Warsaw, POLAND Tel. +48-22-8740109, office@jamris.org

Copyright and reprint permissions Executive Editor

Editorial Board: Chairman: Janusz Kacprzyk (Polish Academy of Sciences; PIAP, Poland) Plamen Angelov (Lancaster University, UK) Zenn Bien (Korea Advanced Institute of Science and Technology, Korea) Adam Borkowski (Polish Academy of Sciences, Poland) Wolfgang Borutzky (Fachhochschule Bonn-Rhein-Sieg, Germany) Oscar Castillo (Tijuana Institute of Technology, Mexico) Chin Chen Chang (Feng Chia University, Taiwan) Jorge Manuel Miranda Dias (University of Coimbra, Portugal) Bogdan Gabryś (Bournemouth University, UK) Jan Jabłkowski (PIAP, Poland) Stanisław Kaczanowski (PIAP, Poland) Tadeusz Kaczorek (Warsaw University of Technology, Poland) Marian P. Kaźmierkowski (Warsaw University of Technology, Poland) Józef Korbicz (University of Zielona Góra, Poland) Krzysztof Kozłowski (Poznań University of Technology, Poland) Eckart Kramer (Fachhochschule Eberswalde, Germany) Andrew Kusiak (University of Iowa, USA) Mark Last (Ben–Gurion University of the Negev, Israel) Anthony Maciejewski (Colorado State University, USA) Krzysztof Malinowski (Warsaw University of Technology, Poland)

Andrzej Masłowski (PIAP, Poland) Tadeusz Missala (PIAP, Poland) Fazel Naghdy (University of Wollongong, Australia) Zbigniew Nahorski (Polish Academy of Science, Poland) Antoni Niederliński (Silesian University of Technology, Poland) Witold Pedrycz (University of Alberta, Canada) Duc Truong Pham (Cardiff University, UK) Lech Polkowski (Polish-Japanese Institute of Information Technology, Poland) Alain Pruski (University of Metz, France) Leszek Rutkowski (Częstochowa University of Technology, Poland) Klaus Schilling (Julius-Maximilians-University Würzburg, Germany) Ryszard Tadeusiewicz (AGH University of Science and Technology in Kraków, Poland)

Stanisław Tarasiewicz (University of Laval, Canada) Piotr Tatjewski (Warsaw University of Technology, Poland) Władysław Torbicz (Polish Academy of Sciences, Poland) Leszek Trybus (Rzeszów University of Technology, Poland) René Wamkeue (University of Québec, Canada) Janusz Zalewski (Florida Gulf Coast University, USA) Marek Zaremba (University of Québec, Canada) Teresa Zielińska (Warsaw University of Technology, Poland)

Publisher: Industrial Research Institute for Automation and Measurements PIAP

If in doubt about the proper edition of contributions, please contact the Executive Editor. Articles are reviewed, excluding advertisements and descriptions of products. The Editor does not take the responsibility for contents of advertisements, inserts etc. The Editor reserves the right to make relevant revisions, abbreviations and adjustments to the articles.

All rights reserved ©

1


JOURNAL of AUTOMATION, MOBILE ROBOTICS & INTELLIGENT SYSTEMS VOLUME 3, N° 4, 2009

CONTENTS SPECIAL ISSUE

41

ISpace – a tool for improving the quality of life A.R. Varkonyi-Koczy, A.A. Tóth

Inter-Academia 2009 Guest Editor: Ryszard Jabłoński 7

46

Simulation of Coulomb-coupled protein-based logic B. Rakos 49

EDITORIAL

Seebeck coefficient measurement by Kelvin-Probe force microscopy H. Ikeda, F. Salleh, at al.

R. Jabłoński 9

Femtosecond photonics for three-dimensional high density optical data storage Y. Kawata, M. Tsuji, at al.

52

Elevated giant magneto-impedance in amorphous metallical Co-dased alloys L.V. Poperenko, D.Yu. Manko

55

An Xps study of RF sputtered Ti1-xFexO2-thin films D. Luca, R. Apetrei, at al.

58

Single-electron transport characteristics in quantum dot arrays due to Ionized dopants D. Moraru, M. Ligowski, at al.

12

Sol-gel glassy antireflection GeO2 – SiO2 – Ag – Re films for solar cells and IR-devices D. Kovalenko, V. Gaishun, at al.

15

Multi-scale simulation of hybrid silicon nanoelectromechanical (NEM) information systems H. Mizuta, M.A.G. Ramirez, at al.

18

The assessment of "Jatropha" as raw material for BDF S. Matsuda, B. Rudyanto, at al.

62

Nanohills in SiGe/Si structure formed by laser radiation A. Medvid, P. Onufrijev, at al.

21

Investigation into electrochemical discharge machining of micro holes L. Kudła

65

Effect of substrate temperature on the crystal properties of LiMn2O4 films prepared by RF magnetron sputtering M. Isai, K. Nakamura, at al.

25

An introduction of high-precise 3D measurement system and its applications T. Hashimoto, M. Kaneko, at al. 69

29

Dye-sensitized solar cells based on SnO2 nanorod and surface treatment with Mg(II) film J. Bai, K. Murakami

New optical equipment in 3D surface measuring K. Wenzel, Á. Antal, at al. 33

Dynamics of laser-induced melting and modification of the surface of semiconductors by nanosecond laser pulses V. Gnatyuk, T. Aoki, at al.

Analysis of current noise in MOSFET-based chargetransfer device H. Inokawa, V. Singh, at al. 76

37

Radius measurement of cylindrical surfaces based on analyzing the interference pattern obtained by scanning the surface with focused laser beam R. Jabłoński, J. Mąkowski

2

72

Epileptic burst measurement using microelectrodes equipped on a cryogenic microprobe for minimally invasive brain surgery of intractable epilepsy treatment T. Yamakawa, T. Yamakawa, at al.


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

127

80

R&D of novel medicinal materials for curing cancer: sugar modified Gd-DTPA MRI contrast agents and Phospha Sugar Anti-cancer agents J. Yamashita, M. Yamashita, at al.

Mechanism of nanostructure formation on a surface of CdZnTe crystal by laser irradiation A. Medvid', A. Mychko, at al. 130

Detection of individual dopants in singe-electron devices - a study by KFM observation and simulation M. Ligowski, D. Moraru, at al.

84

Analyses of the genes involved in disease development to construct disease-resistant plants by genetic engineering S. Tsuyumu, H. Hirata

134

Impurity-concentration dependence of seebeck coefficient in silicon-on-insulator layers M.F. Mohd Salleh, K. Asai, at al.

87

Powder injection moulding of micro parts D. Biało, A. Skalski

137

CdTe X-ray imaging device using vertical thin film field emitter array Y. Tsunekawa, M. Nakagawa, at al.

91

Influence of synthesis conditions on the properties of FexCoyOz- nanoparticles in SiO2 sol-gel film V.V. Sidsky, V.E. Gaishun, at al.

140

Formation of “black silicon” on a surface of Ni/Si structure by Nd:YAG laser radiation A. Medvid', A. Karabko, at al.

93

Low-range tilt sensing with MEMS accelerometers S. Łuczak, D. Kołodziej 143

96

High speed signal processing for photon counting X-ray detection B. Shinomiya, A. Koike, at al.

Testing of the three axis magnetometers for measurements of the earth magnetic field R. Szewczyk, J. Salach, at al. 147

99

Thermionic vacuum arc diagnostic using emissive probe V. Tiron, C. Aniculaesei, at al.

Fabrication of photonic crystal structures by laser lithography V. Mizeikis, K.K. Seet, at al. 150

102

Investigation of low temperature plasma capabilities to modify the structure and function of bio-polymers J. Motrescu, T. Hara, at al.

Plasmonic nano-imaging with metallic nanolens A. Ono 105

Overhauser effect and anisotropy of electron spin g-factor in GaAs / AlGaAs quantum wells T. Ito, P. Verma, at al.

153

Study of low temperature inactivation of spores using microwave excited air plasma M. Nagatsu, M.K. Singh, at al.

157

Testing of carbon fibers as tool electrodes in micro electrical discharge machining L. Kudła, A. Trych

108

Quintet-gated FEA to form crossover beam T. Tagami, M. Takeda, at al. 160

112

Measurements of plasma diffusion coefficient in PilotPSI device using Katsumata probe M.L. Solomon, I. Mihaila, at al.

ZnO films: properties determined by electronic microscopy and ellipsometry M. Rakov, L. Poperenko, at al. 163

115

Growth and application of ultra-long multi-walled carbon nanotube S. Sayaka, M. Okada, at al.

Tribological properties of multilayered vacuum coating A. Rogachou, A. Popov, at al. 118

Tools and technique of monitoring resource distribution for design simulation of organization of information processing O.M. Demidenko

166

Influence of powerful laser radiation on formation of pores in Si by electrochemical etching A. Medvid, P. Onufrijevs, at al. 169

121

Probe radius correction methods – review and comparison of existing methods A. Rak, A. Woźniak

Application of sol-gel process for preparation of functional materials V.E. Gaishun, O.M. Demidenko, at al. 124

172

The growth and properties of ZnO films grown using PA-MOVPE with DMZn and N2O T. Nakano, K. Nishimoto, at al.

One dimensional kinetic model of CMM passive scanning probes G. Krajewski, A. Woźniak

3


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

175

2009

224

Application of dielectric layers in surface plasmon resonance sensors S. Kurlov, T. Lebyedyeva, at al.

Design of multi-layer sputter-deposited anode to reduce catalyst loading for liquid DMFC H. Saito, T. Nakashima, at al. 228

179

Small batch size robot programming with human in the loop B. Takarics, G. Sziebig, at al.

Numerical simulation of blood flow and cholesterol distribution for middle cerebral artery with coarctation M. Naito, K. Mizoguchi, at al. 184

A versatile model for music creation utilizing tension states H. Ruuska, Y. Takebayashi 187

A New Man-Machine Interface for iSpace applications A.A. Toth, A.R. Várkonyi-Kóczy 191

Synthesis and evaluation of novel MRI contrast agents of chemically modified Gd-DTPA complexes with sugars M. Sugiyama, M. Yamashita, at al. 195

The design of an insulin pump - concept of closing the loop H. Hawłas, K. Lewenstein 199

Acoustic waves in ceramics with the electro induced anisotropy S.A. Khakhomov, S.D. Barsukov, at al. 202

Reduction of random telegraph signal (RTS) noise in CMOS image sensors using histogram analysis M.A. Mustafa, S. Itoh, at al. 204

Linearized settling error calibration for a pipeline A/D converter using non-slewing amplifiers S.-W. Jun, K. Yasutomi, at al. 207

Effect of maximum interaction of circularly polarized electromagnetic waves with the molecule of DNA A.P. Balmakov, I.V. Semchenko, at al. 210

Irreversible blocking of polar excitations on frog sciatic nerve using semiconductor pulse laser irradiation Y. Hirayama, T. Yamakawa, at al. 213

Color calibration method providing uniform distribution of color difference throughout the whole vision gamut M. Kretkowski, Y. Shimodaira, at al. 217

Simplified calibration methods for the measuring of LCD monitor charasteristics A. Molnar, K. Samu 220

Introduction of three-dimensional measurement technologies developed and applied in Suzuki Motor Corporation Y. Yamage, T. Aoki, at al.

4

N° 4

DEPARTMENTS 232

IN THE SPOTLIGHT 234

EVENTS


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

SPECIAL ISSUE

Inter-Academia 2009 8th International Conference on Global Research and Education

Guest Editor: Ryszard Jabłoński


Journal of Automation, Mobile Robotics & Intelligent Systems

VOLUME 3,

N° 4

2009

Editorial Oscar Castillo*, Patricia Melin

Special issue on Inter-Academia 2009 8th International Conference on Global Research and Education The Inter-Academia is an already well established initiative uniting various universities in their efforts in promoting mutual exchange of information and improving education, mostly by forming international research teams. It has progressed considerably over the 9 years of its existence. The 8th International Conference on Global Research and Education – Inter-Academia 2009 is the continuation of a successful series of focused, international conferences which started in Bratislava, Slovakia in 2002 and since that time had been held in various European countries and Japan. During that time the Inter-Academia Board was founded of significant researchers and scholars from among the Inter-Academia members and supporters. The Board is acting on behalf of Inter-Academia, promoting its philosophy, pointing out new research areas and topics, proposing new solutions and also applying for sponsorship. The goal of the Inter-Academia conferences is to bring together experts from different areas to give an overview of the state-of-the-art and to present new research results and prospects for the future development. This year the Inter-Academia meeting was organized by The Faculty of Mechatronics, Warsaw University of Technology, Poland. It gathered many prominent scientists and scholars, and the presented papers covered a wide spectrum of topics, notably on the combination of various modern technologies brought together to reduce complexity through the adaptation of interdisciplinary techniques in manufacturing. The main topics selected for Inter-Academia 2009 are: Nanotechnology, Biomedical Engineering, Material Science, Metrology, Photonics, Mechatronic Products, and E-learning. This Special Issue includes 66 papers which have been selected for inclusion via a strict peer review process of full lengths papers, with emphasis on both the originality and quality. The selected papers were written by top scholars and researchers from 8 countries: Japan, Poland, Hungary, United Kingdom, Romania, Belorussia, Ukraine, and Latvia. I believe that this Special Issue will be an up-to-date source of information on the newest theoretical and applied developments in those areas that can be useful for scholars, researches and engineers, and can form a basis for advanced scientific works and applications in the manufacturing industry. I would like to thank all the authors for their great contributions to this Special Issue and all the reviewers for their careful reviews of the papers.

10 August, 2009 Professor Ryszard Jabłoński Faculty of Mechatronics Warsaw University of Technology Conference Chairman Guest Editor of this Special Issue.

Editorial

7


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

FEMTOSECOND PHOTONICS FOR THREE-DIMENSIONAL HIGH DENSITY OPTICAL DATA STORAGE Yoshimasa Kawata, Masatoshi Tsuji, Wataru Inami

Abstract: We have developed a high density optical data storage by using two-photon absorption process. We have developed a multilayered medium in which photosensitive layers and transparent pressure-sensitive adhesives layer were piled up alternately. We also demonstrate the development of compact high power fiber laser as a light source of high density optical data storage system. Keywords: femtosecond pulse laser, two-photon absorption, confocal microscopy, optical memory.

1. Introduction Femtosecond laser is a very promising tool for opening new applications [1],[2]. It is possible to fabricate inside of transparent materials, produce three-dimensional fine structures with photosensitive resin, study dynamics of very fast response of materials. Many applications are proposed and have been developed by using femtosecond lasers. We have developed three-dimensional (3D) optical memory for ultra-high data storage as an application of femtosecond photonics [3]-[8]. Because 3D optical memory can achieve high recording capacity as much as the numbers of recording layers, it is possible to overcome the density limit determined by the diffraction of light [9]. We have developed a multilayered medium in which photosensitive recording layers and transparent layers are piled up alternately in order to increase signal to noise ratio. We have developed the fabrication technique of multilayered mediums using pressure sensitive adhesive.

2. Fabrication of Multilayered Medium We fabricated multilayered medium by wet coating and laminating process using pressure sensitive adhesives (PSAs). First, the photosensitive layer and the PSA as buffer layer were deposited on release liner films. Coating solutions were cast on release liner films, and then solvents were removed in a dryer. The photosensitive layer and the PSA were laminated together. The two layered sheets are superposed on each other continually up to 20 layers. 1,3,3-trimethylindolino-6'-nitrobenzopyrylospiran was used as photosensitive materials. We observed the axial distribution of the fabricated multilayered medium. Figure 1(a) shows the structure of the 20-layered medium we fabricated. Thickness of photosensitive layers was 1.6 μm and thickness of transparent layers was 4.6 μm. Figure 1(b) shows observation result and the cross section of readout signal with a ref-

lection confocal microscope. We confirmed that each photosensitive layer could be observed clearly, without decreasing signal intensity in deep region. By fabricating layered structures in a recording medium, we can control the spatial frequency distribution of recorded bit in the axial direction, and optimize it for a readout system of the multilayered optical memories [4]-[6]. Extension of the spatial frequency distribution in the axial direction is advantageous for the readout system of the multilayered memories. The distribution is given by convolution of the frequency distribution of the bit intensity and that of the multilayered medium. As a result, because each individual bit in the multilayered medium contains higher spatial frequency components in the axial direction, the spatial frequency distribution of a bit recorded in the multilayered medium easily overlaps the coherent transfer function (CTF) of the reflection confocal microscope which is used as readout system. The recording density can also be increased by using the multilayered medium. When a bit is recorded in a multilayered medium, its size may be confined in a volume smaller than that of the focused spot. It is possible to reduce the interval between individual bits in the axial direction. We demonstrated to record and read bit patterns in the multilayered medium fabricated by laminating process. Figure 2 shows recording and reading results. Random bit patterns were recorded in twenty layers and read out without crosstalk. We recorded the 3D bit patterns with a Ti:sapphire laser (Spectra-Physics: Mai Tai) at 800 nm in mode-locked pulse laser operation and used an oil immersion lens with a NA of 1.3 to decrease spherical aberration produced by refractive index mismatch between the air and the medium. The exposure time is 125 ms. The layer interval between neighboring photosensitive layers was 6.2 μm. The most adjacent bit distance in plane direction is 2.0 μm.

Fig. 1. The cross section of the 20-layered medium fabricated by laminating process (a) Schematic diagram of multilayered medium (b) axial response read with a reflection confocal microscope. Articles

9


Journal of Automation, Mobile Robotics & Intelligent Systems

VOLUME 3,

N° 4

2009

Fig. 3. Experimental setup of mode-locked fiber laser.

Fig. 2. Recording and reading results of twenty-layers data.

3. Compact and High-Power Mode-Locked Fiber Laser for Three-Dimensional Optical Memory We also have developed compact femtosecond fiber as a light source of femtosecond photonics. Mode-locked pulse lasers are very promising as alight source in many applications, such as nonlinear spectroscopy, high-resolution micro-processing, multi-photon microscopy and optical memory, etc. Currently, bulk solid-state lasers such as Ti-Sapphire laser are widely used but their applications are limited because of the large size and worse stability. Recently, passively mode-locked erbium-doped fiber laser is developed. Fiber lasers have potential to be compact and stable light source which can replace bulk solidstate lasers in the near future [10], [11]. Moreover, the spatial property of output beam from fiber lasers is circular profile and very stable. We have developed high-power erbium-doped fiber ring laser using the stretched-pulse mode-locking. The fiber laser is very promising to replace a Ti-Sapphire laser that is used for three-dimensional optical memory. In this paper, we introduce compact and high-power erbium-doped fiber (EDF) ring laser for three-dimensional optical memory.

Figures 4 shows the spectrum of the output pulse measured the output from the coupler. The vertical axis is output power and the transverse axis is wavelength. A Gaussian shape spectrum was obtained in the developed fiber lasers. Figure 5 shows a pulse train of the laser output. The repetition rate of pulse was 40 MHz and the pulse width was supposed as several picoseconds. By chirped pulse compensation, it was conpressed the pulse width to nearly 100 fs. Averaged power was measured by PBS and was about 30 mW. The peak power was estimated to be 0.75 nJ from the averaged power and the repetition rate. The output power is enough for the application of threedimensional optical memory.

Fig. 4. The spectrum of output pulse from monitor port.

Figure 3 shows the configuration of compact high-power fiber laser. The laser cavity was composed with EDF and single mode fiber (SMF). EDF had positive dispersion and SMF had negative dispersion at 1.56 Îźm. Two 980 nm laser diodes (LDs) were combined by polarization beam combiner (PBC) and were used as pump laser. The power of each LD was 450 mW. EDF was pumped through the wavelength division multiplexing (WDM). Output of 1:99 coupler was used as monitor which connect with spectrum analyzer and oscilloscope.

Fig. 5. Pulse train of the developed fiber laser. 10

Articles


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

4. Data Recording of Three-Dimensional Optical Memory Figure 6 shows the 3D recording demonstration of photochromic materials by using the fiber laser we have developed. Two layers was recorded and readout. Figure 6(a) shows the 1st layer and (b) is the 2nd layer. Letter “A” and “B” were recorded in bit sequence for each layer respectively. The spacing between the layers is 40 μm. The recording power used was 4.7 mW with an exposure time of 1/125 s focused into the media. The cross talk in the 2nd layer was observed due to the detection system do not incorporate confocal system. The confocal readout system can improve the contrast of readout data.

[1]

[2]

[3]

[4]

[6]

[7]

Fig. 6. Three-dimensional recording result. (a) 1st layer and (b) 2nd layer.

We have succeeded in recording and reading of bit data in a 20 layers medium fabricated by wet coating and laminating process. We are now developing a hundred layers recording media. It was confirmed that each photosensitive layer could be observed clearly without decreasing the signal intensity by the reflection confocal microscope. Random bit patterns were recorded in ten layers and read out without crosstalk. Asa conclusion we may say that the multilayered medium fabricated by wet coating and laminating process is advantageous as the medium of optical memory. We also have developed the compact and high-power fiber ring laser [12]. We can achieve about 30 mW and pulsed-operation of 40 MHz repetition rate. The fiber configuration has potential to approach 100 mW averaged power and 100fs pulse width by adjustment of the wave plates and the fiber length. The fiber laser is very promising as a light source for three-dimensional optical memory.

2009

References

[5]

5. Conclusions

N° 4

[8]

[9]

[10]

[11]

[12]

Noor A.S.M., Miyakawa A., Kawata Y., Torizawa M., “Two-Photon Excited Luminescence Spectral Distribution Observation in Wide-Gap Semiconductor Crystals“, Appl. Phys. Lett., vol. 92, 2008, p. 161106. Torizawa M., Kawata Y., “Two-Photon-Induced Laser Annealing for Enhancement of Photoluminescence Intensity in ZnSe Crystal”, Opt. Lett., vol. 32, 2007, pp. 3327-3329. Kawata S., Kawata Y., “Three-Dimensional Optical Data Storage Using Photochromic Materials”, Chem. Rev., vol. 100, 2000, pp. 1777-1788. Nakano M., Kooriya T., Kuragaito T., Egami C., Kawata Y., Tsuchimori M., Watanbe O., “Three-Dimensional Patterned Media for Ultrahigh-Density Optical Memory”, Appl. Phys. Lett., vol. 82, 2004, pp. 176-178. Booth, M.J., Schwertner M., Wilson T., Nakano M., Kawata Y., Nakabayashi M., Miyata S., “Predictive Aberration Correction for Multilayer Optical Data Storage”, Appl. Phys. Lett., vol. 88, 2006, p. 031109. Miyamoto M., Ohta A., Kawata Y., Nakabayashi M., “Control of Refractive Index of Pressure-Sensitive Adhesives for the Optimization of Multilayered Media”, Jpn. J. Appl. Phys., vol. 46, 2007, pp. 3978-3980. Chu T.C., Liu W.C., Tsai D.P., Kawata Y., “Readout Signals Enhancements of Subwavelength Recording Marks via Random Nanostructures”, Jpn. J. Appl. Phys., vol. 47, 2008, pp. 5767-5769. Miyamoto M., Kawata Y., Ito M., Nakabayashi M., “Dynamic Layer Detection of Rotating Multilayered Optical Memory”, Jpn. J. Appl. Phys., vol. 47, 2008, pp. 59445946. Parthenoppulos D.A., Rentzepis P.M., “Three-Dimensional Optical Storage Memory”, Science, vol. 245, 1989, pp. 843-845. Tamura K., Doerr C.R., Nelson L.E., Haus H.A., Ippen E.P., “Technique for Obtatining High-Energy Ultrashort Pulses from an Additive-Pulse Mode-Locked ErbiumDoped Fiber Ring Laser”, Opt. Lett., vol. 19, 1993, pp. 46-48. Nishizawa N., Chen Y., Hsiung P., Ippen E.P., Fujimoto J.G., “Real-time, ultrahigh-resolution, optical coherence tomography with an all-fiber, femtosecond fiber laser continuum at 1.5 μm”, Opt. Lett., vol. 29, 2004, pp. 2846-2848. Tsuji M., Nishizawa N.,. Kawata Y., “Compact and HighPower Mode-Locked Fiber Laser for Three-Dimensional Optical Memory”, Jpn. J. Appl. Phys., vol. 47, 2008, pp. 5797-5799.

AUTHORS Yoshimasa Kawata*, Masatoshi Tsuji - Department of Mechanical Engineering, Shizuoka University, Johoku, Naka, Hamamatsu, 432-8561, Japan. E-mails: kawata@eng.shizuoka.ac.jp, tsuji@optsci.eng.shizuoka.ac.jp. Wataru Inami - Division of Gloval Research Leaders, Shizuoka University, Johoku, Naka, Hamamatsu, 432-8561 Japan. E-mail: dwinami@ipc.shizuoka.ac.jp. * Corresponding author

Articles

11


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

ELEVATED GIANT MAGNETO-IMPEDANCE IN AMORPHOUS METALLICAL CO-DASED ALLOYS Leonid V. Poperenko, Dmytro Yu. Manko

Abstract: This paper reviews results of an enhancement of giant magneto-impedance (GMI) in cobalt-rich amorphous glassy alloys. There are several ways of GMI increasing, namely: thermal, cryogenic and laser treatment. The results are explained via structural changes of ribbons surface and magnetio-optical properties. This phenomenon is interpreted via classical electromagnetic terms. The role of a conductive intermediate film in three-layer sandwich structures is also revealed. Such structure consisting of two cobalt-based ferromagnetic films and a conductive inner film of amorphous nickel shows significant increase of GMI ratio in comparison to a single layer. GMI enhancement makes possible to create new types of high sensitive magnetic field sensors. The investigation of evaluation processes after the ribbons treatment substantiates to clear understanding the nature of GMI. Keywords: amorphous metallic alloys, giant magnetoimpedance, laser and thermal annealing, cryogenic treatment, three-layer structures.

1. Introduction Amorphous metallic alloys (AMA) possess improved magnetic properties compared with their crystalline analogues, which makes possible their application in devices of magneto-electronics [1]. Giant magneto-impedance effect in amorphous wires and amorphous metallic alloys ribbons has been investigated after its discovery since 1988. Its origin are associated with specificity of magnetic properties. In terms of a theory of electron transfer phenomena the giant magnetoimpedance is connected with spin-dependent scattering of the conduction electrons. This effect is peculiar to the low-coercivity AMA based on cobalt as well as multilayered films with consecutive ferromagnetic and nonmagnetic layers or granular ferromagnetic materials in the wire matrix. The films based on amorphous magnetic alloys are promising materials for mass production of cheap microelectronic sensors and sensing heads for reading information. There are several ways to increase the GMI ratio of amorphous metallic alloys namely: thermal and laser annealing [2][4] or joule heat flux via alternating current through the sample [5], as well as the influence of deformations waves [6] on the GMI. The influence of thermal and laser annealing, cryogenic treatment on physical parameters is presented in this work, as well as impact of intermediate layer of nickel in three layer structure (amorphous alloy/ nickel/amorphous alloy). Achievement of maximal GMI value wasn't pursed in present work. 12

Articles

The GMI in the sandwich films has potential to be used in developing small sensitive magnetic heads for high density magnetic recording. Considering a real head the effect of in-plane sandwich width on GMI has to be studied. The problem is approached by finding the AC field distribution over the film width under the condition of a weak skin-effect [7]. 1.1. Experimental technique GMI effect has been investigated in magnetic fields up to 100 Oe and a frequency F of alternating current from 1 to 3000 kHz. As-quenched (AC) AMA samples were obtained in the strips with width of 10-20 mm and thickness of 20-25 μm. -6 Samples were annealed in vacuum (10 mm Hg) at temperatures of 350°C, 375°C and 400°C. Laser annealing procedure is described in work [5] namely: laser wavelength was 1064 nm, energy density order was equal to 2.5-3.5kW/cm² and applied external magnetic field was about 100 Oe. GMI was investigated in magnetic fields up to 100 Oe, and frequency alternating current that flew through the sample was varied from 10 to 220 kHz for AMA samples and from 1 to 3000 kHz for sandwiched AMA/Nickel/AMA structures. Magnetic field vector was oriented in the plane of the sample. GMI measurements were performed according to scheme which consists of the connected low-value resistor and investigated sample. The GMI ratio DZ/Z was defined according to the expression: DU/U = (UH=0 –UH)/UH=0 = DZ/Z = (ZH=0 –ZH)/ZH Where ZH=0 and ZH are impedances at H = 0 and the significance H of the magnetic field that corresponds to the largest value respectively. The compositions of the samples were determined by Auger spectroscopy. The three-layer sandwich structures were constructed using two AMA ribbons of the same kind with thicknesses 15 μm and a intermediate layer with width approximately 10 μm. 1.2. Results and discussion Fig. 1 presents GMI ratio of ribbons of AMA Co59Fe5Ni10Si11B15. The dependences show the impact of thermal and laser treatments on GMI ratio values which reach the maximum at a frequency of 100 kHz. These treatments don't change the maximum frequency localization. The relative value DU/U reaches the level of about 8-9%.


Journal of Automation, Mobile Robotics & Intelligent Systems

VOLUME 3,

N째 4

2009

metal-metal and metal-metalloid phases emerge. Moreover such phases are moving towards sample surface. A considerable enhancement of the GMI in three-layer structures can be achieved by separation between the conductive films and the magnetic films, which further decreases the DC resistance [10]. A very high sensitivity to an external field is typical for magnetoimpedance in soft ferromagnetic conductors with well-defined anisotropy. Thus, in CoFe(SiB) amorphous wires having almost zero magnetostriction and a circular domain structure is dominant [10]. Fig. 3 and Fig. 4 show the impact of a intermediate layer of nickel on a GMI ratio in the threelayer structures as compared to as-quenched one for appropriate AMA ribbons. Fig. 1. GMI ratio in amorphous metallic ribbon of Co59Fe5Ni10Si11B15 alloy. The heat treatment of the sample at the temperature of 350째C during 10 minutes leads to the increase of GMI effect more than 1.8 times compared to the as-quenched (AC) ribbon. Higher temperature processing up to 400째C leads to the decrease of GMI ratio DZ/Z and its appropriate value is almost equal to that before the treatment. This phenomenon is obviously connected with the structural changes that were occurred in the ribbons during thermal processing. In that time laser annealing leads to the increase of GMI effect of about 1.7 times in AMA Co59Fe5Ni10Si11B15 sample as well as an AMA Co58Fe5Ni10Si12B16 one that is shown in Fig. 2. In the latter the laser annealing results in the enhancement of GMI ratio almost in two times (Fig. 1). The result of the influence of cryogenic treatment in liquid nitrogen on GMI ratio values of AMA Co58Fe5Ni10Si12B16 ribbon is also shown in Fig. 2. Such behaviour implies that the exchange energy between large magnetic domains plays essential role at the low-frequency range below 1 MHz [8].

Fig. 3. GMI ratio in three layer system of Co63Fe5Ni5Si12B15/Ni/Co63Fe5Ni5Si12B15. As it is seen, the intermediate layer presence leads to a dramatic increase of GMI ratio value nearly two times without physical treatment of the ribbon. Similar behaviour has been observed for another sample with AMA Co58Fe5Ni10Si12B16 ribbons being used (Fig. 4).

Fig. 2. GMI ratio in amorphous metallic ribbon of Co58Fe5Ni10Si12B16 alloy.

Fig. 4. GMI ratio in three layer system of Co58Fe5Ni10Si12B16/Ni/ Co58Fe5Ni10Si12B16.

The clusters, which arise as a result of thermal annealing, have no characteristic sizes and differ in shape [9]. Such surface structure occurs due to evolutionary processes of atoms diffusion towards microcrystalline centres inside amorphous matrix [9]. In the case of laser annealing, size of the clusters weakly dependends on the magnitude and location of microcrystalline centres that exist in the amorphous matrix [9]. It should be noticed, that during thermal annealing

But in this case the enhancement of GMI ratio is much smaller. The enhancement of GMI ratio in such threelayer composite structure may be explained by consideration of their magnetic structures difference and the frequency dependence of their impedance Z. [11] The magnetic moment rotation becomes important under application of an external magnetic field and the value of such magnetic moment rotation is compared to existing intrinsic magnetic moment due to magnetic anisotropy Articles

13


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

field. Then the change of real and imaginary parts of the impedance contributes to GMI ratio. Moreover, the magnetic structure is varied after addition of some intermediate layer and the GMI effect behaviour is essentially changed and GMI ratio is just enhanced [11].

[10]

[11]

2. Conclusion Summarizing, we can conclude that the influence of thermal, laser and cryogenic treatments on GMI ratio and physical characteristics in Co59Fe5Ni10Si11B15 was established. Adding a nickel intermediate layer between to ribbons based on cobalt leads to significant enhancement of the GMI effect. The results can be explained by structural changes of amorphous ribbons surface and difference in magnetic structures and the frequency dependence of their impedance in case of three layer composites.

AUTHORS Leonid V. Poperenko*, Dmytro Yu. Manko - Physics Department, Taras Shevchenko Kiev National University, Kiev, 03187, Ukraine. E-mails: plv@univ.kiev.ua, dmitriy.manko@univ.kiev.ua. * Corresponding author

References [1]

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

14

Marin P., Hernando A., “Applications of amorphous and nanocrystalline magnetic materials”, J. Magn. Magn. Mater., vol. 215-216, 2000, pp. 729-734. Panina L.V., Mohri K., Bushida K., Noda M., “Giant magneto-impedance and magneto-inductive effects in amorphous alloys”, J. Appl. Phys., vol. 76, 1994, pp. 6198-6203. Knobel M., Pirota K.R., “Giant magnetoimpedance: Concepts and recent progress”, J. Magn. Magn. Mater., vol. 242, 2002, pp. 33-40. Bulavin L.A., Kravets V.G., Vinnichenko K.L., Manko D.Yu., „Optical Properties of Amorphous Co-Containing Alloys in the IR Region of the Spectrum and Their Magnetoresistive Characteristics”, Journal of Applied spectroscopy, vol. 68, no. 5, 2001 pp. 599-604. Brunetti L., Tiberto P., Vinai F., Chiriac H., “High frequency giant magnetoimpedance in joule-heated Cobased amorphous ribbons and wires”, Mater. Sci. Eng., vol. 304-306, 2000, pp. 961-964. Kurlyandskaya G.V., Barandiaran J.M., Vazquez M., Garcia D., Dmitrieva N.V., ”Influence of gemetrical parameters on the giant magnetoimpedance response in amorphous ribbons”, J. Magn. Magn. Mater., vol. 215216, 2000, pp. 740-742. Makhnovskiy D.P., Panina L.V., “Size effect on magneto-impedance in layered films”, Sensors and Actuators, vol. 81, 2000, pp. 91-94. Hye-S K., Heebok L., Kyeongsup K., Seong-Cho Y., YongKook K., “Temperature dependence of the magnetoimpedance effect in nanocrystalline Fe84Zr7B6Cu1Al2 alloy”, Journal of Alloys and Compounds, vol 326, 2001, pp. 309-312. Kravets V.G., Petford-Long A.K., Portier X., Poperenko L.V., Kolesnik M., “The optical and magneto-optical properties and magnetoresistance of amorphous

Articles

N° 4

2009

CoFeNiSiB alloys”, J. Magn. Magn. Mater., vol. 217, 2000, pp. 129-138. Panina L.V., Mohri K., “Magneto-impedance in multilayer films”, Sensors and Actuators, vol. 81, 2000, pp. 71-77. Wang X.Z., Yuan W.Z., Li X.D., Ruan J.Z., Zhao Z.J., Yang J.X., Yang X.L., Sun Z., “Enhancement of giant magnetoimpedance in composite wire with insulator layer”, J. Magn. Magn. Mater., vol. 308, 2007, pp. 269-272.


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

AN XPS STUDY OF RF SPUTTERED Ti1-xFexO2-d THIN FILMS

Dumitru Luca, Radu Apetrei, Marius Dobromir, Vasile Dascaleanu, Cristian-Mihail Teodorescu

Abstract: Introduction of small amounts of dopant Fe cations in titania was mainly intended for improving the oxidative power of the host TiO2 surface, by shifting the absorption edge of the material towards the blue side of the visible range. Apart from this, other possible applications of Fecontaining TiO2 materials have been foreseen in the field of diluted ferromagnetic semiconductors for spintronics. As a general remark, while pristine and/or low-level doped titania materials remained a hot subject in basic research, and environment- and energy related applications, reports on Ti1-xFexO2 materials (x > 0.1) scarcely occurred in literature. We have initiated recently an investigation of heavily doped titania thin films within a wider iron composition range. Here, we report on the result of an XPS study of the RF sputtered oxygen-deficient Ti1-xFexO2-d thin films (x = 0.15 - 0.62) and discuss the connection of the results with the macroscopic properties of the materials. Keywords: Ti1-xFexO2-d thin films, XPS, AFM, Fe2O3, hematite, magnetite.

films, 1 to 4 sintered pellets of Fe2O3, 2 mm in diameter, 1 mm thick, were placed in successive deposition runs on top of the facing-up TiO2 target, within the intensive sputtering area. Film surface morphology was evaluated from the AFM images taken using a NT-MDT Solver Pro 7 atomic force microscope, operated in tapping mode. X-ray Photoelectron Spectroscopy (XPS) measurements have been done to derive surface elemental composition and chemical state of O, Ti and Fe. The measurements have been done in a Physical Electronics PHI5000 VersaProbe XPS system using the mono-chromated Al Ka radiation (hn = 1486.6 eV). The take-off angle of the photoelectrons was 45°. All the XPS peak positions in the survey spectra were calibrated with respect to the C 1s peak at 284.6 eV. Information on the films structure was derived from XRD patterns (Dron 2 diffractometer using Cu Ka radiation, l = 1.540 Å). The band gap values were calculated from transmittance data (Perkin Elmer) using the standard procedure [6].

3. Results and Discussion 1. Introduction Pristine and doped TiO2 materials have been under intensive research for decades, leading to well-documented applications in environment- and energy-related fields [1]-[3] High efficiency titania materials based on anatase TiO2 are nowadays utilized as photocatalytic, bactericidal and super-hydrophilic materials, or as key components in gas sensors and dye-sensitised solar cells. On the other hand, diluted magnetic semiconductors prepared by doping titania with small amounts of ferromagnetic impurities (Co, Fe) are now under intensive research for possible applications in spintronics [4]. While the information in the literature is mainly concerned with the effect of low-level substitutional Fe doping of the titania thin films, here we extend our approach towards full characterization of Ti1-xFexO2 materials in relation with their nano-structure, optical and electric properties.

The 200 nm thick film samples have been labeled as Fe0, and Fe1 to Fe4, in accordance with the number of Fe2O3 pellets used for film preparation, as described in the previous section. The main characteristics of the samples are shown in Table 1. A 1 μm × 1 μm detail in the AFM surface picture of the Fe3 sample is shown in Fig. 1, as relevant example. Small grains spread within a rather smooth surface occur in all the investigated samples. The AFM images showed that the surface roughness gradually diminished by introduction of iron in the films.

2. Experimental details 300 nm thick films of Ti1-xFexO2-d (TFO) have been grown on soda lime glass and Si(100) substrates in a home-built RF magnetron sputtering facility (13.56 MHz, -5 2.5×10 mbar base pressure) [5] under the conditions of constant forward RF power (80 W) and Ar gas discharge -3 pressure (5.5×10 mbar). A 7.62 cm diameter ceramic TiO2 disk target (K. J. Lesker) was used to grow the reference film (x = 0). In all the deposition experiments, the film substrates were kept at 250°C. To fabricate TFO

Fig. 1. A 3-D AFM image of the Fe3 film surface. A peak value of 13 nm was found for the average surface roughness of the reference sample (Fe0). The TFO films' average roughness remains below this value, fluctuating between 5 and 8 nm, upon increasing Fe concentration in the films (see Table 1). Articles

15


Journal of Automation, Mobile Robotics & Intelligent Systems

Fig. 2. High resolution scan of the O 1s peak (sample Fe3).

VOLUME 3,

N째 4

2009

The position and the shape of the O 1s core level peak remain practically unchanged in all the investigated samples. This peak can be deconvoluted into: (a) a main 2+ component, located at BE = 530.8 eV, related to the O-Ti bonding and (b) a weaker one (BE = 531.6 eV), shifted with 0.5 eV towards higher binding energies, compared 3+ with BE = 531.1 eV, specific to the O-Ti bonding [2] (see Fig. 2). This shift is related to additional effect of the socalled surface component, reported in ref. [3] due to adsorbed OH- radicals in the surface. Along with the main TiO2 oxide component, smaller amounts of TiO and Ti2O3 are present in the films. This conclusion is evident from the core level Ti 2p3/2 XPS spectra (an example of the deconvolution is shown in 2+ 3+ Fig. 3). The normalized peak areas of the Ti and Ti 4+ signals with respect to the Ti main peak vary from 0.68: 0.40:1.00 for x = 0.15; to 0.21:0.30:1.00 for x = 0.29 to 0.13:0.11:1.00 for x = 0.46 and 0.07:0.13:1.00 when x = 0.62.

Fig. 3. High resolution scan of the Ti 2p peak (sample Fe3). The XPS data showed that the Fe/Ti atomic ratio of the as-prepared films increased monotonically by increasing the number of the Fe2O3 pellets (samples Fe0 to Fe4) as shown in Table 1. All the samples were oxygen deficient, as no O2 was introduced in the discharge gas during deposition to compensate for oxygen depletion [7]. Even in the reference sample (Fe0), d is 0.66. As seen in Table 1, the O/(Ti+Fe) atomic ratio diminishes gradually up to a value of 0.82 in the highest iron content TFO films. The Fe/Ti atomic concentration ratio increases monotonically from 0.00 to 1.60 under the same conditions (see Table 1).

Fig. 4. High resolution scan of the Fe 2p peak (sample Fe3). 16

Articles

Fig. 5. The transmittance spectra of the TFO films. The peak area of the core level Fe 2 p3/2 peak increases by a factor of slightly larger than 4 after increasing the number of Fe2O3 pellets. In all the Fe-containing films, the Fe 2p peak could be deconvoluted into three components: 0 (a) the Fe non-reacted component, whose weight fluctuates around 32% from the total amount of Fe in the 2+ sample; (b) the Fe component weighting approx. 24 % of 3+ the total amount of Fe, and (c) the Fe component - the balance. An example of the de-convolution of the 2p3/2 peak of the Fe3 sample is shown in Fig. 4. Previously, Fe K - edge XANES measurements have been done on lower iron - content Ti1-xFexO2 films (x ranging between 0 an 0.15) [8]. The results for nearestneighbour configuration showed that both magnetite and metal Fe co-exist in the samples with x = 0.15, while the EXAFS spectra at the same edge showed the existence of hematite. This is so, since the Fourier transforms rely on a (distorted) second coordination shell (due to the prend sence of Ti cations), similar to the 2 coordination shell of hematite. Weak anatase A(101) and A(004) diffraction peaks occurred in the XRD patterns of the Fe0 reference film, exclusively (not shown here), while the diffraction features of the TFO samples remained below the detection limit of the diffraction instrument. Increasing the iron content results in dramatic alteration of the optical parameters of the films. The transmit-


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

tance spectra (Fig. 5) of the films show that the threshold, a, of the fundamental absorption of the Ti1-xFexO2 2 films could be described by the expression: a = A(E-Eg) , where A is a constant, E and Eg are the optical band gap of the ferromagnetic film and pure anatase TiO2, respectively. The value of the exponent is a characteristic for the indirect allowed transition dominating over the optical absorption [6]. The optical band gap values, derived according to the procedure by Sreemany and Sen [6], are listed in Table 1. The corresponding shifts in the absorption edge of the TFO materials are equivalent to a shift in the absorption edge in terms of wavelength from 413 nm (sample Fe0) to 564 nm (sample Fe4). Table 1. Sample main data.

Sample

x

Fe0 Fe1 Fe2 Fe3 Fe4

0.00 0.15 0.29 0.49 0.62

R(nm) O/(Ti+Fe) Fe/Ti Eg atomic ratio atomic ratio (eV) 13 8 5 5 8

1.34 1.22 1.10 0.79 0.82

0.0 0.34 0.54 1.33 1.60

3.0 2.7 2.5 2.3 2.2

N° 4

2009

for publication further on.

ACKNOWLEDGMENTS The financial support from the Romanian Ministry of Education, Research and Innovation, through Grant 71-63/2007 (MAMAINCOPAE) is acknowledged. The support from Dr. D. Timpu from the Institute of Macromolecular Chemistry "Petru Poni" in Iasi, Romania, in the AFM measurements is appreciated.

AUTHORS Dumitru Luca*, Radu Apetrei, Marius Dobromir, Vasile Dascaleanu - Alexandru Ioan Cuza University of Iasi, 11 Carol I Blvd. 700506-Iasi, Romania. Phones: +40 232 20 1179, +40 232 20 1171, fax: +40 232 20 1150. E-mails: dumitru.luca@uaic.ro, rapetrei@uaic.ro, marius.dobromir@uaic.ro, dascaleanu@yahoo.com. Cristian-Mihail Teodorescu - National Institute of Materials Physics, Bucharest, 077125 Magurele-Ilfov, Romania. Phone: +40 21 369 0170, fax: +40 21 369 0177. E-mail: teodorescu@infim.ro * Corresponding author

References

4. Conclusion 300 nm thick, amorphous Ti1-xFexO2-d films (with 0 < x < 0.62 and 0.66 < d < 1.18) have been fabricated by RF magnetron sputtering of a titania sintered target. The iron content in the sample has been adjusted using a mosaic configuration, by placing a variable number of Fe2O3 sintered pellets on the magnetron surface in the highrate sputtering area. The XPS results showed that a non-reacted Fe0 component occurs in all the iron-containing as-deposited samples, spread in the amorphous matrix. This is accompanied by a red shift of the absorption edge of approximately 151 nm and a decrease of the band gap from 3.0 to 2.2 eV, in the same range of Fe composition. The oxygen-depletion in the films is related to the important fraction of the high-valence state of titanium and Fe cations. The XPS results suggest that it is quite probably that magnetite, hematite and metallic Fe nanodomains can be present in the films, as spread out in the amorphous material, as demonstrated for samples with x below 0.15 [8]. The current results extend the composition range of the TFO materials above the limit value x = 0.15, which is the maximal value reported in ref. [8] At least for this limit value, where we can compare the results, the current XPS results confirm the former XANES and EXAFS results, i.e.: (a) A predominant TiO2 phase coexist with significant TiO and Ti2O3 phases (but no metal Ti); (b) magnetite, hematite and metal Fe co-exist with hematite. Further information on the local order around the cations species and type of Fe incorporation in the range of x between 0.15 and 0.62 is expected from the last-hour performed XANES and EXAFS measurements at the HASYLAB synchrotron facility in Hamburg, Germany. The data are now under processing, and the results are scheduled

[1]

[2]

[3]

[4]

[5] [6]

[7]

[8]

Thompson T.L., Yates J.T. Jr., „Surface science studies of the photoactivation of TiO2 new photochemical processes”, Chem. Rev., no. 106, 2006, pp. 4428-4453. Luca D., Teodorescu C.M., Apetrei R., Macovei D., Mardare D., „Preparation and characterization of in-creased - efficiency photo-catalytic TiO22xNx thin films”, Thin Solid Films, no. 515, 2007, pp. 8605-8608. Luca D., Macovei D., Teodorescu C.M., „Characterization of titania thin films prepared by reactive pulsed-laser ablation”, Surf. Sci., no. 600, 2006, pp. 4342-4346. Coey J.M.D., Douvalis A.P., Fitzgerald C.B., Venkatesan M., „Ferromagnetism in Fe-doped SnO2 thin films”, Appl. Phys. Lett., no. 84, 2004, pp. 1332-1334. Luca D., Mardare D., Iacomi F., Teodorescu C.M., Appl. Surf. Sci., no. 252, 2006, p. 6122 f. Sreemany M., Sen S., „A simple spectro-photometric method for determination of the optical constants and band gap energy of multiple layer TiO2 thin films”, Mat. Chem. Phys., no. 83, 2004, pp. 169-177. Mardare D., Iacomi F., Luca D., „Substrate and Fedoping effects on the hydrophilic properties of TiO2 thin films”, Thin Solid Films, no. 515, 2007, pp. 6474-6478. Apetrei R., Negrila C., Macovei D., Dascaleanu V., Teodorescu C.-M., Mardare D., and Luca D., “Fabrication and characterization of nano-structured ferro-magnetic Ti1-xFexO2 thin films”. In: Proc. NSTI Nanotech 2009, www.nsti.org, ISBN 978-1-4398-1782-7, vol. 1, pp. 375-378.

Articles

17


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

THE ASSESSMENT OF “JATROPHA” AS RAW MATERIAL FOR BDF Satoshi Matsuda, Bambang Rudyanto, Mutsuo Kojima, Wino Herdiana, Emiko Fujiwara

Abstract: The “second-generation” of bio-fuel made from nonfood biomass has attracted much attention especially after the Hokkaido Toyako G8 summit held in November 2008. Among various plants or biomass materials, “Jatropha” has come to the front as one of the most promising candidates for future bio-diesel production. In fact, there are many plans or projects aiming at large scale cultivating Jatropha for bio-diesel oil production in several African or Southeast-Asian countries. However, it should be pointed out that there are too many unknown factors still remaining that are important in practical use of Jatropha oil, for example, productivity as well as sustainability, long-term impact of its large-scale use on soil quality, practical costs of the oil and so on. In this study, an assessment on the validity of Jatropha production was tried as quantitatively as possible using the data obtained from a field survey in Indonesia conducted by the authors in January and March of 2009. Although LCI data for LCA calculations obtained was very limited, the true reality of the situation about Jatropha could substantially be demonstrated to some extent. The main point was that the oil cost was dominated mainly by labor costs, because many work operations could not be mechanized and inevitably be manual handling tasks. In many cases, the labor costs could not be covered by the income from the Jatropha oil (or seed) due to the low productivity of the biomass, whereas the prime cost of the biodiesel oil was rather high expensive compared with fossil fuel. Keywords: bio-fuel, Jatropha, feasibility study, field survey, labor costs.

1. Introduction 1.1. Background Prevention of global warming is becoming urgent issue and “biofuel” gathering worldwide attention has been put into practical use not only in US, EU, Brazil or other countries but in Japan. As for its impact on environment, however, evaluation still varies including aninfluence on global economy, for example food price crisis that might have been caused by competition with food. In the previous study [1], one of the authors pointed out that “bio-ethanol” from crops is never “carbon neutral“, that means the production and use of the fuel does not lead to a net increase in atmospheric CO2 concentration, nor useful as a countermeasure against global warming. 18

Articles

Since the shortage of food crops affected by the expansion of biofuel production became a critical issue, the “second-generation” of biofuel made from non-food biomass has attracted much attention all over the world. In particular, the following statement in the Hokkaido Toyako G8 summit held in November 2008 accelerated this trend: “use of biofuels with food security and accelerate development and commercialization of sustainable second-generation biofuels from non-food plant materials and inedible biomass” (in “G8 Leaders Statement on Global Food Security” [2]). The target biomass has a wide variety of kind; cellulosic materials such as woody biomass and paddy straw, algae and some kind of oil producing plants, which are considered as sustainable and research is in progress. Among them “Jatropha” has come to the front as one of the best candidates for future biodiesel production [3], partially because the production of bio-ethanol from cellulosic materials has encountered many difficulties in practice. Jatropha is native to Central America and has become naturalized in many tropical and subtropical areas, including India, Africa, and North America as well as Southeast Asian countries. The hardy Jatropha is resistant to drought and pests, and produces seeds containing 2740% of oil [4] (average: 34.4% [5]). However, despite its abundance and use as an oil and reclamation plant, none of the Jatropha species have been properly domesticated and, as a result, its productivity is variable, and the longterm impact of its large-scale use on soil quality and the environment is still unknown [3]. 1.2. Objective In Indonesia, nevertheless, there is a national program of biofuels development in which 20% of diesel consumption will be supplied with biodiesel utilization mainly by Jatropha oil (equiv. 10.22 million kL) by the year 2016 to 2025. In this study, an assessment on the validity of Jatropha production was tried as quantitatively as possible using the data obtained from a field survey in Indonesia conducted by the authors in January and March of 2009.

2. Results and Discussion 2.1. Bio-Energy Productivity The standard planting rate of Jatropha in Indonesia was 2500 to 2800 trees per hectare, producing about 4 ton of seed (as dry matter yield) per hectare per year after 4 years of planting (note: only a little seed can be harvested in the first year, 1-1.5 t-dry seed/ha/yr in the 2nd


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

year, and 2.5-3 t-dry seed/ha/yr in the 3rd year). Generally speaking, one ton of fresh fruit consists of 300 kg of seed and 700 kg of coat, which is equivalent to 200 kg of dry seed. This means that 5 kg of fruit is required for 1 kg of dry seed, thus 20 ton of fruit per hectare per year should be harvested to obtain 4 ton of dry seed. It was not certain whether or not such a high productivity of biomass can be maintained permanently. About 30 wt% of dry seed is crude oil content as mentioned above. However, this does not mean that 1200 kg(=4000×0.30) of oil can be obtained from 4 ton of dry seed because the practical oil yield using a pressing machine was around 50%, meaning that about 150 liter oil per ton of dry seed (i.e. about 600 liter oil per hectare per year) could be obtained. But this value is larger than those of other oil crops for biodiesel [6] (e.x. 480 liter from soybean, and 390 liter from sunflower). Table 1 shows several examples of overall oil yield in bio-fuel production. Table 1. Overall oil yield in bio-fuel production. Raw material Crop(kg/ha) Corn [6] 8665 Soybean [6] 2688 Sunflower [6] 1500 Jatropha 4000

Oil(kR/t) 0.372 0.180 0.255 0.150

Overall (kR/ha) 3.22(EtOH) 0.480 0.389 0.600

If 10.22 million kR of bio-diesel oil (the goal of bio-diesel production for the year 2016 to 2025 in Indonesia) should be obtained from Jatropha, at least 17 million hectare of planting land area is required even if no oil loss can be achieved during bio-diesel production process from Jatropha crude oil. Incidentally, this land area is corresponding to about 46% of Japanese total land area. 2.2. Energy Balance The data required for the estimation of energy inputs and costs in Jatropha production is as follows; Labor, Machinery, Fuels (Diesel, Gasoline and LP gas), Fertilizers (Nitrogen, Phosphorus and Potassium), Lime, Seeds, Herbicides, Electricity and Transport. In addition, there are other items necessary for biodiesel oil production including cost of equipment and utilities (steam, water, electricity and so on). However, obtained data was quite limited in this time of field survey because the production of Jatropha was still testing stage there, and almost all process was operated by human power. Only obtained data was that the electric power consumption of oil pressing machine was 12-15 kWh per 300 kg of dry seed treated, equivalent to 160-200 kWh per hectare per year. Suppose that the heating value of Jatropha oil is 37620 kJ (9000 kcal) per liter and power generation efficiency of small scale electricity generator is about 15% (1 kWh = 860 kcal/0.15 = 5733 kcal), 160-200 kWh is corresponding to 17-21% of Jatropha oil content (600 liter = 5.4 *106 kcal, 160-200 kWh is 0.92-1.15 *106 kcal). In Table 2 for reference, the data of the inputs for bio-diesel oil production from Soybeans in the US [6]. Since the process of diesel oil production is almost the same regardless of the kind of raw materials, the difference between the values or items in Table 2 and those of inputs for bio-diesel produc-

N° 4

2009

tion from Jatropha will be the difference in raw material only, so that the energy input in bio-diesel oil production will be 4078 (=11878-7800) kcal/kg-oil = 3752 kcal/L-oil suppose that the specific gravity of Jatropha oil is 0.92. Thus, the total energy input in bio-diesel oil from Jatropha will be 5282-5642 kcal/L, that is corresponding to 59-64% of Jatropha oil content. This means about 60% or more energy of the Jatropha oil will be consumed to obtain the bio-diesel oil. Table 2. Inputs Per 1000kg of Bio-diesel oil from Soybeans in the U.S. [6] Inputs Soybeans Electricity Steam Cleanup water Space heat Direct heat Losses Stainless steel Steel Cement Total

Quantity 5556 kg 270 kWh 1350000 kcal 160000 kcal 152000 kcal 440000 kcal 300000 kcal 11 kg 21 kg 56 kg

kcal×1000 7800 697 1350 160 152 440 300 158 246 106 11878

2.3. Economical Consideration From the results of the field survey, it was cleared that labor cost was evidently a major part of the expense of Jatropha oil production since almost all process was operated by human power as mentioned above. The data of labor costs concerning the oil production from Jatropha obtained from this time of field survey was as follows; 1) Fertilizer Application: 5 person/ha, 2 days 10 person·day/ha 2) Planting Work: 4-5 person/ha, 10 days 40-50 person·day/ha 3) Mowing Grass: 2 persons/ha, 5 days 10 person·day/ha 4) Harvesting: 4-8 persons/ha, every day harvest: 50-70 kg-fruit/person·day Since 20 ton of fruit per hectare per year should be harvested, 333 person·day/ha·yr is needed if average harvest of 60 kg-fruit/person·day can be assumed (20000/60). Then, the sum total labor should be 393-403 person·day per hectare per year. Here, 350 person·day/ ha·yr person·day is supposed because all work operations is not always required every year. The labor cost of male worker was 30 thousand Rupiah (Rp: 1 US dollar = 10100 Rp, 1 Japanese Yen = 106 Rp, in 2009) per day. Thus, the labor cost should be 10.5 million Rp per hectare per year. On the other hand, the income from Jatropha should be only 4 million Rp per hectare per year because the price of dry seed was about 1000 Rp/kg and the seed yield was 4000 kg/ha·year. In fact, other expenditure of about 1.25 million Rp per hectare per year for purchasing nursery trees and maintenance of the field would be required. This means that Jatropha production is not commercially viable operation for farmers by itself. Articles

19


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

From the standpoint of fuel cost, however, Jatropha oil is not cheep: the price of dry seed was about1000 Rp/kg, and 0.15 kg of crude Jatropha oil can be obtained, meaning that the raw material cost of the oil will be higher than about 6667 (=1000/0.15) Rp/kg = 0.66 U.S. dollar/kg. This value is rather expensive as raw material of diesel fuel since whole production cost consists of both raw material and the facilities. This economical analysis implies that the Jatropha crude oil should be utilized as a heavy oil alternative by direct combustion.

References

3. Conclusion

[5]

According to the results of the survey data and calculations, it was revealed that Jatropha will not be a hopeful candidate for future biodiesel production although the concern about the environmental impacts was not researched or discussed in this study. The main points are as follows; 1) The biomass productivity of Jatropha is rather large (About 20 ton of fruit per hectare per year can be harvested) compared with other oil crops for biodiesel such as soybean and sunflower. But huge land area would be required if the plan of Jatropha oil in Indonesia should be achieved because only 0.6 kL of oil can be obtained per hectare per year in practice. 2) Although the analysis of energy balance could not be achieved completely due to the lack of field data, it was revealed that more than 60% of Jatropha oil content would be consumed for the production of bio-diesel oil. 3) Economical analysis showed that the oil cost was dominated mainly by labor costs because many work operations could not be mechanized and inevitably be manual handling tasks. In many cases, the labor costs could not be covered by the income from the Jatropha oil (or seed) due to the low productivity of the biomass, whereas the prime cost of the bio-diesel oil was rather expensive compared with fossil fuel. ACKNOWLEDGMENTS This work has been financially supported by The Mitsui & Co., Ltd. Environmental Fund.

AUTHORS Satoshi Matsuda* - Department of Materials Science and Chemical Engineering, Shizuoka University, 3-5-1 Johoku, Naka-ku, Hamamatsu, 432-8561 Japan. E-mail: tcsmats@ipc.shizuoka.ac.jp. Bambang Rudyanto - Faculty of Economics and Business, Wako University, 2160 Kanaimachi, Machida-shi, Tokyo, 195-8585 Japan. Mutsuo Kojima - Department of Biological and Environmental Science, Faculty of Agriculture, Shizuoka University, 836 Ohya, Suruga-ku, Shizuoka, 422-8529 Japan. Wino Herdiana - Along Japan Center, JI. Raya Padjajaran no.21, Warung Jambu, BOGOR 16153 Indonesia. Emiko Fujiwara - Japan Tropical Forest Action Network, 1-23-16 Shinzyuku, Shinzyuku-ku,Tokyo,160-0022 Japan. * Corresponding author 20

Articles

[1]

[2] [3] [4]

[6]

N° 4

2009

Matsuda S., “Validity of Bio-Ethanol as a Countermeasure against Global Warming”, J. Environ. Information Sci., vol. 37, issue 5, 2009, pp. 1-6. http://www.mofa.go.jp/policy/economy/summit/ 2008/doc/doc080709_04_en.html Fairless D., “Biofuel: The little shrub that could - maybe”, Nature , no. 449, 2007, pp. 652-655. Achten W.M.J., et.al., “Jatropha biodiesel fueling sustainability?”, Biofuels, Bioproducts and Biorefining, vol. 1, issue 4, 2007, pp. 283-291. Achten W.M.J., ���Jatropha bio-diesel production and use”, Biomass and Bioenergy, vol. 32, issue 12, 2008, pp. 1063-1084. Pimentel D., Patzek T. W., “Ethanol Production Using Corn, Switchgrass, and Wood; Biodiesel Production Using Soybean and Sunflower”, Natural Resources research, vol. 44, issue 1, 2005, pp. 65-76.


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

INVESTIGATION INTO ELECTROCHEMICAL DISCHARGE MACHINING OF MICROHOLES Leszek Kudla

Abstract: The paper presents fundamentals and technology of electrochemical discharge machining (ECDM) of microholes in borosilicate glass and partly in diamond crystals. The both mentioned materials have very useful technical properties but are difficult to machine using conventional or even laser techniques. The ECDM gives another chance to reduce some machining problems under condition of precise selection and control of the process means. Keywords: electrochemical discharge machining (ECDM), micromachining, microholes.

tool electrode, which is thinned to form a tip. The auxiliary electrode is stepped aside from the tool and machined workpiece. Discharges (8) generated on the tool electrode caused a local heating and thermal and chemical erosion of a workpiece material. Selection of the borosilicate glass and diamond as investigated materials has different reasons. The borosilicate glass serves as the substrate for the microsensors due to its adequate physical and chemical properties. Among the others its ability for the anodic bonding is of a great importance [8]. The diamond is widely applied for the technological tools. The both materials are hard, brittle and difficult to machine.

1. Introduction The microholes with dimensions below 0.5 mm are typical structures of many miniaturised components, produced from electrically non-conductive materials. The microholes can be made not only mechanically or by lasers but also by utilising various physical or chemical phenomena [1], [2], [3]. An example may be electrochemical discharge machining (ECDM), executed in electrolytes [4], [5], [6], [7]. The machining technique is also named as electrochemical arc or spark machining (ECAM), or as spark assisted chemical engraving (SACE). The basic of the process is the effect of creating and then electrical breaking of resistive gas layer. If the electrodes are of considerable different sizes, then mentioned phenomena concentrated on the smaller electrode. In some conditions, as a consequence of a large current density, continuous discharges occurred.

5 DC/AC

1 2

3 8 7

2. Fundamentals of ECDM To the machining mechanism may contribute as such processes as melting and vaporisation of material, its dissolution by electrochemical reactions, intensified in high temperature, and various other effects, playing a smaller part, like producing of pressure shocks in electrolyte, generation of thermal stresses and forming of microcracks in workpiece material [4], [5]. When electrical current rises with the applied voltage, the tool electrode, electrolyte (water solutions) and auxiliary electrode construct an electric circuit, and in a first stage pure electrochemical reaction occurs. On the electrodes are liberated bubbles of the hydrogen and oxygen. The non-conductive gas layer on the smaller tool electrode has changeable resistance, depending on layer stability by certain voltage. Only sometimes spars discharges across non-compact gas layer appeared. In following stage the production rate of gas bubbles and their leaving rate are in a balance, generating compact gas layer. The voltage of this stage is critical for creating of the gas column and a start of continuous discharge (a kind of arc discharge) and further efficient machining – Fig. 2. b)

a)

4

TOOL ELECTRODE

TOOL ELECTRODE

WORKPIECE

WORKPIECE

6

Fig. 1. Outline of system for electrochemical discharge machining (description in the text). Machining system consists of a container (1) with the electrolyte (2), tool (3) and auxiliary (4) electrodes, dipped in the electrolyte and supplied by direct (sometimes also alternating) current (5) – Fig. 1. In a special holder (6) the machined workpiece (7) is placed very close to the

Fig. 2. Main utmost stages of phenomena in surrounding of tool electrode: generation of gas bubbles (a) and discharges (b). Articles

21


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

Further increasing of the voltage causes increasing of discharge energy. If U assigned the applied voltage, I the mean current and R resistance of electrolytic part of circuit, the mean heat of the spark (energy per second) Q is the mean energy provided into the system diminished by the energy for Joule heating [4] 2

Q=U×I-R× I

(1)

From various calculations and measurements the energy released in a single discharge is evaluated for 0.2 2 J/mm by the duration of discharge of 0.1 ms. The resistance of electrolytic part of circuit may be expressed as a sum R = RAe + Re + RTe

(2)

where RAe assigned the resistance of the interface auxiliary electrode/electrolyte, Re the resistance of electrolyte and RTe the resistance of interface tool electrode/ electrolyte. The resistance RTe varies significantly in a function of applied voltage, corresponding to the basic phenomena on the tool electrode. The resistance of electrolyte can be evaluated from relationship Re =

l s×A

(3)

(4)

where a is the degree of dissociation, F the Faraday constant, μn and μp the mobility of negative or positive ions correspondingly. In practice the conductivity of electro-2 2 lytes is at a range of 10 ...10 S/m. A very thinned electrolyte has conductivity of 0.1...0.2 S/m. When a workpiece is made of the borosilicate glass the main chemical reaction is as follows SiO2+2NaOH ® Na2SiO3+H2O or SiO2+2NaOH+H2O ® Si(OH)4+Na2O Similar reactions can take place with the other components of the glass (B2O3, BaO and Al2O3). In case of the diamond as a workpiece and by alternating current, the chemical reactions depend on instantaneous polarity of the tool electrode. If the oxygen is released, it acts on the graphite and the carbon dioxide or oxide is produced 3C+2O2 ® CO2+2CO When the tool electrode is cathode the result of reaction of the hydrogen with carbon is methane C+2H2 ® CH4 The advantage of such a polarity is that the hydrogen is neutral against the diamond and reacts only with graphite. 22

Articles

2009

3. Technology of ECDM In few works has been reported electrochemical discharge microstructuration of borosilicate glass 7059 [5], [7]. The electrolyte was 10…20 wt % water solution of NaOH. The direct current has been applied with voltage of 20…60 V. The auxiliary electrode was made of graphite or zinc. The tungsten, tungsten carbides, stainless steel and other metals or alloys were used as a tool electrode. Machining time of a single hole with diameter of 150 μm in a glass plate having the thickness of 1.5 mm was few minutes. In machining process of microholes with diameters of 20÷60 μm in diamond, electrolytes like KNO3 or H2SO4 water solutions (1…10 wt %) are used. Tool electrode may be made of platinum, tungsten and of alloys like Pt-Ir or Ni-Cr [1]. The system is supplied by alternating voltage of relatively large value (80…220 V). The process phenomena caused forming firstly a cavity and then a microhole. The time of machining may be relatively large, up to few hours.

4. Description of experimental conditions For experiments has been completed setup consisting of the X-Y table, feed drive for tool electrode, electrical supply device and observation microscope. The container with the auxiliary electrode and specimen holder has been placed on the table – Fig. 3.

where l represents the distance between electrodes, s the conductivity of electrolyte and A the area of the tool electrode surface. The conductivity of electrolyte is calculated from the equation s = a × F (μn + μp)

N° 4

VOLTAGE CLIP TOOL ELECTRODE

VOLTAGE CLIP

AUXILIARY ELECTRODE

CONTAINER ELECTROLYTE

X-Y TABLE

Fig. 3. View of machining area of experimental setup (the microscope is removed). The auxiliary electrode was made of acid resistive 2 steel. Its single side area was ~140 mm . The tool electrodes with diameters smaller than 0.5 mm and made of tungsten or tungsten carbides have been used in machining tests. The investigated range of the voltage by direct current was 20…32 V. In a first step of the research basics of process have been determined. The observation of phenomena on tool electrode in function of electrical and geometrical variables was carried out. Next, in selected conditions, the value of the critical voltage has been measured. Finally, the trials of machining for chosen conditions and specimens have been done. Except of typical parameters very important appeared a shape of the tool electrode.


Journal of Automation, Mobile Robotics & Intelligent Systems

VOLUME 3,

N° 4

2009

5. Results and discussion of machining experiments When the tool electrode is dipped in the electrolyte, the increasing of the voltage caused generation of the gas bubbles with various sizes. Approximately by U » 20 V the electrochemical process intensified and the bubbles become very small – Fig. 4(a). The continuous discharges begin by U 24…25 V – Fig. 4(b). The mean current increased with growing of the immersion depth but fell down for larger voltage – Fig. 5. The fluctuations of instantaneous values of the current may be relatively large, up to 1 A. The specimens were glass 7059 plates with the thickness of 1.1 mm. During 1 hour machining the temperature of the electrolyte increased from 26 to ~50 °C. a)

b)

Fig. 7. View and profile of ring-shaped groove with outside diameter of 500 μm and maximal depth of 40 μm.

DISCHARGES

Fig. 4. View of gas bubbles by U=20 V (a) and discharges by U=32 V (b) generated on tungsten tool electrode d= 0.36 mm.

0,3 0.30

U=32.0 V U=29.6 V

0,25 0.25

Current I , A

Fig. 6. Examples of cavities with diameters of 120 μm and 220 μm.

One of the possibilities for exclusion of the listed problems consists in modification of the electrode shape – Fig. 8. For minimising of the activity of the side discharges a reverse reduction of the electrode cross-section may be applied. Another way can be partial shielding of the tool with non-conductive and heat resistive material. In any case for machining of through microholes a conical shaped tip seems to be a proper solution.

U=27.1 V U=24.8 V

0,2 0.20 0,15 0.15 0.10 0,1 0.05 0,05

Fig. 8. Special shapes of tool electrodes.

00 0,5 0.5

1 1.0

1,5 1.5

2 2.0

2,5 2.5

Immersing depth g, mm

Fig. 5. Relationship: current versus immersing depth of tool electrode for various voltages. Machining caused significant oversize of the cavity or microhole in comparison with tool electrode diameter. The reason is side sparks in surrounding of electrode surface. When immersing depth rises, the discharges enlarge existing hole. In result a machined surface is conical and irregular. Only shallow cavities have proper geometry – Fig. 6. They were machined with electrodes having sharp tip. If the tip is flat and touches the specimen surface the machining has very special course and particular results are obtained. The discharges occurred only on the side surface of the tool electrode causing removal of material in ring-shaped groove – Fig. 7. The depth and outside diameter of the groove increased with the machining time.

Some examples of the special electrodes, fabricated in laboratory conditions, are shown in Fig. 9 and Fig. 10. They are made of tungsten wires. The conical electrode has been shaped using electrochemical machining in the same setup but by reversed polarity. A spherical tip on the next electrode was done by a short-circuit discharge. For the machining of the last, cylindrical electrode with undercut diameter, abrasive techniques were used. Presented examples reveal a variety of possible design and executing solutions. Their usefulness for a certain application is another question.

Fig. 9. View of tungsten electrode with sharp conical tip. Articles

23


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

(a)

[8]

(b)

Fig. 10. View of various shaped electrodes: with sphere on the end (a) and reduced diameter (b).

6. Concluding remarks The electrochemical discharge machining is interesting technology but its more detailed researches and further development are necessary. Practical implementation of the ECDM technique needs investigation as of basics phenomena as well of process technology and applied technical means.

AUTHOR Leszek Kudla - Ph.D., Department for Precision and Electronic Products Technology, Warsaw University of Technology, Faculty of Mechatronics, Warsaw, 02-525, ul. A. Boboli 8, Poland. E-mail: kudla@mchtr.pw.edu.pl.

References [1]

[2]

[3]

[4]

[5]

[6]

[7] 24

Kudla L., “Non-conventional techniques used for math chining of microholes”. In: Proc. of the 7 Int. Conf. th th Inter-Academia 2008, 15 -18 Sept., 2008, Pecs, Hungary, pp. 186-195. Rajukar K.P., et al., “Micro and Nano Machining by Electro-Physical and Chemical Processes”, Annals of the CIRP, vol. 55, no. 2, 2006, pp. 643-666. Ruszaj A., Non-conventional manufacturing methods of machine components and tools. Edition of IOS, Kraków 1999 (in Polish). Jain V.K., et al., On the analysis of the electrochemical spark machining process”, Int. Journal of Machine Tools & Manufacture, vol. 39, 1999, pp. 165-186. Yang C.T., Ho S.S., Yan B.H., Micro Hole Machining of Borosilicate Glass through Electrochemical Discharge Machining (ECDM). Key Engg Materials Vol. 196, 2001, pp. 149-166. Wüthrich R., Fascio V., “Machining of non-conducting materials using electrochemical discharge phenomenon an overview”, Int. Journal of Machine Tools & Manufacture, vol. 45, 2005, pp. 1095-1108. Maillard P., et al., “Geometrical characterisation of

Articles

N° 4

2009

micro-holes drilled in glass by gravity-feed with spark assisted chemical engraving”, Journal of Micromechanics and Microengineering, vol. 17, 2007, pp. 13431349. Dziuban J.A., Technology and application of micromechanical silicon and silicon-glass structures in microsystems technique. OW PWr, Wroclaw 2004 (in Polish).


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

AN INTRODUCTION OF HIGH-PRECISE 3D MEASUREMENT SYSTEM AND ITS APPLICATIONS Takeshi Hashimoto, Mitsuo Kaneko, András Rövid, Shinji Isono, Tomoyuki Sone, Kazushige Baba, Akira Fukuda, Masamu Aniya, Nozomu Naito, Hiroyuki Enomoto, Pedro Skvarca

Abstract: We have been studying the high-precision three-dimensional measurement system using cameras. In this paper, our system is verified, and its application is reported. The principle of our measurement system is based on the stereo measurement. In our system, reducing the quantization error by numerical processing can realize the highprecision 3D measurement at the long distance. This measurement system has many applications, such as industrial use, ITS, robotics, and many others. Our system has been applied on the observation of the glacier. The Perito Moreno glacier, which is our target, moves very fast and the moving speed is reported as about 2 m/day. Then, it is thought that the important information about the global warming can be obtained by observing this glacier. But, as the width of the glacier is several km, there were no reports of continuously observing the movement of the glacier in high accuracy. As a result, the unique observation results about the daily movement of the glacier were measured. The observation of the glacier flow was carried out for the third straight year. In this paper, the valuable results of the glacier observation and experimental results to verify the precision are discussed. Keywords: Image measurement, 3D measurement, Glacier observation

1. Introduction We have been developing the high precision measurement system using multi-cameras. The accuracy of our developed system was verified at the several circumstances. We applied our system on the measuring Perito Moreno glacier. In this paper, our system is verified in the imitation environment of glacier measurement and the glacier observation results are reported. 1.1. Stereo vision The stereo vision is the basic method to measure the target by using two cameras in principle. After the target image is taken with two cameras, the each line of sight from camera to the target can be calculated using the camera calibration. The intersection of two lines of sight shows the target position. 1.2. Improving measurement accuracy The stereo vision using two cameras contains the error due to discreteness of image sensor devices, such as CCD and CMOS. This error is call as the quantization error. Our research results can show that the measurement accuracy is improved by averaging intersections using mul-

ti-cameras. Concretely, ten cameras were used for one observation target point in this glacier observation. Then each target position is calculated by averaging over 20 measurement values. This process can reduce the quantization error dramatically. Our computational method is described in detail in the reference [6]. 1.3. Experiments in this paper In this paper, two experiments are described. Experiment 1 was carried out to verify the accuracy of positional measurement. In Experiment 2, the observed movements of the glacier are shown as one application of our high-precise 3D measurement system.

2. Experiment 1 2.1. Purpose The purpose of this experiment is to verify the accuracy of positional measurement of the glacier. 2.2. Experimental equipment, placement of cameras and target — In this experiment, we used eight digital cameras, Optio W60 (PENTAX Inc., 10 mega pixels), that were used in the glacier observation. One camera group is made up of four cameras. — The target color is similar to the glacier color. Conretely, the target color is white on the back board. The size of target is small compared with the real glacier. This difference is thought to contribute to improving the error of measurement. — To carry out the camera calibration, the optical position measurement instrument, Total Station was used. The range discrimination and the angle discrimination of Total Station is plus or minus 1.0 mm and plus or minus 20 [arc second], respectively. — We carried out the experiment on the scale of 1/10 compared to the placement of glacier observation (refer to Fig. 2). Therefore, the distance from the camera to the target was 100 m. 2.3. Measurement procedure — Before measurement, we measured three-dimensional coordinate of the target and camera groups by using Total Station. — The target images were photographed with each camera. — The corresponding point of target was extracted on each image. — Then, the position of the target was calculated Articles

25


Journal of Automation, Mobile Robotics & Intelligent Systems

based on the principle of the stereo vision method by using the measured coordinate of each camera pair. 2.4. Results of Experiment 1 Figure 1 shows results of measurement of the target. This graph is drawn as a plan view. In this experiment, the maximum error of this experiment is 9.0 mm. Our previous experiment reported that the maximum error tended to be proportional to the distance between cameras and target. For example, the maximum error at the distance 200 m increased about 1.6 times the maximum error at the distance 100 m. That is, it is inferred that the maximum error at the distance 1000 m increases 8 - 10 times the maximum error at the distance 100 m. As a result, it is thought that the accuracy of positional measurement of the glacier is the plus or minus 70 - 90 mm. The error of plus or minus 70 - 90 mm is under 10 % of the error rate because the movement speed of Perito Moreno glacier is reported to be 1 - 2 m/day. This accuracy is thought to be enough for this glacier observation.

VOLUME 3,

N째 4

2009

ing an optical measuring instrument and GPS is general. Since our system can realize the high-precise 3D measurement, this experimental result and other experimental results can be complemented mutually. 3.2. Observational days From December 2008 to January 2009, about two weeks, at Perito Moreno glacier on Patagonia in Argentina, we measured the movement of the glacier by using ten cameras. We have been trying the high-precision observation of the glacier using the stereo camera system for about five years. 3.3. Experimental equipment, placement of cameras and target Optio W60 has the appropriate features for the glacier observation. The most useful feature is the interval photography function that enables unmanned observation for a week. The other reasons why we selected this camera are its small size, waterproof function and so on. Further, this camera lens does not extend outside of its body. This mechanism can reduce the effect of strong wind. One camera group is made up of five cameras. We set one pair of camera groups at the downstream of the glacier. The positions of camera groups and calibration points were measured by Total Station for the camera calibration. Peaks of the glacier were selected as calibration points. The calibration point and camera group are shown in Fig. 2. Distance from camera groups to calibration point is about 1,000 m. Distance between two camera groups is 470 m.

Fig . 1. The measurement results of the target. 2.5. Discussions of Experiment 1 The accuracy of the positional measurement of the glacier could be confirmed by Experiment 1. For details, the result of Experiment 1 shows that the maximum error is 410 mm by using two cameras and the maximum error is 9.0 mm by using eight cameras, respectively. Accordingly, the maximum error has improved by about 50 times by using eight cameras. The average error is 60 mm by using two cameras and the average error is 3.5 mm by using eight cameras. Thus, the average error has improved by about 20 times by using eight cameras. As a result, it is said that Average Method is effective in accuracy improvement.

3. Experiment 2 3.1. Purpose and background The purpose of this experiment is to measure the movement of Perito Moreno glacier with high precision. Perito Moreno glacier in Argentina is famous as the World Natural Heritage Site. This glacier has fast movement speed and is called "an alive glacier". It is thought that that earth environment change is sensitively reflected on this fast glacier movement. Moreover, some unsolved problems of this glacier should be researched. For the measurement of this glacier, the method of us26

Articles

Fig. 2. Placement of camera groups and peak of the glacier. 3.4. Measurement procedure Optio W60 took pictures of target glacier every one hour for two weeks. An example image is shown in Fig. 3. In this figure, numbered points are targets. No.1 is the calibration point. During taking pictures, the cameras were vibrated by the strong wind. Then, we corrected each image based on the edge of the mountain and the sky. Concretely, the images were shifted by some pixels for compensation of the vibration. 3.5. Correspondence point search In general, the same target on each camera image is called corresponding point. In this research, clear peaks were selected in each camera, and centers of gravity of peak area of the glacier were used as corresponding points. The glacier melting was seen in this observation as shown in Fig. 4. However, the peak melting is small for


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

one day. As a result, when we measured the glacier movement, it is thought that it is appropriate to chase the peak of the glacier.

N° 4

2009

(a)

134 [h] start

(b)

Fig. 3. Target image. 134 [h]

(a) start

(b)

Fig. 5. Glacier movement measured with a high-precision 3-D multi-camera system; (a) target 1; (b) target 2 .

4. Conclusions

Fig. 4. Shape change of target (peak of glacier);(a) first observation; (b) after about one week. 3.6. Results of experiment 2 Figure 5 shows the measurement results of each target movement. The time interval of plots is about half day and about one day in this figure. Measurement errors are smaller than the size of each plot. 3.7. Discussions of Experiment 2 Results of Experiment 2 show that each targets moved almost same directions. These results are similar to both the previous observation results last year and other observation methods. Furthermore, the observed average movement speed of the glacier was 1.6 m/day] in this year. This speed is corresponding to the report which the moving speed of Perito Moreno glacier is 1-2 m/day. Thus, it is thought that our results are valid in general. In this figure, it appears that there is slightly change in a peak movement and the differences in peaks. The flow velocity distribution on the surface of the glacier is thought to generate such a little difference because it decides by various elements about the hydraulic pressure, thickness of ice, and the surface tilt, the temperature of ice.

In this research, it was proven that there is accuracy enough to measure the glacier by Experiment 1. Experiment 2 can show that the unique results of glacier peak movement and our high-precise 3D measurement system can work well at the natural environments. In future, we have to construct the counting system that doesn't receive the wind effect, and securing a more suitable camera installation location in the lacier observation. Finally, the glacier observation at Perito Moreno should be continued and expanded to measure the global and dynamic motion of this glacier. Valuable observation data at Perito Moreno glacier will be expected to contribute the global warming and other science theme. ACKNOWLEDGMENTS This research was partially supported by the Ministry of Education, Science, Sports and Culture, Grant-in-Aid for Scientific Research (C), 18560411, 2006.

AUTHORS Takeshi Hashimoto* - Department of Electrical & Electronics Engineering, Shizuoka University Address: 5-1, 3chome Johoku, Naka-ku, Hamamatsu, 432-8561 Japan. E-mail: tethash@ipc.shizuoka.ac.jp. Mitsuo Kaneko, Shinji Isono, Tomoyuki Sone, Kazushige Baba, Akira Fukuda - Shizuoka University, Japan. András Rövid - Budapest Tech, Hungary. Masamu Aniya - University of Tsukuba, Japan. Nozomu Naito - Hiroshima Institute of Technology, Japan. Hiroyuki Enomoto - Kitami Institute of Technology, Japan. Articles

27


Journal of Automation, Mobile Robotics & Intelligent Systems

Pedro Skvarca - Instituto Antartico Argentino. * Corresponding author

References [1] Seiji Iguchi, Kousuke Satou, The three-dimensional image measurement, Shokodo CO., LTD, 1990. [2] Keiji Taniguchi, The image processing engineering basic course, Kyoritsu Shuppan CO., LTD, 1996. [3] Akira Fukuda, Keita Ikuta, Kaiji Mukumoto, Shuhei Takahashi, Hiroyuki Enomoto, “Study of Flow of Glacier Using GPS Data Logger”. In: Proceedings of the Society Conference of IEICE, vol. 1 (20050907), 2005, p. 294, [4] Hiroyuki Enomoto, Ryouhei Suzuki, Isenko Evgeni, Akira Fukuda, Renji Fukuda, Pedro Skvarca, Hemam Sala, “Glacier Dynamics of Perito Moreno Glacier, Southern Patagonia Icefield”, Japanese Society of Snow And Ice, 2005, pp. 1-40. [5] Takeshi Hashimoto, Koji Kurosu, “A Proposal of Long Distance and Precise Position Measurement Method”, inter-academia 2005, vol. 1, pp.159164, 2005. [6] Takahiro Uesugi, Takeshi Hashimoto, Akira Fukuda, Hiroyuki Enomoto, Pedro Skvarca, “Highprecision observation of Perito Moreno glacier by stereo camera system”, Inter-academia 2007. [7] Martin Stuefer, Helmut Rott, Pedro Skvarca, “Glacier Perito Moreno, Patagonia: climate sensitivities and glacier characteristics preceding the 2003/04 and 2005/06 damming events”, Journal of Glaciology, vol. 53, no. 180, 2007. [8] Sho Matsumoto, Takeshi Hashimoto, Akira Fukuda, Masamu Aniya, Nozomu Naito, Pedro Skvarca, “High-precision observation of Perito Moreno glacier at two observation points by stereo camera system”, Inter-academia 2008.

28

Articles

VOLUME 3,

N° 4

2009


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

NEW OPTICAL EQUIPMENT IN 3D SURFACE MEASURING

Klara Wenzel, Ákos Antal, József Molnar, Bertalan Tóth, Péter Tamas

Abstract: Referring to a Hungarian Government supported healthcare project „Intelligent tool and methodology to monitor and safeguard childhood spine deformities” (SPINE GUARD) the Department of Mechatronics, Optics and Information Engineering at Budapest University of Technology and Economics developed a new complex optical measuring equipment built up a projector and a digital camera. If the computer controlled projector projects an equidistant grid of lines then the camera takes a picture from a different position about the object to be measured and with a special software removing the grid lighted background there will be a virtual Moiré effect. After some picture processing the 3D (2,5D) reconstruction is finished. In an other working way the projector lights a rainbow on the object. The edges of the rainbow lines cut the crosssection of objects. An other special software reconstructs the 3D (2,5D) objects from the camera captured cross-section curves. Presentation is about the principle of measuring and evaluation of results and precision.

structure is overlaid with another structure and the line elements are nearly superimposed. Stereo vision is similar the human vision, taking pictures from different position upon the same picture points the 3D position of points are countable. If we use another aspect of stereo vision, where a projector lights of the body taking a picture the 3D reconstruction is possible by Moiré method as well as by scanner techniques.

2. Measuring Equipment We have built an equipment capable of measurement using Moiré method and scanning method. The equipment is a computer controlled projector connected with a camera (Figure 1).

Keywords: 3D measuring, Moiré method, 3D reconstruction of human body.

1. Introduction Surface reconstruction is one of the most important topics in computer vision due to its wide field of application. Some examples of applications are range sensoring, industrial inspection of manufactured parts, object recognition and 3D map building. There are several different techniques that can be used for optical three-dimensional measurements on object surfaces, such as interferometric, stereovision, coded structured light and moiré methods. These are based on both contact and non-contact procedures and present different sensitivities. Interferometry is an old technique to measure the deviation between two wave fields with a sensitivity of a fraction of the wavelength of the illumination source. Holography is an interference-based technique that represents a wave front reconstruction a process by which the amplitude and phase variation across a wave front may be recorded and subsequently reproduced. Stereovision is based on imaging the scene from two or more points of view, and then finding correspondences between the different images in order to triangulate the 3D position. Coded structured light consists of replacing one of the two cameras by a device that projects a light pattern onto the measuring surface [1]. Moiré methods are based on Moiré effect that occurs wherever a repetitive

Fig. 1. Measuring equipment.

3. Moiré Theory The Moiré phenomenon can be readily observed when superimpose two periodic or quasiperiodic structures. When the two structures have the same or slightly different line spacing and their lines are set approximately parallel, a new coarse pattern appears. This pattern is known as a Moiré Fringe pattern. The spacing and orientation of the Moiré fringes depend on the spacing and orientation of the structures being overlapped whereas the visibility of fringes is related to the width of transparent or black lines with respect to the line spacing of the structures [2]. In Figure 2, Moiré pattern caused by two straight-line gratings with different frequencies tilted with respect to one another is shown. 3.1. Moiré phenomenon appearance Superposition of periodic and/or quasiperiodic patterns in optics frequently results in striking spatial configurations commonly called Moiré patterns. The spatial frequency of these new periodicities may be considerably lower than the original ones. Therefore, they become proArticles

29


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

nounced at low contrasts. Composite patterns can be formed in different ways. For example, addition, subtraction, and multiplication (three of the four basic rules of arithmetic) are easy to display by optical means. Pattern combinations according to any of these can be made, with corresponding composite configurations having different appearances [3]. Multiplicative superimposition of two structures is the most common method for generating moiré patterns. 3.2. Moiré topographical methods Moiré topographical methods can be distinguished as: the basic grating-shadow, the grating-projection, the grating-TV and the synthetic grating methods. Shadow Moiré - contour mapping technique that involves positioning a grating close to an object and observing its shadow on the object through the grating. Thus, the basic grating-shadow method offers the best accuracy and the simplest arrangement because the projected grating and the master grating are identical: they have the highest degree of binding. Shadow Moiré technique presents the disadvantage that the master grating has a similar size to the measured object. Projection Moiré is a contour mapping technique that involves projection of a grating onto an object to produce a shadow grating that is observed through another grating. The projection-type methods offer a lower degree of binding between the phenomenon and the observing grating, larger object size, and more flexibility in adjusting the sensitivity, but there are very rigid demands for the performance of the projection and the master grating. All methods were the master grating is generated by an electronic time varying signal or by a computational process offer the lowest degree of binding. This means complete independence of both gratings in amplitude and phase. The advantages are additional operations like detection, different types of superposition, and elevation detection. Their disadvantage is the limited accuracy of all optoelectronic devices [4], [5]. To cope with the above requirements, projection moiré has been chosen as the measuring method to be enhanced.

N° 4

2009

Figure 3 shows the Moiré picture of a body.

Fig. 3. Processed Moiré picture.

4. Rainbow Measuring by the Equipment Projector lights the body with differently colored light-beams. Colored beams draw the surface curves onto human body. Picture taken from a different position (Figure 4) is proper to reconstruct geometry the back surface of body.

Fig. 4. Rainbow lighted person. If we want to reconstruct the surface we have to calibrate the picture in 3D and calculate 3D coordinates the indentified surface points.

Fig. 2. Moiré pattern caused by two straight-line gratings with different frequencies tilted with respect to one another.

30

Articles

4.1. Calibration Process As we want to achieve necessary accuracy, we have to to calibrate the cross-sections in the photo, in order to develop measuring methods, as well as to analyze errors.


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

Colored light-beams light a planar curve in every position of the frame. Points of curve are defined by processing of pictures. For 3D scanning the plane to plane perspective transformation is bijection. Perspective transformation by homogenous coordinates is a linear transformation [8] projects quadrangle to quadrangle. The matrix of transformation (1) has eight independent coordinates.

N° 4

2009

We are able to filter different colors of the rainbow from the picture (Figure 7). Upon the filtered colors and calibrating rectangles 3D coordinates of the surface points are defined. The surface curves of the person are defined by polynomial regression. [7] Surface curve interpolated by NURBS surface patches (Figure 7). [8]

(1) Corners of a rectangular calibration element are appropriate to define of matrix coordinates (Figure 5). Corners of calibration equipment are ( ), and corners of its picture are ( ) (i=0, 1, 2, 3) then the transformation is shown in (2)

Fig. 7. Colored surface points and reconstructed surface.

5. Conclusion Two methods and an integrated equipment has developed for optical measurement and reconstruction of 3D surfaces. The equipment and methods have been certified in practice, too. Fig. 5. Planar perspective projection. ACKNOWLEDGMENTS

(2)

The authors would like to thank NKTH (Research and Technology Office of Hungarian Government) for the support since this study has been carried out as part of the project TECH_08-A1/22008-0121 (NKTH).

There are eight unknown coordinates and eight equations (3) there [6].

i=0, 1, 2, 3

(3)

Determination of corner coordinates can be defined from calibration photo (Figure 6).

The research was also supported by HUNOROB project (HU0045), a grant from Iceland, Liechtenstein and Norway through the EEA Financial Mechanism and the Hungarian National Development Agency.

AUTHORS Klára Wenzel*, Péter Tamás, József Molnár, Ákos Antal - Department Mechatronics Optics and Information Engineering, Mechanical Engineering Faculty, Budapest University of Technology and Economics, Budapest 1111 Mûegyetem rkp. 3-9. Hungary. E-mail: info@mogi.bme.hu. Bertalan Tóth - Development Engineer Knorr-Bremse Fékrendszerek Kft., Budapest, 1119 Major u. 69. Hungary. E-mail: Bertalan.Toth@Knorr-Bremse.com * Corresponding author

Fig. 6. Calibration.

Articles

31


Journal of Automation, Mobile Robotics & Intelligent Systems

References [1]

[2]

[3] [4]

[5]

[6]

[7]

[8]

32

Salvi J., Pagès J., Batlle J., “Pattern codification strategies in structured light systems”. Pattern Recognition, vol. 37, no . 4, 2004, pp. 827-849. Patorski K., Kujawinska M., “Handbook of the moiré fringe technique”, Elsevier Science Publishers, New York, 1993. Bryngdahl O., “Characteristics of superposed patterns in optics”, JOSA 66, 1976, pp. 87-94. Windischbauer G., “Survey on Applications of MoiréTechniques in Medicine and Biology”. In: G. von Bally: Optics in Biomedical Sciences, New York, SpringerVerlag, 1982, pp. 244-249. D'Acquisto L., Fratini L., Siddiolo A.M., “A modified moiré technique for three-dimensional surface topography”. Meas. Sci. Technol, 2003, vol. 13, pp. 613-622. Kim D.K., Jang B.T., Hwang C.J., ”A Palanar Perspective Image Matching using Point Correspondesand Rectangle-to-Quadrilateral Mapping”. Fifth IEE Southwest Symposium on Image Analysis and Interpretation, 2002. Tamás, P., Halász, M., Gräff, J., “3D Dress Design“. AUTEX World Textile Conference, Portorož, Slovenia, 2005, pp. 436-440. Szirmay-Kalos L., Antal G., Csonka F., “Computer Graphics”. Computer Books Budapest, 2003.

Articles

VOLUME 3,

N° 4

2009


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

DYNAMICS OF LASER-INDUCED MELTING AND MODIFICATION OF THE SURFACE OF SEMICONDUCTORS BY NANOSECOND LASER PULSES Volodymyr Gnatyuk, Toru Aoki, Oleksandr Vlasenko, Olena Gorodnychenko

Abstract: In situ detection and dynamics of laser-induced melting in different semiconductor crystals (CdTe, CdHgTe, GaAs, InSb and SiC) were performed by the time-resolved reflectivity (TRR). The samples were subjected to irradiation with 20 ns pulses of KrF excimer or ruby laser with energy density varied in a wide range. The surface morphology of the crystals was monitored using optical microscopy and time dependences of the temperature of the crystal surface as a function of laser pulse energy density was also calculated. The melting and ablation threshold values were determined and specific features of the laser-induced phase transitions in the surface region of the semiconductors were analyzed. Keywords: semiconductors, laser irradiation, time-resolved reflectivity, melting threshold.

2. Experiment 2.1. Experimental procedures and measurement details The subjects for investigation were the semiconductors which have been widely used as materials for radiation sensors: CdTe, CdHgTe, InSb, GaAs and SiC. The samples were grown by different methods and were subjected to various preliminary surface treatments [4]-[9]. The pulsed radiation source was KrF excimer (l= 248 nm) or a ruby (l= 694 nm) laser. Single pulses of 20 ns duration (FHWHM) were used. The energy density J of incident laser pulses was varied in a wide range. The in situ detection and dynamics of phase transitions in the surface region of the semiconductors were monitored by the TRR technique using a CW red (l= 633 nm) or green (l= 532 nm) laser as a probe beam.

1. Introduction Laser techniques have been advantageously used to monitor different processes in semiconductors and to modify their properties, including the following procedures: surface cleaning, recrystallization and surface region reconstruction, local annealing and doping, formation of interfaces, etc. [1]-[7]. However, some problems on interaction of short laser pulses with semiconductors are still open. It is particularly concerned with the dynamics of laser-induced melting and determination of the threshold values of energy density at which phase and threshold processes start in the surface region of crystals. In order to identify the laser-induced melting threshold of different semiconductors and study the dynamics of laser-induced phase transitions in the surface region of crystals subjected to action of nanosecond pulses of KrF excimer or ruby lasers, we used the time-resolved reflectivity (TTR) technique. This method consists in direct detection and in situ observation of the time evolution of optical reflectivity of the surface region of a crystal illuminated with a CW probe laser beam under irradiation with a nanosecond laser pulse [7]-[14]. In addition, the morphology of crystals was monitored and time dependences of the surface temperature of crystals were calculated at different laser pulse energy densities. The data on the threshold values of energy density of nanosecond laser pulses have principal importance in application of laser procedures in semiconductor surface processing.

Fig. 1. Diagram of the experimental setup of the time-resolved reflectivity measurements. The TRR experimental setup is shown in Fig. 1. The reflected probe laser beam was detected by a high-speed photodetector and after amplifying the signal was traced by a storage oscilloscope.

3. Results and discussion 3.1. TRR measurements As seen from the oscillograms showing the dynamics of the reflection coefficient R of a red probe laser beam from the CdTe crystal surface under irradiation with KrF excimer laser pulses of different energy densities (Fig. 2, curves 1-3), irradiation with J Âł Jm ~ 50 mJ/cm2 results in an increase in R with following decaying with time. The maximum of R(t) is independent of energy density that is evidence of formation of a laser-melted surface layer with higher reflectivity. The contribution of the long-time component to the total relaxation of R(t) increases with Articles

33


Journal of Automation, Mobile Robotics & Intelligent Systems

rising J (Fig. 2, curves 1-3). This can be attributed to melting and following crystallization of a deeper surface layer when laser pulse energy density increases.

VOLUME 3,

N° 4

2009

Pulsed laser processing of CdTe crystals with J = Jm resulted in improvement of surface smoothness [8]. Irradiation of CdTe crystals with KrF laser pulses J > Ja ~ 150 mJ/cm2 modified the dynamics of TRR. The reflection coefficient R(t) increases and falls to a value lower than initial one (Fig. 3, curve 4). This is due to an increase in roughness of the surface because of boiling, evaporation and ablation of an overheated surface layer and following solidification. Thus, the value Jm » 50 mJ/cm2 and Ja » 150 mJ/cm2 can be considered as the melting and ablation thresholds, respectively.

Fig. 2. Dynamics of the reflectivity of the CdTe crystal surface illuminated with a red probe laser beam under irradiation with KrF excimer laser pulses of different energy densities J. The relaxation of R(t) occurs during the time much longer than laser pulse action time (Fig. 2, curves 2-4). This is associated to deterioration of the crystallinity, amorphization and following crystallization of the surface layer [14]. The irradiation of CdTe surface causes an increase in the steady-state reflection coefficient and this is a reason that R(t) does not decay to the initial value R0 (Fig. 2, curve 4). One of difficulties in the TRR measurements was that the absorption depth of a red probe beam was ~1.6 μm however the absorption depth of KrF excimer laser radiation was very shallow (~10 nm) and hence the melted surface layer was very thin. It was reasonable to use a probe laser with a shorter wavelength. Fig. 3 shows the dynamics of TRR of CdTe crystals under irradiation with a KrF laser pulse using a CW green probe laser. The changes in R were about 6-7% in comparison with 3-4% as in the case of a red probe laser. There was no any change in the reflectivity at J < Jm (curve 1) and a sharp increase in R(t) under pulsed laser irradiation with J > Jm indicated melting of a thin surface layer (curves 2 -4).

Fig. 3. Dynamics of the reflectivity of the CdTe crystal surface illuminated with a green probe laser beam under irradiation with KrF excimer laser pulses of different energy densities J. 34

Articles

Fig. 4. Dynamics of the reflectivity of the SiC crystal surface illuminated with a green probe laser beam under irradiation with KrF excimer laser pulses of energy densities J higher than the melting (curve 1) and ablation (curve 2) threshold. The dynamics of the reflectivity of SiC crystals irradiated with KrF laser pulses of energy densities J > Jm and J > Ja is shown in Fig. 4. A jump in R(t) is due to the beginning of melting of the surface, however the decrease and following increase in R up to the second maximum (curve 1) is associated with the interference of the probe beam reflected from the surface and from the moving solid-liquid interface [14]. Curve 2 in Fig. 4 has the similar shape as curve 4 in Fig. 3. Decrease in R(t) with time is associated with ablation of the SiC surface under laser irradiation with J > Ja. The TRR measurements were performed for CdTe, CdHgTe, GaAs, InSb and SiC semiconductors and a ruby laser as a pulsed laser source was also used. The following values of the melting Jm and ablation Ja thresholds have been obtained: — for CdTe and CdxHg1-xTe with x = 0.3-1, Jm » 50 mJ/cm2, Ja » 150 mJ/cm2 (KrF excimer laser pulse) and Jm » 100 mJ/cm2 (ruby laser pulse). In the case of the solid solutions the melting thresholds vary by 10-20% depending on x; — for SiC Jm » 580 mJ/cm2, Ja » 650 mJ/cm2 (KrF excimer laser pulse); — for InSb Jm » 100 mJ/cm2 (KrF excimer laser pulse), Jm » 140 mJ/cm2 (ruby laser pulse); — for GaAs Jm » 400 mJ/cm2 (ruby laser pulse). 3.2. Surface morphology study The surface morphology of semiconductor crystals was monitored by optical and atomic force microscopy [5, 8, 14]. Figs. 5 and 6 show the micrographs of the surface of


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

CdxHg1-xTe (x ~ 0.3) and InSb crystals respectively, irradiated with ruby laser pulses of different energy densities. Irradiation of samples with J > Jm resulted in melting of a thin surface layer, followed by crystallization. The recrystallized material was in the form of local islands filling the whole area of the laser interaction zone when the energy density was increased Figs. 5(b, c) and 6(b, c).

Fig. 5. Micrographs of the surface of CdxHg1-xTe (x ~ 0.3) crystals before (a) and after irradiation with ruby laser pulses of energy density J = 160 mJ/cm2 (b) and J = 180 mJ/cm2 (c).

Fig. 6. Micrographs of the surface of InSb crystals before (a) and after irradiation with ruby laser pulses of energy density J = 150 mJ/cm2 (b) and J = 160 mJ/cm2 (c). The melting thresholds determined from the surface morphology investigations were a little higher than the values obtained from the TRR measurements. However, laser-induced changes in the dynamics of TRR of semiconductors irradiated with nanosecond laser pulses with J > Jm corresponded to the modifications of the morphology and structure of the surface region of the irradiated crystals [4], [5], [8]. 3.3. Calculations of temperature (a)

N째 4

2009

(b)

Fig. 7. Time dependences of the temperature of the surface of the CdxHg1-xTe (x ~ 0.3) crystal under irradiation with KrF excimer (a) and ruby (b) laser pulses with different energy densities. The TRR experimental data were in good agreement with the calculations of the temperature of the semiconductors subjected to pulsed laser irradiation. The simulation was made by solving the time-dependent heat flow equation [9]. As an example, the time distributions of the surface temperature in CdxHg1-xTe crystals under irradiation with KrF excimer (a) and ruby (b) laser pulses are shown in Fig. 7. Curves 2 correspond to the energy densities of KrF excimer (a) and ruby (b) laser pulses when the melting of a surface layer of is attained.

4. Conclusion The TRR is an effective technique to determine the threshold energy densities during laser processing of semiconductors and for in situ study of the dynamics of phase transitions in the surface region of semiconductors subjected to nanosecond laser irradiation.

AUTHORS Volodymyr Gnatyuk* - V.E. Lashkaryov Institute of Semiconductor Physics of National Academy of Sciences of Ukraine, Prosekt Nauky 41, Kyiv 03028, Ukraine, and Research Institute of Electronics, Shizuoka University, 3-5-1 Johoku, Naka-ku, Hamamatsu 432-8011, Japan. E-mail: gnatyuk@lycos.com. Toru Aoki - Research Institute of Electronics, Shizuoka University, 3-5-1 Johoku, Naka-ku, Hamamatsu 4328011, Japan. E-mail: rtaoki@ipc.shizuoka.ac.jp. Oleksandr Vlasenko - V.E. Lashkaryov Institute of Semiconductor Physics of National Academy of Sciences of Ukraine, Prosekt Nauky 41, Kyiv 03028, Ukraine. E-mail: o_vlas@isp.kiev.ua. Olena Gorodnychenko - Taras Shevchenko Kyiv National University, 64 Volodymyrska Str., Kyiv 01033, Ukraine. E-mail: semicondalf2002@rambler.ru. * Corresponding author

Articles

35


Journal of Automation, Mobile Robotics & Intelligent Systems

References [1]

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

[14]

36

Perriere J., Millon E., Fogarassy E., Recent advantages in laser processing of materials, Elsevier: Amsterdam, 2006, p. 472. Berchenko N.N., Yakovyna V.S., Nikiforov Y.N., Izhnin I.I., Kurbanov K.R., “Laser-induced shock waves processing of IIVI solid solutions interface”, J. Alloys Compd., vol. 371, no. 1-2, 2004, pp. 86-88. Medvid' A., Fedorenko L., Korbutjak B., Kryluk S., Yusupov M., Mychko A., “Formation of graded band-gap in CdZnTe by YAG:Nd laser radiation”, Radiat. Meas., vol. 42, no. 4-5, 2007, pp. 701-703. Gnatyuk V.A., Aoki T., Hatanaka Y., Vlasenko O.I., Mozol' P.O., “Application of pulsed laser irradiation in the semiconductor sensor fabrication”, Proceed. 4th Inter. Conf. on Global Research and Education: InterAcademia 2005, vol. 2, 2005, pp. 543-552. Gnatyuk V.A., Gorodnychenko O.S., “Influence of pulsed laser radiation on the morphology and photoelectric properties of InSb crystals”, Semicond., vol. 37, no 4, 2003, pp. 396-398. Gnatyuk V.A., Aoki T., Hatanaka Y., “Laser-induced shock wave stimulated doping of CdTe crystals”, Appl. Phys. Lett., vol. 88, 2006, pp. 242111-3. Aoki T., Gnatyuk V.A., Nakamura A., Tomita Y., Hatanaka Y., Temmyo J., “Study of a CdTe high-energy radiation imaging device fabrication by excimer laser processing”, Phys. Stat. Sol. C, vol. 1, no 4, 2004, pp. 1050-1053. Gnatyuk V.A., Aoki T., Nakanishi Y., Hatanaka Y., “Surface state of CdTe crystals irradiated by KrF excimer laser pulses near the melting threshold”, Surf. Sci., vol. 542, 2003, pp. 142-149. Gnatyuk V.A., Aoki T., Gorodnychenko O.S., Hatanaka Y., “Solid-liquid phase transitions in CdTe crystals under pulsed laser irradiation”, Appl. Phys. Lett., vol. 83, no 18, 2003, pp. 3704-3706. Timoshenko V.Yu., Dittrich Th., Sieber I., Rappich J., Kamenev B.V.. P.K. Kashkarov, “Laser-induced melting of porous silicon”, Phys. Stat. Sol. A, vol. 182, no 1, 2000, pp. 325-330. Ivlev G., Gatskevich E., Chab V., Stuchlik J., Vorlicek V., Kocka J., “Dynamics of the excimer laser annealing of hydrogenated amorphous silicon thin films”, Appl. Phys. Lett., vol. 75, no 4, 1999, pp. 498-450. Gatskevich E., Ivlev G., Prikryl P., Cerny R., Chab V., Cibulka O., “Pulsed laser-induced phase transformations in CdTe single crystals”, Appl. Surf. Sci., vol. 248, no 1-4, 2005, pp. 259-263. Kovalev A.A., Zhvavyi S.P., Zykov G.L., “Dynamics of laser-induced phase transitions in cadmium telluride”, Semicond., vol. 39, no 11, 2005, pp. 1299-1303. Baidullaeva A., Vlasenko A.I., Gatskevich E.I., Gnatyuk V.A., Ivlev G.D., Mozol’ P.E. Nanosystems, Nanomaterials, Nanotechnologies, vol. 6, no 4, 2008, pp. 11671174. (in Russian)

Articles

VOLUME 3,

N° 4

2009


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

RADIUS MEASUREMENT OF CYLINDRICAL SURFACES BASED ON ANALYZING THE INTERFERENCE PATTERN OBTAINED BY SCANNING THE SURFACE WITH FOCUSED LASER BEAM Ryszard Jabłoński, Jerzy Mąkowski

Abstract: A focused laser beam is incident on the edge of cylindrical object. The reflected edge wave is interfering with geometrical wave forming fringe pattern containing the information about the surface local curvature. This can be determined by analyzing the detector output signal.

form fringe pattern containing the information about the local curvature. Having the above in view, the close analysis of detector signal was carried out. The obtained intensity characteristics allow determining the curvature of the surface.

2. Measuring set-up Keywords: radius of curvature, laser diffraction, interference pattern.

1. Introduction In laser-scanning measurement of cylindrical objects extremely complex interfering signals occur. They are due to superimposition of reflected, geometrical and scattered light. The proportions between these components vary in time, and also the total intensity distribution changes. The considerations applying Fraunhofer theory are static and fragmentary, and it may be concluded that the existing solutions for diffraction of 3D bodies do not fit to engineering applications. A rigorous diffraction theory is based on Maxwell's equation and the boundary conditions are associated with the obstacle [1]. The boundary conditions are used to calculate a field scattered by the obstacle. The origins of this scattered field are currents introduced in the obstacle by the incident field. The scattered field is allowed to interfere with the unobstructed field to produce resultant interference pattern. Within the Kirchhoff theory of diffraction there are two possible ways of interpretation the diffraction phenomena. The first one, based on so called “edge wave”, which refers to T. Young idea was later extended by A. Rubinowicz and the second Fresnel diffraction model, is based on the “zones concept”. T. Young interpreted the diffraction phenomena as a result of interference of the geometrical wave propagating in free space with “the edge wave” A. Rubinowicz [2]. Theoretically it proved the possibility of division the Kirchhoff diffraction field into two components. One of them has the character of incident wave and the second represents the wave created by interaction between primary field and the edge of object, and this interaction is of reflection type. These mathematical considerations are commonly accepted in the scientific world, but were never proved experimentally. In the paper the geometrical wave is the undisturbed part of incident wave. The edge wave is the wave generated by the edge of an object (it is the wave reflected by the fragment of cylindrical surface considered to be as the edge of the object). These two waves interfere and

The outlook of the experimental set-up is presented in Fig.1. (it was described in [3]). The auxiliary components like: system assuring parallel laser travel, vibration protection set-up, dark chamber, detector electronics, computer, etc, are not shown in the figure.

Fig. 1. Experimental set-up. The unit (1) contains laser, beam forming optics (beam expander with spatial filter) and scan lens. The transformed laser beam is directed to the object (2) and the generated intensity distribution pattern is measured by detector unit (3) and then processed by computer. Computer controls the step motors (4) and (5) by means of interface in/out LC-012-1612. The motor (4) scans the laser unit and motor (5) scans the detector. Laser and detector are supplied by separate power sources. The experiment was limited to the edge effect – it means only to the area located nearby the shadow border line (where the strong diffraction effect is expected). In order to avoid the inaccuracies caused by the instabilities of angular deflection, the entire laser head is scanned parallel to beam axes (in 2.5m steps). The detector unit (fixed to the rotary table, coaxially with object axis) is composed of photodiode, aperture 0.3 mm and electronic circuit. It was calibrated with the power meter (LaserMate-Q, Coherent) and obtained signal (in V) is proportional to the measured light intensity.

Articles

37


Journal of Automation, Mobile Robotics & Intelligent Systems

3. Position of object edge in relation to the laser beam waist. The determination of measuring area (location of measured object) is essential in all measuring instruments. In reference to the notations used in Fig. 2: the areas A and B represent two theoretical halves of laser beam. C is the shadow area and in D area the interference fringes appear.

VOLUME 3,

N° 4

2009

mum displacement was defined for 1/3 of beam diameter (it corresponds to 50 m for 75 mm distance from the laser head). Fig. 3. presents the exemplary detector signal recorded by scanning the detector from point 1 to point 2. We named the course with points A(n) as “a maximum of the first order interference fringe”. This course is determined by scanning the edge of an object withh laser beam in small steps. Out of the obtained plot of courses (Fig. 4) it is easy to select the one with maximum amplitude. This procedure was described in deteils in [5].

Fig. 2. Geometry of measuring area.

Fig. 4. Detector output signal obtained by scanning the edge of an object with laser beam. “The maximum of the first order interference fringe” is indicated as A.

Fig. 3. An exemplary detector signal recorded by scanning the detector from point 1 to point 2. In the designed system the beam waist is 2w0 = 60 mm and its position in reference to the scanning lens is 65 mm. In experiments f 0.9, 2, 2.5, 3, 4 mm cylinders were used, placed perpendicularly to the laser beam axes in the distance 57-85 mm from the laser head (scanning lens). The measurements were taken in 1mm step starting with 57 mm distance and the measuring quantity was the visibility of interference fringes. In each position the laser beam was screened by the edge of an object and the fragment of resulting interference pattern was recorded by scanning the detector from point 1 to point 2 (see notations in Fig. 2). The results were discussed in [4]. Depending on the position of an object the interference pattern changes significantly. Fig. 3 shows the exemplary plots of measurement data recorded by scanning the detector from point 1 to point 2. The course with points A(n) (reference course) represents the measurement data taken for such position of laser beam where the beam axes is tangent to the object surface. The course with points B(n) optimum course, presents the data taken for beam axes being parallel displaced of about 1/3 of a beam size from the previous position. That choice is justified by the following reasoning: with the transverse parallel displacement of laser beam the difference B1-A0 is increasing (what is considered as advantage), but the further displacement results are in gradual decreasing (and even disappearing) of the 3rd– order maximum (what is obvious disadvantage); the opti38

Articles

Fig. 5. The courses A(n) and B(n) for cylindrical object of radius 0.45 mm.

Fig 6. The courses A(n) and B(n) for cylindrical object of radius 2.0 mm. At the distances below 65 mm the diffraction effects are visible on the left side of the graphs. The most


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

b) The phenomenon can be analyzed with different wavelengths. c) The incident wave has plane wave front of unlimited widths. d) The cylinder surface was treated as unlimited set of light-reflecting sharp edges. e) Each elementary ray reflects from consecutive sharp edge. f) The reflected ray interferes with geometrical ray, Fig.7 (this is exactly the case discussed by Rubinowicz [2]). g) The results are presented graphically. h) Phase shift between geometrical wave and reflected wave was assumed only on the basis of path difference of these waves. The phase shift due to the reflection was not taken into account.

uncertain signal was detected at the positions close to the beam waist (it did not show any interference). When the distance 65 mm was exceeded the stronger and more distinct pattern appeared. The graph for 75 mm distance contains most information. There are clearly visible 6 interference fringes and also visibility of fringes is high. Taking this object position as optimum, the series of measurement were performed with cylindrical objects of radius 0.45, 1.0, 1.25, 1.5 and 2.0 mm. The two exemplary results, for radius 0.45 mm and 2.0 mm, are shown in Figs. 5 and 6. Table 1 shows xy coordinates of points A(0), B(0), A(1), B(1). We analyzed the results trying to define the peculiarities related to curvature (such as: number of fringes, fringes gradient, attenuation coefficient) and finally we selected the parameter called “interference signal gain”. This parameter is given by relationship:

The obtained simulation results are presented in Fig. 8 and can be summarized as follows:

Si = Y(0A) – Y(1B) where: Y(0A) – is the amplitude of detector signal level corresponding to the maximum of the first interference fringe in basic course. Y(1B) – is the amplitude of detector signal corresponding to the maximum of the first interference fringe in optimum course. The performed experiments prove the possibility of noncontact radius measurement of cylindrical objects with the use of interferometric phenomena. The significant increase of interference signal, parameterized by Si, enables to measure the radius (curvature) with the resolution 10 μm. The interference maximum appears at the maximum of reference beam what enables the accurate determination of an edge of an object.

Fig. 7. Geometry of simulation experiment.

1. The envelope of obtained signal (low frequency) is in good agreement with experimental results. 2. Each cycle in high frequency signal corresponds to one wavelength path difference of interfering waves. This path difference is bigger for larger diameter of object. 3. The presented results expand the Rubinowicz's considerations on the origin of edge wave [2].

4. Simulation Due to the very compound form of the above interference/diffraction phenomena, we decided to explain these effects by simulation. In simulation we took the following assumptions: a) The investigated objects are the cylinders up to 5 mm diameter.

Table 1. xy coordinates of points A(0), B(0), A(1), B(1) and parameter Si. Radius

Course

X(0)

Y(0)

X(1)

Y(1)

X(0A)-X(1B)

Y(0A)-Y(1B)

0,45

A B A B A B A B A B

16100 11050 14650 9400 16250 10750 17100 11650 15200 9800

5,53 5,77 5,59 5,78 5,69 5,62 5,84 5,28 5,94 4,82

10850 5850 9150 3300 10650 4650 10500 5550 9,450 3750

4,79 3,38 4,77 3,14 4,74 3,00 4,48 3,04 4,02 2,04

10250

2,16

11350

2,45

11600

2,69

11550

2,80

11450

3,90

1,0 1,25 1,5 2,0

Articles

39


Journal of Automation, Mobile Robotics & Intelligent Systems

Fig. 8. Low and high frequency signal for cylinders of radius 0.5mm and 2.5 mm.

5. Conclusion 1. Diffraction angle on the edge depends on the local radius. 2. The radius of an edge can be determined on base of analyzing the interference pattern. 3. The phase shift of the edge wave in relation to the geometrical wave has two components: optical path difference and phase shift on reflection. 4. The developed system enables the measurement of phase shift on reflection for incident angle close to 90 deg.

AUTHORS Ryszard Jabłonski*, Jerzy Makowski – Faculty of Mechatronics, Warsaw University of Technology, Św. A. Boboli Str. 8, 02-525 Warsaw, Poland. E-mail: yabu@mchtr.pw.edu.pl * Corresponding author

References [1] [2]

[3]

[4]

[5]

40

Born M., Wolf E., Principles of optics, Pergamon Press, Oxford, 1980. Rubinowicz A., Die Beugungswelle in der Kirchhof-fschen Theorie der Beugung, Springer Verlag &PWN, Warsaw 1966. (in German) Jablonski R., Makowski J., “New approach to fringe pattern analysis obtained by scanning polished metal cylinders with Gaussian beam” Key Engineering Materials, Measurement Technology and Intelligent Instruments VIII, Trans Tech Publications 2008, Zurich, pp. 267-270. Jablonski R., Makowski J., “Metrological aspects of laser scanning system for measurement of cylindrical objects”, Mechatronics, 2009, Springer Verlag (in press). Jablonski R., J. Makowski J., “Comparison of fringe pattern obtained by scanning polished metal cylinders and sharp-edged objects with laser beam”. In: Proc. Int. th th Conf. ASPE/ICPE, 19 -24 Oct. 2008, Portland Oregon USA, pp. 352-256.

Articles

VOLUME 3,

N° 4

2009


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

ISPACE – A TOOL FOR IMPROVING THE QUALITY OF LIFE Annamária R. Várkonyi-Kóczy, András A. Tóth

Abstract: Intelligent Space or iSpace is a new kind of computing system aiming at improving the environments of humans, creating a natural and easy to use solution. Its main feature is that the intelligence is not implemented separately in the actors, but it is distributed in the whole space. In this paper, we summarize the concepts of iSpace and some of those methods which can advantageously be used for developing a new man-machine interface in iSpace applications. Keywords: Ubiquitous computing, intelligent Space, Image Processing, Intuitive User Interface

1. Introduction A new paradigm in user-machine interaction is the so called “Ubiquitous Computing” approach introduced in [1]. A Ubiquitous Computing system consists of more devices communicating with each other, thus it is a distributed system. As their name suggests, Ubiquitous Computing systems are present everywhere in an environment such as an office or a home and their task is to enhance that environment. The main goal of Ubiquitous Computing systems, however, is to offer such an interface to the user that should be so natural and easy to interact with, that users become unaware of the fact that they are using a computing system. An implementation of the Ubiquitous Computing paradigm is the Intelligent Space (iSpace) [2], which is an area, such as a room, equipped with intelligent sensors and agents. The main feature of the iSpace is that the intelligence itself is not present in the agents but it is distributed in the whole space. Thus, the architecture of the artificial agents, such as robots, is quite simple as they are coordinated by the intelligent sensors. Another important feature of the Intelligent Space is its capability of observing what is happening in it and to build models of the environment. In case of necessity the iSpace is also capable to interact with the environment in order to achieve some kind of change or give information to its users. In this paper, authors discuss the main features of iSpace and also summarize methods which will be used in developing a new human-machine interface which is intuitive and easy to use. By this, users become able to issue orders to the iSpace assuming simple gestures of their hands and/or issuing movements with them. The paper is organized as follows: Section 0 describes the most important features and the architecture of the iSpace. In Section III we give an overview of the methods

that are used in the interface. Finally, Section IV concludes the article.

2. The Intelligent Space A. Purposes and features of the iSpace Intelligent Space (iSpace) is an intelligent environmental system originally developed at Hashimoto Lab, University of Tokyo, Japan. The main and ultimate goal of the Intelligent Space is to build an environment that comprehends human interaction and satisfies them [3]. This means that the system should be easy to use for the people in it: they should be able to express their will through intuitive actions and there should be no need for them to learn how the system is to be used. Another important requirement is that the system should be human centered, that is it should not be disturbing for people: for instance there should be no need of portable devices in order to interact with the space, and the installation of intelligent devices into an existing area should not alter that area overly. Beyond supporting humans, the iSpace should provide an interface also to its artificial agents, e.g. to the robots providing physical services to the humans using the iSpace. The most characteristic feature of the iSpace is that the intelligence is distributed in the whole space, not in the individual agents. Thus, there is little logic in the robots which are controlled by the iSpace. The other fundamental capability of the iSpace is that it is able to monitor what is going on in the space and to build models based on this knowledge. The system is also capable to react with its environment and provide information or physical services to its users. The physical services are provided by the robots. The iSpace is an active information space: that means that the clients are able to request information from the system which is provided to them by the active devices. The iSpace is thus a so called soft environment: it has the capability to adapt itself to its clients [2]. There are several applications currently developed or planned for the iSpace. These include the positioning and tracking of humans [3], the localization of mobile robots [3], the control of robots [3], and finding paths for them by using itineraries taken by people [4]. B. Architecture of the iSpace There are several requirements that the hardware and software architecture of the iSpace has to satisfy. As stated in [4] these are modularity, scalability, ease of integration, low cost, and ease of configuration and maintenance. Articles

41


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

Modularity is important in order to ensure the possibility of run-time reconfiguration when a component is added or removed from the system. Scalability is also relevant, because the iSpace must be able to adapt itself to environments of various sizes and shapes; furthermore it should be possible to integrate local systems into larger ones. Ease of integration means that it should be easy to integrate existing intelligent components into the system. Low cost is to be understood on the components: it is important, because the iSpace is built of many smaller components, thus the economic factor is determined by the cost of the components. Finally, it is expected that it should be easy to set up and maintain an existing system. Thus, the iSpace should also have the capability to learn by itself. Software tasks are implemented as distributed individual processes and can be categorized into three types: sensor and actuator servers, intermediate processing, and application processes. The task of the sensor and actuator servers is the preprocessing of the information and the delivery of the preprocessed information to the network. Sensor fusion, temporal integration, and model building occurs at intermediate processing level. The tasks performed at this level might require real-time capabilities, thus components performing them should be located near the sensor. Finally, the application processes are those that perform the actual applications of the iSpace. At this level only low amount of data is processed and a slower reaction time is sufficient. Functions at this level are required to be easily portable and maintainable by the user. The architecture satisfying all these requirements is described in [3]. The basic elements of the iSpace are the artificial clients and the distributed intelligent sensors. The formers include robots, which provide the physical services to the human users of the iSpace, and the monitors, which furnish them information.

N° 4

2009

These parts are shown in Fig. 1. The advantage of the architecture based on the DIND's is that the iSpace can easily be constructed and modified by using them. Indeed, an existing space can be turned into an intelligent space just by installing DIND's. The modularity of the architecture makes the renewal of functions easy, and thanks to the network an update can be effectively applied to all DIND's. The other advantage of the network is that it facilitates resource sharing.

3. Overview of the methods helping in developing iSpace by a new intelligent interface A. Camera Calibration Cameras taking input pictures can be described by means of so called extrinsic and intrinsic parameters. The former mean the position and orientation of the camera in space. The latter are those parameters that are independent of locations, i.e. focal length, distortion, and skew. The transformation for the so called pin-hole camera model can be described in terms of these parameters: x = P × X = K × R × [I½-C] × X

(1)

where x denotes the image location of the point, X stands for the spatial location, and P is the camera matrix. The matrices in the product yielding the camera matrix are the camera calibration matrix (K), the rotation matrix between the camera coordinate frame and the world coordinate frame (R), the identity matrix (I), and the center of the camera coordinate frame (C). The camera calibration matrix consists of the following values:

(2) where (px, py) represent the image center, ax and ay stand for the horizontal and vertical size of pixels, and s denotes the skew. For a more elaborate description of camera parameters see e.g. [5]. The process of estimating these parameters is commonly referred as camera calibration in literature. There are more known approaches for calculating these parameters. One possible way is based on the correspondence between spatial- and picture points. Another possible solution commonly adopted relies only on picture points. In this latter case, however, the points have to satisfy some other constraint, for instance they have to lie on the same plane in a predefined pattern. Fig. 1. Structure of a DIND. DIND (Distributed Intelligent Networked Device) is the term for the intelligent sensor, which is the basic building element of the iSpace. This device consists of three main components: a sensor which monitors the dynamic environment, a processor, whose task is to deal with sensed data and to make decision, and a communication part through which the device is able to communicate with other DIND's or with the artificial agents of the iSpace. 42

Articles

B. Histogram Based Skin Detection Image data can be stored using various parameters. The well-known RGB color space uses red, green, and blue color components. The components of the HSV (Hue, Saturation, Value) color space in contrast uses quantities that are more related to the every day color description. Hue is the parameter we would define as “the color”, saturation means the concentration, and value is the measure of brightness. As stated in [6] this color space is less


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

sensitive to lighting changes than RGB, and the hue value is invariant for the various skin "colors". In fact, it is only the saturation that increases for dark skinned people. The histogram back projection procedure [6] locates areas with a given color model. Let's consider HSV: backproj(x, y) = H(hue(x, y))

N° 4

2009

dure usually produces more candidate points in the intervals containing real peaks and valleys, for each interval one real peak/valley has to be chosen, discarding the others.

(3)

where backproj means the single channel backprojection image, H is the hue-histogram, and hue is the single channel hue plane of the input image. This means that given the color model (i. e. the histogram) of a given object (in this case human skin), for each pixel on the image the value of the corresponding histogram bin is written back to the image. Thus the probability distribution of skin is obtained. C. Feature point extraction A common procedure in literature adopted to extract feature points is corner detection. Some known methods are those of Harris [7], Förstner [8], He and Yung [9], the SUSAN Corner Detector [10], and the Fuzzy Corner Detector [11]. Another approach, which considers curvature extrema, is the extraction of peaks and valleys as described in [12]. This latter algorithm works in the following way: First, the contour points of the blobs are extracted in counter clockwise order. After this, the k-curvature (see e.g. [12]) for each contour point is computed as follows:

(4) where C[i] denotes the picture coordinates of the ith contour point. Thus the k-curvature is defined as the angle between the vectors (C[i], C[i-k]) and (C[i], C[i+k]). Contour points whose k-curvature is less than a given threshold are classified as peak/valley candidates. For these points the rotation direction of the above mentioned vectors is also computed based on the sign of the cross product of the two vectors, thus the convexity of the corners can be determined.

Fig. 3. Weighting fuzzy functions of the environments. D. Feature point matching Feature points in a stereo image pair can reliably be matched using the epipolar constraint x'T × F × x = 0

(6)

where x' and x stand for the corresponding points in the right and left image, respectively and F denotes the fundamental matrix, which is a matrix mapping points in one of the stereo images unambiguously into a line in the other stereo image, and a similarity measure of the neighborhoods of the feature points (see [5] and [13]). The algorithm described in [13] is a fuzzy logic based matching algorithm, which works in the following way: first the candidate pairs (i.e. points lying within a given fuzzy neighborhood of their corresponding epipolar lines) are located. Then for each candidate point pair detected in the stereo image pair, the sum of differences of their neighborhoods weighted by a two dimensional fuzzy set is calculated: (7)

where I(x,y) and I'(x',y') mean the intensity in the left and right images, respectively and μa and μb denote the membership functions used as weighting functions. The environments of the feature points are denoted as w and w'. These values are shown in Figs. 2 and 3. The pairing which yielded the smallest sum of weighted difference is considered to be matching. Fig. 2. Environments of matched points. l' denotes the epipolar line corresponding to x. convexity(p)= sgn[(C[i-k] - C[i])´(C[i+k] - C[i])] . (5) convexity(p) determines whether the pth extremum is convex or concave. Feature points at convex extrema of the contour are referred to as peaks whereas those at concave extrema are referred to as valleys. Since this proce-

E. Three Dimensional Reconstruction In order to locate the matched feature points in space, the relative position of the cameras must be known beforehand. This can be achieved by calibrating the cameras and extracting the extrinsic parameters. Then the position of points can be computed using one of the methods described in e.g. [5]. The Direct Linear Transform (DLT) is based on the fact that the projection equation (1) can be transformed in Articles

43


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

the following way: A × X = 0,

(8)

where X stands for the location of the spatial point and A is defined as:

.

(9)

Here (x,y) and (x',y') denote the picture coordinates of the matching feature points, piT and p'iT stand for the ith and jth rows of the left and right camera matrices respectively. Taking the Singular Value Decomposition (SVD) of A yields the coordinates of the spatial point corresponding to the image points: A = U × D × VT

(10)

where U, D and V are the matrices yielded by the SVD. The last column of V represents the coordinates of the spatial point.

N° 4

2009

One iteration of the hybrid tracking algorithm consists of three steps: in the first one the new position of the blob is estimated based on its previous position and its speed. The next step is the measurement of the position achieved by using the mean shift algorithm. The area located by the mean shift procedure is compared to the object model by their Bhattacharyya distance, which is also described in [15]. If the Bhattacharyya distance exceeds a given threshold, the mean shift window is temporarily enlarged and the searching procedure is repeated from the position of the previous frame. This part of the algorithm deals with the case when the tracker loses the tracked object for instance due to a sudden change in the speed or direction of the movement.

4. Coclusions In this paper the concept of iSpace is summarized. Different image processing and computer vision methods are also presented which can serve as a basis for developing a hand gesture and -movement controlled man-machine interface in iSpace. The new interface is natural and comfortable to use, thus can contribute to the improvement of the quality of life of humans in smart environments. ACKNOWLEDGMENTS

F. Blob Tracking There are various known approaches in the literature for implementing the tracking of moving objects on a video sequence. One of them is the Closed World Tracking, proposed in [14]. As stated there, “a closed-world is a region of space and time in which the specific context of what is in the region is assumed to be known”. Here context means a method for representing knowledge in a dynamic multi-object tracking problem. The solution matches blobs in subsequent frames by comparing their size, position, speed and color. For each of these features it calculates the so called match-score matrix which is a matrix containing the distances in feature space of the objects and the blobs in a frame. Then, in order to obtain the best matches, it finds the minima of the weighted sum of the four match-score matrices. Another approach for tracking a moving object is the CAMSHIFT (Continuously Adaptive Mean Shift) algorithm [6] based on the mean shift algorithm which is a gradient based searching procedure that operates on a back projection. It iteratively moves a fixed size search window from a given start position, until movement converges to zero. In each step, the new location of the center of the search window is the mode of the area under the window. The CAMSHIFT algorithm enhances the mean shift algorithm by dealing with the temporal change of the probability distribution. This procedure also iteratively moves the search window like the CAMSHIFT but it also changes its size according to the zeroth moment of the area under the search window. The advantages of the CAMSHIFT procedure are its low computational cost and its robustness with relation to outliers. A third tracking method is the hybrid tracking algorithm proposed by the authors of the iSpace ( see [15]). This procedure also relies on the probability distribution. 44

Articles

This work was sponsored by the Hungarian National Scientific Fund (OTKA K 78576).

AUTHORS Annamária R. Várkonyi-Kóczy* - Institute of Mechatronics and Vehicle Engineering, Budapest Tech, Integrated Intelligent Space Japanese-Hungarian Laboratory. E-mail: koczy@mit.bme.hu. András A. Tóth - Integrated Intelligent Space JapaneseHungarian Laboratory. E-mail: bamota@gmail.com. * Corresponding author

References [1] [2]

[3]

[4]

[5]

[6]

M. Weiser, The Computer for the Twenty-Frst Century, Scientific American, 1991 pp. 94104,. Lee J-H., Hashimoto H., “Intelligent Space”. In: International Conference on Intelligent Robots and Systems 2000 (IROS 2000), vol. 2, 2000, pp. 1358-1363. Lee J-H., Morioka K., Ando N., Hashimoto H., “Cooperation of Distributed Intelligent Sensors in Intelligent Environment”, IEEE/ASME Transactions on Mechatronics, vol. 9, no. 3, 2004. Appenzeller G., Lee J.-H., Hashimoto H, “Building Topological Maps by Looking at People: An Example of Cooperation between Intelligent Spaces and Robots”, Intelligent Robots and Systems, vol. 3, issue 7, 1997, pp. 1326-1333. Várkonyi-Kóczy A.R., A Tóth, ISpace - a Tool for Improving the Quality of Life, Journal of Automation, Mobile Robotics and Intelligent Systems. Special issue for Inter-Academia 2009, vol. 3, no. 4, 2009, Várkonyi-Kóczy A.R., “Autonomous 3D Model Reconstruction and Its Intelligent Application in Vehicle Systh tem Dynamics”. In: 5 International Symposium on Inte-


Journal of Automation, Mobile Robotics & Intelligent Systems

[7]

[8]

[9]

[10]

[11]

[12] [13]

[14]

[15]

VOLUME 3,

N° 4

2009

lli-gent Systems and Informatics (SISY 2007), Subotica, Serbia, 24th-27th August 2007, pp. 13-18. Intille S.S., Davis J.W., Bobick A.F., “Real-Time ClosedWorld Tracking”. In: Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, June 1997, pp. 697-703. Bradski G., Darrell T., Essa I., Malik J., Perona P., Sclaroff S., Tomasi C., et al., Intel OpenCV Library, software available online at: http://sourceforge.net/projects/opencvlibrary Bouguet J-Y., Camera Calibration Toolbox for Matlab, software available online at: http://www.vision.caltech.edu/bouguetj/calib_doc/ Segen J., Kumar S., Shadow Gestures: 3D Hand Pose Estimation Using a Single Camera, IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 1999. Várkonyi-Kóczy A.R., “Fuzzy Logic Supported Corner Detection”, Journal of Intelligent & Fuzzy Systems: Applications in Engineering and Technology, vol. 19, issue 1, 2008 pp. 41-50. Malik S., Real-time Hand Tracking and Finger Tracking for Interaction, CSC2503F. Project Report, 2003. Várkonyi-Kóczy A.R., “Autonomous 3D Model Reconstruction and Its Intelligent Application in Vehicle System Dynamics”. In: 5th International Symposium on Intelligent Systems and Informatics (SISY 2007), Subotica, th th Serbia, 24 -27 August 2007, pp. 13-18. Intille S.S., Davis J.W., Bobick A.F., “Real-Time ClosedWorld Tracking”. In: Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, June 1997, pp. 697-703. Kazuyuki M., Lee J-H., Kuroda Y., Hashimoto H., “Hybrid tracking Based on Color Histogram for Intelligent Space, Artificial Life and Robotics”, vol. 11, no. 2, 2007, pp. 204-210.

Articles

45


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

SIMULATION OF COULOMB-COUPLED, PROTEIN-BASED LOGIC BalĂĄzs Rakos

Abstract: In this theoretical work, I investigate photoswitchable proteins as possible building blocks of computing, and signal processing architectures of the future. Coulomb-coupled, photon-pulse controlled macromolecular arrays are proposed and explored with the aid of a simple, mixed quantum-classical, nanoelectromechanical model. I show chains for digital signal propagation, and possibly universal logic gates, therefore computing architectures can be constructed from such proteins. Furthermore, I explore the basic molecular properties needed for their realization, and I show that photoswitchable proteins are possible candidates for this purpose. Keywords: nanotechnology, nanoelectronics, molecular electronics, photoswitchable protein.

1. Introduction The continuously increasing need for improving computer power results in a steady increase of the number of electronic components in microprocessors (Moore's law). In order to realize fast, low-cost, low power consuming and dissipating nanometre-size electronic devices, new concepts must be explored by taking advantage of effects arising on the nanometre-scale. Since the size of such devices is expected to eventually reach the molecular level, an emphasis must be given on the investigation and implementation of novel computing architectures based on molecules. Since Coulomb forces are the strongest intermolecular forces in molecular arrays, Coulomb-coupling [1] is a possible way for the integration of molecule-based electronic elements. The concept of photon-pulse driven, polymer-like signal processing arrays has been presented previously [2], and their advantages have been discussed in the case of two-state quantum systems with mechanical vibration in one variable [2]-[4]. Since proteins are low-cost macromolecules that can be ordered in self-assembled monolayers, and can be engineered to provide desirable properties, they are possible candidates for the realization of molecular computers. It has been demonstrated in the case of Dronpa that such molecules can be reversibly photo switched between states even at the single-molecule level [5]. In this paper, I extend the model presented in [2] to be suitable for the simulation of dipole-dipole coupled photoswitchable proteins placed next to each other. Furthermore, I simulate digital signal propagation in a onedimensional chain consisting of closely placed proteins (Fig. 1). I show that such photoswitchable proteins have 46

Articles

potential advantages in molecular computing.

protein 1

protein 2

protein 3

...

Fig. 1. Diagram of a chain arrangement of closely placed proteins.

2. Modelling For the simulations I used a simple quantum-classical dynamical model based on [2], where both molecular electron and proton transfers are taken into account. For the sake of simplicity I assumed that the molecule can be described as a two-state system, and I presupposed that the potential energy surfaces (PES) as the function of the reaction coordinate, and the damping parameters are known. The Hamiltonian matrix of the molecule depends on the reaction coordinate q: (1) The equations describing the coupled nanoelectronic and nanomechanic behaviour of the molecules assuming the simplest dissipation model are the following:

(2)

(3)

(4) where is the three-dimensional quantum-coherence-vector, Vnn, Ven are the nucleonnucleon and electron-nucleon potential energies (they can be determined from the PES), respectively. The aforementioned equations were used in [2] for the simulation of two-state quantum systems with one-dimensional nuclear vibration, therefore there the reaction coordinate, q represented the distance between the nuclei (bond length), p corresponded to the momentum, and M described the mass of the nucleus. In this study q is used in a more general sense, since the choice of the reaction coordinate depends on the nature of the reaction (e.g. if


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

q is the bond length then p is the momentum, M is the mass of the atom, if q is chosen to be the bond angle then p is the angular momentum, M is the moment of inertia), which depends on the choice of the photoswitchable protein under investigation. The a parameter characterizes the mechanical, t represents the electronic relaxation. The

(5)

matrix is the Bloch matrix. Since in our case the internal molecular dynamics cannot be described using solely the adiabatic approximation, surface hopping [6] between the two electronic states was incorporated in the model as well. According to Rabi's theory, the external electromagnetic field (photon pulse) interacts with the dipole moment of the molecule adding a time-varying component to the off-diagonals of H in equation (1). Coulombcoupling with the neighbouring molecules changes the difference between the energy levels of the molecule: (6) where DE and DE0 are the differences between the two energy levels with and without the influence of the neighbours, respectively, Dμ is the transition dipole moment of the molecule, μi is the dipole di moment of the i-th neighbour, and di is the distance between the molecule and its i-th neighbour.

N° 4

2009

lation, and the distance between proteins was set to 3 nm. All of these data are within realizable domains. The difference between the two energy levels of the protein depends on the dipole moments of the neighbouring molecules (see equation (6)), and since the dipole moments of the two states of a photoswitchable protein can differ significantly, the states of the neighbours can strongly influence the photon frequency required to switch the corresponding protein, thereby improving the selective nature of the switching, which is a major advantage of such macromolecules. In the case of other kinds of molecules, where only the transition dipole moments between ground and excited states play a role in switching, the coupling effect is much smaller, since the transition dipole moments are usually significantly smaller than the dipole moment difference between the two stable states of photoswitchable proteins. If I assume that in the beginning all of the proteins are in state 1, the process is the following: first the entire chain is subjected to pulse 1 1 with frequency w1 (the subscript shows the state of the molecule, the superscript is '1' or '2', if the neighbour(s) of the protein is in state 1 or state 2, respectively) that switches the first member of the chain to state 2. Then the 12 chain is irradiated by pulse 2 with w1 that switches the 2 second protein to state 2. The last pulse with w2 switches the first molecule back to state 1. The demonstration of the photoswitchable protein majority and Fredkin gates is straightforward by taking into account the afore-presented simulation and the discussions in [2].

3. Results and discussion The energy (frequency) of the photon pulse required to switch a photoswitchable protein from state 1 to state 2 usually significantly differs from that of the pulse, which switches it back to state 1 (selective switching). It has been already shown in the case of Coulomb-coupled, two-state molecules with one-dimensional nuclear vibration that such selective switching permits the realization of universal logic gates (Fredkin gate, and majority gate combined with inverting molecules) [2]. Since photoswitchable proteins can be switched selectively as well, they also permit the realization of such logic gates. In the following I will discuss the potentials of such proteins in the realization of computing and signal processing architectures with the aid of our model. 3.1. Pulse-driven photoswitchable protein chain The following example demonstrates the process of loading a bit on the second protein of a chain consisting of identical photoswitchable proteins (Figure 2) similarly to [2] that used the same example structure in the case of diatomic molecules. The potential energy curves of the reversibly photoswitchable protein are displayed in Figure 2(a). The molecule has two stable ground state configurations (state 1, state 2), the two states have different dipole moments (μ1=500 Debye, μ2=1000 Debye), it can be switched to the excited states of the two different configurations by photons with different energies (in the example DE1=2 eV, DE2=3 eV). The transition dipole moments were set to Dμ1=Dμ2=20 Debye during the simu-

Figure 2. Loading a bit to the second protein of a chain consisting of identical proteins. a): Potential energy curves of the protein. The red and blue arrows show the reaction paths from state 1 to state 2 and vice versa, respectively. The dashed circle highlights the place, where surface hopping occurs between the two states. b): Three subsequent photon pulses load a bit onto the second protein. Articles

47


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

3.2. Other energy-level arrangements Figure 3(a) displays an arrangement in which the influence of the states of the neighbours on the energy level difference of the corresponding molecule can be further improved. Let us assume that state 2 of molecule B is originally not stable, and the dipole moment of state 1 (μ1) is greater than that of state 2 (μ2). The influence of molecule A on the energy difference between the two states of B is described by (7) where DE0 is the energy difference without the influence of molecule A, dAB is the distance between the molecules. Since DE depends on the state of molecule A, by properly choosing the parameters of the molecules, a stable state 2 can be induced if molecule A is switched from state 1 to state 2, since the dipole moment of state 2 of molecule A is greater than that of state 1, thereby lowering the energy of state 2 of molecule B with respect of state 1. Note that in this case molecule B can be switched from state 1 to state 2 only if molecule A is in state 2. A main advantage of this arrangement is that the influence of the states of the neighbours on the energy levels of the protein is even stronger than in the previous case (section 2.1), since the energy difference between the states of molecule B depends on the product of μA and μB, where both μA and μB can be significantly higher than the transition dipole moment in equation (6).

N° 4

2009

4. Conclusion and future outlook In this paper, I discussed the potential of photoswitchable protein molecules in the realization of macromolecular computing, and digital signal processing arrays. For the simulations I used a simple quantum-classical nano-electromechanical model. Such proteins are potential candidates of future nanometre size computing architectures since they are cheap, can be assembled in monolayers, can be engineered to provide desired properties. Furthermore, they can provide stronger influence on each other during Coulomb-coupling. The authors hope that this study encourages further experimental studies needed for the realization of such computing architectures. Future studies should include the identification of already existing or artificially developed proteins with the desired properties as well as the application of nanotechnology for the fabrication of these structures.

ACKNOWLEDGMENTS The author would like to express their gratitude to Á. I. Csurgay for his comments and suggestions during finalization of the manuscript. The authors wish to thank the Norwegian Financial Mechanism and the Hungarian Research Fund (OTKA-NNF 78703, TO 72338), and the Mecenatura grant for their financial support.

AUTHOR Balázs Rakos - Department of Automation and Applied Informatics, Budapest University of Technology and Economics, Budapest, 1111, Hungary. E-mail: balazs.rakos@gmail.com.

a)

b) References [1]

Figure 3. a) illustrates the effect of inducing a stable state 2 in molecule B by switching molecule A to state 2. b) displays the energy levels of molecules in a one-dimensional molecular chain. Only the ground-state energy levels are shown.

[2]

[3]

This kind of arrangement also permits signal propagation and the realization of logic gates. Figure 3(b) shows the ground-state energy levels of a molecular chain consisting of an input protein, and two other molecules with different parameters. The sequence of propagating a bit along the chain by subsequent photon pulses is the following: the input molecule is switched to state 2, which induces a stable state 2 in the second molecule in the chain; then the second molecule is switched to state 2, which induces a stable state 2 in the third molecule; the third molecule is switched to state 2, a stable state 2 is induced in the fourth molecule; the input molecule is switched back to state 1; the fourth molecule is switched to state 2; the second molecule is switched back to 1, and the process goes on until the signal reaches the other end of the chain.

48

Articles

[4]

[5]

[6]

Porod W., Lent C.S., Bernstein G.H., Amlani I., Snider G. I., Merz J.L., “Quantum-dot cellular automata computing with coupled quantum dots”, International Journal of Electronics, vol. 23, no. 5, 1999, pp. 549-590. Csurgay Á.I., Porod W., Rakos B., “Signal processing by pulse-driven molecular arrays”, Int. J. Circ. Theor. Appl., vol. 31, 2003, pp. 55-66. Rakos B., Csurgay Á. I., Porod W., “Recovering pure states in two-state quantum systems”, Superlattices and Microstructures, vol. 34, 2003, pp. 503-507. Rakos B., Porod W., Csurgay Á. I., “Computing by pulsedriven nanodevice arrays”, Semicond. Sci. Technol., vol. 19, 2004, pp. 472-474. Habuchi S., Ando R., Dedecker P., Verheijen W., Mizuno H., A. Miyawaki, Hofkens J., “Reversible single-molecule photoswitching in the GFP-like fluorescent protein Dronpa”, Proc. Nat. Acad. Sci., vol. 102, no. 27, 2005, pp. 9511-9516. Tully J., Preston R., J. Chem. Phys., vol. 55, 1971, pp. 562.


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

SEEBECK COEFFICIENT MEASUREMENT BY KELVIN-PROBE FORCE MICROSCOPY Hiroya Ikeda, Faiz Salleh, Kiyosumi Asai

Abstract: In order to measure the Seebeck coefficient of nanometer-scale thermoelectric materials, we propose a new technique in which the thermoelectric-motive force (TEMF) is evaluated by Kelvin-probe force microscopy (KFM). In this study, we measured the Seebeck coefficient of an n-type Si wafer. The surface-potential difference between the high- and low-temperature regions on the Si wafer increases with increasing temperature difference. This indicates that the TEMF can be measured by KFM. The Seebeck coefficient evaluated from the surface-potential difference is 0.71Âą0.08 mV/K, which is close to that obtained by the conventional method. Keywords: Seebeck coefficient, Kelvin-probe force microscopy, nanostructure.

1. Introduction The introduction of nanometer-scale structures into thermoelectric materials has been expected to lead to breakthroughs for enhancing the thermoelectric figureof-merit [1]-[4]. A number of researchers are engaged in characterizing nanometer-scale thermoelectric materials [5]-[8]. However, it is very difficult to measure the thermoelectric characteristics of these materials because of the very small dimensions. External disturbances such as lead-wire contact essential to the conventional thermoelectric-motive force (TEMF) measurement affect accurate evaluation. Recently, probe microscopy techniques have attracted significant attention for the thermoelectric characterization of nanometer-scale materials [9], [10]. In these techniques, a probe is contacted with the sample surface and thermoelectric characterization is performed on the basis of the vertical temperature difference (i.e. normal to the sample surface). Therefore, contact of the metallic probe with the sample surface cannot be avoided, and specially-customized equipment is needed. We propose a new technique using Kelvin-probe force microscopy (KFM) for Seebeck coefficient measurement. Using this technique, it is possible to obtain the workfunction difference between the cantilever and the sample, that is, the Fermi energy of the sample relative to that of the cantilever metal. Consequently, TEMF can be obtained from the Fermi energies at the high- and lowtemperature regions. This allows for evaluation of the Seebeck coefficient of the sample. One of the crucial advantages to be emphasized of this technique is that the cantilever never touches the sample surface during the measurement. Therefore, the TEMF measurement is

not perturbed by external factors such as the metallic probe and the lead wire. Another advantage is that we can use commercial KFM equipment by adjustments of the sample holder. In the present report, we demonstrate the use of KFM for the measurement of the Seebeck coefficient of a bulk Si wafer and show that KFM can be a powerful tool for characterizing the Seebeck coefficient of nanometer-scale materials.

2. Experimental Figure 1 shows a schematic of the experimental setup. Two Cu plates were placed side by side with a gap of 4 mm. A resistive heater was attached to one of the Cu plates. An n-type Si wafer, with an impurity concentra18 2 tion of 1x10 cm-3 cut to a size of 5 x10 mm , was bridged over these Cu plates and attached using conductive Ag-paste. By heating one side of the sample, a temperature difference is produced in a plane parallel to the sample surface. Two K-type thermocouples were directly attached to the sample surface. Time evolution of the surface potential was measured by KFM equipment (Seiko Instruments Inc. SPI 3800N) and monitored by a digital multimeter (HIOKI HiLOGGER 8430), simultaneous with temperature measurement at the high- and low-temperature regions on the sample surface. The KFM cantilever was made of Si coated with Au. Surface-potential measurements were carried out in a vacuum chamber (Seiko Ins-5 truments Inc. SPA-300HV) with a pressure of 2.5x10 Pa.

Fig. 1. Schematic diagram of the apparatus for Seebeck coefficient measurements by KFM.

3. Results and Discussions The time evolution of the surface potential in the high-temperature region and of the temperature at the high- and low-temperature regions on the n-type Si wafer are shown in Figs. 2(a) and 2(b), respectively. Unfortunately, the KFM measurement could not sufficiently follow the time evolution of the surface potential due to overflow in the z-gain during the elevation of the temperature. In this study, therefore, we heated the sample step-by-step and measured the surface potential only at Articles

49


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

each stable-temperature region. During the elevation of the temperature, the cantilever was kept away from the sample. By this procedure, a pulse-like signal was obtained for the surface potential and staircase-shaped curves were obtained for the temperature time-evolution, as shown in Fig. 2. Similarly, the surface potential at the low-temperature region was also obtained. It is seen in Fig. 2(a) that the surface potential is relatively stable during the KFM detection. However, the surface potential has a finite value even during the measurement break. This is likely due to parasitic potential originating from the equipment, which needs to be subtracted from the surface-potential signal.

N째 4

2009

evaluation of the TEMF, the additional potential fadd must be removed from the detected surface potential VKFM.

Fig. 3. Energy bands of the KFM measurement setup (a) before and (b) after application of a voltage to remove the Coulomb force between the cantilever and the sample surface VKFM, and (c) simple equivalent circuit of the band diagram with VKFM. For this purpose, an equivalent circuit, shown in Fig. 3(c), is assumed for the band diagram of Fig. 3(b), where two capacitors are connected in series. In this circuit, two equations,

Fig. 2. Time evolution of (a) surface potential at the hithtemperature region and (b) tempera-tures at the high- and low-temperature regions on an n-type Si wafer.

VKFM = fSi + fadd

(1)

CSifSi = Caddfadd

(2)

are valid, where CSi and Cadd are the capacitance of vacuum and additional capacitors, respectively. These are expressed as [11]

CSi = Moreover, the influence of the Schottky contact at the n-type Si/Ag paste interface must also be taken into account. Figures 3(a) and 3(b) show schematic energyband diagrams of our KFM measurement setup. Before the application of a voltage to remove the Coulomb force between the cantilever and the sample surface (VKFM), our setup has the band diagram shown in Fig. 3(a) wherein all the Fermi energies of the Cu plate, the Ag paste, the n-Si and the Au coating the Si cantilever are identical. Hence, the Si band bends at the surface and the bottom, and the vacuum level slopes between the cantilever and the sample, representing the existence of the Coulomb force. After applying VKFM, the vacuum level becomes flat (no Coulomb force), and there is an energy difference -eVKFM between the Fermi energies of the cantilever and the Ag paste. The potential difference VKFM consists of voltage drops at vacuum (fSi) and the Si bottom (fadd). Since a Fermi energy difference between the cantilever and the sample -efSi is necessary for 50

Articles

e0 d

Cadd =

(3)

eeSiND , d(Vbi - fadd)

(4)

where d is the gap distance between the cantilever and the sample, and ND the impurity concentration in Si, e0 and eSi are the dielectric constant in vacuum and Si, and Vbi is the built-in potential at the n-type Si/Ag interface, which is expressed by Vbi = fAg - cSi - (EC - EF)/e ,

(5)

where ecSi and efAg are the electron affinity of Si and the work function of Ag, respectively. The surface potential fSi is obtained by the equation fSi = cSi + (EC - EF)/e - fAu ,

(6)


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

where efAu is the work function of Au. The theoretical fSi for DT=0K (T=297K) is calculated using Eq. (6) to be 18 -3 1.016V for ND=1x10 cm since the system is in thermal equilibrium under this condition. Then, assuming that the vacuum gap d=9 nm, fadd and VKFM are estimated to be 0.219 and -1.235V, respectively. Here, the potential difference between the measured and calculated KFM values for DT=0K is likely to be due to a background potential fbg originating from the measurement equip ment. Therefore, the corrected KFM value can be defined ~ as VKFM=VKFM - fbg unless the background potential fbg strongly depends on the temperature. From Eqs. (1)-(6), the true surface potential of Si is represented as: ~

fSi =

VKFM 1 + CSi

~

2(fAg - fAu - VKFM) eeSiND

,

(7)

which is shown in Fig. 4 for the high- and low-tempeH L rature regions ( fKSi and fKSi ) as a function of temperature difference. The surface-potential difference between these temperature regions is found to increase with increasing temperature difference, indicating that the TEMF can indeed be measured. Since the values of the surface potentials appear to lie on straight lines, the Seebeck coefficient is constant in this temperature range. From the gradients of these straight lines, the Seebeck H coefficient was estimated to be S=DV/DT=(fKSi L fKSi )/DT =0.71±0.08 mV/K, which is close to the value measured by a conventional method of S=0.89 mV/K [12]. This result indicates that the KFM technique indeed has the ability to evaluate the Seebeck coefficient.

2009

in this study stands on a lot of assumptions. This makes the validity of the resultant value doubtful, especially in nanometer-scale measurements. We have to measure the Seebeck coefficient for samples with Ohmic contact in future. ACKNOWLEDGMENTS This work was financially supported by a Grant-in-Aid for Scientific Research (19560701 and 21360336) from the Japan Society for the Promotion of Science.

AUTHORS Hiroya Ikeda*, Faiz Salleh, Kiyosumi Asai - Research Institute of Electronics, Shizuoka University, Johoku 3-5-1, Naka-ku, Hamamatsu 432-8011, Japan. E-mail: ikeda@rie.shizuoka.ac.jp. * Corresponding author

References [1]

[2]

[3]

[4]

[5]

[6]

[7]

[8]

Fig. 4. Surface potentials at the high- and low-temperature regions on the n-type Si wafer as a function of temperature difference. [9]

4. Conclusions We have developed a new technique using KFM for measuring the Seebeck coefficient of nanometer-scale thermoelectric materials. In our present experiment, the Seebeck coefficient of an n-type Si wafer was estimated to be S=0.71±0.08 mV/K, which is close to that obtained by a conventional method. This indicates that the Seebeck coefficient can indeed be measured by KFM with no contact between the probe and the sample, leading to realization of accurate Seebeck-coefficient measurement for nanometer-scale materials. However, some problems also come significant. The severest point is that the Seebeck coefficient evaluation

N° 4

[10]

[11] [12]

Hicks L.D., Dresselhaus M.S., ”Effect of quantum-well structures on the thermo-electric figure of merit”, Phys. Rev. B, vol. 47, no. 19, 1993, pp. 12727-12731. Hicks L.D., Dresselhaus M.S., ”Thermoelectric figure of merit of a one-dimensional conductor”, Phys. Rev. B, vol. 47, no. 24, 1993, pp. 16631-16634. Balandin A.A., Lazarenkova O.L., ”Mechanism for thermoelectric figure-of-merit enhancement in regimented quantum dot superlattices”, Appl. Phys. Lett., vol. 82, no. 3, 2003, pp. 415-417. Simkin M.V., Mahan G.D., ”Minimum thermal conductivity of superlattices”, Phys. Rev. Lett., vol. 84, no. 5, 2000, pp. 927-930. Harman T.C., Taylor P.J., Walsh M.P., LaForge B.E., ”Quantum dot superlattice ther-moelectric materials and devices”, Science, vol. 297, 2002, pp. 2229-2232. Li D., Wu Y., Kim P., Shi L., Yang P., Majumdar A., ”Thermal conductivity of indi-vidual silicon nanowires”, Appl. Phys. Lett., vol. 83, no. 14, 2003, pp. 2934-2936. Hochbaum A.I., Chen R., Delgado R.D., Liang W., Garnett E.C., Najarian M., Majumdar A., Yang P., ”Enhanced thermoelectric performance of rough silicon nanowires”, Nature, vol. 451, 2008, pp. 163-167. Boukai A.I., Bunimovich Y., Kheli J.T., Yu J.K., Goddard W.A. III, Heath J.R., ”Silicon nanowires as efficient thermoelectric materials”, Nature, vol. 451, 2008, pp. 168-171. Williams C.C., Wickramasinghe H.K., ”Microscopy of chemical-potential variations on an atomic scale”, Nature, vol. 344, 1990, pp. 317-319. Lyeo H.K., Khajetoorians A.A., Shi L., Pipe K.P., Ram R.J., Shakouri A., Shih C.K., ”Profiling the thermoelectric power of semiconductor junctions with nanometer resolution”, Science, vol. 303, 2004, pp. 816-818. Sze S.M., Semiconductor Devices, Physics and Technology, John Wiley & Sons, 1985, Chap. 5. Salleh F., Asai K., Ishida A., Ikeda H., ”Seebeck coefficient of ultrathin silicon-on-insulator layers”, Appl. Phys. Express, vol. 2, 2009, pp. 071203-1-3.

Articles

51


Journal of Automation, Mobile Robotics & Intelligent Systems

VOLUME 3,

N° 4

2009

SINGLE-ELECTRON TRANSPORT CHARACTERISTICS IN QUANTUM DOT ARRAYS DUE TO IONIZED DOPANTS Daniel Moraru, Maciej Ligowski, Juli Cha Tarido, Sakito Miki, Ryusuke Nakamura, Kiyohito Yokoi, Takeshi Mizuno, Michiharu Tabe

Abstract: Single charge manipulation for useful electronic functionalities has become an exciting and fast-paced direction of research in recent years. In structures with dimensions below about 100 nm, the physics governing the device operation turn out to be strikingly different than in the case of larger devices. The presence of even a single charge may completely suppress current flow due to the basic electronelectron repulsion (so called Coulomb blockade effect) [1]. It is even more exciting to control this effect at the level of single-electron/single-atom interaction. The atomic entity can be one donor present in silicon lattice with a Coulombic potential well. In principle, it can accommodate basically a single electron. We study the electrical behavior of nanoscale-channel silicon-on-insulator field-effect transistors (SOI-FETs) that contain a discrete arrangement of donors. The donors can be utilized as “stepping stones” for the transfer of single charges. This ability opens the doors to a rich world of applications based on the simple interplay of single charges and single atoms, while still utilizing mostly conventional and well established fabrication techniques. In this work, we distinguish the effects of single-electron transport mediated by one or few dopants only. Furthermore, we show how the single-electron/single-donor interaction can be tuned by using the external biases. We demonstrate then by simulation and experiment the feasibility of single-electron/bit transfer operation (single-electron turnstile).

2. Coulomb blockade in dopant-induced quantum dots We fabricated and investigated doped-nanowire SOIFETs containing a large number of dopants in the channel. The device structure is schematically shown in Fig. 1(a). The nanowire channel was patterned by an electron-beam lithography technique after doping with phosphorus by diffusion from a spin-coated silica film. The dimensions of the channel are estimated to be about 10 nm height, 50 nm width, and 100 nm length. Doping concentration is expected to be on the order of 1×1018 cm-3 which suggests that the number of dopants in the channel is very roughly about 50. Figure 1(b) shows the channel containing randomly distributed dopants (as a result of the uncontrollability in the doping process). Figure 1(c) shows a simulated possible potential landscape created by superposition of the Coulombic potentials of all the dopants in the channel. The fact that the interdopant distance is about 10 nm [(Nd)-1/3] and thus larger than Bohr radius for phosphorus in Si (~3 nm) is another hint that individual dopants may work as QDs. Due to the relatively large number of dopants and to the device geometry, it is plausible that a multiple-QD array will be formed between source and drain. This can be understood from typical source-drain current (Isd) vs front gate voltage (Vfg) characteristics, as shown in Fig. 1(c), which contain strong oscillations of the current. These oscillations are due to the Coulomb blockade in the dopantinduced QDs.

Keywords: single dopant, silicon nanowire, single-electron transport, single-electron transfer.

1. Introduction Single-electron tunneling and the Coulomb blockade effect have been observed in a variety of materials [1]. Silicon devices are, however, preferable due to the welldeveloped Si technology. Furthermore, utilizing naturally-formed Coulombic wells introduced by individual atoms (dopants) allows us to overcome present limitations of nanolithography. This extreme case of incorporating the physics of single-electron/single-atom interaction into useful electronic devices is the focus of this paper. In the following, we will briefly describe the basic physics of this interaction and the possibilities of controlling the device parameters for realizing applications such as single-electron turnstile [2].

52

Articles

Fig. 1. (a) Device structure for doped-nanowire FETs investigated. (b) Nanoscale channel randomly doped with phosphorus. (c) Dopant-induced potential landscape simulated for one possible dopant arrangement. (d) Measured Isd-Vfg characteristics exhibiting typical Coulomb oscillations.


Journal of Automation, Mobile Robotics & Intelligent Systems

The theory of Coulomb blockade is well established nowadays for metallic QDs. In ultrasmall (nanometerscale) QDs coupled to the environment through tunnel junctions, the presence of even one electron in the QD significantly raises the dot potential blocking further transport. This increase is related to the charging energy Ech=e2/2C, where e is the elementary charge and C is the dot capacitance. For a donor-induced QD, the situation is roughly similar. However, donors can in principle accommodate only one electron (which will practically neutralize the donor). [3] This reasoning can be extended further to arrays of dopants, which have been acknowledged as more attractive for various applications. Roughly speaking, when a nanowire contains many dopants, electrons “looking” into the channel from the source Fermi level will first see the lowest-energy potential valleys (most probably serially arranged). It is important to note that these lowest-energy valleys are expected to be the result of the strong confinement potential due to individual dopants. As gate voltage is increased, the channel potential is continuously lowered. Current will flow by successive single-electron tunneling events and then electrons will become “trapped” inside the dopant-QDs and transport will be blocked. This will give rise to a current peak, from which we can extract information about the structure of the QD array itself. As gate voltage is even further increased, it is expected that conduction path will change so that it incorporates other dopants located in the next lowest potential valleys. This mechanism suggests the possibility of characterizing and utilizing doped nanowires still based on single-electron/single-dopant interaction. One of many possible applications consists of single-electron turnstile mediated by few-dopant arrays.

3. Finding the optimal structure for singleelectron turnstile - statistical simulations Single-electron turnstile is defined as the ability of a device to shift one electron between two electrodes during every cycle of an ac gate voltage. The conditions required to achieve this operation can be understood from the charge stability diagram of the device (i.e., contour map of source-drain current (Isd) as a function of sourcedrain bias (Vsd) and front gate voltage (Vfg)). Such a stability diagram is shown in Fig. 2(a) for a 3-QD system with uniform parameters. It contains zero-Isd rhomboidal regions (so-called Coulomb diamonds) inside which a fixed number of elementary charges reside in the QD array. Consecutive Coulomb diamonds correspond to charge configurations that differ only by one electron. For single-electron turnstile, the basic requirement is an overlap between two such consecutive stable regions. An ac gate voltage crossing this overlap periodically may lead to singleelectron/cycle transfer from source to drain as follows. During the high level of the pulse, one electron is injected into the channel from one electrode (e.g., source – the source-side arm of the “turnstile” is open while the drainside one is still closed). Then, during the low level, the electron is removed towards the opposite electrode (e.g., drain - the drain-side arm is opened this time). We found that the required overlap is strongly dependent on the array parameters. Figure 2(b) shows, as ex-

VOLUME 3,

N° 4

2009

ample, few Isd-Vsd characteristics simulated under ac-Vfg bias for different inter-dot coupling (inter-dot capacitances). The other parameters are all fixed. We can see that by adjusting the inter-dot coupling we find a current plateau aligned at e×f (where f is the ac frequency). This is a clear mark of single-electron turnstile. Controlling this parameter is a key aspect of our experimental results, as shown in Section 4.

Fig. 2. (a) Simulated stability diagram for a 3-QD array (first two stable regions are marked). (b) Simulated Isd-Vsd characteristics under ac-gate operation for different interdot coupling (inter-dot junction capacitances). (c) Statistical results of single-electron turnstile operation for a large number of QD arrangements. 3-QD arrays (particularly with larger central dot and/or larger inter-dot capacitances) are found to be stable turnstile devices. When dopants are utilized as QDs, it is important to consider their positions as random (following basically a Poisson distribution). Therefore, it is necessary to understand the effect of parameter dispersion on the ac behavior of multiple-QD arrays. Our statistical analysis [4] of a variety of configurations suggests that single-electron turnstile is achievable in devices containing more than 3 QDs with a fairly high chance (as shown in Fig. 3(c)). It is even more important though that we were able to find an optimal structure which would be very stable against parameter fluctuations. This consists of a 3-QD array commonly coupled to a gate. When the central dot is made larger than the outer ones, the probability of realizing single-electron turnstile is practically 100%. These findings are promising for developing very stable single-electron turnstile devices based on single-electron/single-dopant interplay.

4. Electrical characterization and tuning of dopant arrays Simulation results suggest the importance of controlling the device parameters for single-electron turnstile. This is consistent with our previous experimental observations of single-electron turnstile operation in few-dopant nanowire SOI-FETs [5]. In this paper, we attempt to control the QD array parameters based on the specific properties of our SOI-FET structure. The advantage of the SOI-FET is that it allows the characterization of the transport through the nanoscale channel with four external Articles

53


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

biases: source and drain, front gate, and back gate (metalized substrate). We focus on identifying at first the necessary conditions for single-electron turnstile, i.e., an overlap between adjacent Coulomb diamonds in the charge stability diagram. Figure 3(a) shows the stability diagram for one device measured at 17 K with zero back gate voltage (Vbg). Coulomb diamonds can be observed and it is expected that the number of electrons inside the QD array changes one by one as Vfg is increased. However, the necessary overlap cannot be clearly identified. It can be concluded that the device parameters are not optimal for single-electron turnstile operation.

N° 4

2009

5. Conclusions We have shown that single-electron/single-dopant interplay can be controlled electrically in doped-nanowire FETs. This allowed the adjustment of the device charge stability diagrams for operation as single-electron turnstile. Further insight can be provided by the ability of directly monitoring the dopant arrangement. This has been also recently demonstrated by our group utilizing a low-temperature Kelvin probe force microscope [7]. The combination of these techniques can become an essential frame for development of single dopant electronics field [8]. ACKNOWLEDGMENTS The authors thank D. Nagata for support in experiments and H. Ikeda for discussions. This work was partially supported by Grants-in-Aid for Scientific Research (16106006 and 18063010) from the Ministry of Education, Culture, Sports, Science, and Technology of Japan.

Fig. 3. (a)-(c) First stable charge regions as a function of Vbg. (d) Isd-Vsd characteristics under ac-Vfg pulses. The ef plateau is observed only for the “optimal” case of Vbg=-5 V. We suggest that back gate voltage (Vbg) can be utilized to tune the stability diagrams appropriately. Figures 3(a)-3(c) show the influence of Vbg on the stability diagrams. The structure of the stable regions is significantly changed by changing Vbg. Most importantly, it becomes now possible to identify an optimal overlap between the first and second observable Coulomb diamonds. This adjustment is due to the strong electric field vertically across the channel created by the large Vbg. The Vbg-induced field is felt differently by dopants located at different depths in the channel. Although not very straightforward, this means that in fact the inter-dopant coupling can be modulated by Vbg [6]. After tuning the charge stability diagrams using Vbg, we apply ac Vfg pulses to the front gate. The high and low levels of the Vfg pulse were set as described above inside the first and the second observable stable regions. The device is actually operated inside the zero-dc-current areas (inside the Coulomb diamonds). However, as shown in Fig. 3(d), under ac-Vfg operation, we found that Isd is in fact not zero and, more strikingly, for Vbg=-5 V it is very close to e×f (where f is the ac frequency). This means that during each Vfg cycles precisely one electron is transferred between source and drain. Single-electron transfer is thus achieved by modulating the inter-dopant coupling with Vbg [6]. Furthermore, this operation takes place by single-electron tunneling in single-dopant-QD arrays. This ability is very promising for developing a new class of electronics based on single-electron/single-atom interaction. 54

Articles

AUTHORS Daniel Moraru*, Maciej Ligowski, Juli Cha Tarido, Sakito Miki, Ryusuke Nakamura, Kiyohito Yokoi, Takeshi Mizuno, Michiharu Tabe - Shizuoka University, Research Institute of Electronics, Johoku 3-5-1, 432-8011 Hamamatsu, Japan. E-mail: daniel@rie.shizuoka.ac.jp. Maciej Ligowski - Warsaw University of Technology, Division of Sensors and Measuring Systems, A. Boboli 8, Warsaw 02-525, Poland. * Corresponding author

References [1]

[2]

[3] [4]

[5]

[6]

[7]

[8]

Averin D.V., Likharev K.K., in “Single charge tunneling”, ed. H. Grabert and M. Devoret, Plenum, New York, 1992, p. 311. Geerligs L.J., et al., “Frequency-locked turnstile device for single electrons”, Phys. Rev. Lett., vol. 64, no. 22, 1990, pp. 2691-2694. Kohn W., Luttinger J.M., “Theory of donor states in silicon”, Phys. Rev., vol. 98, no. 4, 1955, pp. 915-922. Yokoi K.K., Moraru D., Ligowski M., Tabe M., “Single-gated single-electron transfer in nonuniform arrays of quantum dots”, Jpn. J. Appl. Phys., vol. 48, 2009, pp. 024503. Moraru D., Ono Y., Inokawa H., Tabe M., “Quantized electron transfer through random multiple tunnel junctions in phosphorus-doped silicon nanowires”, Phys. Rev. B, vol. 76, no. 7, 2007, pp. 075332. Moraru D., Ligowski M., Yokoi K., Mizuno T., Tabe M., “Single-electron transfer by inter-dopant coupling tuning in doped nanowire silicon-on-insulator field-effect transistors”, Appl. Phys. Exp., vol. 2, 2009, pp. 071201. Ligowski M., Moraru D., Anwar M., Mizuno T., Jablonski R., Tabe M., “Observation of individual dopants in a thin silicon layer by low temperature Kelvin probe force microscope”, Appl. Phys. Lett., vol. 93, no. 14, 2008, pp. 142101. Tabe M., Nuryadi R., Moraru D., Burhanudin Z.A., Yokoi K., Ikeda H., “Si multidot FETs for single-electron transfer and single-photon detection”, Acta Physica Pol., vol. 113, no. 3, pp. 811-814.


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

SOL-GEL GLASSY ANTIREFLECTION GEO2 – SIO2 – AG – RE FILMS FOR SOLAR CELLS AND IR-DEVICES Dmitry Kovalenko, Vladimir Gaishun, Alina Semchenko, Vitaly Sidsky, Nikolay Aleshkevich, Vasili Vaskevich

Abstract: Optically homogeneous double-coated of GeO2-SiO2 films, sodoped by ions of silver and cerium were fabricated with use sol-gel of a method, using tetraethylorthogermanate (TEOG) and tetraethylorthosilicate (TEOS) as precursors for germania and silica, respectively. According to data IR-spectroscopy of a coat are glassy. Refractive indices for the materials were obtained the following: GeO2, n = 1.687 at thickness d = 530.5 nm and SiO2, n = 1.512 at d = 998.2 nm. They demonstrate good adhesion to the glass, germanium, silica and metal substrates. Introduction of Re-earth ions m allows to control refractive index within the limits of from 1,2 up to 1,6, that is important for deriving the multilayer antirefractive coats for the infra-red optics. Introduction silver nanoparticles allow using the yielded coats for magnification of efficiency of solar cells. Keywords: GeO2-SiO2 films, antirefractive film, solar cells, IR-devices.

1. Introduction Series of works are devoted to investigation of germaniumsilcate materials (glasses and films). Silica glass doped with GeO2 has been investigated due to of the ultraviolet (UV)-induced refractive index change that is responsible for the formation of distributed Bragg gratings [1]. Also wide-band gap transparent materials are at demand for antireflection coatings of night vision devices, solar cells and manufacturing of planar wave-guides. GeO2 is widely used for these purposes. Now widely used vacuum deposition and colloidal methods for fabrication of GeO2 films with application in solar cells and planar wave-guides. Use direct sol-gel of a method and organic compounds of germanium, can provide high optical quality and an opportunity of more flexible check of properties of films as compared with other existing methods [2], and also it is essential to simplify production engineering of their deriving. The films doped with europium demonstrate the intensive luminescence about 600 nm. This issue is of importance for possible blue shift of the spectral range for silicon-based solar cells those possess the maximum sensitivity at 600-620 nm. Silver ions and nanoparticles in the germania matrix are efficient absorbers at 300-500 nm due to the plasmon resonance band, and 3+ they can pass the light energy to Eu emission centers [3]. In the present work, we have developed the sol-gel technique for preparation of GeO2, GeO2-Re2O3, GeO2-Ag, GeO2-Re2O3-Ag (Re=Eu, Ce, etc.) films by incorporation of Ag, Eu and Ce-compounds into precursor sol followed by

deposition of films onto silicon and germanium wafers. Homogeneous and transparent films of thickness in the range 0.2-2 μm were produced after the heat treatment at 5000°C in air with good adhesion to the surface of glasses, silicon and germanium wafers. Thus, sol-gel derived germania coatings incorporated with silver and rare earth elements show perspective optical features aimed at the application for antireflective film and materials for solar cells sensitization. 1.1. Subsections The germanium films with silver nanoparticles and rear earth ions were produced by spin coating on silica wafers from precursor sols prepared by mixing tetraethylorthogermanate (TEOG) in water–ethanol solution. HNO3 was used as the catalyst of TEOG hydrolysis. The solution was self-heated up to temperature 30°C due to exothermic hydrolysis and polycondensation reactions. A ripening of sols passed at 22±2°C for 4–5 days in closed vessels. Ripened sols were stable at ~20°C for 2 months. Silver was incorporated by adding of AgNO3 (0.5–1 wt.%) in the precursor sols. Also was added EuNO3 (0.5–1 wt.%) and CeCl3 (0.5–1wt.%). After the spin-coating step, samples were heated up to 500°C in air for 15 min. Synthesis of film-forming solution

TEOG+C2H5OH+H2O

TEOS+C2H5OH+H2O

Ripening of sol (hydrolysis, polycondensation)

Layer on substrate by spin-coating method

Add salts of Ag, Ce, Eu in initial sol

Heat treatment in the air at T=500°C

Layer SiO2 film on the GeO2 film Heat treatment in the air at T=500°C

GeO2 film

SiO2 film

substrate

GeO2 film substrate

Fig.1. Scheme of formation GeO2-Re-Ag and GeO2-SiO2-ReAg antireflection film. Articles

55


Journal of Automation, Mobile Robotics & Intelligent Systems

By sol-gel method also were the two-layer antireflective coat (Fig. 1.) is obtained. And for decrease of reflectivity index it is possible to synthesize nanosized systems of two types; in one case the multilayered antireflection system consists from layer with periodically changed of a refractive index which thickness can be identical. In other case the system consists from too layers with difference in a refractive index, but thickness of such layers are changed. The second layer was hydrophobic SiO2 solgel film. SiO2 film also will be obtained from film-forming (sol) solution, prepared by hydrolysis of tetraethylorthosilicate. Sol superimposed on the surface by spin-coating method. Additively, SiO2 films were deposited as protective layers (Fig. 1). Refractive indices for the materials were obtained the following: GeO2, n = 1.687 at thickness d = 530.5 nm and SiO2, n = 1.512 at d = 998.2 nm. They demonstrate good adhesion to the glass, germanium, silica and metal substrates. 1.2. Subsections Refractive index of clear GeO2 films was equal 1,6. Preliminary theoretical calculations have shown, that the coefficient of reflection of such films R makes no more than 3 % Introduction in initial sol salts of silver, cerium and europium will allow decrease value of refractive index up to 1.2. In Fig. 2 are shown infrared spectrums of a transition of GeO2 sol - gel of films superimposed on polished plates of single-crystal silicon wafers. There is an expressed –1 absorption band in the field of 3300 sm , the bound with –1 presence of hydroxyl groups OH. Bands of 1008-1040 sm largely depend on bridge bonds of germanium with oxy–1 gen Ge-O-Ge, and bands of 1059-1156 sm correspond to the valent antisymmetric oscillations Ge-O-Ge. As a result of heat treatment of films developing process of hydroxyl groups essentially varies, and basic GeO2 bands store position and intensity.

Fig. 2. IR-spectrum of GeO2 -Re-Ag film. According to IR spectroscopy Ag nanoparticles and Re-earth ions not influence on structure of GeO2 film. The absorption spectra of GeO2-SiO2–Re–Ag film have typical plasmon absorption peak of Ag nanoparticles with size 10-20 nm. In the range 1000-2600 nm films have 56

Articles

VOLUME 3,

N° 4

2009

a minimum sign of absorption D = 0,044. Such films cane used us antireflection films.

Fig. 3. Absorption spectra of GeO2-SiO2 -Re-Ag film. For checkout of an efficiency of GeO2-Re-Ag films coats have been superimposed on a surface of a solar battery. Figure 4 presents dependence of a voltage on a solar battery from a wavelength and a composition of antireflection film.

Fig. 4. Dependence of voltage solar battery form wavelength and composition of antireflection films. From figure follows, that the given coats can serve for increase of work efficiency of a solar device. The efficiency of a solar device is increased and the working area of a solar device expanded in wavelength diapason.


Journal of Automation, Mobile Robotics & Intelligent Systems

VOLUME 3,

N° 4

2009

2. Conclusion Changing thickness and sign of a refractive index of antireflection GeO2-SiO2-Ag-Re film allows displacing a minimum of reflection in various sites of a spectrum. Application double-layer antireflection film allows removing almost completely reflection of light from a surface of a detail. Now an important problem is reception of antireflections coats for IR-techniques (thermal imager, night vision equipment etc.). In particular, for germanium lenses. The transmission of germanium plates in the range from 3 up to 12 microns makes 70 %. Drawing antireflection coat allows increasing transition up to 98 %.

AUTHORS Dmitry Kovalenko*, Vladimir Gaishun, Alina Semchenko, Vitaly Sidsky, Nikolay Aleshkevich, Vasili Vaskevich - Advanced Materials Research Laboratory, F. Skorina Gomel State University, 246699, Gomel, Belarus, E-mail: dkov@gsu.by * Corresponding author

References [1] [2]

[3]

Jing L., et al., “Non-Crystalline Solids” Journal of 353, 2007, pp. 4128–4136. Gurin V.S., Alexeenko A.A., Prokopenko V.B., Kovalenko D.L., Formation and properties of sol-gel films and glasses with ultrafine metal and semiconductor particles, J. Inclusion Phenom. Macrocycl. Chem., 1999, pp. 291-297. Kovalenko D., Gurin V., Alexeenko A. Bogdanchikova N., Prokopenko V., Melnichenko I.M., “Features of spectroscopy and formation process of silica solgel films doped with silver nanoparticles”, J. Alloys Compounds, 2002, pp. 208-210.

Articles

57


Journal of Automation, Mobile Robotics & Intelligent Systems

VOLUME 3,

N° 4

2009

MULTI-SCALE SIMULATION OF HYBRID SILICON NANO-ELECTROMECHANICAL (NEM) INFORMATION SYSTEMS Hiroshi Mizuta, Mario A.G. Ramirez, Yoshishige Tsuchiya, Tasuku Nagami, Shun-ichiro Sawai, Shunri Oda, Masakuni Okamoto

Abstract: This paper presents emerging NEM hybrid systems for advanced information processing and describes our recent attempts of developing new multi-physics simulation technologies for these NEM systems at micro-, nano- and atomscales. Keywords: multi-scale simulation, NEMS, suspended-gate, non-volatile memory, phonon.

riety of new hybrid NEM-MOS devices have recently been studied for logic, memory and sensing applications (see Table 1). A typical NEM hybrid device is a suspended-gate field-effect transistor (SG-FET) which features a movable gate suspended over the MOS channel with an air gap. The SG-FETs have already attracted an increasing interest due to the extremely abrupt switching with the subthreshold slope much smaller than the theoretical limitation of 60 mV/dec for MOSFETs. Table 1. A variety of NEM – MOS hybrid devices.

1. Introduction Silicon VLSI technology, developed and matured over the past decades, has been fully exploited to build the vast technology area of micro-electromechanical systems (MEMS). The MEMS market is projected to grow with the rate of 30 – 40 % per annum and reach ten billion dollars in 2015. In parallel with such a rapid expansion of the MEMS market, there have also been continuous efforts at making the MEMS smaller in order to boost the operating frequency to GHz and beyond. Figure 1 shows a recent miniaturization trend of semiconductor-based MEMS, superposed on the CMOS downscaling trend (‘More Moore’).

Fig. 1. Miniaturization trend of semiconductor-based resonators shown superposed on the CMOS downscaling trend from International Technology Roadmap for Semiconductors 2008. Apparently, the MEMS technology has already entered a sub-μm regime and proceeds rapidly towards a nanometer-scale regime. The appearance of high-speed nanoelectromechanical systems (NEMS) tempts to consider the hybridization of the NEMS and conventional silicon electronic devices because we expect such hybrid systems enhance scaling of functional density & performance while simultaneously reducing the power dissipation beyond the conventional CMOS-based systems. A va58

Articles

For the memory applications, two types of novel highspeed and non-volatile NEM memory devices are currently researched to go beyond conventional Flash memory. As regards the sensing applications, an in-plane resonant suspended gate (RSG) MOSFET and a vibrating-body FET have been proposed very recently to achieve extremely high mass sensitivity. Along with such a rapid progress of the hybrid NEM devices, various attempts have been made to develop a new simulation technology suitable for the hybrid NEM devices. For the conventional MEMS, we have a number of powerful 2D/3D simulation tools based on the finite element method (FEM) analysis. In order to design and analyse our hybrid NEM-MOS systems, however, we need a new simulation technology, which solves the nonlinear mechanical equation electrostatically coupled to the carrier transport equations simultaneously. Furthermore, we have to build a multi-scale simulation, which facilitates to simulate the NEM hybrid systems at three different scales, i.e., microscale, nanoscale and atomscale (see Fig. 2). In the following sec-


Journal of Automation, Mobile Robotics & Intelligent Systems

VOLUME 3,

N° 4

2009

tions, we focus on hybrid NEM-MOS memories and nanophononic systems and present our recent attempts of developing the micro-, nano- and atom-scale simulation technology.

Fig. 3. 2D hybrid FEM simulation results for the selfbuckling NEM memory with the FG of 1 μm in length and stored charge density of 8.5x10-8 C/m2.

Fig. 2. Multi-scale simulation technology for hybrid NEM information devices and circuits.

2. Micro- and Nano-scale hybrid simulation of Si NEM-MOS systems As the first attempt of the NEM-MOS hybrid devices, we have researched on a self-buckling floating gate (FG) NEM memory [1],[2]. This memory device features a buckled SiO2 FG with a layer of embedded Si nanodots as charge storage (see Memory in Table 1). The buckled FG may be flip-flopped via the gate electric field, and its structural states are sensed via a change in the drain current of the MOSFET underneath. In order to design and analyse this NEM memory, we developed nanoscale hybrid FEM simulation that solves two- (or three-) dimensional Navier’s equation for structural mechanics analysis of the self-buckling FG and Poisson’s equation and the drift-diffusion equation for MOSFET electrical analysis simultaneously. The numerical simulation was conducted by using a general-purpose FED package COMSOL Multiphysics. Because three nonlinear equations need to be solved self-consistently in multiple degrees of freedom, it is inevitable that the numerical computation is timeconsuming. Therefore 2D simulation is preferably used unless 3D natures are remarkable with the designed structure. Figure 3 shows the 2D carrier distribution calculated for the NEM memory at the OFF (a) and ON (b) states as well as the out-of-plane FG displacement (c) and the MOSFET readout current (d) hysteresis as a function of gate voltage. The threshold voltage shift and the ON/OFF current ratio were found approximately 1.6 V and 5 10 at Vg = 0 V. These results clarify that the bistable states of the NEM memory can be indeed read via the drain current. The associated switching voltage is around 10 V, which should be reduced further.

As an alternative NEM memory architecture, we recently proposed a suspended gate Si nanodot memory (SGSNM) [3], [4]. The SGSNM consists of a MOSFET as readout, silicon nanodots as a FG, and a clamped suspended gate (SG), which is isolated from the FG by an air gap and a thin tunnel oxide (Fig. 4(a)). For the programming (P) process, a negative gate voltage is applied, and the SG is pulled-in on the FG layer, resulting in electron injection from the SG into the FG. For the erasing (E) process, a positive voltage is applied, and the stored electrons are extracted from the FG. The SGSNM architecture enables to avoid an unfavourable trade-off between the ON/OFF current ratio and the P/E voltages and therefore facilitates low-power operation compared to the first NEM memory.

Fig. 4. A schematic suspended-gate silicon nanodot memory (SGSNM) (a) and the hysteresis in the suspendedgate to substrate capacitance (c) associated with the pullin/pull-out operations (b). In order to analyze the SGSNM cell, we developed the Microscale hybrid equivalent circuit simulation (the bottom of Fig. 1), which enables a large-scale cell array simulation in the future. It is, however, vital to build Articles

59


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

a compact model of the SG including the tunnel injection and release processes. Therefore we first conducted a 3D FEM simulation to obtain full pull-in / pull-out characteristics for the SG (Fig. 4(b), (c)). The SG structural parameters were optimized in order to make the programming voltage as low as 6 V. The SG voltage dependences of the SG displacement and associated capacitance CSG were then calculated (Fig. 4(c)). It should be noted that the CSG-VSG curves show a remarkable hysteresis due to the SG stiction (static friction) onto the oxide. The calculated CSG-VSG hysteresis curve was modelled by using simple analytical formula. The same processes were conducted for negative gate voltages as well. Secondly, a compact model for the variable tunnel resistance was constructed based on the numerical simulation of quantum-mechanical tunnel current through the top gate oxide. The tunnel current density-voltage characteristics were calculated for a SiO2 tunnel barrier by solving the 1D Schrödinger equation. The developed compact models were introduced into SmartSpice by using Verilog-A. By using the developed hybrid modelling, the P/E/R processes were successfully analyzed as shown in Fig. 5. The SG voltage and drain voltage waveforms are shown in Figs. 5(a) and (b), and the transient memory node voltage and MOSFET readout current are shown in Fig. 5(c) and (d), respectively. By assuming a 7-nm-thick tunnel oxide, the results show that the SGSNM achieves the P/E times of as short as 1.7 nsec, suitable for fast & non-volatile RAM applications.

N° 4

2009

ner described in Section 2. However, we will need the atom-scale simulation framework for extremely scaled NEM systems whose electromechanical properties are described via low-dimensional phonons (nanophonons). Various new electron transport phenomena associated with nanophonons have been reported very recently, such as phonon blockade of singleelectron tunnelling and quasi-ballistic transport due to suppressed optical phonons [6], which may be exploited to develop novel functional information devices. As the first step towards the atom-scale NEM hybrid simulation, we developed the ab initio simulation of ‘nanophonons’ for the H-terminated freestanding ultrathin Si films of 3 to 10 atomic layers in thickness. The Si(0 0 1) 2x1 dimer structures were formed on the film surface, and the entire atomic structures were optimized by using the DFT (density-functional theory) simulator SIESTA. The atomistic properties of nanophonons were then calculated by using VIBRATOR, which is associated with SIESTA and based on the ab initio force-constant method.

Fig. 6. Ab-initio simulation of nanophononic spectra for a hydrogen-terminated ultrathin Si film of 5 atomic layers in -1 thickness. A phonon bandgap of 17.35 cm is formed both in the (110) and (1-10) directions.

Fig. 5. Program/Erase/Read signal waveforms simulated for the SGSNM cell.

We observed for the first time that remarkable phonon bandgaps are formed for the films thinner than 7 atomic layers (see Fig. 6) [7]. The formation of the phonon bandgaps is caused by the Si dimers formed on the film surfaces in the (1-1 0) or (1 1 0) direction. Such nanophononic properties are quite different from those for bulk silicon and may be further investigated as one of approaches to nanoscale thermal management and novel energy transfer interactions in the ‘Beyond CMOS’ information devices.

4. Conclusion 3. Towards atom-scale NEM systems By downscaling the hybrid NEM systems towards a nanometer regime, we may explore a variety of novel functional devices such as NEM-SET (single-electron transistor). A suspended-gate SET (SGSET) is an apparent extension of the SGFETs which is expected to achieve a dramatic modulation of the Coulomb oscillation period as well as extremely fast switching. We have recently developed a hybrid NEM-SET equivalent circuit simulation successfully [5] by implanting the SET analytical model and the NEM compact model into SmartSpice in a similar man60

Articles

Various attempts have been made to develop new micro- and nano-scale multi-physics simulation technologies suitable for the emerging NEM-MOS hybrid devices. The micro-scale hybrid circuit simulation with compact NEM models developed by using 3D FEM simulation provides a fast and practical approach and has been applied to the NEM memory devices successfully. The preliminary atom-scale simulation revealed the formation of phononic bandgaps in atomically thin Si films.


Journal of Automation, Mobile Robotics & Intelligent Systems

VOLUME 3,

N° 4

2009

AUTHORS Hiroshi Mizuta*, Yoshishige Tsuchiya - NANO Group, School of Electronics and Computer Science, University of Southampton, UK, and SORST JST (Japan Science and Technology). E-mail: hm2@ecs.soton.ac.uk Mario A.G. Ramirez - NANO Group, School of Electronics and Computer Science, University of Southampton, U.K. Tasuku Nagami, Shun-ichiro Sawai - Quantum Nanoelectronics Research Center, Tokyo Institute of Technology, Japan. Shunri Oda - Quantum Nanoelectronics Research Center, Tokyo Institute of Technology, Japan, and SORST JST (Japan Science and Technology). Masakuni Okamoto - Advanced Research Laboratory, Hitachi Ltd., Japan. * Corresponding author

References [1]

[2]

[3]

[4]

[5]

[6]

[7]

Tsuchiya Y., et al., “Nanoelectromechanical nonvolatile memory device incorporating nanocrystalline Si dots,” J. Appl. Phys., issue 100, 2006, 094306. Nagami T., et al.: “Three-dimensional numerical analysis of switching properties of high-speed and non-volatile nano-electro-mechanical memory,” IEEE Trans. Electron Devices, issue 54, 2007, 1132. Ramirez, M.A.G., et al., “Suspended gate silicon nanodot memory”, ESSDERC/CIRC Fringe, Edinburgh, 2008, 19. Ramirez, M.A.G., et al., “Hybrid circuit analysis of a suspended-gate silicon nanodot memory (SGSNM) cell”, to be published at MNE2009, Ghent, 2009. Pruvost B., et al., “Design optimization of NEMS switches for suspended-gate single-electron transistor applications”, IEEE Trans. Nanotechnology, no. 8, 2009, pp. 174-184. Mori N., et al., “Quasi-ballistic electron transport through silicon nanocrystals”, to be published at EDISON 16, Montpellier, 2009. Sawai S., et al., “Atomistic study of phonon states in hydrogen-terminated Si ultra-thin films”, IEEE Silicon Nanoelectronics Workshop, Honolulu, 2008, M0200.

Articles

61


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

NANOHILLS IN SIGE/SI STRUCTURE FORMED BY LASER RADIATION Artur Medvid', Pavels Onufrijev, Klara Lyutovich, Michael Oehme, Erich Kasper, Igor Dmitruk, Iryna Pundyk, Ivan Manak, Dainis Grabovskis

Abstract: Formation of self-assembling nanohills induced by irradiation of nanosecond Nd:YAG laser pulses on the Si0.7Ge0.3/ Si hetero-epitaxial structures is reported. The atomic force microscope study of the irradiated surface morphology has shown a start of nanohills formation after laser irradiation of the intensity I=7.0 MW/cm2. The giant "blue shift" of photoluminescence spectra with maximum intensity in region of 700-800 nm (1.76 - 1.54 eV) is explained by the Quantum confinement effect in the nanohills. The maximum of this photoluminescence band slightly shifts to shorter wavelengths with the increase of the intensity of laser pulses used for sample treatment. Appearance of the 300 cm-1 Ge-Ge vibration band in Raman scattering spectra for sample irradiated with I=20.0MW/cm2 is explained by Ge phase formation. Formation of the Ge-rich phase is explained by localization of Ge atoms drifting toward the irradiated surface under the thermal gradient due to strong absorption of laser radiation.

Si nanoparticles and associated surface state. Another effect has been observed in [5], for pure Ge crystal decrease of QDs diameter till 4 nm leads to “blue shift” of PL spectrum maximum position up to 1.1 eV [5] in comparison with PL spectrum of bulk crystal. Therefore in this paper we will show that the main role in control of PL spectrum and its intensity is QCE with small influence of Ge content. From application point of view in optoelectronics, the investigation of light-emitting diodes based on Si1-xGex structure has been of much interest due to the possibility to change the radiation wavelength in near infrared region of spectrum (~1.5 μm) by varying the concentration of solid solution components [4]. This structure is in good compatibility with Si technology. Therefore this study is focused on formation of optical properties of nanostructures induced by laser radiation on the surface of Si1-xGex/Si hetero-structures.

2. Experiment Keywords: nanohills, SiGe, laser, hetero-epitaxial structure.

1. Introduction Nowadays, nanostructures are the most investigated object in solid-state physics, especially concerning the quantum confinement effect (QCE) in quantum dots (QD) [1], quantum wires [2] and quantum wells [3]. In the case of nanosize structures the energy band diagram of a semiconductor is strongly changed leading to crucial change of properties such as electrical, optical, mechanical and thermal. It is known that in indirected band-gap semiconductors such as Si and Ge radiative electronhole recombination efficiency strongly enhances in nanostructures due to QCE [4]. Moreover, shift of photoluminescence (PL) spectrum toward high energy of spectrum, so called “blue shift”, has been predicted [4] and observed in Ge [5] and Si [6] single crystals. A new flexible possibility is predicted to change the semiconductor basics parameters into QDs of Si1-xGex solid solution both by change of x and QDs diameter. Increase both content of Ge atoms - x, and diameter of QDs leads to the same effective shift of PL spectrum toward low energy of spectrum, so called “red shift”. It has been shown that increase of x from 0.096 to 0.52 leads to shift of maximum position in IR part of PL spectrum toward low energy (0.3 eV) [7]. The same, “red shift” of PL spectrum on 0.7 eV has been observed for nanoparticles with diameter 5-50 nm and x = 0.237 - 0.75 in visible part of spectrum [8]. Authors explain this result by the incorporation of the Ge atoms into 62

Articles

Crystal Si1-xGex alloys were grown on Si(100) wafers by Molecular Beam Epitaxy (MBE). Si1-xGex films were grown by MBE on top of a 150 nm thick Si buffer layer on Si. Alloys containing 30% Ge were used in the experiments. The surface of a Si1-xGex/Si structure was irradiated by 15 ns pulses of a Nd:YAG laser (wavelength 1064 nm, power 1MW). The spot of laser beam of 3 mm diameter was moved by 20 μm steps over the sample surface. The experiments were performed in ambient atmosphere at pressure 1atm, at room temperature (T=20 °C) and 80% humidity. The structural and optical characteristics of Ge nanostructures were studied by atomic force microscope (AFM), photoluminescence, excited by 488 nm radiation of Ar ion laser and Raman scattering (RS), excited by 514.5 nm radiation of an Ar ion laser.

3. Results and Discussion The three-dimensional surface morphology of Si0.7Ge0.3 /Si hetero-epitaxial structure recorded by AFM measurements after irradiation by the Nd:YAG laser at intensities of 7.0 MW/cm2 (a) and 20.0 MW/cm2 (b) is shown in Fig. 1. In Fig. 1(a) are seen the nanohills of the average height of 11 nm formed by laser radiation at the intensity of 7.0 MW/cm2. Similar nanohills of the average height of 27 nm seen in Fig. 1(b) have been obtained by irradiation intensity of 20 MW/cm2. Due to higher irradiation intensity they are more compact in diameter and higher. The three-dimensional surfaces morphology of the same spots as in Fig.1 a) and b) are shown in Fig.1 c) and d).


Journal of Automation, Mobile Robotics & Intelligent Systems

VOLUME 3,

N° 4

2009

Fig. 1. Three-dimensional AFM images of Si0.7Ge0.3/Si sur-faces irradiated by the Nd:YAG laser at intensity a)7 MW/ cm2 and b) 20 MW/cm2 and two-dimensional surface mor-phology of the same spots of structure at intensities: (c) 7.0 MW/cm2 and (d) 20.0 MW/cm2.

Fig. 2. PL spectra of Si0.7Ge0.3/Si hetero-epitaxial structures before and after irradiation by Nd:YAG laser radiation at intensities 2.0 MW/cm2, 7.0 MW/cm2, and 20.0 MW/cm2. PL spectra of the Si1-xGex/Si hetero-epitaxial structures with the maxima at 1.60 -1.72 eV obtained after laser irradiation at intensities of 2.0 MW/cm2, 7.0 MW/cm2 and 20.0 MW/cm2 are shown in Fig. 2. The spectra are unique and unusual for the material, because, depending on Ge

concentration, the band gap of SiGe is between 0.67 eV and 1.12 eV [7]. As seen from Fig. 2, the Si1-xGex structure emits light in the visible range of spectrum and the intensity of PL increases with the intensity of irradiation. After irradiation of the Si1-xGex/Si hetero-epitaxial structure by the laser at intensity of 7.0 MW/cm2 the surface structure begins to look as spots on unwetting material, for example, it looks like water spots on a glass, Fig. 1(c). It means that laser radiation induces segregation of Ge phases at the irradiated surface of the material. This conclusion is in agreement with data from paper [11] where it was shown that Ge phase starts formation at 50% concentration of Ge atoms in SiGe solid solution. According to the Thermogradient effect [9], it is supposed that laser radiation initiates the drift of Ge atoms toward the irradiated surface of the hetero-epitaxial structure. The maximum of the PL band at 1.70 eV is explained by the QCE [10]. Position of the observed PL peak compared with the bulk material shows a significant “blue shift”. The maxima of PL spectra of the Si1-xGex/Si hetero-epitaxial structure slightly shift to higher energy when the laser intensity increases from 2.0 MW/cm2 to 20.0 MW/cm2, which is consistent with the QCE too. Our suggestions, Articles

63


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

concerning to Ge phase formation, are supported by the Raman spectra. After laser irradiation at the intensity of 20.0 MW/cm2 a Raman band at 300 cm-1 appears in the spectrum. This band is attributed to the Ge-Ge vibration and is explained by formation of a new Ge phase [11] in the Si1-xGex/Si hetero-epitaxial structure. The following model is proposed for explanation of dynamics of nanostructures formation.

[4]

[5]

4. Model After irradiation of a Si1-xGex/Si hetero-epitaxial structure by strongly absorbed laser radiation the large gradient of temperature occurs (108 K/m) [9]. It causes the Ge atom drift towards the irradiated surface of the semiconductor. The creation of Ge phase leads to the formation of Ge/Si heterostructure. As a result of difference in Ge and Si lattice constant the mechanically stressed Ge layer occurs on the irradiated surface. The plastic deformation of mechanical stresses in Ge layer takes place with a growth of cone-like nanostructures according to modified Stransky-Krastanov's method.

[6] [7]

[8]

[9]

5. Conclusion 1. Formation of nanohills by laser irradiation of the Si0.7Ge0.3/Si hetero-epitaxial structure is shown to be possible. 2. Photoluminescence spectra of the Si1-xGex/Si hetero-epitaxial structure with nanohills are explained by the Quantum Confinement effect. 3. Formation of a new phase of crystalline Ge nanohills is found on the surface of Si0.7Ge0.3/Si heteroepitaxial structures after laser irradiation at intensities exceeding I=2.0 MW/ cm2.

AUTHORS Artur Medvid'* - Riga Technical University, 14 Azenes Str., Riga, LV-1048, Latvia; Institute of Semiconductor Physics National Academy of Science of Ukraine, 45 Pr. Nauki, 252650, Kyiv-28, Ukraine. E-mail: medvids@latnet.lv. Pavels Onufrijevs, Dainis Grabovskis - Riga Technical University, 14 Azenes Str., Riga, LV-1048, Latvia. Klara Lyutovich, Michael Oehme, Erich Kasper - Universitat Stuttgart, Pfaffenwaldring 47, 70569 Stuttgart, Germany. Igor Dmitruk, Iryna Pundyk - Kyiv National Taras Shevchenko University, Kyiv-03022, Pr. Acad. Glushko, 2-1, Ukraine. Ivan Manak - Belarusian State University, Prosp. Nezavisimosti, 4, 220030, Minsk, Belarus. * Corresponding author

References [1] [2]

[3] 64

Alivisatos A.P., “Semiconductor Clusters, Nanocrystals, and Quantum Dots”, Science, vol. 271, 1996, p. 933. Xia Y., Yang P., Sun Y., Wu Y., Mayers B., Gates B., Yin Y., Kim F., Yan H, “One-Dimensional Nanostructures: Synthesis, Characterization, and Applications”, Advanced Materials, vol.15, no. 5, 2003, pp. 353-389. Fowler A.B., Fang F.F., Howard W.E., Stiles P.J., “Varia-

Articles

[10] [11]

N° 4

2009

tion of the Shubnikov-de Haas amplitudes with ionic scattering in silicon inversion layers”, Phys. Rev. Lett. vol. 16, 1966, p. 901. Emel'yanov A.M., Sobolev N.A., Mel'nikova T.M., Abrosimov N.V., “SiGe Light-Emitting Diodes and Their Characteristics in the Region of Band-to-Band Transitions”, Solid state phenomena, vol.108, 2005, p. 761. Medvid' A., Dmytruk I., Onufrijevs P., Pundyk I., “Quantum confinement effect in nanohills formed on a surface of Ge by laser radiation”, Phys. Stat. Sol. C, vol. 4 (8), 2007, pp. 3066-3069. Chen P.J., Tsai M.Y., Chi C.C., Perng T.P., J. of Nanoscience and Nanotechn, no. 7, 2007, p. 3340. Sun K.W., Sue S.H., Liu C.W., “Visible photoluminescence from Ge quantum dots”, Physica E, vol. 28, 2005, p. 525. Mooney P.M., Jordan-Sweet J.L., Ismail K., et al., “The growth of high-quality SiGe films with an intermediate Si layer”, Appl. Phys. Lett., vol. 67, 1995, p. 2373. Medvid' A., “Redistribution of the Point Defects in Crystalline Lattice of Semiconductor in Nonhomogeneous Temperature Field”, Defects and Diffusion, vol. 210, 2002, p. 89. Efros Al.L., Efros A.L., Phys. and Techn.of Semicond., vol. 16, 1982, pp. 1209-1214. Kamenev B.V., Baribeau J.-M., Lockwood D.J., Tsybekov A., “Optical properties of Stranski-Krastanov grown three-dimensional Si/Si1-xGex nanostructures”, Physica E, vol. 26, 2005, p. 174.


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N째 4

2009

EFFECT OF SUBSTRATE TEMPERATURE ON THE CRYSTAL PROPERTIES OF LIMN2O4 FILMS PREPARED BY RF MAGNETRON SPUTTERING Masaaki Isai, Koichi Nakamura, Takayuki Hosokawa, Takaya Izumichi, Satoshi Sekikawa

Abstract: The LiMn2O4 films for Li secondary batteries have been prepared by a RF magnetron sputtering method. The LiMn2O4 powder was used as a target material. The deposition rate, XRD, and surface morphology were investigated as a function of substrate temperature (Tsub). It was found that the deposition rate had a peak at around the Tsub of 200 째C. This research was also designed for preventing the target material being oxidized during the reactive sputtering process. A quartz tube was inserted between shutter and target. It seems that the quartz tube is effective to prevent oxidation of target material.

tal electrode. This process has complex procedures and induces thick films. This is less attractive in the point of energy density than the deposition process proposed in this paper. In order to solve these problems, various deposition methods have been introduced. [10]-[26] In our early studies, Mn3O4 films have been successfully prepared with preventing the oxidation of evaporant during the recative evaporation process. [16-22] But, unfortunately, charge-discharge curves could not be measured. It seems that this is due to the deficiency of Li ions move between positive and negative electrodes during charge and discharge processes.

Keywords: LiMn2O4 thin film, RF magnetron sputtering, Li secondary batteries.

1. Introduction The lithium (Li) secondary batteries have been used for cell phones, video cameras, etc. The revolution of energy sources for electric vehicles has been started by using fuel cells and Li secondary batteries. These devices require the specific energy density more than 100 Wh/kg and the power density 40 W/kg . [1] Various materials have been used for the positive electrodes of Li secondary batteries. [1], [2] Recently, various manganese oxides have been investigated as a positive electrode material. Manganese is less-toxic and abundant material as compared with Cobalt. We have been focusing especially on the Li-Mn-O defect-spinelstructure (so called defect-spinel-structure). [3], [4] If this structure is adopted as a positive electrode, the operating voltage and theoretical capacity are 3-4 V and 148213 mAh/kg, respectively. [3], [4] The defect-spinel-structure is defined by Mn3O4, Li4Mn5O12, and l-MnO2 triangle in the Li-Mn-O phase diagram as shown in Fig. 1. [4] Our goal is to prepare films with this structure. These defect-spinels are considered to have high structural stability upon insertion and desorption of Li ions. The defect-spinel- structure described above could be obtained through the reaction between Li and l-MnO2 or Mn3O4. [5], [6], [16]-[23]. Figure 2 shows the manganese-spinel-structure of LiMn2O4. There are vacancies (16c) near the 8a sites. Li+ ions can diffuse alternatively, for example, from 8a to16c and then to 8a. [7] All of the materials for the positive electrodes have been prepared by a sintering method.[3]-[9] These powders have to be mixed with some binders and high electric conductivity materials like carbon black to apply the me-

Fig. 1. Li-Mn-O phase diagram.

Fig. 2. Spinel structure of LiMn2O4. In this study, a RF magnetron sputtering method was used to prepare manganese oxide films which involve Li atoms. [25]-[26] The effect of quartz tube on preventing Articles

65


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

(a)

2. Experiment

(b)

3. Results and Discussion The crystal structure of films was identified as LiMn2O4 by X-ray diffraction (XRD) method. Figure 4 (a) an (b) show the dependence of deposition rate on the Tsub at the RF power of 100 and 200 W, respectively. At the RF power of 100 W, the deposition rate is increased as increasing the Tsub as shown Fig. 4(a). The deposition rate is decreased at the Tsub higher than 250°C.

Substrate Temperature (°C)

Deposition Rate (D/s)

LiMn2O4 films were prepared on 0.2 mm-thick aluminum (Al) substrates by a magnetron sputtering apparatus as shown in Fig. 3. The RF powers of 100 and 200W were used in this study. This apparatus has three targets in the vacuum chamber. So, it is easy to prepare multi-layereddevices without breaking vacuum. The deposition condition is shown in Table 1. The deposition time and Ar flow rate were fixed at 60 min and 4sccm, respectively. The substrate temperature (Tsub) was varied from 150 to 300 °C. In order to obtain crystallographic characteristics, Xray diffraction (XRD) measurements were performed with a RIGAKU RINT Ultima II. The film thickness was measured by a gravimetric method.

2009

Deposition Rate (D/s)

oxidation of target material was examined. This research was also designed for preventing the target material being oxidized during the reactive sputtering process. A quartz tube was inserted between shutter and target. It was intended that a quartz tube could act to prevent the oxidation of target material during sputtering process. The purpose of this work is to investigate the effect substrate temperature (Tsub) of the crystal properties of LiMn2O4 films. The effect of quartz tube on preventing the oxidation of target material was also investigated.

N° 4

Substrate Temperature (°C)

Fig. 4(a) and 4(b). Dependence of deposition rate on the substrate temperature (Tsub). The Ar flow rate was fixed at 4sccm. At the RF power of 200 W, the deposition rate is increased as increasing the Tsub as shown Fig. 4(b). The deposition rate is decreased at the Tsub higher than 300°C. Fig. 5(a) and 5(b) show the XRD data of the LiMn2O4 films prepared at the RF power of 100 and 200W, respectively. (a)

(b) Fig. 3. Magnetron sputtering apparatus. Table 1. Deposition condition of LiMn2O4.

vacuum [Pa] deposition time [min] Ar flow rate [sccm] parameter: RF power [W] parameter: Tsub [°C] target

66

Articles

5 60 4 100,200 RT,150,200,250,300 LiMn2O4

Fig. 5(a) and 5(b). The XRD data of the LiMn2O4 films prepared at the RF power of 100 and 200W, respectively The parameter is Tsub.


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

The Fig. 5(a) and (b) show that the (111) peaks were disappeared at the Tsub higher than 250°C. It seemed that the XRD peaks were disappeared due to the deliquescence of LiMn2O4 species during keeping films in air after deposition. They must be kept in vacuum until XRD measurements. This research was also designed for preventing the target material being oxidized during the reactive sputtering process. [25] A quartz tube was inserted between shutter and target. Figure 6 shows the effect of the quartz tube on the deposition rate. In this case, the oxygen gas was introduced. The Ar to O2 flow ratio was 3:1. The total flow rate was fixed at 4 sccm. The RF power was 100 W and the deposition time was varied from 30 to 90 min. The deposition rate was increased as increasing the deposition time in the case of installing the quartz tube. It seems that the quartz tube is effective to prevent oxidation of target material. A detailed study are now in progress.

References [1]

[2]

[3]

[4]

[6]

[7]

[8]

Fig. 6. The effect of the quartz tube on the deposition rate. In this case, the oxygen gas was introduced.

The LiMn2O4 films were successfully prepared by a magnetron sputtering method. A quartz tube was introduced to prevent the oxidation of target material. It was found that the deposition rate was decreased at the Tsub of 300°C. A quartz tube was inserted between shutter and target. It seems that the quartz tube is effective to prevent oxidation of target material.

[9]

[10]

[11]

ACKNOWLEDGMENTS A part of this study has been supported by Mikiya Science and Technology Promotion Society. The support from the Grant-inAid for Scientific research (C) in the Ministry of Education, Culture, Sports, Science and Technology in Japan was also quite helpful. Several parts of vacuum apparatus have been made by K. Sahara, M. Iwasawa, A. Isogai, T. Matsuno, T. Okamoto and T. Kamio in the Manufacturing Center of Shizuoka University.

[12]

[13]

[14]

AUTHORS Masaaki Isai* - Faculty of Engineering, Shizuoka University, Graduate School of Engineering, Shizuoka University 3-5-1 Johoku, Naka, Hamamatsu, Japan 432-8561. E-mail: temisai@ipc.shizuoka.ac.jp. Koichi Nakamura, Takayuki Hosokawa - Graduate School of Engineering, Shizuoka University 3-5-1 Johoku, Naka, Hamamatsu, Japan 432-8561. Takaya Izumichi , Satoshi Sekikawa - Faculty of Engine-

2009

ering. Shizuoka University 3-5-1 Johoku, Naka, Hamamatsu, Japan 432-8561. * Corresponding author

[5]

4. Conclusion

N° 4

[15]

[16]

Desilvestro J., Haas O., “Metal oxide cathode materials for electrochemical energy storage: A Review”, J. Electrochem. Soc., vol. 137, 5C ,1990. Koksbang R., Barker J., Shi H., Saidi M.Y., “Cathode materials for lithium rocking chair batteries”, Solid State Ionics, vol. 84, issue 1,1996. Thackeray M.M., de Kock A., Rossouw M.H., Liles D., Bittihn R., Hoge D., “Spinel electrodes for the Li-Mn-O system for rechargeable lithium batteries”, J. Electrochem. Soc., vol. 139, 1992, p. 363 f. Gummow R.J., de Kock A., Thackeray M.M., “Improved capacity retention in rechargeable 4V lithium/lithium manganese oxide (spinel) cells”, Solid State Ionics, vol. 69, issue 59 ,1994. Hunter J.C., “Preparation of a new crystal form of manganese dioxides: l-MnO2”, J. Solid State. Chem., vol. 39, 1981, p. 142 f. Thackeray M.M., Johnson P.J., de Picciotto L.A., Bruce P.G., Goodenough J.B. “Electrochemical extraction of lithium from LiMn2O4”, Mat. Res. Bull., vol. 19, no. 19, 1984, pp. 179 f. Thackeray M.M., David W.I.F., Bruce P.G., Goodenough J.B., “Lithium insertion into manganese spinels”, Mater. Res. Bull., vol.18, 1983, p. 461 f. David W.I.F., Thackeray M.M., De Picciotto L.A., Goodenough J.B., “Structure refinement of the spinel-related phases Li2Mn4O4 and Li0.2Mn2O4”, J. Solid State Chem., vol. 67, 1987, p. 316 f. Xia Y., Yoshio M., ”An investigation of lithium ion insertion into spinel structure Li-Mn-O compouds”, J. Electrochem. Soc., vol. 143, 1996, p. 825 f. Bando Y., Horii S., Tanaka T., “Reactive condensation and magnetic properties of ion oxide films”, Jpn. J. Appl. Phys., vol. 17, 1978, 1037 f. Striebel K.A., Deng C.Z., Wen S.J., Cairns E.J., “Electrochemical behavior of LiMn2O4 and LiCoO2 thin films produced with pulsed laser deposition”, J. Electrochem. Soc., vol. 143, 1996, p. 1821 f. Shokoohi F.K., Tarascon J.M., Wilkens B.J., Guyomard D., Chang C.C., “Low temperature LiMn2O4 spinel films for secondary lithium batteries”, J. Electrochem. Soc., vol. 139, 1992, p. 1845 f. Rougier A., Striebel K.A., Wen S.J., Cairns E.J., “Cyclic voltammetry of pulsed laser deposited LixMn2O4 thin films”, J. Electrochem. Soc., vol. 145, 1998, p. 2975 f. Liu P., Zhang J.G., Turner J.A., Tracy C.E., Benson D.K., “Lithium-manganese-oxide thin-film cathodes prepared by plasma-enhanced chemical vapor deposition”, J. Electrochem. Soc., vol. 146, 1999, p. 2001 f. Isai M., Yamaguchi K., Iyoda H., Fujiyasu H., Ito Y., “Oxygen gettering effect during the reactive evaporation of manganese films”, J. Mater. Res., vol. 14, 1999, p. 1653 f. Isai M., Ichikawa H., Shimada T., Morimoto K., Fujiyasu H., Ito Y., “Priority of the Mn deposition rate in the reactive evaporation conditions”, Jpn. J. Appl. Phys., vol. Articles

67


Journal of Automation, Mobile Robotics & Intelligent Systems

[17]

[18]

[19]

[20]

[21]

[22]

[23]

[24]

[25]

[26]

68

39, 2000, p. 6676 f. Isai M., Ichikawa H., Takahashi H., Fujiyasu H., Ito Y., “An idea of overcoming the oxidation of Mn metal in the crucible during the reactive evaporation process”, Electrochemistry, vol. 68, 2000, p. 963 f. Isai M., Shimada T., Matsui T., Fujiyasu H., “A new technique keeping off the Mn evaporant from oxygen atmosphere during the reactive evaporation process”, Jpn. J. Appl. Phys., vol. 40, 2001 , p. 5069 f. Isai M., Fujiyasu H., “Prevention against oxidation of Mn evaporant during reactive evaporation process”, Jpn. J. Appl. Phys., vol. 40, 2001, p. 6552 f. Isai M., Nagashio Y., Tatei T., Fujiyasu H., “Stabilization of deposition rate of Mn oxide films by using SUS cell”, Trans. Mater. Res. Soc. Jpn., vol. 30, 2005, p. 1031 f. Isai M. “Research on reactively evaporated Mn oxide films”, Trans. Mater. Res. Soc. Jpn., vol. 31, 2006, p. 1025 f. Isai M., “Preparation of manganese oxide films for Li secondary batteries”, Trans. Mater. Res. Soc. Jpn., vol. 31, 2006, p. 1037 f. Isai M., Chonan Y., Tojo Y., “Preparation of lithium manganese oxide films for Li secondary batteries”, Trans. Mater. Res. Soc. Jpn., vol. 32, 2007, p. 1203 f. Isai M., Chonan Y., Tojo Y., “Preparation of lithium manganese oxide films for Li secondary batteries”, Trans. Mater. Res. Soc. Jpn., vol. 33, 2009, p. 1313 f. Hosokawa T., Nakamura K., Hosoe S., Sakai S., Isai M. “Effect of doughnut-type plate on crystal properties of LiMn2O4 films prepared by RF magnetron sputtering”. In: 7th Int. Na. Con. on Glob. Res. and Edu. In New Methods in Edu., 15th-18th September 2008, Pecs, Hungary, p. 392 f. Isai M., Nakamura K., Hosokawa T., Sakai S., Hosoe S. “Preparation of LiMn2O4 films by RF magnetron sputtering method”, Trans. Mater. Res. Soc. Jpn., vol. 34, 2009, p. 355f.

Articles

VOLUME 3,

N° 4

2009


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

DYE-SENSITIZED SOLAR CELLS BASED ON SNO2 NANOROD AND SURFACE TREATMENT WITH MG(II) FILM Jingyi Bai, Kenji Murakami

Abstract: Homocentrically grown SnO2 nanorods were synthesized by a hydrothermal method, whose diameter and length are around 20 nm and 100-200 nm, respectively. The photovoltaic properties of dye-sensitized SnO2 and MgO-treated SnO2 films were investigated. A light-to-electricity conversion yield of 0.8% was achieved by applying the nanorods as a thin film layer for the dye-sensitized solar cells. Treatment with MgO on SnO2 surface improves the photovoltage, fill factor and cell efficiency owing to preventing a recombination of the electrons injected into SnO2 with acceptors in the electrolyte. Keywords: dye-sensitized solar cell, tin dioxide film, nanorod, spray pyrolysis deposition, surface treatment, magnesium oxide.

1. Introduction Dye-sensitized solar cells (DSCs) are based on a sensitization of mesoporous, nanocrystalline metal oxide films to visible light by an adsorption of the molecular dyes. Most of DSCs utilize nanoporous electrodes made from TiO2 [1]-[6]. However, some applications require more positive conduction band potential than that of TiO2. Tin oxide (SnO2) is an attractive alternative to TiO2 due to its higher electronic conductivity and electron mobility [7]. Such faster electron transport dynamics could minimize the interfacial charge recombination losses in DSCs and improve the cell performances. Hydrothermal method, one of the solvothermal methods, enables syntheses of new materials with different morphologies. In particular, the method is useful to produce the finely dispersed, nanocrystalline materials. In this paper, the synthesis of single crystalline SnO2 nanorods is described through the hydrothermal route with SnCl4 as a precursor. The spray pyrolysis deposition technique was applied to form the metal oxide films for DSCs. The films were analyzed by using the x-ray diffraction (XRD), field emission scanning electron microscopy (FE-SEM). Photovoltaic performances of the assembled DSCs were also investigated. Furthermore, an effect of surface treatment of SnO2 with a MgO coating is discussed on the cell performances.

2. Experimental 2.1. Preparation of dye-sensitized solar cells NaOH (10 mol/L) was slowly added into SnCl4 (1 mol/L) solution drop by drop followed by stirring in an alcohol-water (1:1) mixture. Then, the mixing solution

was transferred into the teflon-lined autoclave. The autoclave was heated to 200°C for 24 h. The product including SnO2 nanorods was centrifuged and washed with the distilled water and alcohol, and then dried at 60°C in an oven. The product was mixed with acetylacetone, non-ionic surfactant and acetic acid followed by agitation ultrasonically for 60 m. The suspension was sprayed pneumatically onto a fluorine doped SnO2 -coated glass (FTO glass) substrate heated at 130°C and it was sintered in air at 500°C for 1 h. The MgO- treated SnO2 was prepared similarly to the procedure for SnO2 nanorods except that a magnesium acetate aqueous was added into the SnO2 nanorod precipitate and kept at room temperature for 12 h. After the FTO glass plate with SnO2 film was soaked into the solution of the N719 (n-Bu4N) 2 [Ru(Hdcbpy)2 (NCS)2] dye in acetonitrile and tert-butanol (1:1) for 12 h, it was assembled into a sandwich-type cell, where the counter electrode was a platinumsputtered FTO glass. A drop of electrolyte solution (0.M LiI, 0.05 M I2, 0.5M tert-butyl pyridine, 0.6M dimethylpropylimidazolium iodide in methoxyacetonitrile) was inserted into the assembled cell. 2.2. Characterization The morphology of the nanoporous SnO2 films and nanorod materials were observed by using the FE-SEM (JOEL, JSM-6320F). The XRD (Rigaku, RINT-2200) profiles enabled a characterization for crystalline phases of nanoporous SnO2 powder. Energy-dispersive X-ray spectroscopy (EDS equipped with FE-SEM) measurements were carried out to determine a component of the nanomaterials. I–V characteristics of the cells at AM 1.5 (1000 2 W/m ) were measured by using the calibrated solar cell evaluation system (Jasco, CEP-25BX).

3. Results and discussion 3.1. SnO2 film It is found from the XRD profiles that only the diffraction peaks of tetragonal rutile SnO2 are detected for all the sprayed films. Figure 1 shows the FE-SEM images of sprayed films. As shown in the figure, the surface is very rough and such morphology is originated from the homocentrically grown nanorods, whose diameter and length are around 20 nm and 100-200 nm, respectively. As a result, the film is mainly composed of disordered networks of the SnO2 nanorods. Articles

69


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

A

B

N° 4

2009

SnO2 films, where JSC is a short circuit current, VOC a open circuit voltage, FF a fill factor and h energy conversion efficiency. In order to consider an effect of the surface treatment of SnO2 with MgO, the performances of DSCs made of MgO-treated SnO2 films with different Mg/Sn ratios are also listed in the table. Although the performances of DSC with SnO2 nanorods-based film are very poor, the surface treatment of MgO improves the efficiency greatly from 0.8 to 1.63%. Very low JSC for the SnO2 nanorods-based film may be attributed to the smaller surface area available for dye anchoring. Increasing the amount of Mg tends to increase VOC and FF, while both JSC and h decrease. The surface treatment with MgO seems to form a barrier on the SnO2 surface preventing a recombination between the SnO2 and the acceptors in electrolyte. The trace amount of MgO, therefore, is very effective to improve the performances as shown in the table.

Fig. 1. (A)FE-SEM photographs of the sprayed SnO2 film (B) Magnified image of (A). 3.2. Mg-modified SnO2 film It is concluded from the XRD profiles that the surface treatment of SnO2 with MgO has no significant influence on a grain size of nanoparticles. Surface morphology of the Mg-modified SnO2 film is also similar to that of the non-treatment surface as shown in Fig. 2. The EDS measurements cannot reveal an existence of Mg due to a very small amount of magnesium composition.

Fig. 2. FE-SEM photograph of the sprayed Mg-modified SnO2 film. 3.3. Photovoltaic performances Typical I–V characteristics of the fabricated solar cells are illustrated in Fig. 3. An open circuit voltage and a fill factor of the cell are improved greatly by the surface treatment of SnO2 with MgO. Table 1 summarizes the photovoltaic performances of the DSCs made of the sprayed 70

Articles

Fig. 3. I-V characteristics of the dye-sensitized SnO2 nanorod and Mg-modified SnO2 nanorod solar cells. Table 1. Photovoltaic performances of the SnO2-based DSCs with and without Mg-modification. -2

Sample

JSC (mA×cm ) VOC (V)

FF

h (%)

SnO2 Mg/Sn=0.01 Mg/Sn=0.04 Mg/Sn=0.08

8.50 7.23 3.38 2.75

0.40 0.61 0.67 0.70

0.81 1.63 1.12 1.07

0.24 0.37 0.49 0.56

AUTHORS Jingyi Bai - Research Institute of Electronics, Shizuoka University, Naka-ku Hamamatsu, 432-8011, Japan. Kenji Murakami* - Research Institute of Electronics, Shizuoka University, Naka-ku Hamamatsu, 432-8011, Japan. E-mail: rskmura@ipc.shizuoka.ac.jp. *Corresponding author

References [1]

[2]

O'Regan B., Grätzel M., “A low-cost, high-efficiency solar cell based on dye-sensitized colloidal TiOV2 films”, Nature, vol. 353, no. 6346, 1991, pp. 737-740. Hagfeldt A., Didriksson B. , Palmqvist T., Lindstrom H., Sodergren S., Rensmo H., Lindquist S.-E., “Verification


Journal of Automation, Mobile Robotics & Intelligent Systems

[3]

[4]

[5]

[6]

[7]

VOLUME 3,

N° 4

2009

of high efficiencies for the Grätzel-cell. A 7% efficient solar cell based on dye-sensitized colloidal TiO2 films”, Sol. Energy Mater. Sol. Cells, vol. 31, no. 4, 1994, pp. 481-488. Nazeeruddin M.K., Pechy P., Grätzel M., “Efficient panchromatic sensitization of nanocrystalline TiO2 films by a black dye based on trithiocianato-ruthenium complex”, Chem. Commun., vol. 18, 1997, pp. 1705-1707. Cao F., Oskam G., Searson P.C., “A Solid State, Dye Sensitized Photoelectrochemical Cell”, J. Phys. Chem., vol. 99, no. 47, 1995, pp. 17071-17073. Gerfin T., Grätzel M., Walder L., “Molecular and Supramolecular Surface Modification of Nanocrystalline TiO2 Films: Charge-Seperating and Charge-Injecting Devices”, Prog. Inorgan. Chem., vol. 44, 1996, p.345. Park N.G., Schlichthorl G., van de Lagemaat J., Cheong H.M., Mascarenhas A., Frank A.J., “Dye-Sensitized TiO2 Solar Cells: Structure and Photoelectrochemical Characterization of Nanocrystalline Electrodes Formed from the Hydrolysis of TiCl4”, J. Phys. Chem. B, vol. 103, no. 17, 1999, pp. 3308-3314. Jarzebski Z., Marton P., “Physical Properties of SnO2 Materials”, J. Electrochem. Soc., vol. 123,S 1976, pp. 299C-310C.

Articles

71


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

ANALYSIS OF CURRENT NOISE IN MOSFET-BASED CHARGE-TRANSFER DEVICE Hiroshi Inokawa, Vipul Singh, Hiroaki Satoh

Abstract: Low-frequency current noise in the MOSFET-based charge-transfer devices was evaluated at room and cryogenic temperatures. Excessive noise, whose power was 25-50 times larger than that of room temperature, was observed at 20 K, and this hindered the accurate transfer of charge. Interestingly, the noise power was proportional to the square of the pulse frequency that drove the gates in these devices, and an expression for the noise was proposed to correlate it with the frequency, gate capacitance and fluctuation of the threshold voltage at the gate. Keywords: low-frequency noise, MOSFET, charge-transfer device.

1. Introduction The charge-transfer device consisting of two serially connected MOSFETs is very important as an element of switched capacitor circuit [1] and as a single-electron turnstile or pump [2]. Since the low-frequency noise in the MOSFET becomes increasingly conspicuous as its size is reduced, it is worthwhile to analyze the noise behavior of the charge-transfer device with small dimensions. In this paper, Low-frequency current noise in the charge-transfer devices fabricated by 65-nm bulk CMOS process is reported in comparison with the ordinary noise of a MOSFET in direct-current (DC) operation. The noise in the transfer current at a low temperature is evaluated, and is found to show anomaly.

Fig. 2. Charge-transfer (Id -Vds) characteristics at 295 K. L=70 nm, W=110 nm, S=170 nm. Figure 1(a) shows the planar view of the fabricated device. Two lower gates cross the Si active area between n+ - doped source/drain regions. An upper gate made of metal covers the active area to electrically induce an inversion layer that serves as a source/drain extension. The gate length L, channel width W and gate spacing S were 60-80 nm, 100-120 nm and 160-180 nm, respectively, and the gate oxide thickness was 5 nm. For the charge-transfer operation, as shown in Fig. 1(b), pulses with a phase delay of 180° were applied to the Vlg1 and Vlg2 to alternately turn on and off the respective channels. The current was measured with Agilent 4156C, and the data acquired in 600 s with the sampling period of 0.6 s were Fourier transformed to obtain the noise power spectrum.

2. Experiments 3. Results and Discussions The charge transferred in one cycle of the pulse is expressed as [3]

a)

Q = CS'Vd + Clg1(Vtho1 - Vlg1L) + Clg2(Vlg2L - Vtho2)

where CS'=Clg1=Cug=Cb=Cs=Cd, Vlg1L and Vlg2L are the lower levels of the Vlg1 and Vlg2 pulses, respectively, and Vtho1 and Vtho2 are the threshold voltages of FET1 and 2, respectively, when source/drain is not biased. Note that the noise in the form of the fluctuation in Vtho1 and Vtho2 led to the fluctuation in Q, and was reflected to the noise in the transfer current Id=Qf, where f is the pulse frequency. Specifically, the noise in the transfer current can be expressed as

b)

dId = Ö2dVthClg f Fig. 1(a) Device structure, and (b) Equivalent circuit when operating as a charge-transfer device. 72

(1)

Articles

(2)

where it is assumed that the noises in the Vtho1 and Vtho2


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

are independent and have the common standard deviation dVth, and Clg1=Clg2=Clg. On the other hand, the noise in the DC current in the subthreshold region can be related to the Vth fluctuation by dId = Id (10dVth/S-1) Âť Id (ln10)dVth /S

N° 4

2009

to dId (see eqs. (2) and (3)), and the difference in dVth itself. Since the lower gates are frequently switched between inversion and accumulation states in the charge transfer operation, the emission and capture times for the traps are largely reduced, and dVth in the low-frequency range may become smaller [4].

(3)

where S is the subthreshold swing in the unit of volt per decade. Figure 2 shows the charge transfer (Id-Vds) characteristics at 295 K for various pulse frequencies. Since the slope of the curve (i.e. conductance) was proportional to the frequency, CS' was found to be 4.0 aF. The current noise spectra have been evaluated at a fixed Vds=0.2 V for various current levels (i.e. various frequencies) as shown in Fig. 3(a). The current noise spectra for DC operation were also obtained with the same device by applying high voltage (2 V) to one lower gate and adjusting the other to attain similar current levels to that of charge transfer measurement (Fig. 3(b)). The spectra for charge-transfer operation showed larger slopes, whereas those for DC operation showed gentler slopes. This difference may be related to the charging and discharging dynamics of the traps modified by the gate pulse in the case of the charge-transfer operation, as is expected from the cycled-gate experiment [4].

Fig. 4. Total noise power vs. current for DC and charge-transfer operations at 295 K. L=70 nm, W=110 nm, S=170 nm. The data points corre-sponding to the spectra in Fig. 3 are labeled with a, b, c, etc. a)

a)

b)

b)

Fig. 3. Noise power spectra in (a) charge-transfer, and (b) DC operations at 295 K. L=70 nm, W=110 nm, S=170 nm. The total noise powers were obtained by integrating the entire spectrum above 5 mHz, and have been plotted with respect to the average current in Fig. 4. The noise power in the transfer current is comparable to or a little smaller than that of the DC current. This can be attributed to the difference in the conversion factors from dVth

Fig. 5. Charge-transfer (Id-Vds) characteristics at 20 K. (a) L=60 nm, W=100 nm, S=160 nm, (b) L=80 nm, W=120 nm, S=180 nm. For noise characterization, plateau areas in (a) 0.3~0.45 V, and (b) -0.1~0.1 V are analyzed by linear regression.

Figure 5 shows the charge-transfer characteristics at low temperature, where current quantization can be expected as a result of Coulomb blockade [2]. However, the current plateaus do not perfectly match the multiple of ef and large noises were found to be superimposed. For the simple characterization of the noises, plateau areas were chosen, straight lines were drawn by least-square method, Articles

73


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

and the square of the deviation from the straight lines were averaged to obtain the noise power, plotted as a function of current in Fig. 6. Fifty times larger noise was observed at 20 K. This excessive noise is estimated to come from the larger dVth, since other terms in eq. (2) are rather independent of temperature. The cause of the large Vth noise is not clear at this moment, but the emission and capture times for the traps may become longer at a low temperature, leading to the increase in the low-frequency component of the noise (see eq. (15) in [5]).

Fig. 6. Total noise power vs. current for charge-transfer operation at 20 and 295 K. Excess noise is observed at low temperature.

N째 4

2009

son between DC and charge-transfer operations, as in Figs. 4 and 6, is still acceptable as long as the device structure and the operation condition are symmetric for lower gate 1 and 2, because the current is proportional to f without an offset in such a case. From the slope of the transfer current in Fig. 8, the Clg1 is calculated to be 2.2 aF based on eq. (1), and this leads to the dVth as large as 7.6 mV at 20 K.

Fig. 8. Noise power and transfer current with respect to the lower level of the Vlg1 pulse. Transfer current can be changed by the asymmetry in Vlg1 and Vlg2 pulses, but the noise power is kept constant as long as the pulse frequency is the same. From the slope of the transfer current, Clg1 can be calculated to be 2.2 aF.

4. Conclusions

Fig. 7. Total noise power vs. pulse frequency for the chargetransfer operation at 20 and 300 K. These data are obtained from another device with L=60 nm, W=100 nm and S=170 nm, but the same tendency has been ob-served generally.

Low-frequency current noise in the MOSFET-based charge-transfer device has been evaluated. It was found that, at 295 K, the noise power in the transfer current was comparable to or a little smaller than that in the DC current. At 20 K, on the contrary, 25-50 times larger noise power was found in the transfer current. In accordance with the proposed expression, the noise power in the transfer current was demonstrated to be proportional to the square of the pulse frequency, and in some device the threshold voltage fluctuation for the lower gates was estimated to be as large as 7.6 mV at 20 K. The reduction of the traps, related to the threshold voltage fluctuation is especially significant for accurate single-electron transfer at low temperatures. ACKNOWLEDGMENTS

In order to see the behavior of the noise precisely, dependence on the pulse frequency at a fixed drain bias was evaluated as shown in Fig. 7. In this case, a single device is measured under a consistent condition both at 20 and 300 K. The noise power is proportional to the square of the gate pulse frequency at both temperatures, supporting eq. (2). The noise at 20 K is 25 times larger than that at 300 K, and the excessive noise at the low temperature is confirmed. The noise power was also evaluated when the Vlg1 and Vlg2 pulses became asymmetric. As indicated by eq. (1), the transfer current can be reduced by increasing the Vlg1L while keeping the Vlg2L constant. It is remarkable that the noise power does not change even when the current changes, indicating again the validity of eq. (2). Note that the use of the current as the basis of compari74

Articles

The authors would like to express sincere thanks to Akira Fujiwara, Yukinori Ono, Katsuhiko Nishiguchi for fruitful discussions. They are deeply indebted to Keisaku Yamada of Waseda University, Toyohiro Chikyo of National Institute of Materials and Science, Tetsuo Endoh of Tohoku University, Hideo Yoshino and Shigeru Fujisawa of Semiconductor Leading Edge Technologies, Inc. for their cooperation in the device fabrication. This work was supported by MEXT KAKENHI (19310093).

AUTHORS Hiroshi Inokawa*, Vipul Singh, Hiroaki Satoh - Research Institute of Electronics, Shizuoka University, 3-51 Johoku, Naka-ku, Hamamatsu 432-8011, Japan. E-mail: inokawa06@rie.shizuoka.ac.jp. * Corresponding author


Journal of Automation, Mobile Robotics & Intelligent Systems

VOLUME 3,

N° 4

2009

References [1] [2]

[3]

[4]

[5]

Razavi B., Design of Analog CMOS Integrated Circuits, McGraw-Hill: New York, 2001, chap. 7 and 12. Fujiwara A., et al., "Current quantization due to singleelectron transfer in Si-wire charge-coupled devices ", Appl. Phys. Lett., vol. 84, no. 8, 2004, pp. 1323-1325. Inokawa H., et al., "Capacitive Parameter Extraction for Nanometer-Size Field-Effect Transistors," On: Int. Conf. Solid State Devices and Materials (SSDM2007), Tsukuba, Japan, pp. 874-875. Dierickx B., Simoen E., "The decrease of â&#x20AC;?random telegraph signal'' noise in metal-oxide-semiconductor field -effect transistors when cycled from inversion to accumulation", J. Appl. Phys., vol. 71, no. 4, 1992, pp. 2028-2029. Machlup S., "Noise in Semiconductors: Spec-trum of a Two-Parameter Random Signal", J. Appl. Phs., vol. 25, no. 3, 1954, pp. 341-343.

Articles

75


Journal of Automation, Mobile Robotics & Intelligent Systems

VOLUME 3,

N° 4

2009

EPILEPTIC BURST MEASUREMENT USING MICROELECTRODES EQUIPPED ON A CRYOGENIC MICROPROBE FOR MINIMALLY INVASIVE BRAIN SURGERY OF INTRACTABLE EPILEPSY TREATMENT Toshitaka Yamakawa, Takeshi Yamakawa, Satoru Ishizuka, Zimin Lev Grigorievich, Michiyasu Suzuki, Masami Fujii

Abstract: A microprobing system, which has the functions of measuring the intracranial EEG(IC-EEG) and freezing brain tissue, is proposed for the minimally invasive brain cryogenic surgery of intractable epilepsy treatment. Two f76 μm platinum electrodes were equipped on a f0.8 mm cryogenic probe. Epileptic burst, which was evoked on a brain sample of a rat, was measured by the electrodes. The freezing function was confirmed with the experiments with sliced hippocampus samples of a rat. Keywords: epilepsy, cryogenic surgery, minimally invasive, microelectrodes, CMOS IC.

1. Introduction About 50 million people worldwide have epilepsy at any one time. The lifetime prevalence of epilepsy (i.e. the number of people presently in the world who have epilepsy now or have had it in the past, or will experience it in the future) is approximately 100 million people. The mean prevalence of active epilepsy, which causes continuing seizures and needs treatments, is 0.82 % of the general population around the world [1]. In 70% of cases of the epilepsy patients’ seizures can be controlled with medications, even though it cannot be cured. However, up to 30% of the patients do not respond to the medications even with the best (strongest) available medicines. In that case, surgeries are applied to remove the brain tissue (neocortex or hippocampus, usually) of the epileptogenic focus. The removing area of the brain tissue is determined by the detected epileptogenic focus. However, the detection accuracy is up to several centimeters. Thus the risk of the side and after effects cannot be avoided. In this paper, a cryogenic microprobe, microelectrodes, and intracranial EEG(IC-EEG) sensing CMOS interfaces are proposed to be compatible with the minimally invasive cryogenic surgery of intractable epilepsy treatment. The proposed microprobe system has the functions of measuring the IC-EEG and freezing the brain tissue around the microprobe’s tip for few millimeters.

2. Cryogenic Microprobe System Figure 1 shows the block diagram of the proposed minimally invasive cryogenic microprobe system. This system mainly consists of 4 blocks: the cryogenic microprobe, the EEG instrumentation amplifier, the brain stimulation current source, and the thermocouple amplifier.

76

Articles

Fig. 1. Proposed cryogenic microprobe system. The system operation is briefly explained below. 1) The cryogenic microprobe is inserted to the predetected epileptogenic focus, which is the breakout area of the epileptic burst. 2) In order to confirm that the probe is inserted to the correct location, IC-EEG is measured by the EEG amplifier with SW1 and SW2 opened. If the epileptic burst is observed, the probe-inserted position is determined to be an epileptogenic focus. 3) The refrigerant gas flows into the probe through the inner pipe, then the temperature of the probe’s tip falls. Hence the brain tissue around the tip is frozen. 4) The brain stimulation current is applied in order to confirm that the brain tissue of the epileptogenic focus is adequately neutralized. When the epileptic burst does not occur after the stimulation, the epileptogenic focus no longer exists. The details of the EEG amplifier and the cryogenic microprobe are explained in the following subsections. 2.1. Cryogenic Microprobe The multifunctional microprobe and its cross sectional view is shown in Fig. 2. The needle pipe consists of the microelectrodes and coaxial pipes. The microelectrodes are made of platinum and acts as stimulating/measuring electrodes (ELECTRODE 1, ELECTRODE 2). The outer pipe is made of stainless steel SUS304 and acts as a negative thermocouple electrode (ELECTRODE 3). The inner pipe is made of Kovar (Fe54%-Ni29%-Co17%) and acts as a positive thermocouple electrode (ELECTRODE 4) and a refrigerant guide to evaporate it. Thus the tip of the needle pipe acts as rapid cooling/freezing probe and also acts as temperature measuring probe. The diameter of the tip is 0.8 mm. The refrigerant cylinder is filled with the


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

refrigerant gas, and is connected to the inner pipe of the needle pipe. Here, HFC(hidrofluoro-carbon)-152a is used as the refrigerant gas, since it has relatively high boiling point (-24 °C). The freezing ability of the microprobe has already been confirmed in the experiments with saline solidified by gelatin [2].

Fig. 2. The proposed cryogenic microprobe and its cross sectional view.

N° 4

2009

3. Results 3.1. Epileptic Burst Measurement Using the Microelectrodes Typical market-available microelectrodes are not suitable for this application because of the sharp tip shape. On the other hand, even though several types of microelectrode array mounted on a flexible film have been developed (e.g. [5]), they are not suited to attach 0.8 mmdiameter rounded surface of the cryogenic probe. The proposed microelectrodes were equipped on the surface of the cryogenic probe by using Teflon-coated platinum wires, which have the diameter of 76 μm. Heat-shrinkable tubes and a biocompatible bond were used to fix the electrodes on the surface of the cryogenic microprobe. The tip of the electrodes was polished with sanding papers (#600, 1000, 2000) and with a diamond polisher of #8000, in order to strictly flatten the contacting surface. The microscopic photograph of the probe’s tip is shown in Fig. 4.

2.2. EEG Amplifier Two instrumentation amplifiers (IA1 and IA2) are used in the proposed system: the EEG amplifier and the thermocouple amplifier. The same circuit design is adopted for both instrumentation amplifiers except the value of the gain-controlling resister RG. Figure 3 shows the circuit diagram of the instrumentation amplifier.

Fig. 4. Microscopic photograph of the probe's tip. For the measurement of epileptic bursts, epilepsy was formed on the CA3 area of a male Wister rat’s sliced hippocampus sample by intermittent pulse current stimulation through Electrode 1. The epileptic discharge was observed by using the electrodes, as shown in Fig. 6.

Fig. 3. The circuit diagram of the instrumentation amplifier. The instrumentation amplifier adopts the 3-Op-amp architecture, which is commonly used for the high CMR (Common Mode Rejection) amplifiers [3]. The output voltage Vout in Fig. 3 is given by the following equation (1) when R1=R2, R3=R4, and R5=R6 are satisfied. (1) Fig. 5. Epileptic burst measured by the microelectrodes. In the depth EEG measurement, 100~300μV of the epileptic burst can be measured by depth electrodes (the electrodes are equipped on a 1mm-diameter catheter, electrodes distance: 5 mm ~10 mm) [4]. The gain is set to 60~80dB in the proposed EEG amplifier. The thermocouple amplifier is designed to have a gain of 80dB, since the Seebeck coefficient of the thermocouple, which appears at the probe’s tip, was 47μV/K. For the proposed instrumentation amplifiers, a bipolar supply voltage is necessary. The element values used in the circuits are shown in Sect. 3.

3.2. Cooling and Freezing of the Epileptogenic Focus Focal brain tissue cooling and freezing were performed by the cryogenic probe. The probe’s tip was vertically attached to the epileptogenic focus evoked on CA3 area. Figure 7, Fig. 8, and Fig. 9 show the epileptic spikes measured before cooling, after cooling of 30 seconds, and after cooling of 150 seconds, respectively. When after cooling of 4 minutes, the epileptic spikes disappeared. These results suggest that the proposed multifunctional microprobe might be useful for surgical treatment of epilepsy. Articles

77


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

±1.5 V is applied. The circuit in Fig. 10 is used as the Op amp (A1, A2, and A3) in the circuit in Fig. 3. The bias voltage Vbias of the Op amp is set to 0.5 V. The epileptic burst measured by the proposed microelectrodes (the upper part of Fig. 11) was used as a differential input of the circuit in Fig. 3. The epileptic burst was successfully amplified with the gain of 72dB (x4000), as shown in the lower part of Fig. 11.

Fig. 6. Epileptic spikes measured before cooling.

Fig. 10. The differential input (upper) and the simulated output signal amplified by the circuit in Fig. 3 (lower).

4. Conclusion Fig. 7. Epileptic spikes measured after cooling of 50 seconds.

A system for intracranial EEG measurement and for minimally invasive cryosurgery of intractable epilepsy treatment was proposed in this paper. The epileptic burst was measured by the proposed microelectrodes in the experiments using hippocampus samples of a rat. Experimental results showed that the epileptic spikes were suppressed by cooling performed by the microprobe. The operation of the EEG amplifier was confirmed by the HSPICE simulations. ACKNOWLEDGMENTS This research is partially supported by the Japanese Ministry of Education, Science, Sports and Culture, Grant-in-Aid for Specially Promoted Research, No.20001008.

Fig. 8. Epileptic spikes measured after cooling of 150 secondes. 3.3. HSPICE CMOS Circuit Simulation Results of the EEG amplifier

AUTHORS Toshitaka Yamakawa* - Department of Electrical and Electronic Eng., Faculty of Eng., Shizuoka University, 3-5-1, Jyohoku, Naka-ku, Hamamatsu, 432-8561, Japan. E-mail: ttyamak@ipc.shizuoka.ac.jp. Takeshi Yamakawa, Satoru Ishizuka, Zimin Lev Grigorievich - Department of Brain Science and Engineering, Graduate School of Life Science and Systems Eng., Kyushu Institute of Technology. Michiyasu Suzuki - Department of Neurosurgery, Graduate School of Medicine, Yamaguchi University. Masami Fujii - Department of Neurosurgery, Yamaguchi University Hospital. * Corresponding author

Fig. 9. Op amp used in the simulation. References HSPICE simulations were performed to confirm the operation of the proposed circuits. Level 52 HSPICE device parameters of a 0.18 μm 2-poly 4-metal CMOS process were used in the simulation. The bipolar supply voltage of 78

Articles

[1]

[2]

World Health Organization, “Epilepsy fact sheet”, WHO, http://www.who.int/mediacentre/factsheets/fs999/en /index.html, 2009. Yamakawa T., Grigorievich Z.L., L’vovich Z.Y., “Micro


Journal of Automation, Mobile Robotics & Intelligent Systems

[3]

[4]

[5]

VOLUME 3,

N° 4

2009

Freezing Probe and It’s Control”, Japanese patent, Application No. TOKUGAN2008-102974, 2007. Kitchin C., Counts L., A Designer’s Guide to Instrumentation Amplifiers, 3rd edition, Analog Devices, Inc., Norwood, MA, 2006. Niedermeyer E., Silva F.L., Electroencephalography: Basic Principles, Clinical Applications, and Related Fields, 5th ed., Lippincott Williams & Wilkins, Philadelphia, PA, 2004. Sandisona M., Curtis A.S.G., Wilkinson C.D.W., “Effective extra-cellular recording from vertebrate neurons in culture using a new type of micro-electrode array”, Journal of Neuroscience Methods, vol. 114, no. 1, 2002, pp. 63-71.

Articles

79


Journal of Automation, Mobile Robotics & Intelligent Systems

VOLUME 3,

N째 4

2009

R&D OF NOVEL MEDICINAL MATERIALS FOR CURING CANCER: SUGAR MODIFIED GD-DTPA MRI CONTRAST AGENTS AND PHOSPHA SUGAR ANTI-CANCER AGENTS Junko Yamashita, Mitsuji Yamashita, Michio Fujie, Kazuhide Asai Takuya Suyama, Satoru Ito, Valluru Krishna Reddy, Manabu Yamada, Keisuke Ogawa, Nobuhisa Ozaki, Satoki Nakamura, Takashi Aoki, Gang Yu, Kengo Aoshima, Tatsunori Kato, Nao Kamikage, Keita Kiyofuji, Yasuo Takehara, Harumi Sakahara, Hisao Takayanagi, Tatsuo Oshikawa, Sophie Laurent, Carmen Burtea, L. Vander Elst, Robert N. Muller

Abstract: Novel Sugar Dendritic Gd-DTPA Complexes for MRI Contrast Agents were prepared and evaluated by in vitro and in vivo methods. The sugar dendritic MRI contrast agents have a good blood vesse pool character, and draw blood vessels and liver cancer remarkably clearer than the clinically using Gd-DTPA. Phospha sugar derivatives or phosphorus heterocyclic derivatives provided by functional groups such as epoxide, bromide, etc., were prepared and evaluated by MTT in vitro method. These phospha sugar derivatives showed excellent activities against leukemia cells as well as solid cancer cells in fashions of (i) higher activity, (ii) wider spectra, (iii) higher selectivity and specificity distingushing healthy and cancer cells, etc., compared with the molecular targeting chemotheraputic anti-cancer agent, Gleevec. Keywords: MRI contrast agent, sugar-ball-dendrimer, Gd (III)-DTPA complex, phospha sugars, phospholanes, anticancer agent, tumor.

1. Introduction Cancer is one of the most serious diseases, and the disease is expected to be more and more serious if the innovation in cancer therapy will not be realized hereafter. To innovate in cancer therapy it is very important that medicinal materials or technologies to find tumors safely at the very early stage (early diagnosis) and to cure tumors by improving the quality of life (QOL) of the patients. To develop and realize such medicinal materials, highly functionalized MRI contrast agents [1], [2], which provide clearer images of very small cancers, are required. The currently quite often used MRI contrast agent is GdDTPA (Magnevist) which is safe and potential MRI contrast agent, however, the MRI contrast agent has poor characters for imaging blood vessels and cancers. To improve the poor characters of Gd-DTPA to draw cancers as well as blood vessels (Magnetic Resonance Angiography; MRA), Gd-DTPA was chemically modified by sugars. The results are described in the first part of this paper. Molecular targeting chemotherapeutic agents play one of the most important role in curing cancers. Among chemotherapeutic agents Gleevec (Imatinib) is one of the most commonly used medicines. Gleevec has potential activity against cancers, especially leukemia cells, nevertheless, Gleevec has lower activity towards some of leukemia cells. Gleevec is also used for solid tumors, however, to cure larger tumor tissues by Gleevec some times faces lower efficiency for the complete cure. 80

Articles

Therefore, new researches to develop alternative anticancer agents to Gleevec are steadily demanded. Phospha sugar is one of sugar analogs which have a phosphorus atom in place of the ring oxygen atom of normal sugars and is assigned to a category of pseudo sugars. Phospha sugars are not found in nature yet, and then all of them reported until now are chemically synthesized. On the other hand, the alternative pseudo sugars, such as aza-, carba-, and thia-sugars [3], [4], having a nitrogen, carbon, and sulfur atom, respectively, in the hemiacetal ring of sugars, are widely known in nature and are also chemically synthesized and modified extensively. Many of them are known to have important biological activities. Sugar starting materials, which have an oxygen atom in the hemiacetal ring, are basically and usually used to prepare pseudo sugars. To prepare phospha sugars from sugar staring materials is rather difficult compared with the other pseudo sugars, therefore, less kinds of phospha sugars are prepared and a little is known about the character of phospha sugars compared with aza-, thia- sugars, etc. As an alternative preparative method, we have developed phospha sugar chemistry starting from phosphorus heterocyclic compounds, e.g., 2-phospholene derivatives, by chemical modification at their reactive sites [3], [4]. Addition of bromine to the unsaturated C=C double bond of the starting 2-phospholene derivatives produced 2-bromo- or 2,3-dibromophospholane derivatives. Substitution reaction of 2-bromophospholane derivatives, which correspond to 2-bromo-2-deoxyphospha sugar derivatives, with amine nucleophiles gave N-glycosides of phospha sugar derivatives. Further, nucleic acid bases such as uracil were introduced into 2-phospholene 1-oxide derivatives by the cyclization reaction of acrylamide derivatives to prepare phospha sugar nucleosides [3], [4]. We are continuously searching biological activity for these phospha sugars or phosphorus heterocycles by in vitro and in vivo bio-assays. In the second part of this paper, we will deal with the successful preparation of many kinds of phospha sugars or phospholane derivatives from 1-phenyl-3-methyl-2-phospholene 1-oxide and the related derivative. The biological activities of the prepared phospha sugars or phospholane derivatives were evaluated by MTT in vitro method for leukemia cell. These deta will be reported in this paper.


Journal of Automation, Mobile Robotics & Intelligent Systems

2. Results and discussion 2.1. Sugar-Ball-Dendritic MRI contrast agents 2.1.1 Preparation of Sugar-Ball-Denderimers of Gd-DTPA-Dn-Sugar structure Currently, one of the most often used clinical MRI contrast agent is Gd-DTPA complex (Gadolinium Diethylenetriamine pentaacetic acid: Magnevist) (Fig. 1), whose molecular size is small and then the contrast agent penetrates the blood vessel. To make the MRI contrast agent remain in the blood vessels so as to draw clearer blood vessels (Magnetic Resonance Angiography (MRA)) and tumors, sugar ball dendritic structure of Gd-DTPA complexes were designated and synthesized. These dendritic Gd-DTPA complexes are generally represented in this paper as Gd-DTPA-Dn-Sugar. The reaction to prepare Gd-DTPA-D1-Glc(OAc) (four peracetylated glucose derivative) is exemplyfied in Scheme 1 [1], [2]. The product of the alikaline hydrized Gd-DTPA-D1-Glc(OH) is represented here as DEN-OH.

Fig. 1. Gd-DTPA (Magnevist).

VOLUME 3,

N째 4

2009

2.1.2 In vivo evaluation of DEN-OH as the MRI contrast agents to draw MRA and tumour The Gd-DTPA-D1-Glc(OAc) and DEN-OH were subjected to in vivo evaluation by using rats. The MRA by DENOH drew blood vessel clearly as shown in Fig. 2 and tumors on the liver of rats were also drawn quite clear images as shown in Fig. 3.

Fig. 2. MRA of rat at 30 min after injection (Left: by GdDTPA; Right: by DEN-OH).

Fig. 3. MRI of liver cancer of rat (Left: by Gd-DTPA at 3 min after injection; Right two: by DEN-OH at 3 min and 30 min after injection). Fig. 2 shows that DEN-OH draws boold vessels remarkably clearer than Gd-DTPA. And Fig. 3 shows that DEN-OH draws the liver cancer quite clearly. These results strongly indicate that Gd-DTPA-Dn-Sugar must be novel good MRI contrast agent for early stage tumor drawing. 2.2. Phospha sugar anti-cancer agents 2.2.1 Preparation of phospholenes, phosphorus heterocyclic compounds The McCormack reaction of 1,3-dienes with phosphorus chlorides, e.g., phosphorus tichloride, phenylphosphonus dichloride, afforded 2-phospholene derivatives (Scheme 2), which were used as the starting materials of phospha sugar derivatives or phospholanes.

Scheme 2. Synthesis of 2-phospholenes (McCormack Reaction).

Scheme 1. Preparation of Gd-DTPA-D1-Glc (OAc). The hydrized product of Gd-DTPA-D1-Glc(OAc) is simpely represented here as DEN-OH.

2.2.2 Preparation of phospha sugars or phospholanes 1,2-Anhydro-phospha sugars or 2,3-epoxy-1-phenylphospholane 1-oxides were prepared by an epoxidation of 2-phospholenes with sodium peroxide as shown in Scheme 3. The epoxidation reaction was stereospecific and stereoselective and gave essentially threo epoxide. The threo epoxide was defined by the two oxygen atoms Articles

81


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

of epoxide and phosphoryl locate on the same side of the sugar ring skeleton. The 1,2-dibromo-1,2-dideoxy-phospha sugars or 2,3-dibromo-1-phenylphospholane derivatives were prepared by an addition reaction of bromine to the double bond of 2-phospholenes as shown in Scheme 4 [3,4]. Ph

Na2O2

Ph

EtOH

Scheme 3. Epoxidation of 2-phospholenes with sodium peroxide.

N째 4

2009

Fig. 5 shows that the dibromo derivative was also active against U937 cell lines. The diastereomers of the dibromide showed different activities, and they are much more active against the U937 cell lines than Gleevec. These findings strongly indicate that the phospha sugars must be quite active and wide spectral anti-tumour agents. The dibromide showed that the phospha sugar were active not only against solution cancer (leukemia cells) but also against solid cancer (stomach cancer). Flow cytomertry for the preliminary mechanistic study indicated that these phospha sugars induced apoptosis for leukemia cells. Further studies on the optimization of the structure-activity of phospha sugar derivatives against cancer and the mechanistic studies are under progress.

3. Conclusion Scheme 4. Preparation of 1,2-dibromo-1,2-didoxy-phospha segars. The substitution, addition, homologation reactions, etc., were carried out to prepare new phospha sugars or phospholane derivatives. 2.2.3 Evaluation of phospha sugars or phospholanes by in vitro MTT method MTT method of phospha sugars against leukemia cell lines, K562 and U937, were carried out for in vitro evaluation as the anti-tumor agents [5]. The results are shown in Fig. 4 and Fig. 5. Fig. 4 shows that some of phospha sugars, e.g., bromohydrin, epoxide, dibromo derivatives, were active and many of the other derivatives were inactive.

Fig. 4. MTT evaluation of phospha sugars as anti-tumor agents against K562 cells.

Fig. 5. MTT evaluation of phospha sugars (dibromophospha sugars) as anti-tumor agents (comparison with Gleevec) against U937 cells. 82

Articles

The novel Gd-DTPA-Dn-Sugar structured MRI contrast agents could image quite small sized tumors, and then could be used for MRI contrast agents for MRA and initial stage tumor drawing. The novel phospha sugars could kill the leukemia cells in (i) high activity, (ii) selective and specific manner, (iii) wide spectra, by induction of apoptosis of cancer cells. Together with these novel medicinal materials, early stage findings and early stage chemotherapeutic treatment to cure cancers should be realized in the near future.

4. Experiment 4.1. Synthesis of 2,3-dibromo-3-methyl-1phenylphospholane 1-oxide To CH2Cl2 (10ml) solution of 3-methyl-1-phenyl-2phosholene 1-oxide (0.27 g, 1.4 mmol) and Mn(IV) dioxide (0.24 g, 2.8 mmol; 2.0 eq.) was added drop wise CH2Cl2 (10 ml) solution of bromine (0.40 ml, 7.8 mmol; 5.6 eq.) and the reaction mixture was stirred for 8 h at room temperature. The reaction was quenched by addition of saturated sodium sulfite aqueous solution. The aqueous mixture was extracted with chloroform (10 ml x 3). The organic layer was neutralized with saturated NaHCO3 aqueous solution, washed with saturated NaCl solution and dried over with anhydrous sodium sulfate. The solvent of the filtrate was evaporated under a reduced pressure to give an oily mixture of product. The mixture was purified by column chromatography on silica gel by using chloroform and methanol (30 : 1) as the eluent to give 2,3-dibromo3-methyl-1-phenylphosholane 1-oxide (0.37 g) in 75% yield; m.p. (Shimadu Simultaneous DTA-TG Apparatus (DTG-60A50AH)) 189.20 째C; b.p. 280.24 째C; TLC (Silica gel: Wako Chromato Sheet and/or Merk Kieselgel 60; Eluent: CHCl3 : MeOH = 20 : 1), Rf = 0.42; MS (MALDI-TOFMS: GL Science (Voyager-DE Porimerix); Matrix: a-Cyano4-hydroxy-cinnamic acid (m/z)), 349.29 (M - H+ (Molecular peak - 1); isotope peaks: 349.29, 351.29, and 353.29) and 351.29 (M + H+ (Molecular peak + 1); isotope peaks: 351.29, 353.29, and 355.29); IR (JASCO FT/IR 410 (KBr)): 1126 cm-1 (P=O), 748 cm-1, 1396 cm-1 (C-Br); 1HNMR (JEOL JNM-AL300 (300 MHz) and Hitach R90H (90 MHz); Solvent: CDCl3, d(ppm)); 1.67 (s, 3H, CH3), 2.362.46 (m, 2H, H-4), 2.97-3.02 (m, 2H, H-5) 4.28-4.31 (m, 1H, C-2), 7.51-7.70 (m, 5H, Ph-H). HPLC (Apparatus:


Journal of Automation, Mobile Robotics & Intelligent Systems

VOLUME 3,

N° 4

2009

JASCO HPLC Set (JASCO 860-CO, 880-PU, 875-UV, RI-930, and 807-IT; Column: Silica gel (Analysis: Wakopac, Wakosil F 4.6 mm Ă&#x2014; 250 mm, Eluent: CHCl3 : MeOH = 30 : 1, Flow rate: 0.5 ml/min), RT (retention time: min) values of diastereo isomers were 8.1, 9.1, 9.9, and 11.5.

AUTHORS Junko Yamashita, Mitsuji Yamashita*, Michio Fujie, Kazuhide Asai, Takuya Suyama, Satoru Ito, Valluru Krishna Reddy, Manabu Yamada, Keisuke Ogawa, Nobuhisa Ozaki, Takashi Aoki, Gang Yu, Kengo Aoshima, Tatsunori Kato, Nao Kamikage, Keita Kiyofuji Graduate School of Science and Technology, Shizuoka University, Hamamatsu 432-8561, Japan. E-mail: tcmyama@ipc.shizuoka.ac.jp. Yasuo Takehara, Michio Fujie, Harumi Sakahara, Satoki Nakamura - Faculty of Medicine, Hamamatsu University School of Medicine, Hamamatsu 431-3192, Japan. Hisao Takayanagi - Pharmaco-chemical Reserach Center, Tanabe Mitsubishi Pharmacy, Yokohama 227-0033, Japan. Tatsuo Oshikawa - Department of Materials Chemistry, Numazu College of Technology, Numazu 410-8501, Japan. Sophie Laurent, Carmen Burtea, L. Vander Elst, Robert N. Muller - NMR Laboratory, Department of Organic Chemistry, University of Mons-Hainaut, B-7000 Mons, Belgium. * Corresponding author

References [1] Takahashi M., Hara Y., Yamashita M., et al., Tetrahedron Lett., vol. 41, 2000, pp. 8485-8488. [2] H. Lammers, M. Frederic, R. N. Muller, et al., Inorg. Chem., vol. 36, 1997, pp. 2527-2538. [3] Yamashita M., Reddy V.K., Rao L.N., Haritha B., Maeda M., Suzuki K., Totsuka H., Takahashi M., Oshikawa T., Tetrahedron Lett., vol. 44, 2003, p. 2339-2341. [4] Totsuka H., Maeda M., Reddy V.K., Takahashi M., Yamashita M., Heterocyclic Commun., vol. 10, 2004, p. 295-300. [5] Nakamura S., Yamashita M., Yokota D., Hirano I., Ono T., Fujie M., Shibata K., Niimi T., Suyama T., Maddali K., Asai H., Yamashita J., Iguchi Y., Ohnishi K. Investigational New Drugs, in press (2009)

Articles

83


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

ANALYSES OF THE GENES INVOLVED IN DISEASE DEVELOPMENT TO CONSTRUCT DISEASE-RESISTANT PLANTS BY GENETIC ENGINEERING Shinji Tsuyumu, Hisae Hirata

Abstract: Many genes responsible for the disease development were identified from transposon-tagging, micro-array, and proteomics analyses. Here, we introduce especially the genes required for the initiation of pathogenic life cycle, the suppression of otherwise induction of resistance responses, massive production of virulence factors in plant pathogenic bacteria and an unique plant gene responsible for the development of canker symptom. From these findings we came to raise some new strategies to construct disease-resistant plants by genetic engineering.

1. Introduction Plants can be considered to be basically resistant to the attack by any plant pathogens by inducing a series of effective resistance responses. However, the pathogens have been evolved to escape from the induction of this resistance responses exceptionally in their host plant species. This escape can be achieved by suppressing the elicitation of resistance responses. Thus, all of the plants are known to have the gene sets required for such complex interactions between the pathogens and their compatible and incompatible plants. So, it is important to identify such broad-sense pathogenicity genes and the plant genes required for the cross talk among these genes.

for the control of plant pathogens. In fact, the mutants deficient in these regulatory genes and/or in their regulons in Xanthomonas axonopodis pv. citri and X. oryzae pv. oryzae (pathogen of citrus canker and that of rice bacterial wilt, respectively) were shown to fail in eliciting disease development. Thus, many strategies using genetic engineering to cause the disturbance of these regulatory mechanisms are expected to provide the new strategy of constructing disease-resistant plants. 2.2. Genes involved in suppression of induced resistance in plants Representative avr (avirulence) genes such avrXa7, avrXa10, avrPto, avrB, and avrRps4, which are known to be responsible for elicitation of the cultivar specific resistance response in the host plants containing each of their cognate resistance genes, were surprisingly found to suppress the elicitation of the resistance response in the plants carrying no cognate resistance gene [7], [8] , [9]. In other words, many of avr genes were shown to have dual functions (namely elicitor and suppressor of resistance response)(Fig. 1).

2. Results and Discussions 2.1. Genes required for the entrance of the pathogens into pathogenic life cycle The soft-rotting bacterial group such as Pectobacterium carotovorum subsp. carotovorum and Dickeya dadantii (syn. Erwinia carotovora subsp. carotovora and Erwinia chrysanthemi, respectively) commonly possess the genes of two component regulatory systems such as phoP/phoQ [1,2,3], and other regulatory genes such as slyA [4] and pir [5], [6] for these pathogens to enter into the pathogenicity life cycle. These regulatory genes were known to be important for the pathogens to adjust to the micro-environments typically found in plants (especially in intra-cellular areas of plants) such as low concentration of Magnesium, acidic pH, and low concentration of sugars. The genes and their translational products under the control of these regulators have been identified using microarray and proteomic analyses. Thus, the new strategy to control the disease incidence by blocking the entry of pathogen into “disease life cycle” may be now possible. The blockage of disease development at this stage may be most effective strategy 84

Articles

Fig. 1. Strategy to escape from suppression of resistance responses. Pathogen (in this case bacteria) inject (instead of onject) the effectors into plant cells via arrow (ie. Type III Secretion System, T3SS). Some elicit the resistance response and others such as most of avr genes suppress the resistance response. Thus, the block of this suppression should lead to elicitation of the resistance response even in compatible plants. Furthermore, we found that some of these Avr proteins can bind to a specific protein of specific plant. For example, Apl1 (the translational product of apl1, a member of the avrBs3 gene family of X. axonopodis pv. citri (a causative agent of citrus canker), was shown to bind


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

specifically to pectin methyl esterase (PME) of Citrus but not of tomato and tobacco [10]. Binding of Apl1 to PME was shown to alter the localizations of both Apl1 and PME (analyzed the localization using their fusions with fluorescent protein(s) using confocal laser microscope) and to be involved in alteration of the activities of both proteins. Thus, the blockage of the suppressor function of Avr protein may be released by binding to its these binding proteins so that the resistance response should remain effective to escape from the disease development (Fig. 1). 2.3. Bacterial genes involved in massive production of virulent factors In soft rot causing bacteria, the major virulent factor was shown to be pectate lyase (PL). In fact, the purified PL was shown to macerate the plant tissues, the major symptom of soft rot disease, and the treatment of plant blocks with PL inhibitors resulted in failure for pathogen to show soft-rot symptom. We found that multiple regulatory mechanisms are involved in the complex regulation of PL: (1) “Product Induction Mechanism”[11], in which induction can be initiated by the metabolic product of the substrate; (2) “Self Catabolite Repression”, in which the break down products of the substrate act as the signal for the repression at high concentration [12], (3) “Hyper-induction by plant components”, in which plant components (turned out to be sugars) at low concentration in addition to the inducer of product induction mechanism are responsible hyper-induction of PL [13]. Beside these regulatory mechanisms, the cell density dependent induction by homo-serine lactone, the posttranscriptional regulation by determining the stability of mRNA and others have also been found. Thus, the failure of these regulations especially of hyper-induction of PL by genetic modification of plants, by application of chemicals and by cultivation and postharvest devices may also be promising ways of controlling the soft rot diseases. 2.4. An unique plant genes involved in hyperplasia symptom development

N° 4

2009

As telomerase, which is the enzyme responsible to stop the reaction of shortening the end part of chromosome (i.e. telomere) upon each cell division, has been thought to be involved in cancer development in animals. In the case of citrus canker, we found that telomerase is also involved in the development of canker symptom where abnormal cell divisions and cell elongation are seen as in the canker cells in animals [14], [15]. Namely, when the expression of TART (gene for telomerase) was suppressed by introducing the RNA interference (RNAi) constructs in Citrus, the development of canker symptom was severely affected. Thus, this enzyme was thought to be good candidate for the control of Citrus canker disease, though we have to keep in mind that this enzyme is known to be involved in normal physiology of plants too. Thus, the special caution not to disturb these normal functions.

3. Conclusion From the facts that complex cross talks between plants and pathogens are required for the establishment of pathogenicity, many targets for the control of plant diseases were considered. Here, we showed the findings the genes involved for the pathogens enter into disease cycle and the regulatory genes involved in complex regulation of virulent factors were the good candidates to device new control measures of plant bacterial diseases. Also, we showed the involvement of telomerase of plants are involved in hyperplasia diseases such as a canker symptom. Thus, the control of telomerase may be effective way for the control such diseases. ACKNOWLEDGMENTS The data presented here have been supported mostly by Grantin-Aid for Scientific Research and a grant for Promotion of Science from the Ministry of Education, Science, Sports and Culture of Japan. Useful discussions with N.T. Keen, J.E. Leach, A. Collmer, A. H. Chatterjee, C.-H. Yang, T. Shiraishi, H. Kaku, A. Bagdonove, I. Toth. We would certainly appreciate the postdoctoral research associates and students in our laboratory for their enthusiastic research abilities.

AUTHORS Shinji Tsuyumu and Hisae Hirata - Institute for Genetic Engineering and Faculty of Agriculture, Shizuoka University, 836 Ohya, Suruga-ku, Shizuoka city, Japan #4228659. E-mail: tsuyumu@agr.shizuoka.ac.jp. * Corresponding author

References [1]

Fig. 2. Apl1 is trafficked from X.a. pv. citri via T3SS into a Citrus cell and binds to PME (pectin methyl esterase). This binding leads to elicit canker symptom with the help of telomerase.

[2]

Haque M.M., Tsuyumu S., “Virulence, resistance to magainin II, and expression of pectate lyase are controlled by the PhoP-PhoQ two-component regulatory system responding to pH and magnesium in Erwinia chrysanthemi 3937”, J. Gen. Plant Pathol., vol. 71, 2005, pp. 47-53. Haque M.M., Tsuyumu S., “Virulence, accumulation of acetyl-coenzyme A and pectate lyase synthesis are controlled by PhoP-PhoQ two-component regulatory system responding to organic acids in Erwinia chrysanArticles

85


Journal of Automation, Mobile Robotics & Intelligent Systems

[3]

[4]

[5]

[6]

[7]

[8]

[9]

[10]

[11] [12]

[13]

[14]

[15]

86

themi 3937”, J. Gen. Plant Pathol., vol. 71, 2005, pp. 133-138. Venkatesh B., Babujee L., Liu H., Hedley P., Fujikawa T., Birch P., Toth I., Tsuyumu S., “The Erwinia chrysanthemi 3937 PhoQ sensor kinase regulates several virulence determinants”, J. Bacteriol., vol. 188, no. 8, 2006, pp. 3088-3098. Haque M.M., Kabir M.S., Aini L.Q., Hirata H., Tsuyumu S., “SlyA, a MarR family transcriptional regulator, is essential for virulence in Dickeya dadantii 3937”, J. Bacteriol., 2009. (In press) Nomura K., William N., Kuwaiti K., Tsuyumu S., “The pir gene of Erwinia chrysanthemi EC16 regulates hyperinduction of pectate lyase virulence genes in response to plant signals”. In: Proc. Nat. Acad. Sci., vol. 95, 1998, pp. 14034-14039. Nomura K., Nasser W., Tsuyumu S., “Self-regulation of Pir, a regulatory protein responsible for hyperinduction of pectate lyase in Erwinia chrysanthemi EC16”, Mole. Plant-Microbe Interact., vol. 12, no. 5, 1999, pp. 385390. Fujikawa T., Yamashita T., Tsuyumu S., “Hypersensitive response suppression by type III effectors of plant pathogenic bacteria”, J. Gen. Plant Pathol., vol. 72, 2006, pp. 176-179. Fujikawa T., Ishihara H., Leach J.E., Tsuyumu S., “Suppression of defense response in plants by the avrBs3/ pthA gene family of Xanthomonas spp.”, Mol. Plant-Microbe Interact., vol. 9, 2006, pp. 342-349. Fujikawa T., Ishihara H., Leach J.E., Tsuyumu S., “Suppression of defense response in plants by the avrBs3/ pthA gene family of Xanthomonas spp.”, Mol. Plant-Microbe. Interact., vol. 19, 2006, pp. 342-349. Tsuyumu S., “Bacterial genes involved in interactions with plants”. In: Genomic and Genetic Analysis of Plant Parasitism and Defense, eds. Tsuyumu, Leach, Shiraishi, Wopert, APS Press, St. Paul, 2004. Tsuyumu S., “Inducer of pectic acid lyase in Erwinia carotovora”, Nature, no. 269, 1977, pp. 237-238. Tsuyumu S., ”Self-catabolite represion” of pectate lyase in Erwinia carotovora.”. J. Bacteriol., vol. 137(2), 1979, pp. 1035-1036. Tsuyumu S., et al., “Factors involved in the pathogenicity of Xanthomonas campestris pv. citri”. In: Molecular Aspects of Pathogenicity and Resistance, eds. by Mills H., Kunoh N., Keen T., Mayama S., APS Press, St. Paul, 1996, p. 105-114. Tsuyumu S., Fujikawa T., Komai K., Kimura S., Komatsu S., Hirata H., “Plants are basically resistant to any disease”, Soil Microbiol., vol. 62, no. 2, 2008, pp. 102105. Ishihara H., Uchida S., Masuda Y., Tamura K., Tsuyumu S., “Increase in telomerase activity in citrus inoculated with Xanthomonas axonopodis pv. citri.”, J. Gen. Plant Pathol., vol. 70, 2004, pp. 218-220.

Articles

VOLUME 3,

N° 4

2009


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

POWDER INJECTION MOULDING OF MICRO PARTS

Dionizy Biało, Andrzej Skalski

Abstract: This paper presents and discusses individual phases of forming micro parts using the injection moulding method. Tests involving the preparing of feedstock for the process of injection micro moulding were conducted. The rheological properties for binder-powder compositions were also defined. Tests were conducted for 316L stainless steel and iron metal powders and Al2O3 ceramics with granularity ranging from 0.135 μm to 16 μm. Binders consisted of polyethylene, paraffin and wax. Technological parameters of the process of injection micro moulding are presented, including the impact of individual factors on the filling of the mould cavities. Moreover, results of tests with respect to the shrinking of the micro parts as well as of the structure of the material following sintering are presented. Keywords: powder, injection moulding, debinding, micro parts, shrinkage.

1. Introduction Currently, several methods of micro machining are applied to make of micro parts and microstructures. The most important of these include micro-milling, etching, LIGA technology, erosive and laser micro-machining, plastic micromoulding, etc. [1], [2], [3]. One of the more promising methods is forming of microelements through injection. Such a method of fabrication makes possible the making of micro-elements with complex forms at high replication accuracies in the form of large-scale series at high efficiency that are competitive with respect to other fabrication methods. The essence of the injection process for forming elements using powders – Powder Injection Moulding (PIM) – involves the preparing of feedstock composed of special thermoplastic binder and powder, preparing the granulate, and forming the parts by way of injection using plastic injection moulding machines [4],[5]. Subsequently, the part is subjected to a debinding process, i.e. the removal of binder, and sintering. The process of micro part injection moulding – μPIM and μMIM (Micro Powder Injection Moulding, Micro Metal Injection Moulding) – is significantly more difficult than any macro injection moulding of products due to the small mass of the micro parts (<0.01 g) and different thermal conditions of the process. It is necessary to use injection moulding machines of special design and operation (different plastifying and feedstock feeding systems). The filling of micro-channels and micro-cavities with injection feedstock in the case of μPIM is coupled with numerous problems. A small batch of injected feedstock

moves along narrow channels and fills mould cavities of very small volume. Feedstock flow is difficult when subject to such conditions, it solidifies quickly and it is more difficult to replicate the details of the design of the micro parts being made. The present paper looks at the influence of mould temperature and the powder content of the injection feedstock on the course of the filling of the injection mould micro channels and cavities. Metal and ceramic powders of varied granularity (from 0.135 to 16 μm) in quantities of from 45% to 60% of the volume were used in the making of the feedstock. Results of measurements of shrinkage of the received micro parts following the process of debinding and sintering as well as the structure of the materials in metallographic micro-sections is presented.

2. Tests A schematic diagram of the process of micro part injection moulding is presented in Figure 1. As in the case of the moulding of macro parts, it consists of the following phases: preparing the injection feedstock, which consists of binder and powder, injecting the feedstock into the mould, debinding (removal of binder), and sintering. Preparing the injection feedstock includes the making of the binder followed by its mixing with powder. Making binder involves the precise hot mixing of its components, i.e. paraffin, wax, and polyethylene (Table 1). Such a composition guarantees the thermoplastic binder properties that are vital in subsequent phases of the injection moulding process. The ready binder should have the following qualities: it must bond well with powder, facilitate easy injection of the feedstock into the mould, and individual components should be easy to remove during debinding.

Fig. 1. Schematic diagram of the process of powder injection moulding. Articles

87


Journal of Automation, Mobile Robotics & Intelligent Systems

VOLUME 3,

N° 4

2009

Table 1. Binder composition. Material

Percentage share

Paraffin Carnauba wax PE Stearic acid

69% 10% 20% 1%

Mixing was performed in a 2Z type mixer with a system for measuring the mixing torque M as well as the temperature of the feedstock Tm. Table 2. Powders used in the tests. Powder Stainless steel Carbonyl iron Al2O3 Al2O3

Symbol

Granularity

Shape

316L OM HQ TM M

16 μm 4 μm 1 μm 0.135 μm 0.66 μm

spherical spherical spherical irregular irregular

Injection feedstock containing binder and powders of varying granularity (Table 2) was prepared. Powder content Vp ranging from 45% to 60% by volume was used. Mixing was conducted until the mass became uniform. A very important parameter during mixing is temperature. Figure 2 presents changes in the mixing torque as a function of mixing time. At the start, the torque M demonstrates significant oscillations in value, which is linked with mechanical breaking up as well as the adding of powder following the dissolving of the binder components. The stabilizing of the torque at a defined value signals the proper state of mixing of the feedstock ingredients. The tests made it possible to establish that the time needed to achieve a uniform feedstock amounted to appropriately 40 minutes.

Fig. 3. Impact of the mould temperature on inflow into micro channels of various crosssections for feedstock with OM (4 μm) powder at temperature of Tm = 125°C and a pressure of p = 60 MPa. of the mould cavities. In preliminary tests a proprietary mould to define the filling of the micro channels was used. It is a replacement for the spiral test used in traditional injection moulding [6], [7]. What was defined was the inflow of the material L [mm] with respect to the 2 cross-section of the channel S [mm ]. The impact of the temperature of the mould on the inflow is presented in Figure 3. The following results were received during tests: the temperature of the injection feedstock and of the injection pressure p at a mould temperature of Tf £ 30°C have little impact on inflow into the micro channels. Inflow of the micro channels improves significantly upon raising the temperature of the mould Tf > 50°C. In this case, the influence of the pressure p and feedstock temperature Tm was also observed. The received results were applied in preparing speciments for bending test (Figure 4a), tensile test (Figure 4b), and for gearwheel (Figure 4c).

Fig 4. Examples of parts made, prior to sintering. a – bars for bending tests: 2x2x10, 1x1x12 mm b – specimens for tension tests: 0.5x0.5x5 mm c – gearwheels: Dz = 1 mm, m = 0.15, z = 8

88

Fig. 2. Graph of mixing torque M as a function of mixing time t for the feedstock with HQ powder at Vp = 55%.

Problems with inflow also occur during forming specimens for tensile tests. Figure 5 depicts the impact of mould temperature on the inflow into the mould cavities. Raising the temperature of the mould in the 25°C–75°C range results in decreased viscosity of the feedstock, which significantly improves the filling of the mould cavities.

The received feedstock was subjected to the process of injection. A special proprietary injection-moulding machine with thermostatically controlled moulds was used. It makes possible the injection of very small quan3 3 tities of material in a range from 1 mm to 300 mm . Tests were conducted on the influence of temperature of the injected feedstock Tm and mould Tf, injection pressure p, and feedstock powder content Vp on the quality of filling

Fig. 5. Appearance of the micro specimens for tensile tests as moulded at various mould temperature and established injection conditions: Tm = 115°C, p = 60 MPa, and Vp = 60%.

Articles


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

Table 3. Injection parameter values for the proper fabrication of micro parts using feedstock with iron powder with a granularity of 1 μm. Powder contint

Tf

Tm

p

45% 50% 55% 60%

25°C 25°C 50°C 75°C

125��C 125°C 125°C 125°C

60MPa 60MPa 60MPa 60MPa

Table 3 compares the values of optimal injection parameters for the proper making of the specimens out of feedstock with iron powder with a granularity of 1 μm and varied content Vp. As the content of powder increases it is necessary to increase the temperature of the mould. However, too high a mould temperature is not recommended because the efficiency of the process decreases significantly. Too high level of heating influences increasing both the mould heating time and cooling time. After injection the specimens are subjected to debinding and sintering processes. The debinding process (removal of binder) consists of two phases. The first phase involves the elimination of easily melted components by rinsing them out using solvents. Tetrachloroethylene was used. Following this phase, the specimens become very porous. The time required to conduct this phase is dependent on the size of the specimen as well as its powder content Vp. The conducted tests demonstrate that approximately six hours is needed in order to properly remove paraffin from the specimen. The second phase of the debinding is conducted together with the sintering phase. The value of the debinding and sintering temperature for individual materials is presented in Table 4.

Fig. 6. Influence of the type and granularity of powders on the shrinkage of parts following the process of sintering at a content of Vp = 55%. The size of powder particles also influences the structure of the received parts. Figure 7 presents metallographic micro-sections for specimens made of the discussed materials (of varying granularity). Parts made of powder with a granularity of over a dozen micrometers (316L) are very porous. Less porous are specimens of OM iron powder (4 μm), and specimens of HQ iron powder (1μm). Ceramic powder TM with lowest granularity (0.135 μm) forms a uniform structure, without pores, following sintering.

Table 4. Temperature of thermal debinding and sintering. Powder

316L

OM

HQ

M

TM

Debinding 550°C 650°C 650°C 550°C 550°C Sintering 1250°C 1110°C 1110°C 1600°C 1450°C The remaining binder undergoes thermal breakdown at temperatures in excess of 500°C. Evacuation of the products of binder breakdown is made possible by the pores left after rinsing in solvent. Sintering takes place after the total removal of the binder. During sintering the specimens achieve their required mechanical qualities and also undergo shrinkage. Literature often describes the relationship, where the finer the powder, the easier it undergoes sintering because it has greater surface energy. For the same reason, dimensional changes in the specimen following sintering are dependent on the size of the powder used. Figure 6 shows the dependence between shrinkage and powder granularity. The finer the powder used, the greater the shrinkage of the finished product with respect to the mould cavities. Such information is very useful in designing injection mould cavities.

Fig. 7. Metallographic micro-sections of materials following sintering, made of feedstock of at a fixed content Vp = 55% ­ a) 316L (16 μm), b) OM (4 μm), c) HQ (1 μm), d) TM (0.135 μm).

3. Conclusions 1. The quality of the filling of mould cavities is primarily influenced by the temperature of the mould and the feedstock powder content coefficient. 2. The course of the inflow into the mould is also influenced by the granularity of the powder used. The finer the powder, the greater the viscosity of the feedstock and the more difficult the fillings of the micro-cavities are. 3. Powder granularity also influences the degree of shrinkage. The smaller the granularity the greater the shrinkage of the product and the less its porosity is.

Articles

89


Journal of Automation, Mobile Robotics & Intelligent Systems

ACKNOWLEDGMENTS This paper is a fragment of Research Program No. 0309/T02/ 2006/31 financed by the Polish Ministry of Science and Higher Education.

AUTHORS Dionizy Biało*, Andrzej Skalski - Institute of Metrology and Biomedical Engineering, Warsaw University of Technology, ul. św. A. Boboli 8, 02525 Warsaw, Poland. E-mail: dbialo@mchtr.pw.edu.pl * Corresponding author

References [1]

[2]

[3]

[4]

[5]

[6]

[7]

90

Oczos K.E., “Micro Parts Forming: Characteristics of Micro Machining Methods,” Mechanik [Mechanic], 5­6, 1999, pp. 309-327 (in Polish). Poiter V., et al., “Micro Powder Injection Moulding,” EURO PM2000, October 1820, 2000, Munich, Germany, vol. PIM, pp. 259-264. Hasselbach J., et al., “Investigation on the International State of Art of Micro Production Technology,” Euspen International Topical Conference, Aachen, Germany, 19th–20th May, 2003, pp. 11-18. Piotter V., et al., “Micro Powder Injection Moulding: New Developments and Results,” Proceedings of the Euro PM 2008: Powder Injection Moulding, vol. 2, pp. 331335. Imgrund P., Haack J., Rota A., “Current Developments in Micro Moulding of Functional Materials”, Proceedings of the 7th International Conference, Bremen, May 2007, pp. 115-118. Stanek M., Manas M., Draga T., Manas D., “Measurement of Polymer Fluidity”, Proc. of the IV International Congress on Precision Machining 2007, 25th–28th September, 2007, Poland, vol. 1, pp. 285-289. Biało D., Skalski A., Paszkowski L., “Certain Aspects of the Injection Moulding of Micro Parts from Metal Powders”, Rudy i Metale Niezelazne – Metalurgia Proszkow, no. 4, 2008, pp. 241-245 (in Polish).

Articles

VOLUME 3,

N° 4

2009


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

INFLUENCE OF SYNTHESIS CONDITIONS ON THE PROPERTIES OF FEXCOYOZ - NANOPARTICLES IN SIO2 SOL-GEL FILM Vitaly Valeryevich Sidsky, Vladimir Evgenyevich Gaishun, Oleg Mikhailovich Demidenko, Tatyana Alexandrovna Savitskaya, Vladimir Vasilyevich Panykov, Alina V. Semchenko

Abstract: Synthesis conditions influence on structure of FexCoyOz nanoparticle in SiO2 - sol-gel coating is discussed. The samples are synthesized by hybrid sol - gel method including following stages: TEOS hydrolysis, introduction into sol of Co(NO3)2 × 6H20 and Fe(NO3)3 × 9H20, deposition of film-forming sol by spin-coating on silica substrate, heat treatment at the temperature from 200 up to 900 °C. The Synthesis conditions influence on a specific structural property of sol gel films is investigated by AFM. Keywords: nanoparticle, AFM, FexCoyOz, phase change, primitive and face-centered cubic lattice.

1. Introduction Unique physical characteristics of nanoparticles arising due to surface or quantum-size effects, are object of intensive researches now [1]. The special place in this investigation is taken by the magnetic characteristics of nanoparticles. Most difference (sometimes very essential) between compact magnetic materials and the same are detected and the theoretical base which is capable to explain the majority of observed effects. A number of general methods of nanoparticles obtaining [2] is developed now; the majority of them can be used for obtaining of magnetic nanoparticles. But the synthesys of magnetic nanoparticles has the essential singularities; they can be formulated as the requirements to know-how of synthesizing of magnetic nanoparticles. It is necessary to synthesize the nanoparticles of the given size and shapes, at all events, the dispersion on the sizes should be small (5-10 %) and can be controlled. Differently, changing the sizes, shapes, structure of nanoparticles, it is possible in definite limits to control the magnetic characteristics of materials on the base of magnetic nanoparticles. All this allows to use the materials including nanoparticles in perspective systems of a record and storage of information, for creation of new permanent magnets, in systems of magnetic cooling, as magnetic sensors etc. Using the sol - gel technology it is possible to synthesize the sol - gel matrix with different structure possessing much higher thermostability and chemical stability than polymer matrixes. The main purpose of the research is the investigation of magnetic properties sol-gel films including ferromagnetic FexCoyOz nanoparticles. Control of the nanoparticles size in sol-gel coatings by doping at various sol-gel process stages by different ratio of such elements as iron and cobalt will allow to use these films m in electronics,

as data carriers with high density, and also as sensors of electromagnetic radiation.

2. Experimental The formation of sol-gel film takes place as a result of hydrolysis of film-form solution chemical agents and their interplay to surface layer of a substrate. For example, film of a silicic acid derivated as a result of a full hydrolysis of solutions applicable alkyne compounds, are consolidated at the expense of covalent linkages Si-O-Si. The formation of such band was conducted by a way of hydrolysis and polycondensation of teraetoxysilane (TEOS) and alkyl-displaced alkoxysilane and ferrum carbonate and cobalt nitrate in organic solution. The main stages of sol-gel process are showed in Fig. 1. Then the film-form solutions were deposited on a substrate by spin-coating or dip-coating. Silicon wafers and quartz glass were applied as the substrates. Just after the deposition the samples were heat treated stepwise in a muffle during 20 minutes at the temperature from 200 to 900 °C.

Fig. 1. The main stages of sol-gel process. FexCoyOz sol-gel films were transparent and homogeneous with light green colour before heat treatment and from golden up to red after heat treatment at 500 °C within 15 minutes. The analysis of phase changes in the synthesized matrixes has been carried out using the XRD method. The examination of doped film surfaces has been performed by AFM (SOLVER P 47 - PRO «NT-MDT») which allows for studying structural formations in the nanometer scale. The magnetic characteristics of the formed silicate sol-gel films doped with FexCoyOz nanoparticles have been studied by ballistic method of construction of hysteresis Articles

91


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

loop of ferromagnetic materials at different strength of external magnetic field.

3. Results In Figure 2 the surfaces images of FexCoyOz SiO2 sol-gel films on the silicon plane (100) are shown. Films are deposited by dip-coating and annealed on air at 800 °C. As can be seen, sol - gel coating with different ratio between ferrum and cobalt compaunds, have the brightly expressed features of structure. The sample in Figure 2(a) has not the expressed features of structure. At the ratio of ferrum and cobalt compounds equal 1:1 on the AFM image are clearly visible nanoparticles with the mean size 50-60 nm and cubical shape. With a large degree of accuracy all particles have the identical size. The size of nanoparticles increase with the increasing of cobalt content. The main role on the forming of cubical nanoparticle shape is played the crystalline pattern with the symmetry of the silicon plane (100)). By XRD data, the type of structure of nanoparticles is face-centered (the temperature heat treating is 800 °C). The unit cell of FexCoyOz nanoparticle, erased in the SiO2-film doped with ferrum and cobalt compounds (1:1), is shown in Figure 3. The image of the unit cell is obtained through the program Powder Cell 2.0 [3].

N° 4

2009

The hysteresis loop for FexCoyOz -doped film-forming sol is characterized by the magnetization Bs = 12 Gs, the coercive force Hc = 68 a/m, and the residual magnetic induction Br about 2 Gs. The hysteresis loop of xerogel differs from that of film-forming sol by the magnetization value Bs =14 Gs.

4. Conclusion SiO2-films with FexCoyOz nanoparticles are synthesized by sol-gel on silicon substrate under heat treatment temperature 800 °C. At the ratio of ferrum and cobalt compounds 1:1 the nanoparticles with the mean size 50-60 nm and cubical shape are arise. The size of nanoparticles increase up to 500 nm with the increasing of cobalt content. The main role on the forming of cubical nanoparticle shape is played the crystalline pattern with the symmetry of the silicon plane.

AUTHORS Alina V. Semchenko*, Vitaly Valeryevich Sidsky, Vladimir Evgenyevich Gaishun, Oleg Mikhailovich Demidenko - Advanced Materials Research Laboratory, Gomel State University, 246019, Gomel, Belarus. E-mail: semchenko@gsu.by. Tatyana Alexandrovna Savitskaya, Vladimir Vasilyevich Panykov - Belorussian State University, 220050 Leningradskay st. 14, Minsk, Belarus. * Corresponding author

References [1]

[2]

[3]

Fig. 2. AFM images of FexCoyOz–SiO2 sol-gel films on the silicon plane after heat treatment on air at the temperature 800 °C with the ratio between ferrum and cobalt compounds: a-1:0,25; b-1:1; c-1:0,5; d - 1:2; e-0:1.

Fig. 3. The unit cell of FexCoyOz nanoparticle, erased in the SiO2-film doped with ferrum and cobalt compounds (1:1). 92

Articles

Baraton M.I., “Synthesis, Functionalization, and Surface Treatment of Nanoparticles”. Los-Angeles: Am. Sci., 2002. Gubin S.P., Koksharov Y.A., “Synthesis, structure and properties of colloidal cobalt nanoparticles”. Inorganic Materials, vol. 38, no. 2, 2002, p. 1287. Kraus W., Nolze G., “POWDER CELL – a Program for the Representation and Manipulation of Crystal Structures and Calculation of the Resulting X-ray Powder Patterns”. J. Appl. Cryst., no. 29, 1996, pp. 301-303.


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

LOW-RANGE TILT SENSING WITH MEMS ACCELEROMETERS

Sergiusz Łuczak, Dariusz Kołodziej

Abstract: Ways of adjusting MEMS accelerometers for tilt measurements over a small angular range of few degrees arc or less are considered with regard to achieving a possibly high accuracy. The paper describes some additional mechanical structures applied to overcome the problem of incompatibility between the large measuring range of MEMS accelerometers and the expected small angular measuring range. Keywords: MEMS, accelerometer, tilt, accuracy.

1. Introduction One of the most common applications of MEMS accelerometers are tilt measurements. The typical low-range of the commercial MEMS accelerometers is of ±1÷2 g, which is adjusted quite well to tilt changes of ±90°. In such a case, the user may achieve related accuracy of tilt measurements up to ca. ±0.2° [1]-[3]. However, in many cases, the angular range of tilt measurements is quite small, i.e. of a few degrees arc or less (e.g. while levelling, aligning, or in astronautics), and sometimes the above accuracy may not be sufficient. So, the typical solution would be application of MEMS accelerometers featuring a lower measuring range. The advantages of such approach are discussed by the author in [4]. However, no such devices are commercially available. A once available accelerometer with the smallest measuring range (according to the knowledge of the author) was a MEMSIC MXA6500E with the range of ±0.5 g [5]. Yet, its accuracy was no better than its competitors' featuring a larger measuring range. In the opinion of the author, the reason for not manufacturing MEMS accelerometers with a measuring range much below ±1 g is probably the noise level (being usually of ca. 1 mg what corresponds to a tilt of ca. 0.1° at best), which makes it useless to decrease the measuring range, as the accuracy, dependent on the noise level, cannot be improved. So, at the time being, it is rather impossible to acquire an accelerometer with a measuring range that is adjusted to detection of tilt angles as small as few degrees arc. However, there are still some ways of adjusting the measuring range of the considered device to the range of the tilt angles to be sensed.

2. Mechanical Gears The first approach to solve the problem of measuring tilt over a small angular range is to apply a special mechanical gear, operating e.g. as a compliant mechanism, that would increase the measuring angular range up to ±90°. Then, the problem becomes a typical tilt measurement

over a large measuring range, referred to e.g. in [1]. However, such mechanism must feature very small plays and kinematic errors, otherwise accuracy of the related measurements would be significantly affected. So, the most promising solution are mechanisms based on separate elastic elements (like springs or membranes) or integrated compliant mechanisms. An example in the first group mentioned above may be a twisted torsional suspension spring having a tapeshape. This elastic element was commonly applied in mechanical measuring instruments, like a microkator or an optikator [6]. It can provide a high ratio of movement conversion, e.g. displacement of ca. 0.03 mm can be transformed into a rotation of 180° [6].

Fig. 1. Mechanism with a tape-shape spring. The basic element of the mechanism presented in Fig. 1 is a twisted spring 1 having a tape-shape, which rotates as it gets tensed. A MEMS accelerometer 4 is fixed to the spring in the middle of its length, while its ends are attached to two flat springs 2 and 3, secured in a frame 6. The first spring is loaded with a seismic mass 5. When the whole structure gets tilted (see the thin arrow), the seismic mass 5 keeps its vertical position, and so deflects the flat spring 2, what tenses the tape-shape spring 1, and thus causes its rotation along with the accelerometer 4 (see the thick arrow). Crucial parameters of the mechanical structure (material and dimensions of all the springs, weight of the seismic mass) should be selected in such a way, as to ensure rotation angle of the accelerometer of ±90° while tilted at the maximum value of the measured angle. Even though it is possible to build the proposed mechanism using a conventional technology, there are certain problems to be solved: Articles

93


Journal of Automation, Mobile Robotics & Intelligent Systems

— at least 4 wires must be led to the accelerometer, and each has a significant stiffness (solution: to integrate the sensor with a wireless transmission unit or a signal encoding unit using only two halves of the twisted spring as wires), — the accelerometer is actually rolled within a nonvertical plane (solution: to apply a more complicated formula for calculating the tilt), — the accelerometer may be too heavy for the tapeshape spring (solution: to immerse the whole mechanism in a liquid, and thus improve also its dynamics), — the seismic mass does not keep exactly vertical position due to a restoring moment generated by all the springs (solution: calibration of the whole system compensating for a linear character of this phenomenon, as indicated by experimental studies performed by the author in the past), — assembly of the system in a conventional technology of fine mechanics is very complicated and expensive. As for the second group, many structures may be proposed. A related example is presented in Fig. 2. The basic element of the mechanism presented in Fig. 2 are two parallel springs 1 and 2 fixed to each other at their ends on one side, and coupled with a MEMS accelerometer 6 right there. On the other side, spring 1 is anchored to a stiff frame 4 while spring 2 is connected with a lever 3, supporting a seismic mass 5 (manufactured as an additional element). Members 1–4 constitute one monolithic structure. When the whole structure gets tilted (see the thin arrow), the seismic mass 5 keeps its vertical position, and so shifts the flat spring 2, what causes deflection of both springs (the springs operate as a thermostat bimetal element), and thus rotation of the accelerometer 6 (see the thick arrow). Crucial parameters of the mechanical structure (material and dimensions of the springs, distance between them, weight of the seismic mass) should be selected in the same way as in the previous case, striving for the rotation angle of the accelerometer to be as large as possible.

VOLUME 3,

N° 4

2009

Nowadays, new fabrication technologies, like surface micromachining, bulk micromachining or LIGA process, make it possible to fabricate the proposed mechanism as a monolithic structure of a complicated shape. Regarding its quite large dimensions, exceeding 10 mm, it can be also fabricated using microstructuring of photosensitive glass, a technology developed at the Technical University of Ilmenau, Germany [7]. Quality of such mechanism is much better compare to the traditional approach, where it was to be assembled out of many separate parts. Sometimes, such kind of assembly caused the mechanism to be dysfunctional. A good example may be a constant force escapement designed by Nicolas Dehon for mechanical watches manufactured by the Girard-Perregaux company [8]. As it turned out, only its integrated version manufactured in silicon technology was operating properly. Apart from the new manufacturing technologies, a powerful toll are new materials, like e.g. Shape Memory Alloys (SMA), that, when appropriately shaped, can provide high stiffness generally, and at the same time high compliancy in a desired direction [9]. Other mechanisms of this kind may be considered also. It may be presumed that one of the following elements playing the substantial role would be applied there: — levers, — toothed wheels, — keys, — threads, — springs. For instance, it seems an interesting idea to apply a harmonic drive operating in a reverse direction (i.e. as a multiplying gear instead of a reduction gear), since ratio of such gear may be even 300 [10]. Its efficiency is reported to be of at least 0.8 [10], what suggests that it will not lock itself by reverse operation. The biggest problem is its resistance to motion, which would require a large weight of the seismic mass, and would result in an increase of the overall dimensions of the whole instrument. Moreover, it also introduces an effect of a restoring moment, as described above.

3. Improvement of Accuracy In the case of applying an additional mechanism, the accelerometer gets rotated by an angle j much bigger than the tilt a, what can be expressed as, (1) where q is the ratio of the mechanism, arbitrarily evaluated to be of 3÷90. Then, relation between uncertainties u(a) and u(j) of the related angles, determining the resultant accuracy of tilt measurements, can be expressed by the following equation, (2)

Fig. 2. Monolithic compliant mechanism. 94

Articles

So, the actual tilt angle will be determined q-times more accurate compare to accuracy featured by the


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

accelerometer. Let us assume, as reported above, that the resultant uncertainty of an accelerometer is of 0.2°, and the ratio is of 90. Then the accuracy of the instrument consisting of the accelerometer and the additional mechanism will be of ca. 10 seconds arc, accepting a 5% kinematic error of the mechanical structure. Such accuracy allows measuring of tilt angles within a rational range of ±2' or wider.

References [1]

[2]

[3]

Two cases can be taken into account: — a single-axis tilt sensor operates as a limit gauge only, — a single-axis accelerometer operates as a two-axis tilt sensor.

[4]

This principle will allow the output signal to be significantly magnified, however rather for one or few particular angular positions only (being the limits of the allowable range).

[6]

[5]

[7]

5. Conclusions Low-range MEMS accelerometers are very interesting measuring instruments because of their low price, miniature dimensions, low power consumption and robustness. Despite the fact that their accuracy does not allow measuring tilt within a small range (of few degrees arc or less), application of the proposed additional mechanical structures makes it possible to sense tilt angles as small as few minutes arc. It should be noted that the proposed mechanisms operates within one plane only. So, in the case of twoaxial tilt measurements, two such mechanisms must be applied, oriented perpendicularly with respect to each other. Additionally, the mechanism with a tape-shape spring (see Fig.1) operates properly only within a vertical plane. So in the considered case of two-axial tilt measurements it should be equipped in an additional support, e.g. two suspension springs. Even though, at the time being, it seems that application of additional mechanical structures is the only solution while measuring small tilt angles, it should be expected that in the nearest future metrological parameters of MEMS accelerometers will be significantly improved, especially with regard to the noise level determining their accuracy. Then, it may be expected that accelerometers with a very low range, i.e. much below ±1 g, will be commercially available. In such a case there will be no problem in direct realisation of the measurements of small tilt angles referred to.

2009

and Photonics, Faculty of Mechatronics, Warsaw University of Technology, ul. Boboli 8, 02-525 Warsaw, Poland. E-mail: s.luczak@mchtr.pw.edu.pl. * Corresponding author

4. Bistable Elastic Elements Still another approach to the problem of increasing the operating angular range of the MEMS accelerometer in the considered tilt measurements is application of bistable elastic elements. This solution does not provide many advantages, yet in some cases may be a helpful and simple solution.

N° 4

[8] [9]

[10]

Łuczak S., “Advanced Algorithm for Measuring Tilt with MEMS Accelerometers” [in] R. Jabłoński et al., Recent Advances in Mechatronics, Springer-Verlag: Berlin Heidelberg, 2007, pp. 511-515. Horton M., Kitchin C., “A Dual Axis Tilt Sensor Based on Micromachined Accelerometers”, Sensors, vol. 13, no. 4, 1996, pp. 91-94. Łuczak S., Oleksiuk W., Bodnicki M., “Sensing Tilt with MEMS Accelerometers”, IEEE Sensors J., vol. 6, no. 6, 2006, pp. 1669-1675. Łuczak S., “MEMS Accelerometers in Sensing Small Tilt”, Proc. 6th Polish-German Mechatronic Workshop, Ilmenau, 2007, pp. 25-30. “Low Profile, Low Consumption ±0.5 g Dual Axis Accelerometer with Absolute Analog Outputs MXA6500E”, MEMSIC Inc., North Andover, MA, 2005. Ratajczyk E., „Obliczanie przełożenia czujników mikrokatorowych”, Mechanik, Warsaw, vol. 9, 1968, pp. 502506 (in Polish). Preuß R., K. Zimmermann, “Simulation and Experimental Testing of a Microgripper”. In: Proc. 3rd Polish-German Mechatronic Workshop, Krynica/Kraków, 2000, pp. 166-169. “The Silicon revolution”, Europa Star, vol. AugustSeptember, 2008. Oiwa T., Sugimoto T., “Shape Optimization for Flexure Hinges”, J. Jpn. Soc. Prec. Eng., vol. 66, no. 60, 2000, pp. 955-959 (in Japanese). Tryliński W., Drobne mechanizmy i przyrządy precyzyjne, WNT: Warsaw, 1978, pp. 437-438 (in Polish).

AUTHORS Sergiusz Łuczak*, Dariusz Kołodziej - Division of Design of Precision Devices, Institute of Micromechanics Articles

95


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

TESTING OF THE THREE AXIS MAGNETOMETERS FOR MEASUREMENTS OF THE EARTH'S MAGNETIC FIELD Roman Szewczyk, Jacek Salach, Adam Bieńkowski, Rafał Kłoda, Marcin Safinowski

Abstract:

2. Magnetoresistive sensors

The paper presents experimental setup for testing of the three axis magnetoresisitive magnetometers for measurements of the magnetic field within the range of Earth's magnetic field. Measuring setup integrates three axis Helmholtz coils controlled by the specialized current source. The accuracy of magnetic field generated by the Helmholtz coils is up to 0.6% for the one direction. Presented experimental setup can be also applied for investigation and calibration of the other magnetometers utilizing magnetoresistive, Hall or fluxgate sensors.

Magnetoresistive sensors are used in navigation systems, in security and in measuring systems for industrial applications, especially in non-destructive testing methods. Magnetoresistive sensors, call also magnetrons, use effect of change in a resistance of ferromagnetic thin film as a function of the magnetic field. These changes are a consequence of the extension of a path of carrier charge and the scattering increase of this charge. For low values of the magnetic field this change can be described by equation (1):

Keywords: magnetic sensors, Helmholtz coils, calibration.

rB=r0*(1+a*m H*B)

1. Introduction Magnetic field occurred everywhere on the Earth, irrespective of the latitude or the longitude. The only difference is connected with its value and direction. This difference is determined by geographic location and geologic conditions. To determine the value and direction of Earth's magnetic field one has to use a magnetometer. The most used for this purpose are fluxgate magnetometers and magnetoresistive magnetometers [1], [2]. Usually measurements cover the determination of horizontal magnetic field H= and vertical magnetic field H^. In case like this, both declination q and inclination j of the magnetic field, may be calculated. Declination q is determined as an angle between magnetic and geographic meridian, whereas inclination j is an angle between ground level and direct of line of magnetic field at the given point. Measurements of magnetic field are also realized in the space, near the planets in Solar System as well as on the Moon. Until now, about 200 magnetometers of different types were sent in the space [3]. To assure appropriate accuracy of magnetic field's measurement it is necessary to investigate uncertainty of magnetometers, as well as its sensibility and linearity. Helmholtz coils are the best standard for generation of the magnetic field. It should be indicated that the accuracy of Helmholtz coils depends only on the accuracy of geometry of the coils and the accuracy of the current source. This connects magnetic field standards with other standards determined as physical phenomena. One of the variants of the Helmholtz coils is three axis Helmholtz coils. In such coils, the compensation terrestrial magnetic field can be carried out as well, as it is possible to generate the arbitrary magnetic field vector.

96

Articles

2

(1)

where r0 is a resistivity of the material without the magnetic field, mH is Hall activity charge carrier, constant a is factor describing scattering mechanism and B is magnetic flux density in the material [4]. In practice, magnetoresistor is produced during the vacuum evaporation of meander path. Therefore significant changes of resistance and high sensitivity can be obtained [4]. Most often magnetorestive sensor contains four magnetoresistors connect in Wheatstone bridge, supplied by differential power supply. The most important advantage from magnetoresistive sensors is wide range of measuring magnetic field (up to 1mT), wide range of frequency (from DC up to 1GHz) as well as the possibility of operation in the range of temperature form ­70°C up to 200°C. Moreover, magnetoresisitive sensors can be produced as small as flat devices, what make them useful for many applications. A defect of magnetoresisitive magnetometers is necessity of use sophisticated electronic circuits, especially in configuration with compensation coil, generating magnetic field perpendicular to the sensor. Moreover, high accuracy magnetoresistive sensors are still expensive. An example of recently developed magnetoresistive magnetometer is three axis magnetometer HMR 2300, applied like reference sensor in developed experimental setup.

3. Developed testing setup for magnetometers Figure 1 presents block diagram of the testing setup for investigation of the functional properties of magnetoresistive magnetometers. The measuring setup covers the precision current source BOP ­36A, amperometer APPA-207, three axis Helmholtz coils and reference magnetoresistive magnetometer HMR 2300. Magnetometer is connected to personal computer and works under the Labview software. Figure 2


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

presents visualization of three axis Helmholtz coils used in measuring setup.

N° 4

2009

Figure 3 presents window of program for acquisition and visualization of the results of magnetic field measurements using HMR2300 magnetoresistive magnetometer.

Fig. 1. Block diagram of testing setup for investigation of functional properties of magnetoresistive magnetometers.

Fig. 3. Program for visualization of the results of measurements using magnetometer HMR-2300. Fig. 2. Visualization of the three axis Helmholtz coils for developed experimental setup. Helmholtz coil for experimental setup was produced during the special process covering water cutting. Water cutting process provides high precision in production of large elements, such as Helmholtz coil's bobbins. Value of the magnetic field generated by the Helmholtz coil in its axis is given by the equation (2) [5]:

(2) where x is the distance from the centre of Helmholtz coils to measuring point, a is the distance between coils, n is number of turns for one coil, I is current in coils as well as r is the radius of coil. In order to generate uniform magnetic field, the distance a between coils should be equal to the radius of coil.

Presented in Figure 3 oscillation of the measuring field is the result of disturbances of Earth magnetic field through a ferromagnetic object.

4. Results of accuracy estimation To estimate the accuracy of value of magnetic field generated by developed Helmholtz coils, following assumptions were taken: - accuracy of coil diameter is 0.5 %, - accuracy of current measuring is 0.2 %. To sum up, the accuracy of generated magnetic field is about 0.6 %. This accuracy is almost the same as accuracy of the magnetoresistive magnetometer HMR-2300.

5. Conclusion

Tab. 1. The most important functional parameters of developed Helmholtz coils.

Presented experimental setup enables investigation of functional parameters of the magnetometers, in range of magnetic field up to 100 μT, what is sufficient for magnetometers operating in the range of the Earth's magnetic field. Generation of arbitrary magnetic vector is possible due to application of the three axis Helmholtz coils. Accuracy of generated magnetic field is about 0.6 % for each axis. Such accuracy is sufficient for majority of industrial and navigational applications and sensors.

Oś Radius Coils winding Coil constant Coil constant r(m) turns n k1( A/m / A) k2( mT / A)

ACKNOWLEDGMENTS

X 0.2230 Y 0.2045 Z 0.1860

This work has been supported by the European Union in the framework of European Social Fund through the Warsaw University of Technology Development Programme.

Table 1 presents functional parameters of Helmholtz coils developed for the measuring setup.

25 25 25

80.213 87.469 96.169

100.799 109.917 120.850

Articles

97


Journal of Automation, Mobile Robotics & Intelligent Systems

AUTHORS Roman Szewczyk* - Institute of Metrology and Biomedical Engineering, Warsaw University of Technology and Industrial Research Institute for Automation and Measurements - PIAP, Warsaw, Poland. E-mail: szewczyk@mchtr.pw.edu.pl Jacek Salach - Institute of Metrology and Biomedical Engineering, Warsaw University of Technology, Warsaw, Poland. E-mail: j.salach@mchtr.pw.edu.pl. Adam Bieńkowski - Institute of Metrology and Biomedical Engineering, Warsaw University of Technology, Warsaw, Poland. E-mail: bienko@mchtr.pw.edu.pl. Rafał Kłoda - Institute of Metrology and Biomedical Engineering, Warsaw University of Technology and Industrial Research Institute for Automation and Measurements - PIAP, Warsaw, Poland. E-mail: rkloda@piap.pl. Marcin Safinowski - Industrial Research Institute for Automation and Measurements - PIAP, Warsaw, Poland. E-mail: msafinowski@piap.pl. * Corresponding author

References [1] [2] [3]

[4] [5]

98

Ripka P., Magnetic sensors and magnetometers, Artech House, 2001. Janicke J.M., The Magnetic Measurements Handbook, Magnetic Research Press, 1997. Gordon D.I., Brown R.E., A Review of Magnetic Sensors”, IEEE Transaction on Magnetics, vol. 8, no. 1, March 1972, p. 973. Lenz J.E., “A Review of Magnetic Sensors”, Proceedings of IEEE, vol. 78, June 1990, p. 973. Bansal R. (ed.), Handbook of Engineering: Electromagnetics, Dekker Publ., New York, 2004.

Articles

VOLUME 3,

N° 4

2009


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

FABRICATION OF PHOTONIC CRYSTAL STRUCTURES BY LASER LITHOGRAPHY Vygantas Mizeikis, Kock Khuen Seet, Saulius Juodkazis, Hiroaki Misawa

Abstract: Using femtosecond laser lithography technique, we have fabricated 3D PhC structures having woodpile and spiral architectures in photoresists, and investigated their photonic properties. The fabricated structures exhibit good structural quality, evidenced by long-range periodicity and absence of deformations, and photonic stop-gaps at nearinfrared and infrared wavelengths, evidenced by spectral bands of high reflectivity. These structures can be used as templates for further enhancement by infiltration of metals or high refractive index materials, such as Si. Keywords: photonic crystals, laser lithography.

1. Introduction Photonic crystals (PhC) are periodically structured dielectric/metallic materials, which offer possibilities to control propagation, absorption and spontaneous emission of optical radiation via photonic band gap (PBG) effect [1], [2]. Since PBG occurs at wavelengths close to the PhC lattice period [3], periodic structuring of materials with high resolution is required in order to fabricate functional PhC structures. This is a challenging task, especially in the case of three-dimensional (3D) PhC structures with PBG at optical (visible to infrared) wavelengths. Laser lithography technique is natively suited for 3D fabrication, and therefore is attractive as a tool for fabrication of 3D PhC structures [4]. It is based on drawing in photosensitive media (such as photoresists) by translating focal spot of a tightly-focused laser beam, and blends achievements of microscopy, non-linear optics, and photochemistry [4]. Here we demonstrate fabrication by this technique, called direct laser writing (DLW), of PhC structures having complex 3D woodpile and spiral architectures in photoresists, and describe their structural and photonic properties.

2. Fabrication technique and samples Fabrication by DLW technique involves translation of laser beam focal spot along a pre-defined path, such that linear photomodified regions in 3D space are drawn [x]. The samples were films of SU-8 (NANO™, Microchem) [5] formulation 50, spin-coated to a 50 mm thickness on glass substrates. SU-8 is an epoxy-based negative ultrathick photoresist which is widely used for the singlephoton lithography of high aspect ratio micro-mechanical devices. Single-photon absorption in SU-8 is negligible at wavelengths longer than 400 nm [6]. As irradiation source, Hurricane X (Spectra Physics) Ti:Sapphire laser system with a pulse length Dtpulse=130 fs, a central

Fig. 1. (a) Structural parameters of a woodpile Dd-distance between the rods, Dz-distance between the layers, m - number of layers, (b, c)~SEM images of woodpile structure recorded in SU-8 by DLW, (d) demonstration of linear defect with 90° bend, (e) spiral structure and its main parameters: L- length of the spiral arms, c-lattice period in the z-axis direction, a-period 2D square lattice in the x-y plane, (f, g) SEM images of quare and circular spiral structures. Articles

99


Journal of Automation, Mobile Robotics & Intelligent Systems

VOLUME 3,

N° 4

2009

wavelength lpulse=800 nm, and a repetition rate of 1 KHz was used. Therefore, two-photon absorption is responsible for the photomodification involving photo-acid generation and polymer cross-linking which renders SU-8 stable against subsequent chemical development. The fabrication was performed in Olympus IX71 microscope with oil-immersion objective lens having numerical aperture NA=1.4. 3D drawing was accomplished by translating the sample using a piezoelectric transducer-controlled translation stage along 3D trajectory defined with a few nanometers accuracy using personal computer. More details on the fabrication can be found in the literature [7,8]. The samples were inspected using a scanning electron microscope (SEM), JSM-6700FT (Jeol). Their optical reflection at infrared wavelengths was measured using Fourier-Transform Infra Red (FT-IR) spectrometer (Valor III, Jasco) with a microscope attachment (Micro 20, Jasco) equipped with a 16×, NA=0.5 objective lens.

3. Results and discussion Woodpile architecture [9] and definition of its parameters are shown in Fig. 1(a). Woodpiles can be formed by stacking layers of uniformly spaced dielectric rods along the z-axis direction. Fig. 1(b, c) shows SEM images of the woodpile sample recorded in SU-8 at a pulse energy of 0.55 nJ and a focal spot scanning step of 80 nm. The sample is a perfect parallelepiped with dimensions of (48×48×21) μm, the individual SU-8 rods have smooth surfaces and elliptical cross-sections with diameters of 0.5 mm in the x-y plane and 1.3 mm along the z-axis. Their elongation in the focusing direction by a factor of about ~2.6 is due to the ellipsoidal shape of the focal region. Fig. 1(d) demonstrates woodpile structure with waveguide type defects formed by missing halves of two rods in the two neighboring top layers. Though refractive index of SU-8 is too low to achieve real waveguiding, this example illustrates that defects with required geometry can be easily created by simply shutting off the laser beam for appropriate time intervals during the recording. Spiral architecture [10], illustrated schematically in Fig. 1(e) consists of spirals that have square or circular shapes with parameters defined Fig. 1(e). Extended spiral structures are generated by centering the spirals on the nodes of a two-dimensional square lattice with period a. The structures are diamond-like, and yield spectrally wide PBGs [10]. Using DLW, these, and even more complex structures consisting of intertwined/circular and phaseshifted spirals can be obtained. Fig. 2(f) shows SEM image of a sample with size of (48×4×830) μm, and parameters a=1.8 μm, L=2.7 μm, and c=3.04 μm, fabricated with pulse energy I=0.6 nJ using the 100 NA=1.35 objective lens. Figure 1(g) shows a circular spiral structure with parameters a= 1.8 μm, L=2.7 μm, c=3.6 μm. Notice the 180° phase shift between the adjacent spirals, which would be impossible to achieve by any other known fabrication technique. Both woodpile and spiral samples have exhibited only minor structural imperfections, such as non-parallelism of the samples' vertical edge directions (about 0.5°), hence they are nearly deformation and shrinkage-free. The structures retain their initial shapes even after being dislodged from the substrates. 100

Articles

Fig. 2. Maxwell's scaling in the measured infrared reflectivity spectra of woodpile (a) and square spiral (b) samples with lattice parameters proportionally scaled down (the lattice parameters and scaling factors are indicated in the plots. Figure 2(a) shows reflection spectra of three samples measured along the z-axis. The measurement direction coincided with z-axis. The structures have lattice parameters proportionally scaled down compared to the first sample (the topmost spectrum) as indicated in the plots. In all spectra two major high reflectance regions can be seen centered at shorter (near 2.0 μm) and longer (near 4.0 μm) wavelengths. Shapes and relative amplitudes of the reflectivity peaks are well reproduced by transfermatrix calculations for model structures with appropriately chosen parameters . Figure 2(b) presents reflection spectra of three spiral samples in the same manner as in Fig. 2(a). Just as for the woodpile samples, pairs of reflectivity peaks are seen in the spectra, and their central wavelengths scale together with the lattice parameters (given in the plots). These peaks are spectrally matched by the dips in the transmission spectra (not shown). The wavelength interval 2.7-3.6 μm is omitted in Fig. 3(b) because it contains some bands of intrinsic SU-8 absorption, which suppress PBGs. The suppression is evident in the topmost spectrum where the short-wavelength peak apparently falls within the absorption band, and therefore is suppressed.


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

Multiple reflectivity peaks in the fabricated samples signify multiple photonic stop-gaps, or forbidden spectral ranges existing along certain directions only. Comparison with photonic band structure calculations indicates that the long-wavelength peaks correspond to the fundamental stop-gaps of these structures, while the short-wavelength peaks represent the second-order stop-gap. Second- and higher-order stop gaps are an evidence of good structural quality achieved, because higher photonic bands are more susceptible to disorder. In addition, they can be exploited for achieving photonic band gap effects in structures with larger structural parameters. Spectral positions of the major reflectivity peaks in these samples depend on the lattice scaling factor in qualitative agreement with Maxwell's scaling, which constitutes clear evidence of their photonic band nature.

[8]

[9]

[10]

N° 4

2009

Seet K.K., Mizeikis V., Matsuo S., Juodkazis S., Misawa H., „Three-dimensional spiral - architecture photonic crystals obtained by direct laser writing”, Adv. Mat., vol. 17, 2005, pp. 541 545. Ho K., Chan C., Sokoulis C., Biswas R., Sigalas M., “Photonic band gaps in three dimensions: New layer-bylayer periodic structures”, Solid State Commun., no. 89, 1994, pp. 413-416. Toader O., John S., “Proposed square spiral microfabrication architecture for large three-dimensional photonic band gap crystals”, Science, no. 292, 2001, pp. 1133-1135.

4. Conclusions The fabricated SU-8 photonic crystal structures have low deformations, sufficient mechanical and chemical robustness, and exhibit substantial signatures of photonic band dispersion. Hence, these structures can be used as templates for the infiltration by other materials with higher refractive index.

AUTHORS Vygantas Mizeikis* - Division of Global Research Leaders (Research Institute of Electronics), Shizuoka University, 3-5-1 Johoku, Naka-ku, Hamamatsu 432-8561, Japan. Tel/Fax: +(81)-53-478-1312. E-mail: dvmzks@ipc.shizuoka.ac.jp. Kock Khuen Seet, Saulius Juodkazis, Hiroaki Misawa Research Institute for Electronic Science, Hokkaido University, CRIS build., Kita 21 Nishi 10, Kita-ku, Sapporo 001-0021, Japan. * Corresponding author

References [1]

[2]

[3]

[4]

[5] [6]

[7]

Yablonovitch E., “Inhibited Spontaneous Emission in Solid-State Physics and Electronics”, Phys. Rev. Lett., no. 58, 1987, pp. 2059-2062. John S., “Strong localization of photons in certain disordered dielectric superlattices”, Phys. Rev. Lett., no. 87, 1987, pp. 2486-2489. Joannopoulos J.D., Johnson S.G., Winn J.N., Meade R.D., Photonic Crystals: Molding the Flow of Light, Princeton University Press, 2008, p. 286. 3D Laser Microfabrication, Principles and Applications, edited by H. Misawa and S. Juodkazis, Willey-VCH, 2007, p. 390. http://www.microchem.com/products/su_eight.htm Witzgall G., Vrijen R., Yablonovitch E., Doan V., Schwartz B., “Single-shot two-photon exposure of commercial photoresist for the production of three-dimensional structures”, Opt. Lett., vol. 23, 1998, pp.1745 1747. Mizeikis V., Seet K.K., Juodkazis S., Misawa H., “Threedimensional woodpile photonic crystal templates for infrared spectral range”, Opt. Lett., vol. 29, 2004, pp. 2061-2063. Articles

101


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N째 4

2009

PLASMONIC NANO-IMAGING WITH METALLIC NANOLENS Atsushi Ono, Prabhat Verma, Satoshi Kawata

Abstract: We proposed a nano-imaging system at optical frequency region with the array of metallic nanorods. We simulated the field distribution of imaging process using the finitedifference time-domain algorithm. It is found that the spatial resolution is 40 nm, which is much beyond the diffraction-limit and is limited by the array pitch. The typical configuration is a hexagonal arrangement with 40 nm periodicity of silver rods of 50 nm height and 20 nm diameter. The image formation highly depends on the coherence and the polarization of the dipole sources, the array pitch, and the source-array distance. The principle of our near-field imaging is based on the longitudinal resonance of the localized surface plasmon along a metallic nanorod. The spectral responses of the device are also investigated. Keywords: plasmonics, near-field optics.

1. Introduction It is common sense that transparent, high refractive, and curved surface materials, like an optical glass, are used as the lens and opaque; high reflective, and flat surface materials like a metal are used as the mirror. Beyond common sense, we proposed metallic nanolens which is constructed by the array of silver nanorods provides an image with subwavelength spatial resolution [1]. Local surface plasmon (LSP) resonance excited on each silver nanorods is utilized for imaging beyond the diffraction limit of light. Electric field enhancement is observed on the metal surface, when light is irradiated to metallic nanostructure [2], [3]. This enhancement is caused by the resonant couplings between electron oscillation and the electric field oscillation of light, is so-called by the surface plasmon. Nanoplasmonics, plasmonic phenomena induced on metallic nanostructure, is very attractive research field. Metallic nanoprobe, metallic nanoshell, and metallic nanospheres have been developed for the local field enhancement, and the optical properties are utilized for plasmonic nanoimaging, plasmonic nanospectroscopy, plasmonic laser, and so on [4]. In this article, we show the imaging mechanism and the imaging characteristics of proposed metallic nanolens. We demonstrated that near-field sources in nanometer scale were plasmonically imaged through the array of silver nanorods by three-dimensional finite-difference time-domain (FDTD) algorithm. Local surface plasmon polaritons of nanorods contribute to the nano-imaging. Furthermore, our simulations show that well-designed metallic nanolens gives color imaging with subwave102

Articles

length resolution [5]. The spatial resolution was 40 nm that was determined by the array pitch. It is expected that plasmonic nanorod array would be applied to optical nanoimaging, nanobiosensor, nanolithography, and nanocommunication.

2. Simulation model The typical configuration of metallic nanolens is the hexagonal arrangement of silver rods of 50 nm height and 20 nm diameter with 40 nm array pitch. The rod diameter and the pitch should be much shorter than the wavelength, while the height can be longer than the wavelength. Drude dispersion formula was applied to FDTD simulation in the reference from experimental value [6] for the consideration of frequency dispersion of silver complex permittivity. The near-field sources in nanometer scale were plasmonically imaged through the metallic nanolens (Fig. 1). Figure 1(a) shows the bird view and the x-z cross sectional view of silver nanorod array. The rod axis is parallel to z axis. Point sources were shaped as letter l as an object (Fig. 1(b)). The point sources were all z-polarized and incoherently oscillating like fluorescent molecules. The oscillation wavelength was 482 nm. Each source is located at the center of a rod. The silver nanorod array is located 10 nm away from the point sources. Figure 1(c) is the intensity distribution at the plane 10 nm away from the top of the rod array. The FWHM of each spot was 30 nm, and letter l is well resolved.

Fig. 1. Subwavelength optical imaging through the metallic nanolens, which is constructed by the array of silver nanorods. (a) model (b) object (c) image.

3. Imaging mechanism Longitudinal resonant mode of LSP excited on each silver nanorods contributes to this super-resolutive imaging. We have examined the field distribution of the elec-


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

tric field components, Ez and Ex, in the vertical cut including a rod center (Fig. 2). It is found that the Ez component is enhanced at the circumference of the top end of the rod and it contributes to provide the hot spot at the image plane as shown in Fig. 2(a). In Fig. 2(b), it is seen that the field of Ex component is enhanced at the side of the rod and fundamental mode of the LSP resonance is excited. The resonant frequency was analyzed by time resolved Fast Fourier Transform (FFT) of the intensity at the image plane which was 10 nm away from the top end of rods on impulse response of z-polarized dipole source (Fig. 3).

N째 4

2009

height were fixed to 20 and 50 nm. Figure 5 shows the peak intensity at the image plane as a function of hexagonal array pitch a. Insets in Fig. 5 show the intensity distribution of the image obtained with the array pitch of 30 and 40 nm. The peak intensity became smaller as the array pitch decreased. When the distance between rods is less than 40 nm, there is the coupling between local surface plasmons in the neighboring rods, and then coupling out or smearing of the peak intensity occurred. Therefore the image became broadened, not forming nanospot, at a less than 40 nm array pitch. When the array pitch is larger than 40 nm, local surface plasmon coupling between rods became negligible and the peak intensity is almost constant that means the spot formed by this nanorod array is as same as the single nanorod case. In conclusion, a pitch of 40 nm is the best pitch for this model, since the image resolution depended on the pitch.

Fig. 2. Intensity field distributions for individual polarization components obtained at the vertical (x-z) cross section of the nanorod array including the centric rod-axis and a zpolarized dipole source. (a) Ez component. (b) Ex component. The size of the scale bar is 50 nm.

Fig. 4. Relationship between the peak intensity (dots) as a function of the source-array distance and exponential fitting for the dot marks (dashed curve). Fig. 3. FFT spectrum of near-field intensity at the image plane. Impulse signal is detected at the image plane through a silver nanorod. A peak is observed at the wavelength of 482 nm.

4. Imaging properties It was investigated that the relationship between the peak intensity at the image plane for the source-array distance to confirm the contribution of near-field components. The peak intensity at the image plane were plotted when the z-polarized dipole source was far from the bottom of the rod array from 2 to 100 nm. Figure 4 shows peak intensity in the image plane as a function of sourcearray distance. The peak intensity decreased in proportional to the source-array distance and it fitted well to an exponential decay curve (dashed curve in the figure). This indicates that a major contribution to plasmon coupling with the dipole radiation is the evanescent components of near-field photons. The array pitch of the nanorod array contributes to the spatial resolution for the optical nano-imaging. Therefore the narrow pitch is better for super-resoltive imaging. The image field distribution dependence on the array pitch was investigated when the rod diameter and

Fig. 5. Hexagonal array pitch dependence on the peak intensity at the image plane. Insets show the intensity distributions of the image obtained with the pitch of 30 and 40 nm.

5. Conclusion In conclusion, subwavelength image transfer process by silver nanorod array was discovered and proposed. The theoretical calculation work was done and the mechaArticles

103


Journal of Automation, Mobile Robotics & Intelligent Systems

nism of the super-resolutive optical imaging through plasmonic nanorod array was clarified by three-dimensional FDTD algorithm. The spatial resolution is limited by not the wavelength but the array pitch. The resolution in the case of used parameters was six times higher than that of conventional diffraction-limited optics. It was found that the image formation depends on the source-array distance and the polarization of the dipole sources, array pitch, and the coherence by the FDTD simulation.

AUTHORS Atsushi Ono* - Division of Global Research Leaders, Shizuoka University, 3-5-1 Johoku, Naka-ku, Hamamatsu 432-8561, Japan. E-mail: daono@ipc.shizuoka.ac.jp. Prabhat Verma - Department of Frontier Biosciences, Osaka University, 2-1 Yamadaoka, Suita, Osaka 565-0871, Japan. E-mail: verma@ap.eng.osaka-u.ac.jp. Satoshi Kawata - Department of Applied Physics, Osaka University, 2-1 Yamadaoka, Suita, Osaka 565-0871, Japan. E-mail: kawata@ap.eng.osaka-u.ac.jp. * Corresponding author

References [1]

[2] [3] [4]

[5]

[6]

104

Ono A., Kato J., Kawata S., “Subwavelength Optical Imaging through a Metallic Nanorod Array”, Phys. Rev. Lett., vol. 95, 2005, no. 267407. Shalaev V.M., editor, Optical Properties of Nanostructured Random Media, Springer, Berlin, 2002. Kawata S., editor, Near-field and Surface Plasmon Polaritons, Springer, Berlin, 2001. Ichimura T., Hayazawa N., Hashimoto M., Inouye Y., Kawata S., “Tip-enhanced coherent anti-Stokes Raman scattering for vibrational nano-imaging”, Phys. Rev. Lett., vol. 92, 2004, no. 220801. Kawata S., Ono A., Verma P., “Subwavelength colour imaging with a metallic nanolens”, Nature Photonics, vol. 2, no. 7, 2008, pp. 438-442. Johnson P.B., Christy R.W., "Optical Constants of the Noble Metals", Phys. Rev. B, vol. 6, no. 4370, 1972.

Articles

VOLUME 3,

N° 4

2009


Journal of Automation, Mobile Robotics & Intelligent Systems

VOLUME 3,

N° 4

2009

OVERHAUSER EFFECT AND ANISOTROPY OF ELECTRON SPIN G-FACTOR IN GAAS / ALGAAS QUANTUM WELLS Tetsu Ito, Wataru Shichi, Masao Ichida, Hideki Gotoh, Hidehiko Kamada, Hiroaki Ando

Abstract: To investigate the dependence of electron g-factor on magnetic field in GaAs / AlGaAs quantum wells time-resolved photoluminescence measurements under a high magnetic field in different experimental configuration, the magnetic field perpendicular (g^) and parallel (g||) to the quantum confinement direction, has been studied. When the angle between the magnetic field and the confinement direction is 45°, the precession frequency varies depending on polarity of magnetic field and the circular polarization + type of excitation light (s or s ). We found that these dependences of the precession frequency exhibit main features of Overhauser effect with an effective magnetic field of 0.5 T that nuclear spins react back on electron spin precession and the g-factor value is not affected by the effective magnetic field. The g^ and g|| values agree well with the results of four-band k·p perturbation calculations. Keywords: quantum well, electron spin g-factor, Overhouser effect.

1. Introduction In quantum structures such as quantum wells (QWs) and dots fundamental properties of the semiconductors are altered by quantum confinement effects. Among these properties Landé g-factor of electrons is one of the most sensitive one susceptible to the quantum confinement. The electron Landé g-factor, which represents spin-magnetic field interaction, has been studied in semiconductor structures [1]-[5]. The electron g-factor in QWs exhibits anisotropy, reflecting symmetry of the quantum confinement [2]-[7]. The electron g-factor (g^) for a magnetic field perpendicular to the direction of quantum confinement is assessed in the Voigt configuration either by means of time resolved photoluminescence (PL) measurements [2], [5], [7] or by pump-probe optical nonlinear measurements [3]-[4]. To precisely assess the electron g-factor (g||) for the magnetic field parallel to the confinement direction an experimental setup, where the angle between the magnetic field and the confinement direction is 45°, is required [3]. In this paper we will discuss nuclear magnetic effects on the spin precession and the electron g-factor using experimental data obtained in the 45° configuration [3], [4], [7].

2. Experimental In experiments to assess the quantum confinement effects accurately and systematically under the same conditions we have used a sample, which consists of QWs having different well widths. In this sample each QW

consists of GaAs well and Al0.35Ga0.65As barrier layers. Time resolved PL measurements were carried out to observe the electron spin precession under a high magnetic field in Voigt configuration at 4 K. The PL traces are obtained by spectrally integrating around the emission peak using a streak camera. The QW sample was irradiated by 2 ps pulse laser with a photon energy of 1.8 eV, and a repetition rate of 80 MHz. Estimated excitation power density 2 was 240 W/cm . The pump pulse is circularly polarized to achieve spin selective excitation. The circular polariza+ tion components s and s in PL were selectively measured by using l/4 wave plate and polarizer. Figure 1 shows typical time evolution of optical anisotropy, defined as + s - s , obtained for the QW with 15 nm well width. A simple spin relaxation is observed under no magnetic field as shown in Fig. 1 by dotted line. The spin relaxation time is assumed to be around 800 ps. With increasing the magnetic field we have observed an oscillation caused by electron spin precession. We have estimated the spin precession frequency from the oscillation in the PL time evolution. The precession frequency increases in proportion to the applied magnetic field, as shown in Fig. 2. The electron g-factor values are derived from the proportionality constant for each QW with different well width. The measured g-factor value in Voigt configuration, g^ is shown in Fig. 3 as a function of well width with filled circles. Dotted line indicates the g-factor value for the GaAs bulk sample. The observed value is very close to the reported value of -0.44 [1], [8]. The g-factor value increases monotonically with a decrease in the well width crossing the zero level between the well widths of 8 and 6 nm. To precisely assess the electron g-factor (g||) for the magnetic field parallel to the confinement direction we employed an oblique experimental configuration [3]. In the oblique configuration magnetic field was applied so that the angle between the directions (polarities) of quantum confinement and of magnetic field may be 45° or 135°. Figure 4 shows measured precession frequency as a function of magnetic field for the QW with a well width of 15 nm. In negative magnetic field we plotted the precession frequency with negative sign. The precession frequency is proportional to the magnetic field with same proportionality coefficients. However, the precession frequency has difference at the same magnetic field de+ pending on the polarization type of excitation light (s or s ). From the proportionality coefficient the electron g-factors in 45° configuration, g45, has been obtained. The dependence of the g|| value on well width is shown in + Figure 3. Here, circularly polarized s light is used for optical excitation. The values of g|| were assessed based on measured g^ and g45 values using the relational Articles

105


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

2

2

2

equation, g|| = 2g45 - g^ [3]. To explore the physical origin of the dependences of precession frequency on the polarization type, temporal change in the spin precession frequency has been assessed. We performed time-resolved PL measurements in picosecond scale. The PL traces were measured at 0, 10, and 20 minutes after changing + excitation polarization from s to s . Exposure time was 2.5 minutes and applied magnetic field was +3 T. The sample was kept illuminated during the measurements. While the change in the precession oscillation from 0 to 10 minutes was evident, the difference was negligible from 10 to 20 minutes. Since the precession oscillation did not seem to be averaged out in exposure time of 2.5 minutes the response time of this effects is considered to be longer than a few minutes.

+

N° 4

2009

the Bohr magneton. When the electron g-factor ge is negative and the hyperfine interaction constant A is positive, as in the GaAs / Al0.35Ga0.65As QW having 15 nm well width, the direction of effective field is anti-parallel to the nuclear spins. The precession frequency lowering caused + by the change in the excitation polarization from s to s (Fig. 4) is consistent with the main feature of effective magnetic field. From the intercept of the fitted lines in Fig. 4 the effective magnetic field of BN = 0.5 T is evaluated. The time scale of building up of nuclear polarization caused by the Overhauser effect, is reportedly from a few minutes to several ten of minutes [3], [4]. The observed transient change in the precession oscillation is also within this time region. Finally we discuss the anisotropy in the electron gfactor. The calculated results obtained by four-band k·p perturbation method [6] are shown in Fig. 3 with solid curve (g^) and dashed curve (g||). The trend of the experimental results is well reproduced by the theoretical analysis. The anisotropic electron g-factor in QWs can be attributed to the spin dependent coupling efficiency of valence states to conduction electron states via k·p perturbation [6].

-

Fig. 1. Time traces of optical anisotropy, defined as s - s , measured for the GaAs/Al0.35Ga0.65As QW with a well width of 15 nm.

Fig. 3. Measured values of g^ (filled circles) and g|| (open circles) as a function of well width for GaAs / Al0.35Ga0.65As quantum wells. Solid and dashed curves are calculated g^ and g|| values, obtained by using four-band k•p perturbation theory.

Fig. 2. Precession frequency as a function of magnetic field for various well widths of GaAs/Al0.35Ga0.65As QWs.

3.Discussions In general nuclear spin effects and electron-electron spin interaction are well known that affects the precession frequency. Especially the nuclear spin alignment induced by the Overhauser effect gives considerable effects to the precession frequency through hyperfine interaction [3], [4]. In the oblique configuration a net electron spin component S||, parallel to the applied magnetic field, persists over the lifetime of the spin precession. The hyperfine interaction AI×S|| between the net electron spin and the nuclear spin builds up a nuclear spin alignment áIñ. This reacts back on the electron spins as an effective magnetic field BN = AáIñ / gemB, which modifies the total magnetic field experienced by the electron spins. Here mB is 106

Articles

Fig. 4. Measured precession frequency as a function of magnetic field for the GaAs / Al0.35Ga0.65As quantum well with a well width of 15 nm. The + and - signs mean the polarity of the magnetic field. Filled and open circles are the results + for the excitation light of circular polarization s and s , respectively.


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

4. Conclusion We have studied the dependence of electron spin gfactor on magnetic field in GaAs / AlGaAs QWs by time resolved PL measurements. In the 45° configuration we have found that precession frequency varies depending on the polarity of applied magnetic field and the circular polarization of the excitation light. These dependences of precession frequency can be explained by the Overhauser effect with an effective magnetic field of 0.5 T. The values of g^ and g|| are qualitatively reproduced by a four-band k·p perturbation calculations. Our results indicate that the nuclear field effects are observed not only in the pump-probe nonlinear measurements but also in simple PL measurements at oblique configuration as an effective magnetic field. When we evaluate the electron g-factor from the proportionality factor in the dependence of the precession frequency on external magnetic field, the effect of the nuclear field is not affect on the electron g-factor.

[6]

[7]

[8]

N° 4

2009

electron spin g-factor in semiconductor quantum well structures”, Phys. Stat. Sol. C, vol. 3, no. 10, 2006, pp. 3496-3499. Ivchenko E.L., Kiselev A.A., “Electron g factor of quantum wells and superlattices”, Sov. Phys. Semicond., vol. 26, no. 8, 1992, pp. 827-831. Ito T., Shichi W., Nishioka Y., Ichida M., Gotoh H., Kamada H., Ando H., “Dependence of electron spin gfactor on magnetic field in quantum wells”, J. Limin., vol. 128, no. 5-6, 2008, pp. 865-867. Kosaka H., Kiselev A.A., Baron F.A., Kim K.W., Yablonovichi E., “Electron g factor engineering in III-V semiconductors for quantumcommunications”, Electron. Lett., vol. 37, no. 7, 2001, pp. 464-465.

ACKNOWLEDGMENTS The authors would like to thank Professor Hidenori Mimura for his encouragements and helpful suggestions.

AUTHORS Tetsu ItoA,B*, Wataru ShichiC, Masao IchidaB,C, Hideki GotohD, Hidehiko KamadaD, Hiroaki AndoB,C A Division of Global Research Leaders, Shizuoka University, 3-5-1 Johoku Naka-ku Hamamatsu 432-8561, Japan. E-mail: ditto@ipc.shizuoka.ac.jp. B Quantum Nano-Technology Laboratory, Konan University, Konan University, 8-9-1 Okamoto Higasinada-ku Kobe 658-8501, Japan. C Graduate School of Natural Science, Konan University, Konan University, 8-9-1 Okamoto Higasinada-ku Kobe 658-8501, Japan. D NTT Basic Research Laboratories, NTT Corporation, 31 Morinosato Wakamiya, Atsugi-shi, Kanagawa 2430198, Japan. * Corresponding author

References: [1]

[2]

[3]

[4]

[5]

Snelling M.J., Blackwood E., McDonagh C.J., Harley R.T., Foxon C.T.B., “Exciton, Heavy-hole, and electron g factors in type-I GaAs/AlxGa1-xAs quantum wells”, Phys. Rev. B, vol. 45, no. 7, 1992, pp. 3922-3925. Hannak R.M., Oestreich M., Heberle A.P., Ruhle W.W., Kohler K., “Electron g factor in quantum wells determined by spin quantum beats”, Solid State Commun., vol. 93, no. 4, 1995, pp. 313-317. Malinowski A., Harley R.T., “Anisotoropy of the electron g factor in lattice-mached and strained-layer III-V quantum wells”, Phys. Rev. B, vol. 62, no. 3, 2000, pp. 2051-2056. Salis G., Fuchs D.T., Kikkawa J.M., Awschalom D.D., Ohno Y., Ohno H., “Optical manipulation of nuclear spin by a two-dimensional electron gas”, Phys. Rev. Lett., vol. 86, no. 12, 2001, pp. 2677-2680. Ito T., Shichi W., Morisada S., Ichida M., Gotoh H., Kamada, Ando H., “Quantum confinement effects on Articles

107


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

STUDY OF LOW TEMPERATURE INACTIVATION OF SPORES USING MICROWAVE EXCITED AIR PLASMA Masaaki Nagatsu, Mrityunjai K. Singh, Ying Zhao, Yuya Fujioka, Akihisa Ogino

Abstract:

2. Experimental

In the present study, we investigated the inactivation characteristics and contribution of different inactivating factors generated in a low temperature and low pressure nitrogen, oxygen and air-simulated plasma for the inactivation of Geobacillus stearothermophilus spores. We used three optical filters i.e. thin quartz (l>180 nm), lithium fluorine (LiF, l>120 nm) and Pyrex glass (l>300 nm) plates to identify the most efficient wavelength range. The effect of optical radiations alone was studied through placing a small isolated and evacuated chamber with spore sample inside the plasma chamber.

The experimental setup used for the sterilization tests consists of a stainless steel cylindrical vacuum chamber having a diameter of 400 mm and a height of 400 mm with a microwave launcher fed from a 2.45 GHz microwave generator, as shown in Fig. 1 [5]. The launched microwave power varied from 0.2-3 kW. A small chamber of 150 mm in length and 70mm in width was placed inside the processing part of main experimental setup. The top of small chamber was covered with a filter which is 30 mm in diameter and 1.5 mm of thickness to stop the radicals and let the VUV/UV go through. As already mentioned, plasma emits electromagnetic radiation ranging from far ultraviolet to the infrared, and the far-UV (100 nm<l <200 nm) is very important component of VUV emission from plasma, since these photons have energy which greatly exceeds that of all chemical bonds in organic molecules.

Keywords: plasma sterilization, low temperature plasma, microwave plasma, air plasma, VUV, radicals.

1. Introduction The low temperature plasma processing offers a potential approach to sterilize the exposed as well as wrapped medical instruments, because many medical instruments and the wrapping materials are made of polymers. Such materials can be easily damaged by thermal treatments i.e. dry heat, steam autoclave techniques. Other low temperature techniques such as ethylene oxide (EtO) and gamma irradiation etc. pose the formation of toxic by-products and material degradation. The plasma sterilization technique overcomes many inherent limitations of the conventional techniques through low temperature treatment of the heat-sensitive objects, short treatment time, safe operation and no toxicity after processing etc. There are several mechanisms which may be responsible for the sterilization: interaction of UV radiation with the DNA of the cell (below l=275 nm), the photons have enough energy to break bonds of C-C (3.8 eV) or C-H(4.5 eV) in a solid and are known to be able to induce strand breaks in the deoxyribonucleic acid [1], removal of the material of the cells by reactive species (oxygen atoms) and the interaction of these two mechanisms [2]. In our previous work, a six-log reduction in spores could be achieved only several minutes irradiation with a lowpressure oxygen/air simulated surface-wave plasma and the chemical etching reaction from the reactive oxygen radicals and UV emission from excited nitrogen atom and molecules make more efficient inactivation rate [3]-[5]. In this work, we follow up to previous study, in order to investigate the contribution of various effecters special VUV/UV radiation in the inactivation of bacterial cells by plasma.

108

Articles

Slot antenna

Punched metal plate

Fan

Plasma

z z Plasma

2.45GHz Microwave

LiFor glass

Biological indicator

Small metal chamber Punched plate

Plug valve

Fig. 1. Microwave excited surface-wave plasma device [5]. So, the LiF filter was used to get the emission spectra above 120 nm. The Biological Indicators (BI, for short, Geobacillus stearothermophilus spores with a population of 2.5Ă&#x2014;106 on a small stainless disc) was put in small chamber and treated with surface-wave plasma (200 sccm flow rate at 10 Pa pressure and 700 W microwave power with on/off time as 20/60 sec). The substrate stage is about 15 cm below the quartz window. Two different pressures in the small chamber were compared each other, one is opening the plug valve to make the same pressure as surrounding condition. The other is closing the plug valve and pumping down about 10-3 Pa in order to prevent any other plasma producing radical in the small chamber and the gas absorption for UV photons. After plasma treatment, the spores were removed from the stainless disc and putting them into culture tubes. They


Journal of Automation, Mobile Robotics & Intelligent Systems

were transferred to Petri dish and incubated at 50-55째C for 48 h. In order to clarify and better understanding the mechanism of sterilization, we use optical emission spectrometry to evaluate the characteristic of plasma under filter and SEM to analyze the changing of spores shape.

VOLUME 3,

N째 4

2009

because some spores stacked together and formed a multilayered structure as protected layer for UV and radicals.

3. Results and discussions For plasma diagnostics, optical emission spectroscopy (OES) measurements were performed using the VUV to IR monochromator systems VM-502 (Acton Research Corporation) to measure the vacuum ultraviolet part of the spectrum in the wavelength range from 100 nm to 500 nm. The MgF2 window was used as the vacuum window in the port between the vacuum chamber and OES device. Figure 2 shows VUV/UV and visible light emission spectra from pure oxygen plasma at a power of 700 W and pressure of 10 Pa used in sterilization experiment. The most significant oxygen lines of these spectrum are the 130.5 nm line which correspond to the transition 3s 3S--2p4 3P [6] and 777.2 nm line which correspond to the transition 3p5p--3s3S [7]. From the Fig. 3, there were three main regions of high efficacy in the case of air-simulated gas plasma, one was the radiation from 200 nm to 250 nm wavelength range was identified to be NO-g radiation, from 300 nm to 400 nm wavelength range was identified to be N2 second positive system, and near the VUV range there was an important Nitrogen line that was around 120 nm (transition 3s 4P--2p3 4S). The short wavelength VUV/ UV penetrates into the protective layer of the microorganisms, and destroys DNA [8]. But the longer wavelength region is comparably in-effective than the shorter wavelength UV. It should be noted that the longer wavelength emission shows smaller antibacterial effect [9].

Fig. 2. Emission spectrum of pure oxygen plasma at 700 W microwave power and 10 Pa pressure. Figure 4 shows the colony count results of the Geobacillus stearothermophilus spores irradiated with pure oxygen plasma and air simulated plasma. As shown in the Fig. 4, the case of direct plasma exposure provides survival plots with 3 different linear segments. After 2-min treatment, only 1% Geobacillus stearothermophilus colony was observed and there was no colonies growing on the agar plate at all after 5-min treatment time in the air simulated plasma and 10-min treatment time in the pure oxygen plasma. From the second linear segment, it took a longer time for inactivating the rest 1% survivors, that

Fig. 3. Emission spectrum of air simulated plasma at 700 W microwave power and 10 Pa pressure.

Fig. 4. Survival curves of Geobacillus stearothermophilus spores by oxygen plasma and air simulated plasma at 700 W microwave power.

Fig. 5. Survival curves of Geobacillus stearothermophilus spores by oxygen plasma with a LiF filter at 700 W microwave power. The experimental results show that the sterilizing efficiency of air simulated plasma is better than oxygen plasma. Previous studies about plasma emission spectra indicated that the effect of UV emission in air simulated plasma under identical electrical discharge conditions is stronger than that in oxygen plasma, because there were NO-g system UV emission, N2 second positive system UV emission and VUV emission due to nitrogen atom in air simulated plasma besides VUV due to oxygen atom. VUV/UV Articles

109


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

acts synergistically with the reactive radicals and accelerate the elimination rate of microorganisms.

N° 4

2009

most no change of the size after 20 min treatment. It also shows that there is no significant erosion of the microorganisms when only VUV/UV radiation exists in the case of vacuum and the VUV/UV is only important factors responsible for inactivation. It is interesting to note that the spore sizes slightly decreased when oxygen gas was filled at 10 Pa in the small chamber with LiF filter, as shown in Fig. 7(c). The size of the spores decreased from original after 30 min treatment. It means there is some erosion of the microorganism due to the oxygen radicals excited by the VUV emission.

4. Conclusion

Fig. 6. Survival curves of Geobacillus stearothermophilus spores by air simulated plasma with a LiF filter at 700 W microwave power. The effects of VUV/UV radiation emitted from the plasma results were shown in Fig. 5 and Fig. 6. In Fig. 5, the colony forming units of the BIs, which were set inside the small chamber with a LiF filter at different pressure conditions, were compared as a function of oxygen plasma treatment time. We calculated the D value of the first segment of survival curves as approximately 4.5 min and 3.5 min. After 50 min treatment time, there were no colonies growing on the agar plate at all under a LiF filter. Compared with the pressure effect on VUV/UV sterilization with LiF filter, we can find that the vacuum condition was more efficient than the low-pressure condition, because in the low-pressure case, gas molecules can absorb VUV/UV photons and reducing the intensity of VUV/UV reaching the BIs. In Fig. 6, the number of colony forming units for air simulated plasma case was shown. In this case, the low-pressure condition was more efficient than the vacuum condition and the number of survivors decreased by more than two orders of magnitude. In the vacuum case, only VUV/UV emission contribute to the sterilization, on the other hand, in low-pressure case, there were some radicals produced by VUV inside the small chamber and their synergistic action to the spore mortality causes an additional role in inactivation of spores.

In this research, we discussed the effects of VUV/UV radiation and radicals on low-temperature sterilization in surface-wave plasma. The results showed that the VUV/ UV photons act synergistically with the radicals make more efficient inactivation of the Geobacillus stearothermophilus spores by air simulated plasma directly. We also tested the inactivation experiment depending on the VUV/UV radiation in two different pressures to evaluate the role of VUV/UV radiation in plasma based CFUs and SEM analysis. From the present study, it was concluded that The VUV/UV emission plays an important role for sterilization and there is no erosion of the microorganisms when only VUV/UV radiation exists. ACKNOWLEDGMENTS This work was partly supported by Grant-in-Aid for Scientific Research from Japan Society for the Promotion of Science and partly supported by the 21st century COE program by JSPS.

AUTHORS Masaaki Nagatsu*, Mrityunjai K. Singh, Ying Zhao, Akihisa Ogino - Graduate School of Science and Technology, Shizuoka University, Hamamatsu, 432-8561, Japan. E-mail: tmnagat@ipc.shizuoka.ac.jp. Yuya Fujioka - Graduate School of Engineering, Shizuoka University, Hamamatsu, 432-8561, Japan. * Corresponding author

References [1]

[2]

(a)

(b)

(c)

Fig. 7. SEM images of (a)untreated Geobacillus stearothermophilus spores and those irradiated by oxygen plasma with LiF filter for (b)20 min in vacuum and (c)30 min in O2 gas. We also carried out the SEM analyses of spores treated by the plasma to investigate the effects on the change of spores shape by VUV/UV radiation. Figure 7 shows the SEM images of the spores before and after pure oxygen plasma treatment with filter. In the vacuum case, there was al110

Articles

[3]

[4]

[5]

Munakata N., Sato M., Hieda K., “Inactivation action spectra of bacillus subtilis spores in extended uv wavelength (50-300nm) obtained with synchrotron radiation”, Photochem. Photobiol., no. 54, 1991 pp 761768. Kylian O., Hasiwa M., Rossi F., “Effect of low-pressure microwave discharges on the pyrogens bioactivity”, IEEE Trans. on Plasma Science, no. 34, 2006 pp 26062610. Nagatsu M., Terashita F., Nonaka H., Xu L., Nagata T., Koide Y., “Effects of oxygen radicals in low-pressure surface-wave plasma on sterilization”, Appl. Phys. Lett., vol 86, 2005 211502-1211502-3. Nagatsu M., Terashita F., Koide Y., “Low- temperature sterilization with surface-wave-excited oxygen plasma”, Jpn. J. Appl. Phys., no. 42, 2003, pp. L856-L859. Singh M.K., Ogino A., Nagatsu M., “Inactivation Factors of Spore Forming Bactria Using Low-pressure Microwave


Journal of Automation, Mobile Robotics & Intelligent Systems

[6]

[7]

[8]

[9]

VOLUME 3,

N° 4

2009

Plasmas in N2 and O2 Gas Mixture”, New J. Phys. (in print). Piper L.G., Clyne M.A.A., Monkhouse P.B., “Electronic Energy Transfer Between Metastable Argon Atoms and Ground-State Oxygen Atoms”, J. Chem. Soc., Faraday Trans., no. 78 1982, p. 1373. Ban K., Mutoh H., Katayama K., Akitsu T., “Antibacterial effect of oxygen/nitrogen radicals and UV-radiation in microwave-excited plasma”, IEEE International Conference on Systems, Man, and Cybernetics, vol . 2, 2006 pp. 1431-1436. Philip N., Saoudi B., Crevier M.-Ch., Moisan M., Barbeau J., Pelletier J., “The respective roles of UV photons and oxygen atoms in plasma sterilization at reduced gas pressure: the case of N2-O2 mixtures”, IEEE Trans., vol. 30, 2002 pp. 1429-1436. Nicholson W.L., Munakata N., Horneck G., Melosh H.J., Setlow P., “Resistance of Bacillus endospores to extreme terrestrial and extraterrestrial environments” Microbiology and Molecular biology Review, vol. 64 2000 pp. 548-572.

Articles

111


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

ZNO FILMS: PROPETIES DETERMINED BY ELECTRONIC MICROSCOPY AND ELLIPSOMETRY Mykhailo Rakov, Leonid Poperenko, Vasyl Tkach, Iryna Yurgelevich

Abstract: Zinc oxide is a good material for application in nanoand optoelectronics due to its notable features, for example, large band gap. In this work ZnO films deposited by reactive magnetron sputtering at various pressure of residual gases and different temperatures of the substrate are investigated. The spectral dependences of ellipsometric parameters D and Y of the films are determined by ellipsometry. The effective values of optical constants n and k are calculated. The roughness and the texture of the surfaces are obtained by Atomic-Force Microscopy (AFM) and Scanning Electronic Microscopy (SEM). One indicates that the refractive index decreases when reducing the pressure of residual gases, and the roughness decreases when elevating the temperature of the substrate. Thus, the behavior of some properties of the films at various conditions of deposition is determined. Keywords: reactive magnetron sputtering, Beattie’s method, semi-infinite medium, AFM, SEM, nanostructures.

1. Introduction Nowadays zinc oxide is widely used in optoelectronics due to following features [1]: 1) Effective recombination due to excess of strongly coupled excitons; 2) Large band gap (3,37 eV) and big exciton binding energy (60 meV); 3) Piezoelectricity because of non-central symmetry; 4) The ability to change conductivity in presence of various gases; 5) Possibility of low-temperature growth (less then 1000 K), high radiative stability. Therefore, ZnO films are used as: transparent electric contacts, ultraviolet light emitters, luminescent matter, piezoelectric transistors and gas sensors. Besides, Al doped zinc oxide is used in sun batteries. In addition, zinc oxide is probably the richest family of nanostructures between uniform-sized ones (including carbonic nanotubes) so it can be used in nanoelectronics as well. The properties of ZnO films strongly depend on conditions of production so the task is to analyze that influence. In this work, properties of ZnO films deposited at various pressures of residual gases and various temperature of the substrate are investigated.

2. Experiment technique Five samples were produced by reactive magnetron sputtering at SiO2 substrates. The O2 and Ar partial pres112

Articles

sures were 1.7×10-3 mbar and 7.4×10-3 mbar, respectively. The thickness of the substrate was about 210 nm. The samples 1-4 were deposited at room temperature of the substrates whereas the substrate for sample #5 had the temperature of 550 °C. The pressures of residual gases were: 2.7×10-5 mbar (#1), 3.4×10-6 mbar (#2), 2.2×10-7 mbar (#3), 9.7×10-6 mbar (#4) and 2.5×10-7 mbar (#5). Such a vacuum was achieved by using liquid nitrogen traps for chilling the water vapor. The ellipsometric measurements for samples #1, 3 and 5 were done by Beattie’s method [2] within the spectral range 1-4.3 eV at room temperature. Thereto, two light sources were used: deuterium lamp for the range 3.4-4.3 eV and incandescent lamp for the range 1-3.4 eV. The method is based upon the following. The light transmits the system “polarizer-sample-analyzer”. Then the photoelectronic multiplier with the recording system measures the light intensity for the azimuth of the polarizer YP=45°C and the azimuth of the analyzer YA=0°C, 45°C and 90°C. If the obtained intensities are denoted as I1, I2 and I3, the ellipsometric parameters D and Y are calculated as [2]: ,

.

The Atomic-Force Microscopy done on NANOSCOP-3-a gives the information about the roughness and the texture of the surfaces (the resolution is about 5 nm). The Scanning Electronic Microscopy done on Carl Zeiss EVO SO XVP determines the composition of the films and enables us to observe the nanostructures (the resolution is about 1 nm).

3. Experimental results

The spectral dependences of cosD and tgY for samples #1, 3 and 5 were obtained. Their shape has moderate similarity whereas the values are different. The typical graphs for cosD and tgY are shown in the Fig. 1. One can divide these in two parts. The oscillations are observed in the range of 1-3.4 eV so this range is idle. As band gap is 3.37 eV, the absorption begins and the effective values of optical constants n and k in the range of 3.4-4.3 eV can be calculated. The model of semi-infinite medium is used to this effect. The appropriate equations are [3]


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

In these equations,

The spectral dependences of n and k within the range 3.4-4.3 eV for three aforementioned samples are shown in the Fig. 2. As the precision of the experiment was moderate, some dependences came out as nearly similar ones.

N° 4

2009

The images of the surfaces obtained by AFM are shown in Fig. 3. The surfaces of samples #1 and #3 are nearly similar but the structure of sample #3 is a little more disordered. On the other hand, the image of sample #5 indicates clearly granular structure.

a)

a)

b) b)

Fig.1. Typical spectral dependences of cosD (a) and tgY (b) within the range 1-4.3 eV. a) c)

b) Fig. 3. The images of surfaces obtained by AFM for the films deposited at: a) RT, 2.7×10-5 mbar; b) RT, 2.2×10-7 mbar; c) 550 °C, 2.5×10-7 mbar. SEM shows clear crystallites in sample #5 (Fig. 4,b) and nanostructures in samples #2,3 and 4 (the typical image is Fig. 4,a and it corresponds to sample #4). Besides, sample #5 has the excess of carbon in it (~15%).

Fig.2. Spectral dependences of n (a) and k (b) for the films #1,3 and 5. Articles

113


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

Finally, as the thickness is several hundreds of nanometres, the model of semi-infinite medium is quite applicable for description of optical constants of the films.

a)

5. Conclusions 1) The refractive index decreases when reducing the pressure of residual gases; 2) The roughness decreases when elevating the temperature up to 550°C; 3) The clear granular structure is formed at elevated temperatures; 4) No clear dependence between the roughness and the pressure of residual gases has been indicated yet as well as between the refractive index and the temperature; 5) The absorption coefficient doesn’t depend on the pressure; 6) The nanostructures appear at low temperature and at high vacuum.

b)

AUTHORS Mykhailo Rakov*, Leonid Poperenko, Vasyl Tkach, Iryna Yurgelevich - Taras Shevchenko National University of Kyiv, Department of Physics, Kyiv, 03022, Ukraine. E-mail: mikho89@mail.ru. * Corresponding author Fig. 4. The images of surfaces obtained by SEM for the films deposited at: a) 550 °C, 2.5×10-7 mbar; b) RT, 9.7×10-6 mbar.

References [1]

The roughness of the surfaces obtained by AFM is in the following Table 1. [2]

Table1. #

Ra, nm

Rz, nm

1 2 3 4 5

114,839 53,5895 176,054 83,0918 33,7302

350,021 128,879 445,631 221,366 108,834

Stand. dev., nm 139,415 62,3263 206,451 96,2661 41,3531

Rmax, nm 700,255 257,439 890,11 454,436 216,924

4. Discussion According to ellipsometry data, the refractive index decreases when reducing the pressure of residual gases. It seems to be due to more disordered structure of the films. On the other hand, AFM does not show clear dependence between the pressure of residual gases and the roughness of the surfaces. Besides, the absorption coefficient seems to have no dependence upon the pressure. Both AFM and SEM show the crystallites in sample #5 (the size has the value of about 100 nm). The roughness strongly decreases when elevating the temperature up to 550 °C. It is probably due to evaporation of the surface. Unfortunately, the experiment does not show true behavior of the refractive index and absorption coefficient at elevated temperature. The nanostructures in samples #2,3 and 4 mean that if the film is for nanoelectronics, it must be produced at low temperature and at rather high vacuum. 114

Articles

[3]

Ellmer K., Klein A., Rech B., Transparent conductive zinc oxide: Basics And Applications In Thin Film Solar Cells, Springer: Berlin, 2008. Borbat A.M., Gorban I.S., Okhrimenko B.A., SubbotaMelnik P.A., Shaikevich I.A., Shishlovskii A.A., Optical measurements, Tekhnika: Kiev, 1967. Rzhanov, K.K. Svitashev, A.I. Semenenko, V.K. Sokolov, Basic of ellipsometry, Nauka: Novosibirsk, 1979.


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

TRIBOLOGICAL PROPERTIES OF MULTILAYERED VACUUM COATING

Aleksandr Rogachou, Aleksandr Popov, Dmitri Piliptsov, Maksim Yarmolenko, Aleksandr Rogachev, Nikolai Fedosenko

Abstract: The results of studies show the deposition conditions of the titanium-nitride based coatings and their structure influence on the tribological characteristics of the dry friction and in the diesel fuel. Tribological characteristics of multilayered systems on the basis of diamond-like coatings (DLC) have been studied. The deposition of the systems onto small-size cutting tools is found efficient for improving their service characteristics. When depositing the coatings to the cutting tool surface the microrelief of edges is not distorted and the service life increases 1.3 - 1.5 times. The morphology and tribological behaviour of composite and multi-layer polymer coatings based on polytetrafluoroethylene (PTFE) and polyurethane (PU) are studied after deposition by means the method of electron beam dispersion. It is shown possibility to produce fine-film systems based on polyurethane with polytetrafluoroethylene as filler that possess high tribological performance. Keywords: diamond-like coatings, polytetrafluoroethylene, polyurethane, friction.

1. Introduction Deposition of thin hard coatings from the gaseous phase onto friction surfaces is an effective method of improving their wear resistance and reducing the friction coefficient [1]. Vacuum coatings from polymers, carbides, nitrides and oxides of transition metals are the most promising and their structure and tribological characteristics have been studied sufficiently well [1], [2]. Yet, their performance varies within a broad range and is a function of the coating structure and thickness, sliding velocity, and magnitude of stress in the contact zone [1]. This demands the development and use of multilayered coatings in friction units. Each layer of such coatings fulfils a definite function and their combination produces integrity of high performance characteristics. Hence, it is essential to investigate the tribological behaviour of the coatings and effects of deposition conditions in order to identify the optimal performance and the most effective parameters of the thin film, for solving tribological problems, such as hardening and recovery of precision friction units.

2. Experimental Flat substrates made from hardened ball bearing steel 100 CR 6 with hardness »7.9 GPa were used as tribotest specimens. First they were polished to roughness Ra = 0.063 m and then multilayered coatings were deposited. Titanium nitride coatings (TiN) were produced by con-

densation at ion bombardment with or without magnetic separation of the drop phase at nitrogen pressure 6×10-2 Pa. The authors have studied DLC, carbon coatings alloyed with titanium (Ti+C), titanium nitride (TiN) coatings obtained using the magnetic separation of the plasma flow, and laminated coatings (Ti-DLC) consisting of alternating titanium and DLC layers. Titanium-alloyed carbon coatings were formed by the simultaneous deposition of carbon and titanium onto the tool surface. As a rule, the layers were deposited during one cycle. Copper coatings were deposited by arc evaporation with the arc current 90 A. Solid lubricant layers from polymerized products of electron beam dispersion of PTFE, PU were obtained using the methods described in [3], the conditions were the following: the energy of electrons - 1.5 keV, the rate of coating deposition - 0.8 m/min, the surface temperature was equal to the ambient temperature. Before the deposition of the multilayered coating the surface of the specimen was bombarded with titanium ions having the energy 1.5 keV. A multiple wave microinterferometer MII-11 served to evaluate the coating thickness by measuring the height of the step on the reference specimen and also by gravimetry. The thickness and the rate of deposition of PTFE, PU coatings were registered with a quartz resonator directly during the layer depositing. The ball reciprocated on the flat without lubrication and in the diesel fuel during tribotests. The sliding velocity in the centre of the friction track was 0.01 m/s. A ball of 6 mm in diameter made of 100 CR 6 steel served as a counterbody, in some cases it was also coated. The friction coefficient was recorded during the dynamic contact; the wear rates of the counterbody and the coating were registered after each test.

3. Results and discussion The properties of the coatings produced by CIB, including their tribological characteristics, are known to depend strongly on the concentration of the drop phase in the flux governing the coating surface morphology. The accomplished tribological studies of the multilayered titanium-nitride coatings with a total thickness up to 2 μm, when produced under optimal process conditions, have shown sufficiently low friction coefficient against 100 CR 6 steel and high wear resistance (Figure 1, Table 1).

Articles

115


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

Figure 1. Friction coefficient as a function of number of cycles in diesel fuel for specimens with different coatings: 1 - original specimen from 100 CR 6 steel; 2 - TiN; 3 - TiN + PTFE; 4 - TiN + Cu; 5 - TiN + Cu + PTFE.

N° 4

2009

The results of the tribotests given in the table 2 show that the layered coating consisting of alternating titanium and carbon layers has the best wear resistance. In this case the wear rate is < 0.006·10-11 m3/m and during the one-hour operation of the machine under a contact pressure of 276 MPa stable friction coefficient of 0.36 and contact resistance of 50 mOhm are kept. A higher wear resistance of the coatings compared, for example, to the multilayer coatings based on DLC and TiN is evidently explained by carbide synthesis occurring in interphase layers. The authors of [4] [5] pointed out the probability of such reactions.

Table 1. Tribological Characteristics of Friction Pairs with Multi-Layered Coatings at Friction in Diesel Fuel. Coating composition

Counterbody wear rate

Pressure in contact zone

Counterbody linear wear

100 CR 6 TiN* TiN* + Cu TiN* + Cu + PTFE TiN TiN + PTFE TiN + Cu TiN + Cu + PTFE

I [10-9] 26.60 73.50 66.50 58.50 4.15 6.50 3.00 2.60

p [MPa] 10.19 3.60 3.97 4.52 63.70 40.70 88.10 99.50

Dh [μm] 5.20 14.70 13.30 11.70 0.83 1.30 0.60 0.53

Figure 1 shows that the friction coefficient of the original uncoated specimen stabilizes after 300 cycles at a level 0.34–0.36. The friction coefficient of the specimen with the TiN coating 0.21 m thick is 0.18, provided it is produced under optimal process conditions. Table 1 shows that the wear rate of the counterbody (100 CR 6 steel ball) is strongly determined by the presence of the drop phase in the TiN coating. When drops are 10–20 μm in size the wear of the counterbody almost triples compared with the original specimen. When the plasma flux is separated magnetically (dimensions of the drops in the coating are 4 m) the wear of the counterbody decreases three times compared with the wear in a steel–steel friction pair. When a two-layer coating is produced by consecutive deposition of TiN and Cu layers within one process cycle (Cu coating has the thickness d = 0.074 μm), the friction coefficient decreases after 700 cycles still more and the wear rate of the counterbody diminishes 2.1 times. In the case of a two-layer TiN + PTFE coating (d = 0.41 μm for the PTFE coating) the friction coefficient stays below the friction coefficient of the specimen coated with TiN, the wear rate of the counterbody is 1.6 times lower. When the fluoropolymer coating is 0.1–0.5 μm thick, its wear resistance is maximal, besides it is transferred to the counterbody relieving local stresses at the real contact spots. Coating with three layers (TiN, copper, and PTFE) reduces the friction coefficient still more and wear rate 2.5 times compared with the one-layer TiN coating. The wear of the counterbody is approximately 10 times less than that of the original 100 Cr 6 steel specimens. 116

Articles

Fig. 2. Hole drilling time ts as a function of number of drillings n, drill diameter d and number of layers: a - d = 0.8 mm; 1 - uncoated; 2 - TiN; 3 DLC; 4 - (C+Ti); 5 - 4 layers (TiDLC). The TiN-coating adjacent to the substrate is seen to have the granular structure with the grain size » 0.07 μm. The titanium and DLC layers have the fine structure with pronounced blurred phase boundaries of diffusion nature. Good tribological behaviour of the multilayer coatings is confirmed by the life tests of the coated drills (Fig. 2). It is seen from the curves in Figure 2 that when depositing TiN-coatings by the common technique the drilling time increases more rapidly compared to the uncoated drill. It results apparently from the fact that the cutting edges damaged in the arc spraying of the titanium nitride coating get blunt fast. Table 2. Tribological Characteristics of Coatings. Coating No TiN Ti–DLC TiN–(DLC–TiN) TiN–(Ti–DLC)

Number

h

of layers k – 1

[μm] – 1.1

f

Iv

0.52 0.59

[10-11 m3/m] 1.4 0.75

11 0.1–0.2 0.22 1(4) 0.8–(0.05–0.05) 0.18 1(7) 0.8–(0.01–0.05) 0.36

<0.2 0.2 <0.006

The morphology and tribological behaviour of composite and multi-layer polymer coatings based on polytetrafluoroethylene and polyurethane are studied after deposition by the method of electron beam dispersion. The analysis of the results proves that the coatings produced by concurrent dispersion of the mixture of the polymers (curves 5) have a rather low and stable (what is essential) friction coefficient (its variations during repeated tests did not exceed 25%) compared to other thin films. It has been observed that PTFE in the multi-layer coatings produced by consecutive application of PU and PTFE is prac-


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

tically fully removed from the friction zone after 80 friction cycles. This is the reason why the monolayer PTFE coatings have a relatively short service life. Yet the friction coefficient after such modification of the rubbers is rather low, especially at the initial stages of friction. Morphological changes in the coating during friction were determined with scanning electron microscopy. It was established that friction of rubbers protected with composite polymer-polymer coatings is accompanied with smoothing of the coatings, especially of the PU drop formations, resulting in appearance of a system of folds perpendicular to the direction of movement of the indenter. The friction process produces no cracks in the coating and it is an essential feature of wear of such coatings. No wear of the rubber surface layer was detected after 103 friction cycles that would always be the case if elastomers modified by application of a PTFE coating alone are worn.

N° 4

2009

AUTHORS Aleksandr Rogachou*, Dmitri Piliptsov, Maksim Yarmolenko, Nikolai Fedosenko - Francisk Skorina Gomel State University, Sovetskaya Str. 104, 246019, Belarus, E-mail: rogachevav@mail.ru. Aleksandr Popov, Aleksandr Rogachev - Belarusian State University of Transport, Kirova Str. 34, 246653, Belarus. * Corresponding author

References [1]

[2]

[3]

[4]

[5]

Matsevityi V.M., Lyubchenko A.P., Kazak I.B., Journal of Friction and Wear, vol. 4, no. 17, 1996, pp. 93-96. (in Russian) Rogachev A.V., Yarmolenko M.A., Tszyan Syao Hun, Lude Lu, Journal of Friction and Wear, no. 2, vol. 25, 2004, pp. 74-77. (in Russian) Rogachev A.V., Kazachenko V.P., Schebrov A.V., Collection of Papers: Polycom-98, Gomel, 1998, pp. 59-65. (in Russian) Popov A.N., Rogachev A.V., Kazachenko V.P., Abstracts of Intern. Scientific Conf. "Materials and Technologies 2000", Gomel, 2000, pp. 79-80. (in Russian) Rogachev A.V., Popov, Kazachenko V.P., Sidorskii S.S., Materials, Technologies and Tools, no. 2, vol. 5, 2001, pp. 77-80. (in Russian)

Fig. 4. Dependence of friction coefficient f on number n of cycles of abrasion of modified rubbers. 1 - Rubber unmodified; 2 - Duplex PU coating (layer thickness - 0.25 μm) + PTFE (0.25 μm); 3 - Duplex PU-PTFE coating (ratio between the components - 1:1; 0.25 μm) + PTFE (0.25 μm); 4 Duplex PU-PTFE coating (2:1; 0.25 μm) + PTFE (0.25 μm); 5 - Monolayered PUPTFE coating (2:1; 0.5 μm).

4. Conclusion It is demonstrated that the friction pairs, whose components have multi-layered TiN + Cu, TiN + Cu + PTFE coatings demonstrate the best tribological characteristics when the coatings are deposited under optimal conditions with account for the temperature criterion and the concentration of the drop phase. Tribotests of the coatings have shown that the friction coefficient remains stable in the range » 0.14–0.18. The wear of the counterbody is minimal in the diesel fuel. The deposition of the diamond-like coatings and multilayer systems on their basis onto small-size cutting tools is shown to be highly efficient for improving their performance. The multilayer DLC - and titanium-based systems have the highest wear resistance. Their deposition does not lead to the damage of the cutting edges that provides increase 1.35–1.47 times in the life. Our study determined the main morphological features of deposition of composite PTFE- and PU-based coatings. It was established that binary polymer-polymer coatings structured as a PU-matrix containing PTFE particles have the best tribological characteristics. Multi-layer coatings have a higher friction coefficient and fail more intensively in the process of dynamic contact as compared to composite coatings. Articles

117


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N째 4

2009

TOOLS AND TECHNIQUE OF MONITORING RESOURCE DISTRIBUTION FOR DESIGN SIMULATION OF ORGANIZATION OF INFORMATION PROCESSING Oleg Demidenko

Abstract: Usage of the distributed information databases (DIDB) is the most spread way of allocation of service of clients and support of cooperation between large corporations or banks. At usage DIDB in real-time mode (the actual data are accessible to all valid users at the moment of inquiry to reading or upgrade) most essential is the factor of capacity of networks. Change of the physical characteristics of modern networks is very expensive procedure. Moreover, if a gorge by transmission becomes data links, their replacement also requires essential time, that can reduce in the additional losses. In such conditions the mechanism of backup is applied to support of continuity of the computing process (CP). Except for data links a gorge in operation with DIDB can be intermediate sites, gateway servers, routers and other equipment. Frequently these functions can fulfil the same computational capabilities, which are used for processing DIDB. In this case (if the backup is not stipulated by network architecture) it is necessary to reallocate a working load (WL) between other network nodes on time of exception of a site for realization of adaptive operations or its replacement. Keywords: monitoring, local area network, information.

1. Introduction To reallocate WL and to not deform the characteristic of operation with DIDB the inspection of the computing process (CP) in a site of the local area network (LAN) is necessary, which is eliminated from the network, and also those sites and workstations that will fulfil its functions. The research CP in computing environment (CE) is possible as from the point of view of the developers of personal computers (PC) and software, and designers of the local area network. If the first are interested by influence those either others structural or functional changes in CE on its general efficiency and productivity, the second consider CE in a context of its interaction with other components of the network. Thus for the designers of networks the execution times of the tasks on workstations and servers (are important depending on implementation of structures of databases (DB)), and also sizes of streams of the connecting information between them. It is offered to create models CE for the first and second user groups on identical principles. In this case network in relation to the PC is represented as one of its devices, that is stipulated by features of collection of statistics in modern CE, having as operating system Windows XXXX (the functions of call to the network register only, that entitles to gather 118

Articles

statistics about behaviour of the network as well as about other units CE).

2. A technique of usage of resources of monitoring at designing and maintenance of the local area network For all resource types we shall define the following basic performances: kij - probability of transition of the process k from a resource i on a resource j (ij, i,j = CPUNet); hi (i = CPUNet) - load of a resource, tki - average operating time of the process k on a resource i. Any of these components has the following algorithm of behaviour. The processor (CPU). Any process captures a resource CPU, the part of a resource of the RAM and accesses to the peripheral memory. The resource CPU always allocated completely to any process, that is CPU is always identified in a couple with any process. Purely resource allocation CPU is organized by operating system (program - manager). Obviously, the resource CPU disperses on all processes and for its selection on the defined time the system of pilot signals formed on the requests of the users is used. The peripheral memory (HDD) is simulated as a place of allocation of the database. Therefore the call to the peripheral memory is imitated as operation with the information block of the defined size Qhdd, with which the readings and records are made. The RAM (Mem) is considered as a resource with maximum parameters, which is completely selected (allocated) for the accessing unit phase task (PT). For it we allocate only moments, when the system initiated a beginning or termination with memory (selection and release of memory). These processes in CE, as a rule, are short-term. Therefore for memory it is not meaningful to define option values tki. Therefore it is difficult to make output about character and features of its usage. This circumstance allocates memory from the remaining resource types. The video resource (Video) belongs to group of standard resources. Its difference consists only in a way of monitoring, when some various functions of call to Video but no one are fixed. Resource of the network (Net) is simulated, as well as remaining standard resources (for example HDD). The functions of call to a resource (in this case to network protocols) register only, that entitles to consider a network resource as internal in relation to CE and possessing all by the same basic performances, as remaining resources. The interaction of components CE can be presented as implementation of processes on resources or competition of processes for resources. All main resources for implementation CP are allocated PT at interaction


Journal of Automation, Mobile Robotics & Intelligent Systems

of components CE. Let's select two main parameters of inter-process communication. The first parameter represents average execution time of the process k on a resource i tki. The concrete values of these times are set by a matrix of allocations M(tki). The second parameter is the matrix of probabilities of capture by the process of various resources Mk(Pij). With the help of these two parameters there is a possibility for the half-markovsky of representation of processes accordingly definition CP as the weighed graph, in which the tops are resources, and weights of arcs will be probabilities of transitions of the process from one resource to another. The specificity of watching by the system of monitoring of the moments of control transfer at implementation CP is those, that there is only possibility to register, for example, beginning and extremity of the disk operation, pre-emptive some times by processes of sort System. These processes fulfil the call to resources of other user's processes (purely implementation of multitasking), but we cannot authentically tell which of them. The following two characteristics therefore are entered. "Complete time" tki, meaning time between the beginning and extremity of the obvious operation above a resource and "exact times" tki, defining complete time minus an operating time System from complete (tki = tkitsys). Correlating these times for model and real system it is possible to speak about adequacy to model (i.e. thus criterion of adequacy of measurements) is determined. The monitoring of interaction of program components in operating system WINDOWS 95 is carried out by fixing the moments of control transfers and result of the subordinate processes. It is known that any process generated by the user, has the rights of use by all system resources, but has no any information on other user's processes fulfilled by the system simultaneously with it. This mechanism is intended for protection of processes against failures of operation of the parallel tasks. CP in multitask operating system consists of periodic transmissions of the rights of use by system resources from one process to the other. Depending on executable functions each of processes uses system resources on its own algorithm. Depending on the requirement to serving resources the processes can be divided into the following types: The auxiliary tasks (are used only RAM and processor). An interactive process of exchange with the user (usage of a subsystem of video output) is added. The background tasks, which do not require(demand) output to the screen (demand processing by the file - server, operation from the remote DB). Processes operating a complete spectrum of equipment. Each process captures a resource of the CPU, the part of a resource of the RAM and accesses to the peripheral memory. The resource of the CPU is always allocated completely to any program module that is the CPU is always identified in a couple with any process. Purely resource allocation of the CPU is organized by the program manager. It is obvious that the resource of the CPU as though disperses on all processes and for its selection on the defined time the system of pilot signals formed on the users requests. The system of monitoring watches the moments of switching between processes realized by the manager, of the tasks.

VOLUME 3,

N째 4

2009

For this purpose used VMM tools under a title Call When Thread Switched. The pointer to the function of callback Thread Switeh CallBack (OldThread, NewThread) is transmitted to this tool. The manager of the tasks calls this function at switching threads, which fulfils the following operations: fixes a switching time; brings in a log record about completion of processor quantum for old thread; calculates the identifier of the current process; brings in a log record about the beginning of quantum of the processor for new thread. During initialization the monitor "inserts" "hooks" on main information data highways of an operating system. The data gathered by these sensors, characterize specialized operations of control transfer between processes. The obtained information is only statistical and is authentic only at the reliable system of identification of processes - sources of service requests. According to one of main principles of construction of modern operating systems, the vertical levels of operating system should be isolated from each other. It means that they should know nothing about a structure of inner patterns each other. The interaction between them should be carried out through the documentary interfaces. Besides, for support of compatibility and expandability, the lower layers should not know about existence of top levels at all. Therefore at development of algorithms of monitoring there was a problem coupled to identification of the process, fulfilled any of watched operations, which was necessary for deciding at writing the driver of the system of monitoring. The driver can determine identifiers of "thread" and process, but cannot determine a name of the process. It was possible to write the program operating library ToolHelp, which enumerates all "threads" and processes in the system. But thus the identifiers, obtained by it would not coincide with values obtained by the driver. It is stipulated by that the identifiers of a level of a kernel differ from identifiers of the user's mode Win32. Therefore it was necessary to find correspondence between these identifiers. For solution of this problem the following way was offered. The driver submitives user's processes "service" returning current system identifiers of "thread" and the process. But to find correspondence of identifiers for all processes, it was necessary to fulfil the given procedure in a context of each process. To hit in a context of each process it is possible with the help of usage of global system "traps". "Traps" are stored in dynamic loaded libraries and after registration in the system are connected to each executable process. In this case "trap" function can not fulfil any operations that are it simply returns handle. Let's remark that all necessary operations should be fulfilled at the moment of initialization and termination of the library. So, during initialization of "trap" the following operations are fulfilled: loading of the driver of the system of monitoring; function call of obtaining of system identifiers of a thread and process; definition of a name of the process; finding and obtaining of a descriptor of the window of the program of collection of statistics; a dispatch to this message box about creation of the process containing item of information on the process. Besides during termination of the library the addiArticles

119


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

tional operations are fulfilled: finding and obtaining of a descriptor of the window of the program of collection of statistics; a dispatch to this message box about termination of the process. The driver of the system of monitoring writes the items of information on all watched events in the local buffer. The given buffer is in unloaded memory. Therefore its size is limited to 4 thousand records. The necessity of layout of the buffer in unloaded memory is coupled to the requirements about minimum effect of the system of monitoring on CP. As the speed of filling of the buffer varies over a wide range, given from buffers should be periodically read out and be saved in the appropriate file. The write operation of the buffer on the disk could be carried out and from limits of the driver, but the given way is ineffective, that is stipulated by that fact, that in the file of an event log the unique identifiers of the process should be brought. But the same identifiers obtained by the driver, can be repeated. Therefore, each time is necessary to fulfil the special procedure of identification of processes, which algorithm is circumscribed above. But it is impossible to fulfil this procedure at a level of the driver. Therefore the task of collection of statistics was necessary for realizing as the separate program. Thus this program fulfils the following operations: the start of the driver of the system of monitoring; the periodic reading given from buffers of the driver; identification of processes in the system; the record of a sequence of events in a log. The program of collection of statistics cooperates with the driver of the system of monitoring and program of identification of processes. The operation of the main unit is made permanently in a background. The program of collection of statistics is the main "launched" unit of the system of monitoring. The given program is installed in the menu of autoload Windows. By default it does not output on the screen of any additional items of information, except for the list of triggered processes and quantity of events per one second. It is also possible to install an output mode of the list of events on the screen. But such mode renders essential influence on a response time of the system, because the majority of events, registered by the monitor, will be coupled to mapping of a log on the screen. Besides there is a possibility of suspension of system operation of monitoring. On a program termination of collection of statistics happens: out swapping of the driver from memory, file closing of a loge and termination of watching of system events. The program of collection of statistics at start registers appearance in the "trap" system, responding for identification of processes in the system. Then at start of the new process it receives all items of information on it, including its system identifiers. On the basis of these items of information the table of correspondence of identifiers is created. After each reading given from the driver there is a conversion of identifiers to the help of the table of correspondence. As to write the data in a log to the disk after each reading requires many resources, these data in the be-ginning are stored in the buffer of the program of collection of statistics, which approximately in 10 times more buffer of the driver. 120

Articles

N째 4

2009

Furthermore, the data record in a log happens only after filling this buffer. The system of monitoring brings the items of information on operation of user's processes in a special magazine, the information from which in further is used at problem solving of simulation. All items of information are stored in a log in the binary format and can be independently interpreted by the user. For obtaining common view about character of operation of the computing process there is a possibility of a display image of a path of system operation. The program of graphics representation of results of monitoring therefore was developed. The express train analysis of results of monitoring can be carried out under the state transition diagram and on a summary matrix of probabilities of transitions between service devices. This statistics allows building the Kiviata diagrams, describing load of each system resource by separate components WL.

3. Results of approbation of a technique and resources of monitoring of the local area network The system of monitoring has passed experimental approbation in computer centre of faculty "the Automated systems of information processing" of the Gomel State University named after F. Scorina. At trials 18 computers of the various configurations were involved. By results of experiments the system of monitoring has produced good metrics by criterion of a similarity of the characteristics of the real system and characteristics of systems simulated by results of the analysis of statistics (of deviation about 10 Operating characteristics: - a size of consumed memory - 213 Kilobytes. Is used for storage calculated statistics; a required disk space - 10 MB per hour (at maximum loading of thesystem); - of CPU on monitoring - 2 of statistics - 0.3 session). The analysis of statistics is made by the contributor with the help of the separate program. The technique allows constructing a picture CP, established by the concrete user CE, depending from sort WL and characteristics CE. At design simulation there is a possibility of manipulation both characteristics CE, and structure WL, that is reached by means of the detailed analysis of a log of statistics. In result the contributor receives toolkit for planning adaptive operations of the local area network.

AUTHOR Oleg Demidenko - Gomel State University, Gomel, 246019, Belarus. E-mail: demidenko@gsu.by.

References [1]

Maximey I.V., Levchuk V.D., Eskova O.I., et al., Technology of setting of imitative experiments on PC at research of computer networks. The theses of the reports of international conference devoted memory of the academician Chunihina S.A., Gomel, 1995.


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

APPLICATION OF SOL-GEL PROCESS FOR PREPARATION OF FUNCTIONAL MATERIALS

Vladimir Evgenyevich Gaishun, Oleg Mikhailovich Demidenko, Yanina A. Kosenok, Dmitry L. Kovalenko, Alina V. Semchenko, Wieslaw Strek

Abstract: A development of high-performance materials for electronics, optics, ceramics fabrication is restrained by traditional modes of their production. Basic technologies for materials include high-temperature physical and chemical processes and require special conditions to attain desired properties of final products. Advanced preparation methods for materials with new features are feasible on the basis of colloid-chemical processes and nanochemistry. In this respect the sol-to-gel transformation followed by solidification and chemical modification are of great interest to attain a variety of different functional properties of materials. Keywords: sol-gel, colored silica film, rust-proofing inorganic covers, colloidal nanosized silica.

of initial materials. They also have a lower synthesis temperature. The sol-gel technology is believed as energy and resource saving process. Is one more advantage is the simplicity of necessary equipment [1].

2. Results and Discussions 2.1. Colored silica sol-gel films on soda-lime glasses and plastic substrates Glass tiles are generally used for decoration purposes. There are two types of sol-gel coatings: inorganic and hybrid inorganic-organic ones. Especially, when the safety regulation matters, the inorganic sol-gel coatings are useful. The inorganic sol-gel coatings are based mostly on metal oxides result in minimum harmful fumes under fire. Inorganic sol-gel coatings in building industry, like glass tiles, are developed in many countries [2].

1. Introduction Sol-gel process in comparison with traditional methods of deriving of vitreous structures has unique advantage: it allows preparing for the same composition, such as silicon oxide, as considerably distinguished shapes - fibrils, films, monoliths, xerogels, aerogels by changing only some experimental conditions. Parameter of processing which needs to be driven, viscosity, pH of sol and concentration of oxides which should be in fixed limits. Distinctivenesses sol-gel of production engineering in comparison with traditional methods of forming of composite materials are: • High chemical homogeneity and purity component, containing in a final material on a molecular scale; • Pliability and controllability of process; • Forming of silicate and ceramic matrixes at lower temperatures; • The high reactive capacity of porous xerogels and subsequent them solid; phase transferring in glass or an aerogel, without of a stage of a melting, allows to gain the broad audience of chemical combinations. Sol-gel technology is very simply, cheep, ecologically. For example, this methods yields high-purity and activated silica glasses for optics, optoelectronics and laser optics at lower temperatures eliminating the fusion stage. The sol-gel transition occurs due to polycondensation, hydrolysis, gel formation followed by heat treatment producing dense gels. Unlike fused glasses, gel ones contain less impurities that results from the quality

Fig. 1. Absorption spectra of silica sol-gel films doped with organic dyes.

Fig. 2. Transmission spectra of silica sol-gel films doped with metal oxides.

Articles

121


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

Hybrid coatings would have both the advantages from inorganic-organic materials. In this work, we present the films doped with organic dyes (rhodamine, coumarine, nigrosine, methylene blue) and metal (Fe, Co, Fe-Co) oxides [3]. Thickness of these films synthesized on plastic substrates at temperatures 100-1500 °C and soda-lime glasses at 5000 °C was 0.2-5 μm. The films possess good adhesive properties and high stability with respect to mechanical attrition those fit the industry standards. The refraction index and factor of dispersion are the same as for conventional silica films. Absorption bands in the visible can be varied depending on type and concentration of organic dyes (Fig. 1) or inorganic dopants (Fig. 2). 2.2. Rust-proofing inorganic covers By the sol-gel technique we have developed the technology to produce protective and rust-proofing inorganic silica covers on the metal surface. Precursor solutions for film-forming solution will be prepared by hydrolyse of silicon organic compound. That is conventional procedure used for silica sol-gel films. films are produced by spin- or dip-coating on a substrate at room temperature. Then films are heated at 400 °C in air. Advantages of protective and rust-proofing covers on the metal surfaces: 1. Resistant to against mechanical attrition the coat is formed at temperature 400 °C. 2. Inductivity quantity – 10.5-8.4. 3. Mechanical stability to a temperature cycling up to T = 700 °C. 4. Humidity and corrosion resistance in aggressive lead-acid and an alkaline condition. At temperature of aggressive medium up to 50 °C. 2.3. Silica gel glasses, doped trivalent rare-earth ions for fiber optics applications Sol-gel silica glasses doped with rare-earth ions are an important class of optical materials with applications including solid-state lasers, optical waveguides, fiber amplifiers and fiber optic sensors. The silica gel glasses can be obtained by the sol-gel process. The samples were prepared from tetraethoxysilane, water, ethanol, fumed silica and soluble in water or ethanol the salts of the RE3+ (Er3+, Nd3+, Sm3+ et al.) elements. Sol-gel glasses usually contain a considerable amount of OH- groups, which reduce the efficiency of luminescent emission. A significant reduction in hydroxyl ion concentration is possible by thermal treatment of xerogel in the fluore-containing atmosphere. The samples were then vitrified in air. Then rods were formed from small samples of the doped glass, which were subsequently used in the drawing of PCS-type fibers with the core diameter of 200 μm. The properties of these fibers are more interesting: their ability of being fiber sensors and pumped amplifiers. The glasses synthesized by the sol-gel method have almost all properties of silica glass (Table 1), possess high optical quality and can be made large enough to their practical application. The technology may contain 122

Articles

N° 4

2009

additional stages to reduce the concentration of OHgroups if necessary. The above-described sol-gel method can be used to fabricate silica glass blanks of various shapes including blanks of the glass activated by rareearth elements which can be applied as light filters and blanks for optical fiber drawing. Table 1. Comparative characteristics of fused and sol-gel glasses. Characteristic

Sol-gel glasses 2,201 Class A 1,458

Fused silica glasses 2,203 Class B 1,458

Density, g/cm3 Chemical resistance Refraction index Concentration of OH- groups, wt.% 0,0005-0,1 0,03-0,05 Linear thermal expansion coefficient, K-1 0,55×10-6 0,55-0,58×10-6 Microhardness, MPa 7000-8500 6860-8850 Young’s modulus, GPa 7,0±0,1 7,3 All matrices synthesized by the sol-gel method show the effect of the sensibilization of the luminescence of rare-earth ions by silver ions because of the above-described features of the matrices. For example, the amplification of luminescence reaches 2-5 times in Sm-Ag- and Eu-Ag-containing glasses and 2-3 times in Sm-Cu- and Eu-Cu-containing systems compared to single-activated glasses. Table 2. Dependence of intensity of peak luminescence of Sm-Ag-containing glass on its composition. Glass composition SiO2

Sm2O3

Ag2O

96,9 96,0 99,3 99,0

2,0 2,0 0,2 1,0

1,1 2,0 0,5 -

Amplification of peak luminescence Ilum (lexc = 280 nm; lreg = 280 nm) 1,1 2,0 0,5 -

Fig. 3. Luminescence spectra of Eu- and Eu-Ag-containing glasses. Complex centers with a high efficiency of the luminescence of Eu2+ in the luminescence band with the maxi-


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

mum at 450 and 590 nm appear in Eu-Al-containing glasses because of the above-described reasons. These centers transform effectively UV radiation into the yellowblue spectral band. 2.4. Colloidal nanosized silica for mechanochemical polishing of single-crystalline silicon A colloidal dispersion, referred to as a sol, contains very fine particles (1-100 nm in diameter) that remain in suspension for very long times. In colloidal silica, the particles are amorphous and they have a negative electrical charge. Owing to electrostatic repulsion of particle from each other the stability of dispersion is increased. They are nearly insoluble in the dispersing medium. Colloidal nanosized silica is meant for a finishing polish of optic and electronic products. In particular, it is used at the stage of finishing polish of single-crystalline silicon wafers. It is an ultra-disperse colloid system based on silicon dioxide. It is a liquid of the color of the milk without any visible mechanical inclusions. The obtained product has following characteristics: • Density of suspension 1.09-1.10 g/sm3 • pH at 20 0C 5.5-7.0 • Contents SiO2 13-17 wt% • Size of particles 30-80 nm Viscosity 1.5-1.7 MPa×s

N° 4

2009

AUTHORS Vladimir Evgenyevich Gaishun*, Oleg Mikhailovich Demidenko, Yanina A. Kosenok, Dmitry L. Kovalenko, Alina V. Semchenko, Wieslaw Strek - Advanced Materials Research Laboratory, Gomel State University, 246019, Gomel, Belarus. E-mail:gaishun@mail.ru. * Corresponding author

References [1]

[2]

[3]

Brinker C.J., Scherer G.W., Sol-Gel Science: The Physics and Chemistry if Sol-Gel Processing, Academic, New York, 1990. Wojtach K., Laczka M., Cholewa-Kowalska K., Olejniczak Z., Sokolowska J., J. Non-Cryst. Solids, no. 353, 2007, pp. 2099-2103. Prokopenko V.B., Gurin V.S., Alexeenko A.A, Kulikauskas V.S., Kovalenko D.L., J. Phys. D, vol. 33, 2000, pp. 3152–3155.

Colloidal nanosized silica possesses a polishing capability for optic and electronic products at a normal speed of a material lift. The time of processing of single-crystalline silicon wafers is 30 minutes at the primary stage of mechanochemical polishing and the material removal has compounded 30-35 microns. The time of processing is 15 minutes at the secondary stage of mechanochemical polishing and the material removal has compounded 5-7 microns. These suspensions on the basis nanosized particles of fumed silica have high stability, small of an arising static electricity on polish, high efficiency, good selectivity, ease of usage, the minimum impurity by ions of metals and are usable at planar stage of metallical layers by production of integrated circuits.

3. Conclusion Sol-gel method offers the advantage of a relatively simple production procedure of the vitreous material. This method is known to produce materials from solutions either in bulk, coating, films, fibers or powders. Silica sol-gel materials are characterized by the lowered maintenance of the impurity, the caused cleanliness of initial materials, and low temperature of synthesis; besides a sol-gel method gives more ample opportunities of influence on physical and chemical parameters of final products. Sol-gel technology is resource- and energy-saving, allows receiving ready products or perform with form and sizes to finished articles (rational preparations) with the big percent of an output suitable and small quantity of waste products. Its advantage is also simplicity of the used equipment.

Articles

123


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N째 4

2009

GROWTH AND PROPERTIES OF ZNO FILMS GROWN USING PA-MOVPE WITH DMZN AND N2O Takayuki Nakano, Kazuki Nishimoto, Masatomo Sumiya, Shunro Fuke

Abstract: ZnO films were grown by plasma-assisted metal organic vapor phase epitaxy (PA-MOVPE) using dimethylzinc (DMZn) and N2O gases. The crystallinity and surface morphology of ZnO films were investigated using X-ray diffraction (XRD) and scanning electron microscopy (SEM), respectively. The crystallinity of ZnO improved (a) when the N2O flow rate was increased, because vapor phase reaction was suppressed while surface migration was enhanced, and (b) when the growth temperature was increased, because surface diffusion was enhanced. In the PA-MOVPE method with N2O gases and DMZn, both the N2O flow rate and crystal growth temperature strongly affected the crystallinity of ZnO films. Keywords: ZnO, PA-MOVPE, epitaxial growth, DMZn, N2O, temperature modulation growth.

1. Introduction Recently, wide band-gap semiconductors have attracted much attention due to the ever-increasing commercial demand for short wavelength light emitting devices (LED) [1]. ZnO is a good candidate for these devices due to its wide band gap of 3.37 eV and to its extremely high exton binding energy of 60 meV at room temperature (RT), which in principle should allow efficient excitonic lasing mechanism at RT [2]. ZnO thin films have been prepared by numerous methods, such as sputtering [3], ion beam-assisted reactive deposition [4], metal organic vapor phase Epitaxy (MOVPE) [5], plasma-assisted MOVE (PA-MOVPE) [6], molecular beam epitaxy (MBE) [7] and pulsed laser deposition (PLD) [8]. Much progress has been achieved in experimental and theoretical investigations of ZnO, and origins of luminescence have been proposed. Among film-preparation methods, PA-MOVPE has an advantage in growing high quality films due to its versatility in controlling various thermodynamic interactions. A disadvantage, however, is that adduct formed by vapor phase reaction is deposited on the substrate surface and obstructs the crystal growth. To obtain highquality ZnO films, different precursors have therefore been adopted in the growth methods of ZnO films. Here, we report the characteristics of ZnO grown on a-plane sapphire with DMZn and N2O as precursors.

2. Experimental ZnO films were deposited on a-plane sapphire substrates at 800 째C for 90 minutes by PA-MOVPE with dimethylzinc (DMZn) and N2O as Zn and O precursors, respectively. The carrier gas for DMZn was N2, and the chamber 124

Articles

pressure was about 9 x 10-2 Torr. N2O flow rates were used 20 ccm. The properties of the ZnO film were evaluated using interference microscopy, X-ray diffractometer (XRD), photoluminescence (PL) and field emission scanning electron microscopy (FE-SEM).

3. Results and discussion 3.1. High quality ZnO epitaxial growth ZnO films were grown at various N2O flow rates (5 to 20 ccm). Figure 1 shows SEM images of ZnO surface, and Fig. 2 shows the X-ray rocking curve (XRC) measurement results. At N2O flow rates of 5 and 10 ccm (Figs. 1a and b, respectively); the ZnO film was formed by aggregation of granular crystals, indicating that the ZnO growth model for these two flow rates is the Volmer-Weber type. At the lower flow rate (5 ccm), the crystal grain had irregular shape (Fig. 1a), whereas at the higher flow rate (10 ccm), the crystal grain on the surface had hexagonal facet (Fig. 1b). This result indicates that adhesive growth occurred when the N2O flow rate was 5 ccm. When the N2O flow rate was increased to 15 and 20 ccm (Figs. 1c and 1d, respectively), the surface became flat, indicating that increasing the N2O flow rate promoted surface migration.

Fig. 1. SEM images of ZnO films at different N2O flow rates: (a) 5, (b) 10, (c) 15, and (d) 20 ccm. In Fig. 2, XRC measurement results confirmed that the crystallinity improved when the N2O flow rate was increased. Table 1 shows the partial pressure of N2O, DMZn, N2 at various N2O flow rates studied here. At low N2O flow rate (5 to 10 ccm), the partial pressure of DMZn increased and the reaction rate in the vapor phase increased. Therefore, adduct that was generated by vapor phase reaction and adhered to the surface, inhibited the crystal growth. When the N2O flow rate was increased, the partial pres-


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

sure of DMZn decreased, thus suppressing the vaporphase reaction, resulting in a flat ZnO surface. These results suggest that the advantage of PA-MOVPE with N2O is the suppression of the vapor phase reaction by lowering the DMZn partial pressure.

N° 4

2009

of hexagonal grains appeared on the ZnO film at the high growth temperature of 840 °C. Above 840 °C, the ZnO thin film starts to evaporate, and the facet in which the evaporation is difficult to be generated has appeared on the surface. The change in XRC-FWHM of the ZnO films is evidence of the consideration for these SEM images. Surface migration is apparently promoted by the increase in growth temperature because the crystallinity also improved with increasing growth temperature. When the growth temperature was increased further (> 840 °C), the evaporation of ZnO increased and the crystallinity deteriorated. The sensitivity of the surface roughness to evaporation of the film explains why the surface roughness increases faster than the deterioration in crystallinity.

Fig. 2. Dependence of ZnO film crystallinity on N2O flow rate evaluated using XRC-FWHM. Table 1. Partial pressure of N2O, DMZn, and N2 at various N2O flow rates studied here. N2O flow rate PN O (ccm) (×10-2 Torr) 2

5.0 10 15 20

1.450 4.840 5.150 6.300

PDMZn (×10-2 Torr)

PN (×10-2 Torr)

0.114 0.078 0.058 0.041

7.436 5.082 3.792 2.659

2

Fig. 3. SEM images of ZnO films at different growth temperature: (a) 600 °C, (b) 800 °C, (c) 820 °C, and (d) 840 °C. Next, the dependence of the ZnO film on growth temperature was evaluated at an N2O flow rate of 20 ccm. Figure 3 and 4 show SEM images and XRC measurement results, respectively. At a growth temperature of 600 °C, the ZnO film was actually crystal grains. At 800 °C, the ZnO film became continuous and flat. At 820 °C, the film was the flattest because surface migration was promoted by the increase in growth temperature. In the high-temperature growth region (800-840 °C), Zn atoms can reach stable sites due to the increased surface migration, thus improving the surface flatness. Moreover, a high density

Fig. 4. Dependence of crystallinity of ZnO films on growth temperature evaluated using XRC-FWHM. 3.2. Temperature modulation ZnO epitaxial growth To produce LED, p type ZnO films were fabricated in which N atoms of N2O gas were incorporated into ZnO. In the low temperature growth region, incorporation efficiency of N atoms is high [9]. However, fabrication of high quality ZnO films at low temperature growth is difficult due to be not obtained sufficient surface migration. ZnO films with incorporated N atoms and with high quality crystallinity were therefore fabricated by temperature modulation growth involving low temperature (LT) growth (at 400 °C, 500 °C, 600 °C or 650 °C) and high temperature (HT) growth (820 °C, based on our results discussed in 3.1) at every 10 nm. Figure 5 shows SEM images of ZnO films grown using temperature modulation growth, and Fig. 6 shows and the XRC measurement results for these films. In Fig. 5a, the ZnO film grown at 400 °C in the LT growth region shows aggregation of granular crystals, indicating that this growth temperature did not sufficiently promote surface migration. When the growth temperature exceeded 500 °C in the LT growth region, the ZnO surface become flat and continuous. These results indicate that surface migration was sufficiently enhanced when the growth temperature was higher than 400 °C, thus improving the lattice for high-quality ZnO buffer layer. In addition, the ZnO surface grown at 600 °C and 650 °C in the LT growth region had good surface morphology which cannot observe the roughness of surface in SEM.

Articles

125


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

ration, and enhanced the near-band edge luminescence. These results indicate that the temperature modulation growth was able to realize high-quality crystal growth in the low temperature region where incorporation of N atoms into the film was possible.

4. Conclusion

Fig. 5. SEM images of ZnO films grown using temperature modulation growth at different LT growth temperature: (a) 400 °C, (b) 500 °C, (c) 600 °C, and (d) 650 °C. The XRC measurement results (Fig. 6) confirm that the crystallinity of ZnO films grown by LT growth was the highest when the growth temperature was 600 °C. Moreover, the crystallinity was drastically improved by annealing the film in air at 900 °C for 30 min. Apparently this annealing causes the atoms in the interstice to move and thus reach stable sites.

Results reported here show that gas flow rate and growth temperature significantly affect the surface diffusion in epitaxial growth by PA-MOVPE using N2O and DMZn. High-quality ZnO films can therefore be fabricated by controlling the gas flow rate and growth temperature. Temperature modulation growth can be used to fabricate high-quality crystal growth in the region where incorporation of N atoms into the film is possible. Fabrication of p-ZnO by using PA-MOVPE seems to be possible by the examination of further process.

AUTHORS Takayuki Nakano* - Department of Electrical and Electronic Engineering, Faculty of Engineering, Shizuoka University, 3-5-1 Johoku Naka-ku Hamamatsu, 4328561, Japan. E-mail: ttnakan@ipc.shizuoka.ac.jp Kazuki Nishimoto - Department of Electrical and Electronic Engineering, Faculty of Engineering, Shizuoka University. Masatomo Sumiya - National Institute for Materials Science, Tsukuba 305-0044, Japan. Shunro Fuke - Department of Electrical and Electronic Engineering, Faculty of Engineering, Shizuoka University. * Corresponding author

References [1] [2]

Fig. 6. Dependence of ZnO film crystallinity on LT growth temperature and on annealing, evaluated using XRCFWHM.

[3] [4] [5]

[6] [7] [8] [9]

Fig. 7. Measured PL of ZnO films as-grown at 600 °C LT growth temperature and after annealing. Figure 7 shows the measured PL of ZnO films grown at 600 °C by LT growth. The PL intensity in the near-band edge was increased by the annealing. The annealing apparently stabilized both atomic position and configu126

Articles

Nakamura S., Mukai T., Senoh M. Iwasa N., Journal of Applied Physics, no. 76, 1994, 8189. Tang Z.K., Wang G.K.L., Yu P., Appl. Phys. Lett., no. 72, 1998, 3270. Krupanidhi S.B., Sayer M., J. Appl. Phys., no. 56, 1984, 3308. Zhamg D.H., Brodie D.E., Thin Solid Films, no. 238, 1994, 95. Fujimoto E., Sumiya M., Ohnishi T., Watanabe K., Lippmaa M., Matsumoto Y., Koinuma H., Appl. Phys. Express, no. 2, 2009, 045502. Minegishi K., Koiwai Y., Kikuchi Y., Yano K., Kasuga M., Shimizu A., Jpn. J. Appl. Phys., no. 36, 1997, p. 1453 f. Segawa Y., Ohtomo A., M. Koinuma M., Z.K. Tang, P. Yu, and GK.L. Wong, Phys. Stat. Sol. B, no. 202, (1997) 669. Ryu Y.R., Zhu S., Budai J.D., Chandrasekhar H.R., Miceli P.F., White H.W., Appl. Phys. Lett., no. 88, 2000, p. 20 f. Tsukazaki A., Onuma T., Ohtani M., Makino S., Sumiya M., Ohtani K., Chichibu S.F., Fuke S., Segawa Y., Ohno H., Koinuma H., Kawasaki M., Nature Material, no. 4, 2004, 42.


Journal of Automation, Mobile Robotics & Intelligent Systems

VOLUME 3,

N° 4

2009

MECHANISM OF NANOSTRUCTURE FORMATION ON A SURFACE OF CDZNTE CRYSTAL BY LASER IRRADIATION Artur Medvid', Aleksandr Mychko, Volodymyr A. Gnatyuk, Sergiy Levytskyi, Yuriy Naseka

Abstract: Self-organizing nanometer size structures are observed on the surface of CdZnTe crystal irradiated by strongly absorbed Nd:YAG laser irradiation (LR) at intensity range of 4-12 MW/cm2. According to this model the Thermogradient effect has the main role in the interaction process of LR and semiconductors. The surface state of the CdZnTe under the influence of Nd:YAG laser irradiation has been studied. A state has been defined by atom force microscope. The influence of LR on the photoluminescence spectra has show that at a threshold intensity of LR I=4MW/cm2 creation of nanometer size structure begins. A graded band gap structure with optical window was formed at the top of nano-hills. Keywords: self-organizing structure, Nd:YAG laser, CdZnTe crystal.

1. Introduction The semiconductor solid solutions of Cd1-xZnxTe are very promising materials for production of optoelectronic devices and detectors for registration of ionizing radiation [1]-[2]. This is caused by the big width of band-gap and high atomic number of zinc and also by possibility to produce this material with high specific resistance r » 1010W-cm. This material allows creating detectors with big active region capacity, low leak currents and absence of polarization effects. The quality of semiconductor detectors used for registration of ionizing radiation is characterized by the level of that unbalanced charge collection, which is formed under the impact of irradiation quanta. For quantitative charge collection process description the quantity that is equal to the production of mobility value (m) on charge carrier life time value (t), that is [3], is used. It defines the length of charge carrier free run that means the volume of active memory. In spite of progress, which is reached in the solution of charge collection problem, and therefore in efficiency of irradiation registration [4]-[7], the increase of charge carrier free run length is active region of the detector during registration of ionizing radiation is the topical task for Cd1-xZnxTe detectors. As is generally known the impurities and defect composition and also the molar portions of CdTe in CdZnTe (0<x<1) can influence the values of m and t. Our previous research works of the influence of laser radiation on electro-physical characteristics of CdZnTe crystals has shown, that at high intensities of laser radiation can change the defect composition, irradiated surface morphology and distribution of Zn atoms in crystal takes place, influen-

cing the change of m and t values. All these facts evoke the interest in research of such interaction of laser radiation with CdZnTe crystal, which results in the best efficiency of charge collection at g and X-ray radiation registration. 1.1. Experimental details High-purity raw materials CdZnTe were used. A single crystal of Cd1-xZnxTe (x = 0.1) has been grown by HighPressure Vertical Zone Melting method. The grown crystals were cut into wafers with dimensions 10x10x1 mm3. The experiments were carried out at the Research Laboratory of Semiconductor Physics in Riga Technical University, Latvia. The changes of the structure and optical properties of the near-surface layer of undoped Cd1-xZnxTe (x=0,1) crystals caused by powerful laser pulses were studied. A Qmodulated Nd:YAG laser was used as the pulsed source of strongly absorbed radiation with wavelength l = 0.532 μm and intensity from 4 to 12 MW/cm2. Pulse duration was 10 ns. Photoconductivity measurements were used to study the optical properties of the near-surface layer of Cd1-x ZnxTe crystals before and after irradiation. The surface structure of the ideally clean CdZnTe surface before and after laser processing was analyzed with atomic force microscopy (Veeco Digital Insruments CP-II). To prevent the evaporation of Cd, the surface of Cd1-x ZnxTe crystals was coated with a thin (0.3 m) film of SiO2, which is transparent to the wavelength of laser radiation [8]. The laser radiation is absorbed by the Cd1-xZnxTe (x=0.1) crystal within a thin near-surface layer of the thickness d ~ a-1 » 10-4-10-5 cm (absorption coefficient a = 6·104 cm-1 [9]). Therefore, the main effects occur in a thin near-surface layer, the depth of which is determined by the length of thermal diffusion (0.1-0.7 μm [10]) and the absorption coefficient. Intensity of the laser radiation was below the threshold of thermal destruction of the SiO2/ Cd1-xZnxTe structure. The irradiated surface was treated in aqueous solution of HF with aim to remove the SiO2 film before the study of the surface morphology with the atomic force microscope. 1.2. Experimental results and discussion Morphology of the semiconductor surface before and after irradiation was studied to find out the influence of high absorbed laser radiation on electro-physical properties of Cd1-xZnxTe. The morphological changes were found on the surface of the sample which was irradiated by laser beam with Articles

127


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

laser intensity 12 MW/cm2 (Fig. 1(a)). The changes observed were the cone-like ~10 nm high nanostructures, formed on the microstructure peaks (Fig. 1(b)). Irradiation of Cd1-xZnxTe (x=0.1) crystals by Nd:YAG laser at intensities below the threshold intensity of 4 MW/cm2 had not changed the surface morphology. The gene-ration of nanostructures began at intensities I ³ 4 MW/ cm2. The explanation of nanostructure formation is shown at Fig. 2. The main role in the initiation of this process has thermogradient effect (TGE) [11]. In conformity with TGE atoms of Zn drift, in toward the minimum of the temperature, that is the bulk of in the sample, and atoms of Cd drift in opposite direction where the maximum of T to the irradiated surface. The concentration of Zn atoms at the irradiated surface decreases as a result of this movement. The evaluation of our results using theoretical and experimental work [12] has shown that the average concentration of Zn atoms at the sample surface decreases [13]. During laser irradiation Zn atoms move in the bulk of the sample substituting Cd atoms, which move toward the irradiated surface of the sample.

N° 4

2009

tion of nanostructures of the irradiated surface (Fig. 2) according to modified Stranskova-Krastanova model. The surface layer of the sample is characterized by high radiation stability, because the modified near surface layer contains more atoms of Cd, which have larger atom weight than atom of Zn. A built-in electric field, generated by graded band gap, is directed in the bulk of the sample due to decrease of surface recombination speed. a)

correct???? in text is Cd1-xZnxTe

a) b)

Fig. 2. A model of nanostructure formation a) profile of irradiated crystal b) change of graded band gap in nano-hills. As a result the carrier collection in this structure (CdZnTe) increases, which is proved by photoconductivity parameter change. (Fig. 3) The photoconductivity value of CdZnTe samples was recorded at temperature T=300K before and after irradiation by Nd:YAG laser. The photoconductivity curves show, that at laser intensity, that is less than 4MW/cm2 the shift of “red border of spectral sensitivity” to the longer waves (smaller energies) and the decrease of surface recombination speed was observed. The irradiation by larger energy causes the shift of “red border of spectral sensitivity” to the shorter waves and the increase of the intensity of the photocurrent.

b)

Fig. 1. Atomic force microscope images of the Cd1-xZnxTe (x=0.1) surface: a) before irradiation; b) after irradiation at the intensity of 12 MW/cm2. The two layers are formed near the semiconductor surface: the top layer consists of CdTe crystal and the lower layer - ZnTe crystal. A mismatch value of there crystal lattice for CdTe and ZnTe crystals is equal to 5.8% [14], that's why the mechanical stress between the layers of CdTe and ZnTe takes place. The relaxation of this mechanical stress is reached by plastic deformation. This deformation leads to crea128

Articles

Fig. 3. Photoconductivity curve of CdZnTe detector sensitive element before and after irradiation by Nd:YAG laser.


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

The shifts of the photoconductivity curve could be explained by earlier meant results. Irradiation of the sample with intensity of 4MW/cm2 the “red” curve shift is caused by the graded band gap formation with strict optical window, which was found in our previous work [15]. The sample irradiation with larger intensity causes the formation of nanostructure with open optical window. As a result the photocurrent value of the irradiated element increases in comparison with not irradiated sample, which indicates the mt parameter value increase two times.

[8]

[9]

[10]

2. Conclusion Studies of the effect of highly absorbed laser radiation on the optical properties of the Cd1-xZnxTe (x=0,1) compound have revealed the formation of nano-structures on the surface of the semiconductor under irradiation by the Nd:YAG laser within the intensity range of 4-12 MW/cm2. A graded band gap structure with optical window is formed on top of the nano-hills. The spectral photo-sensitivity indicates that mt parameter value increased two times.

AUTHORS Artur Medvid', Aleksandr Mychko* - Riga Technical University, Riga, 14 Azenes Str., LV-1048, Latvia. E-mails: medvids@latnet.lv, mychko@latnet.lv. Volodymyr A. Gnatyuk, Sergiy Levytskyi - Institute of Semiconductor Physics, Pr. Nauki, 41, Kyiv, 03028, Ukraine. Yuriy Naseka - Institute of Semiconductor Physics, Pr. Nauki, 41, Kyiv, 03028, Ukraine. * Corresponding author

[11] [12]

[13]

[14] [15]

N° 4

2009

truments and Methods in Physics Research Section A, vol. 387, 1997, pp. 259-263. Medvid' A., Hatanaka Y., Korbutjak D., Fedorenko L., Krilyuk S., Snitka V., „Generation of the A-centers at the surface of CdTe(Cl) by YAG:Nd laser radiation”, Appl. Sur. Science, vol. 197-198, issue 1-2, 2002, pp. 124129. Bell R.O., Toulermonde M., Siffert P., “Calculated temperature distribution during laser annealing in silicon and cadmium telluride”, Appl. Phys. A., vol. 19, no. 3, 1979, pp. 313-319. Ahmatov A., Emel'yanov V.I., “Powerful laser interaction at semiconductors and metals surface: nonlinear optical effect and nonlinear optical diagnostic”, Physics-Uspekhi., vol.147, no. 4, 1985, pp. 675-745. Tauc J., Photo- and Thermoelectrical Phenomena in Semiconductors, Moscow, IL, 1962. Reno J., Jones E., „Determination of the dependence of the band-gap energy on compozition CdZnTe”, Physical Review B, vol. 45, no. 3, 1992, pp. 1440-1442. Medvid' A., Fedorenko L., Korbutjak B., Kryluk S., Yusupov M., Mychko A., “Modification of band-gap in surface layer of CdZnTe by YAG:Nd+3 laser radiation”. In: Proceeding of SPIE 5th International Conference “Advanced optical materials and devices, vol. 1-5, 2007, 65961A. Physicochemical properties of semiconductor materials, Academic press: Moscow, 1979. Medvid' A., Fedorenko L., Korbutjak B., Kryluk S., Yusupov M., Mychko A., “Formation of grated band-gap in CdZnTe by YAG:Nd laser radiation”, Radiation Measurements, vol. 42, 2007, pp. 701-703.

References [1]

[2]

[3]

[4]

[5]

[6]

[7]

Belogorokhov A.I., Lakeenkov V.M., Belogorokhova L.I., “Optical properties of Cd1-xZnxTe (0<x<0.1) single crystals in the infrared spectral region“, Semiconductors, vol. 35, no. 7, 2001, pp. 773-776. Johnson S.M., Kalisher M.H., Ahlgren W.L., James J.B., Cockrum C.A, „HgCdTe 128×128 infrared focal plane arrays on alternative substrates of CdZnTe/GaAs/Si”, Appl. Phys. Lett., vol. 56, no.10, 1990, pp. 946-948. Schlesinger T.E., Toney I.E., et al., „Cadmium zinc telluride and its use as a nuclear radiation detector material”, Materials Science and Engineering, vol. 32, 2001, pp. 103-189. Squillante R., Entine G., Frederick E., et al., „Development of two new M--n CdTe sensor”, Nucl. Instr. and Meth. A., vol. 283,1989, pp. 323-329. Redus R.,. Squillante M.R, „Electronics for high resolution spectroscopy with compound semiconductors”, J. Lund. Ibid., vol. 380,1996, pp. 312-317. Patt B.E., Iwanczyk J.S., Vilkelis G., et al. „New Y-ray detector structures for electron only charge collection utilizing high-Z compound semiconductors”, J. Nucl. Instr. and Meth. A., vol. 380, 1996, pp. 276-281. Chapuis C., Durouchoux P., „The electrical segmention of Ge detectors as an impruvement imiging capabilities in the hard X-ray and gamma-ray range” Nuclear InsArticles

129


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

DETECTION OF INDIVIDUAL DOPANTS IN SINGE-ELECTRON DEVICES - A STUDY BY KFM OBSERVATION AND SIMULATION Maciej Ligowski, Daniel Moraru, Anwar Miftahul, Juli Cha Tarido, Takeshi Mizuno, Michiharu Tabe, Ryszard Jabłoński

Abstract: Single electron devices (SEDs) are candidates to become a keystone of future electronics. They are very attractive due to low power consumption, small size or high operating speed. It is even possible to assure compatibility with present CMOS technology when natural potential fluctuations introduced by dopant atoms are used to create quantum dots (QD). However, the main problem of this approach is due to the randomness of dopant distribution which is characteristic for conventional doping techniques. This leads to scattered characteristics of the devices, which precludes from using them in the circuits. In these work we approach the problem of correlating the distribution of QD's with the device characteristics. For that, we investigate with a Kelvin probe force microscope (KFM) the surface potential of Si nanodevice channel in order to understand the potential landscape. Results reveal the features ascribable to individual dopants. These findings are supported also by simulation results. Keywords: single dopant, Kelvin Probe Force Microscope, single-electron transfer.

1. Introduction Single-electron transport through individual impurity atoms has attracted wide interest for applications such as quantum computing [1] or single-electron transfer schemes [2], [3]. So far, researches focused either on heavilydoped field-effect transistors (FETs) [4] or on isolated impurities in nanoscale channels [5], [6]. In both cases, however, natural potential fluctuations caused by ionized dopant atoms were utilized. This is a very attractive idea since it assures compatibility with present CMOS technology. In Fig. 1(a) we can observe the schematic view of the device. The source and the drain are connected by a doped nanowire which is so small that it consists of only a few dopant atoms. These dopants introduce natural potential fluctuations (seen in Fig. 1(c)) which under appropriate biasing conditions create quantum dots separated by tunnel barriers (see Fig. 1(d)). Electrons can be accumulated in these quantum dots in the proximity of the dopants. However, since the dots are very small they can accumulate only one or few electrons. Therefore electrons can tunnel through the tunnel barriers to enter the dot only if previous electrons have been already removed. This phenomenon is known as Coulomb blockade and it is a principle of single electron tunneling operation [7].

130

Articles

Fig.1. Single electron tunneling in Si nanodevice; (a) device structure; (b) Nanodevice channel - doped Si nanowire; (c) surface potential profile of the channel; (d) effective structure of quantum dots and tunneling barriers in the device. It is natural to assume that in above described systems the distribution of dopant atoms in the channel influences the device parameters and therefore is of great importance. Unfortunately, when conventional doping techniques are used, only a random dopant distribution can be achieved. Therefore, monitoring the dopant distribution is a necessary step towards constructing a standardized SED with predictable characteristics.

2. The resolution limits of KFM Kelvin Probe Force Microscope (KFM) [8] seems to be the most appropriate tool for dopant distribution monitoring due to its high resolution and sensitivity [9]. There are admittedly several reports of shallow dopant profile or carrier concentration measurements by scanning tunneling microscopy (STM) or scanning capacitance microscopy (SCM). However, STM techniques, which are mainly used so far, allow investigating only several top-most layers [10]. SCM, on the other hand, in principle is not suitable tool for single dopant detection and thus achieved results are of much lower resolution [11]. In KFM, surface potential map is obtained by measuring the potential offset between a probe tip and a surface. This offset causes an electrostatic force: F=

1 ¶C 2 V 2 ¶z

where total potential offset is given by V = VCPD -(VDC +VAC sinwt) The VCPD is the measured contact potential difference, while VAC and VDC are the voltages applied for the measurement purpose. As a result of applying an AC


Journal of Automation, Mobile Robotics & Intelligent Systems

voltage the cantilever vibrates with the frequency f0. When surface potential changes it changes the electrostatic force, which changes a fundamental resonant frequency according to the equation [12]: 1 f '0 = 2p

¶F ¶z m*

k-

» f0(1-

1 ¶F ) 2k ¶z

The f0 is a primary oscillation frequency, k is a spring constant, m* is an effective mass and ¶F/¶z is a force gradient. According to Zerweck et al. even as small as 5mV potential difference can be detected. For this signal still a frequency shift can be resolved [12]. In practical solutions, however, this high resolution is decreased by the electron screening effect. This well known phenomenon leads to the reduction of the local surface potential fluctuations and thus precludes from detecting individual dopant atoms. To overcome this problem, we propose a low temperature measurement system. According to the Fermi-Dirac distribution the probability that the electron occupies the electronic state with energy E is given by F(E): F(E) =

VOLUME 3,

N° 4

2009

atom, reduced by screening and distance dependence. We exclude interface trapped charges or surface charge influence, since similar potential fluctuations cannot be observed in the buried oxide (BOX) region. It can be also noticed that the local potential maxima are close to the value of 0 V. This corresponds to the fact that during all the measurements the source and drain were grounded and Fermi level of the top Si layer remained at ground level. Furthermore, we conclude that the channel was partially ionized and hence conductive. As it can also be seen, the potential landscape consists of regions with lower potential (quantum dots) separated by a regions with higher potential (barriers). This image lets us assume that in the case of very thin and narrow channels a one dimensional quantum dot array can be formed.

1 1 + e(E-EF)/kT

where k is the Boltzmann constant, T is the temperature in Kelvin and EF is the Fermi level. It can be noticed that at low temperatures the number of free carriers changes abruptly around Fermi level, which means that they can be easily depleted. On the contrary, at room temperature the change is smoothed and screening occurs.

3. KFM observations of individual dopant atoms In this research we utilized KFM to visualize the potential profile of the doped Si layer. For that purpose a thin silicon-on-insulator field-effect transistor (SOI-FET) without the top gate was investigated by KFM at low temperature (13K). The substrate, which was used as a gate, was 18 -3 a boron doped p-Si (1x10 cm ) and the whole top Si re17 -3 gion was doped with phosphorus (5x10 cm ). Both source and drain were grounded, while back gate voltage (Vg) was set to -3V to deplete free carriers from the channel. The resultant KFM electronic potential image with the corresponding topography profile in the inset are presented in Fig. 2(a), while the simplified measurement setup is shown in Fig. 2(b). Local potential fluctuations induced by randomly distributed dopant atoms can be observed in the channel surface potential. In Fig. 2(c) we can see hypothetical shape of potential well induced by one donor placed at the surface. Fig. 2(d)-(f) show profiles of chosen potential wells observed in KFM image which are ascribed to individual dopants. The size and shape of observed features match with expected phosphorus ion induced fluctuations. The circular shape of the potential dips suggests the point charge. To compare their size we refer to the diameter of the potential well 15 mV above its bottom. The lateral diameter in the range of 5-10 nm is in a good agreement to the Bohr radius of phosphorus in silicon (» 3 nm) [13]. Moreover potential difference of 25 mV seems to reflect potential well, induced by a single phosphorus

Fig. 2. (a) Surface potential map of the channel area obtained by KFM at 13K (corresponding topography profile in the inset, z-range 10nm); (b) Measurement setup and structure of back-gated SOI-FET device; (c) Surface potential dip induced by single donor atom; (d)-(f) The profiles of local potential fluctuation induced by individual phosphorus atoms measured by KFM. To prove the results, further examination of dopantinduced potential fluctuations was performed. This time a boron-doped acceptor-type sample was investigated. 16 -3 The surface potential of 1μm thick p-Si (1x10 cm ) epi18 -3 taxial layer (epi-layer) placed on p-Si (1x10 cm ) substrate was investigated by KFM at 13K. Figure 3(a) shows the resultant KFM electronic surface potential map and the topography profile in the inset. The measurement setup and sample structure can be observed in Fig. 3(b). Opposite sign of detected potential fluctuations for a p-Si epi-layer in comparison to the previous sample (n-Si channel SOI-FET) can be noticed in Fig. 3(c). Furthermore, as the doping concentration of a p-Si epi-layer is in the 16 -3 range of 1x10 cm and assumed depth sensitivity of the KFM is 20nm, we expect to have 1 or 2 dopants visible in 2 the measured area of 75×75 nm . This number can vary due to random distribution and local ionization ratio of acceptors. It can be noticed, however, that for low doped Articles

131


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

p-Si epi-layer, most of the surface potential is flat and oly two ionized acceptors are clearly visible, while for n-type channel, as dopant concentration is much higher, the whole surface potential is modulated by the donor ions and remains non-uniform in the entire measured area.

N° 4

2009

There is a good correspondence between the KFM results shown in Fig. 2(a) and the simulation results shown in Fig. 3(a). It is clear that features ascribable to point charges can be distinguished in the surface potential although smooth potential changes modulated by clusters of donors can be seen as well. The diameter and shape of potential fluctuations induced by individual dopants are shown in Fig. 3(c). It can be noticed, however, that some point charges visible in the image seem to introduce shallower potential dips. To explain this discrepancy we simulated the dependence between the surface potential introduced by a single dopant and the depth on which the dopant is placed, as shown in Fig. 3(d).

5. Conclusions

Fig. 3. (a) Surface potential map obtained by KFM at 13K, topography profile in the inset (z-range 10nm); (b) Measurement setup and structure of boron doped p-Si sample; (c) Local potential fluctuation profiles induced by individual boron atoms.

4. Surface potential simulations For comparison with experimental data, a simple simulation of surface potential was performed. In this simulation the dopants where randomly placed in a 200×200×200 nm cubic volume with the concentration of 18 -3 1x10 cm . Next, the potential superimposing the Coulombic potentials of all the donors introduced into the mesh was calculated in every point of the cubic mesh [13]. The obtained top-most layer (surface) potential is drawn in Fig. 3(a), while Fig. 3(b) shows the schematic view of simulated structure.

Recently improved capabilities of LT-KFM open an opportunity to directly observe dopant induced potential fluctuations at the surface of MOSFETs under operation. So far, no appropriate method for direct detection dopant atom positions in MOSFET has been proposed. Using LTKFM we have detected the individual dopant atoms in thin SOI layer. These findings suggest that LT-KFM may become the best tool for precise dopant position mapping, which is crucial for future nanodevices [14]. By studying the surface potential (basing on KFM and simulation results) it may be possible to obtain an optimum shape of the device channel. Our goal is to study the correlation between number of dopant-induced QDs and electrical characteristics of short or long channel devices. ACKNOWLEDGMENTS The authors thank D. Nagata, S. Miki, R. Nakamura for support in experiments, and H. Ikeda for discussions. This work was partially supported by Grants-in-Aid for Scientific Research (16106006 and 18063010) from the Ministry of Education, Culture, Sports, Science, and Technology of Japan.

AUTHORS Maciej Ligowski*, Ryszard Jablonski - Warsaw University of Technology, Division of Sensors and Measuring Systems, A. Boboli 8, Warsaw 02-525, Poland. E-mail: maciej@ligowski.com. Maciej Ligowski*, Daniel Moraru, Anwar Miftahul, Juli Cha Tarido, Takeshi Mizuno, Michiharu Tabe - Shizuoka University, Research Institute of Electronics, Johoku 35-1, 432-8011 Hamamatsu, Japan. * Corresponding author

References [1] [2]

Fig. 4. (a) Surface potential map of the of the Si block obtained by simulation; (b) Schematic view of a simulated structure; (c) Local potential fluctuation profiles induced by individual dopant atoms or clusters; (d) The dependence between donor depth and corresponding surface potential. 132

Articles

[3]

[4]

Kane B.E., "A Silicon Based Nuclear Spin Quantum Computer", Nature (London), vol. 393, 1998, pp. 133. Moraru D., Ono Y., Inokawa H., Tabe M., "Quantized electron transfer through random multiple tunnel junctions in phosphorus-doped silicon nanowires", Phys. Rev. B, vol. 76, 2007, p. 075332. Yokoi K., Moraru D., Ligowski M., Tabe M., "Single-Gated Single-Electron Transfer in Nonuniform Arrays of Quantum Dots", Jpn. J. Appl. Phys., vol. 48, 2009, p. 024503. Smith R.A., Ahmed H., "Gate controlled Coulomb bloc-


Journal of Automation, Mobile Robotics & Intelligent Systems

[5]

[6]

[7]

[8] [9]

[10]

[11]

[12]

[13] [14]

VOLUME 3,

N° 4

2009

kade effects in the conduction of a silicon quantum wire", J. Appl. Phys., vol. 81, 1996, p. 2699. Ono Y., et al., "Conductance modulation by individual acceptors in Si nanoscale field-effect transistors", Appl. Phys. Lett., vol. 90, 2007, p. 102106. Sellier H., et al., "Transport Spectroscopy of a Single Dopant in a Gated Silicon Nanowire", Phys. Rev. Lett., vol. 97, 2006, p. 206805. Averin D.V., Likharev K.K., in "Single charge tunneling", ed. Grabert H. and Devoret M., Plenum, New York, 1992, p. 311. Nonnenmacher M., et al., "Kelvin probe force microscopy", Appl. Phys. Lett., vol. 58, 1991, p. 2921. Ligowski M., Moraru D., Anwar M., Mizuno T., Jablonski R., Tabe M., "Observation of individual dopants in a thin silicon layer by low temperature Kelvin probe force microscope", Appl. Phys. Lett., vol. 93, no. 14, 2008, p. 142101. Nishizawa M., et al., "Simultaneous measurement of potential and dopant atom distributions on wet-prepared Si(111):H surfaces by scanning tunneling microscopy", Appl. Phys. Lett., vol. 90, 2007, pp. 122118. Goragot W., Takai M., "Measurement of Shallow Dopant Profile Using Scanning Capacitance Microscopy", Jpn. J. Appl. Phys., vol. 43, 2004, p. 3990. Evans G.J., Mizuta H., Ahmed H., "Modelling of Structural and Threshold Voltage Characteristics of Randomly Doped Silicon Nanowires in the Coulomb-Blockade Regime", Jpn. J. Appl. Phys., vol. 40, 2001, p. 5837. Zerweck U., Loppacher C., Otto T., GrafstrĂśm S., and L. M. Eng. Moraru D., Ligowski M., Yokoi K., Mizuno T., Tabe M., Appl. Phys. Exp., vol. 2, 2009, pp. 071201.

Articles

133


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N째 4

2009

IMPURITY-CONCENTRATION DEPENDENCE OF SEEBECK COEFFICIENT IN SILICON-ON-INSULATOR LAYERS Faiz Salleh, Kiyosumi Asai, Akihiro Ishida, and Hiroya Ikeda

Abstract: We measured the Seebeck coefficient of P-doped ultrathin silicon-on-insulator (SOI) layers with thicknesses of 6-100 nm. The dependence of the coefficient on the impurity concentration was investigated, and was shown to be in good agreement with that of bulk Si. In addition, it was found to decrease with increasing impurity concentration, as is usual in semiconductor materials. However, for doping levels above 3.5x1019 cm-3, the Seebeck coefficient was observed to increase. This is likely to be due to the influence of an impurity band. Keywords: Seebeck coefficient, ultrathin silicon-on-Insulator layers, nanostructure, impurity band.

1. Introduction In recent years, thermoelectric devices have attracted considerable attention due to their ability to produce electric power from waste heat, and as a means of tackling issues related to global warming problem. In addition, it is expected that such devices can be utilized in noiseless, vibration-free refrigerators with zero greenhouse gas emission. In the present paper, we investigate the Seebeck coefficient S of P-doped ultrathin silicon-oninsulator (SOI) layers from the viewpoint of refrigeration. Such a thermoelectric material can be used as a cooling system for Si-based devices such as central processing units (CPUs) or field emission displays (FEDs).

2. Experimental

could be produced in a plane parallel to the sample surface. A couple of probes and two K-type thermocouples were directly attached to the sample surface. The time evolution of the thermoelectromotive force was measured by a digital multimeter (KEITHLEY 2700) equipped with a switching module (KEITHLEY 7700), simultaneously with the temperatures at the high- and low-temperature regions. The Seebeck coefficient was evaluated from the thermoelectricmotive force (DV=VH-VL) and the temperature difference (DT=TH-TL) by S=DV/ DT. The SOI wafer consisted of a top Si layer (SOI layer), a buried oxide (BOX) layer and a p-type Si substrate, and it was cut to a size of 10x10 mm2. The SOI layer was thinned to a thickness of 6 nm to 100 nm by repeated thermal oxidation and HF etching. P atoms were doped into the SOI layer by thermal diffusion to produce an n-type Si layer. The impurity concentration ranged from 2x1017 to 5x1019 cm-3, determined by a four-probe method at room temperature. The thickness of the BOX layer was 400 nm and the impurity concentration of the p-type Si substrate was ~1016 cm-3.

3. Results and Discussions An example of the absolute value of Seebeck coefficient of the SOI wafer is shown in Fig. 2, as a function of temperature. From this figure, the Seebeck coefficients obtained during increasing and decreasing temperature of 300-400 K are identical. This indicates that the measured Seebeck coefficient is valid. The Seebeck coefficient in this temperature region is nearly constant. Therefore, the average value of Seebeck coefficient can be estimated from the gradient of the graph of thermoelectro-motive force vs. temperature difference.

Fig. 1. Schematic diagram of the apparatus for Seebeck coefficient measurement. A schematic diagram of our experimental setup is shown in Fig. 1. Two gilded Cu-plates were placed side by side with a gap of 1 mm between them. Resistive heaters were placed beneath the plates and could be heated individually. The sample for measuring was placed across the gap, in contact with both of the plates. Therefore, by controlling the heater current, a temperature difference 134

Articles

Fig. 2. Absolute Seebeck coefficient of SOI wafer measured during increasing and decreasing temperature as a function of temperature.


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

The absolute value of the average Seebeck coefficient of the SOI layers is shown in Fig. 3, as a function of impurity concentration. The numbers adjacent to our SOI data indicate the SOI layer thickness. In this figure, the results for n-type Si wafers obtained from our measurements and reported in the literature [1]-[3] are also shown. The solid line is the theoretical Seebeck coefficient obtained from our calculations based on the electron transport [4]. From Fig. 3, the values of the Seebeck coefficient appear to lie on a curved line (indicated by a broken line in Fig. 3) and to be independent of the SOI layer thickness. In addition, the values for bulk Si wafers are very similar to those for SOI layers, which implies that a Si film as thin as 6 nm has the same Seebeck coefficient as a Si wafer. Hence, this facts suggests that an ultrathin Si Film with a nanometer-scale thickness can replace the bulk Si thermoelectric material.

N° 4

2009

ved inside a sphere occupied by an impurity atom. The impurity band width DE is defined as the energy difference between the band edges obtained from boundary conditions for the wave functions. After obtaining the band width, a Gaussian curve is used to represent the actual DOS across the impurity band, with the maximum lying at the energy level corresponding to the original impurity level [8]. We can express the impurity concentration ND as DEND = ND 2Ö2ps where s is a parameter characterizing the extent of the Gaussian function. This expression corresponds to assuming that the Gaussian function has a triangular shape with a base of DE and a height of ND/Ö2ps. The DOS function for the impurity band can then be represented by pi(E) =

(

2END E-ED exp - 4p DE DE

)2

where ED is the donor level and the conduction band edge EC is set to zero.

Fig. 3. Absolute Seebeck coefficient of SOI wafers as a function of impurity concentration. The Seebeck coefficient of Si wafers obtained from our measurements and reported in the literature [1-3] are also shown. The solid line represents the calculated value and the broken line is an eye-guide. Below 2x1019 cm-3, the absolute value of the experimental Seebeck coefficient of SOI layers decreases with an increase in the impurity concentration, as is generally seen in semiconductor materials [5]. However, as seen in the figure, the theoretical curve does not fit well to the experimental data. This is considered to be due to the influence of phonon drag originating from momentum transfer from the phonon system to the electron system by electron-phonon scattering [2]. The influence of such phonon drag is known to become weak at high impurity concentration [2]. On the other hand, SOI layers with impurity concentrations above 3.5x101019 cm-3 are found to exhibit unusual behavior in their Seebeck coefficients. The inset in Fig. 4 shows a magnified graph of this region. It is clearly seen that the absolute value of Seebeck coefficient increases with increasing impurity concentration. Since phonon drag is no longer significant at these impurity concentrations [2], additional factors such as the influence of an impurity band must be considered [6]. In order to confirm the influence of an impurity band, the density of states (DOS) was calculated for highly-doped bulk Si, based on a Baltensperger model [7]. This model assumes a regular close-packed lattice built of hydrogen-like impurities where the Schrödinger equation is sol-

Fig. 4. (a) Density of states calculated for an impurity concentration of ND=5x1017, 5x1018, 5x1019 cm-3, based on the Baltensperger model, and (b) calculated Fermi energy and impurity-band DOS at the conduction-band edge as a function of impurity concentration. The ionization energy is set to ED=44 meV for phosphorus atoms and the Fermi energy is evaluated from the charge-neutrality condition. The wave function includes the confluent hypergeometric series F(l+1-n, 2l+2; 2rs/naB) in the hydrogen-like model [7], where aB is the Bohr radius and rs is the mean Articles

135


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

radius of the sphere occupied by an impurity atom. The indices n and l are the principal and the orbital quantum numbers, respectively. In this paper, the impurity band width is computed under the condition that the principal quantum number n is approximately equal to unity for l=0, which corresponds to a 1s band [9]. Figure 4(a) shows the calculated DOS for ND=5x1017, 5x1018, 5x1019 cm-3, in the case of P atoms (ED= -44 meV). The Fermi energy for, ND=5x1019 cm-3 evaluated from the charge-neutrality condition, is also indicated by an arrow. Below ND=5x1018 cm-3, it is likely that the DOS overlap between the impurity band and the conduction band is very small. On the other hand, for ND=5x1019 cm-3, the DOS overlap becomes significant and the Fermi energy lies near the conduction band edge EC. The impurity band undoubtedly influences the Seebeck coefficient at higher impurity concentration since the Seebeck coefficient is strongly dependent upon the DOS distribution around the Fermi energy [10]. The calculated Fermi energy and the impurity-band DOS at EC are shown in Fig. 4(b) as a function of impurity concentration. The values of ED for P atoms and EC are also shown in this figure. It is clearly seen that the Fermi energy rises with increasing the impurity concentration and crosses the maximum of the impurity band DOS (ED) at ND=5x1019 cm-3. In addition, the impurity band DOS at an energy of EC also abruptly increase above ND=5x1019 cm-3. These facts indicate that above ND=5x1019 cm-3, a continuous distribution in DOS is produced by the overlap between the impurity and the conduction band. The range of impurity concentrations where the DOS distribution drastically changes is in good agreement with the region of abrupt increase in the Seebeck coefficient shown in Fig. 4. Consequently, the complicated band structure in heavily doped Si, which is quite different from that in weakly doped Si, may be the origin of the Seebeck coefficient enhancement at higher impurity concentration.

4. Conclusion We have investigated the Seebeck coefficient of SOI layers thicknesses of 6-100 nm and found it very similar to that of bulk Si for SOI layers above 6 nm. An enhancement of the Seebeck coefficient was observed at higher concentrations (ND=1x1019 cm-3). This phenomenon is believed to be related to the formation of an impurity-band, and based on calculations of the impurity-band DOS, its influence is likely to be significant above ~1x1019 cm-3, which is consistent with the experimental result. ACKNOWLEDGMENTS This work was financially supported by Grants-in-Aid for Scientific Research (19560701 and 21360336) from the Japan Society for the Promotion of Science.

AUTHORS Faiz Salleh*, Kiyosumi Asai, Hiroya Ikeda - Research Institute of Electronics, Shizuoka University, Johoku 3-5-1, Naka-ku, Hamamatsu 432-8011, Japan. E-mail: f0930131@ipc.shizuoka.ac.jp Akihiro Ishida - Department of Electrical and Electronic Engineering, Shizuoka University, Johoku 3-5-1, Naka136

Articles

N° 4

2009

ku, Hamamatsu 432-8561, Japan. * Corresponding author

References [1] [2] [3]

[4]

[5] [6]

[7] [8] [9]

[10]

Geballe T.H., Hull G.W., “Seebeck effect in Silicon”, Phys. Rev., vol. 98, no. 4, 1955, pp. 940-947. Weber L., Gmelin E., “Transport properties of Silicon”, Appl. Phys. A, vol. 53, no. 2, 1991, pp. 136-140. Yamashita O., “Effect of metal electrode on Seebeck coefficient of p- and n-type Si thermoelectric”, J. Appl. Phys., vol. 95, no. 1, 2004, pp. 178-183. Ishida A., Cao D., Morioka S., Inoue Y., Kita T., “Seebeck effect in IV-VI semiconductor films and quantum wells”, J. Electron. Mater., vol. 38, no. 7, pp. 940-943. Mahan G.D., “Good Thermoelectrics”, Solid State Phys., vol. 51, 1979, pp. 81-157. Heremans J.P., Jovovic V., Toberer E.S., Saramat A., Kurosaki K., Charoenphakdee A., Yamanaka S., Snyder G.J., “Enhancement of thermoelectric efficiency in PbTe by distortion of the electronic density of states”, Science, vol. 321, no. 5888, 2008, pp. 554-557. Baltensperger W., “On conduction in impurity bands”, Philos. Mag., vol. 44, no. 359, 1953, pp. 1355-1363. Brody T.P., “Nature of the valley current in tunnel diodes”, J. Appl. Phys., vol. 33, no. 1, 1962, pp. 100-111. Erginsoy C., “Energy states of overlapping impurity carriers in semiconductors”, Phys. Rev., vol. 88, no. 4, 1952, pp. 893-894. Mott N. F., Davis E. A., “Electronic Processes in Non-Crystalline Materials”, (Clarendon Press, Oxford, 1979), pp. 52.


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N째 4

2009

ACDTE X-RAY IMAGING DEVICE USING VERTICAL THIN FILM FIELD EMITTER ARRAY Yuki Tsunekawa, Masashi Nakagawa, Yoichiro Neo, Hisashi Morii, Toru Aokia, Hidenori Mimura, Masayoshi Nagao, Tomoya Yoshidab, Seigo Kanemaru

Abstract: We have demonstrated the novel CdTe X-ray imaging device that consists of a Schottky CdTe diode and a Vertical Thin Film Field Emitter Array (VTF-FEA). The Schottky CdTe diode was fabricated by deposied indium (In) thin film on a Cl-doped p-type (p-type) CdTe substrate and Sb2S3 on the opposite side of the CdTe substrate. VTF-FEA was fabricated by ion induced bending (IIB) and etch-back method. The signal current was successfully detected by the recombination of the stored on the surface holes with the emitted electrons from FEA, and it was clarified that it depended on the X-ray intensity. And the imaging device succeeded in obtaining the Cu plate transmission image.

generated in the CdTe diode. The holes were drifted by applying field and are stored on the Sb2S3 surface. Electrons emitted from FEA recombine with the holes on the Sb2S3 surface, and simultaneously the output current, which is proportional to the amount of the stored holes, flows in the output circuit. X-ray images are obtained by addressing the FEA in sequence. In the case that the electron beam irradiates the Sb2S3 without X-ray irradiation, the dark current of the CdTe diode flows in the output circuit.

Keywords: Vertical Thin Film Field Emitter Array, Schottky CdTe diode, ion induced bending, etch-back method.

1. Introduction A photon-counting mode X-ray imaging device has a lot of interests, such as high-sensitivity and energydiscrimination. We have studied about photon-counting mode X-ray image sensor, which was constructed with CdTe and already developed for the line sensor [1]. To realize the imaging devices with high spatial resolution, it is important to reduce the pixel size and to arrange the pixels in high density array. However, it is difficult to make CdTe device miniaturized, because the fineness of the process significantly degrades the device performance. Therefore, the imaging device that could detect X rays by using FEA was proposed. In this method, it reads the signal of each pixel one by one with the electron beam. The signal charge generated by X rays is recombined by the emitted electron from FEA. Therefore, the resolution of the image is decided by the spot size of the electron beam. It is unnecessary to make CdTe detector small. It is demanded to fabricate the detector with high resolution. In this report, VTF-FEA was fabricated by using an Ion Induced Bending (IIB) process [2]. And we have constructed an imaging device with VTF-FEA and the CdTe detector. And the principle inspection was performed.

Fig. 1. Operation principle of the CdTe X-ray image sensor using a FEA.

3. Experiments The Schottky CdTe diode was fabricated by depositing In on a CdTe substrate and Sb2S3 on the opposite side of the CdTe substrate. Figure 2 shows a current-voltage characteristic of the Schottky CdTe diode. The dark current was less than 10 nA at a reverse bias voltage of 100 V.

2. Operation principle Figure 1 shows operation principle of the CdTe X-ray image sensor using a VTF-FEA. Sb2S3 is a landing material for the electron beam and acts as the electron blocking layer. The In side of the CdTe diode was positively biased with respect to the FEA. When an electron beam irradiates the Sb2S3, the Sb2S3 layer is charged up and the CdTe diode is reversely biased by the bias voltage. Then, when Xray irradiates the Schottky diode, electron-hole pairs are

Fig. 2. Current-voltage characteristic of the Shottky CdTe diode. Articles

137


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

The VTF-FEA was fabricated by using IIB process and etch-back method [3]. Figure 3 shows the fabrication process of the matrix-structured VTF-FEA. Mo thin film of 20 nm was used as emitter and was irradiated with 25 keV Ar ion, 3.0Ă&#x2014;1015 ion/cm2. 100 nm thickness Nb film was used as the gate electrode. Gate electrode was opened by the etch-back method. Figure 4 shows SEM images of a one-pixel and an emitter tip of VTF-FEA. Table I described the specific of VTF-FEA. Figure 5 shows an electron emission characteristic of one the VTF-FEA. They were measured at high vacuum around 10-6 Pa. Threshold voltage was 25 V, and anode current of 500 nA achieved at a gate voltage of 70 V.

N° 4

2009

Table 1. The specific of VTF-FEA.

Substate size Pixel number Size of 1 pixel Pixel pitch Emitted number

7.5 mm x 7.5 mm 10 x 10 pixels 150 mm x 150 mm 0.4 mm 2000 tips/pixel

Figure 6 shows a schematic of the experimental setup for the CdTe X-ray imaging device. The VTF-FEA and CdTe diode were set in the center of a vacuum chamber. All experiments were carried out under an ultrahigh vacuum below 10-6 Pa. X-ray was irradiated on the CdTe diode through the glass viewing port.

Fig. 6. Schematic of the CdTe X-ray image device. Fig. 3. Fabrication process of the matrix-structured VTFFEA.

3. Results and discussions Figure 7 shows the output current characteristics under the exposure of electron beam with and without X-ray irradiation. The matrix-structured VTF-FEA and CdTe diode were set in the center of a vacuum chamber. A 1 mm thick Cu plate was set in front of the CdTe diode. And Cu plate beclouded the half area of VTF-FEA. The voltage of the mesh electrode was 500 V. The voltage of the cathode electrode, extraction gate and focus electrode were 0 V, 55 V, and 0 V, respectively. The anode voltage was 30 V.

Fig. 4. SEM images of a one-pixel and emitter tip.

Fig. 5. Electron emission characteristic of the VTF-FEA pixel. 138

Articles

Fig. 7. Output current (IA) characteristic under the exposure of electron beam with and without X-ray irradiation.


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

The X-ray tube voltage was changed from 10 kV to 90 kV by 20 kV step. The X-ray tube current was kept 40 μA constantly. As shown in Fig. 6, the output current was obtained only with X-ray irradiated. Then the output current depended on the X-ray tube voltage, and it indicates that the output current is promotional to the X-ray intensity. Thus, the presence of the copper sheet can be detected, and the principle operation of X-ray imaging was demonstrated. Figure 8 shows the X-ray transmission image for Cu plate. The X-ray tube voltage was 30kV. And the X-ray tube current was 40 μA. As shown in Fig. 8, the two areas were successfully divided by reading output signal.

N° 4

2009

References [1]

[2] [3]

Aoki T., Ishida Y., Makino Y., Ohashi G., Y. Tomita, Morii H., Temmyo J., Hatanaka Y., Proc. IDW/AD’05, 2005, pp. 2061-2064. Yoshida T., et al., J. Vac. Sci. Technol., B 24, 932 (2006). Hashiguchi G., Mimura H., Fujita H., J. Appl. Phys., vol. 34, 1995, pp. L883-L885.

Fig. 8. (a) a schematic of the transmission objects and FEA setup and (b) obtained transmission image of Cu plate with 1mm thickness.

4. Conclusion We have demonstrated the novel CdTe X-ray imaging device that consists of a Schottky CdTe diode and a VTFFEA. The VTF-FEA was fabricated by Ion Induced Bending (IIB) process and etch-back method. The output current was successfully detected by the recombination of holes with electrons from the FEA, and clearly depended on the X-ray tube voltage, indicating that the output current |is promotional to the X-ray intensity. Furthermore the transmission image was successfully obtained. ACKNOWLEDGMENTS The authors would like to thank S. Yamashita and K. Matsubara for their technical assistance. The work was supported, in part, by Shizuoka University 21st COE (Center of Excellence) Program and Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Technology and Science, Japan.

AUTHORS Yuki Tsunekawa*, Masashi Nakagawa, Yoichiro Neo, Hisashi Moriia, Hidenori Mimuraa - Research Institute of Electronics, Shizuoka Univ., 3-5-1 Johoku, Nakaku Hamamatsu, Shizuoka 432-8011, Japan. Tel/Fax: +81-53478-1319. E-mail: tsunekawa@nvrc.rie.shizuoka.ac.jp. Masayoshi Nagao, Tomoya Yoshida, Seigo Kanemaru Advanced Industrial Science and Technology, 1-1-1 Uezono, Tsukuba, Ibaraki, 305-8568, Japan. Toru Aokia * Corresponding author

Articles

139


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

FORMATION OF “BLACK SILICON” ON A SURFACE OF NI/SI STRUCTURE BY ND:YAG LASER RADIATION Artur Medvid', Aliaksandra Karabko, Pavels Onufrijevs, Edvins Dauksta, Anatoly Dostanko

Abstract: We have shown the possibility to form a new type of material known as “black silicon”. After irradiation of a Si sample surface, covered with 30 nm thick Ni layer, by Nd: YAG laser beam at intensity 4.5 MW/cm2 the “black silicon” was formed. The formation and self-organization of conelike microstructures on the Ni/Si surface has been detected by scanning electron microscope (SEM). Light is repeatedly reflected between the cones in the way that most of it is absorbed, therefore the surface becomes like a “black body” absorber. The micro-chemical analysis performed on SEM has shown that the microstructures contain NiSi2. This was approved by presence of LO phonon line in Raman back scattering spectrum. The control of micro-cone shape and height was achieved by changing the laser intensity and number of pulses. Keywords: “black silicon”, self-organization, microstructures, Nd: YAG laser.

2. Experiment To produce a sample of 30 nm nickel film of the 99.99 % purity has been deposited onto the Si substrate by DC magnetron sputtering method at room temperature. Before the sputtering the surface of Si was subjected to ion cleaning in Ar ambient. Afterwards the Ni/Si layered structure was exposed to the microsecond Nd:YAG laser beam at basic frequency, various threshold intensities and pulse numbers in a scanning mode. The structural and optical characteristics of microcones where studied by scanning electron microscopy (SEM) and Raman spectroscopy. SEM measurements were performed on the Carl Zeiss LEO 1455 VP device with Rontek EDXA attachment that have also allowed analyze the composition of the sample. Raman spectra of the Ni/Si sample were obtained in the various configurations, including a backscattering mode, using the laser excitation wavelength of 532 nm at room temperature. The spectral resolution was 1 cm-1 and the beam diameter on the sample surface was about 1 mm.

1. Introduction “Black silicon” [1] is a new type of material, which absorbs from 96% to 98 % of the incident light generating hundreds times more current than ordinary Si (60%). Therefore it can be an excellent material for solar cell production [2]. In addition this material could be also used in infrared detector production - a novel application of Si. Single crystal Si is transparent for infrared radiation, but “black silicon” is able to absorb it .The surface micro structuring of Si by femtosecond laser induced plasma [3] or chemical vapor deposition with catalytic metal on Si [4] is used for “black Si” formation. We propose another method - the micro-cone formation by the irradiation of Ni/Si layer structure by basic frequency of microsecond Nd:YAG laser. (a)

(b)

3. Results and Discussion Fig. 1 presents Ni/Si surface subjected to laser irradiation at different conditions. After Ni/Si structure irradition by Nd:YAG laser beam a various degree of damage is observed on the surface of the Ni/Si layered system such as the appearance of cracks and formation of regular cone-like structures. At the lowest intensity and number of laser pulses applied, the surface of Si is stongly cracked and nickel is present in the form of drops. The average drop size is about 0.5-0.8 μm and they are quite evenly distributed off the zones, where microcones are formed. Microcones appear not on a regular basis and their sizes vary from a tiny of 10 mm up to a bigger ones of 40 μm. The surface of microcones is covered with net dendride (c)

Fig. 1. SEM image of Ni/Si surface irradiated by Nd:YAG laser at intensity 3.15 MW/cm2 a) about 3 laser pulses per point, and at intensity 4.5 MW/cm2; b) about 10 laser pulses per point, c) about 22 laser pulses per point. 140

Articles


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

structures. The Nd:YAG laser intensity threshold at which the self-organization of cone-like microstructures was observed on the surface of Ni/Si layer system was 3.15 MW/cm2. The control of micro-cone shape and height was achieved by changing the intensity of laser radiation and a number of pulses. The increase of the laser intensity up to 4.5 MW/cm2 and number of pulses from 3 up to 10 does not result in the appearance of more microcones although the redistribution of nickel can be observed. The nickel net covering the microcones is moved up the cone side, leaving the bottom of the cone without any nickel. The further increase of the laser intensity and number of pulses leads to the formation of cone-like microstructures with most of Ni on the top of the cone and maximal height of the cone about 80-100 μm (Fig.1). The irradiated region of the structure with maximal intensity 4.5 MW/cm2 and 22 pulses becomes black (Fig.1). Moreover we can notice that microcones appear regularly and mostly of the same shape and height. No dendrite structure is found on the surface of microcones, the cone side is pure Si. During the research of the irradiated sample surface by SEM the “metallic islands” with a diameter less than 1 μm have been found. The micro-chemical analysis by means of energy-dispersive X-ray attachment has shown that these islands consist of pure Ni or Ni and Si composition NiSi2. It means that it is possible to form the new phase NiSi2 in our experiment. We assume that NiSi2 is formed via melting and recrystallization of a part of Si substrate. A one more evidence of NiSi2 formation is the appearance of a phonon line at 225 cm-1 which is attributed to Ni-Si vibration in Raman back scattering spectrum (Fig.2). We eliminate the possibility of NiSi phase formation as Ni-Si vibrations for NiSi posses at least two cha-

N° 4

2009

racteristic first-order peaks at 196 and 214 cm-1 [6]. At the same time slight oxidation of the structures is found after laser irradiation in comparison with untreated samples and the most oxidized regions follow NiSi2 formation.

4. Model of micro-cone formation Proposed mechanism of micro-cone formation is explained by two steps. In the first step Ni film melts after irradiation by laser beam and the influence of surface tension forces leads to Ni “metallic island” formation. At higher laser beam intensities islands transforms in to sphere-like particles containing Si and Ni. In the second step - Si melts outside each island, evaporates, accumulating on Ni and after diffusion through Ni causes microcone growth [5].

5. Conclusion We have demonstrated the possibility to form the “black Si” on the surface of Ni/Si structure by Nd:YAG laser radiation. The shape and height of micro-cone structure strongly depends on Nd:YAG microsecond laser intensity and number of laser pulses, that allows controlling this parameters. We have found that microcones are formed as a regular structure at the laser intensity and number of pulses leads to the formation of cone-like microstructures It was found that it is possible to form a silicide phase, namely NiSi2, on the Si surface by Nd:YAG laser radiation. ACKNOWLEDGMENTS We would like to acknowledge the valuable help of S.V. Gusakova in SEM measurements.

Fig. 2. Raman spectrum of the Ni/Si sample surface irradiated by Nd:YAG laser: A. - 4.5 MW/cm2, 10 laser pulses; B. - 3.15 MW/cm2, 3 laser pulses; C. - 4.5 MW/cm2, 22 laser pulses. Articles

141


Journal of Automation, Mobile Robotics & Intelligent Systems

AUTHORS Artur Medvid' - Riga Technical University, 14 Azenes Str., Riga, LV-1048, Latvia; Institute of Semiconductor Physics National Academy of Science of Ukraine, 45 Pr. Nauki, 252650, Kyiv-28, Ukraine. Aliaksandra Karabko, Anatoly Dostanko - Belarusian State University of Informatics and Radioelectronics, Minsk, Belarus. Pavels Onufrijevs* - Riga Technical University, 14 Azenes Str., Riga, LV-1048, Latvia. Edvins Dauksta - Riga Technical University, 14 Azenes Str., Riga, LV-1048, Latvia. * Corresponding author

References [1]

[2] [3]

[4]

[5]

[6]

142

Halbwax M., Sarnet T., Delaporte Ph., Sentis M., Etienne H., Torregrosa F., Vervisch V., Perichaud I., Martinuzzi S., “Micro and nano-structuration of silicon by femtosecond laser: Application to silicon photovoltaic cells fabrication”. Thin Solid Films, 2008, vol. 516, pp. 67916795. http://www.technologyreview.com/energy/21611/ Liu S., Zhu J., Liu Y., Zhao L., “Laser induced plasma in the formation of surface-microstructured silicon”. Materials Letters, 2008, vol. 62, pp. 3881-3883. Jeon M., Uchiyama H., Kamisako K., “Characterization of Tin-Catalyzed Silicon Nanowires Synthesized by the Hydrogen radical-assisted Deposition method”. Materials Letters, 2009, vol. 63, pp. 246-248. Jeon M., Kamisako K., “Aspect of aluminumcatalyzed silicon nanowires synthesized at low temperature and effect of hydrogen radical treatment”. Journal of Alloys and Compounds, 2009, vol. 476, pp. 84-88. Zhao F.F., Chen S.Y., Shen Z.X., Gao X.S., Zheng J.Z., See A.K., Chan L.H., “Applications of micro-Raman spectroscopy in salicide characterization for Si device fabrication”. Journal of Vacuum Science and Technology, 2003, vol. B 21, pp. 862-867.

Articles

VOLUME 3,

N° 4

2009


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N째 4

2009

HIGH SPEED SIGNAL PROCESSING FOR PHOTON COUNTING X-RAY DETECTION B. Shinomiya, A. Koike, H. Morii, T. Okunoyama, Y. Neo, H. Mimura, Toru Aoki

Abstract: Recently practical X-ray measurement systems are demanded energy distinction function. Photon-counting CdTe semiconductor detectors have a high energy resolution in a low count rate condition at room temperature. However, the energy resolution is decreased by pile-up phenomenon in a high count rate condition. In conventional signal processing, processing time estimated X-ray photon energy from the pulse waveform is about tens of microseconds. This time is depended on the pulse decay time. This paper purposes to maintain the high energy resolution by changing the signal-processing algorithm, which derived the pulse rise height of the output waveform from the CdTe detector in a high count rate condition. As a result, the pulse rise time required to estimate the pulse rise height was short about 100 ns at incident X-ray energy 60 keV. As the result of energy spectrum by using this data, the FWHM of about 11keV (at 60 keV) when the count rate of 500 kcps. This result shows the possibility that the photon counting sensor has application for the high count rate imaging without decrease of the high energy resolution.

this digital crest value is treated as an incident X-ray photon energy. The CdTe semiconductor detector used conventional signal processing has a high energy resolution in a low count rate condition at room temperature. However, the energy resolution is decreased by pile-up phenomenon in high count rate. Therefore, in this paper, we tried proposing a new signal processing (pulse rise height signal processing) used without degradation of energy resolution even if pile-up is occurred.

Fig. 1. Block diagram of conventional signal processing of the photon-counting.

Keywords: X-ray, CdTe, photon counting.

2. Principle 1. Introduction X-ray radiography is a method which insight information of the object can be inspected non-destructively. Recently, there is a demand of obtaining material information such as atomic number and electron density in addition to insight shape information. The material discriminated X-ray computed tomography system which for obtaining was had been already reported by using conventional micro-focused X-ray tube and photon counting type CdTe X-ray imager with X-ray energy information [1]-[4]. A photon-counting method is used for obtaining the X-ray energy information and atomic number mapping images were estimated by photon energy information. A photon-counting imaging method make usually slow imaging method because the photon-counting have to treat the incident X-ray photons individually. Therefore, there is a request for using a photon-counting in a high incident rate for fabrication the high-speed X-ray imager and the real-time X-ray imager with energy information. Figure 1 shows block diagram of conventional signal processing of photon counting method by multichannel analyzer (MCA). Waveform from preamplifier is shaped to integral waveform with decay curve by analog integrator inside MCA. Crest value of the integral waveform is converted analog information into digital information, and

2.1. Estimation method of X-ray phone energy The pulse rise height which changed shape by preamplifier is proportional to the x-ray energy. We considered that we can obtain the each photon energy by using the pulse rise height even if pile-up is occurred, and this signal processing is adapted for more high count rate condition than for conventional signal processing. It is because that the pulse rise time is considerably faster than the pulse decay time. The time which obtained the X-ray energy is considerably faster than conventional signal processing used the area of pulse by using pulse rise height. Experimental system was used the CdTe n-p-n diode detector (2x2x0.5 mm-thick) (electrode: n-In, p-Au) under 200 V bias at room temperature, the preamplifier (Clear Pulse) and Am-241 radioisotope as a checking source. The output of preamplifier was connected to digital storage oscilloscope (DSO: Lecroy WS452), and each data saved in oscilloscope was analyzed. 2.2. Analytical method for obtaining pulse rise height In Fig. 2, solid line shows the data obtained from DSO, and dash line shows 50 ns moving average line of solid line. For obtaining the pulse rise height, the two points which before and after pulse rising are had to decide. Articles

143


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

Therefore, we decide the two points which gradient value is about 0 in before and after pulse rising from the moving average line. The pulse rise height was obtained by assigning those different two points which gradient value is about 0, as shown allow line in Fig. 2.

N° 4

2009

In Fig. 4(a), (b), those two energy spectrums show that the conventional signal processing by MCA and the pulse rise height signal processing have same high energy resolution, and two energy spectrums (b), (d) show that non pile-up condition and pile-up condition have same high energy resolution of 4 keV by using the pulse rise height signal processing. On the other hand, energy resolution of 20 keV in Fig. 4(c) was 5 times as large as Fig. 4(a). The FWHM were estimated by energy proofed from channel. This experiment showed high energy resolution was maintained up to about 50 kcps condition by using the pulse rise height signal processing. But count rate was only 50 kcps, our aimed count rate condition is over 1 Mcps. Therefore, next experiment condition was high count rate condition in over 50 kcps.

Fig. 2. Analytical method for obtaining pulse height. Solid line is the oscilloscope data obtained from the preamplifier. Dash line is 50ns moving average line of the continuous line.

-

3. Experiment and result We carried out three kinds of experiments. First experiment shows comparison the pulse rise height signal processing and conventional signal processing used MCA in two condition of non pile-up and in pile-up. Second experiment is run in a high count rate condition of less 1 Mcps by using the pulse rise height signal processing. In third experiment, atomic number and electron density was obtained by using the pulse rise signal processing. 3.1. Experiment in two condition of non pile-up and in pile-up We carried out experiments in a non pile-up and a pileup by using the pulse rise height signal processing and conventional signal processing used MCA. Preamplifier was used Clear Pulse-5102B. (Rise time is about 500 ns, the pulse attenuation constant is 60 Îźs.) Fig. 3(a) was output waveform from DSO in a few kilo cps of non pile-up condition. Fig. 3(b) was output waveform from DSO in about 50 kcps of pile-up condition.

48 36

-

24 12 0

-

Fig. 3. Those two figures show digital signal waveform from oscilloscope when Am241 was used. Fig.(a) was output waveform from DSO in a few kilo cps of non pile-up condition, gamma source was Am of 3.7Mbq x 1, distance of detector and gamma source was 200 mm. Fig.(b) was output waveform from DSO in about 50k cps of pile-up condition, gamma source was Am of 3.7Mbq x 3, distance of detector and gamma source was a few millimeter. 144

Articles

Fig. 4. Energy spectrums used Am241 in non pile-up and pile-up.(a)(c) two figures were used the conventional signal processing, and (b)(d) two figures were used the signal processing of the pulse height. (a)(b) two figures is in non pile up shown (a) Fig. 3, and (c)(d) two figures is in pile-up shown (b) Fig. 3.


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

3.2. Experiment in high count rate condition We carried out experiment by using the pulse rise height signal processing in a high count rate condition. In order to measure energy spectrum in a high count rate condition, white X-ray tube source was used together with Am241 which as checking source. This experimental block diagram is shown in Fig.5, the fast preamplifier was Clear Pulse-5015S. (Rise time is about 100 ns, the pulse attenuation constant is 60 μs), and measurement time of each energy spectrum was 30 ms.The voltage of white X-ray tube was 40 kV, and Am241 has main peak energy of 59.5 keV.

Fig. 5. Schematic of measurement system. X-ray tube voltage was 40kV. X-ray tube current was 0, 20, 50, 80 μA. The distance of X-ray tube and detector was 170 mm, Am241 and detector was a few millimeter. 300

300

250

250

200

200

150

150

100

100

50

50

0

0

300

300

250

250

200

200

150

150

100

100

50

50

0

0

Fig. 6. Energy spectrums used white X-ray and Am241 in high count rate condition by a using fast preamplifier (CP5015S). Figure (a) is used only Am241. Figure (b), (c) and (d) was used Am241 and white X-ray. X-ray tube vol-tage is 40 kV. Preamplifier is CP5015S (Rise time is about 100 ns by using Am241. The pulse attenuation constant is 60 μs). In Fig. 6, the count rate was become higher with increase of the X-ray tube current. Those energy spectrums shown in Fig. 6 show the gradation of energy resolution with a high count rate and unoriginal counts at high energy region with high count rate. We considered that the origin of degradation of energy resolution with a high count rate and the origin of increase of unoriginal counts at high energy region with high count rate are increase of the overlap of previous and next pulse at the leading edge with high count rate. However, when count rate was 500 kcps in Fig. 6 (c), energy resolution was 11.0keV. This energy resolution was only 2.3keV lager than the reference energy resolution 8.7keV which used only Am241 as shown in Fig. 6(a). We couldn’t find so large degradation of energy resolution up to 500 kcps condition. From a result of this experiment, in 300 kcps which is ten times from experiment condition used the conventional signal

N° 4

2009

processing, degradation of energy resolution degrease from reference energy resolution is less than 1.0 keV, and energy resolution is not over 9.7 keV. In next experiment, we tried to obtain effective atomic number in 300 kcps condition for checking the accuracy of atomic number. 3.3. Obtaining effective atomic number and electron density Energy spectrum was measured by using the pulse rise height signal processing in 300 kcps condition, and effective atomic number and electron density was calculated by using the calculation technique of dual energy X-ray CT (DXCT). The DXCT’s theory is as follows [5], [6]. 3.3.1. Measurement effective atomic number and electron density In this section, effective atomic number and electron density of Al plate were obtained by used expressions calculated effective atomic number and electron density of DXCT from μ of experimental value. Fig. 7 shows schematic of measurement system. Detector and preamplifier (fast amp: CP5015S) were same as for previous experiment 3.3, and incident X-ray energy was obtained by using the pulse rise height signal processing. X-ray tube voltage was 150 kV. X-ray tube current was 10 μA. Sample was 20 mmthick Al plate.

Fig. 7. Schematic of measurement system. X-ray tube voltage was 150 kV. X-ray tube current was 10 μA. Sample was 20 mm-thick Al plate. The distance of X-ray tube and sample was 500 mm, sample and detector was 300 mm. Table1 shows that results of experiment for obtaining effective atomic number and electron density, and effective atomic number was 13.1 and electron density was 1.21×1024 cm-3, count rate was 300 kcps, energy resolution estimated from previous experiment 3.3 in this count rate was about 10 keV, total photon count (10~150 keV) was 60 k counts, Minimum measurable energy was about 10 keV, and two energy range used in the calculation technique of DXCT were 35-50 keV and 85-90 keV. Table.1 result of experiment for obtaining effective atomic number and electron density of Al plate. Effective atomic number Electron density Measurement time Count rate Total count Energy resolution Utilizable energy range Energy range used in DXCT

13.1 1.21×1024 cm-3 0.2 s 300 kcps 60k counts About 10 keV 10~150 keV 35-40 keV, 85-90 keV

Articles

145


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

3.3.2. Estimation of a relationship between effective atomic number and total photon count Fig. 8 shows effective atomic number and electron density when total photon count was decreased from 60k counts. As total count lower than about 36k counts, error of effective atomic number was bigger than 13: atomic number of Al. Energy spectrums in total count rate condition of 60k counts and 15k counts were shown in Fig. 9. In Fig. 9, shapes of spectrums transmitted Al plat conformed in each total count condition. On the other hand, Fig. 9 shows different shape of spectrums transmitted no sample in each total count condition between 25 keV and 55 keV. The differences of shape cause increase in error of atomic number to theoretical atomic number. We consider, by choosing appropriate two energy range, there is a possibility that high calculation accuracy obtained effective atomic number and electron density is obtained in a low total photon count condition.

AUTHORS B. Shinomiya*, A. Koike, H. Morii, T. Okunoyama, Y. Neo, H. Mimura, Toru Aoki - Research Institute of Electronics, Shizuoka University, Hamamatsu, Japan. E-mail: shinomiya@nvrc.rie.shizuoka.ac.jp. * Corresponding author

References [1]

[3] [4]

[5]

Fig.9. Energy spectrums in total count rate condition of 60k counts and 15k counts.

4. Conclusion We have proposed the decrease of degradation of energy resolution with high count rate by using the signal processing of using pulse rise height in pile-up condition. We could obtain about the same energy resolution (4 keV) as conventional signal processing in non pile-up and we could maintain this energy resolution even if pileup is occurred. Moreover, we showed that energy resolution 11.0 keV in about 500 kcps, and we couldn’t find so large degradation of energy resolution with the high count rate. So, there is a possibility used the photon counting in the high count rate condition without decrease of the high energy resolution. Effective atomic num146

Articles

2009

ber of Al plate sample was shown 13.1 in 300 kcps condition and a possibility that obtaining higher-accuracy effective atomic number with fewer total photon count. Those results are indicated the possibility that the photon counting sensor has application for high speed imaging.

[2]

Fig. 8. Change of effective atomic number and electron density when total photon count was decreased from 60k counts.

N° 4

[6]

Nilaura M., Nakamura A., Aoki T., Hatanaka Y., Phys. Stat. Sol. B, no. 229, 2002, pp. 83-87. Aoki T., Ishida Y., Sakashita D., Gnatyuk V.A., Nakamura A., Tomita Y., Hatanaka Y., Temmyo J., ”Development of energy discriminated CdTe imaging detector for hard Xray”. In: Proc. SPIE 5540, 2004, pp. 196-205. Aoki T., et al., IDW/AD’05, 2005, p. 4283. Zou W., Nakashima T., Onishi Y., Koike A., Shinomiya B., Morii H., Neo Y., Mimura H., Aoki T., “Atomic Number and Electron Density Measurement Using Conventional X-ray Tube and a CdTe Detector”. Japanese Journal of Applied Physics, vol. 47, no. 9, 2008, pp. 7317-7323. Ohno Y., Torikoshi M., Tsunoo T., Hyodo K., “Dual-energy X-ray CT with CdTe array and its extension”. Nuclear Instruments and Methods in Physics Research. Section A, no. 548, 2005, pp. 72-77. Torikoshi M., Tsunoo T., Endo M., Noda K., Kumada M., Yamada S., Soga F., Hyodo K., “Design of synchrotron light source and its beamline dedicated to dual-energy X-ray computed tomography”, Journal of Biomedical Optics, vol. 6(3), 2001, pp. 371-377.


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N째 4

2009

THERMIONIC VACUUM ARC DIAGNOSTIC USING EMISSIVE PROBE

Vasile Tiron, Constantin Aniculaesei, Gheorghe Popa

Abstract: In this paper, experimental results are presented on study of the copper thermionic vacuum arc using emissive probe. Experimental results show that plasma potential is direct related to the discharge voltage and it shows nonlinear distribution from the anodic melted spot towards to the plasma extremity.

For ignition and maintaining of the discharge is necessary that thermoelectrons coming from tungsten filament to be focused and accelerated towards the anode so as to cause melting and evaporation anodic material and to ensure in interelectrodic space a sufficiently high pressure vapours of the anodic material [5].

Keywords: thermionic vacuum arc, emissive probe.

1. Introduction Thermionic vacuum arc (TVA) is an electrical discharge burned in the vapours of a metal anode, vapours produced by bombarding anodic material with an energetic thermoelectronic beam [1]. The main advantage of this method is the easy control of the ion energy and the fact that during deposition of the metallic thin layer this is continuously bombarded with energetically ions without the need to bias the substrate [2]. Another advantage of the method is the presence of a high energetic electron density around the anode which provides a high rate of deposition, and that this method is very useful in deposition of refractory or ceramic materials [3]. These benefits correlated with discharge that occurs under high vacuum conditions, the deposited layer is characterized by high purity and hardness, good adhesion, low roughness and high compaction [4]. 1.1. Experimental set-up The experimental setup consists of a stainless steel chamber, equipped with various ports for attachment of electrical and optical diagnostic systems, in which the electrodes of the discharge are placed. To obtain a high vacuum condition it is used a pumping system consisting in a preliminary vacuum pump and a turbo-molecular pump. Electrodes system consists of an anode and an electronic gun with role of cathode (Figure 1). The anode is shaped as a crucible which contains the material to be evaporated, in this case copper. Cathode is made in the form of a coil with 4 loops of tungsten wire (0.5 mm thickness) surrounded by the Wehnelt cylinder used for termoelectrons focus. Both, the cathode and the vacuum vessel are grounded, while the anode is at a very high potential to them. Electronic gun is mounted on a linear displacement system that allows modification of the distance between electrodes during operation of termionic vacuum arc. The angle between the cathode axis and normal direction to anode surface is in this case fixed at value of 60째.

Fig. 1. Experimental setup. 1.2. Results and discussions Thermionic vacuum arc can be established in vacuum conditions between a heated cathode and an anode placed at small distances in front of the cathode. For a convenient applied d.c. voltage across the cathode and anode space, a melted spot appears on the anode surface and a continuous evaporation of the anode material from this melted spot is established due to the accelerated electrons emitted from the cathode and incident on anode. Consequently, in vacuum, a steady state density of the metal vapors appears in the interelectrodic space. At further increase of the applied high voltage, suddenly a bright discharge appears in the interelectrodic space in the vapours of the anode material, with a simultaneous decrease of the voltage drop between electrodes and with a significant increase of the current. The anode fall must be high enough to ensure the vapor production at the anode after ignition. The cathode fall, which depends on the relative position of the electrodes and the cathode temperature, must ensure enough ions in proximity of the cathode. Since the cathode is connected to the grounded vessel, the potential drop between the interelectrodic plasma and the vessel is about the cathode fall. Ions produced in the interelectrodic plasma arrive at the metallic walls of the vessel with the energy represented by plasma potential (hundreds of volts).

Articles

147


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

Fig. 2. Plasma potential versus discharge voltage measured at a fixed position (3 cm from anode).

N° 4

2009

(arc current, arc voltage, filament temperature, relative position of the electrodes and anode material) this dependence presents different sectors with different evolution rate of plasma potential. Branch ab corresponds to the weakly ionized plasma where the plasma potential increases linearly from vessel wall towards anode. Next branch (bc) corresponds to the transition region where plasma potential increases rapidly to a high value which corresponds to a double layer. Branch cd corresponds to the bright plasma area where the plasma potential is almost constant close to the anode potential. This area is like a â&#x20AC;&#x153;fireballâ&#x20AC;? where the termoelectrons are trapped and forced to excite and ionize the metallic atoms evaporated from anode. a)

The ions energy depends on the operating parameters like arc voltage, current intensity through filament and the geometrical parameters, like the relative position of the electrodes. Thus, plasma potential is an important parameter in thin layer deposition process. To determine the plasma potential an emissive probe was used. In the Fig. 2 the plasma potential, measured as floating potential of the emissive probe placed in a constant distance from the anode (3 cm), is presented versus discharge voltage. a) b)

Fig. 4. Axial distribution of plasma potential for a constant arc current and different voltage arc (a) and for different values of arc current (b).

b)

The spatial distribution of plasma potential strongly depends on operation parameter like arc voltage and arc current and its value is direct related to the voltage arc (fig. 4). For plasma processing (thin film deposition) is necessary to produce plasma with high energetic ions and high ionization degree. Consequently, is useful to operate the arc discharge at high anode voltage and low current value.

2. Conclusion Fig. 3. TVA plasma image (a) and corresponding axial distribution of plasma potential (b), arc current - 180 mA, arc voltage - 750 V; dashed line - emissive probe path. In Fig. 3b is illustrated the spatial distribution of plasma potential measured on normal direction to the anode surface. Depending on the operating parameters 148

Articles

The TVA is a valuable tool for basic physical investigation related to plasma assisted deposition process due to its flexibility and accessible range of operation parameters. The energy of ions is direct related to plasma potential which can be fully controlled and changed even during deposition.


Journal of Automation, Mobile Robotics & Intelligent Systems

VOLUME 3,

N° 4

2009

ACKNOWLEDGMENTS This work was supported by Romanian Ministry of Education, Research and Innovation under grant PC “CAPACIF” no. 72223/2008.

AUTHORS Vasile Tiron* - Research Department, Faculty of Physics, “Al. I. Cuza” University, Iasi, 700506, Romania. E-mail: vasile.tiron@uaic.ro. Constantin Aniculaesei, Gheorghe Popa - Faculty of Physics, “Al. I. Cuza” University, Iasi, 700506, Romania. E-mails: aniculaeseiconstantin@yahoo.com, ghpopa@uaic.ro. * Corresponding author

References [1]

[2]

[3]

Ehrich H., Schuhmann J., Musa G., Popescu A., Mustata I., „Adhesive metal films obtained by thermionic vacuum arc (TVA) deposition”, Thin Solid Films, vol. 333, 1998, pp. 95-102. Musa G., Ehrich H., Mausbach M., „Studies On Thermionic Cathode Anodic Vacuum Arcs”, J. Vac. Sci. Technol., vol. A12, no. 5, 1994. pp. 2887-2895. Kuncser V., Mustata I., Lungu C.P., Lungu A.M., Zaroschi V., Keune W., Sahoo B., Stromberg F., Walterfang M., Ion L., Filoti G., „Fe-Cu granular thin films with giant magnetoresistance by thermionic vacuum arc method: Preparation and structural characterization”, Surface & Coatings Technology, vol. 200, no. 1-4, 2005, pp. 980 983.

Articles

149


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

INVESTIGATION OF LOW TEMPERATURE PLASMA CAPABILITIES TO MODIFY THE STRUCTURE AND FUNCTION OF BIO-POLYMERS Iuliana Motrescu, Takuya Hara, Akihisa Ogino, Shigeyasu Tanaka, Taketomo Fujiwara, Hirokazu Kawagishi, Shinya Kodani, Gheorghe Popa, Masaaki Nagatsu

Abstract:

2. Material and Method

The possibility to modify biopolymers using low temperature plasma is investigated. Two types of plasma are employed in this study: an atmospheric pressure plasma and a low pressure microwave plasma. In both cases changes of the structure and function of the exposed molecules are reported, the effects being stronger for the low-pressure plasma irradiation.

Low temperature plasma processing of proteins would take advantage of all the features offered by these plasmas. In order to elucidate the mechanisms involved in plasma – protein interaction, we focused on simpler molecules due to the complicated structure of proteins, which makes it almost impossible to figure out the modifications that occurred under plasma exposure [5], [6]. Our study involved amino acids and peptides. A peptide is a bio-molecule similar to protein, made of up to 50 amino acids chained by peptide bonds. As bio-molecules used to test low temperature plasma treatment potency, we have chosen a nonapeptide, Arginine Vasotocin (AVT or [Arg8] Vasotocin). [Arg8] Vasotocin is a peptide Cys-Tyr-Ile-Gln-Asn-Cys-Pro-Arg-Gly-NH2 with a disulfide bond between Cys1 - Cys6, having the chemical structure C43H67N15O12S2. AVT is the progenitor of all vertebrate neurohypophyseal hormones. Among other functions, Arginine Vasotocin is involved in the osmoregulation process in some non-mammals [7], [8], [9]. The samples were prepared by solving AVT amorphous powder in pure water, placing the solution on small Silicon plates and drying them, all of these in order to obtain a homogenous treatment of the sample. Atmospheric pressure plasma and microwave excited surface wave low-pressure plasma were used to investigate the effect of cold plasmas on these bio-molecules in different conditions. In both cases plasma was optically and electrically diagnosed. For the atmospheric pressure plasma we considered two different regimes of the discharge due to the nature substrate sample: conductive (silicon) or dielectric (glass). The modifications produced on the bio-molecule by both types of discharges were tested by means of presence of biological function, structure modification investigated by X-Ray Photoelectron Spectroscopy (XPS), and Time of Flight Mass Spectrometry (TOF-MS). The configuration used to produce the atmospheric pressure plasma, presented in Figure 1, is very simple: a high voltage pulse is applied on an electrode placed around a glass tube. The discharge gas in our experiments was Helium. The low-pressure plasma was produced in a stainless steel chamber of 17 cm height and 25.5 cm diameter using Argon as discharge gas and 500 W microwave power [10]. The samples were placed at about 8 cm inside the chamber. For all the experimental conditions the treatment time was 10, 20, 30 minutes, respectively. In each case control samples were prepared for each set of exposed items.

Keywords: biopolymers, biomolecules, low temperature plasma, modification, plasma processing.

1. Introduction Applications involving low temperature plasmas have been developed in the recent years. All of them take advantage of the cold plasmas features of being suitable for heat resistant materials treatment, being non-toxic, a better alternative to chemical modification, and many others. Low-pressure plasmas, but also atmospheric pressure plasmas, have been employed in such applications [1]-[4]. The aforementioned have even more advantages, since they don’t need a vacuum system and have a simpler electrode configuration, being easier to manipulate. Biopolymers are polymers produced in the living organisms and involved in biological processes. Well known molecules among them are DNA, enzymes, starch. We are interested in proteins, which are biopolymers containing amino acids as small repetitive structure. Proteins are very common in living organisms from bacteria spores to human body, participating in almost every process in the body, and having various functions from enzymatic to forming the cytoskeleton of cells. The primary structure is chain of amino-acids linked by peptide bonds, the secondary structure is formed by regularly repeating local structures, involving hydrogen bonds, while the tertiary structure consists of several secondary structures stabilized by non-local interactions as hydrophobic, salt bridges, hydrogen bonds, disulfide bonds, etc. The organization process for the tertiary structure is called folding. Some proteins have even a quaternary structure where some sub-protein subunits bind together. The four levels of spatial organization make proteins bio-molecules with a complicated three-dimensional structure. Although the atomic species present in proteins are limited, and there are only 20 essential amino – acids involved in protein structure, the numerous type of bonds and number of amino – acid residues make protein molecules very complex and very stable.

150

Articles


Journal of Automation, Mobile Robotics & Intelligent Systems

Fig. 1. Atmospheric pressure plasma device.

3. Results and discussions

VOLUME 3,

N° 4

2009

The interpretation of the obtained results is quite difficult since AVT molecule has a very complicated structure. The main type of bonds in this molecule (that we should correlate with the spectra for the control sample) are: -NH2, C-C(=O)-N, C-C, C-N-C, N-H, C-CH3, C-S, one benzene with a OH , C-O, and a penta cycle with simple bonds between four Carbon atoms and a Nitrogen. A rough analysis indicates that atmospheric plasma produces modification of Vasotocin, small changes after 10 and 20 minutes, and more profound after 30 minutes of exposure. XPS analysis indicated a strong reduction of S 2p peak possible indicating the break of disulfide bond in AVT molecules. The peaks corresponding to N-C=O and C-N from C 1s spectra exhibited a decrease after plasma irradiation (Figure 2), confirmed by N 1s spectra for the treated and control samples. The bonds involving oxygen in AVT samples also modified, -OH and O=C peaks decreasing.

3.1. Atmospheric pressure plasma treatment We investigated the optical emission spectra for both regimes produced in the case of atmospheric discharge and we concluded that for the silicon substrate, the relative intensity of the light emitted by the plasma is almost twice stronger than for the glass substrate. Moreover, in the former case the emission is recorded due to more species, compared to the latter case (mainly radicals containing oxygen). These findings made us decide that the discharge produced when the irradiated substance is placed on conductive sample is more suitable for bio-molecule structure modification. Vasotocin samples (on silicon substrate) were treated at an applied voltage of 10 kV peak to peak for 10, 20 and 30 minutes. The XPS spectra of control and 30 minutes samples are presented in Figure 2.

Fig. 3. TOF-MS spectra of AVT treated and control. Treated Vasotocin kept its biological function after atmospheric plasma treatment and the mass was also conserved as TOF-MS measurements indicated.

Fig. 2. XPS spectra of AVT after and before AP treatment.

3.2. Low pressure plasma treatment In the other set of experiments, microwave excited surface wave plasma was produced. Samples were prepared on Silicon substrates and exposed for the same time intervals. The evaluated results indicated a stronger effect than for the atmospheric pressure plasma, which was expected due to higher particles energies and maybe also presence of a larger range of radiations including UV. Same tests were employed as for the other samples. Sulphur S 1s peak caused by the presence of disulfide bond was not present after sample irradiation even for 10 minutes. The intensity of Nitrogen 1s spectrum strongly decreased, indicating the breaking of C â&#x20AC;&#x201C; N bondings, fact confirmed by the fitting of C 1s peak. Disulfide bond causes a peak around 165 eV for the control Vasotocin sample but it doesnâ&#x20AC;&#x2122;t exist anymore for the treated vasotocin. Results indicate that the structure of the molecule was modified by rearranging of its three dimensional structure. Plasma processing did not fragmentize Articles

151


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

the peptides as TOF-MS spectra of treated and control samples show (Figure 3). The conformational change induced by plasma processing is also proved evaluating the biological function of Arginine Vasotocin molecules after treatment. The procedure consists in measuring the time evolution of the water volume flow through a small part of frog abdominal skin. The results are shown in Figure 4, in blue for the control sample, and in red for the 30 minutes processing. As measurements indicate, the exposed molecules don’t contribute anymore to osmoregulation and we assume that this is due to the conformational change caused by plasma treatment.

[2]

[3]

[4]

[5]

[6]

[7]

Fig. 4. Evaluation of AVT biological function. [8]

4. Conclusions Our study proved that low temperature plasmas are able to produce modifications of biopolymers, such as peptides. Plasma processing determined the rearrangement of the three dimensional structure without altering molecular mass, which means without fragmentize the molecule. Moreover, the changes induced by the low-pressure plasma exposure caused the lost of biological function of osmoregulation. The new conformation was not able to bind the receptor and produce the opening of the water channels in frog belly. Further investigations are necessary to improve the treatments and also elucidate the produced modifications and the interaction mechanisms. ACKNOWLEDGMENTS This work has been partly supported by a Grant-in-Aid for Scientific Research from Japan Society for the Promotion of Science (JSPS).

AUTHORS Iuliana Motrescu, Takuya Hara, Akihisa Ogino, Shigeyasu Tanaka, Taketomo Fujiwara, Hirokazu Kawagishi, Shinya Kodani, Masaaki Nagatsu* - Graduate School of Science and Technology, Shizuoka University, Hamamatsu, 432-8561, Japan. E-mail: tmnagat@ipc.shizuoka.ac.jp. Gheorghe Popa - Faculty of Physics, Alexandru Ioan Cuza University, Bd. Carol I 11, 700506, Iasi, Romania. * Corresponding author

References [1]

152

Nagatsu M., Terashita F., Nonaka H. Xu L., Nagata T., Koide Y., "Effect of oxygen radicals in low-pressure

Articles

[9]

[10]

N° 4

2009

surface-wave plasma on sterilization", Appl. Phys. Lett. No. 86 , 2005, p. 211502. Eto H., Ono Y., Ogino A., NagatsuM., "Low Temperature Internal Sterilization of Medical Plastic Tubes Using Linear Dielectric Barrier Discharge", Plasma Process. Polym., no. 5, 2008, pp. 269-274. Ogino A., Noguchi S., Nagatsu M., "Effect of Plasma Pretreatment on Heparin Immobilization on Polymer Sheet", J. Photopolym. Sci. Technol., vol. 22, no. 4, 2009, pp. 461-466. Morent R., De Geyter N., Verschuren J., De Clerck K., Kiekens P., Leys C., “Non-thermal plasma treatment of textiles”, Surf. Coat. Technol., no. 202, 2008, pp. 34273449. Gao C.H., Herald T.J., Muino P.L., “Application of a Plasma Reactor to Modify Egg Ovalbumin for Improved Solubility”, J. Food Sci., no. 1, 2001, pp. 89-94. Mogul R., Bol'shakov A.A., Chan S.L., Stevens R.M., Khare B.N., Meyyappan M., Trent J.D., “Impact of LowTemperature Plasmas on Deinococcus radiodurans and Biomolecules”, Biotechnol. Prog., no. 19, 2003, pp. 776-783. Tanaka S., T. Hasegawa T., Tanii H., Suzuki M., "Trends in Compartaive Endocrinology and Neurobiology", eds. H. Vaudry, et al., Ann. N.Y. Acad. Sci. no. 1040, 2005, pp. 483-485. Malvin G.M., “Vascular effects of Arginine Vasotocin in toad skin”, Am. J. Physiol., no. 265, 1993, R426. Donald J.A., Trajanovska S., “A perspective on the role of natriuretic peptides in amphibian osmoregulation”, Gen. Comp. Endocrin., no. 147, 2006, pp. 47-53. Motrescu I., Ogino, Tanaka S., Fujiwara T., Kodani S., Kawagishi H., Popa G., Nagatsu M., “Modification of Peptide by Surface Wave Plasma Processing”, submitted to Thin Solid Films.


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

TESTING OF CARBON FIBRES AS TOOL ELECTRODES IN MICRO ELECTRICAL DISCHARGE MACHINING Leszek Kudla, Anna Trych

Abstract: The aim of this paper is to present the outcomes of the experiments concerning micro electrical discharge machining with the use of carbon fibres as tool electrodes. Key properties of carbon fibres are discussed and the procedure of such electrodes fabrication is also described. Conducted tests indicate promising perspectives for machining with such electrodes by means of shaping microcavities, but at the same time they reveal problems. A longitudinal machining as well as experiments with different electric parameters were presented. Finally, the influence of discharge energy is being considered.

along the axis of the fibre. A percentage of carbon in the final fibre varies from 92 up to almost 100 per cent. Carbon fibres can be also made of other precursors such as pitch or cellulose. The process of receiving filaments from these precursors is likewise a very complex one and the properties of obtained products are strongly determined by an applied procedure. Taking into consideration a process of EDM the most important properties are: electrical resistivity, heat stability, thermal conductivity or heat capacity. Selected properties of carbon fibres are presented in Table 1. Table 1. Properties of carbon fibres.

Keywords: electrical discharge machining (EDM), micromachining, carbon fibres electrodes.

1. Introduction Materials used in the process of Micro Electrical Discharge Machining (μEDM) are the same as the materials used in EDM e.g. tungsten, sintered tungsten carbides, sometimes also copper, brass, graphite and various composites even based on diamond [1]. Nonetheless, the procedure of application of these materials into a micro scale is needed. Sometimes special preparations must be undertaken to produce microelectrodes [2]. While searching for materials, which could be used in this process, it is essential to pay attention to some key properties that are required. Among them are: especially good electrical conductivity, thermal and wear resistance in electrical discharge conditions. Thus, the usage of carbon fibre seems to be adequate for described purpose. Primarily the evaluation of a possible use of carbon fibres as tool electrodes in EDM was described in paper [3].

2. Important properties of carbon fibres with regard to their application in EDM The carbon fibres are well known for their very good mechanical properties. They are often used to produce carbon fibre reinforced composites. Because of this, they can be used to replace metals in many uses. The carbon fibres have diameters between 5 and 10 micrometers. They are mainly made of polyacrylonitrile (PAN) filaments as a precursor by its oxidation and pyrolysis. The whole process of how the carbon fibre is made is extremely complex and long. The main process operations of PAN precursor are thermal cyclization (ring closure) and dehydrogenation that consists of replacing CH groups with C atoms and CH2 with CH groups. At the end of the processing the carbon fibre (or graphite fibre) is composed of carbon atoms, which are bonded together in chains and aligned

Property Young's modulus Symbol E [Unit] [GPa] Value

230-490

Tensile strength Rm [GPa]

Electric resistivity rR [Wm]

Heat stability T [°C]

~3

0.18×10-8

3000 *)

*) in protective atmosphere or in vacuum, sublimation temperature 3650 °C. The carbon fibre is approximately 3 times stronger than similar steel wire. Noticeable is the very low specific electrical resistance and good thermal stability. These properties are considered to be very advantageous for the μEDM process.

3. Fabrication procedure of carbon fibre electrodes To use carbon fibres as tool electrodes they must be specifically adapted. Because of their dimension and properties it is difficult to manipulate them and to machine a work piece. For the purpose of the experiments carbon fibre electrodes having a length shorter than 1 mm were fabricated - Figure 1.

Fig. 1. Properly fabricated tool electrode with carbon fibre length ~0.7 mm. Articles

153


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

The fabrication procedure was as follows: at first, a brass shank was prepared (f1x20 mm); next, the shallow hole was drilled in the front of this shank. The obtained hole was filled with a soft alloy of good plasticity and electric conductivity. Then, a conical tip was formed mechanically. Afterwards, a microhole was drilled in the tip and a single carbon fibre was inserted. Finally, it was fixed into the tip by plastic deformation of the connecting alloy and the end of the fibre was trimmed - Figure 2. In the process of fabrication of electrodes the most difficult task was to achieve a provision of axial position and parallelism to shank surface.

4. Experiments 4.1. Initial successful trials with the use of carbon fibre electrodes The first μEDM trials were conducted with short carbon fibre electrodes for sinking of microcavities - Figure 3. They revealed many problems, especially with regard to the provision of axial position of the carbon fibre. The material used in all experiments as a specimen was a gauge block of thickness of 1.33 mm made of hardened chromium steel GCr15. As a dielectric fluid cosmetic kerosene was applied. Independent variables were: voltage U, resistance R, capacitance C, polarity. Controlled variables included average current intensity of discharges I and in some cases total time of sinking t. Obtained cavities were identified by microscope observations and with the use of the Talysurf 10 Taylor Hobson measurement system intended for examination of micro-geometry of surfaces.

2009

According to Palatnik formula: RER = C r lC T 2

(1)

were: C [J/(kg·K)] - heat capacity, r [kg/m3] - density, lC [W/(m·K)] - thermal conductivity, T [°C] - melting temperature erosive wear resistance was calculated for electrode materials - Table 2. Table 2. Erosive wear resistance according to Palatnik formula. Material RER (x109)

Fig. 2. Tip of a newly fabricated carbon fibre electrode.

N° 4

Cu 1620

W 6029

Zn 53.8

Ni 582

Erosive wear resistance (RER) calculated for carbon varies from 2.22×109 to 217×109 (depending on assumed values for carbon or graphite). This rate is much lower than on other electrode materials. Therefore, a solution for electroplating by Ni layer or other may be applied. It seems to be a promising concept for improvement of erosive wear resistance. Nevertheless, the diameter of tool electrodes is bound to be bigger than in the previous concept. Another solution would be a usage of longer electrodes supported by a special guide-eye. In this concept the electrode could be long enough to machine assumed length of cavity. A further step is a quasi-continuous electrode made of a carbon filament to avoid the rapid wear problem Figure 5. Such electrode could be long enough to machine cavities continuously.

TE

DIELECTRIC FLUID

WP f0 fn

Fig. 5. Drive concept of quasi-continuous electrode.

Fig. 3. Shape of microcavities machined by carbon fibre electrode, influence of electrode inclination (a), depth g=1.6÷2.4 μm (b) [4]. Next critical problem is erosive wear resistance Figure 4.

4.2. Investigations concerning longitudinal sinking These are not only round holes that are requisite in micro technologies. Some microelements must have special shape cavities. With the use of other materials as electrodes the kinematics of such process is similar to milling. With the use of carbon fibre as an electrode it is troublesome to machine such cavities due to its whippiness. Special procedures have been applied to machine the longitudinal cavity - Figure 6.

20 μm

Fig. 4. Tip of 7 μm carbon fibre electrode after sinking. 154

Articles

Fig. 6. Cavity after longitudinal sinking.


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

The procedure was as follows: sinking of electrode into low depth in the material and then another shallow hole very close to the previous one. In such a way that all round holes might connect with each other to create cavity. Parameters used in this experiment were U=100 V, R=22 kW, C=47 pF. The whole procedure took about 17 minutes. 4.3. Polarity of the carbon fibre electrodes The experiments with reversed polarity were conducted to check whether the use of carbon fibres should have a significant influence on the efficiency. The tests have shown that reverse polarity (tool electrode as a cathode) - just like with the use of other electrode materials - has no beneficial influence on the process. The consumption of an electrode material proved to be higher and the efficiency of machining - lower in case of application of the reversed polarity. Parameters used in these experiments are presented in Table 3. Table 3. Parameters used in tests. No 1 2 3 4

U [V] 100 120 120 100

R[kW] 22 22 22 22

C[pF] 47 47 47 47

Polarity reverse reverse normal normal

Depth [μm] 1.2 2 2.8 1.7

4.4. Influence of the discharge energy on the process efficiency Further experiments were conducted to find interdependence between discharge energy and process efficiency. At first theoretical discharge energy was calculated. It was assumed that during discharge the energy constitutes about 0.7 of its calculated value. The theoretical value of energy was calculated on the basis of voltage and capacitance given for each trial with the use of the following formula E=0.7

CU2 2

(2)

Over the test given parameters were distributed evenly. The resistance value of R=22 kW was constant.

N° 4

2009

mena occurring during electrical discharge machining the electrode could vibrate at the beginning of the process. The bottom of the cavity is also uneven and very rough, so the measured depth of about 1.5 μm is only approximate. The relationship between depth and applied energy is presented in Figure 8.

Fig. 8. Relationship between depth of cavity and applied energy. In contrast with other electrode materials for the higher discharge energy the lower values of depth were obtained. Nevertheless, it appears that in above discussed case a diameter of cavity as well as the time of effective machining should have also been taken into consideration. However, in the presented experiments only a total of time of sinking was measured with no regard of whether the effective machining took place, or not. Therefore, the obtained relationship can only be considered as an approximation and needs further detailed verification.

5. Conclusions Conducted experiments lead to following conclusions. The use of carbon fibres electrode is troublesome in preparation of short tool electrodes as well as because of low erosive wear resistance. Nonetheless, other properties seem to be adequate for EDM. Of particular advantage are such properties of carbon fibres as their small diameters and mass production. Preliminary experiments have shown that use of carbon fibres as tool electrodes could potentially find its application in μEDM. Conducted test showed that reverse polarity has no advantageous influence on the machining process. Investigations concerning longitudinal sinking reveal that a special procedure must be applied due to a whippiness of fibres. Owing to a very low erosive wear resistance it is essential to apply longer electrodes to the process. Possibility of using metal coatings or the system of continuous electrodes, which is now under investigation, seems to be very promising. This could prove to be the right solution to the problem of low erosive wear resistance and limited length of short electrodes. ACKNOWLEDGMENTS

Fig. 7. 3D imagining and Profile of obtained cavity during the experiment. Obtained cavities were identified with the use of a measurement system intended for examination of micro-geometry of surfaces (Veeco optical profilometer). Example of a cavity is shown in Figure 7. The edge of the cavity is very irregular. Due to a variety of sudden pheno-

The authors gratefully acknowledge Ms. Anna Pakuła M.Sc. from Warsaw University of Technology for the help during optical measurements of samples.

AUTHORS Leszek Kudla - Ph.D., Department for Precision and Electronic Products Technology, Warsaw University of Articles

155


Journal of Automation, Mobile Robotics & Intelligent Systems

Technology, Faculty of Mechatronics, Warsaw, 02-525, ul. A. Boboli 8, Poland. E-mail: kudla@mchtr.pw.edu.pl. Anna Trych* - PhD student, Warsaw University of Technology, Faculty of Mechatronics, Warsaw, 02-525, ul. Boboli 8, Poland. E-mail: a.trych@gmail.com. * Corresponding author

References [1]

[2]

[3]

[4]

156

Uhlman E., Röhner M., “Investigation on Micro Electrical Discharge Machining using Diamond-based Tool Electrodes”. Proc. of Euspen Conf., May 2008, Zürich, Switzerland, vol. II, pp. 478-482. Yamazaki M., Suzuki T., Mori N., Kunieda M., “EDM of micro-rods by self-drilled holes”. Journal of Materials Processing Technology, no. 149, 2004, pp. 134-138 Kudla L., “Discussion of Properties of Carbon Fibres as Tool Electrodes for Micro-Electric-Discharge Machining”. Proc. of Scientific School of Erosive Machining Techniques KBM PAN, vol. 14, 2008, pp. 107-114. (in Polish) Trych A., “Investigation into the process of micro electro discharge machining”, Master's Thesis, Faculty of Mechatronics, Warsaw University of Technology, June 2008, pp. 62-63. (in Polish)

Articles

VOLUME 3,

N° 4

2009


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N째 4

2009

QUINTET-GATED FEA TO FORM CROSSOVER BEAM Tomoya Tagami, Masafumi Takeda, Shun Horie, Tomoya Yoshida, Masayoshi Nagao, Yoichirou Neo, Toru Aoki, Hidenori Mimura, Seigo Kanemaru

Abstract: In this report, we have described about new field emission array (FEA) structure built-in micro-column. This newly proposed FEA has multi-gated structured and is designed to form a crossover. We had already demonstrated the quad- and quintet- gated FEA. A crossover formation by was successfully observed for the first time. by the quintetgated FEA Furthermore electron beam lithography was demonstrated.

the load of lens was reduced in results. Figure 2(a) shows the fabricated the quad-gate FEA in which einzel lens consisted of upper four electrodes. The focusing characteristics with this structure were simulated by SIMION 8.0 as shown in figure 2(b). It was cleared that the needed voltage at intermediate electrode of einzel lens was reduced over 60V at crossover point. (a)

1. Introduction We have developed FEA with focusing function. To overcome an electrical field reduction during focusing in conventional double-gated FEA, the volcano-structured double gated FEA was proposed. It was demonstrated that emission current could be kept constantly under focusing operation. But it was found that a potential barrier surly was built under strong focusing operation and any emitted electron could not achieve to the anode in the result [1]-[3]. In order to form fine electron beam crossover, multi-gated FEA was designed. In our previous report, the quad-gated FEA was demonstrated its superior performance [4]. But quad-gated FEA need high operation voltage which is nearly breakdown of an insulator layer. To improve the characteristics, quintet-gated FEA was designed. And in this report, we talk that a clear crossover formation was successfully observed. And electron beam lithography without any other electrical optics was demonstrated.

(b)

2. Quintet-gated FEA

Fig. 1. Schematic of multi-gated FEA. Quintet-gated FEA was designed to improve focusing property. Figure 1 shows the schematic of the improved multigated FEA. In order to improve the multi-gated structure, the thickness of insulator was increased to 1000 nm, comparing with quad-gated FEA. And intermediate electrode at einzel lens consists of two gates. With this structure, effective thickness of the lens was increased and

Fig. 3. The simulated results of focusing characteristics and applying voltages at intermediate electrode.

3. Experiment Figure.3 shows the fabrication processes of the quintet-gated FEA. We used etch-back method [1-4]. By using this technique, multi-gated structure was exactly selfalignment integrated without any other alignment. Articles

157


Journal of Automation, Mobile Robotics & Intelligent Systems

VOLUME 3,

N° 4

2009

Fig. 3. Fabrication processes of quintet-gated FEA. Figure 4 shows the scheme of measurement system. All experiments were carried out under an ultra vacuum below 10-7Pa. To estimate the electron beam spot sizes, phosphor screen was used as anode and located 1 mm apart from the quintet-gated FEA. The irradiated area on screen was measured by charge-coupled camera.

Fig. 5. The focusing characteristics and observed irradiated area by focused electron beam.

Fig. 6. A schematic of electron beam lithography setup and exposure condition. Fig. 4. A schematic of measurement chamber and examination condition. Figure 5 shows the focusing characteristics. When focusing voltage is changed from 100V to -10V, it was found that anode current kept constantly. It indicated that any electrical field reduction at tip and building a potential barrier were removed. And the insets show the sequence of formation crossover processes under focusing operation. As shown in Fig. 5, the bright spot gradually became small and it was minimized by applying 13V at intermediate gate. Under over focusing condition by increasing the strength of lens, the spot size started spread again. Judging from this spot size behavior, it was indicated that electron beam crossover was formed. Figure 6 shows electron beam lithography system with the quintet-gated FEA. ZEP-520 of the thickness of 50 nm was coated on Si substrate as resist. The Si substrate was set in front of 10 mm apart from the FEA. Si substrate could be moved by stepping motor with the speed of 3Îźm/sec. The acceleration voltage of 5 kV was applied to Si substrate.

158

Articles

Fig. 7. the SEM images of developed resist pattern. Dose and focusing conditions were specified.


Journal of Automation, Mobile Robotics & Intelligent Systems

VOLUME 3,

N째 4

2009

Figure 7 shows the SEM images of after electron beam exposure and developing resist on Si. As a result, under focusing operation, the width of developed areas was narrowed. The electron beam lithography was successfully demonstrated. But the patterned sized was relatively difficult from experiment in Fig. 5 and simulation result. Further considerations are needed.

4. CONCLUSION The quintet-gated FEA has been fabricated and shows the excellent characteristics. The reduction of electric field and building potential barrier under strong focusing condition were completely removed. And clear forming process of electron beam crossover could be observed for the first time with lower operation voltage, comparing with quad-gated FEA. ACKNOWLEDGMENTS The authors would like to thank S. Yamashita and K. Matsubara for their technical assistance. The work was supported, in part, by Shizuoka University 21st COE (Center of Excellence) Program and Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Technology and Science, Japan.

AUTHORS Tomoya Tagami*, Masafumi Takeda, Shun Horie, Yoichirou Neo, Toru Aoki, Hidenori Mimura, Seigo Kanemaru - Research Institute of Electronics, Shizuoka University, 3-5-1 Johoku, Nakaku, Hamamatsu, Shizuoka 432-8011, Japan. Tel/Fax: +81-53-478-1319. E-mail: tagami@nvrc.rie.shizuoka.ac.jp. Tomoya Yoshida, Masayoshi Nagao - National Institute of Advanced Industrial Science and Technology, Uezono, Tsukuba, Ibaraki, 305-8568, Japan. * Corresponding author

References [1] [2]

[3] [4]

Soda T., Nagao M., Kanemaru S., Neo Y., Aoki T., Mimura H., JJAP, vol. 47, no. 6, 2008, pp. 5252-5255. Neo Y., Takeda M., Soda T., Nagao M., Yoshida T., Kanemaru S., Sakai T., Hagiwara K., Saito N., Aoki T., Mimura H., J. Vac. Sci. Technol. B,vol. 27, issue 2, pp. 701-704. Neo Y., Soda T., Nagao M., Yoshida T., Aoki T., Mimura H., Appl. Phys. Express, no. 1, 2008, 053001. Nagao M., Yoshida T., Kanemaru S., Neo Y., Mimura H., Jpn. J. Appl. Phys., no. 48, 2009, 06FK02 (4 pages).

Articles

159


Journal of Automation, Mobile Robotics & Intelligent Systems

VOLUME 3,

N째 4

2009

MEASUREMENTS OF PLASMA DIFFUSION COEFFICIENT IN PILOT-PSI DEVICE USING KATSUMATA PROBE Marius Lucian Solomon, Ilarion Mihaila, Viorel Anita, Claudiu Costin, Gheorghe Popa, Hennie van der Meiden, Richard Al, Gerard van Rooij, Niek Lopes Cardozo, J체rgen Rapp, Codrina Ionita, Ronald St채rz, Roman Schrittwieser

Abstract: Cross-field transport of particles and energy is a major issue in magnetized fusion-relevant plasmas, due to its relevance for plasma confinement. In this paper Katsumata probe measurements performed in Pilot-PSI linear magnetized plasma device are presented. The plasma diffusion coefficient in perpendicular direction to the magnetic field was estimated from the measurements. Keywords: Katsumata probe, magnetized plasma, diffusion coefficient.

vacuum chamber. The maximum value of the axial magnetic field can be 1.6 T. At the end of the vacuum vessel, face to face with the plasma source there is installed a water-cooled solid target. Copper, carbon or tungsten is used as target materials. The main goal of this set-up is to study the interaction of high density magnetized plasmas with solid targets made from the same materials which will be used to realize the divertor in ITER tokamak. PilotPSI is also the prototype for the new plasma fusion experimental set-up Magnum-PSI. The plant shall obtain particle fluxes at target surface similar with particle fluxes at the divertor surface in ITER, using confining magnetic fields up to 3 T.

1. Introduction For studying plasma diffusion of magnetized plasmas, different experimental methods can be applied. One of these is to use the Katsumata probe [1]. Katsumata probe measurements performed in a linear magnetized plasma device called Pilot-PSI are presented in this contribution. This device was designed for investigations of plasmasurface interaction at ITER relevant parameters [2]. The device is fully operational at FOM-Institute for Plasma Physics Rijnhuizen, the Netherlands. Plasma diffusion coefficient in normal direction to the magnetic field was determined by Katsumata probe measurements using the model recently proposed by Brotankova et al. [3].

A Katsumata probe was used to determine the diffusion coefficient perpendicular to the magnetic filed direction, in Pilot-PSI. The Katsumata probe consists of a tungsten collector having the diameter of 1.6 mm that can be moved inside of a ceramic tube, which is bigger in diameter. A schematic drawing of a Katsumata probe is shown in Figure 2.

2. Experimental set-up The scheme of Pilot-PSI device [2] is shown in Figure 1. The cylindrical vacuum vessel made of stainless steel is 1 m long and 40 cm in diameter. The plasma source of the discharge is a cascaded arc, installed in the center of the frontal flange of the vacuum chamber. Fig. 2. Schematic drawing of a Katsumata probe.

Fig. 1. General scheme of Pilot PSI set-up. Five magnetic coils distributed along the vessel are used to create rather uniform magnetic field inside the 160

Articles

The depth position of the tungsten collector inside the ceramic tube was marked down with h having origin (h = 0) at the level of the ceramic tube. The probe had radial mobility and was inserted inside the vacuum chamber at 3.4 cm in front of the target. The fluctuations of the floating potential were measured at different radial positions of the probe and also for different depth of the collector inside the ceramic tube. The radial position had a range from 0 to 2 cm, referenced to the axis of the vacuum vessel. Like collector position inside the ceramic tube, the depth h had a range from 0 to 5 mm. The fluctuations spectrum show that, as the collector is retracted inside the ceramic tube, the signal amplitude corresponding to certain frequencies (in the range of hundreds of KHz) has an exponential decay.


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

3. Results and discussions Spectra of the floating potential fluctuations of the collector is shown in Figure 3 for a radial position of the Katsumata probe R = 0.4 cm and depths of the collector inside the ceramic tube as parameter (h between 0÷5 mm). The method used to process the data for a Katsumata probe is the one described by J. Brotankova et al. in the paper [3]. Fluctuations spectra obtained for each depth (h = 0÷5 mm) were smoothed using twenty adjacent points on each graph. After graphs smoothing for all h values, the amplitude A dependence versus h for certain frequencies was extracted.

Fig. 5. Dependence of 1/L2 on the frequency.

Fig. 3. Amplitude spectrum of the fluctuations on the collector for different h position. Then, the graph of ln(A) versus h values is represented, for each different constant frequency in the range from 30 KHz to 120 KHz (Figure 4). From this drawing one can notice that for values of h smaller than 1.5 mm the amplitude logarithm decreases linearly and this means that the real signal registered by the Katsumata probe collector diminishes exponential. The relation between diffusion length L inside the ceramic tube [3] and the corresponding frequency is: 1/L2 = p f/D

In Figure 5 it can be observed that the dependence of 1/L2 on the frequency f is almost linear and, according to the above mentioned relation, the graph's slope is equal witch p/D. Thus, the transversal diffusion coefficient can be determined. For the measurements achieved in the experimental conditions of pressure p = 7.4 Pa, magnetic field B = 0.4 T, current through the plasma source IS = 90 A, gas flow QH2 = 3 slm (standard liters per minutes), the obtained transversal diffusion coefficient varied between 0.23 m2/s in the center of the plasma column (at the radial position R = 0 cm) and 0.74 m2/s at the edge of the plasma column (at the radial position R = 1 cm). This coefficient slowly increases from the center of the plasma column towards the edge. These values are in good agreement with the Bohm diffusion [4]. These values are in the same range as Bohm diffusion coefficient. For example, for an electron temperature of 1.7 eV measured on the axis of Pilot-PSI in a magnetic field of 0.4 T [2], the transversal Bohm diffusion coefficient is 0.27 m2/s

4. Conclusions Katsumata probe measurements can be used to estimate plasma diffusion coefficient in Pilot-PSI linear magnetized plasma device. The obtained values for transversal diffusion coefficient were in the range 0.23 to 0.74 m2/s. These values are in good agreement with the Bohm diffusion and also there are similar with the one obtained nowadays in tokamak devices [5]. ACKNOWLEDGMENTS This work, supported by the European Communities under the contract 1EU-3/11.08.2008 of Association EURATOM-MEdC, was carried out within the framework of the European Fusion Development Agreement. The views and opinions expressed herein do not necessarily reflect those of the European Commission.

Fig. 4. Decay of ln A inside the ceramic tube for constant frequency of the fluctuations. Here D is the diffusion coefficient of the plasma inside the ceramic tube, in perpendicular direction to the magnetic field lines.

AUTHORS Marius Lucian Solomon*, Ilarion Mihaila, Viorel Anita, Claudiu Costin, Gheorghe Popa - Faculty of Physics, Al. I. Cuza University, Association EURATOM-MEdC,11 Carol I Blvd., 700506-Iasi, Romania. E-mail: marius_solomon@plasma.uaic.ro. Hennie van der Meiden, Richard Al, Gerard van Rooij, Niek Lopes-Cardozo, Jurgen Rapp - FOM-Institute for Articles

161


Journal of Automation, Mobile Robotics & Intelligent Systems

Plasma Physics Rijnhuizen, Association EURATOM-FOM, Trilateral Euregio Cluster, P.O. Box 1207, 3430 BE Nieuwegein, The Netherlands. Codrina Ionita, Ronald Stärz, Roman Schrittwieser Institute for Ion Physics and Applied Physics, University of Innsbruck, Association EURATOM-ÖAW, Technikerstr. 25, A-6020 Innsbruck, Austria. * Corresponding author

References [1] [2]

[3]

[4] [5]

162

Katsumata I., Okazaki M., Japan J. Appl. Phys., vol. 6, 1967, p. 123. van Rooij G.J., Veremiyenko V.P., Goedheer W.J., de Groot B., Kleyn A.W., Smeets P.H.M., Versloot T.W., Whyte D.G., Engeln R., Schram D.C., Lopes Cardozo N.J., Appl. Phys. Lett., vol. 90, 2007, p. 121501. Brotankova J., Adamek J., Stockel J., Martines E., Popa G., Costin C., Schrittwieser R., Ionita C., van Oost G., van de Peppel L., Czech. J. Phys., vol. 56, no. 12, 2006, pp. 1321-1327. Chen F.F., Introduction to plasma Physics and controlled fusion, Plenum Press: NewYork, 1984. Wesson J., Tokamaks, Clarendon Press: Oxford, 1987.

Articles

VOLUME 3,

N° 4

2009


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

GROWTH AND APPLICATION OF ULTRA-LONG MULTI-WALLED CARBON NANOTUBE Sayaka Shimizu, Morihiro Okada,Yoku Inoue, Youitirou Neo, Hiroshi Kume, Toru Aoki, Hidenori Mimura

Abstract: We found the easy and efficient synthesis method of the vertically aligned ultralong multi-walled nanotubes using iron chloride powder. The 2.1-mm-long bulk nanotubes can be grown by conventional thermal chemical vapor deposition on bare quartz surface with the single gas flow of acetylene for 20 min. In addition to the high growth rate, the bulk of carbon nanotubes is easily spun into the yarn by pulling it out, and the present method also provides the coating ability with nanotubes as a new functionality of this nanomaterial. © 2008 American Institute of Physics. Keywords: CNT, CVD, yarn, FeCl2.

1. Introduction Since their discovery in 1991 [1], carbon nanotubes (CNTs) have been regarded as a key nanomaterial for a wide range of applications [2]. Their small tubular structure is responsible for diverse features, such as high mechanical strength [3], strong electric properties [4], good heat conductance [5], and high electron emission [6], which are of interest for academic researchers and industries [7]. However, despite several efforts [8]-[10], a suitable mass-synthesis method for CNTs that promotes widespread practical applications is not yet to be established. Here, we present a simple and easy way for mass synthesis that yields ultra-long, vertically aligned multi-walled CNTs (MWCNTs).This method requires no additional process for catalyst preparation predeposition and only requires iron chloride powder and acetylene gas used. We found that high dehydrogenation activity of iron chloride on acetylene increases the growth rate of CNTs compared to conventional predeposited metal catalysts. The lengths of obtained MWCNTs ranged up to the millimeter scale, and they can easily be spun into yarn by hand with the naked eyes. In addition, this method can be used to coat the entire surface of a target with MWCNT using iron chloride in vapor phase as a growth initiator. The most attractive feature of the proposed method is that it is extremely simple, and therefore, could be used in any laboratory for bulk production of MWCNTs. Vertically aligned CNTs (VACNTs) were synthesized using a conventional thermal chemical vapor deposition (CVD) system. A smooth quartz substrate was placed at a center of horizontal quartz tube furnace (40 mm in diameter and 30 cm in length) with iron chloride (FeCl2) powder (99.9%, Kojundokagaku Laboratory) using a quartz boat. As a pretreatment, the quartz substrate was cleaned using ethanol. Typically, a thin metallic film deposited

on a substrate is widely used as a catalyst; however, in the proposed method, such a film need not be predeposited. During heating, the sample was maintained at vacuum of 5×10-2 Torr, and once the optimal growth temperature was reached, it was purged with acetylene (98%, Japan Air Gases) gas using a mass flow controller. CVD growths were carried out at a furnace temperature of 820 °C at 5 Torr. Hereafter, we refer to the proposed method as “chloride mediated CVD” (CM-CVD). Features such as using a pure gas flow and not using a metallic film on the substrate are peculiar to CM-CVD and could lead to significant reduction in the mass-production costs of CNTs.

2. Result and discussion A VACNT sample is shown in Fig. 1(a). Densely grown CNTs are vertically aligned on a quartz substrate(10×10 mm2). The top (grown) surface is macroscopically flat, i.e. the CNT length is uniform throughout the substrate (Fig. 1(a)).] Figure 1(b) compares the samples. From Figs. 1(a) and 1(b), one can observe the vertically aligned growth throughout the surface, including the side and back surfaces.

(a)

(b)

Fig. 1. (a) VACNT samplel (b) SiO2 substrate. In CM-CVD, the stability of the VACNTs grown on a substrate depends on the growth conditions, such as the growth time and pressure. It is found that the 1.5 mm-long CNTs adhere strongly to the substrate, while the VACNTs grown under 4 Torr detached easily, leaving a very clean surface.

(a)

(b)

Fig. 2. (a) Pinching and pulling out 50cm CNTl, (b) extended CNT.

Articles

163


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

For this type of VACNTs, one can easily spin a CNT yarn using tweezers or an adhesive tape [11]. As seen in Fig. 2(a), we obtained a 50 cm CNT yarn by pinching and pulling out CNTs [12]. Figure 2(b) shows that during the pulling, a tensioned yarn extended up to 1 m 50 cm in length. In this study, we propose the CNT growth model in CM-CVD. To understand the reaction process, the transition of the partial pressure of released gases was measured using a quadrupole mass analyzer (MKS instruments, PPT-C200-F2T), as shown in Fig. 3. In the mass analysis, a part of the released gas was sampled, and the operating pressure was reduced to 1×10-6 Torr. After the beginning of the growth process with the acetylene (C2H2, 26 amu) feed, HCl (36 amu) was generated for a very short time as a trigger signal. After some time, hydrogen (2 amu) increased for 15 min, and then decreased gradually. We believe that the generation of the two gases (HCl and H2) relate to “bud beak” and “CNT growth”, respectively. HCl generation can be expressed as FeCl2 + C2H2 ® FeC2 + 2HCl

(1)

Since FeCl2 is completely vaporized at 820 °C [13], therefore, FeC2 molecules or related carbon-rich iron carbide species are considered to nucleate into nanoparticles as a result of multiple collisions, and to be deposited on the surfaces in the heated zone. Then, the FeC2 nanoparticles segregate into graphene layers as follows [14]. FeC2 ® FeC2-x + xC

(3)

Then, acetylene is highly decomposed at the root of CNTs due to the high dehydrogenation activity of iron 164

Articles

2009

chloride [16]. Reactions of acetylene at the iron chloride surfaces are expressed as follows: FeCl2 + C2H2 ® FeC2 + 2HCl

(4)

In the reaction, acetylene is decomposed, and FeC2 phase is again considered to form. Since FeC2 highly exceeds a solubility limit of carbon in iron, carbon is expected to rapidly segregate as follows: FeC2 ® FeC2-x + xC

(5)

By this process, the CNTs grow because of this carbon segregation. Here, Eq. (5) is the same with Eq. (2). This reaction cycle is expected to repeat rapidly, which means the ultra-long VACNTs are grown in a short amount of time. This reaction cycle, where iron chloride acts as catalyst, is one of the most important discoveries of this study. For the CNT coating, a vapor phase reaction of iron chloride with acetylene is essential. By observing the heating process through a viewing port, we found that FeCl2 started to vaporize at 550 °C, and completely vaporized over 800 °C, when the local vapor pressure reached 70 Torr. During this heating process, FeCl2 vapor spreads across the entire heated space. Since acetylene flow causes the sublimation of FeC2 nanoparticles on the entire surfaces, CNTs can be grown everywhere, even on micronspaced quartz wool. This growth initiation via vapor phase FeCl2 is another important finding of this study.

(2)

We believe that CNT growth in CM-CVD is initiated during the segregation of the graphene layers from carbonrich iron carbide (FeC2) as predicted by Jourdain et al. [15]. Once budlike CNT structures are formed, the growth is triggered. The subsequent hydrogen generation reflects the carbon supply for the growth of CNTs because hydrogen generation is the result of only the decomposition of acetylene. There are two possible reactions that result in the decomposition into hydrogen. One is a Fe-catalyzed dehydrogenation reaction, and the other is the reaction mediated by iron chloride. However, from our conventional CVD experiments, we found that small amounts of hydrogen are generated during the growth on a Fe predeposited quartz substrate, and the hydrogen generation is terminated within 3 min. In addition, the length of the CNTs grown using this simple thermal CVD method was less than 300 μm. Therefore, these results indicate that the observed high-speed growth is mediated by iron chloride. At first, iron chloride surfaces may appear on the catalyst nanoparticles at the root of CNTs due to the adsorption of the remaining FeCl2 and/or chlorination of surface. Fe atoms with the remaining HCl, according to the following well-known reaction: Fe + 2HCl ® FeCl2 + H2

N° 4

Reaction

Fig. 3. Sampled partial pressure of released gases as a function of growth time. Numbers denote the mass unit, where 2, 26 are left axis and 36 is right axis.

3. Conclusion In summary, we established a simple one-step growth method of aligned bulk CNTs mediated by iron chloride. The CM-CVD offers a potentially viable VACNT mass production method. In the present system, we can grow 1 g of VACNT for 2 h process time including 15 min of growth time, and material cost performance is as low as US $ 0.5/g. If a large number of flat plates are arranged in a growth chamber, then a significant amount of VACNTs can be easily obtained from the entire surface by highspeed coating growth. Although our VACNTs length has not exceeded the world record [17] (18 mm: reported by


Journal of Automation, Mobile Robotics & Intelligent Systems

VOLUME 3,

N° 4

2009

University of Cincinnati researchers), the rate of “production per process time and cost [g/h-$]” might be significantly high. We believe that this very simple one-step growth technology, based on bifunctionality of FeCl2, will contribute to the development of new CNT applications. ACKNOWLEDGMENTS This work was supported by the AGH University of Science and Technology under Grant No. 11.11.120.612.

AUTHORS Sayaka Shimizu*, Morihiro Okada, Yoku Inoue, Youitirou Neo, Hiroshi Kume, Toru Aoki, Hidenori Mimura - Research Institute of Electronics, Shizuoka University, 3-5-1 Johoku Hamamatsu 432-8011 Japan, Tel: +81-53-478-1356, Fax: +81-53-478-1356. E-mail: sayashimizu@nvrc.rie.shizuoka.ac.jp * Corresponding author

References [1] [2] [3] [4]

[5] [6] [7] [8]

[9] [10] [11] [12]

[13] [14] [15] [16] [17]

Iijima S., Nature (London), 1991, no. 354, p. 56. Baughman R.H., Zakhidov A.A., de Heer W., Science (London), 2002, no. 381, p. 678. Treacy M.M.J., Ebbesen T.W., Gibson J.M., Nature (London), 1996, no. 381, p. 678. Ebbesen T.W., Lezec H.J., Hiura H., Mennett J.W., Ghaemi H.F., Thio T., Nature (London), 1996, no. 382, p. 54. Kim P., Shi L., Majumdar A., McEuen P.L., Phys. Rev. Lett., 2001, vol. 87, p. 215502. de Heer W.A., Chatelain A., Ugarte D., Science, 1995, vol. 270, p. 1179. Saito Y., Uemura S., Carbon, 2000, vol. 38, p. 169. Christen H.M., Puretzky A.A., Cui H., Belay K., Fleming P.H., Geohegan D.B., Lowndes D.H., Nano Lett., 2004, vol. 4, p. 1939. Eres G., Puretzky A.A., Geohegan D.B., Cui H., Appl. Phys. Lett., 2004, vol. 84, p. 1759. Hata K., Futaba D.N., Mizuno K., Namai T., Yumura M., Science, 2004, vol. 306, p. 1362. Jiang K., Li Q., Fan S., Nature London, 2002, vol. 419, p. 801. See EPAPS Document No. E-APPLAB-92-084821 for CNT yarn beingspun by pinching and pulling out from the VACANT sample using tweezers. CRC Handbook of Chemistry and Physics, 84th ed., edited by Lide D.R. (CRC, Boca Raton, 2003). Kosugi K., Bushiri M.J., Nishi N., Appl. Phys. Lett., 2004, vol. 84, p. 1753. Jourdain V., Kanzow H., Castignolles M., Loiseau A., Bernier P., Chem. Phys. Lett., 2002, vol. 364, p. 27. Laude T., Kuwahara H., Sato K., Chem. Phys. Lett., 2007, vol. 434, p. 78. 18-mm-long VACNTs was issued by press release from National Science Foundation. (http://www.nsf.gov/ news /news_summ.jsp?cntn_id=108992&org=NSF)

Articles

165


Journal of Automation, Mobile Robotics & Intelligent Systems

VOLUME 3,

N째 4

2009

INFLUENCE OF POWERFUL LASER RADIATION ON FORMATION OF PORES IN SI BY ELECTROCHEMICAL ETCHING Artur Medvid, Pavels Onufrijevs, Leonid Fedorenko, Mykola Yusupov, Edvins Dauksta

Abstract: The influence of strongly absorbing N2 laser radiation on pores formation on a surface of Si single crystal has been investigated using optical microscope and atomic force microscope. After irradiation by the laser and subsequent electrochemical etching in HF acid solution morphological changes of the irradiated parts of a surface of Si were observed. At the same time, pores formation on the nonirradiated parts of Si surface took place. The porous part of the Si surface is characterized by strong photoluminescence in red part of spectra with maximum at 1.88 eV. Suppression of the pores formation by laser radiation is explained with inversion of Si type condition from p to n. This fact is explained by Thermogradient effect - generation and redistribution of the intrinsic defects in gradient of temperature. It was shown that the depth of p-Si layer on n-Si substrate depends on intensity of laser radiation and it increases with intensity of laser radiation. The results of the investigation can be used for optical recording and storage of information on surface of semiconductors. Keywords: porous Si, laser, optical storage, chemical etching.

1. Introduction The discovery of clearly visible photoluminescence (PL) at room temperature from porous silicon (por-Si) in 1990 [1] has opened the way to worldwide intensive studies on its optical and transport properties and to the numerous technological applications of por-Si in microelectronics [2] optoelectronics [3] and biology [4]. There are many methods for fabrication of por-Si. The main method of fabrication of por-Si is electrochemical etching of a Si crystal in HF solution with water or ethanol [5]. In this process, on the Si surface deep channels of pores form, with depth of several micrometers and some nanometers in diameter [6]. Sometimes laser radiation (LR) is used for controlling light emission spectrum or/and for stabilizing pores' properties. It is known that in order to form the por-Si layers the sample can be 1) irradiated by Nd:YAG laser fundamental frequency before stain etching [7]; 2) irradiated by the laser after formation of pores [8]; or 3) chemically etched in situ [9]. It is known that formation of pores on n-type Si is difficult due to deficit of holes. Formation of p-n junction by laser beam on a surface of p-Si(B) [10]-[12], p-Ge [13], p-InSb [14]-[18], InAs [19] and p-CdTe(Cl) [20] crystals have been shown. In our previous paper [11] we have shown the possibility to transform p-type Si in to n-type Si by LR which makes it possible to control the rate of chemical reaction by LR. 166

Articles

It was shown that the use of laser for inversion of Si crystal from p-type to n-type makes possible suppression of pore formation by laser beam. Using p-type to n-type laser inversion in Si crystal the possibility of pore formation suppression by laser beam was shown. This phenomenon can be used for optical writing. PL intensity of porous areas depends on the current density and the maximum of spectra, which is located in the range of 1.5-2.5 eV. Si is not reacting with hydrofluoric acid (HF) if electric current is not applied. The holes play main role in the pore formation in electrochemical reaction. Usually Si surface is passivated with H atoms and acid ions have no influence on silicon structure. Reaction between Si surface and acid solution starts after electric current is applied and Si-H bonds are broken. Once H atoms are detached from Si atoms, acid ions (F-) start to bind with the unprotected Si atoms and form SiF26 . In our previous paper [11] we have shown the possibility to transform p-type Si into n-type Si by LR, due to large temperature gradient. Temperature gradient effect (generation and redistribution of the intrinsic defects (interstitials and vacancies) in gradient of temperature [12] induced the drift of the vacancies to the irradiated Si surface. Therefore, it is possible to control the speed of chemical reaction, the distribution of pores and their size by LR. The aim of this study is to show the influence of powerful laser radiation on the formation of pores on a surface of p-Si exposed to electrochemical etching method.

2. Experiments and discussions The experiments were carried out on (100) p-type Si(B) commercial wafer. At the first stage the samples were irradiated by pulsed N2 laser (l = 337 nm, t=5 ns) at laser intensities I1(area 1) > I2(area 2) > I3(area 3) > I4=70MW/cm2 (area 4) as shown in Fig.1. No morphological changes have been detected by optical microscope and atomic force microscope (AFM) studies after irradiation of the samples. At the second stage, the samples were electrochemically etched for 10 minutes at the current strength 50 mA/cm2 in HF solution (48%) with ethanol in proportion 1:2. Electrochemical etching was carried out in ambient daylight conditions, but it's not enough to affect n-type silicon which was formed on the surface of the irradiated Si sample. The ntype Si is not reacting with HF acid ions unless holes are generated on the surface of the sample, but ambient light is not enough to generate reasonable hole concentration.


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

PL has been not observed on the irradiated and etched sample areas. PL spectrum of the non-irradiated surface was shifted to the red part of spectra with maximum at l = 656 nm (1.88 eV) and PL was observed by a naked eye at room temperature under UV light as shown in Fig. 4. PL intensity is increasing along with current strength from 20 mA/cm2 to 50 mA/cm2, due to pores density on the sample surface. AFM study has shown the formation of pores on the etched surface except irradiated areas where n-type Si formed as shown in Fig.2. After irradiation and etching, the area 4 peals to pieces with peace thickness of 1 μm and porous Si is found below the n-Si layer. Further experiments show that the next area 3 start to peel off after subsequent electro chemical etching of the sample used in the earlier experiments. The thickness of the intact Si layer is up to 1.5 μm. These results show that n-type silicon is formed on surface layers in depth less than 1.5 μm. It means that n-type Si thickness depends on laser radiation intensity.

N° 4

2009

Fig. 3. PL spectrum of the irradiated (2) and non irradiated (1, 3) areas.

3. Conclusion The possibility to control the electrochemical activity by laser beam has been shown. The depth of formation of p-n junction at the surface of p-Si crystal can be controlled by intensity of laser radiation. One more proof of ptype Si transformation into n-type Si by the Nd:YAG laser beam has been provided. Possibility of optical information recording and storage on p-Si surface by the Nd:YAG laser beam has been shown.

AUTHORS Artur Medvid' - Riga Technical University, 14 Azenes Str., Riga, LV-1048, Latvia; Institute of Semiconductor Physics National Academy of Science of Ukraine, 45 Pr. Nauki, 252650, Kyiv-28, Ukraine. Pavels Onufrijevs, Edvins Dauksta* - Riga Technical University, 14 Azenes Str., Riga, LV-1048, Latvia. E-mail: edvins.dauksta@rtu.lv. Leonid Fedorenko, Mykola Yusupov - Institute of Semiconductor Physics, 41 pr. Nauki, Kyiv, Ukraine. * Corresponding author Fig. 1. Irradiated areas with laser intensities at I1(area 1) > I2(area 2) > I3(area 3) > I4=70MW/cm2 (area 4).

Fig. 2. 3D AFM image (a) por-Si formed by N2 laser at in-tensity I = 20.0 MW/cm2 and subsequent electrochemical etching in HF acid solution, (b) the surface of por-Si for-med by electrochemical etching in HF acid solution. Articles

167


Journal of Automation, Mobile Robotics & Intelligent Systems

References [1]

[2]

[3]

[4]

[5] [6]

[7] [8]

[9]

[10]

[11] [12] [13] [14] [15] [16] [17]

[18] [19]

[20]

168

Canham L.T., „Photoluminescence in porous silicon obtained by hydrothermal etching”. Appl. Phys. Letl., vol. 57 no. 1, 1990, p. 1046. Chen Li, Huailin Liao, Li Yang, Ru Huang, „Highperformance integrated inductor and effective crosstalk isolation using post-CMOS selective grown porous silicon (SGPS) technique for RFIC applications”. SolidState Electronics, vol. 51, 2007, p. 989. Adamo R., Anopchenko A., Bettotti P., Cazzanelli M., D'Amato E., Daldosso N., Ferraioli L., Froner E., Gaburro Z., Guider R., Hossain S.M., Navarro-Urrios D., Pitanti A., Prezioso S., Scarpa M., Spano R., Wang M., Pavesi L., „Low dimensional silicon structures for photonic and sensor applications”. Applied Surface Science, vol. 255, no. 3, 2008, p. 624. Memisevic J., Korampally V., Gangopadhyay S., Grant S.A., „Characterization of a novel ultra-low refractive index material for biosensor application”. Sensors and Actuators B. Chemical, 2009. (in press) Halimaoi A. In: Properties of porous Silicon, edited Leigh Canham, INSPEC: London, 1997. H. Nishitani, H. Nakota, „Relaxation spectra of photoluminescence from porous silicon obtained by chemical etching of laser-modified silicon”. Appl. Phys., vol. 31, 1992, p. L1577. Allongue P. In: Properties of porous Silicon, INSPFC: London , 1997. Fedorenko L.L., Sardarly A.D., Kaganovich E.B., Svechnikov S.V., Dikiy S.P., Baranets S.V. Fiz. Tekh. Poluprovodn., vol. 31, 1997. Mada Y., Ione N., „Thermogradient mechanism of pn junction formation by laser radiation in semiconductors”. Appl. Phys. Lett., vol. 48, 1986, p. 1205. Jung H., Kwong D.L., „Visible electroluminescence from stain-etched porous Si diodes”. Appl. Phys. Lett., vol. 62, 1992, p. 12. Blums Y., Medvid' A. Phys. Stat. Sol. A, vol. l47, 1995, p. K91. Medvid' A., Onufrijevs P., Fedorenko L., Rimshans J., Dauksta E. Mater. Sc. & Appl. Chem., vol. 16, 2008, p. 70. Fujisawa I. Japan J. Appl. Phys., vol. 19, 1980, p. 2137. Fedorenko L.L., Bolgov S.S., Malyutenko V.K., Ukr. Phys. J., vol. 20, 1975, p. 2040. Bogatyrev V.A., Kachurin G.A. Sov. Phys. Semicond., vol. 11, 1977, p. 100. Medvid' A., Fedorenko L., Snitka V. Appl. Sur. Sc, vol. 14, 1999, p. 280. A. Vasiliev, Konov V., Korshunov A., Laser Processing and Diagnostics, edited Banerle D. (Springer-Verlay, Berlin, New York, Tokyo, 1984. Kurbatov L., Stojanova I. Reports of Acad. Sc. USSR, vol. 268, 1983, p. 594. Medvid' A., Litovchenko V.G.,Korbujak D.V., Fedorenko L.L., et al., Radiation Measurements, vol. 33, 2001, p. 725. Medvid' A. Defect and Diffusion Forum, vol. 210-212, 2002, p. 89.

Articles

VOLUME 3,

N° 4

2009


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

PROBE RADIUS CORRECTION METHODS – REVIEW AND COMPARISON OF EXISTING METHODS Adam Rak, Adam Woźniak

Abstract: Scanning technology has been becoming more common then ever. Scanning offers new and effective possibilities of measurement. Nowadays planning, production and assembling without high accurate metrology is impossible. Measurement of small, curved elements became much easier with great development of mechanical components of measure machines, such as: guideways, transducers, bearings, servomechanisms. All this improvement made possible to collect with good accuracy points in high density. Algorithms, and computer software were greatly improved as well. Especially, many efforts were put on probe radius correction algorithms development. In this paper a review and a comparison of probe radius correction methods are shown.

Many methods use a set of indicated measured points to estimate the correction direction. Such algorithm was described by Shuh-Ren Liang and Alan C. Lin in [3].

Keywords: coordinate measuring machine (CMM), scanning probe, probe tip radius correction. Fig. 1. Principle of scanning performance.

1. Introduction CMM scanning probes are used to locate and measure points, which will then form a three-dimension element. The measured points are being located by means of contact between a stylus tip and a surface of an element. Inductive transducers, located inside a probe, are registering a movement of stylus tip in X, Y, Z coordinates. As shown in Fig. 1, the data from transducers in addition with coordinates of scanning probe are actual position of stylus tip. A scanning probe, in fact, acts as a small coordinate measuring device. During scanning on CMM the computer is registering a center of probe tip, which was calibrated before the process. The coordinates of measured points are registered during probing. The measured data, called indicated points, is not giving real information about shape and dimension of an element. Only, so-called corrected measured point is an approximation of a real point on the surface, which is described in [1]. The stylus tip radius correction is an offset vector of norm equal to the effective stylus tip radius which is added to the indicated measured point. As many researches have shown, the biggest influence on measurement error in scanning process on CMM, has the correction of indicated points, particularly when a curved surface is being measured, which was described in [2]. There are many correction methods, which can be divided into three different groups: probe radius correction using information about position of indicated points, probe radius correction based on information from force generators in scanning probe, probe radius correction using information from CAD model.

It was proposed to convert massive data points into numerous, connected triangular meshes and then calculate unit normal vector to each triangular mesh. The process starts with determining base and target trajectory. The one with fewer points is called base trajectory. Next step is to find corresponding points between each trajectory.Once the corresponding points are found, a connection between them is implemented and triangular meshes are formed. Trajectories cannot cross each others. After that, the program calculates unit normal vector to each mesh. Finally, each point is shifted in its normal direction with the value of probe radius. Another simple method to calculate the correction direction is to connect successive points with a straight line and then calculate unit normal vector to determinate line. Alternative, it is possible to connect every second point of trajectory and calculate normal vector to the line. A big disadvantage of this method is that very often the corrected points are shifted in chaotic direction, which is not corresponding to measured surface, particularly curved or sharp shape, such as cutting edge. Much research was put on this problem, and as a result, a new correction method was proposed in [4]. Normal vector is not calculated in this algorithm, as well as the measured shape is not smoothed by any mathematical model, such as NURBS and the points are not manipulated to look more correct. It is also possible to perform validity check of corrected points. The principle of algorithm is that the profile is defined by an envelope of a family of circles formed by probe tips. The corrected measured points are Articles

169


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

situated somewhere on arcs of circles and are found by means of special mathematical equations and also fuzzy logic knowledge, which is described in [5]. Another method, proposed by Y. C. Lin and W. I. Sun in [6] is based on multi-cross-product method. The authors determine the directions of four normal vectors and their average for the compensated point from cross products of the four nearest tangential vectors. Tangential vectors are calculated from stylus tip centre and its four nearest points. Similar method, described by Jagoda and others in [7], determines eight normal vectors as cross product of stylus tip centre and eight nearest points. Normal vector of indicated point is the average of eight vectors calculated before. Another approach for stylus tip radius correction was proposed by Jae-jun Park and others, whose idea was totally different from previous [8]. The authors designed a new measuring probe with an internal elastic structure equipped with strain gauges, so that the contact force between the tip and measured shape can be estimated and used for the calculation of the stylus tip correction vector. Another example of new approach for stylus correction is an idea of Aoyama, Kawai, Kishinami, who proposed a potentiometric spherical tip, which is capable of detecting the point of contact with an electrically conducting object surface [9]. The touched point can be estimated by current flowing at electrodes distributed on a tip, which is covered with a thin resistance film.

2. Comparison of probe radius correction methods The measurements were carried out on Accura Zeiss CMM, equipped with Vast Gold active scanning probe and Calypso software. Each point was recorded every 0,01 mm with measuring tip which radius was 2,001296 mm. Scanning speed was 1 mm/sec. There were two radius correction methods built in the machine. One method was based on information of contact force from scanning probe and is called VAST correction. The information about contact force generated in scanning probe, so that the tip was always in contact with the work piece, was used to determine the correction direction. Another method was using NURBS splines and the calculations were made by Calypso program - this method is called Calypso correction. Trajectories made by measure tip were transformed into splines, which enabled to calculate normal vector of each spline. Moreover, two simple programs for calculating corrected points were created and used for the comparison. To correct an indicated point, the first program was using next or previous point's coordinates to calculate the correction vector. Correction direction was perpendicular to a line created by connecting indicated point with next or previous point. The second program, for calculating the correction direction is using every second point's information. Additionally, the correction was carried out on STEM software for the stylus tip radius compensation in CMM scanning measurement, described in [10], according to the algorithm from [5]. The measured elements had free form surfaces and constant sections. Indicated points were corrected by five methods. The results are shown in Fig. 2 - 4. The 170

Articles

N° 4

2009

indicated and corrected points are marked as follows: indicated points - black rectangle, VAST correction white rectangles, STEM correction white circles, Calypso correction - white triangles, every another and every second point white stars and pentagons. Measured profile, as well as corrected points was put on common chart, as shown in Fig. 2. Corrected profile is systematic, except for the edge, where the corrected points are irregular, which was shown in details in Fig. 3. VAST corrected points are put in wrong order, besides the scatter of corrected points is about 0,1 mm on the edge. When the profile is straight, VAST corrected points are given with smaller scatter. STEM, Calypso and correction based on previous or next or every second point position results are shown in Fig. 4. STEM and Calypso corrected points are close to each others, the distance is less then 1 μm. STEM corrected points are put in order, while Calypso, and correction based on previous or next or every second point position points are disordered. The scatter of Calypso and STEM corrected points is about 2 μm, while correction based on previous or next or every second point position has a scatter of about 4 μm.

3. Conclusion In this paper a review and comparison of probe radius correction methods is shown. Each group of correction method has some characteristic results. Methods based on mathematical calculations, such as STEM or Calypso gives smooth profile without loops. The programs use some special algorithms to eliminate problems with incorrect point order. VAST correction results are irregular and scattered. The reason of such effect is the friction between measuring tip and the surface. An additional force component is added to the contact force and it results in big, irregular scatter and rough profile. Big influence on this effect has the speed of scanning, as well as material of measured work piece. The biggest differences between correction method results are on the edges of profile. The difference between VAST correction and STEM, Calypso, or correction based on previous or next or every second point is about 0,2 mm. Correction based on previous or next or every second point occasionally gives results, which are chaotic. Corrected profile may have loops or the points may be in wrong order, as shown in Fig. 4.

fig. 3

Fig. 2. Measured profile with indicated and corrected points.


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

[4]

[5]

fig. 4

[6]

Fig. 3. Comparison of correction methods, an edge of measured element.

[7]

[8]

[9]

[10]

N° 4

2009

Woźniak A., Mayer R., Bałaziński M., “Stylus tip envelop method: corrected measured point determination in high definition coordinate metrology”. International Journal of Advanced Manufacturing Technology, no. 42, 2009, pp. 505-514. Woźniak A., Bałaziński M., Mayer R., “Application of fuzzy knowledge base for corrected measured point determination in coordinate metrology”. In: Proceedings of North American Fuzzy Information Processing Society NAFIPS 2007, San Diego, USA, 24th-27th June, 2007, pp. 135-139. Lin Y.C., Sun W.I., „Probe Radius Compensated by the Multi-Cross Product Method in Freeform Surface Measurement with Touch Trigger Probe CMM”, International Journal of Advanced Manufacturing Technology, no. 21, 2003, pp. 902 - 909. Jagoda J., Karbowski K., Komenda Z., „Ocena dokładności wybranych metod korekcji promieniowej końcówki pomiarowej” - Master's thesis, Cracov University of Technology, 2006, in Polish. Park J., Kwon K.,Cho N., “Development of a coordinate measuring machine (CMM) touch probe using a multiaxis force sensor”, Meas. Sci. Technol., vol. 17, pp. 2380- 2386 doi: 10.1088/0957-0233/17/9/002. Aoyama H., Kawai M., Kishinami T., “A new method for detecting the contact point between a touch probe and a surface”, CIRP Ann., vol.38, pp. 517-520. Łoś A., Woźniak A., “STEM - program do obliczania punktów pomiarowych skorygowanych w pomiarach skaningowych na WMP", Przegląd Elektrotechniczny, ISSN 0033-2097. vol. 84, no. 5, 2008, pp. 208-211. (In Polish)

Fig. 4. Comparison of correction methods. ACKNOWLEDGMENTS Research co-funded as a research project of the Ministry of Science and Higher Education of Poland. A. Woźniak is a beneficiary of the Homing program awarded by the Foundation for Polish Science.

AUTHORS Adam Rak*, Adam Woźniak - Division of Metrology and Quality Engineering, Institute of Metrology and Biomedical Engineering, Faculty of Mechatronics, Warsaw University of Technology, 02-525 Warsaw, ul. Św. A. Boboli 8, Poland. E-mail: adamrak2@wp.pl. * Corresponding author

References [1]

[2]

[3]

ISO 10360-6:2001 Geometrical Product Specifications (GPS) - Acceptance and reverification tests for coordinate measuring machines (CMM). Wozniak A., Mayer R., Bałazinski M., “Investigation into precise measurement of cutting tool edges using coordinate measuring machine”. In: CIRP 2nd International Conference on High Performance Cutting, Vancouver, Canada, 12th-13th June, 2006, Proceedings. Shuh-Ren Liang, Alan C. Lin: “Probe radius Compensation for 3D data points in reverse engineering”, Computers in Industry, no. 48, 2002, pp. 241 - 251. Articles

171


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

ONE DIMENSIONAL KINETIC MODEL OF CMM PASSIVE SCANNING PROBES Grzegorz Krajewski, Adam WoĹşniak

Abstract: Scanning probes, also known as measuring probes, are one of the most used solution for coordinate measuring machines (CMM), because of ability to collect large amount of measuring points as well as fast measurements. Fast measurement and accuracy of measuring system have recently determined two critical parameters which decide about effectiveness of the machine. Dynamic is the most significant factor limiting accuracy of measurement. Majority of research into scanning CMM does not separate the performance of the probing system from the other error sources of the machines, so a real error influenced by the probing system is not deeply determined. This paper presents the scanning probe model regardless the machine links. Determined model makes possible to better understand the probing process with higher speeds and optimize the measuring process. Finally (future work), optimization algorithms will be proposed and applied to the probing system and the result will be improved accuracy. Keywords: scanning probes, CMM, dynamics, model.

1. Introduction Modern manufacturing process, especially in aerospace and automotive industries, requires more sophisticated measuring devices. Verification process became inessential part in modern production, so the accurate measurement, high number of data collection and measuring velocity plays important role and cannot be achieved simultaneously at the satisfactory level. Dynamics became the dominant source of potential measurement error. Scanning probes are commonly used in CMM machines. The main asset of this solution is a great number of data to be collected which influences on the final accuracy of measurement. Another advantages of this inspection is possibility to measure in continuous way number of points when the machine is moving. One of the significant limitations of scanning speed determines the dynamic effects of the whole measuring system [1]-[3]. The Probe itself, machine, controller, calibration process, must be inter-related, as well as the measurement techniques have to be designed to ensure the better performance of the measurement. This article focuses mainly on the scanning probes and their influence on accuracy of the measurement. It is crucial to recognize kinetics properties of scanning probes. Well known method of research, based on shape artifacts [4]-[6], demands advanced and complicated equipment [7]-[16]. International standards also provide so172

Articles

me recommendations [17]-[19] concerning this type of measurements. Regardless of the scanning properties of measurement it is very important to create a theoretical model of scanning probe with mathematically calculated parameters of scanning. This model allows optimizing the mechanical properties of probes as well as measuring algorithms, which finally leads to improved accuracy with high dynamics of measurements. Scanning probes consist of three perpendicularly mounted guideways which are parallel to machine axes to perform the axis movement of stylus tip. Each guideway is incorporated with length measuring system with range up to a few millimeters and a typical resolution of 0.1 um. In fact, small range of scanning probes is compensated by machine movements for achieving continuous contact between stylus tip and material. There are two types of scanning probes on the market, assive and active ones. A passive probe is based on simple measurement of tip displacement, while machine is moving. Contact force is controlled and evaluated using three (one axes each) parallel electronic springs which give an electrical signal proportionally to the acting force. Active sensors represent a different approach to measurement. Active scanning sensors use a motorized mechanism that is able to control the stylus movement and modulate the contact force with the component.

2. Kinetic model of scanning passive probes Scanning probe itself can be easily modeled using the mechanical components. A model consist of mass which represent mass M of (stylus, stylus carrier, transducers, guide ways, springs). Probing system also can be described by stiffness K1 of probe mechanism and stiffness of stylus, and damping C1. Taking these factors into consideration we can easily evaluated simple probe model (Fig.1). a)

b)

Fig. 1. CMM passive scanning probe: a) picture [20], b) one dimensional mechanical model.


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

As the first part of simulation model one dimensional probe has been evaluated. It means that model cover only one degree of freedom probe which theoretically can measure with only one direction. It takes into consideration only probe mechanics regardless effect of stylus stiffness and stylus mass. Proposed model can be described using a mathematic formula as the second order differential equation: (1) where: - time, t x(t) - probe response,

quency, and

2009

4. This configuration allowed measuring the free vibration of scanning probes. The initial values of measurement where executed using external forces. Unknown parameters where evaluated on the basis of free vibration characteristics of probe mechanism using “additional mass methods” [21]. The evaluated parameters allow for determining the dynamic properties such as stiffness of probing system, damping and masses. These parameters state the dynamic properties of the probing system, which have influence on accuracy of measurement especially with increased scanning speed.

3. Experiments

- second derivative of x(t), - first derivative x(t),

N° 4

free undamped fredamped frequency.

The equation (1) has the following solution: (2)

This type of measurement evaluated for two types of scanning probes: Vast XXT manufactured by Zeiss, and SP600 manufactured by Renishaw. Vast XXT is described as passive probes with maximum permissible error of scanning MPETij = 2.7 μm, according ISO standards. SP600 is also typically passive probes with accuracy MPETij = 3.4 μm. Determined parameters of model can be easily adapted to a mathematical formula. The solution of equation 1 gives “real” probe response for different forcing functions. The result of VAST XXT and SP600 output performance simulations for forcing function “sin(pt)(1-Heaviside(t-1)) were presented in the Figures 3 and 4.

Initial value considerations, x(t)=P and x'(t)=0, parameters A i y can be expressed[21]:

(3)

(4)

Fig. 3. Simulation of forcing response for VAST XXT probe.

Fig. 2. Measurement of mechanical parameters of the scanning probes: 1 - laser head, 2 - machine table, 3 - stationary optics , 4 - retroleflector, 5 - scanning probe. For the experimental evaluation of unknown parameters A, M, K1, C1, y, w special measuring stand based on laser interferometer was prepared and described in Fig. 2. A small retroreflector with known mass 1 was attached to the probe carrier (styli were removed). Laser head 3 and stationary optics 2 were located on machine table

Fig. 4. Simulation of forcing response for SP600 probe. Articles

173


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

An excitation function applied to the proposed model, which represents the measuring surface (sinus), only gives symmetric errors which are proportional to the maximum of excitation function. Proposed model applied to the evaluated probes show that error is bigger than Maximum Permissible Error MPETij specified by manufacturers of probe. For VAST XXT MPETij specified in datasheet equal 2.7 μm but obtained by the model is closely 3.4 μm. Bigger difference between obtained results and MPETij represents SP600 probe. For SP600 maximum permissible error describing accuracy of probe according ISO 10360-4 standards equals 3.4 μm. Simulations show that error equals 12 um, which is over than expected.

4. Conclusions Proposed kinetic model and simulation represent only mechanical properties of probe transducers which can be valuable for probe designers and manufacturers. Deep knowledge about scanning probes principles is inessential, so the prospective model should provide information about wider group of measuring surface especially with the high speed curvature performed for high speed scanning. It is necessarily to develop model for future simulation of additional dynamic effect as stylus bending, friction between stylus tip and surface, micro- and macro geometry of measured surface. Improved model allows to compensate dynamic effects of scanning probes operation.

[7]

[8]

[9]

[10]

[11]

[12]

[13]

[14]

ACKNOWLEDGMENTS Research co-funded as a research project of the Ministry of Science and Higher Education of Poland. A. Woźniak is a beneficiary of the Homing program awarded by the Foundation for Polish Science.

AUTHORS Grzegorz Krajewski*, Adam Woźniak - Division of Metrology and Quality Engineering, Institute of Metrology and Biomedical Engineering, Faculty of Mechatronics, Warsaw University of Technology, 02-525 Warsaw, ul. Św. A. Boboli 8, Poland. E-mail: gkrajewski@gazeta.pl. * Corresponding author

References [1] [2]

[3]

[4]

[5]

[6]

174

Bosch J.A., Coordinate Measuring Machines and Systems, Marcel Dekker, Inc. New York, Hong Kong, 1995. Butler C., “An investigation into the performance of probes on coordinate measuring machines”, Industrial Metrology, vol. 2, no. 1, 1991 s. 59-70. Ratajczyk E., Współrzędnościowa technika pomiarowa, Oficyna Wydawnicza Politechniki Warszawskiej, Warszawa, 2005. (in Polish) Morse E.P., Artifact selection and its role in scanning probes evaluation, The UNC of Carolina at Charlotte, USA, 2002. Weekers W.G., Schellekens P.H.J., “Compensation for dynamic errors of coordinate measuring machines”, Measurement, vol. 20 ,1997, p. 197-209. Woźniak A., Mayer R., “Micro-shape master artifact for testing the dynamic performance of CMM scanning pro-

Articles

[15]

[16]

[17]

[18] [19]

[20] [21]

N° 4

2009

bes”. In: 8th International Scientific Conference Coordinate Measuring Technique, Bielsko-Biała, Poland, April 2008, pp. 107-112. Liang R., Jusko O., Lüdicke F., and Neugebauer M., “A novel piezo vibration platform for probe dynamic performance calibration”, Measurement Science and Technology, vol. 12, 2001, pp. 1-6. Bendeli A., Duruz J., Thwaite E.G., “A surface simulator for the precise calibration of surface roughness measuring equipment”, Metrologia, vol. 10, 1974, pp. 137-43. Woźniak A., „Metody badania dokładności sond skaningowych WMP”. In: VII Szkoła-Konferencja „Metrologia wspomagana komputerowo”, Waplewo, May 2005, Papers vol. 3, pp. 151-156. (in Polish) Woźniak A., “New method for testing the dynamic performance of CMM scanning probes”, IEEE Transactions on Instrumentation and Measurement, vol. 56, no. 6, 2007, pp. 2767-2774. Woźniak A., “Application of piezotranslator for the dynamic testing of scanning probes in coordinate measuring machines”, Mecatronica (Societea Romania de Mechatronica), no.1, 2004, pp. 93-95. Woźniak A., Dobosz M., “Methods of testing of static inaccuracy of the CMM scanning probe”, Metrology and Measurement Systems, vol. X, no. 2, 2003, pp. 191-203. Woźniak A., “An investigation into the dynamic performance of CMM scanning probes”, Elektronika, no. 8-9, 2004, pp. 256-258. Woźniak A., “Application of piezotranslator for the dynamic testing of scanning probes in CMM”. In: The 7th International Conference on Mechatronics and Precision th th Engineering, Bucharest, Romania, 27 -29 May 2004 , pp. 245-246. Woźniak A., “Preliminary dynamic error characteristics of CMM scanning probe”. In: 2nd Int. Conf. on Global Research and Education Inter-Academia, Warsaw, Poland, September 2003, vol. 1, pp. 201-204. Woźniak A., Dobosz M., Ratajczyk E., “Testing the inaccuracy of CMM scanning probe”. In: International Symposium on Metrology and Quality Control, Cairo, Egypt, 24th-27th Sept. 2001, pp. 57-63. PN-EN ISO 10360-4: 2002. Specyfikacja geometrii wyrobów (GPS). Badania odbiorcze i okresowe współrzędnościowych maszyn pomiarowych (CMM). Part 4: CMM stosowane w trybie pomiaru skaningowego. (in Polish) VDI/VDE 2617 1989. Accuracy of Coordinate Measuring Machines. Part 3. Dusseldorf. ANSI/ASME B89.1.12M 1990. Methods for Performance Evaluation of Coordinate Measuring Machines, American Society for Mechanical Engineering New York. Renishaw, General Catalogue 2009. Arczewski K., Pietrucha J., Szuster J.T., Drgania układów fizycznych, Oficyna wydawnicza Politechniki Warszawskiej: Warsaw, 2008. (in Polish)


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N째 4

2009

DESIGN OF MULTI-LAYER SPUTTER-DEPOSITED ANODE TO REDUCE CATALYST LOADING FOR LIQUID DMFC Hiroyuki Saito, Tsuneyoshi Nakashima, Katsuki Nakase, Masao Sudoh

Abstract: Superior catalyst utilization of direct methanol fuel cells (DMFCs) may be obtained by localized catalyst loading on reaction sites. The objective of this work is to improve the catalyst utilization by multi-layer structure and reduction of loading catalyst. Multi-layer anode consisted of sputter-deposited Pt-Ru catalyst layer and the support layer of Nafion-carbon-Isopropanol ink (NCI). Single layer anode consisted of sputter-deposited Pt-Ru catalyst layer and the layer of carbon-glycerin ink (CG). Multi-layer (1~4 layers) and single-layer (0.04, 0.10 and 0.24 mg cm-2) were evaluated by using electrochemical measurement and SEM images. Three-layer anode provided 50.9 W g-1, 3.4 times as mass activity of conventional paste method anode. Methanol residues stripping voltammetry revealed that electrochemical surface area (ECSA) was increased with the number of layers. Additionally, single-layer anode (0.04 mg cm-2) provided over 150 W g-1. These results suggested that reduction of loading catalyst per unit layer and multilayer structure enhanced catalyst utilization. Keywords: liquid DMFC, low catalyst loading, sputter-deposition, multi-layer anode.

1. Introduction Liquid Direct Methanol Fuel Cells (l -DMFCs) is prospective as power source for portable application because of high energy density and simple structure. However the catalyst is very expensive and it is necessary to reduce the amount of catalyst and to improve catalyst utilization. These may be realized in a sputter-deposited electrode with only active reaction sites [1]. In the previous works, sputter-deposited Pt cathode with Nafion-carbonbutyl acetate (NCB) layer showed high mass activity for oxygen reduction [2]-[4]. Mass activities could be improved by preparation of thin catalyst layer and an increase in ratio of catalyst contacted with a protonic conductor. Such a thin layer and low loaded catalyst layer could be prepared by sputtering method. In this study, multi-layer anodes and single-layer anodes with sputter-deposited Pt-Ru with NCI or CG ink were investigated electrochemically at anodes for l -DMFC to clarify effect on the mass activity and reaction sites [5].

2. Experimental The multi-layer anodes were prepared by sputtering Pt-Ru and spreading NCI ink, which consisted of 5 wt% Nafion solution (Wako Pure Chemical Industries), carbon powder (Vulcan XC72R) and Isopropanol (Wako Pure Che-

mical Industries, > 99 %). First, NCI ink was spread on the carbon paper TGPH-90 (Toray) and drying at 333 K. Second, Pt-Ru catalyst was loaded on carbon paper TGPH-90 by sputtering method (ULVAC, RFS-200). These processes were repeated 1 to 4 times and multi-layer anodes were prepared. The total amount of Pt-Ru loading were 0.24 mg cm-2 in all multi-layer anodes. The amount of loading catalyst in each layer of n layers anode; 0.24/n mg cm-2. For example, the amount of loading catalyst in each layer of 3 layers anode; 0.08 mg cm-2. The single layer anodes were prepared to investigate effects of the amount of catalyst loading by sputtering Pt-Ru and spreading CG ink, which consisted of carbon powder and glycerin (Wako Pure Chemical Industries). First, CG ink was spread on the carbon paper TGPH-90 and drying at 573 K. Second, Pt-Ru catalyst was loaded on carbon paper TGPH-90 by sputtering method. The amount of loading catalyst were 0.04, 0.10 and 0.24 mg cm-2. The paste method anode was prepared by spreading the mixture, which consisted of Pt-Ru/C (Tanaka Kikinzoku Kogyo) and 5 wt% Nafion solution dispersed by n-butyl acetate (Wako Pure Chemical Industries, 99 %), on the carbon paper TGPH-90 and drying at 333 K. This process was repeated to achieve the Pt-Ru loading of 2.0 mg cm-2. The paste method cathode was prepared by spreading the mixture, which consisted of Pt/C (Tanaka Kikinzoku Kogyo) and 5 wt% Nafion solution dispersed by n-butyl acetate, on the carbon cloth with support layer and drying at 333 K. This process was repeated to achieve the Pt loading of 2.0 mg cm-2. Nafion 117 (Du pont) used as an electrolyte membrane in DMFC was boiled in 3 wt% hydrogen peroxide (H2O2, Wako Pure Chemical Industries, > 30 %) solution for 1 h and rinsed in boiling deionized water for 1 h. Then it was boiled 0.5 mol L-1 sulfuric acid (H2SO4, Wako Pure Chemical Industries, > 95 %) for 1 h and rinsed in boiling deionized water for 1 h. The pretreated membrane, the anode and cathode were assembled by hot-pressing them under 398 K and 10 MPa for 2 min to ensure good contact among the cell components. The schematic diagrum of the experimental apparatus is shown in Figure 1. All electrochemical measurements were performed at 338 K. In cell polarization curves studies, 1.0 M methanol solution was supplied at 1.25 mL min-1 into the anode and dry air was supplied at 62.5 mL min-1 into the cathode. In anode impedance and stripping voltammetry studies, humidified hydrogen was supplied at 30 mL min-1 into the cathode that was used as reference electrode [6]. Geometrical area of electrodes were 6.25 cm2.

Articles

175


Journal of Automation, Mobile Robotics & Intelligent Systems

VOLUME 3,

N째 4

2009

Fig. 1. Schematic diagram of the experimental apparatus.

3. Results and discussion 3.1. SEM images The cross section of multi-layer anodes (1 layer and 3 layers) observed by SEM were given in Figure 2. Pt-Ru particles were located on white lines in Figure 2. A number of white lines were observed in cross section of 3 layers anode (Figure 2 (b)). It could be confirmed that Pt-Ru multilayer structure was formed by repeating process of spreading NCI ink and sputtering Pt-Ru. A sputtering method was well suited to form thin catalyst layer and to deposit catalyst locally. Multi-layer structure by using sputtering method enables to build thin catalyst layer and there was a possibility that ohmic resistance and mass transfer resistance were decreased.

efficiency. In comparison of mass activity with multi-layer and paste method anode, all multi-layer anodes had higher mass activity than paste method anode. Mass activity of 3 layers anode was the highest (50.9 W g-1) and it was approximately 3.4 times higher than that of paste method anode. These result may suggest that an increase in active Pt ratio in multi-layer anodes. The number of layers affected on mass activity in multi-layer anodes. Pt-Ru was only active as a catalyst when it was in contact with the electrolyte, conductive support and fuel. To increase the number of layer divided catalyst layer by Nafion as electrolyte and carbon as conductor. This may reduce inactive catalysts with an increase in amount of Pt-Ru contacted with Nafion and carbon. However, 4 layers anode had lower mass activity than 3 layers anode. There was a possibility that mass transfer was inhibited by permeating large amount of NCI ink into substance.

Fig. 3. Mass activity (V-W curves) for multi-layer anodes and paste method anode. Fig. 2. SEM images of multi-layer anodes. (a) mono-layer anode, (b) three-layer anode. 3.2. Electrochemical measurements Figure 3 shows mass activity of multi-layer anodes and paste method anode to evaluate the catalyst utilization 176

Articles

A comparison of the in situ stripping behavior of adsorbed methanol residues for multi-layer anodes was shown in Figure 4. The potential of the stripping peaks were almost equivalent in all multi-layer anodes. The Electrochemical surface area (ECSA) for anodic catalyst is calculated by using the following equation [7].


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

(1)

Fig. 4. In situ stripping voltammetry of methanol residues at multi-layer anodes. The ECSA data for the multi-layer anodes calculated from Eq. 1 are summarized in Table 1. ECSA increased as the number of layers increased. In spite of same catalyst loading, 4 layers anode had ECSA about two times 1 layer anode. As a result, division of catalyst layer was effective to increase ECSA and to reduce size of Pt-Ru particle for sputtering method.

N째 4

2009

Rm represents the resistance of membrane, and R1 and C1 are charge transfer resistance and double layer capacitance, respectively. R2 and CPE1 are related to mass transfer. L1 is related to the adsorbed inter mediate. The fitting data well agreed with experimental data. At high and medium frequency region, the arcs of 2, 3 and 4 layers anodes were smaller than that of 1 layer anode. In Table 2, R1 was decreased and C1 was increased as the number of layers increased. These results showed enhancement of three phase boundary by multi-layer structured. However, mass transfer resistance R2 of 4 layers anode was larger than other multi-layer anodes. It seems that permeating large amount of NCI ink into substance caused increase of R2 and cell performance decreased. Figure 7 shows mass activity of single-layer anodes to evaluate the catalyst utilization efficiency. Mass activity of single layer anodes was incresed with decreasing amount of loading catalyst. Mass activity of single-layer 3 (0.04 mg cm-2) was over 150 W g-1. ECSA data for single layer anodes are summarized in Table 3. In the same manner as mass activity, ECSA was increased with decreasing the amount of loading catalyst. These results suggested that size of Pt-Ru particle was decreased and active site was increased by reduction of the amount of loading catalyst.

Table 1. ECSA data for multi-layer anodes. 1 layer ECSA (m2g-1) 9.32

2 layers

3 layers

4 layers

15.84

18.85

20.14

Fig. 6. Equivalent circuit of anode polarization. Table 2. Impedance parameter of multi-layer anodes.

R1 (Wcm2) R2 (Wcm2) C1 (mF cm2)

1 layer

2 layers

3 layers

4 layers

2.977 0.5142 4.882

2.674 0.5193 7.906

2.488 0.4889 11.56

2.275 0.7835 16.87

Fig. 5. Nyquist plots for multi-layer anodes and paste method anode. Plots; experimental data, lines; fitting data. Loading current density; 40 mA cm-2. Effects of the number of layers on anodic polarization were further investigated by AC impedance analysis at 40 mA cm-2. Nyquist plots shown in Figure 5 were analyzed using an equivalent circuit given by Figure 6 [8] and impedance parameters were summarized in Table 2. Here,

Fig. 7. Mass activity (V-W curves) for single-layer anodes. Articles

177


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

Table 3. ECSA data for single-layer anodes. Single 1 layer Loading catalyst (mg cm-2) 0.24 ECSA (m2g-1) 17.32

[8]

Single 2 layers

Single 3 layers

0.10 34.35

0.04 79.85

4. Conclusion SEM cross section images clarified that thin multi-layer structure was formed by repeating process of spreading NCI ink and sputtering Pt-Ru. In 3 layers anode, mass activity by sputtering method was 3.4 times higher than that by paste method. ECSA increased with increasing number of layers. AC impedance analysis clarified that charge transfer resistance was reduced and double layer capacitance was increased by the increment of the number of layers. Reduction of the amount of loading catalyst per unit layer enhanced mass activity and ECSA.

AUTHORS Hiroyuki Saito, Tsuneyoshi Nakashima, Katsuki Nakase, Masao Sudoh* - Department of Materials Science and Chemical Engineering, Shizuoka University, 3-5-1 Johoku, Naka ward, Hamamatsu, 432-8561, Japan, E-mail: tcmsudo@ipc.shizuoka.ac.jp. * Corresponding author

References [1]

[2]

[3]

[4]

[5]

[6]

[7]

178

Cha S.Y., Lee W.M., “Performance of Proton Exchange Membrane Fuel Cell Electrodes Prepared by Direct Deposition of Ultrathin Platinum on the Membrane Surface”, J. Electrochem. Soc., vol. 146, no. 11, 1999, pp. 40554060. Makino K., Furukawa K., Okajima K., Sudoh M., “Optimization of sputter-deposited platinum cathode for direct methanol fuel cell”, Electrochemica Acta, vol. 51, no. 5, 2005, pp. 961-965. Makino K., Furukawa K., Okajima K., Sudoh M., “Performance of Sputter-deposited platinum cathode with Nafion and carbon loading for direct methanol fuel cells”, J. Power Sources, vol. 166, no. 1, 2007, pp. 30-34. Sudoh M., Nakase K., Tauchi M., Makino K., “Design of Thin-layered Membrane Electrode Assembly Prepared by Sputtering Method for Direct Methanol Fuel Cells”, ECS Transaction, vol. 11, no. 1, 2007, pp. 1397-1406. Haug A.T., White R.E., Weidner J.W., Huang W., Shi S., Stoner T., Rana N., “Increasing Proton Exchange Membrane Fuel Cell catalyst Effectiveness Through Sputter Deposition”, J. Electrochem. Soc., vol. 149, no. 3, 2002, pp. A280-A287. Arico A.S., Baglio V., Blasi A.D., Modica E., Antonucci P.L., “Analysis of the high-temperature methanol oxidation behavior at carbon-supported Pt-Ru catalysts”, J. Electroanal. Chem., vol. 557, 2003, pp. 167-176. Lai C.-M., Lin J.-C., Hsueh K.-L., Hwang C.-P., Tsay K.C., Tsai L.-D., Peng Y.-M., “On the accelerating Degradation of DMFC at High Anodic Potential”, J. Electrochem. Soc., vol. 155, no. 8, 2008, pp. B843-B851.

Articles

N° 4

2009

Hsu N.-Y., Yen C.-H., Jeng K.-T., Chien C.-C., “Impedance studies and modeling of direct methanol fuel cell anode with interface and porous structure perspectives”, J. Power Sources, vol. 161, no. 1, 2006, pp. 232-239.


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

NUMERICAL SIMULATION OF BLOOD FLOW AND CHOLESTEROL DISTRIBUTION FOR MIDDLE CEREBRAL ARTERY WITH COARCTATION Masashi Naito, Kensaku Mizoguchi, Youhei Takagi, Yasunori Okano

Abstract: In order to understand of coarctation growth mechanism in a blood tube, numerical analysis for blood flow and cholesterol distribution in a blood tube was carried out. Numerical results showed that back flow existed behind the coarctation, and it was found that high blood pressure (HBP), and nonelastic conditions increased the cholesterol concentration behind the coarctation. Keywords: numerical simulation, blood flow, coarctation, cholesterol distribution.

2. Model description and numerical procedure 2.1. Model description Figures 1 and 2 show an analysis area and schematic model, respectively. In this study, the middle cerebral artery (MCA) that had a coarctation with ratio of 30 % was simulated under different vessel conditions; (a) is NBP and elastic, (b) is NBP and nonelastic and (c) is HBP and elastic.

1. Introduction Because of the recent development of medical equipments, early detection of vessel disease in a brain and cure can be realized. However, judgment of disease condition strongly depends on doctorâ&#x20AC;&#x2122;s skill and experience, and an objective criterion for treatment is strongly demanded. Therefore, it is required to investigate causes and prevention of the disease from engineering view points such as fluid dymanics and mass transfer. However, it is difficult to model the property of vessel experimentally. Therefore, numerical simulatrion by using high performance computers must be a useful tool for it. A coarctation, which is inflamed vessel wall, is a vessel disease. This phenomenon blocks the blood flow, and the growth of coarctation causes other diseases. In general, it is well known that cholesterol distribution plays an important role in the growth of coarctation. Torii et al. [1] have reported the effect of blood pressure on aneurysm. Their model included the effect of vessel wall movement on normal blood pressure (NBP) and HBP. Kim and Ley [2] considered the cooling effect of blood over inflamed atherosclerotic plaque. However, the blood pressure effect and cholesterol distribution were not considered in their analysis. Sugawara et al. [3] investigated the relation between the intravascular pressure and the diameter of the carotid artery in six patients. These results showed the relationship between pressure and diameter was relatively linear throughout the cardiac cycle. Therefore, it was found that movement of vessel wall was expressed by a mathematical equation.

Fig. 1. Analysis area and coarctation point.

Fig. 2. Schematic model of MCA with coarctation. 2.2. Govering equations The governing equations for the flow in the blood tube are the following continuity (1), Navier-Stokes (2), (3) and diffusion (4) equations; ,

The final goal of this study is the development of simulation code for understanding of coarctation growth, which can be used for protection of blood tube desease. As the first step, the two-dimensional simulation code that can analyze the effect of blood flow and cholesterol distribution on a coarctation has been developed.

(1)

,

(2)

,

(3)

Articles

179


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

.

(4)

These equations were transformed into a dimensionless form as follows:

N° 4

2009

2.4. Movement of vessel wall For the movement of vessel wall, the following three steps were considered. Firstly, stress at the vessel wall was calculated by using the following equation: (14)

,

(5)

,

(6)

,

(7)

where DP is expressed as DP = Pblood - Pbase. Pblood is time dependent blood pressure value, which was determined by the inlet flow. Therefore, the blood pressure is expressed by using Fourier series equation (Eq. (10)). a0, an and bn are expreseed as follows: NBP (Normal blood pressure)

.

(8) ,

(15)

Dimensionless numeber are as follows: ,

,

,

, ,

,

,

,

.

(16)

(9)

They were discretized by the finite difference method, and solved by HSMAC method. Physical properties of blood were used for the calculation and Schmidt number was assumed to be 1000. Because vessel diameter was relatively large and flow rate was high, blood was assumed to be Newtonian fluid [1]. Mesh number and time step were set to be (x,y) = (200,50) and 1.0×10-6 second, respectively.

.

(17)

HBP (High blood pressure) ,

(18)

2.3. Boundary condition Velocity in the inflow boundary was given by Fourier series expansion by using the flow rate obtained by ultrasound Doppler [4]. Fourier series expansion of inflow boundary can be written as follows:

,

(19)

.

(20)

(10) where a0 is the modification coefficient; an and bn are the Fourier cosine coefficient and Fourier sine coefficient of the n-th term; t is the time. a0, an and bn are expreseed as ,

(11)

In this study, the physiological ranges for NBP and HBP are assumed to be 70-120 mmHg and 100-180 mmHg, respectively. Pbase was set to be 100 mmHg. This value shows radius of standard vessel. Figure 3 (b) shows the pressure waveforms for the HBP and NBP conditions. Secondary, strain can be obtained as folows: (21)

,

(12) E denotes a Young’s modulus which was assumed to be 105. Thirdly, deformation value of vessel wall is calculated by using radius of vessel and strain as follows:

.

180

(13)

(22)

Figure 3(a) shows inlet timedependent blood flow. In this case Reynolds numeber changes between 80 and 380.

Figure 4 shows time dependent deformation value of vessel. The Maximum deformation value of vessel is 0.424 mm (HBP) and 0.228 mm (NBP).

Articles


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

0.5 Velocity [m/s]

a)

0.3 0.2 0.1 0.2

0.4 0.6 Time [s]

0.8

1.0

0.2

0.4 0.6 Time [s]

0.8

1.0

180 Pressure [mmHg]

b)

150 120 90 60 30 0

0.0

Deformation of vessel wall [mm]

Fig. 3. Inlet timedependent blood flow (a) and blood pressure wave form (b). 0.45 0.30 0.15 0

0

2009

3. Results and discussion

0.4

0.0 0.0

N° 4

0.2

0.4 0.6 Time [s]

0.8

1.0

Fig. 4. Time dependent deformation value of vessel wall HBP and NBP

Figure 5 shows the effect of vessel condition on velocity distribution at peak systole. The inlet maximum velocity is called as “peak systole” (t = 0.1 s). The results show fast velocity area appears on the top of the coarctation and reverse flow appears behind the coarctation. When the vessel diameter expands at peak systole, distance between the coarctation and vessel wall becomes wider. However, the vessel diameter in the case (b) keeps constant value. In (b), distance between the coarctation and vessel wall becomes narrow and the fastest velocity appears above the coarctation. Figure 6 illustrates the effect of vessel condition on cholesterol distribution at peak systole. Mostly cholesterol is transported by blood flow because of high Schmidt number system. Figure 5 shows that reverse flow appears behind coarctation area. Therefore, cholesterol is transported by the reverse flow and cholestreol accumulated these area. Also, (c) has the highest cholesterol concentration. Figure 7 shows the effect of vessel condition on velocity distribution at peak diastole. The inlet minimum velocity is called as “peak diastole”. When the vessel diameter decreases at peak diastole, distance between the coarctation and vessel wall becomes narrow. Fast velocity area appears on the top of the coarctation and the reverse flow appears behind the coarctation, same as the case of peak systole as shown in Figure 5. Figure 8 indicates the effect of vessel condition on cholesterol distribution at the peak diastole. Cholesterol is transported by the reverse flow similar to the case of peak systole, and this phenomenon enhances growth of coarctation. Also, cholesterol concentration was increased under HBP and nonelastic conditions. Effect of vessel condition on cholesterol distribution in one cycle is shown in Figure 9.

a)

a)

b)

b)

c)

c)

Fig. 5. Effect of vessel condition on velocity distribution (left) and its enlarged view in “behind coarctation area” shown in Fig. 2 at peak systole (right). (a) NBP and elastic vessel, (b) NBP and nonelastic vessel and (c) HBP and elastic vessel. Articles

181


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

a)

b)

c)

Fig. 6. Effect of vessel condition on cholesterol distribution in the blood tube (left) and in “beh ind coarctation area” shown in Fig. 2 at peak systole (right). (a) NBP and elastic vessel, (b) NBP and nonelastic vessel, and (c) HBP and elastic vessel.

a)

a)

b)

b)

c)

c)

Fig. 7. Effect of vessel condition on velocity distribution (left) and its enlarged view in “behind coarctation area” shown in Fig. 2 at peak diastole (right). (a) NBP and elastic vessel, (b) NBP and nonelastic vessel and (c) HBP and elastic vessel.

a)

b)

c)

Fig. 8. Effect of vessel condition on cholesterol distribution in the blood tube (left) and in “behind coarctation area” shown in Fig. 2 at peak diastole (right). (a) NBP and elastic vessel, (b) NBP and nonelastic vessel, and (c) HBP and elastic vessel. 182

Articles


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

This result shows that cholesterol concentration changes between the peak systole (1.1 s) and the peak diastole (1.3 s) in the cases of (a) and (c). Because the blood flow suppresses the coarctation, cholesterol concentration becomes high value between peak systole. However, the coarctation expands at the peak diastole to decrease blood flow rate. Therefore, dead space appears behind the coarctation area and cholesterol concentration decreases in this area. Vessel diameter and back flow velocity of (b) remained constant values and they were larger than those of (a) (Figures 5 and 7). Therefore, cholesterol concentration of (c) is higher than that of (a).

4. Conclusions. A simulation code for calculating velocity distribution and cholesterol distribution in a blood tube with coarctation under various vessel conditions has been developed. The numerical results showed the followings: (1) Reverse flow appeared behind the coarctation. (2) Cholesterol was transported to behind of the coarctation area by the reverse flow. (3) HBP and noelastic vessel wall had a significant effect on the blood flow and cholesterol distribution.

AUTHORS Masashi Naito* - Department of Material Sciences & Chemical Engineering, Shizuoka University, 3-5-1 Johoku, Naka-ku, Hamamatsu, 432-8561, Japan. E-mail: f0930200@ipc.shizoka.ac.jp. Kensaku Mizoguchi - Department of Material Sciences & Chemical Engineering, Shizuoka University, 3-5-1 Johoku, Naka-ku, Hamamatsu, 432-8561, Japan. Present address; Science Academy of Tsukuba, Tsukuba International Congress Center, 2-20-3 Takezono, Tsukuba, 305-0032, Japan. Youhei Takagi - Department of Material Sciences & Chemical Engineering, Shizuoka University, 3-5-1 Johoku, Naka-ku, Hamamatsu, 432-8561, Japan. Yasunori Okano - Energy System section, Gradate School of Science and Technology, Shizuoka University, 3-5-1 Johoku, Naka-ku, Hamamatsu, 432-8561, Japan. * Corresponding author

N° 4

2009

flow”, Int. J. Numer. Meth. Fluids, no. 47, 2005, pp. 603-607.

Nomenclature a0 : modification coeffcients [-] an : Fourier cosine coefficient [-] bn : Fourier sine coefficient [-] c : cholesterol concentration [ mg/dL ] c0 : base cholesterol concentration (140) [ mg/dL ] C : dimensionless cholesterol concentration [ - ] D : diffusion coefficient [ m2/s ] E : Young’s modulus [ Pa ] h : thickness of vessel wall [m] Lmovement : movement of vessel wall [m] n : term number of fourier series [-] p : pressure [ Pa ] Pi : timedependent blood pressure [ mmHg ] Pbase : baseline blood pressure value [ mmHg ] Pblood : blood pressure value [ mmHg ] r : radius of vessel [m] t : time [s] u : velocity in horizontal direction [ m/s ] ui : inlet velocity of blood flow [ m/s ] U : dimensionless velocity in horizontal direction [ - ] v : velocity in vertical direction [ m /s ] V : dimensionless velocity in vertical direction [ - ] x : coordinate in horizontal direction [-] X : coordinate in horizontal direction [-] y : coordinate in vertical direction [-] Y : coordinate in vertical direction [-] [-] e : strain [ N/m2 ] DP : difference of blood pressure [s] Dt : time step [ m2/s ] n : kinematic viscosity [N] s : stress [-] Sc : Schmidt number

References [1]

[2]

[3]

[4]

Torii R. et. al., “Numerical investigation of the effect of hypertensive blood pressure on cerebral aneurysm Dependence of the effect on the aneurysm shape”, Int. J. Numer. Meth. Fluids , no. 54, 2007, pp. 995-1009. Kim T., Ley O., “Numerical Analysis of the cooling Effect of Blood Over Inflamed Atherosclerotic Plaque”, Trans ASME J. Biomech. Engineering, no. 130, 2008, 031013. Sugawara M., et. al., ”Relationship between the pressure and diameter of the carotid artery in humans”, Heart Vessels, 15, pp.49-51 (2000). Oshima M., et. al., ”Modeling of inflow boundary conditions for image - based simulation of cerebrovascular Articles

183


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

A VERSATILE MODEL FOR MUSIC CREATION UTILIZING TENSION STATES

Heikki Ruuska, Yoichi Takebayashi

Abstract: We investigates novel methods for creating systems for assisting musical composing. In this presentation we describe some previously neglected problem areas in musical structure analysis: knowledge-based computational expectation mapping and narrative-based structure. We also investigate how results from musical pattern analysis might be used for predicting human cognitive processing of general commonsense domains in everyday life. We conclude with a statement of how investigating music can lead to insights in other fields of cognitive science.

nation of some or all of timbre, typical melody, typical rhythm, or relative pitch. C. Musical pieces are placed in a setting – this can be a subset of the tonal world of Western classical music or Gamelan rhythmic repertoire, or a pitch range – with both tacit and explicitly suggested rules and discrete pitch and rhythmic blocks as basic units. An illustration is given in Figure 1.

Keywords: musical composition, music analysis, cognitive science, computer-generated music.

1. Introduction This paper aims to examine musical listening and composition processes from a cognitive science standpoint: what kinds of cognitive processes music engages, what kinds of effects it causes, and how those processes and effects could be simulated on a computer. We propose a tension-release model for analyzing expectations aroused by progressions in common musical pieces, and a computational model for analyzing scalable musical structures from an affective viewpoint. We test our theories on classical Western music and traditional Japanese music, and propose a framework for software to assist in music creation. Most current research on music analysis and composition assistance is based on mathematical analysis of repeating patterns, applied to datasets (existing pieces of music represented according to some schema), often coupled with classic Western music theory [1],[3],[4], [5],[6]. We view music as an interdependent projection and compliment of non-musical mental activities. This view necessitates comparing and relating musical activities to non-musical behavior, which is the topic of chapter 2.

2. Music as a Cognitive Process 2.1 Music as Stories Music and stories share similar structural aspects, and making the analogy allows for some insights. We limit our inspection to common practice music which shares these elements. A. Musical pieces have a beginning and an end. B. Musical pieces commonly have some protagonists and antagonists (instruments or parts), which act with or against each other. These are identified by a combi184

Articles

Fig. 1. Rough story model for a musical piece. 2.2 Narrative Centers Taking spoken stories as a base analogy, there are some basic observable facts regarding story frame structure of most music. — Musical cultures have taken a long time to build. This suggests an evolution has taken place where the musical culture has adjusted itself to its surroundings – in cognitively and culturally comfortable position. — Most pieces in common practice music are 5-15 minutes long. Longer pieces typically consist of shorter parts. This observation leads to suggest that 1) short pieces are not cognitively satisfactory: they either have too little material or too few repeats, and 2) long pieces either overload the cognitive machinery or bore the listener with too many repeats. — Most pieces have a few melodic or rhythmic patterns which are repeated, possibly with variations, throughout the piece. This could be due to an inability to remember patterns with a single effort, similarly to natural language. — Most music is pulsating with discretely observable beats: commonly less than 30 or more than 200 per minute, which suggests a cognitive linkage with human heartbeats and their associated mental states.


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

3. Tension State Model 3.1 Overview We have constructed a tension-release model for examining basic musical processing. The model treats each musical event as a signal which calls for resolution, causing expectations in the listener. These expectations are organized in layers for convenience, roughly reflecting familiarity with the musical matter but not necessarily implying strict structural hierarchy. 3.2 Patterns and Progressions Two ways people process music are by dividing it into discrete blocks (phrases and parts) and/or linear patterns (accelerating, slowing beats). For explaining listeners’ attentive interest in a piece of music, we propose the following tension-release model: each invocation of an identified pattern creates a mental tension which can have one or multiple expected release patterns. Tension and release patterns can be grouped into multiple layers. Table 1 illustrates three rough classes for patterns. Table 1. Tension-controlling Patterns (classes). Immediately observable Physical Properties Rising – lowering pitch (Pm) Hammering - continuous melody (Mm) Consonance - dissonance (Cm) Style-specific Knowledge Harmonic Progressions (Hm) Overall Piece Structure: forms (Fm) Explicit-relation Knowledge Comparison with other pieces, explicit quotations (Qm) Abstract structural references (Am)

Fig. 2a). Nocturne Op. 15 Nr. 3 (excerpt). Cresc

A«

1

Even

Even

B

2

2

ª « 1

2

Dim

2

ª « 1

2

2

Even

Even Ev ª 1

12

2

« ª 2

Ev «

2

2

2

«ª Ev ª 12

2

Ev « 1

2

2

Even

ª « 1

2

2

ª « 1

2

2

ª « 1

2

2

ª « 1

2

2

Cresc

Dynamics: All

Even

Pitch: Part 1

ª « 1

2

2

Pitch: Part 2 1

Part Interleaving

Pulse Pattern

Fig. 2b). Western Music Tension Model. Figure 2b depicts a partial analysis (Pm, Mm, Fm) of the said Nocturne. The following is a tension-release analysis of rhythmic changes at phrase end. Western music is typically constructed to an even pulsating rhythm, with fluctuations occurring in between or at the end of beginning of phrase blocks. (Tension: (M(Pulse=C))). Actual interpretation of the piece, which may or may not be written down in the score, often includes a slight “breath” at the end of the phrase (marked * in Fig. 2b). As a general pattern, phrase endpoints are often accompanied by a ritenuto. This creates a new level expectation: the phrase is going to end soon. (Tension: (M(Rit)) => Tension: (F(PhraseEndSoon=t))). And, if the phrase does actually end, creates an emphasis on it. (Tension:(F(PhraseEndSoon)) and Release:(F(PhraseEnd)) => (Tension:(F(Emphasis=Phrase?))). In some pieces, the next phrase may be precluded by a slight acceleration to balance the rhythm. 3.4 Case Study 2: Traditional Japanese Music

In the following sections, we use the tension-release model to analyze existing music. The abbreviations in brackets will be used. 3.3 Case Study 1: Western Tonal Music Western classical tonal music has been widely analyzed, not only by music theorists and composers but also by mathematicians and AI researchers (eg. Lerdahl [3], [4], Levitt [1], Mazzola [5]). However, current theoretical analysis fails to answer questions such as why certain phrases and/or their combinations cause pleasure, frustration, excitement and other emotive reactions. Western music is often composed of equal-length patterns divided into clear blocks: these blocks can further be subdivided into phrases, which also are often of equal length. The equal-length and scalable system creates default assumptions for when each phrase begins and ends. Once a phrase has been introduced, a repetition of its beginnings will create an expectation of something similar following. Figure 2a and 2b provide an illustration of this: an excerpt from Nocturne in g Op. 15 No. 3 by Frederick Chopin.

Fig. 3a). Rokudan no Shirabe (excerpt). Cresc

Dim

Cresc

Dim

Cresc

Even

A Even

Even

ª«

ª«

Dynamics

Effects

E.pat

ª«

Even

Pitch

Pulse Pattern

Figure 3b). Japanese Music Tension Model. Articles

185


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

Japanese traditional music, not very well known in the West, has methods for creating variations and expectations quite different from those used in Western classical music. For an introduction to the style, the reader is encouraged to refer to Malm 2000 [7]. The use of pentatonic scales, alongside a tendency to not to compromise individual sound qualities over ease of modulation in instrument evolution, have resulted in a musical culture where complex harmonies and scale modulation are not as readily usable as in Western music. Instead, expressive power of individual instruments, complex rhythmic nuances and constant fluctuation of tempo are used to create tensions and variation. Figure 3a and 3b show a part of a classical Japanese piece Rokudan no Shirabe (Six Variations) by Kengyo Yatsuhashi (16??). The excerpt is taken from the fourth variation. An accelerative pattern is noticeable from the pulse line. (Tension:(M(Acc=C))). This would generally continue until the end of the section, with slight pauses (rest-stops) in between. Further local tensions are caused by jumps in pitch, which often signify phrase change. (Tension: (F(PhraseEnd)) associate Tension:(P(HighJump))).

4. Experimental Music Creation 4.1 Related Research Much previous research exists on electronic composing assistance technologies. Beside now-common tone synthesis and automated digital alteration of timbre and pitch, there is software for chord completion, melody generation and improvisation. Most of these are based on mathematical formalization, statistical pattern analysis, or rule-based evaluation of commonly occurring chord, rhythmic and melodic patterns (for example [?], [/], [?], respectively). However, the methods are limited to analysis of tonal structure in Western music. 4.2 Current Data Currently, we have 24 tension pattern types for classical Western music and 14 types for traditional Japanese music, gathered from three compositions (total of 437 measures according to notation, or 115 annotated phrases). Whereas the number of music samples is limited, the in-depth analysis has enabled us to create a suggestive grammar for improving annotated ad hoc improvisations. As noted in Section 2.3, a full musical piece should be at least a few minutes long, which makes meaningful evaluation impossible with the current data set.

5. Discussion and Future Work 5.1 Performance and Education Assistance Ability of computers to engage in collaborative musicianship is poor: a machine works in a pre-programmed fashion, without ad hoc stopping or taking of breaths. An anticipating, backtracking model for mid-level analysis should help in a more interactive communication between a synthesizer and musicians in a live performance. 5.2 New Music and Cultural Aspects What kind of cognitive machinery is associated with 186

Articles

N° 4

2009

processes in learning music? We take by assumption that one of the reasons people like music is because it gives them a challenge-release framework, like stories. Some basic characteristics remain the same, but by inclusion of new patterns, people can find new insights and ways to think about what they are and have been doing.

6. Contributions We have presented a case for cognitive affect analysis of musical processing. To partially explain human attraction to music, we have constructed a model where musical patterns create musical goals in human minds, which require resolutions. Interest in following the piece to the end depends on how satisfactorily these resolutions are carried out. Called tension-release model, we have analyzed existing musical pieces for tension-creating patterns on a more concise scale than done before, and found it useful for explaining cognitive effects of variations within a piece of music.

AUTHORS Heikki Ruuska* - Graduate School of Informatics, Shizuoka University, Hamamatsu 432-8011, Japan. E-mail: hruuska@acm.org. Yoichi Takebayashi - Graduate School of Science and Technology, Shizuoka University, Hamamatsu 432-8011, Japan. E-mail: takebay@inf.shizuoka.ac.jp. * Corresponding author

References [1] [2] [3] [4] [5]

[6]

[7]

Schwanauer, Stephan M., Levitt, David A., Machine models of music, MIT Press, 1993. Minsky, Marvin, The Emotion Machine, Simon & Schuster, 2005. Lerdahl, Fred, Jackendoff, Ray S. , A Generative Theory of Tonal Music, MIT Press, 1982. Lerdahl, Fred, Tonal Pitch Space, Oxford University Press, 2001. Mazzola G., ”Mathematical Music Theory-Status Quo 2000”. In: Perspectives in Mathematical and Computational Music Theory. EpOs Osnabruck, 2004. Hirata, Keiji, Aoyagi, Tatsuya, “Computational Music Representation Based on the Generative Theory of Tonal Music and the Deductive Object-Oriented Database”, Computer Music Journal. MIT Press, 2003. Malm, William P., Traditional Japanese Music and Musical Instruments, Kodansha International, 2000.


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

A NEW MAN-MACHINE INTERFACE FOR ISPACE APPLICATIONS András A. Tóth, Annamária R. Várkonyi-Kóczy

Abstract:

2. The Intelligent Space

Intelligent Space or iSpace is a new kind of computing system aiming at improving the environments of humans, creating a natural and easy to use solution. Its main feature is that the intelligence is not implemented separately in the actors, but it is distributed in the whole space. In this paper, we present a hand gesture and movement recognition system, whose purpose is to be used as an intuitive interface for the iSpace.

Today, with the spread of machine intelligence, “smart environment” became a popular tool for humans collecting information, bearing, forming the environment, getting assistance etc. Intelligent Space (iSpace) is an intelligent environmental system offering ambient intelligence for improving the comfort and safety of everyday life as well as for achieving personalised healthcare and independent living for disabled persons. The main and ultimate goal of such systems is to build an environment that is human centered, comprehends human interaction, and satisfies them [3]. This means that the system should be easy to use for the people in it: they should be able to express their will through intuitive actions and there should be no need for them to learn how the system is to be used. The most characteristic feature of the iSpace is that the intelligence is distributed in the whole space, not in the individual agents. Furthermore, iSpace is able to monitor what is going on in the space and is able to build models based on this knowledge. The system is also capable to react with its environment and provide information or physical services to its users. There are several applications currently developed or planned for the iSpace. These include the monitoring of physiological functions of humans, the positioning and tracking of humans [3], the localization of mobile robots [3], the control of robots [3], and finding paths for them by using itineraries taken by people [4], etc. For detailed description of the structure and operation of iSpace, see [2], [3], [4], and also the paper of the authors of this paper published at IA'2009 [5].

Keywords: Ubiquitous computing, intelligent space, manmachine interaction, smart environments, image processing, intuitive user interface.

1. Introduction A new paradigm in user-machine interaction is “Ubiquitous Computing” [1]. The main goal of Ubiquitous Computing systems is to offer such an interface to the user that should be so natural and easy to interact with, that users become unaware of the fact that they are using a computing system. Intelligent Space (iSpace) [2] is a special intelligent implementation of the Ubiquitous Computing paradigm, which can be any smart area, such as a room, a railway station, an underpass, road crossing, or even a town, etc. equipped with intelligent sensors and agents. The main feature of the iSpace is that the intelligence itself is not present in the agents but it is distributed in the whole space. Thus, the architecture of the artificial agents, such as robots, is quite simple as they are coordinated by the intelligent sensors. Another important feature of the Intelligent Space is its capability of observing what is happening in it and to build models of the environment. In case of necessity the iSpace is also capable to interact with the environment in order to achieve some kind of change or give information to its users. In this paper, authors introduce a new human-machine interface to iSpace applications which is intuitive and easy to use. By this, users become able to issue orders to the iSpace assuming simple gestures of their hands and/or issuing movements with them. The rest of this paper is organized as follows: Section 2 describes the most important features and the architecture of the iSpace. Section 3 details the set up and the operation of the proposed man-machine interface. Section 4 analyzes the performance of the experimental system. Finally, Section 5 concludes the paper and outlines possible improvements.

3. The new hand gesture and movement controlled human-machine interface A. Overview of the system At current stage of the development, the procedures described in this section assume that the whole field of view of the sensor is filled with a homogeneous background and the user makes hand gestures and movements before this background. No other parts of the body or other objects are visible on the image delivered by the sensor; however the absence of the hand is permitted. Hand gesture recognition and the three dimensional modeling of the hand works with the input of two cameras. For this procedure it is presumed that the hand is fully visible on both input images or not visible at all and that the hand is steady. A further assumption is that stretched fingers are well separated and that the hand stays parallel to the cameras. Articles

187


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N째 4

2009

valleys using the fuzzy based matching algorithm [6]. Finally, the DLT method is adopted in order to perform the triangulation which will yield the three dimensional coordinates of the feature points. C. Hand gesture estimation The matched peaks and valleys (see [5]) are used not only to build a spatial model but to estimate the current hand gesture, as well. Currently, three hand gestures have been defined which are called hand gesture A, B and C, respectively. Gesture A represents a hand with all five fingers stretched out. Gesture B is a closed fist. Finally, thumb, index and middle fingers are stretched out in case of gesture C, whilst the other two are bent. For each predefined hand gesture the number of peaks and valleys is stored and the count of respective feature points on the processed frame is compared to the stored count of the predefined gestures, as summarized in Table 1. The hand gesture estimator can yield five types of result which are: no hand on the picture, gesture A, gesture B, gesture C, other type of gesture. Fig. 1. Gesture estimation and three dimensional reconstruction workflow. Hand tracking and movement classification uses only the input of one camera. In the case of these procedures we assume that the hands are moving slowly or they are steady. Hands visible on the picture never touch or overleap with each other. Furthermore, at most two hands are to be seen on the input sequence. In this way the software part of the system consists of two main components: the first one processes two images of a stereo camera pair and has two functions: it classifies hand gestures and yields three dimensional locations of feature points of the hands. The second takes only the video stream of one camera as an input and performs the tracking of a slowly moving hand. Furthermore, it tries to match the shape of the movement with one of the predefined shapes. In the current implementation two predefined shapes are considered: a line and a circle but in the future other shapes will be defined, as well. The workflows of these two components are illustrated in Figure 1 and 2, respectively. The first step of both components is the extraction of skin regions. To achieve this, histogram back projection is performed as described in [5]. The next steps to retrieve areas containing skin include applying a threshold to the back projected image and locating the connected components, discarding those having the area below a given threshold. B. The 3D model The three dimensional model of the hand captured by the sensors consists of the spatial location of the feature points of the hand. Three steps are executed in order to build the model: first feature points are extracted separately in both images. Then, these feature points are matched and finally, three dimensional coordinates are calculated using the known camera matrices. The matching occurs separately for the peaks and the 188

Articles

D. Hand tracking In order to track the motions of the hand a simplified approach based on the one presented in [7] is developed. We consider only the center of gravity of the blobs, which can be computed as follows:

(1) where (xc, yc) denote the center of gravity of the blob and b(x, y) is the intensity at a given position. Table 1. Expected and accepted peak and valley count for each type of hand gesture. Gesture Expected peaks Expected valleys Accepted peaks Accepted valleys

A 5 4 At least 5 At least 3

B 0 0 Exactly 0 Exactly 0

C 3 2 Exactly 3 At least 1

Fig. 2. Block diagram of the hand tracking and movement estimation. The reason for taking only center of gravity into account is that color information is unnecessary, as all blobs represent hands and have thus the same color.


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

Their size will also be approximately the same, as they are about the same distance from the camera. Finally, speed is neglected because it was assumed that tracked hands move slowly. The tracking procedure works in the following way: in each frame it tries to match the current blobs to those of the previous frame by finding the minima of the matchscore matrix of the distances of the blobs of the current frame and the previous one. The same ID is assigned to the matched blobs throughout the whole sequence. A new unique ID is assigned to unmatched blobs (which are supposed to belong to hands that have just entered the scene). The sequence of the centers of gravity is stored for each blob. This sequence is referred to as the motion history of the blob. When a hand leaves the scene, no blobs in the next frames will be matched to it again, and thus its motion history ceases to grow. E. Movement recognition The recognition of predefined movements is achieved on a rule based manner by using some statistical parameters of the motion history yielded by the tracking procedure. For each motion type to be estimated, appropriate thresholds for the statistical parameters have been set. Each motion history belonging to a hand still visible on the image is matched (until the first steady point) to each of the threshold-sets, and if all constraints of a given threshold-set are met, the appropriate motion is predicted. Thus, the possibility that more than one motion type is predicted for the same movement is not intrinsically excluded. The considered parameters for linear motion are the sum and variance of the angles between two subsequent motion vectors, to which upper thresholds have been set and one of the sums of the shift in directions x or y has to be above a lower threshold. In case of the circular motion the parameters taken into consideration are the following: the maximum and minimum values of the angle between the motion vector and axis x must be above and below an upper and lower bound, respectively. The average of this angle also has to be below a given upper threshold. The averages and sums of shift in directions x and y have all to be below an upper bound.

4. The experimental system We have built an experimental set up for testing and analyzing the performance of the new interface. The hardware part of the testing system consists of two web cameras (Genius Look 316) connected to a PC (CPU: AMD Athlon 64 4000+, 1 GB RAM, OS: Windows XP). The cameras are located at one end of a table. At the other end a homogeneous background is mounted. The image processing algorithms are running on the PC and are implemented using OpenCV [8]. For the camera calibration the Camera Calibration Toolbox for Matlab has been adopted [9]. The experimental setup is illustrated in Fig. 3.

N째 4

2009

Fig. 3. The experimental setup (d is the distance of the tip of the index finger from the homogeneous background). B. Functional test We have worked with two types of tests. The purpose of the functional tests is to get the ratio of correct answers of the implemented modules when the given preconditions are met. The first measurement tests whether the hand gesture classifier based on the peaks and valleys count is able to differentiate between the predefined hand gestures A, B, and C and some negative gestures which have different count of peaks and valleys using 50 samples for each gesture type. It was found that this procedure gives acceptable results if the hand lies apart from the homogeneous background and does not lie too near the camera. (In the first case, the contour might become noisy due to the hand region touching the shadow while in the latter one, peaks and valleys may fail to be detected.) The results of the measurement series satisfying these constraints are summarized in Table 2. Table 2. Results of the hand gesture prediction. Input/Response A B C Negative

A 46 0 0 0

B 0 50 0 0

C 0 0 45 0

Negative 4 0 5 0

The second test aims at determining the accuracy of the three dimensional reconstruction. As also stated in [10], due to the lack of ground truth, it is more convenient to measure the jitter of the reconstruction rather than the absolute position itself. We measured the jitter at three different locations (d=22 cm, d=32 cm, d=42 cm, where d is the distance of the tip of the index finger as shown in Figure 3), using 20 samples at each location. It was found that the spatial reconstruction gives little jitter in the directions parallel to the camera planes (i. e. directions X and Y), and it yields higher jitter in the direction perpendicular to the plane though the values are still usable. Results for d=32 cm are summarized in Table 3. Here Deviation means the standard deviation of the vector of the given coordinate component

(2) where X denotes the vector of the X coordinates, avg is the average of the X coordinates, and N is the number of Articles

189


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

the valid samples. At d=42 cm there were three samples where the peak of the index finger failed to get detected. Deviation for Y and Z components is calculated similarly. Table 3. Jitter measurement results for d=32 cm. All the values in the table are shown in mm. Direction X Y Z

Maximum difference 2.75 3.46 13.30

Deviation 0.80 0.80 3.71

The third test demonstrates the performance of tracking. The tracker assigns an ID to each yet unidentified blob entering the scene, which should be kept until the object corresponding to that blob leaves the scene. The tracker meets this requirement in case of the predefined working conditions. The last test measures the performance of the hand movement classifier. Both movements were performed 16 times from various starting points and in various directions. The linear movement classifier recognized linear motion 13 times whilst circle classifier gave the right result 12 times. Thus, the classifiers have an acceptable performance.

ACKNOWLEDGMENTS This work was sponsored by the Hungarian National Scientific Fund (OTKA 78576).

AUTHORS András A. Tóth* - Integrated Intelligent Space Japanese-Hungarian Laboratory. E-mail: bamota@gmail.com. Annamária R. Várkonyi-Kóczy - Institute of Mechatronics and Vehicle Engineering, Budapest Tech, Integrated Intelligent Space Japanese-Hungarian Laboratory. E-mail: koczy@mit.bme.hu. * Corresponding author

References [1]

[3]

[4]

[5]

[6]

[7]

[8]

5. Conclusions and future work In this paper the concept and the experimental setup of a new man-machine interface are introduced. This interface makes able humans to control iSpace by hand gestures and hand movements. The presented experimental setup at current stage is able to classify three simple hand gestures and two basic hand movements. In our future work we plan to extend the interface to be able to understand more hand gestures and hand movement instructions. The reliability of the operation can be increased by applying a more sophisticated hand mo190

Articles

2009

del, by taking into account the position and direction of fingers as well. To make more “natural” the use of the interface we will also examine how could we neglect such limitations required in the first implementation like homogeneous background, non-touching, non-overlapping, and slowly moving hands.

[2]

C. Timing tests The timing tests measure the time necessary to compute the steps described in the previous sections. It is important that these procedures work at real time speed in order to be used for the intuitive interface. For the first subsystem (see Figure 1) the considered steps were contour extraction, peaks and valleys localization, feature point matching, and computation of the spatial points. Each of the first three steps was found to operate in the order magnitude of 10 ms, which is about the frame grabbing rate of commercial 15-30 Hz web cameras. The time needed by spatial reconstruction was significantly shorter: it was found to be in the order of magnitude of 10 μs. The steps considered for the second subsystem (see Figure 2) are blob extraction, blob tracking, and testing for circular and linear movements. Magnitude of blob extraction time is the same as that of frame grabbing, blob tracking and shapes detecting time was 100 μs and 10 μs, respectively. In order to measure the time the standard C routine clock() was used. As the granularity of this timer is larger than the measured intervals, the timings were obtained by averaging more measurements.

N° 4

[9]

[10]

M. Weiser, The Computer for the Twenty-Frst Century, Scientific American, 1991 pp. 94104,. Lee J-H., Hashimoto H., “Intelligent Space”. In: International Conference on Intelligent Robots and Systems 2000 (IROS 2000), vol. 2, 2000, pp. 1358-1363. Lee J-H., Morioka K., Ando N., Hashimoto H., “Cooperation of Distributed Intelligent Sensors in Intelligent Environment”, IEEE/ASME Transactions on Mechatronics, vol. 9, no. 3, 2004. Appenzeller G., Lee J.-H., Hashimoto H, “Building Topological Maps by Looking at People: An Example of Cooperation between Intelligent Spaces and Robots”, Intelligent Robots and Systems, vol. 3, issue 7, 1997, pp. 1326-1333. Várkonyi-Kóczy A.R., A Tóth, ISpace - a Tool for Improving the Quality of Life, Journal of Automation, Mobile Robotics and Intelligent Systems. Special issue for Inter-Academia 2009, vol. 3, no. 4, 2009, Várkonyi-Kóczy A.R., “Autonomous 3D Model Reconstruction and Its Intelligent Application in Vehicle Systh tem Dynamics”. In: 5 International Symposium on Intelli-gent Systems and Informatics (SISY 2007), Subotica, Serbia, 24th-27th August 2007, pp. 13-18. Intille S.S., Davis J.W., Bobick A.F., “Real-Time ClosedWorld Tracking”. In: Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, June 1997, pp. 697-703. Bradski G., Darrell T., Essa I., Malik J., Perona P., Sclaroff S., Tomasi C., et al., Intel OpenCV Library, software available online at: http://sourceforge.net/projects/opencvlibrary Bouguet J-Y., Camera Calibration Toolbox for Matlab, software available online at: http://www.vision.caltech.edu/bouguetj/calib_doc/ Segen J., Kumar S., Shadow Gestures: 3D Hand Pose Estimation Using a Single Camera, IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 1999.


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

SYNTHESIS AND EVALUATION OF NOVEL MRI CONTRAST AGENTS OF CHEMICALLY MODIFIED GD-DTPA COMPLEXES WITH SUGARS Masaki Sugiyama, Mitsuji Yamashita, Gang Yu, Michio Fujie, Keisuke Ogawa, Nobuhisa Ozaki, Takashi Aoki, Sayaka Mizuno, Shingo Okada, Kentarou Tachi, Kengo Aoshima, A. Uma Ravi Sankar, Bitragunta Siva Kumar, Yasuo Takehara, Harumi Sakahara

Abstract: MRI is one of medical diagnostic imaging technologies that can draw the cross section in the body. To obtain a clearer image, Gd complexes are often used as MRI contrast agents. Gd-DTPA (Gd-Diethylenetriaminepentaacetate, Magnevist®) is used in particular as the MRI contrast agents. We prepared and evaluated novel MRI contrast agents that were chemically modified Gd-DTPA with sugars (represented as Gd-DTPA-Sugar) via hydrolysis route for providing specificity to target organs and tissues. Gd-DTPASugar complex showed an excellent potential for the MRI contrast agent (r1=31.2 s-1mM-1). Gd-DTPA-Sugar complexes alternatively prepared by shorter synthetic route without protection/ deprotection (hydrolysis) method showed inferior results (r1=6.3 and 8.1 s-1mM-1) to the hydlized product.

molecular size big, Gd-DTPA-Sugar complexes having extended carbon chains were prepared with shorter synthtic route.

2. Results and discussion 2.1. Synthesis of Gd-DTPA-Sugar complex

Diethylene triamine DETA-2Glc(OH) 2

Keywords: MRI contrast agent, Gd(III)-DTPA, tumor imaging.

1. Introduction MRI is one of a medical diagnostic imaging technology, and it can obtain the cross section of all angles in the body. MR imaging is obtained from the difference of nuclear relaxation time of protons, which are resonated water and fat protons in the body by irradiation in high magnetic field. Therefore, even if the contrast agents need not to be used for MRI, the imaging is possible. But to obtain a clearer image, Gd complexes are used as MRI contrast agents. Gd complexes enhance contrast by shortening T1 relaxation times of water protons. Because T1 relaxation time depends on Gadolinium concentration of MRI contrast agent, r1 relaxivity that divided T1 relaxation time by Gadolinium concentration is used as a guide to contrast intensification of MRI contrast agent. Now, Gd-DTPA is used extensively as MRI contrast agents [1]. However, Gd-DTPA has problems such as that it’s not so high r1 relaxivity, low retention in blood vessels and no specificity in the body. Our laboratory designed novel GdDTPA complexes for that chemistry modified sugar [2]. To give organ and tissue specificity, we focused an attention on function of sugars as organ and tissue specificity, and then chemically modified Gd-DTPA complexes with sugars became the candidates for resolving the problems of the Gd-DTPA. We prepared some Gd-DTPA-Sugar complexes and evaluated in vivo and in vitro. Gd-DTPA-Sugar complex was showed great result. So, Gd-DTPA-Sugar complex was prepared by short route for large quantity synthesis. Then, because r1 relaxivity is improved when

DETA-2Glc(OAc) 4

DETA-2Glc(OAc)-Boc 3

DTPA-DETA-D2-4Glc(OAc) 5

Gd-DTPA-DETA-D2-4Glc(OAc) 6

Gd-DTPA-DETA-D2-4Glc(OAc) 7a

Scheme 1. Synthesis of Gd-DTPA-Sugar complex. Reagents and conditions: (a) D-(+)-Glucono-1,5-lactone, DMF, r.t., 24 h; (b) (Boc)2O, DMF, r.t., 24 h, Ac2O, Et3N, r.t., 48 h; (c) TFA, CH2Cl2, r.t., 4 h; (d) DTPA dianhydride, DMF, Pyridine, r.t., 4 h; (e) GdCl36H2O, 95 °C, 1 h; (f) NaOH aq(1N), H2O, r.t., 24 h. Articles

191


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

A pathway of synthesis of Gd-DTPA-Sugar complex is shown in Scheme 1. Synthesis of Gd-DTPA-sugar complexes, that consist of dendrimer structure, used a convergent method. DTPA dianhydride 1 of dendrimer core was prepared by dehydration-condensation of DTPA. DETA-2Glc(OH) 2 was prepared by reaction of D-(+)-Glucono-1,5-lactone and primary amine groups of diethylenetriamine. DETA-2Glc(OAc)-Boc 3 was prepared by t-Boc protection of secondary amine group by (Boc)2O and acetylation of hydroxyl groups by Ac2O. DETA-2Glc(OAc) 4 of dendrimer terminal was prepared by deprotection reaction of t-Boc group. DTPA-DETA-D2-4Glc(OAc) 5 of ligand was prepared by reaction of DTPA dianhydride 1 of dendrimer core and DETA-2Glc(OAc) 4. Gd-DTPA-DETAD2-4Glc(OAc) 6 was prepared by chelation reaction of DTPA-DETA-D2-4Glc(OAc) 5 and Gadolinium(III) ion. GdDTPA-DETA-D2-4Glc(OH) 7a of Gd-DTPA-sugar complex was prepared by hydrolysis of acetyl groups of Gd-DTPADETA-D2-4Glc(OAc) 6 by NaOH aq(1N). 2.2. Short route syntheses of Gd-DTPA-Sugar complexes

N° 4

2009

Shorter routes of syntheses of Gd-DTPA-Sugar complexes were shown in Scheme 2. DETA-2Glc(OH) 2 and HMTA-2Glc(OH) 8 of Dendrimer terminals were prepared by reaction of D-(+)-Glucono-1,5-lactone and primary amine groups of diethylene triamine and bis(hexamethylene)triamine. DTPA-DETA-D2-4Glc(OH) 9 and DTPAHMTA-D2-4Glc(OH) 10 of ligands were prepared by reaction of DTPA dianhydride 1 of dendrimer core with DETA2Glc(OH) 2 and HMTA-2Glc(OH) 8 of dendrimer terminals. Gd-DTPA-DETA-D2-4Glc(OH) 7b and Gd-DTPA-HMTA-D24Glc(OH) 11 of Gd-DTPA-sugar complexes were prepared by chelation reaction of DETA-D2-4Glc(OH) 9 and DTPAHMTA-D2-4Glc(OH) 10 and Gadolinium(III) ion. 2.3. in vitro evaluation The value of r1 relaxivity, that is caluculated by being divided T1 relaxation time by gadolinium concentration, is used as a guide to contrast intensification of MRI contrast agent, because the relaxation time depends on the gadolinium concentration of MRI contrast agent. Because gadolinium complex formation constants depend on the pH value of the aqueous media and the free gadolinium ion concentration that did not form the complexes have influence on measurements of relaxation time, the media for gadolinium complex preparation were adjusted to pH 7.0 in water, and to the media was added Chelex®100 Resin, stirred for six hours, and thus removal of the free gadolinium ion was performed. The removal of free gadolinium ion was confirmed by the color test by using Xylenol Orange. Gadolinium concentration was measured by an ICP-AES instrument because relaxation time depended on gadolinium concentration of contrast agents. T1 was measured by TD-NMR of 0.47 T at 37 °C. T1 was measured not only in water but also in serum albumin which is the mostly existing protein in blood. Table 1. Comparison of r1 relaxivity. Gd complexes

Scheme 2. Short route syntheses of Gd-DTPA-Sugar complexes 7b and 11. Reagents and conditions: (a) Ac2O, pyridine, 65 °C, 24 h; (b) D-(+)-Glucono-1,5-lactone,DMF, 2 r.t., 24 h, 8 80 °C, 12 h; (c) pyridine, DMSO, 60 °C, 24 h; (d) GdCl36H2O, pyridine, 40 °C, 12 h.

Gd-DTPA 7a 7b 11

2.4. in vivo evaluation The relaxivity constant r1 was calculated by the following expression.

Fig. 1. Rat's MRI when Gd-DTPA (0.1 mmol/kg) was administered. 192

Articles

r1 [s-1mM-1] in H2O in albumin 3.5 3.5 31.2 6.3 6.8 8.1 7.7


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N째 4

2009

Fig. 2. Rat's MRI when Gd-DTPA-DETA-D2-4Glc(OH) 7a (0.05 mmol/kg) was administered.

Fig. 3. Rat's MRI when Gd-DTPA-DETA-D2-4Glc(OH) 7b (0.05 mmol/kg) was administered.

Fig. 4. Rat's MRI when Gd-DTPA-HMTA-D2-4Glc(OH) 11 (0.05 mmol/kg) was administered.

r1 ; relaxivity [s-1mM-1] T1 ; relaxation time [ms] r1H O ; water of relaxivity [s-1mM-1] [Gd3+] ; Gadolinium concentration [mM] 2

The MR images of the rats were drawn by MRI machine at 3.0 T. Concentration of MRI contrast agent were adjusted by normal saline solution, Gd-DTPA solution used was 0.1 mmol/kg, solutions used for Gd-DTPA-DETA-D24Glc(OH) 7a (Fig. 2), Gd-DETA-D2-4Glc(OH) 7b (Fig. 3) and Gd-DTPA-HMTA-D2-4Glc(OH) 11 (Fig. 4) were 0.05 mmol/kg. The each MR image of Fig. 1-4 shows before administration, and then about 1, 5, and 20 minutes after the administration, respectively, from left to right.

3. Conclusion Gd-DETA-D2-4Glc(OH) 7(a) showed great result to give quite higher r1 relaxivity by in vitro, clearer contrast effect, higher retention in blood vessels, and specificity in the liver by in vivo. But Gd-DETA-D2-4Glc(OH) 7(b) being prepared by shorter route showed not so great result.

AUTHORS Masaki Sugiyama, Takashi Aoki - Department of Materials Science, Graduate School of Engineering, Shizuoka University, Hamamatsu 432-8561, Japan. Mitsuji Yamashita*, Nobuhisa Ozaki, Bitragunta Siva Kumar - Department of Nano-Materials, Graduate School of Science and Technology, Shizuoka University, Hamamatsu 432-8011, Japan. E-mail: tcmyama@ipc.shizuoka.ac.jp. Gang Yu - Innovative Joint Research Center, Shizuoka University, Hamamatsu 432-8561, Japan. Michio Fujie - Analytical Instrument Center, Faculty of Medicine, Hamamatsu University School of Medicine, Hamamatsu 431-3192, Japan. Keisuke Ogawa, Kengo Aoshima - Department of Materials Science, Graduate School of Science and Engineering, Shizuoka University, Hamamatsu 432-8561, Japan. Sayaka Mizuno, Shingo Okada, Kentarou Tachi - Department of Materials Chemistry, Faculty of Engineering, Shizuoka University, Hamamatsu 432-8561, Japan. A. Uma Ravi Sankar - Japan Association for the Advancement of Medical Equipment, Hongo, Bunkyo-ku, Tokyo 113-0033, Japan. Yasuo Takehara, Harumi Sakahara - Radio Isotope Laboratory, Faculty of Medicine, Hamamatsu University Articles

193


Journal of Automation, Mobile Robotics & Intelligent Systems

School of Medicine, Hamamatsu 431-3192, Japan. * Corresponding author

References [1] [2]

194

Bongrtz G., Magn. Reson. Mater. Phy., vol. 20, 2007, pp. 57-62. Takahashi M., Hara Y., Yamashita M., et.al, Tetra-hedron Lett., vol. 41, 2000, pp. 8485-8488.

Articles

VOLUME 3,

N째 4

2009


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

N° 4

2009

THE DESIGN OF AN INSULIN PUMP ­ CONCEPT OF CLOSING THE LOOP

Hubert J. HawĹ&#x201A;as, Krzysztof Lewenstein

Abstract: This paper explains what insulin pump is and outlines basic requirements for such device. The material presented in this paper is an attempt to collect actual knowledge about closed loop insulin delivery systems. From scientific point of view we are going to propose the closed loop insulin-dispensing device with control built on the base of predictive neural network system. The blocks of such systems are listed, and main problems to overcome are defined.

Treatment requires that the person be connected to the insulin pump for nearly 24 hours a day, 7 days a week. Thus patient carries the pump with himself/herself mostly all the time and the pump acts almost like a part of his/her body. This indicates the importance of a wellthought out design. Insulin pump is compact mechatronics device, which may be split into functional blocks as show in Fig. 1. The link between the human body and the insulin pump is The Infusion Set. It consists of a needle or cannula and thin tubing connected to a reservoir for insulin (Cartridge).

Keywords: insulin pump, diabetes, artificial pancreas. Infusion set

1. Introduction An insulin pump is a device used for continuous dosage of insulin at a selected rate. An ordinary insulin pump is automatic drug-dispensing device that works in open loop mode. This means it exactly follows programmed drug dispending instructions with no any feedback form the patient body. Even with such easy job to do this device significantly facilitates treatment and improves the lives of diabetic patients. In conventional insulin therapy [1], two types of insulin are needed: long-term and short-term insulin. Long-term type is used for daily insulin demand. Shortterm type is fast acting insulin that is needed to reduce postprandial blood glucose (BG) level and its dose mainly depends on carbohydrates contents in meals. An insulin pump uses only short-term insulin, which is distributed evenly during the day, so it works on a longterm basis (called basal), and a dose to counteract food (called bolus) that works on a short-term basis. This small mechatronics device makes the life of patients more normal. It is important especially in the case of children. Unfortunately, insulin pumps are not easily affordable because of their price, as high as several monthly incomes of an average family in Poland. There is no manufacturer of such pumps in Poland, which is one of the reasons of their high price. Starting production in Poland would lower the price and make the device more affordable, which is one of the reasons why we started our design.

2. Ordinary insulin pumps There are three types of such devices (a) hospital insulin pump, (b) personal insulin pump, (c) implantable insulin pump. This paper regards insulin pump of personal type, which is an iPod-size unit. The whole unit and each of its blocks must fulfil clinical, technical and user requirements

Case (Body)

Cartridge

Pump Processing Module

Monitoring Module

Controls

PSU

CGM Module

Sensor

Fig. 1. Functional diagram of an insulin pump. The Pump Module controlled by Processing Module doses a precise amount of medicine through Infusion Set into the patient's body. There are also other blocks, as Control Module that let user to communicate with device, or Power Supply Unit (PSU), which controls battery level and provides power to whole units. The Monitoring Module may be a function of the Processing Module instead of being a device by itself. It has to monitor proper work of all blocks of the pump. It must detect any malfunction of the whole unit and also has to inform the user about it, e.g. about a disconnected infusion set. Blocks in dashed line are external and optional part of the insulin pump. These modules are planned in a version with closed-loop operation. The Continuous Glucose Monitoring (CGM) Module and a Sensor are devices for measuring level of blood glucose. Our plan is that the glucose sensor would allow creating a closed loop control of insulin dosage making the insulin pump work almost like an artificial pancreas. Articles

195


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

2.1. Basic requirements. Giving priority to the functional specifications is mandatory requirement for reliable and safe operation. The device must have a monitoring system that ensures safe operation in case of hardware failure or improper usage. Not only electrical and mechanical components have to be reliable but working algorithm has to be safe, too. Some other requirements are: miniaturization, energy-efficient design or basal range, duration and frequency, or types (shape) of bolus. 2.2. Patient wish list Some users would like to have some extra features/ functions in their pumps [2] amongst them the remote control is mentioned often. Pumps are quite often hidden under clothing, so using a remote control would help to take a bolus or check an alarm when in public. Remote control in the form of a hand watch is a solution suggested by some of them. Another feature/function that users of the insulin pumps would like to have is the possibility of scanning meal for contents of carbohydrates leading to the automatic insulin dosage. Unfortunately, this feature/function looks sci-fi idea, however, we can create a user-edited library with meals and their contents of carbohydrates and knowing approximate body response to insulin and carbohydrates than calculate the insulin bolus.

3. Closing the loop Just an ordinary insulin pump is a “dumb machine” that does exactly what its wearer says. But we can imagine device that will automatically choose appropriate dose of medicine to correct BG level. In this case we could treat it as an artificial pancreas. But term artificial pancreas popularly used for such devices may be seemed not fully adequate even if we are interested in glycemic control only. There are more functions of pancreas and glycemic control is more complex. The real healthy pancreas produces amylin (IAPP) and glucagons (GCG) along with insulin (INS) to stabilize BG level. Action of glucagon is opposite to that of insulin, which instructs liver cells to take in glucose from blood. Effect of amylin is to slow down rate at which glucose appears in blood, by slowing down digestion. Use of synthetic amylin that is called symlin in treatment of diabetics type 1 may reduce amount of needed insulin even by 50% [3]. Furthermore there are many other factors affecting natural physiological glucose level control, for example neural signals. The function of closed-loop BG level control system based solely on insulin delivery is closer to just eta-Cell working itself than the whole pancreas. Idea of such device is shown in Fig. 2. A healthy pancreas begins secreting insulin when a person smells or thinks about food [4], but its artificial counterpart does it when senses that BG level is rising. The consequence of this approach is that we will be always a step behind the right glucose level, but even if we only manage to reduce events of hyper and hypoglycemia it be a success. The problem is complex but preliminary tests shows that this kind of treatment improves diabetes control [5]. A closed-loop insulin delivery system consists of a three 196

Articles

N° 4

2009

blocks, which are: (i) IDD Insulin Dispensing Device, (ii) CGM Continuous Glucose Monitoring, (iii) Algorithm. insulin in

glucose in

IDD

Food (Carbohydrates)

Algorithm

Patient BG level CGM

Closed-loop insulin delivery system

Glucose demand glucose out

Fig. 2. The closed-loop insulin delivery system diagram. 3.1. Insulin Dispensing Device Insulin Dispensing Device (IDD) dose insulin into patients' blood. It may be a standard insulin pump that is described in sections above. There are no technical problems to overcome with the insulin pump itself, but there is an issue about subcutaneous insulin delivery, which causes delays in absorption. 3.2. Continuous Glucose Monitoring CGM that stands for Continuous Glucose Monitoring is responsible for continuous measuring BG levels, however it may deliver results with short intervals instead of continuous reading. Most of the glucose sensors are placed in the tissue just beneath the skin (minimal invasion sensor) and they measure glucose level not in the blood but in the interstitial fluid and there are some consequences of this approach. First and foremost is that, there is a lag time between change of glucose level in blood and in interstitial fluid. This is not a significant problem when glucose level is not changing rapidly, but it becomes significant after meals when such situation occurs. For example delay for the microdialysis sensor is 7.1 ± 5.5 min. [6] or for the subcutaneous glucose sensor the half-time response was estimated to be 4.0 ± 1.0 min. [7]. But some users report that they experience delay between 10.15 min. This imposes that in case of rapid changes of BG level the event of hyperglycemia or hypoglycemia may be read as euglycemia. Second fact is that CGM has to be calibrated with traditional BG measurement, so there is a need for use both CGM system and from time to time use of traditional glucose meter. Next problem with sensors is that their accuracy drops when measuring low BG levels. “CGM sensors experienced periods of transient loss of sensitivity, particularly during hypoglycemia, identified as sensor readings holding steady at a very low glucose value” [8]. Another point to improve glucose sensor is to make it non-invasive or implantable for long term, because patients are not satisfied with inserting another needle type device in their body. This shows that sensor design needs improvement. 3.3. Algorithm Algorithm is a link between IDD and CGM. An idea is that it will take measurement of BG from CGM and will respond with an appropriate insulin dosing instructions for IDD. Unfortunately, this is not a trivial task.


VOLUME 3,

Journal of Automation, Mobile Robotics & Intelligent Systems

The basal and bolus amount and timing needs to be set basing on BG level, but as mentioned before there is a lag time between the insulin delivery and moment when it reaches blood system and starts working, plus another lag time between change of BG level and response from CGM sensor. Algorithm has to cope with these lags. that's the main problem to overcome. Some of the researchers suggest step back from fully automated closed-loop insulin delivery to the hybrid approach with manual insulin delivery before meal. The result shows that it improves postprandial glycemic excursions [9]. There are two main concepts for glycemic control in the treatment of type 1 diabetes: the Proportional-Integral-Derivative (PID) system and the Model Predictive Control (MPC) system [10]. PID system is known from the industrial sector and it utilizes formula that calculates the insulin dose by proportional, integral and derivative parts of glucose error signal, where error signal is difference between actual and desired BG level. “Model Predictive Control (MPC) systems are highly adaptive methods that utilize mathematical models based on observations of biological behaviour patterns using system identification…” [10]. MPC is also referred as receding horizon control. An idea of algorithm is to solve finite horizon optimal control problem in a current state and over a fixed future interval. For calculation it takes into account current state as an initial state and also future constrains. It applies controls only for next sampling time and again recalculates controls. In effect it provides a solution for each sampling time instead of single solution for all points along the timeline. This allows for parameter optimisation while algorithm is running. “In sillico study effectively highlighted the fact that MPC systems improved glucose regulation over classical PID systems in limiting the oscillation of glucose levels.” [10] “The application of further mathematical models, such as fuzzy control and artificial neural networks, are also promising, but are largely clinically untested.” [10] We have some experience in using neural networks of different kind for computer aided of the medical (cardiological) diagnosis and, what's more important, for prediction individual insulin dosing for diabetics patient [11]. As mentioned before, BG level depends on many factors thus it may be worth to take into account some other parameters than BG level itself. For example, we know that body glucose demand depends (somehow) on physical activity. Physical activity itself is hard to measure, however blood pressure, heart rate and body temperature are easily measurable, and they give some information about physical activity. Neural network control is suitable for control of BG level because it can be highly individualised and trained when running. Neural networks can be tuned to person's metabolism and typical day activity by continuous body monitoring.

2009

an insulin pump and control algorithm that fully or semi automatically will choose an insulin dose. The concept of closed loop insulin delivery system is promising treatment method for persons with diabetes because it has to reduce chronic complication of diabetes mellitus potentially. But before technology will be available to utilization there is a place for improvement in stability and accuracy of CGM sensors. Problems and limitations with delayed action and BG sensing due to subcutaneous insulin delivery/glucose sensing also needs to be overcome. There is a need for reliable and safe control algorithm for a device working in closed-loop mode. We need to trace improvement of CGMs and work done to the closedloop systems. We believe that our work on control algorithm with artificial neural network will bring new information to overall design of such medical device. We have already started consultations with medical specialists to meet medical requirements of an insulin pump. From the scientific point of view, our design will enable us to inter-compare typical insulin treatments to meet individual requirements using a closed loop insulin control based on a glucose sensor and a neural network system.

AUTHORS Hubert J. Hawłas* - Institute of Metrology and Biomedical Engineering, Warsaw University of Technology, Faculty of Mechatronics, Św. Andrzeja Boboli 8, Warszawa, 02-525, Poland. E-mail: h.hawlas@mchtr.pw.edu.pl. Krzysztof Lewenstein - Institute of Metrology and Biomedical Engineering, Warsaw University of Technology, Faculty of Mechatronics, Św. Andrzeja Boboli 8, Warszawa, 02-525, Poland. E-mail: lewenk@mchtr.pw.edu.pl. * Corresponding author

References [1]

[2] [3] [4]