Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download ST_Vol_192_Sept_2015..
Document related concepts
Transcript
Sensors & Transducers International Official Journal of the International Frequency Sensor Association (IFSA) Devoted to Research and Development of Sensors and Transducers Volume 192, Issue 9, September 2015 Editor-in-Chief Prof., Dr. Sergey Y. YURISH IFSA Publishing: Barcelona Toronto Copyright 2015 IFSA Publishing, S. L. All rights reserved. This journal and the individual contributions in it are protected under copyright by IFSA Publishing, and the following terms and conditions apply to their use: Photocopying: Single photocopies of single articles may be made for personal use as allowed by national copyright laws. Permission of the Publisher and payment of a fee is required for all other photocopying, including multiple or systematic copyright, copyright for advertising or promotional purposes, resale, and all forms of document delivery. Derivative Works: Subscribers may reproduce tables of contents or prepare list of articles including abstract for internal circulation within their institutions. Permission of the Publisher is required for resale or distribution outside the institution. Permission of the Publisher is required for all other derivative works, including compilations and translations. Authors' copies of Sensors & Transducers journal and articles published in it are for personal use only. Address permissions requests to: IFSA Publisher by e-mail: [email protected] Notice: No responsibility is assumed by the Publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein. Printed in the USA. Sensors & Transducers Volume 192, Issue 9, September 2015 www.sensorsportal.com e-ISSN 1726-5479 ISSN 2306-8515 Editors-in-Chief: Professor, Dr. Sergey Y. Yurish, tel.: +34 93 4137941, e-mail: [email protected] Editors for Western Europe Editors South America Meijer, Gerard C.M., Delft Univ. of Technology, The Netherlands Ferrari, Vittorio, Universitá di Brescia, Italy Mescheder, Ulrich, Univ. of Applied Sciences, Furtwangen, Germany Costa-Felix, Rodrigo, Inmetro, Brazil Walsoe de Reca, Noemi Elisabeth, CINSO-CITEDEF UNIDEF (MINDEF-CONICET), Argentina Editor for Eastern Europe Editors for Asia Sachenko, Anatoly, Ternopil National Economic University, Ukraine Editors for North America Katz, Evgeny, Clarkson University, USA Datskos, Panos G., Oak Ridge National Laboratory, USA Fabien, J. Josse, Marquette University, USA Ohyama, Shinji, Tokyo Institute of Technology, Japan Zhengbing, Hu, Huazhong Univ. of Science and Technol., China Li, Gongfa, Wuhan Univ. of Science and Technology, China Editor for Asia-Pacific Mukhopadhyay, Subhas, Massey University, New Zealand Editor for Africa Maki K., Habib, American University in Cairo, Egypt Editorial Board Abdul Rahim, Ruzairi, Universiti Teknologi, Malaysia Abramchuk, George, Measur. Tech. & Advanced Applications, Canada Aluri, Geetha S., Globalfoundries, USA Ascoli, Giorgio, George Mason University, USA Atalay, Selcuk, Inonu University, Turkey Atghiaee, Ahmad, University of Tehran, Iran Augutis, Vygantas, Kaunas University of Technology, Lithuania Ayesh, Aladdin, De Montfort University, UK Baliga, Shankar, B., General Monitors, USA Barlingay, Ravindra, Larsen & Toubro - Technology Services, India Basu, Sukumar, Jadavpur University, India Booranawong, Apidet, Prince of Songkla University, Thailand Bousbia-Salah, Mounir, University of Annaba, Algeria Bouvet, Marcel, University of Burgundy, France Campanella, Luigi, University La Sapienza, Italy Carvalho, Vitor, Minho University, Portugal Changhai, Ru, Harbin Engineering University, China Chen, Wei, Hefei University of Technology, China Cheng-Ta, Chiang, National Chia-Yi University, Taiwan Cherstvy, Andrey, University of Potsdam, Germany Chung, Wen-Yaw, Chung Yuan Christian University, Taiwan Cortes, Camilo A., Universidad Nacional de Colombia, Colombia D'Amico, Arnaldo, Università di Tor Vergata, Italy De Stefano, Luca, Institute for Microelectronics and Microsystem, Italy Ding, Jianning, Changzhou University, China Djordjevich, Alexandar, City University of Hong Kong, Hong Kong Donato, Nicola, University of Messina, Italy Dong, Feng, Tianjin University, China Erkmen, Aydan M., Middle East Technical University, Turkey Fezari, Mohamed, Badji Mokhtar Annaba University, Algeria Gaura, Elena, Coventry University, UK Gole, James, Georgia Institute of Technology, USA Gong, Hao, National University of Singapore, Singapore Gonzalez de la Rosa, Juan Jose, University of Cadiz, Spain Goswami, Amarjyoti, Kaziranga University, India Guillet, Bruno, University of Caen, France Hadjiloucas, Sillas, The University of Reading, UK Hao, Shiying, Michigan State University, USA Hui, David, University of New Orleans, USA Jaffrezic-Renault, Nicole, Claude Bernard University Lyon 1, France Jamil, Mohammad, Qatar University, Qatar Kaniusas, Eugenijus, Vienna University of Technology, Austria Kim, Min Young, Kyungpook National University, Korea Kumar, Arun, University of Delaware, USA Lay-Ekuakille, Aime, University of Lecce, Italy Li, Fengyuan, HARMAN International, USA Li, Jingsong, Anhui University, China Li, Si, GE Global Research Center, USA Lin, Paul, Cleveland State University, USA Liu, Aihua, Chinese Academy of Sciences, China Liu, Chenglian, Long Yan University, China Liu, Fei, City College of New York, USA Mahadi, Muhammad, University Tun Hussein Onn Malaysia, Malaysia Mansor, Muhammad Naufal, University Malaysia Perlis, Malaysia Marquez, Alfredo, Centro de Investigacion en Materiales Avanzados, Mexico Mishra, Vivekanand, National Institute of Technology, India Moghavvemi, Mahmoud, University of Malaya, Malaysia Morello, Rosario, University "Mediterranea" of Reggio Calabria, Italy Mulla, Imtiaz Sirajuddin, National Chemical Laboratory, Pune, India Nabok, Aleksey, Sheffield Hallam University, UK Neshkova, Milka, Bulgarian Academy of Sciences, Bulgaria Passaro, Vittorio M. N., Politecnico di Bari, Italy Patil, Devidas Ramrao, R. L. College, Parola, India Penza, Michele, ENEA, Italy Pereira, Jose Miguel, Instituto Politecnico de Setebal, Portugal Pillarisetti, Anand, Sensata Technologies Inc, USA Pogacnik, Lea, University of Ljubljana, Slovenia Pullini, Daniele, Centro Ricerche FIAT, Italy Qiu, Liang, Avago Technologies, USA Reig, Candid, University of Valencia, Spain Restivo, Maria Teresa, University of Porto, Portugal Rodríguez Martínez, Angel, Universidad Politécnica de Cataluña, Spain Sadana, Ajit, University of Mississippi, USA Sadeghian Marnani, Hamed, TU Delft, The Netherlands Sapozhnikova, Ksenia, D. I. Mendeleyev Institute for Metrology, Russia Singhal, Subodh Kumar, National Physical Laboratory, India Shah, Kriyang, La Trobe University, Australia Shi, Wendian, California Institute of Technology, USA Shmaliy, Yuriy, Guanajuato University, Mexico Song, Xu, An Yang Normal University, China Srivastava, Arvind K., Systron Donner Inertial, USA Stefanescu, Dan Mihai, Romanian Measurement Society, Romania Sumriddetchkajorn, Sarun, Nat. Electr. & Comp. Tech. Center, Thailand Sun, Zhiqiang, Central South University, China Sysoev, Victor, Saratov State Technical University, Russia Thirunavukkarasu, I., Manipal University Karnataka, India Thomas, Sadiq, Heriot Watt University, Edinburgh, UK Tian, Lei, Xidian University, China Tianxing, Chu, Research Center for Surveying & Mapping, Beijing, China Vanga, Kumar L., ePack, Inc., USA Vazquez, Carmen, Universidad Carlos III Madrid, Spain Wang, Jiangping, Xian Shiyou University, China Wang, Peng, Qualcomm Technologies, USA Wang, Zongbo, University of Kansas, USA Xu, Han, Measurement Specialties, Inc., USA Xu, Weihe, Brookhaven National Lab, USA Xue, Ning, Agiltron, Inc., USA Yang, Dongfang, National Research Council, Canada Yang, Shuang-Hua, Loughborough University, UK Yaping Dan, Harvard University, USA Yue, Xiao-Guang, Shanxi University of Chinese Traditional Medicine, China Xiao-Guang, Yue, Wuhan University of Technology, China Zakaria, Zulkarnay, University Malaysia Perlis, Malaysia Zhang, Weiping, Shanghai Jiao Tong University, China Zhang, Wenming, Shanghai Jiao Tong University, China Zhang, Yudong, Nanjing Normal University China Sensors & Transducers Journal is a peer review international journal published monthly by International Frequency Sensor Association (IFSA). Available in both: print and electronic (printable pdf) formats. Copyright © 2015 by IFSA Publishing, S. L. All rights reserved. Sensors & Transducers Journal Contents Volume 192 Issue 9 September 2015 www.sensorsportal.com ISSN 2306-8515 e-ISSN 1726-5479 Research Articles Duty-Cycle and Duty-off Factor Measurements Based on Universal Sensors and Transducers Interface (USTI-MOB) IC Sergey Y. Yurish and Javier Cañete 1 Security in Visible Light Communication: Novel Challenges and Opportunities Christian Rohner, Shahid Raza, Daniele Puccinelli, and Thiemo Voigt ............................ 9 Green Walls Utilizing Internet of Things Andrejs Bondarevs, Patrik Huss, Shaofang Gong, Ola Weister and Roger Liljedahl ........ 16 Metrological Array of Cyber-Physical Systems. Part 11. Remote Error Correction of Measuring Channel Yuriy Yatsuk, Mykola Mykyjchuk, Volodymyr Zdeb, and Roman Yanovych ...................... 22 Metrological Array of Cyber-Physical Systems. Part 12. Study of Quantum Unit of Temperature Svyatoslav Yatsyshyn, Bohdan Stadnyk ............................................................................ 30 Experimental and Modelling Study of a Piezoelectric Energy Harvester Unimorph Cantilever Arrays Almuatasim Alomari and Ashok Batra ............................................................................... 37 Determination of Multiple Spring Constants, Gaps and Pull Down Voltages in MEMS CRAB Type Microaccelerometer Using Near Pull Down Capacitance Voltage Measurements R. K. Bhan, Shaveta, Abha Panchal, Yashoda Parmar, Chandan Sharma, Ramjai Pal, Shankar Dutta ............................................................................................................. 44 Mössbauer, VSM and X-ray Diffraction Study of Fe3O4 (NP’s)/PVOH for Biosensors Applications Almuatasim Alomari, Hasan M. El Ghanem, Abdel-Fatah Lehlooh, Isam M. Arafa, Ibrahim Bsoul, Ashok Batra................................................................................................ 53 Larger Selectivity of the V2O5 Nano-particles Sensitivityto NO2 than NH3 Amos Adeleke Akande, Bonex Wakufwa Mwakikunga, Koena Erasmus Rammutla, Augusto Machatine ............................................................................................................ 61 Surface Morphology, Compositional, Optical and Electrical Properties of TiO2 Thin Films S. S. Roy, A. H. Bhuiyan .................................................................................................... 66 Non-destructive Testing of Wood Defects Based on Discriminant Analysis Method Wenshu LIN and Jinzhuo WU ............................................................................................ 74 Research on Electronic Transformer Data Synchronization Based on Interpolation Methods and Their Error Analysis Pang Fubin, Yuan Yubo, Bo Qiangsheng and Ji Jianfei ................................................. 81 Performance Characteristics of GaAs/Al0.32Ga0.68As Quantum-Well Lasers Hadjaj Fatima, Belghachi Abderrahmane, and Helmaoui Abderrachid ............................. 90 Multi-Model Adaptive Fuzzy Controller for a CSTR Process Shubham Gogoria, Tanvir Parhar, Jaganatha Pandian B. ............................................... 96 Ten Top Torque Tips. (White Paper) Bob Dobson, Tony Ingham ................................................................................................ 104 Authors are encouraged to submit article in MS Word (doc) and Acrobat (pdf) formats by e-mail: [email protected]. Please visit journal’s webpage with preparation instructions: http://www.sensorsportal.com/HTML/DIGEST/Submition.htm International Frequency Sensor Association (IFSA). 3 rd IEEE International Workshop on Metrology for Aerospace Florence, Italy, June 22 - 23, 2016 ABOUT THE WORKSHOP HONORARY CHAIR Marina Ruggieri, Italy GENERAL CHAIRS Pasquale Daponte, Italy Robert Rassa, US TECHNICAL PROGRAM CO-CHAIRS Stephen Dyer, US Marcantonio Catelani, Italy PUBLICATION CHAIR Lorenzo Ciani, Italy TREASURY CHAIR Cosimo Stallo, Italy INTERNATIONAL PROGRAM COMMITTEE Domenico Acierno, Italy Carlo Albanese, Italy Giovanni Betta, Italy Erik P. Blasch, US Paolo Carbone, Italy Luigi Carrino, Italy Goutam Chattopadhyay, US Stefano Debei, Italy Murat Efe, Turkey Pietro Ferraro, Italy Jesus Garcia, Spain Domenico Giunta, Netherlands Maria S. Greco, Italy Richard Hochberg, US Satoshi Ikezawa, Japan Stephen Johnson, US Karel Kudela, Slovak Rep. Chin E. Lin, Taiwan Walter Matta, Italy Daniele Mortari, US Aldo Napoli, France Pavel Paces, Czech Jacek Pieniazek, Poland Vasily Popovich, Russia Helena G. Ramos, Portugal Artur L. Ribeiro, Portugal Roberto Sabatini, Australia Nicolas Sklavos, Greece Patrizia Tavella, Italy Fabrizio Francesco Vinaccia, Italy Graham Wild, Australia Ruqiang Yan, P.R. China Ho-Soon Yang, Republic of Korea Mark Yeary, University of Oklahoma, US David Zucconi, Italy LOCAL COMMITTEE Carlo Carobbi, Italy Stefano Manetti, Italy Alberto Reatti, Italy Roberto Singuaroli, Italy Matteo Venzi, Italy Andrea Zanobini, Italy Judy Scharmann (IEEE AESS Executive Assistant) LOCAL ARRANGEMENTS MetroAeroSpace aims to gather people who work in developing instrumentation and measurement methods for aerospace. Attention is paid, but not limited to, new technology for metrology-assisted production in aerospace industry, aircraft component measurement, sensors and associated signal conditioning for aerospace, and calibration methods for electronic test and measurement for aerospace. WORKSHOP TOPICS The main topics include, but are not limited to: • • • • • • • Electronic instrumentation for aerospace Automatic test equipment for aerospace Sensors and sensor systems for aerospace applications Wireless sensor networks in aerospace Attitude - and heading - reference systems Monitoring systems in aerospace Metrology for navigation and precise positioning PAPER SUBMISSION Paper submission will be handled electronically, through the submission page set up on the conference web page: www.metroaerospace.org The best contributions will be awarded, including the Best Student Paper Award and the Best Paper authored and presented by a woman. Special sessions will be organized on specific topics, see online at: www.metroaerospace.org/index.php/program/special-session IMPORTANT DATES January 22, 2016 – Submission of Extended Abstract April 15, 2016 – Notification of Acceptance May 22, 2016 – Submission of Final Paper With the endorsement of FLORENCE Cradle of the Renaissance and appreciated by millions of tourists, Florence has been UNESCO World Heritage Site since 1982. CONTACT US www.metroaerospace.org [email protected] Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 1-8 Sensors & Transducers © 2015 by IFSA Publishing, S. L. http://www.sensorsportal.com Duty-Cycle and Duty-off Factor Measurements Based on Universal Sensors and Transducers Interface (USTI-MOB) IC * Sergey Y. YURISH and Javier CAÑETE Excelera, S. L., Parc UPC-PMT, Edificio RDIT-K2M, c/ Esteve Terradas, 1, 08860, Castelldefels, Barcelona, Spain Tel.: +34 93 4137941 E-mail: [email protected], [email protected], Received: 15 July 2015 /Accepted: 31 August 2015 /Published: 30 September 2015 Abstract: An experimental investigation of metrological characteristics of designed Universal Sensors and Transducers Interface (USTI-MOB) integrated circuit working in duty-cycle and duty-off factor measuring modes is described in the article. The USTI-MOB is based on the novel patented methods for duty-cycle – to – digital conversion. Experiments have confirmed the high metrological performance at low power consumption (0.35 mA current consumption at Vcc = 1.8 V). So, the relative error of duty-cycle – to – code conversion is changed from ±0.08 to ±1.00 % in all specified measuring range of USTI-MOB. Metrological characteristics and functionality make the USTI-MOB very suitable for various sensor systems designs, based on duty-cycle output sensors. In this case the significant time-to-marked reduction and design simplification can be achieved. Copyright © 2015 IFSA Publishing, S. L. Keywords: Universal Sensors and Transducers Interface, USTI-MOB, Duty-cycle measurement, Duty-off factor. 1. Introduction The duty cycle (D.C.) is the ratio between the pulse duration tp and the period Tx of a rectangular waveform (Fig. 1): D.C. = tp Tx ×100 % (1) The physical meaning of duty-cycle is the percentage of one period in which a signal is active. The value, reciprocal to the duty cycle is called 'dutyoff factor': K off = T 1 = x D.C. t p http://www.sensorsportal.com/HTML/DIGEST/P_2700.htm (2) Sometimes the duty-off factor is called 'period-topulse duration ratio' or 'relative pulse duration'. Several sensors' and microcontrollers' manufacturers mistakenly use the “duty-cycle output” term instead of “pulse-width modulated (PWM) output”. In the last one, the information parameter is a ratio between pulse width tp and pulse space ts (tp/ts or ts/tp) but not the ratio between period Tx and tp as for the dutycycle (1). The difference between duty-cycle and PWM informative parameters are shown in Fig. 1. The duty-cycle of pulse signal (D.C.) is widely used as informative parameter of sensors' outputs and in various measuring and DAQ systems. For example, accelerometers from Analog Devices, Kionix and MEMSIC; temperature sensors from Smartec (The Netherlands), magnetic field Hall effect sensor HAL810 from Micronas, Hall Effect Differential Gear Tooth Sensors CYGTS101DC-S 1 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 1-8 from ChenYang-Technologies GmbH & Co. KG, and others [1-6]. The duty cycle of the output signal in such sensors is related to the measurand, the measuring value can be easily computed by means of measuring a duty cycle. In comparison with an analog sensor output signal and even with frequency sensor output signal, the duty-cycle is rather immune to interfering signals, such as spikes [7], and the ratio does not depend on the absolute value of any component [8]. cycle and duty-off factor is described in Section II. The obtained experimental results of measurements are provided and discussed in Section III. The Section IV describes two cases of study: an example of temperature sensor system based on duty-cycle output temperature sensors SMT16030, SMT172 (Smartec); and accelerometer sensor system. The article is concluded in the last section. 2. Method for Duty-Cycle and Duty-off Factor Measurements tp Tx D.C.= tp/Tx tp ts PWM: tp/ts or : ts/tp Fig. 1. Difference between duty-cycle and PWM informative parameters. The D.C. as an informative parameter are also used in different interfacing and readout circuits. So, in the ASIC front-end interface for resistive-bridge sensors based on a relaxation oscillator with frequency and duty cycle output, the D.C. depends on the overall bridge resistance and used as an informative parameter related to the sensor temperature [9]. A capacitive sensor readout circuit that converts capacitance changes of a sensor element to changes of the duty-cycle of a square-wave oscillator is described in [10]. It has achieved a performance of 13-bit effective resolution with a 1-kHz bandwidth. A low-voltage CMOS on-chip design of such readout circuit may also create opportunities for a low-power consumption of the readout circuit [10]. Due to its simplicity and low number of components, the power consumption of the circuit is expected to be significantly smaller than in similar tradition analog readout designs [10]. The designed by authors Universal Sensors and Transducers Interface (USTI-MOB) IC for low power consumption applications contains appropriate measuring modes for duty-cycle, duty-off factor and PWM parameters (pulse width and pulse space) [11]. The aim of this research was to determine the metrological characteristics of designed USTI-MOB at duty-cycle and duty-off factor measurements. The article is organized as follow. In Section I the method for duty-cycle and duty-off factor measurements are described in short. The experimental set-up for duty- 2 Various methods exist to measure the duty-cycle of pulse signal. For example, some simple duty-cycle - to - digital can be based on the classical approach: to measure the pulse width tp and period Tx of signal, then calculate the ratio according to equation (1) and (2). Main error’s components are quantization errors at pulse width and period measurements. Both components can be big enough. If a high accuracy is needed, a very high clock frequency should be used. For example, the approaches described in [12] and [13] are based on the 33 MHz MCS-51 and 16 MHz Microchip types of microcontrollers. Such high clock frequencies are not suitable for applications with the low power consumption, including mobile devices (smart phones and tablets) and IoT. Another approach to measure a duty-cycle is to take random samples of a digital signal (randomsampling method) [14]. The method can be realized very easy by a program-oriented way. But this method is suitable only for low-resolution conversions for which the necessary resolution is a maximum of 9 bits. The method of reading the time-domain sensor signals is described in [15]. It can eliminate the part of quantization error without increase of clock frequency. The method uses the internal clock frequency as 2N times of the signal frequency. So, it means that the period Tx is not changed with the sensor output signal. However, very often, the frequency (period) of signal is changing. In this case this method cannot be used. The Vernier-type method for duty-cycle measurement is described in [16]. The method using two phase-locked loops (PLLs), developed to emulate the Vernier caliper to measure the duty cycle. The method lets to minimize the measuring error and obtain a higher resolution without increasing the clock frequency. But the main unit - the Vernier Caliper Emulator (VCE), needs two different clock frequencies. Such block has a relatively high power consumption. In addition, the VCE must be calibrated by a calibration signal with a known duty cycle and frequency [16]. Sometime, duty-cycle output sensors’ manufactures recommend to convert duty-cycle in voltage, and then, voltage-to-digital by an ADC [17]. Such solution introduces addition component of error due to these two-stage conversion. This error is much Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 1-8 bigger in comparison with the error, which can be achieved at direct duty-cycle – to – digital conversion. The USTI-MOB is based on novel, patented method for duty-cycle and duty-off factor measurement. The method is based on the determination of average pulse width and average period during the conversion time Tq. The last one is determined by the beforehand given quantization error δ for period measurement and equals to the integer number of periods NTx. During this time and each of pulse widths, the pulses of the reference frequency are counted. At the end of conversion time the duty cycle is calculated according to the following equation: _ N_ D.C. = Tx N_ = tp __ (3) Tx x and duty-off factor is calculated according to the formula: K off = N_ Tx N_ tp _ = Tx __ t (4) M04 S C R ; Select phase shift measurement mode ; Start measurement ; Check result status: ‘r’ if ready or ‘b if busy ; Get result in BCD ASCII format Fig. 2. Commands for RS232 communication modes at duty-cycle measurements. p The duty-cycle (and duty-off factor) measurement contains two main components of error: the error due to period and error due to pulse width measurements. The first one can be eliminated due to the described above method for the duty-cycle measurement, which is used in the USTI-MOB. The second component (relative quantization error) can be calculated in the worst case (one period Tx) according to the following equation: δq = The supply voltage of the evaluation board was +14 V dc, provided by the Promax FA-851 power supply. The duty-cycle and duty-off factor of signals generated by the waveform generator were measured by both: the USTI-MOB and Universal Frequency Counter/Timer Agilent 53220A with the ultra high oven stability internal time base. The digital oscilloscope Promax OD-591 monitored the signal's waveforms. Before measurements, the USTI-MOB was calibrated in the working temperature range: +24.4 oC at 45-47 % RH. The measurands were sent to a PC via an RS232 interface implemented with the ST202D IC. The user interface was realized with the help of terminal software Terminal V1.9b running under Windows XP or Windows 7 operation systems. The commands of RS232 communication mode for duty-cycle measurements in the 1st USTI-MOB channel are shown in Fig. 2. 1 ×100 % 4 ×106 × t p (5) 3. Measurement Technique and Experimental Set-Up The diagram of experimental measurement set-up for the USTI-MOB working in duty cycle and dutyoff factor measuring modes is the same as it is shown in [18]. The circuit diagram is similar to the USTI circuit diagram of connection [19]. The difference is only in the voltage of power supply Vcc: +1.8 V for the USTI-MOB and +5 V for the USTI. A square waveform pulse signal whose duty-cycle and duty-off factor must be measured, was fed from the first channel of Waveform Generator Agilent 33500B to inputs FX1, ST1 and FX2, ST2 (the 1st channel of IC) of the USTI-MOB running on a 4 MHz clock. In case of duty-off factor measurement mode the first command must be changed to 'M05'. The dutycycle and duty-off factor can be measured also in the 2nd USTI-MOB channel. It this case it is necessary to use commands 'M14' or 'M15' respectively. Every measurement were consisted of 60 values (sample size). The measurement errors were evaluated from appropriate statistics with the help of NUMERI software [20]. The Waveform Generator Agilent 33500B has the high-stability OCXO timebase (frequency reference ±0.1 ppm of setting ±15 pHz) [21]. The Universal Frequency Counter/Timer Agilent 53220A-010 has the ultra high-stability OCXO timebase (±50 ppb) [22]. 4. Experimental Results The D.C. range of measurements is dependent on the frequency of input signal. The maximal possible signal frequency for the D.C. measurement by the USTI-MOB is 100 kHz, and duty-cycle can be measured in the narrow range D.C. ∼50 %. At 500 Hz frequency, the D.C. can be measured in the wide range from 1 to 99 %. The possible dutycycle values vs. frequencies are shown in Table 1. In the experimental investigation, 50 % D.C. was measured at 500 Hz and 100 kHz of input signal 60 times each. The conventional true value for dutycycle measurement (D.C.=50 %) is shown in Fig. 2. 3 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 1-8 Table 1. Duty-cycle values vs. frequencies. Frequency, kHz < 0.5 1 10 20 30 > 40 Duty-cycle, % 1 ... 99.3 1.5 ... 98 15 ... 80 30 ... 71 46 ... 60 50 theoretical frequency distribution for data from a normal, uniform or exponential population. If S < χ2 max, where S is the sum of deviations between the dataset and the assumed distribution, and χ2 max is the maximum possible allowable deviation in the χ2 distribution, the hypothesis of appropriate distribution can be accepted [20]. The χ2 test has been used at 95 % confidence and the number of intervals grouping of experimental data for histograms were from 3 to 6 [23]. Fig. 2. Conventional true value for duty-cycle measurement (D.C. = 50 % at 500 Hz). Fig. 5. Relative errors for duty-cycle measurements of 500 Hz input signal. The experimental results of duty-cycle measurements are shown in Fig.3-5, and statistical characteristics - in Table 2. Table 2. Statistical characteristics of duty-cycle D.C. measurements at 100 kHz and 500 Hz input signal. Parameter Fig. 3. Results of measurements for duty-cycle of 100 kHz input signal. Number of measurements, n Minimal D.C., (min) Maximal D.C., (max) Sampling Range, D.C., (max) - (min) Median D.C. Arithmetic Mean, D.C. Variance D.C. Standard Deviation D.C. Coefficient of Variation D.C. Confidence Interval at probability P=95 % Maximal Relative Error, δx % Distribution low: - uniform - normal Fig. 4. Results of measurements for duty-cycle of 500 Hz input signal. The χ2 test for goodness of fit test was applied to investigate the significance of the differences between observed data in the histograms and the 4 - exponential 100 kHz 500 Hz 60 60 0.4975 0.5004 0.5129 0.5007 0.0154 0.0004 0 0 0.5049 0.50055 2.4E-0005 2.3E-0008 0.0049 0.0002 102.7083 3334.0845 D.C.∈ [0.5037 ÷ 0.5061] D.C. ∈ [0.5005 ÷ 0.5006] ≤ ±1.00 ≤ ±0.08 χ2 S<> S=7.67 < χ2 S=117.8> χ211 =12 (accepted) (rejected) S=23,69 > S= 162.34 > χ2=9.4 χ2 =7.8 (rejected) (rejected) S= 53066871 > S=4480.89 > χ2= 11 χ2= 9.4 (rejected) (rejected) As it is visible from the Table 1, the relative error is changed from ±0.08 to ±1.00 % in all specified Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 1-8 measuring range of USTI-MOB at duty-cycle measurements. Such metrological characteristics are very well suitable for many sensors applications. 5. Cases Study 5.1 Low Power Consumption Temperature Sensor System According to a new market research report 'Temperature Sensor Market, A Study of major Sensor types (ICs, Thermostat, Thermistor, Resistive Temperature Detectors (RTDs), Thermocouple) & Applications, Global Forecast & Analysis 2011 – 2016', the market size of temperature sensors in the year 2010 was $3.27 billion and is expected to reach $4.51 billion units by 2016, at an estimated CAGR of 5.6 % and $6.05 billion by 2020, growing at a CAGR of 5.11 % between 2014 and 2020 [24]. Temperature sensor utilizes digital technology, which means better in efficiency and sensing performance. Temperature sensors have a significant place in different industry verticals. The major applications of temperature sensors are in petrochemical industry, automotive industry, consumer electronics industry, metal industries, food and beverages industry, and healthcare. The emerging applications of temperature sensor in aerospace and defense industry such as temperature stabilization in satellites and Heat Ventilation Automation and Control (HVAC), have fueled the growth of this market [24]. The demand for reliable, high performance and low cost sensors is increasing leading to the development of microtechnology and nanotechnology, offering opportunities like miniaturization, low power consumption, mass production, etc. The designed low power consumption temperature sensor systems consists of two dutycycle output sensors SMT16030 (Fig. 6) or SMT172 from Smartec [3, 4] and USTI-MOB IC controlled by a microcontroller (in case of sensor systems or smart, digital sensors) or by PC (in case of DAQ systems), Fig. 7. Fig. 7. Low power consumption temperature sensor systems. These temperature sensors are silicon sensors with duty-cycle outputs with linear responses to temperatures -45 0C ... + 130 0C. In applications where multiple sensors are used (more than two), easy multiplexing can be obtained by using a low cost digital multiplexers. The USTI-MOB can work also in so-called master communication mode. In this case no any external microcontroller or PC are necessary. The measuring mode can be selected by the external jumpers, and the USTI-MOD IC continuously forms result of measurement on its RS232 bus at 2800 baud rate. The temperature sensor SMT160-30 has dutycycle changes in an output signal from 10.85 % to 93 % at 1-4 kHz [3] (Fig. 8). Fig. 8. Temperature sensor SMT16030’s output at 24.4 0C: 44.21 % duty-cycle, 2 964.66 Hz at Vcc=4.97 V. Fig. 6. Temperature sensors SMT16030 from Smartec (The Netherlands) on investigation board. The sensor SMT172 has the same range of dutycycle but at 0.5 - 7 kHz (the frequency range is the same as in SMT16030 sensor for Vcc = 4.7-5.5 V) [4]. The highest frequency in both sensors is achieved at lower supply voltage and in the middle of temperature range. In general, the duty cycle of the both output sensors' signals is defined by a linear equation [3, 4]: 5 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 1-8 D.C. = 0.320 + 0.00470 × t , (6) where t is the temperature in 0C. Temperature is then derived from the measured duty cycle. The maximal total accuracy of SMT16030 temperature sensor in TO220 package and in the temperature range from -45 0C to +130 0C is ±1.7 0C. It means, that the duty-cycle on sensor's output can be changed from 0.4347 (absolute error Δt = 0) to 0.4428 (absolute error Δt =±1.7 0C) at +24.4 0C, for example: distributed according to the triangular (Simpson’s) because of the main component of δIC is the quantization error, the total mean root square error of temperature sensor system can be calculate as [25, 26]: 2 2 δ δ σ syst = σ D.C .2 + σ IC 2 = D.C . + IC ≈ 2.3 6 (9) ≈ 0.812 + 0.0612 ≈ ±0.82 % D.C. = 0.320 + 0.0047 × 24.4 = 0.4347 D.C. = 0.320 + 0.0047 × (24.4 + 1.7) = = 0.4428 (7) The relative error for D.C measurement can be calculated as: Δ D.C. = D.C. 0.4428 − 0.4347 =± ×100 % = 0.4347 0.0081 =± ×100 % = ± 1.86 % 0.4347 δ D.C . = ± (8) The average, total relative error of USTI-MOB at duty-cycle measurement for +24.4 0C obtained from the experimental investigation is δIC = ± 0.149 % (Fig. 9 and Fig. 10). The USTI-MOB has been preliminary calibrated at the same temperature and 48 % RH in order to eliminate the quartz crystal’s systematic error and short time temperature instability. The small positive trends (dashed lines) observed in both of cases are due to so-called sensor’s self-heating effect. Fig. 9. Duty-cycle values at 60 measurements for the temperature +24.4 0C. Taking into account many components of error for temperature sensor’s relative error δD.C. which is distributed according to the Gaussian distribution low, and the USTI-MOB’s relative error δIC 6 Fig. 10. The relative error of duty-cycle measurement. In practice, the relative error is more convenient in comparison with the mean root square error, so, it is expediently to calculate the following: δ syst = σ syst × 2.3 ≈ 0.82 × 2.3 ≈ 1.89 % (10) Clear, the error’s component δIC can be neglected in comparison with the δD.C. component, because of the δIC is in one order (and even more) less than δD.C. [23]. So, in this case the error of the temperature sensor system δsyst is determinated only by the sensor’s error itself, and USTI-MOB does not introduce the significant error into the measuring channel. In order to increase the accuracy, the temperature sensor in TO18, HEC or SOIC-8 packages should be used [3]. The new temperature sensor SMT172 from Smatrec has the maximal absolute error for TO-18 package ±0.8 0C in the same temperature range from -45 0C to +130 0C (for other packages the absolute error will be ±1.0 0C) [4]. In this case the relative error calculated by the same manner as in (8) will be ± 0.87 %. In order to decrease the absolute error to ±0.1 0C the following second order equation must be used [4]: T = −2.42 × D.C.2 + 215.63 × D.C. − 68.83 (9) In this case it is recommended to use the USTI IC [18] for accurate duty-cycle measurement instead of Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 1-8 the USTI-MOB IC. The USTI has also the extended rage of frequencies at D.C. measurements (up to 625 kHz), but the increased power consumption (11 mA in comparison with 0.35 mA in the active mode). 5.2 Accelerometers Dual axis, low cost accelerometer fabricated on a monolithic CMOS IC MXD2125 (MEMSIC) provides two outputs that are set to 50% duty cycle at zero g acceleration [27]. It measures acceleration with a full-scale range of ±2 g and a sensitivity of 12.5 %/g. It can measure both dynamic acceleration (e.g. vibration) and static acceleration (e.g. gravity). The duty-cycle outputs are proportional to acceleration: A( g ) = (t p / Tx ) − 0.5 20 % (10) This device is offered from the factory programmed to either a 10 ms period (100 Hz) or a 2.5 ms period (400 Hz). The sensor can be directly interfaced to the USTI-MOB IC. The accelerometer sensor systems is shown in Fig. 11. Fig. 11. Accelerometer sensor systems. Taking into account the low frequency output signal, the USTI-MOB can measure duty-cycles in its two channels with low relative error: < ±0.08 %, which can be neglected in comparison with the accelerometer's error. The dual axis accelerometers MXD2020E, MXD6025 and MXD6125 (MEMSIC); and ADXL202, ADXL210, ADXL212 and ADXL213 (Analog Devices) can be also connected to the USTI-MOB by the same way, as it is shown in Fig. 8. The appropriate equations to calculate acceleration from duty-cycle for accelerometers from Analog Devices, Inc. are shown in Table. 3. Table 3. Equations for acceleration calculation. Accelerometer Equation ADXL202 A( g ) = ADXL210 A( g ) = ADXL212 A( g ) = ADXL213 A( g ) = (t p / Tx ) − 0.5 12.5 % (t p / Tx ) − 0.5 4% (t p / Tx ) − 0.5 12.5 % (t p / Tx ) − 0.5 30 % Ref. [28] [29] [30] [31] Selectable bandwidths for these accelerometer let to use a reasonable frequency for application with the USTI-MOB. 6. Conclusions The experimental investigation of the designed USTI-MOB integrated circuit working in the dutycycle and duty-off factor measuring modes confirms its high metrological characteristics at low power consumption (0.35 mA current consumption at Vcc = 1.8 V). The relative error of duty-cycle – to – code conversion is changed from ±0.08 to ±1.00 % in all specified measuring range of USTI-MOB at dutycycle measurements. Metrological performances (relative error and frequency range) can be improved in four times if to use the USTI IC [1, 19] instead of the USTI-MOB, if the power consumption is not a critical parameter at the design. The optimal trade-off between accuracy, power consumption and communication speed has achieved. It makes the USTI-MOB suitable for application in various duty-cycle output sensors such as temperature sensors, accelerometers, magnetic sensors, Hall Effect gear tooth sensors etc. to produce smart, digital output sensors, DAQ systems or sensor systems. The significant time-to-market reduction will be achieved in such design approach. The USTI-MOB IC will be introduced on the modern market at the end of 2015 by Technology Assistance BCNA 2010, S .L. (Excelera), Barcelona, Spain (http://www.excelera.io). References [1]. S. Y. Yurish, Digital Sensors and Sensor Systems: Practical Design, IFSA Publishing, 2011. [2]. Sensors Web Portal (http://www.sensorsportal.com). [3]. SMT16030 Digital Temperature Sensor, Datasheet, Smartec, March 2015. [4]. SMT172, Datasheet, Smartec, June 2015. 7 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 1-8 [5]. G. de Graaf, R. F. Wolffenbuttel, Light-to-Frequency Converter Using Integrated Mode Photodiodes, in Proceedings of IMTC’96, 4-6 June, 1996, Brussels, Belgium, pp.1072-1075. [6]. D. Hernandez, R. Amador, I. León, K. Kohlhof, Constant temperature anemometer with duty-cycle output conversion, in Proceedings of the IX Workshop IBERCHIP 2003, Habana, Cuba, March 2003. [7]. Gerard C. M. Meijer, Concepts and Focus Point for Intelligent Sensor Systems, Sensors and Actuators A, Vol. 41-42, 1994, pp.183-191. [8]. S. Middelhoek, P. J. French, J. H. Huijsing and W. J. Lian, Sensors with Digital or Frequency Output, Sensors and Actuators, Vol. 15, 1988, pp. 119-133. [9]. V. Ferrari , A. Ghisla, Zs. Kovács Vajna, D. Marioli, A. Taroni, ASIC front-end interface with frequency and duty cycle output for resistive-bridge sensors, Sensors and Actuators A, Vol. 138, 2007, pp. 112–119. [10]. Zeljko Ignjatovic, Mark F. Bocko, An Interface Circuit for Measuring Capacitance Changes Based Upon Capacitance-to-Duty Cycle (CDC) Converter, IEEE Sensors Journal, Vol. 5, No. 3, June 2005, pp. 403-410. [11]. J. Cañete, S. Y. Yurish, Sensors Systems for Smartphones, Tablets and IoT: New Advanced Design Approach, Sensors & Transducers, Vol. 187, Issue 4, April 2015, pp. 1-9. [12]. P. C. de Jong and F. N. Toth, Measuring Duty Cycles with an Intel MCS-51 Microcontroller, Available online at http://www.smartec.nl/pdf/appsmt01.pdf [13]. J. Bauer, Various Solutions for Calculating a Pulse and Duty Cycle, AN1473, Microchip Technology, Inc., 2012. [14]. J. Vuori, Simple Method Measures Duty Cycle, EDN Magazine, March 3, 1997. [15]. G. Chao, G. C. M. Meijer, A Novel Method of Reading the Time-Domain Sensor Signals, in Proceedings of ProRISC, November 29-30, 2001, Veldhoven, The Netherlands. [16]. S. S. Huang and M. S. Young, Method for Designing a Temperature Measurement System Using Two Phase-locked Loops, Review of Scientific Instruments, Vol. 74, No. 8, August 2003, pp. 3826 - 3831. [17]. Eric Jacobsen, Designing a Homemade Digital Output for Analog Voltage Output Sensors, [18]. [19]. [20]. [21]. [22]. [23]. [24]. [25]. [26]. [27]. [28]. [29]. [30]. [31]. Application Note AN1586, Freescale Semiconductor, 2006. S. Y. Yurish and J. Cañete, Universal Sensors and Transducers Interface for Mobile Devices: Metrological Characteristics, Sensors & Transducers, Vol. 188, Issue 5, May 2015, pp. 15-25. Universal Sensors and Transducers Interface (USTI), Specification and Application Note, Technology Assistance BCNA 2010, S. L. (Excelera), 2010. E. Schrufer, Signal Processing: Digital Signal Processing of Discrete Signals, Lybid', Kiev, 1992, (in Ukrainian). 33500B Series Waveform Generators, Data Sheet, Agilent Technologies, Inc., USA, 2012. Agilent 53200A Series RF/Universal Frequency Counter/Timers, Data Sheet, Agilent Technologies, Inc., USA, 2010. P. V. Novitskiy, I. A. Zograf, Errors Estimation for Measuring Results, Energoatomizdat, Leningrad, 1991 (in Russian). Global Temperature Sensor Market worth $4.51 Billion by 2016, MarketsandMarkets, November 2014. N. V. Kirianaki, S. Y. Yurish, N. O. Shpak, V. P. Deynega, Data Acquisition and Signal Processing for Smart Sensors, John Wiley & Sons, UK, Chichester, 2001. Sergey Y Yurish, Ferran Reverter, Ramon PallasAreny, Measurement error analysis and uncertainty reduction for period- and time-interval-to-digital converters based on microcontrollers, Measurement Science and Technology, 16, 2005, pp. 1660–1666. Improved, Ultra Low Noise ±2g Dual Axis Accelerometer with Digital Outputs, MXD2125GL/HL, MXD2125ML/NL, MEMSIC, Inc., USA, 2004. Low Cost 62 g/610 g Dual Axis iMEMS® Accelerometers with Digital Output, ADXL202/ADXL210, Rev. B, Analog Devices, Inc., 1999. Dual-axis Accelerometer Evaluation Board ADXL210EB, Analog Devices, Inc., USA, 2003. Precision ±2 g Dual Axis, PWM Output Accelerometer ADXL212, Analog Devices, Inc., 2011. Low Cost ±1.2 g Dual Axis Accelerometer ADXL213, Rev. A, Analog Devices, Inc., 2010. ___________________ 2015 Copyright ©, International Frequency Sensor Association (IFSA) Publishing, S. L. All rights reserved. (http://www.sensorsportal.com) 8 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 9-15 Sensors & Transducers © 2015 by IFSA Publishing, S. L. http://www.sensorsportal.com Security in Visible Light Communication: Novel Challenges and Opportunities 1 Christian ROHNER, 2 Shahid RAZA, 3 Daniele PUCCINELLI, and 1, 2 Thiemo VOIGT 1 Uppsala University, Dept of Information Technology, Box 337, 75105 Uppsala, Sweden 2 SICS Swedish ICT, Box 1263, 164 29 Kista, Sweden 3 SUPSI, Institute for Information Systems and Networking, Via Cantonale, 6928 Manno, Switzerland 1 Tel.: +46 70 167 9361 1 E-mail: [email protected] Received: 31 July 2015 /Accepted: 31 August 2015 /Published: 30 September 2015 Abstract: As LED lighting becomes increasingly ubiquitous, Visible Light Communication is attracting the interest of academia and industry as a complement to RF as the physical layer for the Internet of Things. Aside from its much greater spectral availability compared to RF, visible light has several attractive properties that may promote its uptake: its lack of health risks, its opportunities for spatial reuse, its relative immunity to multipath fading, its lack of electromagnetic interference, and its inherently secure nature: differently from RF, light does not penetrate through walls. In this paper, we outline the security implications of Visible Light Communication, review the existing contributions to this under-explored space, and survey the research opportunities that we envision for the near future. Copyright © 2015 IFSA Publishing, S. L. Keywords: Visible light communication, Security. 1. Introduction With Visible Light Communication (VLC), visible light is employed as the transmission medium and Light Emitting Diodes (LEDs) can offer high-capacity wireless data transmission capabilities on top of the basic role as lighting devices [1]. LEDs are replacing incandescent light bulbs because of their much higher energy efficiency, superior reliability, and ever dropping price points. As LEDs become increasingly ubiquitous, VLC continues to evolve from its former role as a subfield of Optical Wireless Communication to a candidate physical layer for the Internet of Things (IoT) that attracts the attention of both academia and industry. Nowadays, VLC is primarily viewed as a complement to RF in the face of the looming spectrum crunch: as the radio spectrum becomes increasingly http://www.sensorsportal.com/HTML/DIGEST/P_2717.htm crowded, the superior spectral availability in the visible light range becomes increasingly attractive for the IoT with its billion devices that need to be networked. The bulk of the recent work on VLC has targeted the high end segments of the design space, pursuing the goal of high throughput by means of advanced modulation schemes. Until recently, Gbps range data rates had only been demonstrated with laser diodes [2]; as recently as 2014, a 3 Gb/s link has been demonstrated with a Gallium Nitride LED [3]. Increasing the throughput for visible light is also possible by Multiple-Input-Multiple-Output transceivers as discussed by Azhar, et al. [4] and O’Brien, et al. [5] whereas Komiyana, et al. [6] increase the throughput by using RGB-LEDs with multiple colors such as blue, green and red. Other 9 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 9-15 authors have explored low-end communication links between resource-constrained devices, using simple LEDs for transmission and LEDs or photodiodes for reception [7-8]. Furthermore, smartphone-based VLC between a screen and a camera has also been explored in recent years [9-11]. At the application layer we have a number of interesting approaches making use of visible light ranging from indoor localization [12-13] to underwater networking with light [14]. Localisation is a key enabler of the IoT as many IoT applications require accurate localization information. Visible light has several key properties that we review in Section II; while its spectral availability is certainly the main reason behind the growing interest in VLC, the inherent security that stems from the spatial confinement of light beams is arguably the most captivating difference with respect to RF and, quite possibly, the most underrated. In fact, at the time of writing, there are only a few studies that address security in visible light communication. Mostafa and Lutz address secure VLC link at the physical layer [15] by investigating the achievable secrecy rates for of the Gaussian wiretap channel. Zhang, et al. [16] propose a secure system for barcode-based VLC, i.e., for secure transmission between a screen and a camera. For supporting a secure data exchange, the system requires a fully duplex VLC channel. In this paper, we outline the security implications of visible light and we survey the opportunities for VLC security research that arise in the IoT realm. The remainder of the paper is organized as follows. In Section 2, we present the physical layer properties of visible light. Section 3 discusses how to secure visible light communication whereas the following Section 4 takes up security implications of visible light communication. Finally, Section 5 concludes the paper. 2. Physical Layer Properties of Visible Light VLC was already a key communication tool long before the digital revolution of the past century. Alexander Graham Bell’s photophone, patented in 1880, predated Guglielmo Marconi’s wireless radio by over 15 years before carried human speech by way of mechanically modulated sunlight. Today’s fiberoptic communications networks are based on pulsed light transmitted via glass fibers. IBM Zurich built an optical wireless system as early as the early 1980s, but the technology failed to take off owing to the lack of demand (the Internet was still in its infancy). When wireless communication took off in the 1990s, RF was the wireless medium of choice. Now that the tightly regulated RF spectrum is getting increasingly crowded, VLC is gaining appeal as a much needed alternative to RF for Internet connectivity. VLC’s attractiveness is largely due to the availability of approximately 670 THz of free unlicensed spectrum, which means that very high data rates may be achieved with VLC and, even more 10 importantly, that VLC offers a viable solution to alleviate the spectrum crunch. At the same time, the rise of VLC is also being fueled by the massive uptake of Light Emitting Diodes (LEDs), which are replacing incandescent illumination solutions due to their comparatively high energy efficiency and ever decreasing price points. The key features of VLC that are advantageous compared to RF and that make VLC an attractive infrastructure for the IoT are: • Spectral availability (10,000 times larger than RF’s with an area spectral efficiency (bits/s/m2) that is 1,000 greater [17]); • Free unlicensed spectrum; • Inherent security due to spatial confinement of light beams (light does not penetrate through walls); • Spatial reuse opportunities, also due to spatial confinement; • Immunity to multipath fading; • Due to the limited Field of View of LEDs, VLC is inherently more directional than RF, and today’s commodity hardware may be largely regarded as directional; • The properties above also enable accurate localization [12-13] and gesture recognition based on visible light [43]; • Non-line-of-sight communication is possible thanks to diffused reflection, provided that the receiver has sufficient sensitivity to detect it; • Lack of electromagnetic interference; • Lack of health risks [18]. Most of the academic research on VLC has targeted the high-end portion of the design space, focusing on the achievement of high data rates. Resource-hungry high-end VLC systems have been investigated extensively in a relatively large body of work that has focused on Physical Layer advancements [19-21]. Energy efficiency has not been treated as a first-order problem because the idea is to piggyback on solid state lighting systems so that the communication footprint is negligible compared to the overall lighting footprint. At the time of writing, the provision of Internet connectivity is the most widely cited application of VLC. Dubbed LiFi, VLC-based Internet connectivity is particularly suitable to any application that requires lots of downlink bandwidth and minimal upstream capacity, such as those video/audio download/streaming applications that are taking a massive toll on cellular capacity. The typical architecture is based on Power Line Communication systems to deliver data to light fixtures for VLC forwarding to end devices. This is a particularly advantageous way to offer Internet connectivity in locales where RF is off limits, such as airplanes, operating theaters in hospitals, and hazardous factory environments. In recent years, the huge potential of optical wireless communication for in-house networking has been practically demonstrated in the EU-funded project OMEGA, achieving rates in the order of Gbps with laser diodes [2]. Data rates of the order of hundreds of Mbps can be achieved with white LEDs Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 9-15 by way of resource-rich hardware with strong computational capabilities [22]. Due to the rising popularity of VLC, the IEEE has recently published a VLC standard for local area networks (IEEE 802.15.7) that defines the Physical and Medium Access layers for short range wireless optical communication using visible light [23] in point-to-point communication scenarios, which have been the primary target of all research efforts in this space thus far and that also present the first step towards VLC as an infrastructure for the IoT. 3. Securing Visible Light Communication In real-world Internet of Things deployments, wireless communication is usually protected against unauthorized access to the wireless medium, modification of messages, eavesdropping, and replay attacks. Authentication security services confirm the identity of an entity and grant access to the wireless medium. Confidentiality services ensure that only the participating devices understand the contents of messages. Integrity services ensure that the data is not modified while in transit. Last but not least, freshness security services validate that the received data is not a reply of previously received message but that it belongs to the current secure session. There exist three well-known security mechanisms that can be used to protect VLC: proximity-based protection, steganographic protection, and cryptographic protection. These solutions provide security in fundamentally different ways; the choice of any of these solutions for a real-world deployment depends on the application’s security requirements. 3.1. Proximity-based Protection Proximity-based protection relies on the directionality properties of visible light and the inherent confinement of light beams within enclosed spaces; these properties may be exploited to restrict the communication coverage to a specific area. Finegrained control of light characteristics can limit the flow of communication in a restricted proximity. Such a security solution is acceptable in physically protected environments that offer snoop-free line-ofsight communication. Examples of such environments are enclosed spaces such as rooms and vehicles. Cui, et al. [24] discuss some of the key issues in line-ofsight VLC system design. Ensuring a snoop-free confinement of light signal to a particular source is an open research challenge and having such guarantees offers novel applications and opportunities such as VLC-based access control. 3.2. Steganographic Protection Steganography aims to protect the communication by hiding a message within another message. A possible steganographic protection is hiding secret communication in existing illumination. Unlike cryptographic protection, stenographically protected messages do not seek attention, e.g., from the NSA, and easily pass casual scrutiny. In a typical steganographic protection scheme, the communicating end points share a secret that describes how data is concealed. Steganography mainly addresses confidentiality, but not authentication and integrity. Nevertheless, it is hard for an attacker (without knowing the shared secret) to breach integrity unless the attacker modifies the entire message and hereby also modifies the hidden message. However, if the confidentiality is compromised, the integrity is also compromised since an attacker can identify and alter the hidden message. This is not the case in cryptography. Providing steganographic protection by hiding secret light signals in existing VLC is worth investigating especially for devices that have limited processing and memory resources and cannot afford to run complex and expansive cryptographic operations. 3.3. Cryptographic Protection and Key Generation Unlike steganography, cryptography offers most security services including confidentiality (encryption/decryption), integrity (with hashing and message integrity codes), and authentication (identity validation). In the case of VLC, cryptographic protection can be applied at different layers. The IEEE 802.15.7 standard for VLC already provides confidentiality and integrity security services at the MAC layer. The security is optional and no key management is specified in 802.15.7; however, standardization efforts are being carried out in the new IEEE 802.15.9 WG to provide key management for 802.15.4 and 802.15.7. Schmid, et al. [7] provide MAC and physical layers for LED-to-LED VLC networks but propose the implementation of security at the upper layers. Modern cryptographic protection mainly relies on secret keys and all other operations are known, i.e., security through obscurity is avoided. Key management, however, is one of the hardest problems in cryptography. Solutions have been proposed that exploit the properties of wireless channels to generate keys to secure wireless links [25-26]. For instance, it is possible to exploit channel reciprocity, whereby two closely located receivers experience the same signal envelope in the absence of interference [27]. Since practical channels are never immune to interference, a technique is presented in [25] that does not require identical signal envelopes for the communicating terminals, but only matching deep fades, which are immune to reasonable levels of interference. Because the light carrier wavelength is much smaller than the area of the photodetector, VLC is immune to fast fading and is only subject to slow fading in the form of path loss and log-normal shadowing [1]. Because of VLC’s relative immunity 11 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 9-15 to multi-path fading compared to RF, the effectiveness of schemes based on channel reciprocity for VLC must be thoroughly investigated. In the case of systems using both VLC and RF, it is possible to use the radio for key generation, and then use the generated keys for VLC. 3.4. Chaffing and Winnowing In addition to the three methods explained above for VLC protection, a less known security mechanism called chaffing and winnowing [28] can also be used. It offers confidentiality and authentication services but without requiring any encryption/decryption. It uses shared key and Message Authentication Codes (MACs) to provide authentication and uses the same MACs to offer confidentiality. For confidentiality, it breaks the message into smaller packets and assigns a serial number to each packet. The sender sends the valid packets as well as chaffs (fake packets) that have a valid serial number and message format but a bogus MAC. The receiver records all the packets that have valid MACs and immediately discards the packets that have invalid MACs; this process is called winnowing. The receiver can assemble the valid packets and recover the secret message. While this technique is underused nowadays, it may be worth to investigate the use of chaffing and winnowing in VLC. Steganography and chaffing and winnowing are alternative candidates in situations where export control or other circumstances hinder the use of cryptography. 4. VLC Security: Attacks and Opportunities In this section we highlight opportunities and attacks in the context of VLC security. Opportunities arise through the use of VLC as out-of-band or sidechannel, and the physical properties of light. Attacks known from radio communication get a different flavor in VLC, mainly because of the restricted Field of View of LEDs. This includes jamming, a denial-ofservice attack that is a particular threat to missioncritical IoT systems that must deliver data timely. 4.1. Authentic Channels An interesting concept in visible light communication are visual channels enabled by the transmission between a screen and a camera. These allow users to recognize and verify the captured scene. Visual channels can be used as a secure out-of-band channel for intuitive pairing of devices using twodimensional barcodes, displayed by (or affixed to) at least one of the devices. The barcode represents 1 a.k.a. multi-factor authentication 12 security-relevant information that can be read visually by a camera-equipped device and is used to set up an authenticated channel. Visual channel are considered resilient against active attacks such as man-in-the-middle attacks, and have the property that active attacks are easily detected by the user. The idea of encoding cryptographic information into barcodes was first proposed by Hanna [29] as well as Gehrmann, et al. [30]. This work has be generalized into the concept of visual channels by McCune in his work ‘Seeing-isbelieving’ [31]. Saxena, et al. [32] extends the Seeingis-believing system to achieve mutual authentication using just a unidirectional visual channel, and using visual channel authentication even on devices with limited displaying capabilities (e.g., LEDs). The ability to provide an authentic channel is unique to VLC and is not available in radio communication. 4.2. Out-of Band Channels Out-of-band channels are an important tool to establish security in general, and have been used in particular for authentication purpose [33]. For example, receiving the same (or complementary) information through independent channels imply higher probabilities for message authentication. The potential ubiquity of VLC makes it an ideal candidate to complement a radio communication channel for security purposes, for instance to distribute public keys or a fingerprint thereof to check the authenticity of key material received over the primary communication channel. 1 4.3. Multiple VLC Channels VLC scenarios often include several light sources, potentially offering multiple (out-of-band) channels. If operated interference-free and possibly directed, VLC could create zones in which subsets of the sources can be received. From a security perspective, such zones could be combined with network coding [34] or threshold secret sharing schemes [35] where T out of N linear combinations of data are needed to reconstruct it. This can be used to either increase the probabilities for message authentication (see Section IV-B), to make data only accessible in certain spatial zones, or to require the user to move around in a room to collect the necessary information to re-construct the data. Visible light has the property that the effective intensity of light is additive that is, light from different sources will add upp at the receiver. The received signal will therefore be unique for the location. Besides of being used for localization, this property has been leveraged for distance bounding [36] or key generation [37] in radio communication. Although multi-path fading and dispersion are expected to be Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 9-15 much smaller in VLC, direct on-off modulations will result in distinct timing patterns that can be used for this purpose. 4.4. Denial-of-Service Denial-of-service attacks based on jamming are relatively straightforward to perform on many wireless networks [38]. In particular, low-power radios are notoriously easy to jam even without sophisticated hardware support [39]. There exist approaches to guard low-power radio networks from malicious traffic. In [40], for instance, a central unit detects and corrupts malicious packets so they are not accepted by the unit under attack. Note that this approach, however, can only detect jamming without preventing it [41]. Another option is to detect and map jammed areas to reroute the traffic around these areas [42], but this is only applicable for larger networks. As discussed in Section II today’s VLC can be regarded as directional which makes it easier to defend against the equivalent of jamming attacks on lowpower radios. Fig. 1 presents a scenario where an attacker tries to disturb the sink node from receiving a packet. Note that in this scenario we assume that the attacker uses a directional light source. Furthermore, we assume the attacker knows the position of the node against which it launches the attack, and is therefore able to aim the light beam accurately. Jamming attacks on low-power radios do not need such information and are hence easier to launch. Once the attack is detected, the node under attack could physically shield itself from the attack and a multi-hop visible light network2 could reroute to deliver information via other nodes to the intended sink as shown in Fig. 1. In networks where transmissions are less directional as is the case for most RF communication that often use omnidirectional antennas, shielding in a similar manner would be much more difficult. While in the discussion above we make use of transmitter’s directionality to defend against denialof-service, the same properties also cause problems. For example, as mentioned above, jamming attacks on low-power radio networks can be detected [41]. Due to the multi-path effects and the inherent broadcast nature of radio traffic, a jamming attack on one or several hosts can easily be detected by other nodes that would also experience a higher energy level in the radio channel. These nodes can then take actions such as re-routing of traffic. With today’s directional VLC channels, however, it might not be as straightforward to understand that one or several nodes are exposed to a jamming attack. For example, even light sensors close to the host under attack might not recognize an ongoing attack even though a human present in the same room might be able observe such an attack. 5. Conclusions Thanks to the massive uptake of LEDs for illumination as well as the fear of the RF spectrum crunch, VLC has recently emerged as a hot research area and complement to RF as infrastructure for the IoT. Nevertheless, VLC security has only been investigated in a few studies. In this paper, after reviewing the key properties that make VLC fundamentally different from RF, we have surveyed various solutions from the wired/RF security literature that may be employed successfully to secure VLC. Moreover, we have delved into a survey of opportunities for security research that arise from the uptake of VLC, and we have reasoned about how attacks against VLC may fare. We hope that this paper will serve to stimulate future investigations in VLC security research, which remains an under-explored space whose strategic importance is bound to grow as VLC research and development efforts continue to gain momentum in the IoT realm. References Fig. 1. Visible light DoS defense via shielding and rerouting. [1]. S. Dimitrov, H. Haas, Principles of LED Light Communications, Cambridge University Press, 2015. [2]. OMEGA, the Home Gigabit Access project, http://www.ict-omega.eu. [3]. D. Tsonev, H. Chun, H. Rajbhandari, J. McKendry, S. Videv, E. Gu, M. Haji, S. Watson, A. Kelly, G. Faulkner, M. Dawson, H. Haas, D. O’Brien, A 3 Gb/s single-LED OFDM-based wireless VLC link using a gallium nitride µLED, IEEE Photonics Technology Letters, Vol. 26, No. 7, 2014, pp. 637-640. [4]. A. H. Azhar, T.-A. Tran, D. O’Brien, Demonstration of high-speed data transmission using MIMO-OFDM visible light communications, in Proceedings of the IEEE Globecom Workshops (GC’10), 2010, pp. 1052-1056. 2 We expect to see such networks multi-hop VLC networks in the future which requires, however, a redesign of the protocol stack. 13 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 9-15 [5]. D. C. O’Brien, S. Quasem, S. Zikic, G. E. Faulkner, Multiple input multiple output systems for optical wireless: challenges and possibilities, in Proceedings of the SPIE Optics and Photonics. International Society for Optics and Photonics, Vol. 6304, 2006. [6]. T. Komiyama, K. Kobayashi, K. Watanabe, T. Ohkubo, Y. Kurihara, Study of visible light communication system using RGB LED lights, in Proceedings of the IEEE SICE Annual Conference (SICE), 2011, pp. 1926-1928. [7]. S. Schmid, G. Corbellini, S. Mangold, T. R. Gross, LED-to-LED visible light communication networks, in Proceedings of the Fourteenth ACM International Symposium on Mobile Ad Hoc Networking and Computing (MobiHoc’13), 2013, pp. 1-10. [8]. D. Giustiniano, N. O. Tippenhauer, S. Mangold, Lowcomplexity visible light networking with LED-toLED communication, IFIP Wireless Days’12, Dublin, Ireland, Nov. 2012, pp. 1-8. [9]. T. Hao, R. Zhou, G. Xing, COBRA: color barcode streaming for smartphone systems, in Proceedings of the 10th International Conference on Mobile Systems, Applications, and Services, 2012, pp. 85-98. [10]. W. Hu, J. Mao, Z. Huang, Y. Xue, J. She, K. Bian, G. Shen, Strata: Layered coding for scalable visual communication, in Proceedings of the 20th Annual International Conference on Mobile Computing and Networking (MobiCom’14), ACM, 2014, pp. 79-90. [11]. A. Wang, S. Ma, C. Hu, J. Huai, C. Peng, G. Shen, Enhancing reliability to boost the throughput over screen-camera links, in Proceedings of the 20th Annual International Conference on Mobile Computing and Networking (MobiCom’14), 2014, pp. 41-52. [12]. P. Hu, L. Li, C. Peng, G. Shen, F. Zhao, Pharos: enable physical analytics through visible light based indoor localization, in Proceedings of the Twelfth ACM Workshop on Hot Topics in Networks, 5, 2013, pp. 1-7. [13]. L. Li, P. Hu, C. Peng, G. Shen, F. Zhao, Epsilon: A visible light based positioning system, in Proceedings of the 11th USENIX Symposium on Networked Systems Design and Implementation (NSDI), Seattle, USA, 2014, pp. 331-343. [14]. M. Dunbabin, P. Corke, I. Vasilescu, D. Rus, Data muling over underwater wireless sensor networks using an autonomous underwater vehicle, in Proceedings of the IEEE International Conference on Robotics and Automation (ICRA’06), 2006, pp. 2091-2098. [15]. A. Mostafa, L. Lampe, Physical-layer security for indoor visible light communications, in Proceedings of the IEEE International Conference on Communications (ICC), 2014, pp. 3342-3347. [16]. B. Zhang, K. Ren, G. Xing, X. Fu, C. Wang, Sbvlc: Secure barcode-based visible light communication for smartphones, in Proceedings of the IEEE INFOCOM, 2014, pp. 2661-2669. [17]. A. Sevincer, A. Bhattarai, M. Bilgi, M. Yuksel, N. Pal, LIGHTNETs: Smart LIGHTing and Mobile Optical Wireless NETworks, IEEE Communications Surveys and Tutorials, Vol. 15, No. 4, 2013, pp. 1-22. [18]. M. S. Uddin, J. S. Cha, J. Y. Kim, Y. M. Jang, Mitigation technique for receiver performance variation of multi-color channels in visible light communication, Sensors, Vol. 11, No. 6, 2011, pp. 6131-6144. [19]. M. Afgani, H. Haas, H. Elgala, D. Knipp, Visible Light Communication Using OFDM, in Proceedings 14 [20]. [21]. [22]. [23]. [24]. [25]. [26]. [27]. [28]. [29]. [30]. [31]. [32]. [33]. [34]. of the 2nd International Conference on Testbeds and Research Infrastructures for the Development of Networks and Communities (TRIDENTCOM’06), Barcelona, Spain, 1–3 March 2006, pp. 129-134. R. Mesleh, H. Elgala, H. Haas, Optical Spatial Modulation, IEEE/OSA Journal of Optical Communications and Networking, Vol. 3, No. 3, March 2011, pp. 234-244. S. Dimitrov, S. Sinanovic, H. Haas, Clipping Noise in OFDM-based Optical Wireless Communication Systems, IEEE Transactions on Communications (IEEE TCOM), Vol. 60, No. 4, Apr. 2012, pp. 1072-1081. J. Vucic, C. Kottke, S. Nerreter, K. Langer, J. Walewski, 513Mbit/s visible light communications link based on DMT-modulation of a white LED, Journal of Lightwave Technology, Vol. 28, No. 24, Dec. 2010, pp. 3512-3518. IEEE Standard for Local and Metropolitan Area Networks 15.7: PHY and MAC Standard for ShortRange Wireless Optical Communication Using Visible Light, IEEE Standard 802.15.7. K. Cui, G. Chen, Z. Xu, R. D. Roberts, Line-of-sight visible light communication system design and demonstration, in Proceedings of the IEEE 7th International Symposium on Communication Systems Networks and Digital Signal Processing (CSNDSP’10), 2010, pp. 621-625. B. Azimi-Sadjadi, A. Kiayias, A. Mercado, B. Yener, Robust key generation from signal envelopes in wireless networks, in Proceedings of the 14th ACM Conference on Computer and Communications Security, Alexandria, VA, USA, 2007, pp. 401-410. S. Jana, S. N. Premnath, M. Clark, S. K. Kasera, N. Patwari, S. V. Krishnamurthy, On the effectiveness of secret key extraction from wireless signal strength in real environments, in Proceedings of the 15th Annual International Conference on Mobile Computing and Networking (MobiCom’09), Beijing, China, Sep. 2009, pp. 321-332. R. Wilson, D. Tse, R. Scholtz, Channel identification: Secret sharing using reciprocity in ultrawideband channels, IEEE Transactions on Information Forensics and Security, Vol. 2, No. 3, 2007, pp. 364-375. R. L. Rivest, et al., Chaffing and winnowing: Confidentiality without encryption, CryptoBytes (RSA Laboratories), Vol. 4, No. 1, 1998, pp. 12-17. S. R. Hanna, Configuring security parameters in small devices, [draft-hanna-zeroconf-seccfg-00], 2002. C. Gehrmann, Deliverable Detailed Technical Specification of Mobile Terminal System Security, SHAMAN, 2002. J. M. McCune, A. Perrig, M. K. Reiter, Seeing-isbelieving: using camera phones for human-verifiable authentication, in Proceedings of the IEEE Symposium on Security and Privacy, 2002, pp. 110-124. N. Saxena, J.-E. Ekberg, K. Kostiainen, N. Asokan, Secure device pairing based on a visual channel, in Proceedings of the IEEE Symposium on Security and Privacy, 2006, pp. 313-319. L. M. Feeney, B. Ahlgren, A. Westerlund, A. Dunkels, Spontnet: Experiences in configuring and securing small ad hoc networks, in Proceedings of the 5th Int’l Workshop on Networked Applicances (IWNA5), 2002, pp. 102-106. R. Ahswede, C. Ning, S.-Y. Li, R. Yeung, Network information flow, IEEE Transactions on Information Theory, Vol. 46, No. 4, 2000, pp. 1204-1216. Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 9-15 [35]. A. Shamir, How to share a secret, Communications of the ACM, Vol. 22, No. 11, 1979, pp. 612-613. [36]. S. Brands, D. Chaum, Distance-bounding protocols (extended abstract), EUROCRYPT’93, Lecture Notes in Computer Science, Vol. 765, 1993, pp. 344-359. [37]. R. Blom, An optimal class of symmetric key generation systems, EUROCRYPT’84, Lecture Nodes in Computer Science, Vol. 209, 1985, pp. 335-338. [38]. A. Mpitziopoulos, D. Gavalas, C. Konstantopoulos, G. E. Pantziou, A survey on jamming attacks and countermeasures in WSNs, IEEE Communications Surveys and Tutorials, Vol. 11, No. 4, 2009, pp. 42-56. [39]. A. Wood, J. A. Stankovic, G. Zhou, Deejam: Defeating energy-efficient jamming in IEEE 802.15. 4-based wireless networks, in Proceedings of the 4th Annual IEEE Communications Society Conference on Sensor, Mesh and Ad Hoc Communications and Networks (SECON), 2007, pp. 60-69. [40]. M. Wilhelm, I. Martinovic, J. Schmitt, V. Lenders, WiFire: A firewall for wireless networks, in Proceedings of the ACM Conference on Applications, Technologies, Architectures, and Protocols for Computer Communications (SIGCOMM’11), New York, NY, USA, August 2011, pp. 456-457. [Online]. Available: http://discofiles/publicationsfiles/WMSL11- 2.pdf [41]. D. Giustiniano, V. Lenders, J. B. Schmitt, M. Spuhler, M. Wilhelm, Detection of reactive jamming in DSSSbased wireless networks, in Proceedings of the Sixth ACM Conference on Security and Privacy in Wireless and Mobile Networks, 2013, pp. 43-48. [42]. A. Wood, J. A. Stankovic, S. H. Son, JAM: A jammedarea mapping service for sensor networks, in Proceedings of the 24th IEEE Real-Time Systems Symposium (RTSS’03), 2003, pp. 286-297. [43]. T. Li, C. An, Z. Tian, A. Campbell, X. Zhou, Human Sensing Using Visible Light Communication, in Proceedings of the 21st Annual International Conference on Mobile Computing and Networking (MobiCom), Paris, France, September 2015. ___________________ 2015 Copyright ©, International Frequency Sensor Association (IFSA) Publishing, S. L. All rights reserved. (http://www.sensorsportal.com) 15 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 16-21 Sensors & Transducers © 2015 by IFSA Publishing, S. L. http://www.sensorsportal.com Green Walls Utilizing Internet of Things Andrejs BONDAREVS, 1 Patrik HUSS, 1 Shaofang GONG, 2 Ola WEISTER and 2 Roger LILJEDAHL 1 ITN – Communication Electronics, LiU – Campus Norrköping, SE-60174, Norrköping, Sweden 2 Vertical Plants System Sweden AB, Box 2175, SE–60002, Norrköping. Sweden 1 Tel.: +46 (0) 700896063 1 E-mail: [email protected] Received: 19 August 2015 /Accepted: 20 September 2015 /Published: 30 September 2015 Abstract: A wireless sensor network was used to automatically control the life-support equipment of a green wall and to measure its influence on the air quality. Temperature, relative humidity, particulate matter, volatile organic compound and carbon dioxide were monitored during different tests. Green wall performance on improving the air quality and the influence of the air flow through the green wall on its performance were studied. The experimental results show that the green wall is effective to absorb particulate matter and volatile organic compound. The air flow through the green wall significantly increases the performance. The built-in fan increases the absorption rate of particulate matter by 8 times and that of formaldehyde by 3 times. Copyright © 2015 IFSA Publishing, S. L. Keywords: Internet of things, Wireless sensor network, Green wall, Air quality, Particulate matter, Volatile organic compound. 1. Introduction Green walls are plants grown in vertical systems that can be freestanding but generally attached to internal or external walls. Green walls allow for high density and high diversity vegetating on vertical areas [1]. Internet of Things (IoT) is a global infrastructure for the information society, enabling advanced services by interconnecting (physical and virtual) things based on existing and evolving interoperable information and communication technologies [2]. Green walls are not a part of IoT yet. They are gaining its popularity because of the aesthetic and environmental reasons. Green walls require regular maintenance, which includes monitoring (for example, temperature and relative humidity) and control (for example, irrigation) [3-4]. 16 Several papers indicate early stages of connecting green houses to IoT [5-11]. However, no full integration with IoT could be found. If green walls would be a part of IoT, the status of the green wall and its impact on the air quality can be constantly monitored. However, no research on connecting green walls to IoT could be found so far. 1.1. Indoor Air Quality Indoor air quality shows how polluted the air is. Air quality includes such pollutants as carbon dioxide, particulate matter (PM) and volatile organic compound (VOC). In large cities the air pollution is often a problem [12]. To improve the indoor air quality, air purifiers are used. It is known that plants improve indoor air http://www.sensorsportal.com/HTML/DIGEST/P_2718.htm Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 16-21 quality [13-14]. However it is unknown how effective the general indoor green wall at absorbing carbon dioxide, PM and VOC. It is also unknown how the air flow through the green wall influences its performance. In this paper we present that it is possible to connect green walls to IoT and use IoT to control the equipment of the green wall and measure its influence on the air quality in a controlled environment. 2. The Experiment Setup The experiment was done in a laboratory without windows and with the ventilation system turned off. 2.1. Wireless Sensor Network The wireless sensor network (WSN) developed at Linköping University in Sweden [15-16] is based on the ZigBee specification and is an IoT solution. It provides nodes for monitoring temperature, relative humidity, carbon dioxide, VOC, PM and electricity consumption as well as for automatic control. The architecture of the WSN is shown in Fig. 1. It is a mesh network. The coordinator is responsible for the control of the network and is a gateway between the WSN and USB interface. The coordinator is connected to the Local Server which is an IBMcompatible computer with software responsible for managing the data and tasks, through the USB interface. The Local Server is a gateway between the WSN and the Internet. The main server provides cloud based service for several WSNs at the same time. The WSN consists of the coordinator, routers and wireless sensors. Routers provide the ability to extend the network and increase possible communication paths between sensor devices. The router functionality can be combined together with other functionalities, such as sensing or control, while the router device is powered by mains. Sensor devices are batterypowered wireless devices with low energy consumption and are in the sleep mode most of the time; therefore, they can only be used for sensing and monitoring purposes. Wireless Sensor Network Wireless sensor Local Server Coordinator Router Relay Switch Box Router Cloud Wireless Sensor Wireless Sensor Relay Switch Box User Fig. 1. Wireless Sensor Network architecture. Dashed lines show some possible alternative wireless connections. 2.2. The Green Wall The green wall shown in Fig. 2 is provided by the Vertical Plants System AB in Sweden. The green wall size is 200 × 200 × 18 cm. The green wall has a builtin water pump, water storage compartment and fan for forced air circulation through the plants (see Fig. 2). To create an isolated environment, a greenhouse made of PVC (Polyvinyl chloride) and aluminum was built around the green wall, see Fig. 3. The greenhouse size is 406 × 203 × 223 cm (15.9 m3). As illustrated in Fig. 4, a separation wall with one opening was built inside the green house. The opening is about 20 cm in diameter. If the fan is on, then the airflow goes from the lower compartment to the upper compartment. Fan has five speeds, but only two speeds were used. Table 1 shows the relation between the air flow and the fan speed. Fig. 2. The green wall used for the experiment, provided by Vertical Plants System AB (Sweden). 17 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 16-21 Water pump Fan Light Dehumidifier Feedback Schedule Schedule Fig. 5. Control structure for the equipment on the green wall. 2.4. Sensors Third-party sensors without any wireless functionality were integrated into the WSN. Table 2 lists types and models of sensors used. Fig. 3. The green house that was built around the green wall to create an isolated environment. The horizontall separation wall can be seen on the side of the green house. Fig. 4. Schematical view of the greehouse from the side. Table 1. Air flow depending on the fan speed. Fan speed 2 5 Airflow 28.8 m3/h 82.8 m3/h 2.3. Automatic Control In order to maintain the green wall (see Fig. 2), a life support system should be controlled. It includes lights, the water pump and the fan. Each unit is connected to a relay switch node. Lights are scheduled to be on every day from 7:00 to 21:00. The water pump is scheduled to run for 10 minutes at 7:00, 9:00, 11:00, 13:00 and 17:00. The fan is controlled manually depending on the experimental conditions. As illustrated in Fig. 5, to control the relative humidity inside of the green house, a dehumidifier is used. It is connected to the relay switch node and controlled by a feedback loop to keep relative humidity at 60 %. 18 Table 2. Sensors used in the experiment. Sensor type Temperature Relative humidity Carbon dioxide VOC PM Sensor model SHT21 [17] SHT21 [17] COZIR GC-0012 [18] NanoSense E4000 [19] NanoSense P4000 [19] The sensors used are shown in the Fig. 6. PM and VOC sensors are connected to the wireless nodes through a RS485 to UART (Universal asynchronous receiver/transmitter) converter. VOC and PM sensors themselves have a separate power (12 V). The carbon dioxide sensor is customly built into the enclosure and is connected directly to the wireless node through the UART interface. 3. Measurement Data from the cloud was exported to the Comma Separated Values format and used in MATLAB for further analysis. 3.1. Carbon Dioxide Carbon dioxide from a high pressure carbon dioxide cylinder was injected into the lower compartment of the green house (see Fig. 4). It reaches the concentration of 1167 and 8037 ppm, respectively. Measurements were performed with fan on and off, respectively. 3.2. Volatile Organic Compound Two different substances were used for the test: Universal Glue and Formaldehyde 4 % solution. The Universal Glue contains two VOCs: acetone and methyl acetate. Formaldehyde and acetone are common VOCs. Substances were put on a surface of Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 16-21 seven days. Fluctuations of ±25 ppm can be observed between the night and the day. CO2, ppm 600 cm2 (paper for glue and glass for formaldehyde) inside of the green house. Fig. 7. Carbon dioxide concentration measured over the time. Black – light period. Grey – dark period. As shown in Fig. 8, the PM1.0 concentration reduces at a rate of -1.87 µg·m3·h-1 with fan off, -6.67 µg·m3·h-1 with fan at speed 2 and -20.06 µg·m3·h-1 with fan at speed 5. Fig. 6. Sensors used in the experiments. (a) PM/VOC sensor, (b) carbon dioxide sensor, (c) temperature and relative humidity sensor. 1 – wireless node developed at Linköping University in Sweden, 2 – VOC or PM sensor, 3 – RS485 to UART converter. 3.3. Particulate Matter Ashes from the fireplace were used, as they can be found in many houses and contain PM10, PM2.5 and PM1.0. Ashes were injected into the lower and upper compartments of the green house through the standard computer 80 × 80 mm cooling fan. The air flow generated by the fan effectively distributes particles in the air. Fig. 8. PM1.0 concentration measured over the time. A - fan speed 5. B - fan speed 2. C - fan off. As shown in Fig. 9, it reduces Formaldehyde concentration at a rate of 0.033 ppm·h-1 with fan off, 0.076 ppm·h-1 with fan at speed 2 and 0.09 ppm·h-1 with fan at speed 5. Values are for the case when the light is on. 3.4. Temperature and Relative Humidity The temperature was not controlled, but was monitored. Temperature was measured in the lower and upper compartments, and on the green wall. The relative humidity was controlled by a dehumidifier and should remain constant at a level of 60-70 % when controlled. 4. Result The green wall reduces the carbon dioxide concentration at a rate of 11.46 ppm/h with fan off, see Fig. 7. The carbon dioxide concentration stabilizes in Fig. 9. VOC (Formaldehyde) concentration measured over the time. A - fan off. B - fan speed 2. C - fan speed 5. Dehumidifier is off. 19 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 16-21 Fig. 10 shows that temperature fluctuates between 25.5 and 27.4 °C in the upper compartment of the green house and between 23.4 and 25.2 °C in the lower compartment of the green house. than that in the dark. During our experiment, the formaldehyde absorption rate is about 6 times faster in the light than in the dark, which was confirmed in two tests (see Fig. 9). 6. Conclusions Fig. 10. Temperature measurement in the green house over the time. Fig. 11 shows that the relative humidity is stable and is at the level of 66 % in the upper compartment of the green house and 56 % in the lower compartment of the green house. The green wall evaporates about 137 g of water per hour. The green wall with an active built-in fan to increase air flow significantly improves the air quality. It is effective at absorbing VOC and PM, but not equally effective at absorbing carbon dioxide. The built-in fan increases the PM absorption rate by 8 times and Formaldehyde absorption rate by 3 times. The green wall increases the relative humidity, which is good to use in a dry environment. WSN, which is a part of IoT, reduces the complexity of the experiment setup. Wireless nodes are easy to install and move around. The ability to access the nodes through the Internet makes it easy to control the experiment. Acknowledgements The Norrkoping municipality, the Swedish energy agency and Vinnova in Sweden are acknowledged for financial support of the study. Gustav Knuttson at LiU - Campus Norrköping is acknowledged for the technical support. References Fig. 11. Relative humidity in the green house measured over the time. 5. Discussion In the carbon dioxide measurements (see Fig. 7) it can be noticed that the carbon dioxide concentration reduces in the dark time period also in the beginning of the experiment. This behavior could be caused by sol-called Crassulacean Acid Metabolism (CAM) [20]. CAM involves microorganisms that live in the soil and they do not require light to absorb carbon dioxide. CAM occurs in approximately 6 % of high plant species. However, there is no proof that this was the case during our experiment. Paper [14] states that formaldehyde absorption is not significantly affected with the light intensity; however, there is a considerable difference between light and dark conditions. In light condition the absorption rate should be approximately 5 times faster 20 [1]. Victoria’s Guide to Green roofs, walls and facades. (http://www.growinggreenguide.org). [2]. International Telecommunication Union (http://www.itu.int). [3]. S. Loh, Living walls - a way to green the built environment, in BEDP Environment Design Guide, Institute for BEDP, August 2008, TEC 26. [4]. A. Wood, P. Bahrami, D. Safarik, Green Walls in High-Rise Buildings, Images Publishing, Australia, 2014. [5]. T. Ahonen, R. Virrankoski, M. Elmusrati, Greenhouse Monitoring with Wireless Sensor Network, in Proceedings of the IEEE/ASME International Conference on Mechatronic and Embedded Systems and Applications (MESA’08), pp. 403 – 408. [6]. N. Sakthipriya, An Effective Method for Crop Monitoring Using Wireless Sensor Network, MiddleEast Journal of Scientific Research, 20, Vol. 9, 2014, pp. 1127-1132. [7]. D. D. Chaudhary, S. P. Nayse, L. M. Waghmare, Application of Wireless sensor networks for greehnouse parameter control in precision agriculture, International Journal of Wireless & Mobile Networks, Vol. 3, No. 1, 2011, pp. 140 – 149. [8]. Y. Song, J. Ma, Z. Zhang, Y. Feng, Design of Wireless Sensor Network-Based Greenhouse Environment Monitoring and Automatic Control System, Journal of networks, Vol. 7, No. 5, May 2012, pp. 838 – 844. [9]. I. Matijevics, S. Janos, Control of the Greenhouse’s Microclimatic Condition using Wireless Sensor Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 16-21 [10]. [11]. [12]. [13]. [14]. Network, IPSI BgD Transactions on Internet Research, 2010, July, pp. 35 – 38. A. Lambebo, S. Haghani, A Wireless Sensor Network for Environmental Monitoring of Greenhouse Gases, in Proceedings of the ASEE Zone I Conference, April 3-5, 2014, pp. 1 – 4. O. Postolache, P. Girão, M. Pereira, C. Grueau, H. Teixeira, M. Leal, Greenhouses microclimate realtime monitoring based on a wireless sensor network and GIS, in Proceedings of the XX IMEKO World Congress, September 9-14, 2012, Busan, Republic of Korea, Vol. 1, pp. 1 – 5. Beijing Air Pollution (http://aqicn.org/city/beijing/) R. A. Wood, M. D. Burchett, R. Alquezar, R. L. Orwell, J. Tarran, F. Torpy, The potted-plant microcosm substantially reduces indoor air VOC pollution: i. office field-study, in Water, Air and Soil Pollution, Springer, 2006, pp. 163-180. Kwang Jin Kim, Myeong Il Jeong, Dong Woo Lee, Jeong Seob Song, Hyoung Deug Kim, Eun Ha Yoo, Sun Jin Jeong, and Seung Won Han, Variation in [15]. [16]. [17]. [18]. [19]. [20]. Formaldehyde Removal Efficiency among Indoor Plant Species, HortScience, 45, 10, 2010, pp. 1489-1495. P. Huss, N. Wigertz, J. Zhang, A. Huynh, Q. Ye, S. Gong, Flexible and Reliable Local Manager for Internet of Things, Advanced Science and Technology Letters, Vol. 41, 2013, pp. 1-4. P. Huss, N. Wigertz, J. Zhang, A. Huynh, Q. Ye, S. Gong, Flexible Architecture for Internet of Things Utilizing an Local Manager, International Journal of Future Generation Communication and Networking, Vol. 7, No. 1, 2014, pp. 235-248. Sensirion manufacturer (http://www.sensirion.com) CO2Meter manufacturer (http://www.co2meter.com) Nano-Sense manufacturer (http://www.nanosense.com). Antony N. Dodd, Anne M. Borland, Richard P. Haslam, Howard Griffiths, Kate Maxwell. Crassulacean Acid Metabolism: plastic, fantastic, Journal of Experimental Botany, Vol. 53, No. 369, 2002, pp. 569-580. ___________________ 2015 Copyright ©, International Frequency Sensor Association (IFSA) Publishing, S. L. All rights reserved. (http://www.sensorsportal.com) 21 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 22-29 Sensors & Transducers © 2015 by IFSA Publishing, S. L. http://www.sensorsportal.com Metrological Array of Cyber-Physical Systems. Part 11. Remote Error Correction of Measuring Channel Yuriy YATSUK, Mykola MYKYJCHUK, Volodymyr ZDEB, and Roman YANOVYCH National University ‘Lviv Polytechnic’, Institute of Computer Technologies, Automation and Metrology, Bandera str. 12, Lviv, 79013, Ukraine Tel.: +38-0322-58-23-79 E-mail: [email protected] Received: 30 August 2015 /Accepted: 28 September 2015 /Published: 30 September 2015 Abstract: The multi-channel measuring instruments with both the classical structure and the isolated one is identified their errors major factors basing on general it metrological properties analysis. Limiting possibilities of the remote automatic method for additive and multiplicative errors correction of measuring instruments with help of code-control measures are studied. For on-site calibration of multi-channel measuring instruments, the portable voltage calibrators structures are suggested and their metrological properties while automatic errors adjusting are analysed. It was experimentally envisaged that unadjusted error value does not exceed ± 1 μV that satisfies most industrial applications. This has confirmed the main approval concerning the possibilities of remote errors selfadjustment as well multi-channel measuring instruments as calibration tools for proper verification. Copyright © 2015 IFSA Publishing, S. L. Keywords: Cyber-physical system, Metrological assurance, Multi-channel measuring instrument, Remote errors correction and verification, Code-control voltage measure. 1. Introduction Cyber-physical systems (hereinafter CPSs) are deemed to be an integral part of manufacturing systems, factories, machinery, test facilities, moving objects, vehicles etc. These facilities typically utilize thousands of physical phenomena, whose parameters are constantly changing. Each CPS is comprised of dispersed hardware components and computer software, intended to obtain information about the progress of physical processes in controlled facilities, as well as its storage, transmission, processing and production by control signals. Especially it is needed information on the measured values including the location, value, speed changes, etc. 22 The measurement data, received from controlled objects, would be characterized by the set of metrological parameters. The measuring channels distribution in space, permissible changes in a wide range of operating parameters and inevitable degradation of measuring circuits parameters result in a significant deterioration of the CPS measuring channel performance. Thus, an operative metrological maintenance of measuring channels becomes important [1]. 2. Shortcomings CPS measurement data accumulation and processing is performed by means of multi-channelled measuring instruments (further MCMIs) that consists http://www.sensorsportal.com/HTML/DIGEST/P_2719.htm Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 22-29 of measuring sensors, communicating lines (further CLs), channel commutators (further CCs) and measuring instruments (further MIs) (Fig. 1). The current trends of measuring systems design seems to be the implementation the measuring transducers that transfer the received signals into electrical form aiming the direct computing [2-3]. Whereby, digital measuring information could be obtained with help of the certain methods of processing, transmission, storage, reverse transformation for the control function for CPS units. Fig. 1. Functional scheme of modern multi-channel measuring instruments: CC is the channel commutator; IB is the intrinsic safety barrier; CL is the communicating lines; IAB is the input amplification block; ID is the isolation device; ACS is the analogue control circuit; CNT is the instrument controller. The low level of output sensors signals require input amplification block (further IAB) that scales the previously mentioned signals to normal level for ADC operation and simultaneously converts them in a digital code necessary for MI controllers. Measurements in sparkproof operating conditions and in dangerous environments envisage implementtation of some specific techniques. First, the inner safety barriers must applied at the output sensors of each measuring channel, and the analogue circuit of MI has to be isolated from the digital one (Fig. 1) [4-7]. The interference values often exceed the signal parameters of CC channels. So standard signal transducer (further SST), isolated amplifier (further IA) or isolation device (further ID) it usually applied. The systematic errors that have both significant additive and multiplicative components emerge in measuring circuit of such data acquisition systems (further DASs). Error values increase in DASs with isolated channels; therefore, it is difficult to ensure their operation by considerable time at the certain temperature drift [7-12]. To correct the errors of CPSs, the calibrators of electrical quantities directly connected to measuring channel input instead of sensors are mainly applied. However, these calibrators are large, heavy, and quite expensive; so their implementation is complicated [13]. To provide the remote automatic adjustment, currently the CPS measuring channels with embedded devices are designed. It upraises a problem of automatic errors correction of operating calibrators that have to be inexpensive due to their wide use. 3. Aim of Work The aim of this article is the development of theoretical basis and practical guidance for providing the high accuracy of multi-channelled measuring instruments in operating conditions. 4. Theory and Applied Researches The MCMI scheme (Fig. 1) for measured object without spark and explosive environments and at the common mode voltage lower than the CCs chip breakdown voltage (10 V), is studied. So while gauging spark and explosive objects, it should be used the isolation blocks (further IBs) on the sensor outputs of every measuring channel. It recommends an extra electrical isolating the sensors and MCMI for particular dangerous objects [4-7]. For this purpose, the magnetic, capacitive, or optical means are generally applied in the measuring circuits that considerably decrease the error values at variable operating condition. The emerging ground loop can be quite large (up to several kilo ampere) that causes the common mode voltage up to hundreds of volts. Its values especially increase with CL length between the ground points of both measured facilities and MCMI. Another source of common mode voltage can be leakage currents of power networks that pass through measuring equipment insulation for ground loops measured object. That application point of common mode voltage to the sensor is generally unknown. To exclude above-mentioned drawbacks the relays as CC with switching function "before turning off" for large common mode voltage, can be used. Such scheme practical application is inherent in a significant (up to several millivolts) additive error component (further AEC) caused by contact EMF at temperature drift (up to ten μV/K). Thus, MCMI structure seems to be similar to the design shown in Fig. 1. To reduce significantly errors values caused by CLs and CCs, SST converters or IA are currently applied. Three-wire sensors connection, and screening the CLs as well as MCMI analogue part substantially decreases the common mode voltage [2, 7, 14]. Then the block diagrams of MI significantly differ from MCMI in Fig. 1. 4.1. Metrological Properties Analysis of Classic Multi-Channel Measuring Instruments Output sensors signals are submitted to the CC inputs through IB (if necessary) and CL. In addition to the measuring signal UX, every measuring channel is inherent in own common mode voltage. A differential circuit of IAB is applied for reducing the common mode voltage affects (Fig. 2). 23 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 22-29 the common mode resistance estimated for “i” CCs channel. This error can be determined in a few tens per cent. Taking into account the following values of ratios Z1ixe, Z2ixe<<Zin, Z1en, Z2en, the expression for the equivalent input resistances is defined: Fig. 2. Equivalent scheme of channel commutator and input amplifying block. Relegated to MCMI input the measured voltage Uixn in the “i” on-channel is presented below (at the known common mode voltage Uicm applying point): U ixn = U ix (1 + δ i )+ eie +U iI +U ixc +U ijx +U ijxc , (1) where Uix is the output sensor voltage in i on-measuring channel, eie=eiCL+eiIB+eiCC is the equivalent input offset voltage, eiCL=e1iCL+e2iCL, eiIB=e1iIB+e2iIB, eiCC= +e1iк+e2iк, e1iк, e2iк is the residual voltage of the first and the second on-keys CC respectively (eiCC=0 for MOS FET chip keys), δi=Zixe/Zin, Zixe=Z1ixe+Z2ixe, Z1ixe=Z1ix+Z1iCL+Z1iIB+Z1iCC, Z2ixe=Z2ix+Z2iCL+Z2iIB+ +Z2iCC is the total resistance between common mode voltage applying point and the first and the second ІАВ differential inputs respectively, UiI is the equivalent error value caused by equivalent currents of both IAB differential inputs, Uiec is the equivalent error value caused by common mode voltage of “i” on-channel, Uijx is the equivalent error value caused by the penetration of the measured voltages Ujx from other measuring off-channel, Uijxc is the equivalent error value caused by the penetration of the common mode voltages Ujxc from other measuring off-channel. AEC value UiI, caused by equivalent currents of both differential inputs IAB, is estimated: U iI = U1iI − U 2iI = I1e Z1e − I 2e Z 2e , (2) where I1e = I1in + I i11 + I12e , I 2e = I 2in + Ii21 + I 22e is the equivalent current values of both IAB differential inputs accordingly, I1in, I2in is the input current of both IAB differential inputs concordantly, Ii11, Ii21 is the input reverse current of both CC “on” input keys at i n n i =1 i =1 channel respectively, I12e = I i12 , I 22e = I i22 , Ii12, Ii22 is the output reverse CC current for i on-channel, Z1e, Z2e is the equivalent common mode resistance of both IAB differential inputs accordingly, n is the number of measuring channels. We can accept that value of the input and output common resistance of CCs approximately equal to each other: Zi11=Zi1(1+δi11), Zi12=Zi1(1+δi12), Zi21=Zi1(1+δi21), Zi22=Zi1(1+δi2), where δi11, δi12, δi21, δi22<<1, δi11, δi12, δi21, δi22 are the relative dispersion of 24 Z1e ≅ Z en Z eis b Z Z 2 ixe + 1ixe , 1 + Z en + 2 Z eis Z en a 2 (3) Z 2e ≅ Z en Z eis Z en + 2Z eis b Z Z1ixe + 2ixe , 1 + a 2 Z en (4) where Zen=(Z1e+Z2e)/2, Zen = 0,5Zi1Zc [(n + 1)Zc + Zi1 ] , Z1e, Z2e is the equivalent common mode input resistance accordingly, Z1e = 1 G1e , Z 2 ec = 1 G2 ec , Zeis=Zis+Zicm, Zis, Zicm is the common mode resistance of i on- measuring channel and isolation resistance of common measuring bus (measuring “ground”) relatively MCMI grounding n point respectively, G1e = 1 Z i11 + 1 Z1c + (1 Z i 21 ) , i =1 G2e = 1 Z i 21 + 1 Z 2 c + n (1 Z ) , i 22 Zi11, Zi21 is the i =1 common mode input resistance for “i” Zi22 is the common on-channel, Zi21, mode output resistance for i on-measuring channel, a=Zeis/(Zeis+Zen), b=a/(1+a), Z1ixe=Z1ix+Z1iCL+Z1iIB+Z1iCC, Z2ixe=Z2ix+Z2iCL+Z2iIB+ +Z2iCC, Zin is the differential input resistance, Z1c, Z2c is the common mode input resistance of both IAB differential inputs respectively. Considering the expressions (3) and (4), obtain the AEC UiI caused by equivalent input currents: ( ) U iI ≅ ΔI e Z enb + 2 I ein ΔZ ixe 1 + a 2 (b a ) , 2 (5) where ΔIe=I1e-I2e, Iein=(I1e+I2e)/2, ΔZixe=Z1ixe-Z2ixe. Caused by common mode voltage Uicm at “і” onchannel after sequence of alterations, error Uixc, is determined: ≅ 2 ( + ), (6) where Zixe=Z1ixe+Z2ixe, δixe=ΔZixe/(Z1ixe+Z2ixe) is the relative dispersion of both total input resistance Zixe IAB, δie=(Z1e-Z2e)/2Zen is the relative dispersion of both equivalent input differential resistances IAB, Zisx=Zicm+Zis+Zen/2. caused by Analysis envisages that error value inherent in additive and common mode voltage asymmetrical features and depends on both differential inputs resistances IAB (Fig. 2). For its reduction, one should increase the insulation resistance of the common bus IAB to the applying . point of common mode voltage Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 22-29 The equivalent error Uijx caused by penetration to “i” on-channel measuring voltage Ujx from the all offchannels is equal to: n −1 Z in , U ijx = U jx Z in + Z jp + Z jxe j =1 (7) where =Zjx+ZjCL+ZjIB, Zjx, ZjCL, ZjIB is the inner resistance of sensor, CL and ІВ at j off CC channel respectively. The nature of this relative to the measured voltage in “i” on-channel error is additive. For its adjustment it can be applicable the known automatic methods. Analysis of (7) results in the following; the error of voltage Uijx increases proportionally to the number of measuring channels n. For its reduction within the MCMI classical structure should choose the CC with a maximum high value off-resistances. However, this kind of MCMI accuracy improvement substantially limits imposed by the parameters of chip components. For example, if typical values are equal to ≃109 Ohm, ≃1012 Ohm, ≪ , and the measured voltages values approximately equal to each ≃ і weighting factor significance kijx drops other down in the order of value: to ≅ 0.001 at the number of measuring channels n=2, or to ≅0.01 at number of channels n=12. interference caused Threshold value of AEC by penetration of a common mode voltage of all the other off-channels to the “i” on-channel, gives expression: n −1 U jcm Z ixe Z jp [k1δ ixe + k2δ ie ] , U ijxc = j =1 2 Z jcm Z jp + Z ieci ( ) (8) where Zjp=Z1jp+Z2jp is the “off” keys resistances of j 2Z jcm (Z en + 2Z is ) CC off-channel, Z ieci = , Zjcm is the 2Z jcm + Z en + 2 Z is common mode resistance in j CC off-channel; 2 Z en Z jcm 2 Z jcm + Z en + 2 Z is k1 = , , k2 = Z jcm + Z en + 2 Z is (Z en + 2 Z is )2 δixe=(Z1ixe --Z2ixe)/Zixe is the relative dispersion of equivalent resistance between the applying point of common mode voltage and both IAB differential inputs. Its analysis shows that the AEC value determined by asymmetries of input measuring circuits MCMI in “i” on-channel and input equivalent common mode resistances, depends on the number of n measuring channels. Indeed, equilibration of input measuring circuits is time-dependent. These schemes are symmetric for a particular object and measuring current circuit parameters MCMI in certain working conditions. However, while measuring circuit reconfigures or working conditions changes, this symmetry is broken. In practice, tend to reduce the AEC value ensuring sufficiently high insulation resistance of common bus IAB at applying point . Further minimization of common mode voltage of this error value is possible automatically by AEC adjusting. Analysis of Equations (1), (5)-(8) envisages that the MCMI AEC substantially depends on the number of n measuring channels. This is especially true for equivalent values of input offset voltage, input currents, input impedances IAB and resistances “off” keys CC. For the relay switches, the CC implementation can significantly diminish the equivalent input currents and resistances impact. However, the residual voltage relays significantly increases AEC value. The switching channel speed MCMI has to be small. If the electronic keys apply in the CC, located at the MCMI input, the keys residual voltages are eliminated only. To diminish these errors components, is suggested to set the smart transducer with input IAB as close as possible to the sensor output [2, 7]. It virtually eliminates the errors caused to CL and CC parameters because output signals of such transducers are standard high-level electrical signals that can submitted straight to standard ADC inputs. The problem of MCMI design significantly complicates when the common mode voltage exceeds electric strength of CC keys. Three-wire sensors connecting and respectively reciprocal isolation of measuring channels are recommended for these errors appreciable minimization. 4.2. Analysis of Properties of Isolated Multi-Channel Measuring Instruments The relative isolation of measurement channels is suggested due to several reasons. The first one is necessity to protect the MCMI electrical circuits of the measured object against spark and/or explosive damage (Fig. 3). Additionally it needs to connect IB to the every measuring line. The IB inner resistance value can reach hundreds Ohms. It could cause AEC value magnifycation due to passing the leakage currents through the mentioned resistance. F.i. under regulations, the insulation resistance of power networks has not been less than 40 MOhm. Then the passing leakage current Ір of grounded measuring object does not exceed 220 V / 40 MOhm ≤ 5 µА. This current can produce voltage drop UixiB=Ip(Zix+ +ZiB)=10 mV on the inner resistances (Zix≈ZiB≤1 kOhm) of sensor and IB, which is considered as MCMI AEC. Isolation for every measuring channel permits diminishing the impact of the potential difference that emerges between grounding points of measuring object and MCMI. These potential differences are generated by powerful sources due to the leakage currents passing through resistances of ground point and earth. Their value may reach hundreds of volts (electric transport, melting furnaces, converters) can cause these interferences. It is impossible to exploit switches in such conditions. 25 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 22-29 at “i” “on” measuring channel, kiA is the ІАі transducer coefficient. at “i” on-measuring Input equivalent voltage channel, due to its equivalent common mode voltage , presents as: U ixcA = U ic Fig. 3. Scheme of multi-channel measuring instruments with isolated channel. To minimize the common mode interference, three-wired CL connecting sensors apply. The screen serves as a third wire, which protect two information CL lines between the sensor output and MCMI input (Fig. 3). By the sensor side this screen should be connected to the point of applying the common mode voltage if it available. Moreover, it needs to connect MCMI to the screen at the CL end. MCMI screen must have the high insulation resistance concerning the measuring circuit. The caused CL error significantly rises if CL length substantially grows. Therefore, SST implementation decreases the afore-mentioned error, owing to the low cost chip components [2-3]. Especially it is inherent in the isolated amplifiers. Output voltage IA MCMI (Fig. 3) is high enough for direct interface with standard ADC. Then the relay keys can operate in CC unit. Output voltage IAi is given: expression =( + +e + + + ) (1 + + , )+ (9) where e = e + e + , = + , = + , = + , e , e is the offset voltage ІАі and residual voltage ІВі , is the input accordingly, δinA≅ZixA/ZinA, , current and resistance ІАі respectively, , is the isolation resistances between ІАі input and output, common bus and screen ІАі accordingly, і ≅ / , = + ) ( + +∑ + is the equivalent output voltage IAB, eio is the offset output voltage ІАі, is the equivalent input voltage of “i” on-measuring channel caused by equivalent common = + , is the mode voltage equivalent input voltage of “i” on-measuring channel caused penetration of measuring voltage Ujx other offmeasuring channel, сА is the equivalent input voltage of “i” on-measuring channel caused penetration of equivalent common mode voltage = + other off-measuring channel, e =e +e , , is the resistance and voltage between grounding points of ІАі and measuring object 26 Z 2ixe Z ⋅ iek Z icm + ZiG + Z 3iis Z 2iis (10) Analysis of latter clarifies at minimization of error that should be provided firstly at voltage value small resistance of screen and secondly by high value of the screen insulation resistance concerning the measuring scheme. From comparing the latter equation and Equation (6) we conclude that the error value х caused by common mode voltage at “i” on-measuring channel is reduced in / times. For example, it occurs if IA AD210 type Analog Devices is used and is provided the screen resistance ≤10 Ohm at ordinary values ≃ ≃ of insulation resistance 240 V / 2 µA=1,2·108 Ohm [15]. Also, if select the ≤ 2500 equal to the common mode voltage maximum isolation voltage of the same IA type at + ≃40 МOhm, the equivalent input voltage is ≤2500(10/1.6·108)·(103/1.2·108) ≅ 41 nV. The х latter is negligible for most application cases. Reduced to an IAB input equivalent voltage х at “i” on-measuring channel caused by penetration of measured voltage х all the rest (n-1) CC offchannels is unable to change it comparing with value of obtained from (7). Threshold AEC value caused by penetration of equivalent common mode voltage = + of all off-measuring channels at “i” on-channel compared to the expression (8) / times. For above-given decreases in conditions, adopting AEC becomes negligible mainly. Analysis of (9) envisages that both AEC and MEC input circuit of IA significantly affect the measurement accuracy in working conditions. In order to raise it, the manual zeroing and conversion factor IA specification apply. However, while operating the values of both factors vary substantially, worsening MCMI accuracy. 4.3 Error Correction of Multi-Channel Measuring Instruments Analysis ratio of (1) to (9) helps to identify the AEC significant affect the MCMI metrological properties in working conditions. For their adjustment, manual MCMI zeroing applies [15]. Usually CPS MCMIs are considered as distributed systems, measurable objects of which are located at appreciable distance from each other. So, suppose that it is almost impossible to carry out instrument’s zeroing of every measuring channel at manual mode. To automate the MCMI error adjusting process it Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 22-29 seems to be better the inverting switching input of gauging signals; input polarity switch would be located as close as possible to sensor output [14, 16]. If applicable IA, this switch should be near-by the input amplifier (Fig. 4). During the measurement of i on-channel, the sensor signal Uix is received with the measurement result code Nіх: Nix = 0.5 ( N1ix − N2ix ) = = 0.5kiAk ADC (Uix + ΔiAx ) , (12) where , is the measurement results codes of sensor output signal for direct and reverse polarity of PISXi connection, ∆ = 0.5[( А + Х )∆ Х + ∆ Х ХХ ] is the uncorrected AEC value, Х , ∆ Х is the average value and absolute dispersion of reverse currents keys PISXi respectively, ∆ Х = Х + В + РХ , РХ , ∆ Х is the average value resistance and resistance match between on-keys PISXi respectively. The determined value is transformed in: Nix = N1ik Fig. 4. Multi-channel measuring instruments with remote errors correction: PISX, PISC is the polarity inverse switch of measuring and calibration values accordingly; CU is the control unit; CCVD is the code-control voltage divider; SW is the switch. In working conditions, MEC MCMI is characterized by significant dispersion (up to ± 2 %). Therefore, the problems in application emerge. We propose to perform the MCMI remote calibration basing on the code-control voltage measure (further CCVM) located in every measuring channel. It can be realize due to availability of modern microelectronics. During calibrating the output signal of CCVM Uіk=kE0i feed the measuring channel input, so the sensor measuring output signal Uіx is disconnected. The available set of calibration codes is transmitted to the CCVM from CNT MSMI. Output voltage of CCVM is converted in calibration result code Nіk, where i is the number of channel; k takes the values 1, 2, ..., K (K is the maximum number of calibration codes meanings). While i on-channel has to be calibrated, it sends the Nіk code: Nik = 0.5 ( N1ik − N 2ik ) = = 0.5kiA k ADC (U ik + ΔiAc ) , (11) where , is the measurement results codes of the calibration voltage Uіk=kE0i for direct and reverse polarity of PISC connection, E0i is the reference voltage, is the ADC conversion factor, )∆ +∆ ] is the ∆ = 0.5[( А + , ∆ is the average uncorrected value AEC, value and absolute dispersion of reverse currents keys ,∆ is the average value PISCi respectively, resistance and resistance match between channels “on” keys PISCi accordingly. U Δ −Δ Uix + ΔIAx = N1ik ix 1 + IAx IAc Uik + ΔIAc Uik Uik (13) MCMI MEC depends on performance of the reference voltage Еоі and on the CCVD conversion coefficient k. For the estimation of uncorrected errors limit we take the ordinary values for ADG787 switch [17] ( А =30 nA max, ≃20 nA max, Х ≃ ≃ ≃ 3.35 Ohm max, ∆ Х, ≃0.1 Ohm), ≃0.05 ≃ Х+ В ≤1 kOhm max, ∆ Х ≃ ∆ ≃5·10-2·2·10-8=1 nA, then ∆ ≃4 nV, ∆ ≃0.1 V. By performed while calibrating procedure the AEC uncorrected values become negligible for practical requirements. Then remains the unadjusted AEC, and its value is determined during the measurement by the total resistances of the sensor and IB, and also by the reverse currents differences of on-keys PISX and PISC. Studies envisaged that this difference does not exceed several per cent for modern MOS chips. So, it can be realized accurate MCMIs. To insure high accuracy in working conditions, we propose method of remote calibration. It should be measured the actual output voltage for k different factors of division in every measuring channel at training stage of MCMI (on the step of adjustment). All K values of output voltages к are measured by accurate voltmeter for every of k division factor getting the codes array Nuik. Then the same voltage values are measured by MCMI and received other . In MCMI memory the high-mentioned codes array Nuik by known method is entered and the /Nuik is appropriate calibration coefficients Kik= computed. They are fixed in MCMI memory and further apply at determination of the measurement result code : Nix = KiKUix (14) Reference voltage values К =kEoi of CCVD vary during work. To reduce the impact of these changes, must be selected the stable electronic components, for example, with parameters of 27 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 22-29 reference voltage Еіо ⁄ ≤ ±2 ∙ 10 1/ and CCVD і ⁄ ≤ ±2 ∙ 10 1/ in the temperature range -25...+85 oC [18-19]. Then changing values к and therefore Кік would not exceed ±0.026 %. This range of variation is satisfactory for measurements. The high temperature stability of suggested CCVM structure and the individual calibration possibility while debugging make it possible to obtain reference voltages within wide range, from a few millivolts to nearly reference voltage. The same feature can apply to verify the MCMI metrological properties directly on-site by the portable CCVM. During adjusting it should be accurately gauged the k output measures к for every measuring channel. As regulations require, these points have to be arrange evenly along the measuring range. Value that is close to the maximum measuring range, can be used as a working standard at i on-measuring channel. MCMI on-site verification by means of CCVM excluding measuring sensors, assures the particular possibility of metrological checking of all channels. Portable CCVM is protected against varying operation conditions by implying the protective and preventive methods. Obviously, it needs to develop appropriate software for the prompted method implementation. 4.4. Experimental Investigations of Code Control Voltage Measure A number of MCMIs has been implemented before, and their metrological maintenance is not sufficiently correct. Indeed, for quick calibration already active MCMIs the market offers several types of portable calibrators. Their main drawbacks are complexity and necessity of calibration results correction caused by possible changes of working conditions. Simultaneously calibrators drift themselves, and there emerge the contact EMFs in connection points to MCMI. To avoid them, we suggest the voltage calibrator (further VC) with error self-correction (Fig. 5). The foundation of AEC automatic adjustment bases on two synchronous polarity switches operation that are located at the input and output of calibrator PIS1 and PIS2 respectively. The output voltage VC averaging follows this step. The averaging can be carried out both in digital form and in analogue form when using LPF. Then the digital processing of results is the sum of even number of output signals VC conversions. For PIS1 and PIS2 one polarity of calibrator output voltage Uk1і we receive: U k1i = μiH [E0 H + e1 ]( 1 + δ μ i + δ E ) + e2 , while for the other polarity is defined Uk2і: U k 2i = μiH [E0 H − e1 ]( 1 + δ μ i + δ E ) − e2 , In working conditions, calibrator requires the periodic manual AEC adjustment, which prolongs duration and complexity of metrological works. In such way, we propose to provide the AEC automatic correction. 28 (16) where μіН is the nominal code of CCVD (DAC), Е0Н is the nominal value of reference voltage, δμі, δЕ is the relative error of CCVD and reference voltage respectively; e1, e2 is the AEC buffers of input and output voltages respectively. At averaging, the output voltage value Ukі of calibrator for the current code μі, is determined as: U ki = 0 ,5(U k 1i + U k 2i ) = 0 ,5 μiH E0 H ( 1 + δ μ i + δ E ) (17) Results of modelling of designed scheme coincided with experimental results. In experiments, for calibrator was selected reference voltage with output voltage E0=100 mV, and DAC codes change from 0 to 1 in increments of 0.25. To test the AEC impact on the obtained results we have been submitted е1, е2 = 15 mV from the stable power supply. It was received two sets of experimental results: output CCVM voltage without AEC, UК1, mV and the output CCVM voltage with AEC source, UК2, mV (Table 1). Table 1. Investigation results of code control voltage measure experimental unit. No. 1. 2. 3. 4. 5. Fig. 5. Scheme of portable code control measure with automatic errors correction: PIS1, PIS2 is the first and the second polarity inverse switch, LPF is the low pass filter, G is the correction frequency generator, μ is the control code of CCVD. (15) μН 0 0.25 0.5 0.75 1 UК1, (mV) -0.003 25.007 50.015 75.023 100.031 UК2, (mV) -0.003 25.007 50.014 75.022 100.031 The AEC imitator values are selected a priori more the possible values of equivalent offset voltage amplifiers, which use in the calibrator scheme. Simulator equivalent voltage AEC housed in various characteristic points layout VC, namely the inputs, outputs and all feedback loops of operational amplifiers. Discrete resistor voltage divider is used. The CCVM output voltage is measured by multimeter Picotest M3511A, which has those technical parameters as measurement range DCV 100 mV, Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 22-29 accuracy 0.012 % in 1 year, least significant digit at average 1 µV. If the experiment results analysis shows, that numeric data at the third and the forth columns Table 1 not differ by more than one least significant digit of using voltmeter (±1 μV). This confirms the theoretical assumptions for the possibilities of remote automatic calibration of measuring channels MCMI CFS. 5. Conclusions 1) Basic error factors of multi-channel measuring instruments due to equivalent input voltages and currents shifts, the influence of the switch channels, connecting lines, non-informative parameters of sources of measuring signals, common mode voltages, measured voltage penetration of other disconnected channels are considered. It is shown that the errors inherent in multi-channel measuring instruments with isolated channels can significantly exceed the similar ones of traditional structures. 2) Remote adjustment errors for developed multichannel measuring instruments of CFS are suggested to carry out by means of embedded code-control voltage measures. For both multichannel measuring instruments and embedded code-control voltage measures, the additive error components correction is proposed to perform by inverted switching implementation. For multiplicative error component correction is suggested to implement code-control voltage measures based on stable voltage reference source and DAC multiplier. 3) For on-line errors correction of multichannel measuring instruments, the portable and compact code-control voltage measures with implementation of the input signal double inverting method are suggested. As result, the obtained additive error value does not exceed ±1 μV ensuring the high accuracy and stability of mentioned instruments for CPS operation. Acknowledgement The scientific results, presented in this article, were obtained in the frame of research project number 0115U000446, 01.01.2015 - 31.12.2017, financially supported by the Ministry of Education and Science of Ukraine. References [1]. ISO 10012:2003. Measurement management systems – Requirements for measurement process and measuring equipment. [2]. Smart Sensor Systems, edited by G. Mejer, John Wiley & Sons, Ltd, 2008. [3]. J. W. Gardner, V. K. Varadan, O. O. Awadelkarim, Microsensors, MEMS, and Smart Devices, John Wiley & Sons Ltd, Chichester, England, 2001. [4]. ATEX directive 2014/34/EU, Decision No. 768/2008/EC of the EPC of 9 July 2008 on a common framework for the marketing of products, Brussels, 21.11.2011 under reference COM (2011) 763 final. [5]. Explosion protection. Theory and practice. PHOENIX CONTACT GmbH & Co. KG, Web Portal (http://www.phoenixcontact.com). [6]. C. Lehrmann, D. Seehase, M. Sattler, M. Gruner, Latest news on explosion protection, Ex-Manual, VEM, 2013. [7]. Data Acquisition Handbook, A Reference for DAQ and Analog & Digital Signal Conditioning, 3rd Ed., Measurement Computing Corporation, 2012. [8]. Scott Wayne, Finding the Needle in a Haystack: Measuring small differential voltages in the presence of large common-mode voltages, Analog Dialogue, 34, 1, 2000, pp. 1-4, (http://www.analog.com//) [9]. Instrumentation and Measurement, Analog Devices Inc., 2015, (http://www.analog.com/en/applications/markets/instrumentation-and-measurement.html). [10]. Data Acquisition Systems, Omega Company Products, One Omega Drive, Stamford, CT 06907, 1-888-TC-OMEGA USA, (http://www.omega.com/techref/pdf/dasintro.pdf). [11]. Data Acquisition (DAQ) Fundamentals, Application Note 007, National Instruments Corp., August 1999: (http://physweb.bgu.ac.il/COURSES/SignalNoise/dat a_aquisition_fundamental.pdf). [12]. Data Acquisition, Cole-Parmer Instrument Company, 2015, (http://www.coleparmer.com/Category/Data_ Acquisition/741). [13]. Fluke Multifunction Calibration Tools, Fluke Inc., 2015. (http://en-us.fluke.com/products/multifunctioncalibrators/) [14]. V. Yatsuk, P. Malachivsky, Methods of Increase of Measurement Accuracy, Beskyd-bit edition, Lviv, 2008 (in Ukrainian). [15]. Precision, Wide Bandwidth 3-Port Isolation Amplifier AD 210, Analog Devices Inc., Web Portal (http://www.analog.com/media/en/technicaldocumentation/data-sheets/AD210.pdf). [16]. Yatsuk V., Stolyarchuk P., Mikhaleva M., Barylo G., Intelligent Data Acquisition System Error Correction in Working External Conditions, in Proceedings of the 3rd IEEE Workshop on Intelligent Data Acquisition and Advanced Computing Systems: Technology and Applications (IDAACS’05), Sofia, Bulgaria, September 5-7, 2005, pp. 51-54. [17]. 2.5 Ω CMOS Low Power Dual 2:1 Mux/Demux USB 1.1 Switch ADG787, Analog Devices Inc., Web Portal (http://www.analog.com/media/en/technicaldocumentation/data-sheets/ADG787.pdf). [18]. Datasheet Catalog.com Web Portal (http://www.datasheet catalog.com/). [19]. Catalog ELFA DISTRELEC Web Portal (http://www.online-electronics.com.ua/catalog/). ___________________ 2015 Copyright ©, International Frequency Sensor Association (IFSA) Publishing, S. L. All rights reserved. (http://www.sensorsportal.com) 29 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 30-36 Sensors & Transducers © 2015 by IFSA Publishing, S. L. http://www.sensorsportal.com Metrological Array of Cyber-Physical Systems. Part 12. Study of Quantum Unit of Temperature Svyatoslav YATSYSHYN, Bohdan STADNYK National University ‘Lviv Polytechnic’, Institute of Computer Technologies, Automation and Metrology, Bandera str.12, Lviv, 79013, Ukraine Tel.: +38-0322-37-50-89 E-mail: [email protected] Received: 30 August 2015 /Accepted: 28 September 2015 /Published: 30 September 2015 Abstract: The reference measure of temperature may be embedded in appropriate unit of Cyber-Physical System. Whereas this measure made on the basis of fundamental constants of matter would be installed in such System, the latter will get an extra precision. It is shown that metrologically correct Kelvin redefinition which would be changed by CODATA to 2018 is insufficient to create a Temperature Standard on the basis of fundamental constants of matter. New approach to the mentioned Standard and firstly to the Quantum Unit of Temperature is developed. Copyright © 2015 IFSA Publishing, S. L. Keywords: Cyber-physical system, Quantum unit of temperature, Kelvin redefinition, Quantum standard of SI units. 1. Introduction Cyber-Physical System technologies (further CPS) have to utilize the sophisticated metrology equipment for production lines. This involves the estimation of the comparability of CPS measurement instrument by self-verification. Development of portable, highly-precise devices is able to provide inplace precision measurements. The studied quantum standard may be recommended firstly to apply as intrinsic standard; such a standard does not need permanently recurring measurements against the realization of the SI unit in order to validate its accuracy. Intrinsic standards are important instruments in disseminating accurate measurements in an efficient way for instance in CPS operation. At the end of last century there were successfully implemented the atomic (molecular) standards of SI units. As result of quantum discreteness qualities 30 extend in nanosphere, for example, the ability appears to realize not only measuring instruments but to create also the standards of measurands. Currently temperature remains the last only value among seven main units of International unit system that is still not regulated at the atomic (molecular) and hence much higher level in terms of accuracy. 2. Shortcomings New created CPSs often require self-verification of temperature values to ensure their quality work. The existing standards lose accuracy characteristics by several orders while uploading them to the end user that is actually considered a normal metrological practice. However, such practice cannot be deemed adequate for the advanced machines. So, they need the development of intrinsic standards of temperature that http://www.sensorsportal.com/HTML/DIGEST/P_2720.htm Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 30-36 could be embedded into CPSs ensuring their precision operation. 3. Goal of the Work Goal of the work consists in the provision of the temperature support for Cyber-Physical System operation by the studied temperature-measuring instrument of a new generation, namely considering, after CODATA redefinition of the “Temperature”, the next step in creating the Quantum Unit of Temperature based on the fundamental constants of matter. 4. Primary Thermometry and Quantum Units of Temperature and the relative offset from the CODATA 2010 value is +1.9×10-6 [3]. Considering the mentioned primary methods of temperature determination and having determined Boltzmann constant with a very small error, scientists could develop the unit of a temperature scale due to the energetic/power unit, endued with a certain determination error. Unfortunately, these hard works, details of which were descript earlier in [3, Table 1 (Summary uncertainty budget for a determination of Boltzmann constant by QVNS thermometry)], practically not eliminate the major principal shortcoming. It consists in the necessity to calibrate thermometer at TPW temperature. Other researchers are unable to get rid of traditional calibration and only replace the outdated method by the modern one. Nevertheless the replacement of the temperature measuring instruments for the energy ones will raise especially severe difficulties precisely in the area of ultralow energies gauging [4-5] that can be associated with minimal energy (temperature) unit. 4.1. Quantum Energetic Unit In the chain of leading metrological centers (USA, GB and other countries), through several years the intensive endeavors of elaborating and assuring the unit of temperature scale in the form of a quantum energetic unit (minimal by size a discrete value of energy or heat energy that can be defined, established and fixed by the experimenter) are carried out at the high methodological level [1]. The recommended by Ia. Mills, et al. [2] new format of unit with new definition is the next. The kelvin, K, is the unit of thermodynamic temperature; its magnitude is set by fixing the numerical value of the Boltzmann constant to be equal to exactly 1.380 65 … ×10-23 when it is expressed in the unit s-2m2kgK-1, which is equal to J·K-1. The effect of proposed definition is that the kelvin is equal to the change of thermodynamic temperature that results in a change of thermal energy kT by 1.380 65 … ×10-23 J/K. Then using k rather than TTPW to define kelvin better reflects modern practice in determining thermodynamic temperature directly by primary methods, particularly at very high and low temperatures. The unit of thermodynamic temperature, the kelvin, will be redefined in 2018 by fixing the value of the Boltzmann constant, k. The present CODATA recommended value of k is determined predominantly by acoustic gas-thermometry results. To provide a value of k based on different physical principles, purely electronic measurements were performed by using a Johnson noise thermometer to compare the thermal noise power of a 200 Ώ sensing resistor immersed in a triple-point-of-water cell to the noise power of a quantum-accurate pseudo-random noise waveform of nominally equal noise power. Measurements integrated over a bandwidth of 550 kHz and total integration time of 33 days gave a measured value of k=1.3806514(48)×10-23 J/K, for which the relative standard uncertainty is 3.5×10-6 4.2. Classic Set of Primary Methods of Thermometry Primary thermometry envisages that particular measuring instrument concerns the concrete measurand (T) that can be defined by the calculating the gained results excluding other unknown quantities and applying only fundamental constants of matter as proportionality factors. Secondary thermometry develops the methods of measurement of another kind temperature than thermodynamic one or the methods using any dependence of properties on temperature, and then some points of received dependence are ascribed the certain values of thermodynamic temperature. The classic set of primary methods includes five methods of thermometry: noise, gas, acoustic, optical, and magnetic. Those methods are based on the fundamental physical laws whose mathematical descriptions comprise the thermodynamic temperature. Among them the gas and optical thermometry have gained widest application in the reproducibility of thermodynamic temperature. The last problem is particularly inherent in Nanothermometry where methodical errors rise when the thermal capacities or sizes of the thermometry-processed body and the thermometer become comparable [6]. Noise method of thermometry still remains in a state of metrological elaboration. In the conditions of durable development of nanotechnology with its unrepeatable measurements the uncertainty accumulation causes a substantial decrease in authenticity of information extracted from the received experimental results. From Johnson-Nyquist equation we could state that the concrete and precise determination of rootmean-square voltage and hence electric power implies the usage of a studied object with a priori known value of electric resistance. At precisely defined Boltzmann 31 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 30-36 constant a relative instrumental error of the measured voltage is formed as the sum of notified values: the relative errors of determining object resistance and temperature, and also the relative error of specifying frequency bandwidth of measurement. It indicates the undoubted advantages (resistance and temperature are known with high accuracy) of employing the substance of its sensitive element in systematic noise studies. Recently the considered method is implemented as second one in addition to acoustic gasthermometry method aiming the determination of the Boltzmann constant and trying to introduce the metrologically obvious step in redefining the notion “Temperature” by CODATA. The noise method is increasingly applied in Nanothermometry [7]. Particular attention is paid to the investigation of 1/f γ electrical noise and especially to their connection with changes of entropy of studied thermodynamic system [8], including the nanodimensional system [9]. Similar results were obtained by us in [10]. The deduced equation of noise thermometer transformation function relates the power of electric noise Pel with the thermodynamic temperature Т through the dissipation rate of entropy dS/dt: T dS dN = −eφ = Pel , dt dT ( ) ΔT is the change of monocrystal temperature. It enables to obtain more exact temperature readout at known temperature dependence of Raman shift ν 0 . Raman thermometer with CNT calibration artefacts is characterized by high accuracy and is inherent in possibility of in-place self-calibration [11]. Magnetic method of thermometry is the method of measuring temperatures lower than 1 К. It is based on the temperature dependence of magnetic susceptibility χ of a paramagnetic substance. Gas method of thermometry operates basing on volume expansion effect of substance. Here the measuring temperature changes ΔT are proportional to the changes of Sensitive substance volume. This method seems to be a separate issue that requires a special approach. Acoustic method of thermometry is not considered by the fact that in this article we study only the methods related to electronphonon interaction. (1) where е is the electron charge; N is the amount of charge carriers. Inside the researched substance the latter could be estimated by following the thermodynamic approach. Earlier we have followed from the most probable physical processes in sensitive dS ΔS substance under the given conditions that: , ≈− τ dt here ΔS is the entropy changes taking place as consequence of a relaxation process with constant τ. Raman method of thermometry is a contactless temperature-measuring method. The measurement of the solid body surface temperature with taking advantage of a Raman phenomenon is related to one of few methods of primary thermometry being realized with the help of a thermometer whose state equation could be written in an explicit form, avoiding the involvement of unknown constants dependable on temperature. The given method helps to measure the temperature for objects ranged from 100 nM to 100 µM as well as within this from cryogen till midhigh temperatures, which in addition does not demand calibration before measurement, could be distinguished. For instance, temperature measurements are made by exciting the quantum dots with a laser, to obtain their emission spectra. The similar application of Raman thermometer hits in measuring the temperature and diameter of carbon nanotubes. Raman thermometer in Nanothermometry is the thermometer which due to small diameter of the He-Ne or another laser of continuous action enables to reduce the size of a thermometring zone to tens µm and lesser. Raman method gives opportunity of electron-phonon interaction study. Wave number of 32 optical phonon of Stokes component is strongly dependent on temperature. For example, for silicon monocrystal this dependence in the temperature range 300 ... 400 K is linear: ν 0 cm−1 = 0,025ΔT , where 4.3. Promising Methods of Thermometry Coulomb blockade thermometer is the primary thermometer based on electric conductance characteristics of tunnel junction arrays. The parameter U½=5.439NkBT/e (kB is the Boltzmann constant), the full width at half minimum of the measured differential conductance dip over an array of N junctions together with the physical constants, provide the absolute temperature [12]. So, half width U1/2 depends only on the constants of matter and known parameter N that seems to be quite close to design of primary thermometer based on fundamental constants of matter. A typical Coulomb blockade thermometer is made from an array of metallic islands, connected to each other through a thin insulating layer. A tunnel junction forms between the islands, and as voltage is applied, electrons may tunnel across this junction. The tunneling rates and hence the conductance vary according to the charging energy of the islands as well as the thermal energy of the system. In order for the Coulomb blockade to be observable, the temperature has to be low enough so that the characteristic charging energy (the energy that is required to charge the junction with one elementary charge) is larger than the thermal energy of the charge carriers. For capacitances above 1 femtofarad (10−15 farad), this implied that temperature has to be below about 1 kelvin. This temperature range is routinely reached for example by 3He refrigerators. Thanks to small sized quantum dots of only few nanometers, Coulomb blockade has been observed currently above liquid helium temperature, up to room temperature. Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 30-36 Free electron gas primary thermometer is based on bipolar transistor, temperature of which is extracted by probing its carrier energy distribution through its collector current, obtained under appropriate polarization conditions, following a rigorous mathematical method. The obtained temperature is independent of the transistor physical properties as current gain, structure (homo-junction or heterojunction), and geometrical parameters, resulting to be a primary thermometer. This assumption has been tested using off the row of silicon transistors at thermal equilibrium with water at its triple point. The obtained transistor temperature values involve an uncertainty of a few milli-Kelvins. Further free electron gas primary thermometer has been successfully tested in the temperature range of 77…450 K [13]. Remark: we are inclined to consider in this paper the Transistors, FETs or other micro- and nanodimensional electronic devices as quasidotted objects. Rather complicated Josephson junction noise thermometer is a thermometer, operation principle of which is built on the Josephson Effect and readouts are proportional to thermodynamic temperature (further TT). This thermometer can be obtained by means of Josephson element, which contains two superconducting plates separated by a thin oxide layer. An element is switched in measuring circuit through a point contact between sharpened wire and plate of the superconducting substance. Charge carriers of element may exist as the electrons that are scattered by lattice, and as Cooper pairs that create the superconductivity effect, but are not scattered by a lattice [14]. Due to tunneling, electrons as well as Cooper pairs would flowed through oxide layer resulting in emerging a stationary (DC of superconductivity that not exceed particular value I0 can pass through the oxide layer without voltage drop) and a non-stationary Josephson Effect. The latter consists in emerging the oscillations at frequency f 0 = 2e u 0 , where h is the Planck h constant, while the DC voltage u0 is applied to Josephson element. Therefore it becomes a generator of sinusoidal current. Thermal noise that arises in the Josephson element creates the current pulsations resulting in monochromatic frequency blurs of signal. Since the thermal noise inherent in a normal distribution, then extend of signal frequency mode has a typical bypass of Gaussian curve. Half-width of the spectral line of thermal noise in Josephson element, measured by radio spectrometer, is given by the equation Δ f 0 = 4π k BT r ( 2e 2 Ir ) (1 + ), h u0 where T is the TT, r is the resistance, which shunts the element in measuring scheme, I is the current that passes through the element. 5. Investigation in Creating the Quantum Unit of Temperature Temperature in nanothermometry is the statistically formed value of quantity, determined by the inner energy of a body of sufficient sizes for purpose of applying the thermodynamic consideration to this body. It seems to be one of the fittest terms among the considerable number of temperature definitions which try to identify temperature in nanothermometry. A thermodynamical notion of temperature is related to heat exchange between two systems. The quality of supplying or not to the balance among themselves under some predetermined conditions pertains to all macroscopic systems. The necessity to characterize a state of thermodynamic systems by some specific quantity becomes obvious. So, a notion “Thermodynamic temperature” has been introduced for this purpose. The objective measurement of temperature is possible due to the transitivity of a thermodynamic equilibrium. Therefore there is a possibility to compare the object temperatures among themselves without the objects’ per se contact. Current definition of the unit of thermodynamic temperature, kelvin, is based on a material artifact, namely, the triple-point-of-water temperature [15, p.175]. The latter depends on the isotopic composition, purity etc. and therefore is not precise value. It can be realized with uncertainty about 10-7. Temperature as a physical value that characterizes the inner energy of bodies is not being measured directly nowadays. All usable measuring instruments transform temperature in some other physical value that could be used immediately. Temperature that is defined by indices of a thermometer of concrete type is named the empirical temperature. Developing the apparatus of statistical physics, try to link the term “Temperature” with basic constants of microphysics, on the one hand, and threshold sizes of nanoparticles where this notion is still applicable, on the other hand. The special significance is bestowed to the definition of the minimal particle size where the notion “local temperature” could be adopted, i. e. the temperature at which a part of thermodynamic system remains in a canonical state, and the energetic distribution of electrons corresponds to the exponentially falling oneparametric function. In the world of nanotechnology the combination of measuring technologies and theoretical research is getting more and more significant, since it concerns a single non-repeated measurement. So, everybody try to ensure himself the final measurement result by the next generation of Standards [2] and by introducing a number of metrological procedures such as selfverification, self-validation and self-calibration [16]. It is discussed below the possibility of researching the most contemporary measure of temperature on the basis of fundamental constants of matter with involvement of the Standard of electrical resistance on the basis of Inverse of Conductance Quantum [17] as well as the Standard of voltage based on the Josephson junctions [12] that can produce voltage pulses with 33 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 30-36 time-integrated areas perfectly quantized in integer values of h/2e. The synthesized voltage is intrinsically accurate because it is exactly determined from the known sequence of pulses, the clock frequency, and fundamental physical constants. Thus, we consider the investigation of the electrical resistance value of which is based on Klitzing constant, and of the electrical voltage standard on the Josephson Effect for exact frequencyto-voltage conversion, combined with the clock. As the mentioned resistance we propose to study one of widespread FET constructions, namely the CNTFET with built-in CNT [18] which has to be superconductive. Source and drain have to be manufactured from dissimilar metals that form the thermoelectric pair through the CNT. The latter, being in superconductive state, is inherent in resistance which corresponds to 25812.807 557 ± 0.0040 Ohms, due to transient resistance of contacts. While studying the dissipation of electric power ( I 2 R = U 2 / R ) on such an electric resistance in temperature measurement area: 3 E = U 2 Δt / RKl = I 2 RKl Δt = N k BTa, 2 (2) was noted that we are able to estimate the change of ΔQ Ne = TT T, or substituting this equation by I = Δt Δt (Δt is the time period) we clarify it to: ( Ne)2 h 3 Δt = N kBT , 2 2 (Δt ) e 2 Otherwise, the temperature increment reduced due to one-electron relaxation on phonon of the superconductive CNT junction with source/drain and due to unit time application is defined only by fundamental constants of matter (h and kB); it is equal to 2h·1s./3kB =3.2 ×10-11 K. Hence, the proposed in [2] figures regarding interrelation and inter-definition of basic SI units and the principles of study of the mentioned units through the fundamental constants of matter are modified by the results of the performed study (see Fig. 1 and Fig. 2). Fig. 1. Interrelation and inter-definition of basic SI units: blue arrows show the revealed relationship of the studied unit T with unit I,A (by unit V and unit R) and with unit t,s. (3) when the electrical current is formed per unit time by 3 N conduction electrons that transfer the energy k BT 2 to the atoms of matter. From here the TT jump ΔT at current transmission I through superconductive CNT (cooling is considered to be negligible), is defined as: ΔT = 2 hN 2 hI = ,K 3k B Δ t 3k B e (4) On condition of power supply from Johnston junctions array it appears an opportunity to pass a discrete particular number of electrons through nanotube of FET. Then the resulting value of the temperature increase of atom reduced to one electron that was scattered on its phonon at the unit time is identified as Reduced Quantum Unit of Temperature: ΔT Δt →1 s . N →1 = TUR = 2h 3k B K s. ⋅1[ s.] = = 3.199 49342 ⋅10 −11 ≈ 3.2 ⋅10−11 [ K ] 34 , (5) Fig. 2. Principles of mentioned units study through the fundamental constants of matter: elimination of interrelation between unit m and unit T as well as the emergence (blue arrow) of interrelation between unit I,A and unit T,K. In such a way the Reduced Quantum Unit of Temperature that is independent of kind of matter and recommended in the creation of Temperature standard, can be regarded. It would be the Standard based on a 2 quantum effects (von Klitzing Effect and Josephson Effect) and, having been measured against the SI system of units, has a certain value with uncertainty determined by sum of 2 uncertainties: of Planck constant and of Boltzmann constant [19] which Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 30-36 together make its total relative uncertainty value that equals to 59.2×10-8. The last value also includes the relative standard uncertainty of atomic unit of time that is 5 orders of magnitude smaller (5.9×10-12 [19]) and therefore is neglected at this stage of study. Operating mode is as follows. The studied appliance is propose to supply by short (~10-3 s.) pulse voltage consequences, effect of which is measured at the 2nd stage at power absence. Measuring temperature with minimal methodical error is easiest with the builtin thermocouple. It's enough at manufacturing the CNTFET to make source and drain from two dissimilar conductive metals (f.i. Ni and Cu). Superconductive CNT as the 3rd intermediate body forms a quasi junction of produced thermocouple. 6. Conclusions 1) Advances in Cyber-Physical Systems are impossible without the temperature gauging that demands the continuous development of experimental techniques due to progress in Thermometry and Nanothermometry, namely in creation of Temperature Standard. 2) Expanding the set of Quantum Standards of the SI units towards the study of major pillars of the Temperature Standard on the basis of fundamental constants of matter, becomes possible as a result of emerged opportunities of unique electronic devices, in particular Resistance Standard (on the basis of Inverse of Conductance Quantum) and Voltage Standard (on the basis of Josephson junctions array) combined in addition with the Cesium Frequency Standard. 3) Researching the foundations of creation of the Quantum Unit of Temperature envisages the minimum value of temperature jump caused by electron relaxation on phonon as quasiparticle of atom which is the smallest constituent unit of ordinary matter. 4) It is proved that the Reduced Quantum Unit of Temperature is determined by the electric energy dissipated on CNTFET contacts at passing a current, via ratio of h and kB. The above given RQUT is equal to 3.199 493 42×10-11 K with relative standard uncertainty 59.2×10-8 (at one electron relaxation per unit time). Acknowledgments Authors would like to thank the National University ‘Lviv Polytechnic’ and the Rector, Prof. Yu. Bobalo for the comprehensive support. References [1]. Consultative Committee for Thermometry 2006, Mise en Pratique for the definition of the Kelvin (S’evres, France, Bureau International des Poids et Measures). [2]. Ian Mills, T. Quinn, P. Mohr, B. Taylor, E. Williams, The New SI: units and fundamental constants, Royal Society Discussing Meeting, Jan. 2011. [3]. S. P. Benz, A. Pollarolo, J. Qu, H. Rogalla, C. Urano, W. L. Tew, P. D. Dresselhaus, D. R. White, An Electronic Measurement of the Boltzmann Constant, Metrologia, 48, 3, 2011, pp. 142-153. [4]. M. Hohmann, P. Breitkreutz, M. Schalles, T. Fröhlich, Calibration of heat flux sensors with small heat fluxes, in Proceedings of the Internationales Wissenschaftliches Kolloquium: ‘In Shaping the future by engineering’, Technische Universität, 58, Ilmenau, Germany, 08-12 Sept. 2014, pp. 1-9. [5]. M. Lindeman, Microcalorimetry and transition-edge sensor, Thesis UCRL-LR-142199, US Department of Energy, Laurence Liverpool National Laboratory, April 2000. [6]. S. Yatsyshyn, B. Stadnyk, Ya. Lutsyk, L. Buniak, Handbook of Thermometry and Nanothermometry, IFSA Publishing, 2015. [7]. B. Stadnyk, S.Yatsyshyn, Ya. Lutsyk, Research in Nanothermometry. Part 1. Temperature of Micro- and Nanosized Objects, Sensors & Transducers, Vol. 140, Issue 5, May 2012, pp. 1-7. [8]. V. P. Koverda, V. N. Skokov, Stability of random process with 1/f spectrum for deterministic effects, Journal of Applied Physics, Vol. 83, Issue 4, 2013, pp. 1-5 (in Russian). [9]. F. Gasparyan, Excess Noises in (Bio-) Chemical Nanoscale Sensors, Sensors & Transducers, Vol. 122, Issue 11, November 2010, pp. 72-84. [10]. S. Yatsyshyn, B. Stadnyk, Z. Kolodiy, Development of Noise Measurements. Part 1. Fluctuations and Thermodynamics, Proper Noise and Thermometry, Sensors & Transducers, Vol. 150, Issue 3, March 2013, pp. 59-65. [11]. B. Stadnyk, S. Yatsyshyn, O. Seheda, Yu. Kryvenchuk, Metrological Array of CyberPhysical Systems. Part 8. Elaboration of Raman Method, Sensors & Transducers, Vol. 189, Issue 6, June 2015, pp. 116-120. [12]. CBT – Coulomb Blockade Thermometer. http://metrology.tkk.fi/courses/S-108.4010/2007/CBT_esitelma _1804.pdf. [13]. J. Mimila-Arroyo, Free electron gas primary thermometer: The bipolar junction transistor, Appl. Phys. Lett., 103, 2013, 193509, http://scitation.aip.org/ content/aip/journal/apl/103/19/10.1063/1.4829741 [14]. P. Joyez, D. Vion, M. Götz, M. Devoret, D. Esteve, The Josephson effect in nanoscale tunnel junctions, Journal of Superconductivity, Vol. 12, No. 6, 1999, pp. 757-766. [15]. E. O. Göbel, U. Siegner, Quantum Metrology: Foundations of Units and Measurements, Wiley-VCH Verlag GmbH &Co. KGaA, 2015. [16]. S. Yatsyshyn, B. Stadnyk, Ya. Lutsyk, V. Yatsuk, Metrological Array of Cyber-Physical Systems. Part 1. Challenge of Modernity, Sensors & Transducers, Vol. 186, Issue 3, March 2015, pp. 1-11. [17]. A. J.Giesbers, G. Rietveld, E. Houtzager, et al., Quantum resistance metrology in graphene, Applied Physics Letters, Vol. 93, Issue 22, 2008, pp. 222109-1 – 222109-3. [18]. R. Sahoo, R. Mishra, Simulations of Carbon Nanotube Field Effect Transistors, Internation Journal of Electronic Engineering Research, Vol. 1, Issue 2, 2009, pp. 117-125. 35 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 30-36 [19]. The NIST Reference on Constants, Units and Uncertainty, CODATA Internationally Recommended 2014 Values of the Fundamental Physical Constants, http://physics. nist.gov/cuu/ Constants/index.html ___________________ 2015 Copyright ©, International Frequency Sensor Association (IFSA) Publishing, S. L. All rights reserved. (http://www.sensorsportal.com) 36 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 37-43 Sensors & Transducers © 2015 by IFSA Publishing, S. L. http://www.sensorsportal.com Experimental and Modelling Study of a Piezoelectric Energy Harvester Unimorph Cantilever Arrays Almuatasim Alomari and Ashok Batra Department of Physics, Chemistry and Mathematics (Materials Science Group) College of Engineering, Technology, and Physical Sciences Alabama A&M University Normal, Alabama 35762 USA Received: 8 August 2015 /Accepted: 10 September 2015 /Published: 30 September 2015 Abstract: The electrical output parameters and mode shapes of multiple piezoelectric unimorph cantilever beams (UCB’s) with same thickness and different length reported and examined. Connecting arrays of 5 commercial unimorph beams made of polyvinylidene difluoride (PVDF) in series showed widening in the bandwidth and increasing in the power magnitude of energy harvester comparing to single unimorph beam. The output power was increased from 2 µW to 5 µW and the bandwidth was widened from (47, 55) Hz to (22, 88) Hz. Finite element analysis (FEA) was used to investigate about the first fifth mode shapes of the suggested system using COMSOL multi-physics, with a good agreement between model and experiment. Copyright © 2015 IFSA Publishing, S. L. Keywords: Piezoelectric, Unimorph, PVDF, Mode shapes, Modelling, FEA, COMSOL. 1. Introduction The power that is produced by scavenging vibration energy can be useful in many applications such as, wireless sensor networks, microelectromechanical systems (MEMS), and biomedical sensors [1-5]. Piezopolymer materials (such as PVDF) are getting interest during excellent piezoelectric properties and arbitrary configurations such as, curved surfaces, and be embedded in a MEMS. Increasing the Power and the bandwidth of piezoelectric harvesters have been the topic of most publications in recent years. In order to enhance the maximum output power and operational bandwidth of piezoelectric energy harvesters, researchers have developed a variety of techniques based on, varying shape of structure beam using an L-shaped flexible structure [6-8], adding an additional impendence between the piezoelectric harvester and load resistance [9-12], using dual-mass systems [13], changing the cross-section of a dynamic http://www.sensorsportal.com/HTML/DIGEST/P_2721.htm magnifier [14] using an energy harvester with a dynamic magnifier [15-17], and using energy harvesting cantilever arrays [18-19]. Generally, a single unimorph piezoelectric cantilever is attached to a base excitation is characterized as a single degree of freedom (SDOF) piezoelectric energy harvester system (PEHS). Researchers reported [20-21] that the SDOF is valid only for harvesting energy in a region close to its resonance frequency. Different from single UCB, multiple cantilevers or cantilever arrays integrated in on energy harvesting device as seen in this work can easily achieved continuous wide bandwidth if the geometric parameters of the harvester are appropriately selected [22]. In a continuous effort of increasing a power and bandwidth of a piezoelectric harvester arrays a 3 piezoelectric bimorph cantilevers with the same dimensions but with different tip masses have been developed and studied [18]. The results showed the possibility of the system to work in different frequency ranges, and to widen the overall 37 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 37-43 equivalent bandwidth of the converter array. A multiple piezoelectric bimorphs (PB) with different thickness of piezoelectric layers have been reported and tested experimentally and numerically. They found that the bandwidth of their PB array configuration could be tailored by choosing an appropriate connection pattern (mixed series and parallel connections) [19]. Different energy harvesters with cantilever array were also implemented compatibly with current standard MEMS fabrication techniques [23-26]. A unimorph cantilever beam (UCB) configuration in simple form includes one piezoelectric layer and one shim layer which is commonly showing a single resonance frequency at each natural mode shape. A multifrequency energy harvester devices are easily achieved by integrating multiple cantilevers or cantilever arrays in one cantilever beam [27-28]. We demonstrate in this paper the electrical output parameters and mode shapes of a 5 commercial piezoelectric unimorph cantilever arrays with same thickness and different length integrated in aluminum beam. Fig. 1. Experimental setup used for the frequency response measurements of a unimorph cantilever arrays (photos by A. Alomari, 2015): (1) Shaker with an accelerometer (Bruel&Kjaer 4810) and the cantilever arrays; (2) Laser vibrometer (Microtrack II) MTI; (3) Fixed gain amplifier (Bruel&Kjaer 2718); (4) Control function generator (GF8046 ELENCO); (5) Digital multimeter (Keithley 2110); (6) Variable resistances box; (7) Picoscope; (8) Data acquisition system. 2. Experimental Results and Analysis In this section, we present experimental results of our design of energy harvester. A parametric study was undertaken using the experimental setup shown in Fig. 1. Energy harvesting measurements were carried out initially by attaching a commercial polyvinylidene fluoride (PVDF) unimorph cantilever beam at the front of Teflon base, the configuration shown in Fig. 2(a). The second configuration involved attaching multiple polyvinylidene fluoride (PVDF) unimorph cantilever beams or arrays at the front of aluminum cantilever beam, as shown in Fig. 2(b). Both devices were then connected to a shaker system. The dimensions, electrical, mechanical properties of cantilever beams are shown in Table 1. (a) (b) Fig. 2. Close views of the proposed system tested under base excitation (a) single UCB (b) multiple UCB’s (photos by A. Alomari, 2015). Table 1. Properties of cantilever beams investigated. Type of beam Length (mm) Width (mm) Thickness (mm) Young's modulus (GPa) Density (kg/m3) Aluminum beam Lm=100 wm=24 hm=0.5 Em=69 ρm=2700 Dielectric constant - ε 33T =12 - Piezo strain coefficient (10-12 C/N) Capacitance (nF) - d31=23×10-12 Cp=2.7 - Fig. 3 shows the frequency response functions (FRF) of output voltage and average output power for single UCB at various load resistance. It can be shown from Fig. 3 that there is only one peak which represents the resonance frequency of UCB. The 38 Piezoelectric unimorph cantilever beam PVDF layer Polyester layer Lp=41, 39, 37, 35, 33 Ls=41, 39, 37, 35, 33 wp=16 ws=16 hp=0.25 hs=0.25 Ep=4 Es=8.3 ρp=1780 ρs=1820 energy harvesting bandwidth at resonance frequency of the harvesting beam is between (47, 55) Hz. Fig. 4 shows the frequency response functions (FRF) of output voltage and average output power for multiple of UCB’s at various load resistance. It can be Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 37-43 1.6 R=1 kΩ R=10 kΩ R=100 kΩ R=1 MΩ 1.4 Output Voltage (V) 1.2 increasing RL and the resonance frequency of a Piezoelectric energy harvester in UCB depends on the external load resistance RL. Fig. 6 shows the output voltage and average power for multiple UCB’s at various range of load resistances, RL, from 100 Ω to 10 MΩ obtained for excitation at the first fifth mode shapes. 2.5 R=1 kΩ R=10 kΩ R=100 kΩ R=1 MΩ 2.0 Output Voltage (V) shown from Fig. 4 that there are five peaks which represent the resonance frequency of UCB’s. The energy harvesting bandwidth at resonance frequency of the harvesting beams in this case was between (22, 88) Hz. The maximum power for mode 1 excitation is around 0.29 µW at 22 Hz. For mode 2 excitation, the maximum power output is around 0.63 µW at 30 Hz. The maximum power output for mode 3 excitation is 1.54 µW at 45 Hz. For mode 4 excitation, the maximum power output is around 4.7 µW at 65 Hz. The maximum power output for mode 5 excitation is 5.22 µW at 88 Hz. The resonance frequencies, output voltage, and average power of the first fifth modes at optimum resistance for single UCB and multiple UCB’s from the graphs in Fig. 3 and Fig. 4 are summarized in Table 2 at the end of this article. 1.5 1.0 0.5 0.0 1.0 20 0.8 60 80 100 Frequency (Hz) 0.6 (a) 0.4 6 0.2 R=1 kΩ R=10 kΩ R=100 kΩ R=1 MΩ 5 20 40 60 80 100 Frequency (Hz) (a) 2.5 R=1 kΩ R=10 kΩ R=100 kΩ R=1 MΩ 2.0 Output Power (μW) 0.0 Average Power (μW) 40 4 3 2 1 0 1.5 20 40 60 80 100 Frequency (Hz) 1.0 (b) 0.5 Fig. 4. Experimental data of FRF of multiple UCB’s at various load resistance of output (a) voltage (b) power. 0.0 20 40 60 80 100 Frequency (Hz) (b) Fig. 3. Experimental data of FRF of single UCB at various load resistance of output (a) voltage (b) power. Fig. 5 shows the output voltage and average power for single UCB at various range of load resistances, RL, from 100 Ω to 10 MΩ. It can be seen from Fig. 5 (a) that the output voltage increases with Table 2. Resonance frequency, voltage and power output parameters of multiple UCB’s. Resonance Output Output Cantilever Mode shape frequency voltage power beam type (Hz) (V) (μW) Single st 1 Mode 50 1.41 1.97 UCB st 1 Mode 22 0.17 0.29 2nd Mode 30 0.25 0.63 Multiple 3rd Mode 45 0.39 1.54 UCB’s 4th Mode 65 0.69 4.69 5th Mode 88 0.72 5.22 39 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 37-43 frequency of each vibration mode in Fig. 8 is summarized in Table 3. 1st mode at 22 Hz 2nd mode at 30 Hz 3rd mode at 45 Hz 4th mode at 64 Hz 5th mode at 88 Hz 2.0 8 Output Voltage (V) Output Voltage (V) 2.5 1.5 1.0 0.5 6 4 2 0.0 0 2 4 6 8 10 12 Load Resistance (MΩ) 0 0 2 (a) 4 6 10 Load Resistance (MΩ) 2.5 (a) 2.0 1st mode at 22 Hz nd 2 mode at 30 Hz rd 3 mode at 45 Hz 4th mode at 64 Hz th 5 mode at 88 Hz 8 1.5 Output Power (μW) Average Power (μW) 8 1.0 0.5 6 4 2 0.0 0 2 4 6 8 10 12 0 Load Resistance (MΩ) (b) Fig. 5. Experimental data of (a) output voltage and (b) output power versus load resistance of single UCB at resonance frequency. 0 2 4 6 8 10 Load Resistance (MΩ) (b) Fig. 6. Experimental data of (a) output voltage and (b) output power versus load resistance of multiple UCB’s at first fifth of mode frequency shapes. 3. Modelling Results A 3 dimensional UCB’s with aluminum beam are used for the simulation in COMSOL. The model is designed in COMSOL as shown below in Fig. 7(a). The model consists of 5 UCB’s with different lengths attached at the front of aluminum beam. The lengths of shim layer and piezomaterial are made equal. Using solid mechanics module, one end of the model is fixed and the other end is made to move freely. Meshing a geometry is done using size parameters for free tetrahedral, with a fine mesh near the clamped end as shown in Fig. 7(b). The complete mesh consists of 22954 domain elements, 13676 boundary elements, and 1333 edge elements for a total number of degrees of freedom of 139605. The Eigen-frequency analysis is done with different modes of a suggested model beams as shown in Fig. 8 below. The resonance 40 (a) (b) Fig. 7. (a) Designed model, and (b) Meshing proposed system in COMSOL. Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 37-43 a) b) c) d) e) Fig. 8. Modeling the resonance frequency using COMSOL for multiple UCB’s (a) First mode at 22.98 Hz, (b) Second mode at 30.91 Hz, (c) Third mode at 48.70 Hz, (d) Fourth mode at 72.1 Hz, (e) Fifth mode at 91.7 Hz. 41 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 37-43 Table 3. Experimental, and COMSOL resonance frequency results of multiple UCB’s. Resonance frequency (Hz) Cantilever Mode shape Error beam type EXP. COMSOL (%) 22 22.98 4.45 1st Mode 2nd Mode 30 30.91 3.03 Multiple 3rd Mode 45 48.70 8.22 UCB’s 4th Mode 65 72.10 10.92 5th Mode 88 91.70 4.20 4. Conclusions This paper has tested the effect of attaching a multiple unimorph cantilever beams with same thickness and different length of shim and piezoelectric layers at the front of aluminum beam on the output electrical parameters and bandwidth of both an experimental and modelling level. The experimental results show an increasing in maximum output power and bandwidth of multiple piezoelectric unimorph cantilever beams (UCB’s comparing to single UCB. Connecting arrays of 5 commercial unimorph beams made of polyvinylidene difluoride (PVDF) in series showed increasing in output power from 2 µW to 5 µW and widening in the bandwidth from (47, 55) Hz to (22, 88) Hz. The first fifth mode shapes of five piezoelectric unimorphs cantilever arrays integrated in one aluminum beam are investigated theoretically using COMSOL multiphysics and experimentally, with a good agreement between model and experiment. Acknowledgements The authors gratefully acknowledge support for this work through the National Science Foundation grant #EPSCoR R-II-3 (EPS-1158862). Authors thank Dr. Chance M. Glenn, Dean, College of Engineering, Technology and Physical Sciences and Dr. M. D. Aggarwal, Chairman, Department of Physics, Chemistry and Physics for their keen interest in this work. References [1]. G. K. Ottman, H. F. Hofmann, A. C. Bhatt, G. A. Lesieutre, Adaptive piezoelectric energy harvesting circuit for wireless remote power supply, IEEE Transactions on Power Electronics, Vol. 17, Issue 5, 2002, pp. 669-676. [2]. S. Roundy, P. K. Wright, J. Rabaey, A study of low level vibrations as a power source for wireless sensor nodes, Computer Communications, Vol. 26, Issue 11, 2003, pp. 1131-1144. [3]. G. Park, T. Rosing, M. D. Todd, C. R. Farrar, W. Hodgkiss, Energy harvesting for structural health monitoring sensor networks, Journal of Infrastructure Systems, Vol. 14, Issue 1, 2008, pp. 64-79. 42 [4]. K. A. Cook-Chennault, N. Thambi, A. M. Sastry, Powering MEMS portable devices — a review of nonregenerative and regenerative power supply systems with special emphasis on piezoelectric energy harvesting systems, Smart Materials and Structures, Vol. 17, Issue 4, 2008, pp. 1-33. [5]. Z. L. Wang, Nanogenerators for self-powering nanosystems and piezotronics for smart MEMS/NEMS, in Proceedings of the 24th IEEE International Conference on Micro Electro Mechanical Systems (MEMS’11), Cancum, Mexico, January 2011, pp. 115-120. [6]. Y. Ting, G. Hariyanto, B. K. Hou, S. Ricky, S. Amelia, C.-K. Wang, Investigation of Energy Harvest and Storage by Using Curve-shape Piezoelectric Unimorph, in Proceedings of the IEEE International Symposium on Industrial Electronics (ISlE’09), Seoul, Korea, 5-8 July 2009, pp. 2047-2052. [7]. J. W. Xu, W. W. Shao, F. R. Kong, Z. H. Feng, Rightangle piezoelectric cantilever with improved energy harvesting efficiency, Applied Physics Letter, Vol. 96, Issue 15, 2010, pp. 15-29. [8]. A. Erturk, J. M. Renno, D. J. Inman, Modeling of Piezoelectric Energy Harvesting from an L-Shaped Beam-Mass Structure with an Application to UAVs, Journal of Intelligent Material System Structure, Vol. 20, Issue 5, 2009, pp. 529-544. [9]. N. A. Kong, D. S. Ha, A. Etrurk, D. J. Inman, Resistive impedance matching circuit for piezoelectric energy harvesting, Journal of Intelligent Material System Structure, Vol. 21, Issue 13, 2010, pp. 12931302. [10]. J. R. Liang, W. H. Liao, Impedance matching for improving piezoelectric energy harvesting systems, in Proceedings of the SPIE International Conference on Active and Passive Smart Structures Integrated Systems, San Diego, CA, USA, Vol. 7643, March 2010, pp. 1-12. [11]. W. J. Wu, Y. Y. Chen, B. S. Lee, J. J. He, Y. T. Pen, Tunable resonant frequency power harvesting devices, in Proceedings of the SPIE International Conference on Smart Structures Material Damping and Isolation, San Diego, CA, USA, Vol. 6169, 27 February-1 March 2006, pp. 61690A-1–61690A-8. [12]. A. Badel, D. Guyomar, E. Lefeuvre, C. Richard, Piezoelectric energy harvesting using a synchronized switch technique, Journal of Intelligent Material System Structure, Vol. 17, Issue 8-9, 2006, pp. 831-839. [13]. X. Tang, L. Zuo, Enhanced vibration energy harvesting using dual-mass systems, Journal of Sound and Vibration, Vol. 330, Issue 21, 2011, pp. 5199-5209. [14]. V. K. Sharma, K. Srikanth, A. K. Viswanath, Influence of cross-sectional area of a dynamic magnifier for vibration energy harvesting, in Proceedings of the 4th IRF International Conference, Hyderabad, India, Vol. 36, No. 3, 6 April 2014, pp. 197-205. [15]. W. Zhou, G. R. Penamalli, L. Zuo, An efficient vibration energy harvester with a multi-mode dynamic magnifier, Smart Materials and Structures, Vol. 21, Issue 1, 2012, pp. 015014. [16]. A. Aladwani, M. Arafa, O. Aldraihem, A. Baz, Cantilevered piezoelectric energy harvester with a dynamic magnifier, Journal of Vibration and Acoustics, Vol. 134, Issue 3, 2012, pp. 031004-1– 031004-10. Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 37-43 [17]. O. Aldraihem, A. Baz, Energy harvester with dynamic magnifier, Journal of Intelligent Material System Structure, Vol. 22, Issue 6, 2011, pp. 521-530. [18]. M. Ferrari, V. Ferrari, M. Guizzetti, D. Marioli, A. Taroni, Piezoelectric multifrequency energy converter for power harvesting in autonomous microsystems, Sensors and Actuators A, Vol. 142, Issue 1, 2008, pp. 329-335. [19]. H. Xue, Y. Hu, Q.-M. Wang, Broadband Piezoelectric Energy Harvesting Devices Using Multiple Bimorphs with Different Operating Frequencies, IEEE Transactions on Ultrasonic, Ferroelectrics, and Frequency Control, Vol. 55, Issue 9, 2008, pp. 2104-2108. [20]. A. Erturk, D. J. Inman, On mechanical modeling of cantilevered piezoelectric vibration energy harvesters, Journal of Intelligent Material Systems and Structures, Vol. 19, No. 11, 2008, pp. 1311-1325. [21]. A. Alomari, A. Batra, M. Agarwal, C. R. Bowen, Parametric study and effect of beam material on a unimorph piezoelectric energy harvester with tip mass, Sensors & Transducers, Vol. 190, Issue 7, July 2015, pp. 1-9. [22]. N. Elvien, A. Erturk, Advances in energy harvesting method, Springer-Verlag, New York, 2013, pp. 3-14. [23]. J. Liu, H. Fang, Z. Xu, X. Mao, X. Shen, D. Chen, H. Liao, B. Cai, A MEMS-based piezoelectric power [24]. [25]. [26]. [27]. [28]. generator array for vibration energy harvesting, Microelectronics Journal, Vol. 39, Issue 5, 2008, pp. 802-806. I. Sari, T. Balkan, H. Kulah, An electromagnetic micro power generator for wideband environmental vibrations, Sensors and Actuators A, Vol. 145-146, Special Issue, 2008, pp. 405-413. S. Cheng, Y. Jin, Y. Rao, D. P. Arnold, A bridge voltage doubler AC/DC converter for low-voltage energy harvesting applications, in Proceedings of Power MEMS, Washington DC, USA, 1-4 December 2009, pp. 25-28. H. Yu, J. Zhou, L. Deng, Z. Wen, A Vibration-Based MEMS Piezoelectric Energy Harvester and Power Conditioning Circuit, Sensors, Vol. 14, Issue 2, 2014, pp. 3323-3341. W. Al-Ashtari, M. Hunstig, T. Hemsel, W. Sextro, Enhanced energy harvesting using multiple piezoelectric elements: Theory and experiments, Sensors and Actuators A, Vol. 200, 2013, pp. 138-146 (Selected Papers from the 9th International Workshop on Piezoelectric Materials and Applications in Actuators). L. Tang, Y. Yang, C. K. Soh, Advances in Energy Harvesting Methods, (Ed. N. Elvin, A. Erturk), Springer, New York, 2013. ___________________ 2015 Copyright ©, International Frequency Sensor Association (IFSA) Publishing, S. L. All rights reserved. (http://www.sensorsportal.com) 43 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 44-52 Sensors & Transducers © 2015 by IFSA Publishing, S. L. http://www.sensorsportal.com Determination of Multiple Spring Constants, Gaps and Pull Down Voltages in MEMS CRAB Type Microaccelerometer Using Near Pull Down Capacitance Voltage Measurements R. K. Bhan, Shaveta, Abha Panchal, Yashoda Parmar, Chandan Sharma, Ramjai Pal, Shankar Dutta Solid State Physics Laboratory, Lucknow Road, Timarpur, Delhi–110054, India Tel.: 91-011-23903403 E-mail: [email protected] Received: 11 August 2015 /Accepted: 14 September 2015 /Published: 30 September 2015 Abstract: A simple experimental method based on capacitance voltage (CV) measurements is presented to extract the spring constants (k) different actuation voltages and gaps, in crab type capacitive MEMS accelerometer sensors. It is shown that in addition to main spring action provided by the legs of the structure, the additional spring constants related to the interaction of main spring-proof mass joint and corner region of the proof mass also contribute to change in capacitance. The proposed approach is used in resolving and measuring these model parameters simultaneously because all of them can be extracted from the just one CV measurement. It is found that this additional k varies by more than a factor of 10 across the 6-inch wafer. Furthermore, zero bias capacitance C0, zero bias gap g0, main spring constant k and initial pull down voltage Vpd1 vary by factors of 2.4, 2.38, 30 and 3 respectively. The method also allows us to extract different values of pull down and spring constants associated with different regions of crab structure. The experimental results agree well with the theoretical predictions and reported trends in literatures. The method is routinely applied while fabricating the actual prototype sensors fabricated in our laboratory. Copyright © 2015 IFSA Publishing, S. L. Keywords: Accelerometer, MEMS sensors, Capacitance Voltage, Spring constants, Pull down voltage. 1. Introduction Micro-Electro Mechanical Systems (MEMS) based capacitive microaccelerometers (MA) are matured and used in many systems due to their high sensitivity, low noise, low temperature sensitivity and low power dissipation characteristics [1-3]. Capacitive accelerometers performance is generally superior in low frequency range and they can be operated to achieve high stability and linearity. These accelerometers can be designed based on either change 44 in gap or change in area approaches having their own advantages and disadvantages. The change in gap type accelerometers can be fabricated using variety of process recipes [4]. One of the structures based on change in gap type is called Crab type accelerometer (Fig. 1) in which the area of the proof mass that is varied to achieve the targeted capacitance and gap varies as a result of change in acceleration. Further, four legs of the structure with appropriate spring constants are designed to hold the structure at four corners. http://www.sensorsportal.com/HTML/DIGEST/P_2722.htm Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 44-52 Fig. 1. Top and side view of crab type MEMS micro accelerometer structure showing main proof mass and four legs. As described above, the functioning of these MEMS accelerometers is based on the capacitance change due to change in gap as a result of acceleration and hence design involves design of MEMS capacitor that is sensitive to various noise effects present for low spring constant desired for 3–5 V applications. There is lot of literature existing about design and fabrication such capacitors mainly in the context of RF MEMS switches, varactors and accelerometers [1-7]. Recently we reported sensitive axis-misalignment measurements in crab type micro accelerometers being developed in our laboratory [8]. Young and Boser [9] were the first to develop a parallel-plate MEMS capacitor (varactor) that is similar to crab type accelerometer discussed here. Four such capacitors were connected in parallel to achieve total capacitance of change of 15 %. Claza, et al. [10] reported capacitance voltage (CV) measurements of RF MEMS switches in both the on and off states. The spring constant of the suspensions was used to control the actuation voltage of the different designs. Dec and Suyama [11-12] developed two - and three-plate capacitor designs using the standard polysilicon micromachining process. The spring constant of 39 N/m was achieved by using four curved beams and the measured capacitance was tunable from 1.4 to 1.9 pF. Barker, Muldavin, and Rebeiz [13] developed a parallel-plate varactor having a capacitance ratio of 1.35 and suitable for 20 to 100 GHz applications using a low spring constant support. Zou, et al. [14] developed a widetuning-range parallel plate varactor using a novel electrode design in which top capacitor and actuation electrodes were separated by two gaps. Their measured capacitance had a tuning ratio of 1.55–1.65. However, none of these authors reported analysis of the spring constants in the range when applied actuation voltages are of the order of pull down voltages or even larger. The knowledge of parameters in the said range of operation is important for design and parameter extraction. In all the above cases, it was assumed that the capacitance change is controlled by one spring constant of the structure which is true to a first approximation. However, it is argued here that when the deflection of the top capacitor plate is large i.e. on application of actuation voltage greater than pull down voltage Vpd, the capacitance increases further, with sharp transitions near pull down voltages that are manifested by shoulders in CV curves. Such shoulders have been observed by many workers in the field [5, 14-15] but these will be governed by additional or other spring constants of the structure that come into play. Based on our experimental observations, it is proposed here that in Crab type accelerometer, spring constant of structure will be actually staircase type having three values corresponding to anchored leg, corner joining region of the main proof mass and the leg, proof mass near corner region dominated by proof mass. This will be discussed further in detail. Section 2 discusses the simple theoretical design of microaccelerometer (MA) and importance of spring constant k. Section 3 describes the fabrication process. In Section 4, we discuss the CV measurements. Results and discussions are discussed in Section 5 and finally in section 6 we discuss the brief conclusions. 2. Theoretical The principle of working of an accelerometer can be explained by a simple mass (m) attached to a spring of stiffness (k) that in turn is attached to a casing, as illustrated in Fig. 2. The mass used in MEMS accelerometers is often called proof-mass. The system also includes a method or device for damping the shock or vibration to provide a desirable damping effect. This is called a dashpot having a particular damping coefficient c (Fig. 2) that is normally attached to the mass in parallel with the spring. When the spring mass system is subjected to linear acceleration, a force equal to mass times acceleration acts on the proofmass, causing it to deflect. This deflection is sensed by a suitable means and converted into an equivalent electrical signal. Some form of damping is required, otherwise the system would not stabilize quickly under applied acceleration. Fig. 2. The ideal model of MEMS capacitive microaccelrometer. The design involves the solution of the motion equation where all real forces (the sum of all forces in 45 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 44-52 the z direction) acting on the proof-mass are equal to the inertia force on the proof-mass. Accordingly, a dynamic problem can be treated as a problem of static equilibrium and the equation of motion can be obtained by direct formulation of the equations of equilibrium as follows: + + = (1) ext The static deformation is given by (2) For a case of damped mass spring system, the dynamic behavior of motion as given by eqn. (1) is a second order linear differential equation with constant coefficients given as follows: m , (3) where m = mass of the proof-mass; a = acceleration; x = relative movement of the proof-mass with respect to frame; D = damping coefficient; k = spring constant; Fext = external force applied. It may be seen from the above equations that we need to know the value of spring constant k for the design of micro accelerometer. The proof-mass size and spring constants are selected in such a way that there shall be a capacitance change of measurable range say around1 fF or more for the smallest resolution of acceleration to be detected say 1 mg as a minimum resolution. This limitation comes from the capacitance signal that can be handled comfortably by the signal processing electronics. Generally workers in the field have assumed one value of spring constant for a given material and geometry in the design of an capacitive accelerometer [7]. This approach may be correct as a first order design particularly for a crab type accelerometer. However, it is argued here that same spring constant cannot be used for all the values of deflection in crab type MA. As the applied force or acceleration is increased in z direction (say –z, i.e. downwards), the proof mass gets pulled down further and further and spring action of other regions associated with the main spring start contributing. However near the corners structure cannot deflect or pull down fully because of four springs attached to four anchors upwards. For accurate modeling and particularly for shock test modeling, understanding of this effect is important. This effect can be witnessed by simple CV measurements. CV measurements are routinely used in the laboratory to see whether the fabricated MEMS membrane is indeed movable or not. Experimental evidence of this can be seen in the CV measurements 46 of MEMS varactors as reported by many workers [5, 8-14]. It is shown here that spring constant in general increases with decreasing gap. The crab type MA is modeled as a parallel-plate capacitor and neglecting the incremental increase in capacitance due to fringing fields, the model is a good starting point for understanding the effect of electrostatic actuation on capacitance. Following the assumptions as mentioned in Ref. [7], and equating the applied electrostatic force with the mechanical restoring force due to the stiffness of the beam (F=kx), we find that 1 2 , (4) where W and w are the dimensions in x and y directions, V is applied DC voltage, go is the initial gap or zero bias gap and g is the reduced gap as a function of applied voltage. This equation predicts the variation of gap (or capacitance) of proof mass plate as a function of applied voltage. The upper limit of V in this equation is pull down voltage of the membrane. This can be utilized to calculate the change of capacitance as function of V. This equation will be utilized to simulate the expected trends in CV as a function of g and k. Fig. 3 shows the effect of changing gap g on the capacitance for a fixed spring constant k of 10 N/m. It may be seen from this figure that C increases with increase in V till we reach near pull down voltage limit where C increases drastically because gap g reduces to infinitesimally small value e.g. for g=3 µm, C rises sharply because of approaching pull down voltage Vpd of 3 V. Fig. 3. Effect of varying gap g on the capacitance voltage characteristics for a given spring constant k. This figure shows the effect of V on C for a fixed k. Also, it can be seen from this figure as expected the base value of zero bias capacitance increases with decreasing gap. For the case under consideration this capacitance increases from 1.48 pF to 2.94 pF when the gap is decreased from 6 µm to 3 µm, however the curvature of the CV curve remains same. This Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 44-52 curvature is dependent on the spring constant of the structure. Fig. 4 shows the effect of changing spring constant on the CV for a fixed gap of 4 µm. As expected, it may be seen that as expected all the curves merge at the base value of the capacitance indicating that the gap is fixed thus yielding fixed capacitance at about 2.2 pF. However, as the spring constant increases, the pull down voltage indicated by the sharp increase in the C at the last value of V e.g. 1.5 V for k=1 N/m. This voltage changes from 1.48 V to 4.46 V as the k goes from 1 to 10 N/m. probed for CV measurements. The actual structures of Crab type microaccelerometers were successfully realized using above process. Fig. 5 shows the SEM picture of crab type structure. The pictures clearly showed the released and hanging structures as expected. Fig. 5. SEM pictures of crab type microaccelerometer structure. Fig. 4. Effect of varying spring constant k on the capacitance voltage characteristics for a given or fixed gap of the structure. These figures show the basic behavior of C as function of g or k. It is proposed and shown in this paper that using simple CV measurements, wherein voltage is varied up to pull down value and further beyond, one can determine that how many springs are at play during the actuation of the structure by using Equation (4). We have used this methodology, for extraction of different spring constants in our structure as will be shown in the next section. 4. Capacitance Voltage (CV) Measurements The capacitance voltage measurement was done at wafer level using Kithely Semiconductor Parametric Analyzer 4200 model. Fig. 6 shows the typical photograph of the fabricated wafer diced into two parts that was later diced into two parts (subsequently individual chips) after the CV mapping was completed. 3. Fabrication Process The microaccelerometer structure was realized using a three mask dissolved wafer process (DWP) as discussed elsewhere in our earlier paper Ref. [8]. Briefly the structure consists of 12 μm thick boron doped silicon inertial proof–mass suspended over a glass pit that was created in 7740 Pyrex glass substrate using wet etching. A bottom plate capacitor contact was made of Ti plus Gold that was deposited and patterned. The depth of cavity is about 4-5 microns. Further, in the Silicon wafer deep boron diffusion was carried out at 1175 oC to achieve boron doping ~ 1×1020 atoms/cc over a depth of 12 μm. This heavily boron doping acts as etch stop layer during the final stages of DWP process in chemical etchant. The devices were fabricated using 6-inch wafer process. Before dicing of the individual dies, the devices were Fig. 6. Photograph of a typical wafer showing multiple crab type micro accelerometer structures on the semi diced wafer that are probed for CV measurement. This measurement was further used to verify the deflection or immovability of structures. It can also be used to check whether the structure is properly released or not. If the structure is released properly, then it will move down with increasing actuation voltage which will result in continuously increasing 47 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 44-52 capacitance because of reducing gap till actuation voltage. This is particularly easy to verify in Crab type of structure as will be shown in next sections. 5. Results and Discussions Most of the measured devices showed shallow U-type CV characteristics indicating that the structures are properly released and the devices are responding to change in actuation voltage as per theoretical predictions of Fig. 3 and Fig. 4. However, there is evidence of change in gap, spring constant and stress in the proof mass from device to device across the wafer. The change in gap results in vertical shift of zero bias capacitance, change in spring constants results in change of curvature in CV curve i.e. small k leads to low pull down voltages and sharp change in capacitances and vice versa. Furthermore, there are slight differences in actuation forces in +ve and –ve cycles of the voltages resulting in non-symmetrical CV. This type of behavior is evident in device nos. 23, 8 and 15 because the non-symmetry in CV curves in maximum in these devices. In rest of the devices, this non-symmetry was negligible. This is attributed to association or joining together of proof mass with air (conjunct air) and some form of dielectric charging non-uniformities effect seen in many MEMS capacitors that lead to non-symmetry in CV curves as discussed in Ref. [16]. However, how exactly it is affecting in our case is not understood clearly at present. Next we found that there is a variation among the fabricated devices across the wafer. There were about 13 types of different varying CV curves that we found in our devices that are shown in Fig. 7. 1.89 pF to 3.17 pF indicating that our process needs to control the variation of released gap in these devices. Secondly there is strong evidence of variation in curvature and number of shoulder in these CV curves from device to device. This indicates that we have a variation in spring constants from device to device as well. The most important finding is that there can be more than one shoulder in CV curves near the pull down voltage in both +ve and –ve directions of voltage as can be seen from the sample 8 in Fig. 7. For the case of +ve voltage cycle, we can see from this sample that these shoulders in CV are evident and the onset voltages are 3.4 and 3.45 V respectively. This indicates that another spring constant comes in play beyond this voltage which controls the further reduction of the gap and hence the increase in capacitance. We have utilized these shoulder positions for estimating the multiple spring constants that come into play while the structure is pulled down by actuation voltage. This behavior will be analyzed further for extraction of spring constants in detail by fitting the experimental data to theory. For example, we can see from this figure that within a CV measurement range of 0-5 V we have only one shoulder with onset at 1.46 V for sample 21 and at 4.20 V for sample 23. Next, we try to fit all different variants of CV curves covering minimum to maximum range of capacitance and shoulders in our case. The samples covering the said range are 8, 11, 12 and 23 and will be analyzed further for detailed for parameter extraction. Rest of the devices in terms of their capacitances and number of spring constants that come into play along with their values lie within the measured range of these devices. Type I devices: Fig. 8(a) and Fig. 8(b), show that for both the device numbers of 1 and 11, there is minimum change in capacitance of 0.02 pF from 0 to 5 V, although their zero bias capacitance C0 as well as Cmax at 5 V is different for both the devices. Fig. 7. Results of CV measurements for 13 microaccelrometer devices having different behavior and depicting the complete range of variations observed in our devices. These different types are further categorized into three major types viz. Type I, Type II and Type III as will be discussed subsequently. It may be seen from this figure that zero bias capacitance varies from 48 Fig. 8 (a). The fitting of experimental data for typical CV curves of Type I devices showing no shoulders or evidence of pull down voltage for V<5 Volts. This is the case of minimum change in zero bias capacitance with low g and high k. Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 44-52 Type II devices: Fig. 9 for device 23 shows the different behavior compared to device 1, 11 or 13 in the sense that it shows the evidence of an additional capacitance in parallel to the original capacitance as evidenced by one shoulder near pull down voltage Vpd1 of 4.21 to 4.40 V. The value of zero bias capacitances for these two cases are 2.55 and 4.54 pF respectively as extracted by fitting of experimental data to theory. Similarly the extracted values for g for these capacitances are 3.14 and 1.95 µm respectively. Fig. 8 (b). The fitting of experimental data for the typical CV curves of Type I devices showing no shoulders or evidence of pull down voltage for V<5 Volts. This is the case of minimum change in zero bias capacitance with moderate g but high k. The capacitance values for 5 V are 2.24 and 1.89 pF respectively. This indicates that gap and spring constants are different for both of these devices. Further, there is no evidence of single or double shoulder within the measurement range of 0 to 5 V. However they both show similar change of 0.02 pF in capacitance values. Actual fitting of experimental data shows a value of k=180 N/m for device 1 and 180 N/m for device 11. This suggests that both these curves will behave differently when subjected to same acceleration although their observed curvature in CV curves is similar. Fig. 8 (c) for the device 13 shows the similar behavior as in Fig. 8 (a) and Fig. 8 (b), however it shows the highest change in capacitance of 0.34 pF from 0 to 5 V out of the lot wherein also there is no evidence of single or double shoulder. The zero bias gap capacitance and spring constant estimated from fitting in this device are 3.8 pF and 22 N/m respectively. Fig. 8 (c). The fitting of experimental data for the typical CV curves of Type I devices showing no shoulders or evidence of pull down voltage for V<5 Volts. This is the case of with high g but low k. Fig. 9. The fitting of experimental data for the typical CV curves of Type II devices showing one shoulders and one pull down Vpd1 ~ 4.21 volts. Further, the extracted values for k for current Type II and later Type III have been normalized for full area of the membrane and are of 14 and 230 N/m respectively. However, the spring constant density Kn per unit area has also been given in the summary Table 1 for all the devices. A possible physical model that explains all the observations in totality will be proposed after discussing the results of another trend observed in devices 8 and 21. Type III devices: Fig. 10 (a) and Fig. 10 (b) showing the results for the devices 8 and 21 exhibit even further different behavior compared to Fig. 9 for sample 23 in the sense that we see two additional capacitances acting in parallel to original capacitances as is evident by two shoulders near pull down voltages of Vpd1 and Vpd2 shown in these figures. These capacitances are having different spring constants as can be clearly seen from the different curvatures in three different regions of CV. The transitions of pull down voltages viz. Vpd1 and Vpd2 are clearly resolved and very sharp. For the case of device 8, the values of gas extracted by parameter fitting of experimental data are 3.85, 3.10 and 2.0 µm respectively. Similarly, the extracted values of k are 6.5, 21 and 68 N/m respectively. Incidentally, this device shows the highest change in capacitance i.e. C of 1.08 pF for a V of 3.4 Volts of the lot. This corroborated by its corresponding lowest value of k=6.5 N/m out of the lot. The values of onset values pull down voltages viz. Vpd1 and Vpd2 are 3.40 and 49 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 44-52 4.10 respectively. The results of device 21 Fig. 10(b) shows the similar trends as that of device 8 except the fact that shoulders in CV curve are merged unlike in device 8 and C in this case ranges from 0.02 to 0.03 pF. Further, the values Vpd1 and Vpd2 in this case are 1.40 and 2.0 V respectively and are also lower compared to values of device 8. Table 1. The summary of the measured and extracted parameters for all the different devices. Sample No. 1 11 13 23 8 21 Type Type I Single C, No shoulder (Vpd) in CV Type II Two C’s in parallel, One shoulder (Vpd1) in CV Type III Three C’s in parallel, Two shoulders (Vpd1 & Vpd2) in CV g (µm) 3.95 4.66 k (N/m) 180 100 kn (N/m3) 180e6 100e6 C0 (pF) 2.24 1.89 Cmax (pF) 2.26 1.91 Cmin (pF) 2.24 1.89 C (pF) 0.02 0.02 Vpd1 (V) - Vpd2 (V) - 3.8 22 22e6 2.32 2.66 2.32 0.34 - - 3.40 14 14e6 2.55 3.27 2.55 0.72 1.95 230 230e6 4.54 4.91 4.54 0.37 4.21 to 4.40 - 3.85 3.10 2.50 2.80 2.78 2.76 6.5 21 68 45 150 650 6.5e6 21e6 68e6 45e6 150e6 650e6 2.32 2.85 3.54 3.16 3.18 3.20 3.40 3.47 3.90 3.19 3.20 3.22 2.32 2.85 3.54 3.16 3.18 3.20 1.08 0.62 0.36 0.03 0.02 0.02 3.4 to 3.5 4.10 to 4.22 1.40 2.0 Fig. 10 (a). The fitting of experimental data for the typical CV curves of Type III devices showing two shoulders and two pull down voltages Vpd1 and Vpd2 of ~ 3.40 and 4.10 volts and having well defined shoulders. Fig. 10 (b). The fitting of experimental data for the typical CV curves of Type III devices showing two shoulders and two pull down volts Vpd1 and Vpd2 ~ 1.4 and 2.0 volts and having merged shoulders. 50 It may be recalled here that Vpd values are dependent on both g and k and hence difficult to isolate the reason for their variation. The summary of the measured and extracted parameters for all the different devices are given in Table 1. About 30 functional devices were measured from the same batch (under consideration here) and majority of them showed similar to Type I behavior as mentioned in Table 1 i.e. single type C observed within V<5 Volts, No shoulder or pull down (Vpd) was observed in their CV curves for V<= 5 volts. It may be stressed and mentioned here that Type I devices will show Type II or even Type III behavior if voltage measurement range is extended much beyond to 5 V. Since, our interest and targeted values of Vpd was ~5 V, no detailed measurements and analysis was carried out for V> 5 Volts. Table 1 also shows that across the whole wafer, C0 varies from 1.89 to 4.54 pF i.e. by a factor of 2.4. Initial gap g varies from 1.95 to 4.66 µm i.e. a factor of 2.38. Similarly, spring constant k corresponding to legs of main structure varies from 21 to 650 N/m i.e. by a factor of 30.95. This variation is largest out of all the variables. This is expected because this value is highly sensitive to dimensional control in photolithography and etching. This variation in k and g lead to variations in first pull down voltage Vpd from 1.40 to 4.21 V i.e. by a factor of 3. Furthermore, k varies by more than a factor of 10 within Type III devices. Next, we try to propose a simple physical model which explains our experimental observations. Fig. 11 shows the proof mass of crab structure acting as a top plate of parallel capacitor based physical model which explains our experimental observations. Initially for low values of actuation voltages, the whole of top plate of the capacitor, barring four edges which are tied by the legs of the structure, moves down resulting in Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 44-52 reducing gap and hence increasing capacitance. This movement of the structure is controlled by the spring constant k1 and the gap g1 as shown in this figure. Fig. 11. Proposed physical model with three spring constants and gaps for working of Crab type microaccelrometer under application of actuation voltage resulting in observed CV curves. This continues till we reach the near the onset of pull down voltage of the structure as evidenced by the first shoulder of the CV curve (Fig. 9, Fig. 10(a) and Fig. 10(b). Earlier such shoulders in CV curves were observed by Venkatesh, et al. [5] in variable gap varactors. Next, once the voltage is increased further, the edges that were not responding to this voltage so far because of higher gap will start responding and contribute its capacitance that acts in parallel to the main plate. Hence the overall capacitance increases as evidenced by shoulders in CV curves. However, the magnitude of this capacitance is dependent on spring constant k2 in addition to the gap g2 as shown in Fig. 11. Therefore, the capacitance continues to increase again but this time it is controlled by k2 i.e. spring constant of the right angled leg connected to the edge of the plate & g2 i.e. gap closer to the inner region of the capacitor plate edge having gap g1. This is clearly evident by another branch in CV i.e. in Type II devices. It is difficult to pin point the demarcation of regions of the k’s and g’s. In practice, the variation in k and g will be continuously graded. Similarly, for Type III devices, another capacitance adds its contribution in parallel evidenced by (device 12) third branch of capacitance. However, this is controlled by gap g3 i.e. right near the edge and spring constant k3 i.e. of the main plate. As expected from the geometry of the structure we have k3>k2>k1 and g1<g2<g3. These observations are supported by the extracted parameters as shown in Table 1. In short spring constant increases as we approach from outer spring (i.e. legs) to inner spring (i.e. main capacitor plates) and gap increases in opposite direction i.e. from inner capacitor plate to outer capacitor legs/arms. These proposed trends in CV measurements of MEMS capacitors have been reported in literature although in different context. Such shoulders [8, 14-15] have been observed by many workers in the field. Particularly, in Ref. [15] double shoulder was observed in CV measurements. In short, we have shown that one can measure the different spring constants of the Crab type micro accelerometer by using simple CV measurements and extract different actuation voltages, gaps and spring constants from these measurements. Further, these parameters should be used by the designers of accelerometers while modeling dynamic behavior of these structures for different values of g under consideration. It is clear that it is not one single k that comes in to play, rather it is dependent of values of g under consideration. Higher the g, more the structure gets pulled down and possibly the contribution from the edges cannot be neglected. Particularly this effect has more importance MEMS varactors as discussed by Venkatesh, et al. [5]. Here the gap is totally controlled by the actuation voltage. 6. Conclusions In brief, using the CV measurements, an analytical approach based on fitting of different branches of CV curves is utilized for extraction of gaps, spring constants, pull down voltage parameters for MEMS Crab type accelerometers. The measured trends of our results compare well with measurements of other MEMS devices like varactors, RF switches etc. The method is routinely used in our laboratory for characterization of MEMS sensors in our laboratory. Acknowledgements The authors would like to thank Director, Solid State Physics Laboratory for his continuous support and for the permission to publish this work. Help from other colleagues of MEMS division are also acknowledged. References [1]. Girish Krishnan, Chaitanya U. Kshirsagar, G. K. Ananthasuresh, Navakanta Bhat, Micromachined High-Resolution Accelerometers, Journal of the Indian Institute of Science, Vol. 87, No. 3, Jul.–Sep. 2007, pp. 331-361. [2]. Arjun Selvakumar, A High-Sensitivity Z-Axis Capacitive Silicon Microaccelerometer with a Torsional Suspension, Journal of Micromechanical Systems, Vol. 7, No. 2, June 1998, pp. 192-200. [3]. T. Tsuchiya, H. Funabashi, A z-axis Differential Capacitive SOI Accelerometer with Vertical Comb Electrodes, Sensors & Actuator A: Physical, Vol. 116, No. 3, 2004, pp. 378-383. [4]. S. Dutta, R. Pal, P. Kumar, O. P. Hooda, J. Singh, Shaveta, G. Saxena, P. Dutta, R Chatterjee, Fabrication Challenges for Realization of Wet Etching Based Comb Type Capacitive Microaccelerometer Structure, Sensors & Transducers, Vol. 111, Issue 12, December 2009, pp. 18-24. 51 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 44-52 [5]. Chenniappan Venkatesh, Navakanta Bhat, K. J. Vinoy, Satish Grandhi, Microelectromechanical torsional varactors with low parasitic capacitances and high dynamic range, Journal Micro/Nanolithography, MEMS and MOEMS, Vol. 11, No. 1, Jan.–Mar. 2012, pp. 013006-1–013006-8. [6]. Arjun Selvakumar, A High-Sensitivity Z-Axis Torsional Silicon Accelerometer, in Proceedings of the International Electron Device Meeting (IEDM ‘96’), San Francisco, CA, December 8-11, 1996, pp. 765 – 768. [7]. Gabriel M. Rebeiz, RF MEMS: Theory, Design, and Technology, Chapter 2, John Wiley & Sons, New Jersey, 2003, pp. 1-27. [8]. R. K. Bhan, Shaveta, Imran, Ramjai Pal, Shankar Dutta, An improved analytical approach for estimation of misalignment error of sensing axis in MEMS accelerometers using simple tilt measurements, Sensors and Transducers, Vol. 189, Issue 6, June 2015, pp. 128-136. [9]. D. J. Young, B. E. Boser, A micromachined variable capacitor for monolithic low-noise VCOs, in Proceedings of the IEEE International Conference in Solid-State Sensors and Actuators Workshop, Washington DC, Hilton Head, June 1996, pp. 86-89. [10]. C. Calaza, B. Margesin, F. Giacomozzi, K. Rangra, V. Mulloni, Electromechanical characterization of low actuation voltage RF MEMS capacitive switches [11]. [12]. [13]. [14]. [15]. [16]. based on DC CV measurements, Microelectronic Engineering, Vol. 84, Issues 5–8, May–August 2007, pp. 1358–1362. A. Dec, K. Suyama, Micromachined electromechanically tunable capacitors and their applications to RF IC’s, IEEE Transactions Microwave Theory Techniques, Vol. 46, No. 12, December 1998, pp. 2587-2595. A. Dec, K. Suyama, Microwave MEMS-based voltage controlled oscillators, IEEE Transactions Microwave Theory Techniques, Vol. 48, No. 11, November 2000, pp. 1943-1949. N. S. Barker, G. M. Rebeiz, Distributed MEMS truetime delay phase shifters and wideband switches, IEEE Transactions Microwave Theory Techniques, Vol. 46, No. 11, November 1998, pp. 1881–1890. J. Zou, C. Liu, J. Schutt-Aine, Development of a wide tuning-range two-parallel- plate tunable capacitor for integrated wireless communication systems, Int. J. RF Microwave CA, Vol. 11, No. 5, August 2001, pp. 322-329. H. Nieminen, V. Ermolov, K. Nybergh, S. Silanto, T. Ryhänen, Microelectromechanical capacitors for RF applications, Journal of Micromechanics and Microengineering, 12, 2002, pp. 177-186. Advanced RF MEMS, edited by Stepan Lucyszyn, Cambridge University Press, New York, 2010, Chapter 6.2.3. ___________________ 2015 Copyright ©, International Frequency Sensor Association (IFSA) Publishing, S. L. All rights reserved. (http://www.sensorsportal.com) 52 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 53-60 Sensors & Transducers © 2015 by IFSA Publishing, S. L. http://www.sensorsportal.com Mössbauer, VSM and X-ray Diffraction Study of Fe3O4 (NP’s)/PVOH for Biosensors Applications 1 Almuatasim Alomari, 2 Hasan M. El Ghanem, 3 Abdel-Fatah Lehlooh, 4 Isam M. Arafa, 5 Ibrahim Bsoul, 1 Ashok Batra 1 Department of Physics, Chemistry and Mathematics (Materials Science Group) College of Engineering, Technology, and Physical Sciences Alabama A&M University Normal, Alabama 35762 USA 2 Department of Physics, Jordan University of Science & Technology, Irbid, 22110, Jordan 3 Physics Department, Yarmouk University, Irbid 211-63, Jordan 4 Department of Chemistry, Jordan University of Science & Technology, Irbid, 22110, Jordan 5 Physics Department, Al al-Bayt University, Mafraq 130040, Jordan 1 Tel.: (256)372-8109, fax: (256)372-5622 E-mail: [email protected] Received: 8 August 2015 /Accepted: 10 September 2015 /Published: 30 September 2015 Abstract: In this article, structure and magnetic properties of nano magnetic Fe3O4 (magnetite) nanoparticles functionalized polyvinyl alcoholic (PVOH) have been investigated by X-ray diffraction (XRD), Vibrating sample magnetometer (VSM) and Mossbauer Spectroscopy (MS) for use in biosensor applications. XRD showed an average of cluster sizes using Debye–Scherrer formula are between 10-13 nm. The magnetization data at room temperature shows weak hysteresis loops and the isotherms of the magnetization curves indicate that superparamagnetism superimposed on the paramagnetic behavior exists in all coated samples. The paramagnetic contribution in coated samples was found to perfectly fit a Langevin equation, with an average number of magnetic dipole moments around 20 Bohr magnetons. The results of MS showed that all magnetic components corresponding to iron oxide particles in polymer spectrum split into a number of sextet separated by about 10-35 T. The line width, relative intensity and the values of the hyperfine fields and isomer shifts for the magnetic components of the samples are estimated. It was found that only the Fe3O4 sample is suitable for practical medical applications such as, drug delivery systems and to design artificial muscles due to its sufficiently high value of saturation magnetization and attraction to magnet ability. Copyright © 2015 IFSA Publishing, S. L. Keywords: Nano magnetic Fe3O4 nanoparticles, X-ray diffraction, Debye–Scherrer formula, Vibrating sample magnetometer, Mossbauer Spectroscopy, Langevin equation. 1. Introduction Synthesis of superparamagnetic iron oxide nanoparticles (SPION) with polymers has gained http://www.sensorsportal.com/HTML/DIGEST/P_2723.htm increasing interest for emerging applications as tissue repair, drug delivery and in cell separation, cellular imaging in magnetic resonance imaging (MRI), sensors, imaging agents, storage media and catalysis 53 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 53-60 in biotechnology and biomedical application [1-7]. One of the most important features is to prepare coated particles with iron oxide core shell for use in applications that require high magnetization values at room temperature, nontoxic fine particles and have long time stability with size smaller than 100 nm [8]. Many researchers have studied structure and magnetic properties of iron oxide as metal alloy and produced new spinel iron oxide hybrids [9-11], they have also studied it as amorphous with a short-range crystallinity, where amorphous nature of the atomic arrangements has been observed [12]. Uniformly dispersed amorphous nanoparticles of magnetite in a polyvinyl alcohol matrix have been obtained by ultrasound radiation [13]. In other research composite was prepared by mechanical milling of Fe3O4 / SiO2 material constitutes a mixture of ultrafine Fe-rich spinel particles (magnetite/maghemite) [14]. The preparation of magnetite (Fe3O4) has been typically performed by particle precipitation from the hydrolysis and condensation of iron (II)/iron (III) salts in basic media stable aqueous dispersions of magnetic iron oxide colloids were initially generated by ball milling of large particles in the presence of organic stabilizers [15-18]. Solution methods were also developed to prepare aqueous Fe3O4 sols, it was reported that the particle size of Fe3O4 colloids approximate of 10 nm [19-20]. Dextran coated iron oxide nanoparticles were synthesized by addition of FeCl2 and FeCl3 in the presence of ammonium hydroxide (NH4OH) and the polysaccharide surfactant (Mn = 40,000 g/mol), SEM showed the size of Iron oxide nanoparticles is between 10–20 nm [21]. Polymer coated magnetite nanoparticles were synthesized by in situ precipitation in the presence of poly (vinyl alcohol) (PVOH) (Mn = 20,000 g/mol) from an aqueous mixture of ferric and ferrous chloride salts in an alkaline media [22]. It was reported that the prepared samples showed superparamagnetic Fe3O4 colloid behavior with nanoparticles size is in the range of 4–10 nm using XRD, VSM, and TEM. A comparative study of dextran versus the PVOH surfactants in the precipitation of iron oxide colloids was also conducted [23]. A recent report showed the preparation of PVOH coated Fe3O4 colloids using sonochemical methods from iron (II) acetate precursors yielding superparamagnetic hybrid materials [24]. PVOH–magnetite ferrogels prepared using freezing and thawing cycles showed superparamagnetic properties that can be tailored for drug delivery systems and to design artificial muscles [25]. One of the important material which can be immobilized on magnetic nanoparticles in order to use them for biosensing purposes is Streptavidin [26]. Streptavidin is known for its special affinity towards the vitamin biotin and hence it is suitable for detection of diverse biomolecules in immunoassays, e.g. detection of viral nucleic acids in vitro [27]. This paper is aimed at the study of basic magnetic properties of iron oxide Fe3O4 coated with PVOH and non-coated iron oxide Fe3O4 prepared by 54 low-cost conventional sonication method to determined functionality for use in biosensing and biomedical applications. 2. Experimental Section Polyvinyl alcohol (PVOH,72000g/mol) was suspended in 100 mL of 1,2 ethylenedichloride (C2H4Cl2) in a closed container and subjected to sonication for about 1 h at 60-70 oC. To this solution palmatoyl chloride (C15H31COCl, xxxx g/mol) was added with continuous sonication. The reaction mixture proceeded rapidly after addition of triethylamine base (NEt3) with the elimination of triethylammonium chloride salt. The obtained reaction mixture was left overnight in the closed container. This afford 4.15 g of different amounts of palmatoyl chloride is added to afford 4.15 g of the required modified matrix (palmatoyl-PVOH) with different degree of substitution, see Table 1. Table 1. Relative samples contents of PVOH, C15H31COCl (g) and Number of palmatoyl substituted vinylOH units in poly (palm-g- PVOH) polymer backbone. Sample PVOH (g) C15H31COCl (g) Number of palmatoyl substituted vinylOH units on palmatoyl PVOH polymer backbone S1 S2 S3 S4 S5 S6 1.12 1.76 2.18 2.47 2.69 2.86 3.49 2.75 2.27 1.93 1.68 1.48 1:2 1:4 1:6 1:8 1:10 1:12 To each of the above rapidly stirred solutions 100 mL of aqueous solution containing 1:2 molar ratio of FeCl2:4H2O (1.19 g) and FeCl3:6H2O (3.23 g) was added. The resulting colloidal mixture was sonicated for 30-40 minutes to ensure homogeneous distribution of Fe2+ and Fe3+ in the colloidal solution of the matrix system. The chloride salt of iron was then converted into oxide by adding 5-6 mL ammonia while the solution is under sonication. Immediately the colloidal solution becomes dark indicating the formation of magnetic particles. Sonication continued for ~ 1 h and left for few hours before suction filtration. The obtained materials were vacuum dried at 70 oC. This procedure gives 1.39 g of Fe3O4 tiny particles entrapped into the spaces provided by 4.15 g of the palmatoyl-modified PVOH matrix. In other words, the percent of magnetite in each matrix is 25.1 %. Approximate particle size of samples was determined using X-ray diffraction and Debye– Scherrer formula. The vibrating sample magnetometer has become a widely used instrument Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 53-60 for determining magnetic properties of a large variety of materials: diamagnetic, paramagnetic, ferromagnetic and antiferromagnetic. In this case we used VSM MicroMag 3900, Princeton Measurements Corporation. The value of magnetic field was between 0 to 1 Tesla at different temperatures. The source of γ – ray in Mössbauer device was a 25 mCi of Co57. The computer processing of the spectra showed intensities I of the components (atomic fraction of Fe atoms), hyperfine inductions Bhf, isomer shifts δ, and quadrupole splitting QS. 3. Mathematical Section X-ray diffraction (XRD) is a versatile, nondestructive technique that reveals detailed information about the chemical composition and crystallographic structure of natural and manufactured materials. The Debye–Scherrer formula can be used to determine the size of particles of crystals in the form of powder. The Debye– Scherrer formula can be written as [28]: where M is the total measured magnetization, a is a fitting parameter and χp is the high field paramagnetic susceptibility. Fig. 1 shows the X-ray diffraction patterns of uncoated and coated Fe3O4 magnetite NP’s synthesized by sonication method. All peaks of the uncoated Fe3O4 particles matches exactly the prepared peaks of six coated samples. The calculations of uncoated and coated Fe3O4 particles made on the peak centered at 41o, using Equation (1). The average diameter of the particles assuming spherical Fe3O4 clusters is of the order of 13 nm (nano-sized particles). (1) where D is the mean size of the ordered domains, K is a dimensionless shape factor, λ is the X-ray wavelength (1.54056 Å), β is the line broadening at half the maximum intensity (FWHM). 3.2. Langevin Function The Langevin function can be written as [29]: 1 M = coth(a ) − , Ms a (5) p S1 S2 S3 S4 S5 S6 Fe3O4 Intensity (a. u.) Kλ , β cos θ M*(H, T) =M (H, T) - χ H, 4. Results and Discussion 3.1. X-ray Diffraction (XRD) D= field on the material (Oe), kT: is the thermal energy (eV), χ: is the susceptibility. The reduced magnetization M*(H, T) can be obtained by [30]: (2) where M is the total magnetization (emu/g), Ms is the saturation magnetization (emu/g), a is the ratio of the Zeeman energy of the magnetic moment in the external field to the thermal energy. The Langevin theory also leads to the Curie law. For small a [29]: M = nμ 2 H 3kT (3) χ= nμ 2 , 3kT (4) Therefore: where n is the number of atoms per unit volume, µ is the magnetic moment (emu), H is the acted magnetic 20 40 60 80 100 2θ Fig. 1. X-ray diffraction patterns of all samples with Fe3O4. The magnetization (M) versus the applied magnetic field (H) was carried out at room temperature as shown in Fig. 2. The results showed weak hysteresis loop for all six uncoated samples at room temperature. The corriesive field (Hc) was too low to be measured, while the remnance magnetization (Mr) varies for samples as shown in Fig. 3 (a) for sample (S3), while Fe3O4 showed high value of magnetization compared to other samples as shown in Fig. 3 (b). 55 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 53-60 shown in Fig. 5. The susceptibility, the saturation magnetization Ms and the average magnetic dipole moment for all samples are calculated and tabulated in Table 2. S1 S2 S3 S4 S5 S6 5 8 0 6 -5 -10 -10 -5 0 5 10 H (kOe) M (emu/g) Magnetization (emu/g) 10 Magnetization (emu/g) Mr=0.42 (emu/g) T= T= T= T= T= 2 Fig. 2. Magnetic hysteresis curves of all coated samples. Hc=42 Oe 4 298 323 373 423 473 o K K K o K o K o o 8 0 0 6 2 4 6 8 10 H (kOe) 4 (a) 2 0 -1000 -500 0 500 1000 H (Oe) -2 60 -4 M (emu/g) -6 Hc=51 Oe Mr=5.2 (emu/g) Magnetization (emu/g) (a) 80 40 20 0 -500 0 0 500 2 4 6 8 10 1000 H (kOe) H (Oe) -20 -40 (b) -60 Fig. 4. Selected isothermal total magnetization measurements for (a) sample 3 (S3), and (b) sample Fe3O4 at different selected temperature from 298 to 473 (Ko). (b) Fig. 3. Magnetic hysteresis curves of (a) sample 3 (S3), and (b) Fe3O4. The isothermal magnetization curves of different samples have been determined at temperatures between 298 to 473 Ko. The isothermal curves of samples show a large initial slope and nearly linear behavior for large fields; this suggests that the system contains paramagnetic and apparently superparamagnetic contribution, as shown in Fig. 4. A very good agreement between the reduced magnetization and the Langevin function found as 56 T= 298 oK T= 323 oK T= 373 oK T= 423 oK T= 473 oK 20 60 0 -1000 40 Table 2. The susceptibilities, the saturation magnetization Ms and the average magnetic dipole moment µ. Sample χo χp S1 S2 S3 S4 S5 S6 Fe3O4 2.2 1.46 1.71 1.36 1.37 1.51 15.53 1120 966 1420 1444 1070 2150 5260 Ms (emu/g) 9.86 7.51 6.51 6.11 5.87 5.23 63.46 μ (emu/g) 25.4 24.9 24 20.8 20.1 19.4 30.5 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 53-60 M (Measured) M* Langevin 10 M (emu/g) 8 6 4 2 0 0 2 4 6 8 10 H (kOe) Fig. 5. Magnetization curve of the sample 3 (S3). The best least square fit with Equation (2). The Mössbauer spectra show magnetic ordering with broad magnetic splitting, and superparamagnetic behavior. Hence, the spectra are fitted with (one or more) magnetic sextets and one quadrupole doublet. The fitted Mössbauer spectra are shown in Fig. 6. The Mössbauer parameters are listed in Table 3. Table 3. Hyperfine field Beff, Quadruple Splitting (QS), and Isomer Shift (δ) Results of Mössbauer Spectra for all samples. Sample S1 S2 S3 S4 S5 S6 Fe3O4 Sub spectra Sextet doublet Sextet Sextet Sextet doublet Sextet Sextet Sextet doublet Sextet Sextet Sextet doublet Sextet Sextet doublet Sextet Sextet doublet Sextet Sextet Sextet Sextet Sextet Beff (T) 10.4 44.4 31.3 18.3 46.5 37.9 26.1 46.8 40 19 46.5 26.2 46.1 20.7 49.9 47.6 43.9 39.1 21.7 QS δ (mm/s) (mm/s) 0.27 0.74 0.33 0.36 0.36 0.36 0.71 0.38 0.28 0.28 0.28 0.69 0.32 0.34 0.34 0.34 0.70 0.37 0.50 0.50 0.73 0.38 0.37 0.37 0.73 0.38 0.32 0.32 0.34 0.37 0.33 The spectrum for Sample 1 (S1) is fitted by one broad magnetic sextet with a hyperfine field (Bhf =10 T) and one quadrupole doublet with quadrupole splitting (QS=0.74 mm/s) and relative intensity I %=80 %. The spectrum for Sample 2 (S2) is fitted with three magnetic sextets with an average hyperfine field (Bhf =35.9 T) and one quadrupole doublet with quadrupole splitting (QS=0.71 mm/s) and relative intensity I %=48 %. The spectrum for Sample 3 (S3) is fitted with three magnetic sextets with an average hyperfine field (Bhf =39.9 T) and one quadrupole doublet with quadrupole splitting (QS=0.69 mm/s) and relative intensity I %=65 %. The spectrum for Sample 4 (S4) is fitted with three magnetic sextets with an average hyperfine field (Bhf =37.9 T) and one quadrupole doublet with quadrupole splitting (QS=0.70 mm/s) and relative intensity I %=58 %. The spectrum for Sample 5 (S5) is fitted with two magnetic sextets with an average hyperfine field (Bhf =39.7 T) and one quadrupole doublet with quadrupole splitting (QS=0.73 mm/s) and relative intensity I %=70 %. The spectrum for Sample 6 (S6) is fitted with two magnetic sextets with an average hyperfine field (Bhf =39.2 T) and one quadrupole doublet with quadrupole splitting (QS=0.73 mm/s) and relative intensity I %=63 %. The spectrum for sample Fe3O4 is fitted by five magnetic sextets with an average hyperfine field (Bhf =43 T) without quadrupole splitting. The magnetic ordered phases represented by magnetic sextets correspond to iron atoms in an iron oxide phases (magnetite) with large particle sizes, large enough to have net magnetic moment manifested by magnetic Zeeman splitting but not large enough to have well define magnetic splitting as in bulk magnetite. The quadrupole doublet in the spectra which is found to be around (QS ≈ 0.70 mm/s) could be attributed to iron oxide phase (most probable magnetite as the XRD data shows) with small particle sizes, small enough that the particles behave superparamagnetic (zero net magnetic moment). The relative intensity of the quadrupole doublet is found to be greater than that of the magnetic sextet in the spectra of nearly all the samples. This indicates that the iron oxide phases produced are below the blocking volumes at room temperature or blocking temperatures below room temperature (fine particle sizes), hence, behaving superparamagnetically. In brief; all samples show doublet as indicative of superparamagnetic particles of magnetite and slight hyperfine splitting. The MS parameters are similar to all samples indicating that the iron oxide particles have the same environment for all samples. The slight difference of hyperfine spectrum in samples suggests that small sized particles are produced (fine nanoparticles) when iron oxides nanoparticles were synthesized in presence of PVOH-palmitoyl chloride matrix [30]. 57 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 53-60 S1 S3 Relative Transmission (a.u.) S2 -5 0 5 10 -10 -5 0 5 -10 10 -5 0 Velocity (mm/sec) Velocity (mm/sec) Velocity (mm/sec) S4 S5 S6 5 10 5 10 Relative Transmission (a.u.) -10 -10 -5 0 -10 10 5 Velocity (mm/sec) -5 0 5 -10 10 -5 Velocity (mm/sec) 0 Velocity (mm/sec) Relative Transmission (a.u.) Fe3O4 -10 -5 0 5 10 Velocity (mm/sec) Fig. 6. Mössbauer spectra of samples: S1, S2, S3, S4, S5, S6 and Fe3O4 sample. 5. Conclusions In this research, we report the preparation of iron oxide (Fe3O4) coated with PVOH polymer in different number of palmatoyl chloride relative to hydroxyl group on the backbone. The XRD data used to determine the average size of the Fe3O4 clusters is found to be around 10-13 nm. The magnetization measurement on all samples is 58 carried out at different temperature, revealing that all samples contain superparamagnetic contribution. The paramagnetic saturation magnetization was calculated using Langevin function and found to be between 4-9 emu/g for coated samples and 64 emu/g for Fe3O4 sample. The average magnetic dipole moment was calculated to be around 20 Bohr magnetons. Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 53-60 The Mössbauer data indicate that the samples have superparamagnetic behavior and fine particles. The isomer shift, the relative intensities and quadruple splitting appear to be independent on the number of palmatoyl chloride relative to number of hydroxyl group, and this was confirmed by the value of the slight hyperfine splitting. In short, only the Fe3O4 sample is suitable for practical medical applications such as, drug delivery systems and biosensing purposes due to its sufficiently high value of saturation magnetization and attraction to magnet ability. [13]. [14]. [15]. [16]. References [1]. A. K. Gupta, S. Wells, Surface modified superparamagnetic nanoparticles for drug delivery: preparation, characterization and cytotoxicity studies, IEEE Transaction on Nanobioscience, Vol. 3, Issue 1, 2004, pp. 66-73. [2]. A. K. Gupta, M. Gupta, Cytotoxicity suppression and cellular uptake enhancement of surface modified magnetic nanoparticles, Biomaterials, Vol. 26, Issue 13, 2005, pp. 1565-1573. [3]. H. Gu, K. Xu, C. Xu, B. Xu, Biofunctional magnetic nanoparticles for protein separation and pathogen detection, Chemical Communications, Vol. 9, Issue 9, 2005, pp. 941-949. [4]. F. Shamsipour, et al., Conjugation of Monoclonal Antibodies to Super Paramagnetic Iron Oxide Nanoparticles for Detection of her2/neu Antigen on Breast Cancer Cell Lines, Avicenna J. Med Biotechnol, Vol. 1, Issue 1, 2009, pp. 27-31. [5]. D. K. Kim, et al., Superparamagnetic iron oxide nanoparticles for bio-medical application, Scripta Materialia, Vol. 44, Issue 8-9, 2001, pp. 1713-1717. [6]. Andrea Fornara, et al., Tailored magnetic nanoparticles for direct and sensitive detection of biomolecules in biological samples, Nano Letters, Vol. 8, Issue 10, 2008, pp. 3423-3428. [7]. J. Lodhia, G. Mandarano, N. J. Ferris, S. F. Cowell, Development and use of iron oxide nanoparticles (Part 1): Synthesis of iron oxide nanoparticles for MRI, Biomedical Imaging and Intervention Journal, Vol. 6, No. 2, 2010, pp. 1-11. [8]. Y. Zhang, N. Kohler, M. Zhang, Surface modification of superparamagnetic magnetite nanoparticles and their intracellular uptake, Biomaterials, Vol. 23, No. 7, 2002, pp. 1553-1561. [9]. R. Y. Hong, et al., On the Fe3O4/Mn1-xZnxFe2O4 core/shell magnetic nanoparticles, Journal of Alloys and Compounds, Vol. 480, Issue 2, 2009, pp. 947-953. [10]. A. P. Douvalis, et al., Revealing the interparticle magnetic interactions of iron oxide nanoparticlescarbon nanotubes hybrid materials, in Proceedings of the International Conference on the Applications of the Mössbauer Effect (ICAME’09), Vienna, Austria, 19-24 July 2009, pp. 1-4. [11]. G. A. Al-Nawashi, S. H. Mahmood, A. D. Lehlooh, A. S. Saleh, Mössbauer spectroscopic study of orderdisorder phenomena in Fe3-xMnxSi, Physica B: Condensed Matter, Vol. 321, Issues 1-4, 2002, pp. 167-172. [12]. S. M. Yusuf, et al., Structural and magnetic properties of amorphous iron oxide, Physica B: [17]. [18]. [19]. [20]. [21]. [22]. [23]. [24]. [25]. [26]. [27]. Condensed Matter, Vol. 405, Issue 4, 2010, pp. 1202-1206. C. Yee, et al., Self-Assembled Monolayers of Alkanesulfonic and -phosphonic Acids on Amorphous Iron Oxide Nanoparticles, Langmuir, Vol. 15, No. 21, 1999, pp. 7111-7115. J. H. Yu, C. W. Lee, S. S Im, J. S. Lee, Structure and magnetic properties of SiO2 coated Fe2O3 nanoparticles synthesized by chemical vapor condensation process, Reviews on Advanced Materials Science, Vol. 4, Issue 1, 2003, pp. 55-59. P. S. Stephen, Low viscosity magnetic fluid obtained by the colloidal suspension of magnetic particles, U.S. Patent 3 215 572, 2 November 1965. R. E. Rosensweig, J. W. Nestor, R. S. Timmins, Ferrohydrodynamic fluids for direct conversion of heat energy, in Mater. Assoc. Direct Energy Convers. Proc. Symp. AIChE-I. Chem. Eng. Ser. 5, 1965, pp. 104-118. R. Kaiser, G. Miskolczy, Magnetic properties of stable dispersion of subdomain magnetite particles, Journal of Applied Physics, Vol. 41, No. 3, 1970, pp. 1064-1072. A. E. Berkowitz, J. A. Lahut, C. E. Van Buren, Properties of magnetic fluid particles, IEEE Transaction on Magnetics, Vol. 16, Issue 2, 1980, pp. 184-190. W. C. Elmore, Ferromagnetic colloid for studying magnetic structures, Physical Review Letters, Vol. 54, Issue 4, 1938, pp. 309-310. R. Massart, Preparation of aqueous magnetic liquids in alkaline and acidic media, IEEE Transaction on Magnetics, Vol. 17, Issue 2, 1981, pp. 1247-1248. R. S. Molday, D. Mackenzie, Immunospecific ferromagnetic iron-dextran reagents for the labeling and magnetic separation of cells, Journal of Immunological Methods, Vol. 52, Issue 3, 1982, pp. 353-367. J. Lee, T. Isobe, M. Senna, Preparation of ultrafine Fe3O4 particles by precipitation in the presence of PVA at high pH, Journal of Colloid and Interface Science, Vol. 177, Issue 2, 1996, pp. 490-494. H. Pardoe, W. Chua-Anusorn, T. G. St. Pierre, J. Dobson, Structural and magnetic properties of nanoscale iron oxide synthesized in the presence of dextran, or polyvinylalcohol, Journal of Magnetism and Magnetic Materials, Vol. 225, Issue 1-2, 2001, pp. 41-46. R. Abu-Much, U. Meridor, A. Frydman, A. Gedanken, Formation of a three-dimensional microstructure of Fe3O4-poly (vinyl alcohol) composite by evaporating the hydrosol under a magnetic field, Journal of Physical Chemistry B, Vol. 110, Issue 16, 2006, pp. 8194-8203. P. J. Reséndiz-Hernández, O. S. RodríguezFernández, L. A. Garcia-Cerda, Synthesis of poly (vinyl alcohol)–magnetite ferrogel obtained by freezing–thawing technique, Journal of Magnetism and Magnetic Materials, Vol. 320, Issue 14, 2008, pp. e373-e376. H. L. Liu, C. H. Sonn, J. H. Wu, K. M. Lee, Y. K. Kim, Synthesis of streptavidin-FITC-conjugated core-shell Fe3O4-Au nanocrystals and their application for the purification of CD4(+) lymphocytes, Biomaterials, Vol. 29, Issue 29, 2008, pp. 4003-4011. J. Drbohlavova, et al., Preparation and Properties of Various Magnetic Nanoparticles, Sensors, Vol. 9, Issue 4, 2009, pp. 2352-2362. 59 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 53-60 [28]. B. D. Cullity, S. R. Stock, Elements of X-Ray Diffraction, 3rd ed., Prentice-Hall Inc., Upper Saddle River, NJ, 2001. [29]. B. D. Cullity, C. D. Graham, Introduction to Magnetic Materials, 2nd ed., Addison-Wesley Publishing Company, 2009. [30]. P. V. Finotelli, D. A. Sampaio, M. A. Morales, A. M. Rossi, M. H. Rocha-Leão, Ca Alginate As Scaffold For Iron Oxide Nanoparticles Synthesis, Brazilian Journal of Chemical Engineering, Vol. 25, No. 4, 2008 pp. 759-764. ___________________ 2015 Copyright ©, International Frequency Sensor Association (IFSA) Publishing, S. L. All rights reserved. (http://www.sensorsportal.com) 60 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 61-65 Sensors & Transducers © 2015 by IFSA Publishing, S. L. http://www.sensorsportal.com Larger Selectivity of the V2O5 Nano-particles Sensitivity to NO2 than NH3 1, 2, 3 1 Amos Adeleke Akande, 1 Bonex Wakufwa Mwakikunga, 2 Koena Erasmus Rammutla, 3 Augusto Machatine DST/CSIR National Centre for Nano-Structured Materials, P O Box 395, Pretoria 0001, South Africa 2 University of Limpopo, Department of Physics, P/Bag X1106, Sovenga, 0727, RSA 3 School of Physics, University of Pretoria, Pretoria, 0002, South Africa 1 Tel.: +27 12 841 4771, fax: +27 12 841 2229 1 E-mail: [email protected], [email protected] Received: 26 March 2015 /Accepted: 31 August 2015 /Published: 30 September 2015 Abstract: V2O5 nanoparticles (NPs) were prepared using microwave irradiation technique and characterized using X-ray diffraction (XRD), Raman spectroscopy (RS), Field emission scanning electron microscopy (FESEM). The physiosorption analysis with the aid of Brunauer-Emmiter-Teller (BET) method shows high surface area and relatively high pore diameter. The material’s gas sensing capabilities was tested for NH3 and NO2 keeping operating temperature at 300 K. An increase in electrical resistance were observed for both NH3 (reducing gas) and NO2 (oxidizing gas). This increase in resistance has been explained from the fact that V2O5 possess both n-type and p-type conductivity with NH3 preferring to interact with the n-type phase and NO2 attaching to p-type adsorption sites. The sensitivity of the p-type V2O5 phase to NO2 is found to be 32 times greater than the sensitivity of the n-type V2O5 phase to NH3. The results show that V2O5 is 32 times more sensitive to NO2 than NH3. Copyright © 2015 IFSA Publishing, S. L. Keywords: V2O5, Nano-particles, Selectivity, Sensitivity, NH3, NO2, Oxidizing gas, Reducing gas, n-type, p-type, Conduction band. 1. Introduction Nanoscale materials are very suitable for gas detection at molecular level due to their inherent small size, high conductance and large surface-tovolume ratio [1]. Wide band gap semiconductor metal oxides like SnO2, ZnO, WO3, V2O5 and TiO2 are widely investigated materials for gas sensors application because of their simplicity, easy to synthesize, cost effective and capability of detecting large number of toxic and volatile gases under different conditions [1]. Vanadium has been http://www.sensorsportal.com/HTML/DIGEST/P_2724.htm extensively studied as semiconductor materials because of its ability to exhibit metal-to-insulator phase transition with respect to temperature and pressure [2-3]. Vanadium pentoxide (V2O5) forms the most stable oxide among other binary oxides of vanadium, it exhibits orthorhombic crystallographic structure at TC = 375 °C with a band gap of 2.5 eV [3-4]. Sensing capability of V2O5 nanostructures has been achieved for nitrogen monoxide [5] and nitrogen dioxide and ethanol [5], the material has also been applied in thermo-chromic window, electrochromic window and electrochemical devices [6]. In 61 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 61-65 this current work, we report the chemiresistive properties of V2O5 NPs to NH3 and NOx gases. 2. Experimental 0.5 grams of Ammonia metavanadate NH4VO3 powder (purity 99.99 %) was ultrasonically dissolved in 10 mL distilled water, and 5 moles of N2H2 reagent was added drop-wise. The mixture was transferred into 100 mL Teflon vessel and placed onto the Multiwave 3000 Microwave reactor. The reactor power was set to 600 W, the temperature was maintained at 180 °C, and the reaction was allowed to run for 20 minutes after which the reactor cooled the vessels for 25 minutes. Afterwards, the resultant mixture was collected by filtration and washed repeatedly using isopropanol and acetone in an ultrasonic bath to remove undesired impurities and to minimise particle agglomeration. The final product was dried at 100 °C for 11 hours. The powder was characterized using a Panalytical X’ pert Pro PW 3040/60 XRD equipped with Cu Kα (λ=0.154nm) monochromatic radiation source. XRD patterns were recorded at 45.0 kV and 40.0 mA from 2θ = 5 to 90°. Raman spectroscopic studies were conducted using a Jobin–Yvon T64000 Raman spectrograph with a 514.5 nm excitation wavelength from an argon ion laser. The power of the laser at the sample was low enough (0.384 mW) in order to minimise localised heating of the sample. The T64000 was operated in a single spectrograph mode, with the1800 lines/mm grating and a 100x objective on the microscope. Morphology studies were carried out using a LEO 1525 field emission scanning electron microscope (FESEM). BET analyses were carried out using Micromeritics TriStar II series Surface Area and Porosity instrument and a Micromeritics sample degassing system from USA. Fig. 1. Schematic diagram of KSGA565 KENOSISTEC sensing station illustrating how the gas sensing measurement was performed. Fig. 2. XRD pattern of V2O5 NPs. However, the development of VO2 monoclinic phase was also observed at 2θ=27.6° [PCDPDFWIN CAS No. 710042]. The crystallite size of the particles was calculated using Debye Scherer’s model and found to be 5 nm. The SEM micrograph of V2O5 NPs in the Fig. 3 shows formation of sphere-like colloidal particles. 3. Gas Sensor Test Gas testing measurements were achieved using a set-up (at the university of Cologne Germany) similar to the KSGA565 KENOSISTEC sensing measurement illustrated in Fig. 1. The sensor was prepared by dispersing V2O5 NPs in ethanol and making a thick paste on the interdigitated electrode and the test were performed by measuring change in electrical resistance of different concentration of the analyte using KEITHLEY picoammeter system. 4. Results and Discussion Fig. 2 shows the XRD pattern of V2O5 nanoparticles (NPs) belonging to the orthorhombic phase of V2O5. The broad diffractions at 2θ = 17° are the characteristics of the orthorhombic V2O5 200 reflection [PCDPDFWIN CAS No. 890611]. 62 Fig. 3. SEM micrograph of V2O5 NPs. Raman specrum of V2O5 NPs in Fig. 4 shows Raman active modes of V2O5. A strong peak at lower frequency band 141 cm-1 corresponding to the Bg symmetry while the high frequency vibration at 992 cm-1 corresponds to stretching of O-V-O atoms. Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 61-65 Vibration modes at 282, 402 and 525 cm-1 are resulting from the bending mode of V-O band while the stretching mode at 692 cm-1 corresponds to the motion parallel and perpendicular to ab-plane [3, 7-8]. In agreement with XRD analysis, the development of VO2 monoclinic vibration mode was also observed at 194 cm-1 [9]. Fig. 5. N2 adsorption/dsorption isotherms profile of V2O5 NPs, insert is the BET plot. The specific surface area SBET and pore size/diameter (dpor) of 92.7 m2g-1 and dpor of 12 nm, respectively, were determined by physisorption of nitrogen according to BET theory as shown in Fig. 5. The surface area and the nano-porous studies by SEM and BET revealed the material’s potential in gas and chemical sensor application. The measured electrical response for different NH3 concentrations is presented in Fig 6 (a), with operating or substrate temperature kept at 28 °C. The material showed good response to NH3, Fig. 6 (b) shows linear profile for NH3 sensitivity plot, where Rgas and Rair are the resistance of V2O5 NPs in the presence of the analyte gas and air respectively [10, 11]. The measured electrical response for different NO2 concentrations is presented in Fig. 7 (a) with operating temperature kept at 28 °C. The material showed good response for all conncentrations. Increase in the electrical resistance of V2O5 NPs sensor upon injection of the oxidising gas NO2 shows p-type semiconductor behavour. Fig. 7 (b) is the NO2 sensitivity profile. (a) (a) (b) (b) Fig. 6. (a) The electrical response of V2O5 NPs for different concentrations of NH3, (b) Sensor response as a function of different NH3 gas concentrations. Fig. 7. (a) The electrical response of V2O5 NPs for different concentrations of NO2, (b) Sensor response as a function of different NO2 gas concentrations. Fig. 4. Raman spectrum of V2O5 NPs. 63 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 61-65 The proposed mechanism bases on the fact that V2O5 is known to exhibit both p-type and n-type conductivities [3-4]. The results show that NH3 prefers to interact with the n-type phases of V2O5. In this case the NH3 molecule captures the adsorbed O2ions from the V2O5 surface and oxides to NO2 and H2 as depicted in Fig. 8. This process robs the n-type phase of V2O5 of its electrons leading to less electron population in the conduction band (CB) and hence increase in resistance as shown in Fig. 6 (a). assisted techniques. The SEM image and BET nano-porous analysis revealed the large surface property of the material. Electrical response for different concentrations of these gases showed good response, and the reaction mechanism between the gases and sensor showed that V2O5 NPs is predominantly a p-type semiconductor with a ratio of p-type- to- n-type of 32:1. This suggests that for every one electron in V2O5 there are 32 holes. We conclude from the results that V2O5 is 32 times more selective to NO2 than to NH3. Acknowledgements Support from the India-Brazil-South Africa trilateral cooperation under the National Research Foundation (NRF) grant number HGER24X is acknowledged. Amos A. Akande also acknowledge Dr. Ella C. Linganiso for the sensing measurement. References Fig. 8. Sensing mechanism of V2O5 NPs. As for the proposed sensing scheme for NO2, the results show increase in resistance regardless of the fact that NO2 is an oxidising gas. This is clear evidence that NO2 prefers to interact with the p-type phase of V2O5. In this case NO2 interacts with holes or vacancies in the adsorbed O2-. When such a hole interacts with NO2, the NO2 is broken into N2 and adsorbed O2-. Only N2 is released in the process leaving behind the O2 as adsorbed at the V2O5 surface as O2-. In the process the p-type phase will have gained and electron or lost a hole in the conduction. The loss of holes in the CB of a p-type V2O5 leads to an increase in resistance as shown in Fig. 7 (a). It is worth noting that the magnitude of responses of V2O5 sensor to both NH3 and NO2 are many magnitude different. The sensitivity of the n-type phase to NH3 is calculated as the slope of the response vs concentration curve in Fig. 6 (b). This works out to be 3.47 × 10-3 ppm-1. When the same sensitivity calculation is done for the p-type phase to NO2 form Fig. 7 (b) one finds the value of 0.11. The response of the p-type phase to NO2 is therefore 32 time greater than the sensitivity of the n-type phase to NH3. Therefore, this V2O5 sensor is 32 times more selective to NO2 than to NH3. This further suggests that the V2O5 nanomaterials is predominantly p-type. 7. Conclusions We report gas sensing capability of vanadium pentoxide nanoparticles synthesized using microwave 64 [1]. Vishah B., Samanta S., Singh A., Debnath A. K., Aman Mahajan, Bedi R. K., Aswal D. K., Gupta S. K., Chemiresistive gas sensing properties of nanocrystalline Co3O4 thin films, Sensors & Actuators, B: Chemical, Vol. 176, 2013, pp. 38-45. [2]. Morin F. J., Oxides Which Show a Metal-to-Insulator Transition at the Neel Temperature, Physical Review Letters, Vol. 3, Issue 1, 1959, pp. 34-36. [3]. A. A. Akande, E. C. Linganiso, B. P. Dhonge, K. E. Rammutla, A. Machatine, L. Prinsloo, H. Kunert, B. W. Mwakikunga, Phase evolution of vanadium oxides obtained through temperature programmed calcinations of ammonium vanadate in hydrogen atmosphere and their humidity, Materials Chemistry and Physics, Vol. 151, 2015, pp. 206-214. [4]. Rao M. C., Vanadium Pentoxide Cathode Material for Fabrication of all Solid State Lithium-Ion Batteries - a Case Study, Research Journal of Recent Sciences, Vol. 2, No. 3, 2013, pp. 67-73. [5]. Huotari J., Spetz A. L., Lappalainen Gas sensing properties of pulsed laser deposited vanadium oxides thin film, Proceedings of the 14th International Meeting on Chemical Sensors (IMCS’12), 2012, pp 279-282. [6]. A. A. Akande, K. E. Rammutla, T. Moyo, N.S.E Osman, S. Nkosi, C.J. Jafta, B. W. Mwakikunga, Magnetism variations and susceptibility hysteresis at the metal-insulator phase transition temperature of VO2 in a composite film containing vanadium oxides, J. Magn. Magn. Mater., 375, 2015, pp 1-9. [7]. Mwakikunga B. W., Sidras-Haddad E., Maaza M., First synthesis of vanadium dioxide by ultrasonic nebula-spray pyrolysis, Optical Materials, Vol. 29, 2007, pp. 481-487. [8]. Se-Hee Lee, Cheong H. M, Seong M., Liu J. P., Tracy C. E, Mascarenhas A., Pitts J. R., Deb S. K., Raman spectroscopic studies of amorphous vanadium oxide thin films, Solid State Ionics, Vol. 165, 2003, pp. 111-116. [9]. Petrov G. I., Yakovlev V. V., Raman microscopy analysis of phase transformation mechanisms in Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 61-65 vanadium dioxide, Applied Physics Letters, Vol. 81, No. 6, 2002, pp. 1023-1025. [10]. Watchakun K., Samerjai T., Tamaekong N., Liewhiran C., Siriwong C., Kruefu V., Wisitsoraat A., Tuantranont A., Phanichphant S., Semiconducting metal oxides as sensors for environmentally hazardous gases, Sensors & Actuators, B: Chemical, Vol. 160, 2011, pp. 580-591. [11]. Yu M., Liub X., Wang Y., Zheng Y., Zhang J., M. Li, Lan W., Q. Su, Gas sensing properties of p-type semiconducting vanadium oxide nanotubes, Applied Surface Science, Vol. 258, 2012, pp. 9554-9558. ___________________ 2015 Copyright ©, International Frequency Sensor Association (IFSA) Publishing, S. L. All rights reserved. (http://www.sensorsportal.com) 65 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 66-73 Sensors & Transducers © 2015 by IFSA Publishing, S. L. http://www.sensorsportal.com Surface Morphology, Compositional, Optical and Electrical Properties of TiO2 Thin Films S. S. Roy, A. H. Bhuiyan Department of Physics, Bangladesh University of Engineering and Technology, Dhaka-1000, Bangladesh Tel.: 88-01711983489 E-mail: [email protected] Received: 31 August 2015 /Accepted: 28 September 2015 /Published: 30 September 2015 Abstract: Titanium oxide (TiO2) thin films have been deposited on to glass substrate by spray pyrolysis deposition technique (SPDT). The surface morphological, structural, electrical and optical properties of the asdeposited TiO2 thin films have been investigated as a function of substrate temperature (Ts). The scanning electron micrographs of as-deposited films showed uniform surface of TiO2 thin films. Elemental analysis clearly showed that the grains were typically comprised of both Ti and O in the thin films. Strong diffraction peaks (101) and (200) at 25° and 48° respectively indicating TiO2 in the anatase phase .The peaks were found to shift slightly from their standard positions at higher Ts, and there was some deviation in the lattice parameters. The crystallite size is found to be around 13 nm. The optical transmission of the thin films was found to increase from 73 to 89 % and the band gap energy shifts from 3.64 to 3.40 eV with increase of Ts. The room temperature dc electrical resistivity varies from 42 to 27 ohm.cm for the thin films grown at different Ts. Copyright © 2015 IFSA Publishing, S. L. Keywords: TiO2, SPDT, Ts, Anatase, Optical band gap, DC electrical resistivity. 1. Introduction The increasing demands for microelectronics and microstructural components in different branches of science and technology have greatly expanded the sphere of research of thin films [1-2]. Transparent and conducting oxides (TCOs) such as zinc oxide (ZnO), titanium oxide (TiO2), nickel oxide (NiO), indium oxide (In2Ox), aluminum oxide (Al2O3), tin oxide (SnO2) etc. are used for a variety of applications including architectural windows, solar cells, flat-panel displays, smart windows, polymerbased electronics etc. [3-8]. From these materials, TiO2 shows unique characteristics in chemical inertness, stability to heat treatment and mechanical hardness. Compact TiO2 thin films deposited on conducting glass are used in new types of solar cells: 66 liquid and solid dye-sensitized photo electrochemical devices among other uses [9-10]. TiO2 photocatalyst films having an anatase crystal structure with different thicknesses were prepared by the low-pressure metal– organic chemical vapor deposition (LPMOCVD) to examine the effect of growth conditions on photocatalytic activity [11]. Film thickness was linearly proportional to the deposition time. Structure of the film was strongly dependent on the deposition time. The optimum thickness of TiO2 catalyst film grown by LPMOCVD may locate between 3 and 5 μm. M/TiO2 (M = metal; Ag, Cu) thin films on quartz glass were prepared by radio frequency (rf) magnetron co-sputtering process and the calcination effects on their optical and structural properties were investigated [12]. These films were amorphous below 300 oC, and these were http://www.sensorsportal.com/HTML/DIGEST/P_2725.htm Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 66-73 anatase phase at 300–700 °C. The crystallite size of the anatase phase and the agglomerates of primary particles of the M/TiO2 thin films increase with increasing calcination temperature. Preparation of TiO2 films with superior optical and structural characteristics was attempted using a conventional RF magnetron sputtering by modifying deposition variables, such as the substrate temperature (Ts) and the gas composition in the sputtering ambient [4]. TiO2 thin films with higher refractive index and better homogeneity were obtained with substrate heating between 200 °C and 400 °C. TiO2 thin films were prepared by chemical spray pyrolysis from aqueous solutions and it was shown that it had three different kinds of polymorphous crystalline forms: rutile, anatase, and brookite [13]. The rutile phase is always formed at higher temperatures, while the anatase phase is formed at lower temperatures and transformed into rutile phase above Ts of 800 ºC and the refractive index lies in the range between 2.01–2.29. TiO2 thin films prepared with and without lithium (Li) and niobium (Nb) were uniform, crackfree, stoichiometric, and amorphous when deposited at 300 °C and below; and were polycrystalline anatase when deposited at 400 °C. Films prepared around 200 °C were very porous, but the porosity was decreased as the Ts increased [14]. Optical absorption spectra revealed an indirect band gap of 3.0 eV for amorphous and anatase films and a direct band gap of the same value in rutile. Dark dc conductivity of undoped films was lower than 10-10 (Ohm.cm)-1. The presence of Nb and Li increased the conductivity by 2–3 orders of magnitude, similar to the effect of hydrogen annealing. TiO2 thin films were obtained using the MOCVD method [15], film thickness increased with deposition time as expected, while the transmittance varied from 72 to 91 % and the refractive index remained close to 2.6. It is observed that there are many techniques, including solgel, sputtering, MOCVD, evaporation and chemical vapour or spray deposition, by which the TiO2 films may be deposited on to glass substrates [12-16]. In this study, TiO2 thin films were prepared by the spray pyrolysis deposition technique (SPDT) which is particularly attractive because of its simplicity, fast, inexpensive, and suitable for mass production [16]. So, the aim is to grow TiO2 thin film by SPDT and to study the effect of the Ts on the structural, optical and electrical properties of TiO2 thin films. 2. Experimental Details TiO2 thin films were deposited using homemade spray pyrolysis set up from aqueous solution of titanium chloride (TiCl4). 0.1M of TiCl4 was added with 50 ml water and 50 ml ethanol for precursor solution for TiO2 thin film. The distance between substrate to spray nozzle was 25 cm and air pressure was 1 bar. To enhance the solubility of prepared solution, a few drops of HCl were added. The transparent solution thus obtained and subsequently diluted by ethanol, served as the precursor. The solution was sprayed onto the organic solvent and ultrasonically cleaned glass substrates heated at five different Ts, namely 250, 300, 350, 400 and 450 °C. Ts was recorded using a cromel-alumel thermocouple. The flow rate of the solution during spraying was adjusted at about 1 ml/min and was kept constant throughout the experiment and the spray time was 5 min. For each concentration the reproducibility of the thin films was verified by repeating the experiments several times. The main reaction which leads to the formation of TiO2 may takes place as follows: TiCl4 + O2 (g) → TiO2 (s) + 2Cl2 (g) ↑ . The surface morphology of the films was examined by a HITACHI S-3400N model scanning electron microscope (SEM), the elemental analysis was performed by an electron dispersive spectrometer attached to the SEM, X-ray diffraction (XRD) patterns were recorded by a Philips PW3040 X’Pert PRO X-ray diffractometer. The optical transmission spectra for as-deposited thin films were obtained in the ultraviolet (UV) UV-visible (UV-vis.) and near infrared regions (300-1100 nm) using a UV-VIS spectrophotometer (Model: 1201V, Shimazdu, Japan). Electrical resistivity of the thin films was measured at room temperature by Vander Pauw technique. The temperature dependence of the electrical conductivity was also measured. 3. Results and Discussion 3.1. Surface Morphological and Elemental Analysis SEM images were recorded to examine the surface morphology of the as deposited TiO2 thin films and these images are shown in Fig. 1 (a, b, c, d). The surface of the thin film is uniform and homogeneous. SEM images show that there are no remarkable grains and impurity for the TiO2 thin films. SEM images reveal that sprayed particles (atoms) are absorbed onto the glass substrate into clusters as the primary stage of nucleation. It was observed that the coating was transparent and homogeneous without any visual cracking over a wide area. Multiple coating increased thickness, but did not affect the uniformity of the thin film. SEM micrographs reveal the formation of particles with different shapes and sizes. So, the SEM surface studies of TiO2 films exhibit a smooth and homogeneous growth on the entire surface. EDX analysis revealed the presence of titanium (Ti) and oxygen (O) and confirm the formation of TiO2 thin films through the chemical oxidation route free from impurities (Table 1). This implies that the prepared samples are made up of Ti and O. 67 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 66-73 Table 1. Quantitative results of TiO2 thin films from EDX analysis. Ts (oC) 250 400 450 Element (Atom %) Ti O 55.23 44.63 67.34 32.22 60.12 39.34 3.2. X-ray Diffraction Analysis XRD patterns for TiO2 thin films synthesized at different Ts are shown in Fig. 2. XRD patterns of TiO2 thin films indicate that the sample prepared at Ts of 250 °C was amorphous where these prepared at Ts viz, 300, 350, 400 and 450 °C were crystalline nature and there was no indication of other crystalline by-products. These diffraction patterns show that the TiO2 thin film contains only anatase, which is well known as the most suitable structure for the photocatalysis. The diffraction peaks observed at 2θ values of 25.25, 38.05, and 48.5° correspond to the diffraction lines produced by (101), (004), and (200) planes of the tetragonal structure with anatase phase and the diffraction data were in good agreement with data of 2θ and peak respectively [17-18]. 10 μm (a) 10 μm (b) 10 μm (c) 10 μm (d) Fig. 1. SEM images of TiO2 thin films at Ts (a) 250; (b) 350; (c) 400; and (d) 450 ºC. The (101) surface of TiO2 thin film is energetically the most stable and the predominant crystal face found in polycrystalline samples. Strong diffraction peaks at 25° and 48° are indicating TiO2 in the anatase phase [19]. The intensity of XRD peaks of the sample reflects that the formed are crystalline and broad diffraction peaks indicate the size of crystallite is very small. XRD data of TiO2 crystallite size was obtained by Debye-Scherrer’s formula [20]. The crystallite sizes (D) and the lattice 68 constants (a = b, c) are calculated from the XRD patterns and these values are given in Table 2. No trace of rutile was found in the as-deposited films regardless of deposition temperature. The increase of Ts increases the intensity of the peaks, it is due to the crystallinity of the TiO2 improves progressively. It is noticed that no additional XRD peaks corresponding to Ts> 250 oC. The values of average crystallite size (D) is in the range 12.74–13.05 nm for Ts of 300 to 450 oC. This means that the homogeneity of the thin Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 66-73 o 450 C 0.9 Transmittance, T (%) 0.8 o 400 C Intensity (a.u) high transmittance may be attributed to less scattering due to the decrease in the degree of irregularity in the grain size distribution [22]. (200) (004) (101) film is improved with substrate heating during the film deposition. o 350 C o 300 C 0.7 0.6 0.5 0.4 o 250 C o 300 C o 350 C o 400 C o 450 C 0.3 0.2 0.1 o 250 C 0.0 20 25 30 35 2θ 40 45 200 50 Table 2. XRD data for the TiO2 thin films at various Ts. D (nm) 12.7417 12.8498 12.8702 13.0451 600 800 1000 1200 Wavelength (nm) Fig. 2. XRD patterns of TiO2 thin film synthesized at various Ts. Ts (°C) 250 300 350 400 450 400 a, c (Å) 3.7870, 9.5194 3.7829, 9.0856 3.8072, 8.9276 3.8035, 8.9640 Fig. 3. Optical transmittance vs. wavelength of films at various Ts. The optical band gap of semiconductors is determined using the Tauc formula [23]. Fig. 4 shows the (αhν)2 as a function of hν for the TiO2 thin films deposited at various Ts. The α was found in the order of 105 m-1 which may be suitable for a transparent conducting film. In Fig. 4, it is observed that α first increases slowly in the low energy region and then increases sharply near the absorption edge so the value of the α depends on Ts. 3.3. Optical Properties 3.3.1. Transmittance and Optical Band Gap Transmission spectra were taken in the range of 300 to 1100 nm for TiO2 thin films and are shown in Fig. 3. It is seen from the graph that the value of transmittance is high in the visible and IR region, it is minimum at wavelength ~ 400 nm. Films prepared at Ts of 250 ºC exhibit a transmission of > 65 %, again it is found > 75% in the visible and infrared regions for TiO2 thin films prepared at Ts of 400 ºC. An average of 75 to 85 % transmittance is observed in the wavelength range of 800-1100 nm and below 600 nm transmittance decreases gradually. The transmittance increases from 10 to 15 % with Ts, and shows the highest transmittance of about 88 % for the thin films grown at Ts of 400 °C. The shift in the fundamental absorption edge is due to the structural changes as revealed by XRD analysis. The transmittance of the films is also influenced by a number of effects, which include surface roughness and optical in homogeneity in the direction normal to the film surface. The increase in transmittance may be due to the transition of the TiO2 films from amorphous to polycrystalline structure and relatively ( α h ν ) 2 (m -1e V ) 2 2.00E+015 250 300 350 400 450 1.50E+015 1.00E+015 o C C o C o C o C o 5.00E+014 0.00E+000 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 hν (eV) Fig. 4. Variation of (αhυ)2 with (hυ) TiO2 thin for TiO2 thin films at various Ts. The Eg of the TiO2 thin films decreases when Ts increases. At Ts of 250 °C the Eg was found to be 3.62 eV and a minimum value 3.40 eV was observed at Ts of 400 °C. The value of the Eg for Ts of 450 °C 69 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 66-73 is same as that of 400 oC. It can be seen that a band gap tuning of 0.22 eV occurs when Ts is changed by about 150 °C.by increasing of Ts. It may be due to the surface morphology and structure of the TiO2 thin films no change and no disorderness. The observed band gap values are in good agreement with reported values between 3.2 and 3.9 eV [16]. concentration for neutral defects and stochiometric changes of the films. The conductivity of the prepared TiO2 thin films increases as Ts increases as shown in Fig. 7. 0.24 0.22 The of refractive index, n for TiO2 thin films decreases with Ts, as shown in Fig. 5. N of TiO2 thin film has been obtained to be 2.75 at Ts of 250 °C and it became lowest 2.55 at Ts of 400° C. This value is lower than the reported value 2.80 of TiO2 thin film [24] and it is lower than that of bulk TiO2 and this low value of refractive index may probably be due to the smaller density of the films which is suggested by Arai [25]. Extinction coefficient, k 0.20 3.3.2. Refractive Index and Extinction Coefficient 250 300 350 400 450 0.18 0.16 0.14 0.12 C C C o C o C o O 0.10 0.08 0.06 0.04 0.02 200 400 600 800 Wavelength (nm) 1000 1200 Fig. 6. Variation of extinction coefficient with wavelength at various Ts. 2.8 Refractive Index. n 2.7 2.5 2.4 200 250 300 350o Ts ( C) 400 450 500 Fig. 5. Variation of refractive index with Ts. Resistivity, ρ (Ohm.cm) 2.6 50 48 46 44 42 40 38 36 34 32 30 28 26 24 22 20 250 It is observed that refractive index decreases as Ts increases. It may be due to the decrease of impurity in the film. The variation of extinction coefficient, k with wavelength is shown in Fig. 6. It is observed that k increases with the increase of Ts. The rise and fall in k is directly related to the absorption of light. The k about 0.05 in the range of wavelength 500-1100 nm is very close to the reported value of TiO2 thin films prepared by DC magnetron sputtering method [24]. 3.4. Electrical Properties The variation of the resistivity, ρ of TiO2 thin films as a function of Ts is shown in Fig. 7. It is observed that ρ of the as-deposited TiO2 thin films is decreased with increasing Ts. This behavior indicates the semiconducting nature of the TiO2 thin films. The decrease of resistivity means increase of conductivity with temperature which may due to the increase of carrier mobility or due to increase of carrier 70 o 300 350 400 o Substrate Temperature, Ts ( C) 450 Fig. 7. Electrical resistivity vs. Ts of TiO2 thin films. It may be due to increase in the free path of carrier concentration. At high temperature the mechanism of impurities thermal activation becomes the dominant one. The increase in σ is due to the increase in the crystalline nature as the temperature increases. Consequently, Ti4+ ions have more concentration in the films obtained at high Ts. Regarding the O2- ions, which results in an increase of the free electron concentration after there is a decrease in the resistivity of the films [26-27]. On the other hand, the increase in the conductivity with temperature can also be explained as follows: the grain size increases with temperature which leads to a decrease in grain boundaries and hence resistivity [21, 28]. This is clearly understood from Fig. 2 where one can see a single phase structure of TiO2 thin film. The single phase structure enhances the electron Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 66-73 mobility thus improve the conductivity [29]. So the high electrical conductivity has been found for TiO2 thin films deposited at Ts of 400 oC. The activation energy (ΔE) is calculated from the 0.25 -1.6 -1.8 -2.0 -2.2 -2.4 -2.6 -2.8 -3.0 -3.2 -3.4 -3.6 -3.8 2.0 Figure of Merit (ohm.cm) -1 ln σ ((Ohm.cm) ) -1 slope of the curves lnσ vs. (1/T) in Fig. 8 and the variation of ΔE with different Ts is shown in Fig. 9. The low value ΔE may be associated with the localized levels hopping due to the excitation of carriers from donor band to the conduction band. So the low value of ΔE indicates that the prepared sample is stoichiometric. 0.2238 Ω−1cm−1. The increase in the figure of merit of the TiO2 thin films is mainly due to the increase in the optical transmittance with increasing Ts. The experimental data suggest that Ts of 400 °C is the best Ts with other conditions for depositing highquality TiO2 thin films. o 250 C o 300 C o 350 C o 400 C o 450 C 0.20 0.15 0.10 0.05 250 300 350 400 450 o Ts ( C) Fig. 10. Variation of figure of merit versus Ts. 2.2 2.4 2.6 2.8 3.0 -1 1000/T (K ) 3.2 3.4 4. Conclusions Fig. 8. Electrical conductivity (lnσ) vs. inverse of absolute temperature of TiO2 thin films at various Ts. 0.030 Activation Energy (eV) 0.025 0.020 0.015 0.010 0.005 250 300 350 o Ts ( C) 400 450 Fig. 9. Variation of activation energy versus Ts. of TiO2 thin films. The Figure of merit is well-known as an index for evaluating the performance of transparent conducting films, and it is given by the equation F = (− ρlnT)−1 where ρ is the electrical resistivity and T is the average transmittance in the wavelength range of 800-1100 nm [17]. Fig. 10 shows the figure of merit values of TiO2 thin films deposited at various Ts. The figure of merit for the TiO2 thin films deposited at Ts of 250 - 400 °C were found to be 0.0666 - Transparent and homogeneous TiO2 thin films have been prepared using TiCl4, ethanol and water by employing a simple and inexpensive spray pyrolysis deposition technique. Roughness of TiO2 thin film is decrease with increasing Ts. The XRD pattern of thin films shows a single anatase phase with a strong peak (101) and particle size 13 of nm. The average transmittance of the TiO2 film were about 75 % in the wavelength range of 600-1100 nm, the optical band gap 3.4 eV, and the lowest value of refractive index 2.55 at Ts of 400 °C. The resistivity of TiO2 thin film decreases as Ts increases. The minimum resistivity is found to be 27 Ω-cm for TiO2 thin film deposited at Ts = 400 °C. The highest figure of merit occurred for the film grown at 400 °C with an optical transmittance about 85 % in the wavelength range of 800-1100 nm. Activation energy of TiO2 thin films varies in the range of 0.010 to 0.028 eV for different Ts. The results suggest that high-quality TiO2 thin film can be produced when deposited at a growth temperature of 400 °C. The obtained experimental results indicate the suitability of this material as transparent and conducting window materials in thin film solar cells and gas sensor devices. Acknowledgement The authors would like to thank Bangladesh University of Engineering and Technology (BUET), Dhaka, Bangladesh for supporting this research work. 71 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 66-73 References [1]. A. Kumar, Y. W. Chung, J. J. Moore, J. E. Smugeresky, Surface Engineering: Science and Technology 1, in Proceedings of the Minerals, Metals & Materials Society (TMS) Annual Meetings, Warrendale, 1999, pp. 177. [2]. Gloker D. A., Shah S. I., Handbook of Thin Film Process Technology, Institute of Physics Publishing, Bristol and Philadelphia, 1998. [3]. S. Mondal, K. P. Kanta, P. Mitra, Preparation of Aldoped ZnO (AZO) Thin Film by SILAR, Journal of Physical Sciences, Vol. 12, 2008, pp. 221-229. [4]. S.-H. Jeong, B.-S. Kim, B.-T. Lee, Structural and Optical Properties of TiO2 Films Prepared Using Reactive RF Magnetron Sputtering, Journal of the Korean Physical Society, Vol. 41, No. 1, July 2002, pp. 67-72. [5]. A. K Srivastava, S. Thota, J. Kumar, Preparation, Microstructure and Optical Absorption Behaviour of NiO Thin Films, Journal of Nanoscience and Nanotechnology, Vol. 8, No. 8, 2008, pp. 4111-4115. [6]. G. Kiriakidis, N. Katsarakis, M. Bender, E. Gagaoudakis, V. Cimalla, InO Thin Films, Candidates for Novel Chemical and Optoelectronic Applications, Materials Physics and Mechanics, Vol. 1, No. 2, 2000, pp. 83-97. [7]. H. Luo, M. Jain, Thomas M. M., E. Bauer, A. K. Burrell, Q. Jia, Optical and Structural Properties of Single Phase Epitaxial p-Type Transparent Oxide Thin Films, Advanced Materials, 19, 2007, pp. 3604-3607. [8]. J. Jachon, M. Varghes, K. E. Abraham, Studies on Cu, Fe, and Mn-doped SnO2 Semi Conducting Transparent Films Prepared by a Vapor Deposition Technique, Chinese Journal of Physics, Vol. 45, No. 1, 2007, pp. 84-97. [9]. D. W. Ayele, W.-N. Su, B.-J. Hwang, A Low-Cost Electrode Polytetrafluoroethylene-Framed TiO2 Decorated with Oleic Acid-Capped CdSe Quantum Dots for Solar Cell, Journal of Energy, Vol. 2013, 2013, Article ID 394512. [10]. T. S Kang, D. Kim, K. J. Kim, Time-Dependent Electrochromism of Nanocrystalline TiO2 Films in Propylene Carbonate Solution of LiClO4, Journal of The Electrochemical Society, Vol. 145, 1998, pp. 1982-1986. [11]. S.-C. Jung, S.-J. Kim, N. Imaishi, Y.-I. Cho, Effect of TiO2 thin film thickness and specific surface area by low-pressure metal–organic chemical vapor deposition on photocatalytic activities, Applied Catalysis B: Environmental, 55, 2005, pp. 253-257. [12]. S. W. Ryu, E. J. Kim, S. K. Ko, S. H. Hahn, Effect of calcination on the structural and optical properties of M/TiO2 thin films by RF magnetron co-sputtering, Materials Letters, Vol. 58, No. 5, 2004, pp. 582-587. [13]. R. Ayouchi, C. Casteleiro, R. Schwarz, J. R. Barrado, F. Martin, Optical properties of TiO2 thin films prepared by chemical spray pyrolysis from aqueous solutions, Physica Status Solidi C, 7, No. 3-4, 2010, pp. 933-936. [14]. N. Golego, S. A. Studenikin, M. Cocivera, Spray pyrolysis preparation of porous polycrystalline thin films of titanium dioxide containing Li and Nb, Journal of Materials Research, Vol. 14, No. 3, Mar. 1999, pp. 698-707. [15]. M. I. B. Bernardi, E. J. H. Lee, P. N. Lisboa-Filho, E. R. Leite, E. Longoa, J. A. Varela, TiO2 Thin Film 72 [16]. [17]. [18]. [19]. [20]. [21]. [22]. [23]. [24]. [25]. [26]. [27]. [28]. [29]. Growth Using the MOCVD Method, Journal of Materials Research, Vol. 4, No. 3, 2001, pp. 223-227. A. Gowrimanohari, S. Dhanapandian, K. Santhosh Kumar, T. Mahalingam, Optimization of Deposition Parameters on the Physical Properties of TiO2 Thin Films by Spray Pyrolysis Technique, International Journal of Thin Films Science and Technology, 3, No. 1, November 2014, pp. 1-6. R. Vijayalakshmi, V. Rajendran, Synthesis and characterization of nano-TiO2 via different methods, Archives of Applied Science Research, 4, 2, 2012, pp. 1183-1190. M. Ba-Abbad, A. H. Kadhum, A. Mohamad, M. S. Takriff, K. Sopian, Synthesis and Catalytic Activity of TiO2 Nanoparticles for Photochemical Oxidation of Concentrated Chlorophenols under Direct Solar Radiation, International Journal of Electrochemical Science, 7, 2012, pp. 4871-4888. K. Thamaphat, P. Limsuwan, B. Ngotawornchai, Phase Characterization of TiO2 Powder by XRD and TEM, Kasetsart Journal (Natural Science), 42, 2008, pp. 357-361. E. M. Mahdi., M. Hamdi, M. M. S. Meor Yusoff, P. Wilfred, XRD and EDXRF Analysis of Anatase Nano-TiO2 Synthesized from Mineral Precursors, Advanced Materials Research, Vol. 620, 2013, pp. 179-185. F. Hanini, A. Bouabellou, Y. Bouachiba, F. Kermiche, A. Taabouche, M. Hemissi, and D. Lakhdari, Structural, optical and electrical properties of TiO2 thin films synthesized by sol–gel technique, IOSRJEN, Vol. 3, Issue 11, 2013, pp. 21-28. S. S. Roy, A. H Bhuiyan, J. Podder, Optical and Electrical Properties of Copper Oxide Thin Films Synthesized by Spray Pyrolysis Technique, Sensors & Transducers, Vol. 191, Issue 8, August 2015, pp. 21-27. J. Tauc, Optical properties and electronic structure of amorphous Ge and Si, Materials Research Bulletin, 3, 1968, pp. 37-46. M. Stamate, I. Vascan, I. Lazar, G. Lazar, I. Caraman, M. Caraman, Optical and Surface Properties of TiO2 Thin Films Deposited by DC Magnetron Sputtering Method, Journal of Optoelectronics and Advanced Materials, Vol. 7, No. 2, April 2005, pp. 771-774. G. Wu, M. Takahashi, K. Arai, H. Inaba and K. Minoshima, Extremely high-accuracy correction of air refractive index using two-colour optical frequency combs, Scientific Reports, 3, 2013, 1894. M. A. Malati and W. K. Wong, Doping TiO2 for solar energy applications, Surface Technology, 22, 1984, pp. 305-322. X. Zhao, Q. Zhao, J. Yu, B. Liu, Development of multifunctional photoactive self-cleaning glasses, Journal of Non-Crystalline Solids, Vol. 354, 2008, pp.1424–1430. K. P. Biju, M. K. Jain, Effect of crystallization on humidity sensing properties of so l– gel derived nanocrystalline TiO2 thin films, Thin Solid Films, Vol. 516, 2008, pp. 2175–2180. J. Nowotny, T. Bak and T. Burg, “Electrical properties of polycrystalline TiO2. Prolonged Oxidation Kinetics, Journal of Ionics, Vol.13, 2007, pp. 79-82. Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 66-73 ________________ 2015 Copyright ©, International Frequency Sensor Association (IFSA) Publishing, S. L. All rights reserved. (http://www.sensorsportal.com) 73 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 74-80 Sensors & Transducers © 2015 by IFSA Publishing, S. L. http://www.sensorsportal.com Non-destructive Testing of Wood Defects Based on Discriminant Analysis Method Wenshu LIN and Jinzhuo WU College of Engineering and Technology, Northeast Forestry University, Harbin 150040, China Tel.: (86) 0451-82191853, fax: (86) 0451-82190631 E-mail: [email protected], [email protected] Received: 16 July 2014 /Accepted: 31 August 2014 /Published: 30 September 2015 Abstract: The defects of wood samples were tested by the technique of stress wave and ultrasonic technology, and the testing results were comparatively analyzed by using the Fisher discriminant analysis in the statistic software of SPSS. The differences of defect detection sensitivity and accuracy for stress wave and ultrasonic under different wood properties and defects were concluded. Therefore, in practical applications, according to different situations the corresponding wood non-destructive testing method should be used, or the two detection methods are applied at the same time in order to compensate for its shortcomings with each other to improve the ability to distinguish the timber defects. The results can provide a reference for further improvement of the reliability of timber defects detection. Copyright © 2015 IFSA Publishing, S. L. Keywords: Wood sampling, Defect, Stress wave, Ultrasonic, Discriminant analysis method. 1. Introduction Wood non-destructive testing technology is a new and comprehensive detection method, which can detect and evaluate the physical properties of wood, growth characteristics, mechanical properties, and wood defects without destroying the final value of the wood. Among the non-destructive testing method, any method has its own advantages and disadvantages, and the effects on wood defect detection are different. In addition, wood is a natural material and its properties are very complex. Under different wood properties (water content and density) and different defect types, shape, size and position of the distribution conditions, the accuracy and precision of wood non-destructive testing method have some differences, so the reliability of testing was different. The timber defects testing reliability problems can be seen as wood defects detection rate and misjudgment rate. Discriminant analysis is an important statistical method. It is a multivariate statistical analysis 74 method, and is used to determine the ownership of an objects based on its various characteristic values under the defined classification conditions. The basic principle is based on a certain criterion to create one or more discriminant functions, with large amount of data of the study object to determine the discriminant function coefficients to be determined and calculated discriminant index. Then what kind of a sample belonged to can be determined [1]. Therefore, the sample’s misjudgment rate can be obtained by using discriminant analysis. In this study, the Fisher discriminant analysis was used to distinguish the different classes. Since stress wave and ultrasonic can detect the wood detection accurately, so many researchers focus on studying these two waves, and a series of results have been achieved [2-8]. However, the sensitivity of the defect and defect detection reliability are different for two waves when the timber properties and the impact of external conditions are different. Therefore, each testing method has its own advantages and disadvantages. For the better application of stress http://www.sensorsportal.com/HTML/DIGEST/P_2726.htm Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 74-80 wave and ultrasonic flaw detection on wood, these two detection methods must do in-depth research. At present, for both waves detection in wood properties comparative study is relatively rare, while the detection of internal defects in wood comparative studies have not been conducted. the wood, and two sensors was used to induct the change of wave, and the propagation data was shown on the screen of the laptop. Based on the propagation time and velocity, the wood properties (such as modulus of elasticity and defects) were determined. Wood sample Sensor 2 Sensor 1 2. Materials and Methods Small hammer 2.1. Wood Samples Elm is one of the popular tree species in the northeast of China; therefore the elm (Ulmus rubra) wood samples were used in this study. The elms were obtained from Fangzheng Forestry Bureau located in Heilongjiang Province, and then were delivered to the Wood Manufacturing Factory of Northeast Forestry University and processed to wood samples. The specifications for wood samples were 300 mm×50 mm (length × width × height). The moisture content of the wood samples is around 9 % due to the long time stored after being processed. The wood samples include intact samples and samples with defects, and the number for each type of wood samples was 16. Each group of wood samples was labeled before processing and testing in order to deal with the tested data easily. Wood defects include natural defects (such as wood cracks and decays) and artificial defects (holes and cracks) in this study. Artificial defects were conducive to quantify the extent of the defect. The detail processing methods are as follows: From the intact wood samples, the diameter of hole with 10 mm and 20 mm were produced, and the number of holes ranges from 1 to 3, and the quantities of each type of holes were 16, and the locations of holes along the axis. In addition, a crack was sawn on the intact wood samples, and the depth of crack was 3 mm, and the shape of crack was arc. The locations of the holes and cracks are illustrated in Fig. 1. Fig. 1. Locations of holes and cracks. 2.2. Testing Method 2.2.1. Stress Wave Testing A) Testing Method. The impact stress wave method was used in the study. Two probes were nailed on both ends of the wood sample, and sensors were hanged on the probes (Fig. 2). When hitting the sensor by using a small hammer, stress wave was generated in the interior of Laptop Stress wave testing instrument Fig. 2. Principle of testing wood defects using stress wave technology. B). Testing system. The stress wave detector ARBOTOM imported from Germany was used in the experiment. The ARBOTOM detector is mainly used to measure the internal situation of wood. The propagation velocity of stress wave and wood density is highly correlated, so the ARBOTOM can be used to collect the information of wood internal defects. Before testing, some parameter values should be input, such as the number of sensors, the distance of all sensors, the unit of distance, the height of sensors above ground, the PC port, filtering mode, the name of tree species, and so on. 2.2.2. Ultrasonic Testing A) Testing Method. In this paper, the so-called penetration method was used in the test. On one end of the wood sample, an emission transducer and ultrasonic pulse wave are localized and on the other end a receiving transducer is set up, so that the ultrasound can travel through the wood sample. The received ultrasonic signal is converted into electrical signal and amplified by an amplifier. By the simulated digital converter, the signal is converted into digital information and stored in a computer. After proper processing using specific procedures, we can obtain the ultrasonic parameters such as propagation time, velocity, and amplitude. Based on the propagation parameters, we can judge whether there are inner defects in the wood sample. The principle of testing wood defects using ultrasound is illustrated in Fig. 3. B) Testing System. The testing instrument used in the paper is RSMSY5 ultrasonic tester made by Wuhan Institute of Rock and Soil Mechanics, China. The data collection software in the tester can adjust the testing parameters, signal collection mode, and storage and open functions of the signal. Meanwhile, the wave 75 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 74-80 and frequency forms can be obtained together with the transit parameters including propagation time, velocity, and elastic modulus. These wave forms can be further analyzed and processed. Wood sample Receiving Emission Ultrasound Ultrasonic tester Fig. 3. Principle of testing wood defects using ultrasonic technology. the correct number of specimensof discriminant × 100% total number of specimensin each group (1) thecorrectnumberof specimens tobedetermined ×100% totalnumberof specimens tobedetermined in eachgroup (2) CRBS = CRD= According to Eqs. (1) and (2), the number of specimens correctly classified was 11, 9, 8 and 10 for the four groups of stress wave tested specimens, and the corresponding correct rate of back substitution was 100 %, 81.82 %, 72.73 %, and 90.91 %, respectively. The correct rate of disriminant for the specimens to be determined was 12 / 20 ×100 % = 60 % . The misjudgment rate for the four groups by using the stress wave testing is presented in Table 1. Table 1. The misjudgment rate for the number of holes of specimens by stress wave and ultrasonic testing. 3. Results and Analysis Groups 3.1. Effects of the Number of Holes on the Parameters and Dynamic Elastic Modulus for the Two Waves The specimens were divided into four groups based on the number of holes with 20 mm diameter and intact samples. Each group contains 11 specimens. A total of 20 specimens will be judged with 5 specimens in each group. The specimen’s moisture content, density, wave velocity and elastic modulus were used as the data index in the discriminant analysis. In order to facilitate the calculation, Let X1 represent moisture, X2 represent the density, X3 represent the velocity of propagation, and X4 represent the elastic modulus, and the discriminant analysis method in the multivariate statistical analysis software SPSS was used for the comparisons. 3.1.1. The Discriminant Analysis for Stress Wave Testing Based on the Classification Function Coefficients (Fisher's Discriminant Function Coefficients) table, four groups of linear discriminant functions were derived as follow: y1=-12.050-0.0768X1-38.286X244.329X3+85.322X4, y2=-2.416-0.02444X1+19.061X2+22.348X338.041X4, y3=-3.321+1.604X1+12.231X2+14.742X329.825X4, y4=-5.092-0.577X1+1.463X2-0.134X3-5.092X4. The following functions were used to calculate the correct rate of back substitution and disriminant, where CRBS represents the correct rate of back substitution, and CRD represents the correct rate of disriminant. 76 1 2 3 4 Stress wave testing Ultrasonic testing 1 2 3 4 1 2 3 4 20 % 0 0 - 20 % 0 0 20 % 20 % 0 20 % 0 20 % 0 60 % 0 0 20 % - 20 % 0 20 % 20 % 0 0 40 % - Based on the correct rate of back substitution, we can see that the correct rate of back substitution increased as the number of holes increased. There are misjudgment situations for these two groups: intact and one-hole specimens, two-hole and three-hole specimens, which illustrates that stress wave testing cannot accurately judge the specimens with fewer inner defects. However, the stress wave is able to distinguish the specimens with 3 holes and intact specimens. 3.1.2. The Discriminant Analysis for Ultrasonic Testing Based on the Classification Function Coefficients (Fisher's Discriminant Function Coefficients) table, four groups of linear discriminant functions were derived as follow: y1=-9.177-0.415X1-22.251X2-27.553X3+52.680X4, y2=-2.259+0.509X1+7.700X2+11.231X3-17.719X4, y3=-2.881+1.690X1+10.061X2+13.187X324.747X4, y4=-3.903-1.386X1+2.638X2+1.478X3-7.250X4. Based on Eqs. (1) and (2), the number of specimens correctly classified was 11, 9, 9 and 10 for the four groups of ultrasonic tested specimens, and the corresponding correct rate of back substitution was 100 %, 81.82 %, 81.82 %, and 90.91 %, respectively. The correct rate of disriminant or the specimens to be determined was 13 / 20 × 100% = 65% . The misjudgment rate for the number of holes of specimens by using ultrasonic testing is presented in Table 1. Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 74-80 From the correct rate of back substitution, we can see that the correct rate of back substitution increased as the number of holes increased. There are misjudgment situations for these two groups: intact and one-hole specimens, two-hole and threehole specimens. 3.1.3. Comparative Analysis Between Stress Wave and Ultrasonic Testing When the number of holes is fewer, the correct rate is lower for both waves based on the misjudgment rate for stress wave and ultrasonic detection, but the misjudgment rate is relatively lower by using the ultrasonic testing method. From the correct rate of specimens to be determined, the probability of detection using ultrasonic is greater than stress waves, which illustrates that ultrasonic is more sensitive than stress wave when testing the specimens with holes. The discrimination results figures (Fig. 4) also indicate that the grouping is better when using ultrasonic. Canonical Discriminant Functions 4 3 Group 2 1 Ungrouped Cases 3 1 4 -1 4 -6 -4 -2 3.2.1. The Discriminant Analysis for Stress Wave Testing Based on the Classification Function Coefficients (Fisher's Discriminant Function Coefficients) table, three groups of linear discriminant functions were derived as follow: y1=-3.719+1.565X1-13.435X2-10.729X3+22.296X4, y2=-2.500-2.245X1+7.512X2+3.218X3-5.281X4, y3=-4.333+1.275X1+0.721X2+2.962X3-8.349X4. Based on Eqs (1) and (2), the number of specimens correctly classified was 10, 10 and 11 for the three groups using stress wave, and the corresponding correct rate of back substitution was 90.91 %, 90.91 %, and 100 %, respectively. The correct rate of disriminant for the specimens to be determined was 8/15×100%=53.3 %. The misjudgment rate for the specimens with different hole size by using the stress wave testing is shown in Table 2. 0 2 Table 2. The misjudgment rate for the specimens with different hole size by stress wave and ultrasonic testing. 3 -2 -3 -4 -8 The specimens were divided into three groups based on the sizes of the holes. Each group contains 11 specimens. A total of 15 specimens will be judged with 5 specimens in each group. Group Centroids 2 0 3.2. Comparison of the Effects of the Size of Holes on the Two Waves 2 Groups 1 1 2 3 4 Stress wave testing 1 2 3 20 % 20 % 20 % 40 % 20 % 20 % - Ultrasonic testing 1 2 3 20 % 20 % 20 % 40 % 20 % 0 - (a) Stress wave testing Canonical Discriminant Functions 3 2 Group 2 1 3 0 Group Centroids 1 Ungrouped Cases -1 4 4 -2 3 -3 -4 -5 -8 It is noted that the smaller the size of holes, the lower the correct rate of back substitution based on the correct rate of back substitution. The judgment results are not very good for the specimens with hole size of 10 mm. The misjudgment rate was improved when the hole size increased to 20 mm, however there is still misjudgment situation. Based on the correct judgment rate for the specimens to be determined, the probability of correct classification was also low by using stress wave testing. 2 1 -6 -4 -2 0 2 4 (b) Ultrasonic testing Fig. 4. Discrimination results for specimens with different number of holes by stress wave and ultrasonic testing. 3.2.2. The Discriminant Analysis for Ultrasonic Testing Based on the Classification Function Coefficients (Fisher's Discriminant Function Coefficients) table, three groups of linear discriminant functions were derived as follow: y1=-3.252+1.391X1-7.108X2-7.752X3+13.875X4, 77 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 74-80 y2=-2.407-2.033X1+8.788X2+4.295X3-7.650X4, y3=-4.043+0.992X1+0.239X2+4.149X3-7.586X4. Based on Eqs. (1) and (2), the number of specimens correctly classified was 8, 11 and 11 for the three groups using ultrasonic testing, and the corresponding correct rate of back substitution was 72.73 %, 100 % and 100 %, respectively. The correct rate of disriminant for the specimens to be determined was 9/15×100 %=60 %. The misjudgment rate for the specimens with different hole size by using ultrasonic testing is shown in Table 2. The judgment results are not very good for the specimens with hole size of 10 mm. Similar to the results using stress wave testing, the misjudgment was improved when the hole size was changed to 20 mm. The correct judgment rate of for the specimens (20 mm) to be determined was also higher by using ultrasonic testing. number of specimens correctly classified was 9, 9, 7 and 8 for the four groups with different types of defects, and the corresponding correct rate of back substitution was 81.82 %, 81.82 %, 63.6 % and 80 %, respectively. The correct rate of disriminant for the specimens to be determined was 13/20×100 %=65 %. The misjudgment rate for the types of defects of specimens by using the stress wave testing is shown in Table 3. Canonical Discriminant Functions 4 3 Group 2 2 1 0 Ungrouped Cases 3 1 -1 3.2.3. Comparative Analysis Between Stress Wave and Ultrasonic Testing From the misjudgment rate for stress wave and ultrasonic detection, it is noted that when the size of the hole is smaller, the correct rate is lower for both waves. From the correct rate of specimens to be determined, the probability of detection using ultrasonic is greater than stress waves, especially for the specimens’ holes with diameter of 20 mm. The discrimination results figures (Fig. 5) also indicate that the grouping is better when ultrasonic testing is applied. 3.3. Comparison of the Effects of the Defect Types on the Two Waves Wood defects include decay, hole, and crack. In the study, all the specimens were divided into four groups based on defect types and intact wood. A total of 11 specimens have defects and another 10 specimens are intact wood. Twenty specimens will be judged, with 5 specimens in each group. The moisture content, density, wave velocity, and elastic modulus of the specimens were used as data index in the discriminant analysis method. The stepwise discriminant method in the multivariate statistical analysis software is used to discriminate the types of defects. After stepwise discriminant, wood moisture content was removed from the input data, which illustrated that the discriminant ability by using wood moisture content was less significant. That is to say, the impact of wood moisture content on the judgment of timber defects is smaller compared with other three data index. After discriminant analysis, the 78 3 2 -2 -3 -6 1 -4 -2 0 2 4 (a) Stress wave testing Canonical Discriminant Functions 6 4 Group 2 Group Centroids 2 0 Ungrouped Cases 3 1 3 -2 -4 -4 2 1 -2 0 2 4 6 (b) Ultrasonic testing Fig. 5. Discrimination results of size of holes by stress wave and ultrasonic testing. Table 3. The misjudgment rate for the types of defects of specimens by stress wave and ultrasonic testing. Groups 3.3.1. The Discriminant Analysis for Stress Wave Testing Group Centroids 1 2 3 4 Stress wave testing Ultrasonic testing 1 2 3 4 1 2 3 4 0 0 0 0 0 20 % 0 40 % 0 20 % - 20 % 20 % 0 40 % 0 0 40 % 0 40 % 20 % 0 0 20 % 0 - Based on the misjudgment rate for different defect types using stress wave detection, it is noted that it is easy to misjudge between intact specimens with speciments with crack, or between speciments with decay and with hole. For the former case, it is Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 74-80 possible that the stress wave propagation may not pass the crack part, therefore it will be the same as the intact wood. The latter case showed that the sensitivity is high for both decayed and holed wood when using stress wave testing. However, the stress wave can judge whether there is any defect in the wood specimens or not. Canonical Discriminant Functions 4 3 2 1 After stepwise discriminant, the wood moisture content data index was also removed. It means that the impact of wood moisture content on the judgment of timber defects is smaller compared with other three data index. After the analysis, the number of specimens correctly classified was 10, 10, 9 and 8 for the four groups with different types of defects, and the corresponding correct rate of back substitution was 90.91 %, 90.91 %, 81.82 % and 80 %, respectively. The correct rate of disriminant for the specimens to be determined was 13/20×100 %=65 %. The misjudgment rate for the types of defects of specimens by using the ultrasonic testing is shown in Table 3. From the misjudgment rate for different type of defects using ultrasonic detection, it is easy to misjudge between specimens with decay and holes, which illustrates that the sensitivity is high for the decayed and holed wood tested by ultrasonic. Similar to stress wave testing, ultrasonic can also judge whether there is any defect in the wood specimens or not. 3.3.3. Comparative Analysis Between Stress Wave and Ultrasonic Testing From the correct rate of back substitution and the corresponding discrimination results figures (Fig. 6) for stress wave and ultrasonic detection, it is noted that the testing reliability is better when ultrasonic is used to detect wood decay and holes. In addition, from the misjudgment rate, there are misjudgments between wood decay and holes for both waves; however the misjudgment rate is smaller for ultrasonic testing. It is not quite different for both waves to detect wood cracks, which may be due to the fact that wave propagation didn’t pass the wood cracks. 4. Conclusions The testing results of elm wood specimens by using stress wave and ultrasonic detection were analyzed by using discriminant analysis method in SPSS statistical software, and the following conclusions can be drawn from the study: Group Centroids 2 Ungrouped Cases 0 3 -1 3.3.2. The Discriminant Analysis for Ultrasonic Testing Group 4 1 4 -2 3 -3 -4 -6 2 1 -4 -2 0 2 4 6 (a) Stress wave testing Canonical Discriminant Functions 6 4 Group 4 2 2 3 0 Group Centroids 1 -2 Ungrouped Cases -4 4 -6 3 -8 -10 -6 2 1 -4 -2 0 2 4 (b) Ultrasonic testing Fig. 6. Discrimination results of type of defects by stress wave and ultrasonic testing. 1) The correct judgment rates were 60 % and 65 % for stress wave and ultrasonic testing decayed wood specimens, respectively. With the increase of decay, the probability of detection using ultrasonic increased more significantly compared to stress waves, which indicated that ultrasonic is more sensitive to severe decay than stress wave when testing the holed specimens. 2) Ultrasonic showed very good sensitivity of detection on larger holes. In the experiment, the correct rates of judgment were 53.3 % and 60 %, respectively, for stress wave and ultrasonic testing wood specimens with holes. The testing reliability for both waves was low for the specimens with 10 mm holes, while the reliability was greatly improved for ultrasonic testing when the diameter of holes increased to 20 mm. 3) For the identification of defect types, both stress wave and ultrasonic were able to distinguish whether there are defects within the wood specimens or not. However, it is not very significant for the distinction between holes and decay, especially for stress wave detection. The misjudgment was small when either stress wave or ultrasonic was used to detect the specimens with severe decay and holes, and the discriminant accuracy rate can reach 80 % for both waves, especially for ultrasonic. The grouping 79 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 74-80 situation was not very good for crack detection discriminant, which illustrated that the surface crack detection reliability was not very good both waves. Due to the limited number of wood samples tested, there may be some derivations in the statistical analysis. Therefore, it is recommended that multiple tree species and different wood defects can be considered in the future study. In addition, there may be errors during in the experiment, including the measurement error and human error. Because the impact force is not the same when using the mall hammer to hit the sensors, so there is a great influence on the accuracy for the stress wave detection. For ultrasonic testing, the quality of gap coupling between wood and ultrasonic probe will have impact on the testing results. In order to reduce the error, the instrument itself should be able to transmit and receive signal during the stress wave testing process, which will reduce the influence of human factors. In addition, if coupler is not needed, its impact will be reduced and there will be no pollution on wood products, which will enlarge the application of this technology in non-destructive testing of wood products. Acknowledgements This study was financially supported by the Fundamental Research Funds for the Central Universities (DL13BB19), and the Excellent Research Projects Scholars (415003). for Returned References [1]. W. Xue, SPSS Statistical Analysis Methods and Applications, Beijing, Electronic Industry Press, 2009, pp. 327-349. [2]. H. Wang, X. C. Yang, K. H. Xu, Current situation of research on the non-destructive testing technique for wood defects, Forestry Science and Technology, Vol. 27, No. 3, 2002, pp. 35-38. [3]. Robert J. R., James C. W., Anton T., Stress Wave Nondestructive Evaluation of Wet wood, Forest Products Journal, Issue 7, No. 8, 1994, pp. 79-83. [4]. C. R. Raini, G. W. Francis, M. G. Thomas, et al., Stress-wave Analysis of Douglas-fir Logs for Veneer Properties, Forest Products Journal, Issue 50, No. 4, 2000, pp. 49-52. [5]. R. J. Ross, R. F. Pellerin, NDE of Wood-based composites with longitudinal Stress Wave, Forest Products Journal, Vol. 38, Issue 5, 1988, pp. 39-45. [6]. J. I. Doulop, Testing of poles by acoustic resonance, Wood Science and Technology, Vol. 17, Issue 1, 1983, pp. 31-38. [7]. Olivito R. S., Ultrasonic measurements in wood, Materials Evaluation, Vol. 54 No. 4, 1996, pp. 514-517. [8]. J. Wang, J. M. Biernacki, F. Lam, Nondestructive evaluation of veneer quality using acoustic wave measurements, Wood Science and Technology, Vol. 34, Issue 6, 2001, pp. 505-516. ___________________ 2015 Copyright ©, International Frequency Sensor Association (IFSA) Publishing, S. L. All rights reserved. (http://www.sensorsportal.com) 80 Overseas Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 81-89 Sensors & Transducers © 2015 by IFSA Publishing, S. L. http://www.sensorsportal.com Research on Electronic Transformer Data Synchronization Based on Interpolation methods and Their Error Analysis * 1, 2 Pang Fubin, 1 Yuan Yubo, 1 Bo Qiangsheng and 1 Ji Jianfei 1 2 Jiangsu Electric Power Company Research Institute, Nanjing 211103, China School of Electrical Engineering, Southeast University, Nanjing 210096, China * E-mail: [email protected] Received: 1 August 2015 /Accepted: 31 August 2015 /Published: 30 September 2015 Abstract: In this paper the origin problem of data synchronization is analyzed first, and then three common interpolation methods are introduced to solve the problem. Allowing for the most general situation, the paper divides the interpolation error into harmonic and transient interpolation error components, and the error expression of each method is derived and analyzed. Besides, the interpolation errors of linear, quadratic and cubic methods are computed at different sampling rates, harmonic orders and transient components. Further, the interpolation accuracy and calculation amount of each method are compared. The research results provide theoretical guidance for selecting the interpolation method in the data synchronization application of electronic transformer. Copyright © 2015 IFSA Publishing, S. L. Keywords: Electronic transformer, Data synchronization, Interpolation method, Interpolation error. 1. Introduction In relay protection and measurement control of power system, the synchronization of sampling data from electronic transformer is the key precondition and guarantee for protection device to measure and operate accurately. The output signal of traditional transformer is analog, which is directly sampled by the secondary electrical equipment and all channels are mainly synchronous [1]. With the continuous expanding promotion of digital substation technology, the primary and secondary equipment gradually develop toward small, intelligent and high steady. The electronic transformer, which has small volume, strong anti-saturation, excellent insulating property and vast dynamic range, meets the requirement of electrical engineering development and has been widely used in intelligent substation. http://www.sensorsportal.com/HTML/DIGEST/P_2727.htm When using electronic transformer, the primary electrical quantity is connected to the merging unit via the data sampling device, and then sent to the protecting and controlling devices in bay level. Without unified synchronizing mechanism to each sampling link of electronic transformer, the sampling data of each channel is nonsynchronous, thus resulting in the synchronization problem in the substation [2]. The IEC60044 standard has specified two methods for data synchronization: impulsive synchronization and interpolation methods [3-5]. The frontier method requires that each merging unit includes an external synchronous port for frequency doubling, which increases the cost and encounters the difficulties of real-time signal receiving and sending, thus the system is complicated and costs much. In engineering application, the interpolation method, 81 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 81-89 which maintains synchronization by computing and reduces the cost effectively, is gradually widely adopted in power system to achieve data synchronization. The interpolation method abandons the requirement of synchronization of each sampling channel, and computes the sampling value of all channels at the same time with the prior knowledge of the time delay from data sampling to arriving at the merging unit. The mechanism of interpolation method is relative simple, and the cost of the system is very low by software computing to achieve electronic transformer data synchronization. Currently, there have been many researches on linear, quadratic and cubic interpolation methods and the analysis on the interpolating error of each method [6-14]. But these researches mainly focus on the interpolation error accuracy under certain application background, while the interpolation error varies greatly from each other and the calculation amounts are also different. In this paper, the most general composition form of the signal is considered, and the interpolation error expressions of linear, quadratic and cubic interpolation methods are deduced. Besides, the influencing factors of each method are analyzed and simulated, and the interpolation accuracies and amounts of all methods are compared. 2. Theoretical Analysis of Interpolation Methods Fig. 1. Diagram of interpolation for data synchronization. 2.2. The Common Interpolation Methods and their Interpolation Errors Currently, there are three interpolation methods: linear, quadratic and cubic interpolation methods. The next paragraph will introduce the interpolating principle and error of each method taking the example of sampled signal S1 (t ) . a) Linear interpolation method. The linear interpolation method is the most simple interpolation method which calculates the interpolating value at time t0 with the prior knowledge 2.1. Introduction of the Problem ( t12 , S1 (t12 ) ) Without a centralized and accurate time standard to synchronize all the nodes in the substation, the sampling time of all channels are usually random. Therefore, the sampling data from all channels to the merging unit are generally asynchronous. As it is shown in Fig. 1, the S1 (t ) , S2 (t ) are sampled signals from two independent channels. There is reset signal in the merging unit and when the number of sampling frames reaches to a certain number, the number will be reset to zero and begin to count in a new circle. Thus, combined with the internal crystal oscillator, the merging unit marks the arrival time of S1 (t ) , S2 (t ) . Assume that when the reset signal sets the count number to zero, t11 , t12 , t13 , t21 , t22 , t23 are the arrival time of signal S1 (t ) , S2 (t ) from the sampling link to the merging unit, respectively. The interpolation method is to compute the sampling value of both channels at the time of interpolation time t0 , 2t0 , 3t0 according to the sampling value at t0 , the y1 (t0 ) , the other times. At the sampling time interpolation value of S1 (t ) is interpolation error can be calculated as follows: ε = y1 (t0 ) − S1 (t0 ) 82 of y1_l (t ) = sampling ( t11 , S1 (t11 ) ) , [15]: t − t12 t − t11 S1 (t11 ) + S1 (t12 ) t11 − t12 t12 − t11 (2) The principle of liner interpolation is to consider the interpolation time t0 to be one point on the line composed by ( t11 , S1 (t11 ) ) and ( t12 , S1 (t12 ) ) . According to the Lagrange Interpolation Error Formula, the interpolation error of linear interpolation is as follows [16]: Rl (t ) = S1 (t ) − y1_l (t ) = 1 '' S1 (ξ )(t − t11 )(t − t12 ) , 2 (3) where ξ is the time in interval [t11 , t12 ] , S1'' (ξ ) is the second derivative of signal S1 (t ) at the time of ξ . Consider the most general situation and assume that S (t ) is composed by direct current, steady and transient components, whose form can be expressed as: ∞ (1) points S1 (t ) = I 0 + I n sin(nωt + ϕn ) +IT e n =1 − t τ , (4) Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 81-89 where I 0 is the direct component, ω is the angular frequency of fundamental wave and ω = 2πf = 100π ; I n and ϕn are the amplitude and 2 proportional to N , RT_l (t ) is proportional to IT and inversely proportional to N τ . b) Quadratic interpolation method. The principle of quadratic interpolation method 2 2 initial phase of the nth harmonic wave and n = 1 corresponds to those of the fundamental wave; IT is the initial value of the transient wave and τ is the time constant. It can be derived form Equation (4) can be expressed as: that the second derivative form of S1 (t ) is： time interval, the interpolation quadratic interpolation method is: S1'' (t ) ∞ = − n ω I n sin(nωt + ϕn ) + 2 2 n =1 IT e − ( t13 , S1 (t13 ) ) t τ (5) y1_ q (t ) = τ2 t − IT e τ 1 ∞ 2 2 Rl (t ) = − n ω I n sin(nωt + ϕn ) + 2 2 n =1 τ (t − t11 )(t − t12 ) 2 ∞ 2 ω n In I ≤ n =1 + T2 (t − t11 )(t − t12 ) 2 2τ t= t11 + t12 . Assume that there are N 2 sampling points in each fundamental wave period, then t12 − t11 = 1 , and Equation (6) is converted 50 N to: n =1 2 N + 5 ×10-5 IT N 2τ 2 time interval of adjacent sampling points. The quadratic interpolation method is to consider the interpolation time t0 to be one point on the parabola ( t11 , S1 (t11 ) ) , ( t12 , S1 (t12 ) ) and it is also called parabolic interpolation, whose interpolation error can be expressed as: Rq (t ) = 1 ''' S1 (ξ )(t − t11 )(t − t12 )(t − t13 ) 3! 3∞ 3 ω n In IT n =1 ≤ + 3 (t − t11 )(t − t12 )(t − t13 ) 6 6τ (t − t11 )(t − t12 )(t − t13 ) = μ ( μ − T )( μ − 2T ) = f ( μ ) (10) = RH_l (t ) + RT_l (t ) The derivation of Equation (10) indicates that it (7) gets the maximum value when t = t11 + ∞ where RH_l (t ) = 4.935 n 2 I n n =1 2 , RT_l (t ) = 5 × 10-5 IT N N 2τ 2 are the maximum interpolation errors of harmonic and transient components of linear interpolation method. It can be seen from Equation (7) that ∞ RH_l (t ) is proportional to and By introducing t = t11 + μ , the items in Equation (9) can be obtained as follows: ∞ = T = t13 − t12 = t12 − t11 = 1/ ( 50 N ) is the where (9) 2 ∞ 2 ω n In IT 1 n =1 Rl (t ) ≤ + 2 10000 N 2 2τ 2 4.935 n 2 I n of (8) ( t13 , S1 (t13 ) ) (17), Obviously, Equation (6) gets the maximum value expression t − t11 t − t12 S1 (t13 ), t13 − t11 t13 − t12 composed by (6) when are the sampling points of the same t − t12 t − t13 t − t11 t − t13 S1 (t11 ) + S1 (t12 ) t11 − t12 t11 − t13 t12 − t11 t12 − t13 + By introducing Equation (5) into Equation (3), the interpolation error of linear interpolation method can be obtained as follows: ( t11 , S1 (t11 ) ) , ( t12 , S1 (t12 ) ) , n2 I n and inversely 3± 3 T, 3 then the maximum value of Equation (9) is: ∞ Rq (t ) ≤ 15.192 n3 I n n =1 3 + 5.132 ×10-7 IT N = RH_q (t ) + RT_q (t ), N 3τ 3 (11) n =1 83 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 81-89 ∞ where RH_q (t ) = The unknown variables in the matrix above are as follows: 15.192 n3 I n n =1 3 is the maximum N interpolation error of harmonic components of quadratic interpolation method, and it is proportional ∞ n In to 3 3 and inversely proportional to N ; n =1 RT_q (t ) = 5.132 ×10-7 IT is N 3τ 3 the maximum interpolation error of transient components of quadratic interpolation method, which is proportional 3 3 to IT and inversely proportional to N τ . c) Cubic interpolation method. The idea of cubic interpolation method is: when the interpolation condition is satisfied, the interpolation interval is divided into several sections. Apart from the two boundary nodes, all other inside nodes have continuous first and second order derivative. From the viewpoint of geometric meaning, the cubic interpolation method enables the concave convex non-deformation of the curve, which avoids the Runge Phenomenon effectively. The cubic interpolation is the most simple spline interpolation. Take the signal S1 (t ) for example, when there are n sampling points S1 (t11 ) , S1 (t12 ) ,…, S1 (t1( n +1) ) in g = 6 S1 (t12 ) − S1 (t11 ) − S ' (t ) , 1 11 1 T1 T1 Ti μi = Ti + Ti +1 Ti +1 λi = 1 − μi = T i + Ti +1 6 S1 t1i , t1(i +1) − S1 t1(i −1) , t1i gi +1 = Ti + Ti +1 S1 (t1( n +1) ) − S1 (t1n ) 6 ' g n +1 = T S1 (t1( n +1) ) − T1 n ( ) i = 1, 2, n − 1, (14) where S1 t1i , t1(i +1) represents the first divided difference of S1 ( t ) at the point of t1i and t1(i +1) . The interpolation error of cubic interpolation method is: Rc (t ) = S1 (t ) − y1_ c (t ) ∞ 5 S1(4) (t ) 384 ≤ ∞ T4 ∞ = 20.294 n 4 I n n =1 4 N + 2.08 ×10−9 IT N 4τ 4 = RH_c (t ) + RT_c (t ), (15) interval [t11 , t1( n +1) ] , and the sampling time for each point is t11 < t12 < < t1( n +1) , respectively. The sampling intervals are T1 , T2 ,…, Tn , then the interpolation expression of S1 (t ) in the ith interval is (t ( 1 i +1) −t 6Ti ) n =1 is the maximum N4 ∞ interpolation which is proportional to 3 Mi + ( t − t1i ) 6Ti n4 I n and n =1 3 M i +1 + inversely M i 2 t1( i +1) − t M t − t1i + S1 (t1( i +1) ) − i +1 Ti 2 S1 (t1i ) − 6 Ti T 6 Ti i t ∈ t1i , t1(i +1) , i = 1, 2, n From the Equation above, it can be obtained that the solution of cubic interpolation requires the value of n + 1 unknown variables of M1 , M 2 ,…, M n +1 , which are determined by the matrix as follows: 2 1 M1 g1 μ 2 M g λ1 1 2 2 = μn −1 2 λn −1 1 2 M n +1 g n +1 RT_c (t ) = proportional 2.08 ×10−9 IT N 4τ 4 is to N4 ; the maximum interpolation error of transient components of cubic interpolation which is proportional to (12) 84 where RH_c (t ) = interpolation error of harmonic components of cubic approximated as y1_ c (t ) : y1_ c (t ) = ∞ 20.294 n 4 I n (13) IT and 4 4 inversely proportional to N τ . 3. Comparison of Interpolation Error and Calculation Amounts of all Interpolation Methods Aforementioned the principles and errors of different interpolation methods have been introduced and calculated. In practical application, due to the difference of the component of signal S1 (t ) and the sampling rates, the interpolation errors vary greatly when adopting different interpolation methods. To Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 81-89 explore the differences between interpolation errors and calculation amount of all interpolation methods, further analysis should be carried out to show which method exhibit the highest interpolation accuracy and cost least calculation amount. Suppose that the interpolation error relationship of harmonic components of three interpolation methods are RH_l (t ) > RH_q (t ) > RH_c (t ) , which indicates that the interpolation error of linear, quadratic and cubic interpolation method declines in orders: ∞ n =1 2 N ∞ > 15.192 n3 I n n =1 3 N ∞ > Assume that the interpolation error relationship of transient components of three interpolation methods is RT_l (t ) > RT_q (t ) > RT_c (t ) , the interpolation error of linear, quadratic and cubic interpolation method declines in orders: 3.1. Interpolation Error of Harmonic Components 4.935 n 2 I n 3.2. Interpolation Error of Transient Components 20.294 n4 I n n =1 4 5 × 10-5 IT N 2τ 2 > (17) where max {a, b} represents the larger number of a , b . Especially, when there is only fundamental 2.08 × 10−9 IT N 4τ 4 Nτ > 1.026 × 10-2 (16) ∞ ∞ 3 4 3.078 n I n 1.336 n I n n =1 N > max ∞ n =1 , , ∞ 2 3 n I n I n n n =1 n =1 N 3τ 3 > (19) Obviously, the interpolation error of each method is related with Nτ . The larger Nτ is, the easier Equation (19) is satisfied. Therefore, the rising of sampling rate N or time constant τ will reduce the interpolation error of high order interpolation method. The solution of Equation (19) is calculated as follows: N From Equation (16), it can be concluded as follows: 1) For given harmonic order and amplitude, the increasing of N will make the Equation (16) easier to be satisfied, which indicates that the rising of sampling rates will reduce the interpolation error of higher order interpolation method and promote the interpolating accuracy. 2) For given sampling rate, the in the increasing of n and I n will make the Equation (16) more difficult to be satisfied, which indicates that the rising of harmonic order and amplitude will enlarge the interpolation error of higher order interpolation method. The solution of Equation (16) can be calculated as follows: 5.132 × 10-7 IT (20) Equation (20) indicates that when the condition Nτ > 1.026 × 10-2 is satisfied, the interpolation error of linear, quadratic and cubic interpolation error declines in orders. 3.3. Comparison of Calculation Amount Due to the principle variations of all interpolation methods, the calculation amount of each method is different from each other. From Equation (2), (8) and (12) it can be concluded that for a interpolating point, the linear interpolation costs five times add operation, two times multiply operation; the quadratic interpolation costs fourteen times add operation, twelve times multiply operation; the cubic interpolation costs twenty-one times add operation, thirty times multiply operation, together with the solution of a interpolation polynomial matrix. Therefore, the calculation amount of cubic interpolation is larger than that of the quadratic interpolation, and the calculation amount of the latter is also larger than that of linear interpolation. 4. Calculating Simulation wave component in S1 ( t ) and n = 1 , Equation (17) 4.1. Calculating Model is simplified as: The conclusion aforementioned has made it clear that the interpolation errors of all interpolation methods are no related with direct component. Assume that the expression of the signal is N > 3.078 (18) From Equation (17) it can be derived that when there is only fundamental wave component in the signal, the interpolation error of linear, quadratic and cubic interpolation methods decline in orders. ∞ S1 (t ) = I n sin(nωt + ϕn ) +IT e − t τ , and the initial n =1 phase of each harmonic wave is 85 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 81-89 ϕn = 0, n = 1, 2, , ∞ , and there are N sampling points in the fundamental wave period. To fully exhibit the interpolation error of each method, two hundred interpolation points are calculated in one fundamental wave period to avoid the insufficience of calculating number. Utilizing MATLAB, the influencing factors of interpolation errors of linear, quadratic and cubic interpolation methods are calculated and the calculation amounts are also compared. (a) Comparison of interpolation errors of three methods Ignore the transient component and assume that IT = 0 , when there is only fundamental wave in the signal, I1 = 1 , I n = 0(n ≠ 1) . The interpolation errors of linear, quadratic and cubic interpolation methods are calculated when N = 20 and N = 80 , and the results are shown in Fig. 2 and Fig. 3. (b) interpolation error of cubic method Fig. 2. Interpolation accuracy of each method when N = 20 . Fig. 3. Interpolation accuracy of each method when N = 80 . When there are 20 sampling points in one fundamental wave period, the maximum interpolation error of linear interpolation method reaches to 1.2 %, while those of quadratic and cubic interpolation methods are 0.2 % and 0.0027 %, which indicates that the interpolation accuracy is higher for high order interpolation method. When there are 80 sampling points in one fundamental wave period, 86 4.2. Influences of Sampling Rates the maximum interpolation error of linear interpolation method has declined to 0.078 %, while those of quadratic and cubic interpolation methods are 0.003 % and 0.00001 %. The proportion decrease of three methods are 15.38, 66.67 and 270 respectively for two different sampling rates, which coincides with the proportion decrease results of 16, 64, and 256 shown in Equation (16). Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 81-89 To further analyze the influences of sampling rates on interpolation accuracy, the interpolation error of three methods are calculated when the sampling rates grows from 500 Hz ~ 10000 Hz, and the results are shown in Fig. 4. From the figure it can be seen that the interpolation accuracy is improved with the increasing of sampling rates; when the sampling rate is given, the interpolation error of linear, quadratic and cubic interpolation error declines in orders, which coincides with the conclusion in Paragraph 3.1 that when there are more than 4 sampling points in the fundamental wave period, the interpolation accuracy of three methods gets promoted in orders. With the increasing of sampling rates, the interpolation error of each interpolation method declines rapidly. Since there is only harmonic component error in the error composition, the interpolation errors of linear, quadratic and cubic interpolation method are inversely proportional to the square, cube and biquadratic of the sampling rates, and the declining slope the of interpolation error curve for three methods becomes steeper in orders. three methods when the harmonic order changes from 1 to 10. When the order is small, the interpolation errors of three methods are also small, and the error of linear, quadratic and cubic interpolation method declines in orders; with the increasing of harmonic order, the interpolation errors of three methods grows rapidly and the Equation (16) is no longer satisfied. The interpolation errors of quadratic, cubic interpolation method are higher than that of linear interpolation method, and there is severe error in interpolating results. Fig. 5. Interpolation accuracy of each method for the third harmonic wave when N = 20 . Fig. 4. Interpolation accuracy of three methods at different sampling rates when N = 20 . 4.3. Influences of Harmonic Orders The conclusions in Paragraph 3.1 indicate that the harmonic order has an effect on the interpolation error. To explore the influences of harmonic order on the interpolation accuracy, the interpolation errors of linear, quadratic and cubic interpolation method are calculated when the harmonic order is three ( I 3 = 1 , I n = 0(n ≠ 3) ), N = 20 , and the results are shown in Fig. (5). When there is only the third harmonic order wave, the maximum interpolation error of linear interpolation is 11 %, while those of quadratic and cubic interpolation are 5.2 % and 0.8 %, respectively. Compared with Fig. 2, when the harmonic order grows higher, the interpolation error of each method becomes larger and the interpolation accuracy has declined. Fig. 6 illustrates the interpolation errors of Fig. 6. Interpolation accuracy of each method for the different harmonic order waves. 4.4. Influences of Transient Components In order to explore the influences of transient components on interpolation accuracy, the interpolation accuracy is calculated when the expression of the signal is S1 (t ) = sin100πt + e − t 0.0008 , N = 20 N = 80 , respectively. The and interpolation error is divided into transient component error and total error(including harmonic component error), and the results are shown in Fig. 7 and Fig. 8. 87 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 81-89 Fig. 7. Interpolation accuracy of each method in two conditions when N = 20 . Fig. 8. Interpolation accuracy of each method in two conditions when N = 20 . When the time constant is 800 us and N = 20 , the multiply of the both Nτ satisfies the Equation (20) and the maximum interpolation error of quadratic interpolation method is 2.2 %, which is slightly smaller than the interpolation error of linear method. The cubic method has the least interpolation error and the maximum error is only 0.3 %. When considering the interpolation error of fundamental wave component, the transient component of the first twenty sampling points plays an important role in the interpolation error, and the error presents geometrical attenuation with the increasing of time and gradually the error of fundamental wave component is the major part. When N = 80 , the interpolation errors of all methods become smaller than those of N = 20 when considering transient component error only. The maximum interpolation errors of linear, quadratic and cubic method are 0.28 %, 0.04 % and 0.001 %, respectively. When considering the total interpolation error, the first twenty sampling 88 points are still the major part of the transient component error, and with the increasing of time the fundamental component wave plays an important role in the interpolation error. View the trend as a whole, the interpolation errors are prominently smaller than those when N = 20 . 4.5. Comparison of the Calculation Amount Although the interpolation errors of linear, quadratic and cubic interpolation method decline in orders when the sampling rate is large, the calculation is an important factor to evaluate whether the method is appropriate for the consideration of hardware realization. The operating time of MATLAB reflects the calculation amount. The operating time of the calculation in Paragraph 3.2 when N = 20 is counted 100 times and the average time of calculation is shown in Table 1. Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 81-89 Table 1. Computational complexity comparison between three methods. Method Interpolation points Operating time(ms) Linear 200 5.748 Quadratic 200 10.34 Cubic 200 78.647 From Table 1 it can be seen that the average operating time for linear method is 5.478 ms, for quadratic method is 10.34 ms, which is two times of the linear method; for cubic method, the operating time is 78.647 ms, which is fourteen times of the linear method. Therefore, the promotion of interpolating accuracy is gained by increasing the calculation amount. 5. Conclusion In this paper, the original problem of the electronic transformer data synchronization is introduced first and the common interpolation methods are given. Based on the composition of the signal, the error expressions of linear, quadratic and cubic methods are deduced and the influences of sampling rate, harmonic order and transient component on the interpolation error are calculated and compared. The conclusions are as follows: 1. By increasing the sampling rate, reducing the harmonic order and amplitude, the harmonic component errors of the three methods can reduced; 2. By reducing the initial amplitude of transient wave and increasing the sampling rate and time constant, the transient component errors of the three methods can reduced; 3. When there are more than four sampling points in the fundamental wave period, the interpolation errors of harmonic component of linear, quadratic and cubic method decline in orders; when Nτ > 1.026 × 10-2 , the interpolation errors of transient component of three methods decline in orders; 4. Although the increasing of sampling rate promotes the interpolation accuracy, the calculation amount of linear, quadratic and cubic method increases in orders. References [1]. Yang, Z. X., Ji. J. F., Ao, Yuan, Y. B., Analysis of ECT synchronization performance based on different interpolation methods, Sensors & Materials, 162, 1, 2014, pp. 251-257. [2]. Cao, T. J., YI, Yin X. G., Zhang Z., Discussion on data synchronization of electronic instrument transformers, in Proceedings of the CSU-EPSA, 19, 2, 2007, pp. 108-113. [3]. Yuan Y. B., Gao L., Intelligent Substation Integration Testing Technology and Application, China Power Publishing, Beijing, 2013. [4]. Dong Y. H., Sun T. J., Xu B. Y. Tongjing, Xu Bingyin, Data synchronization based on cubic spline interpolation for electronic instrument transformers, Electric Power Automation Equipment, 32, 5, 2012, pp. 102-107. [5]. Xiang M. J., Gao H. L., An Y. Q., An adaptive interpolation algorithm to improve data synchronization precision, Automation of Electric Power Systems, 36, 8, 2012, pp. 77-82. [6]. Qiao H. X., Huang S. F., Liu Y., Discussion on data synchronization of electronic current transducer based on quadratic interpolation, Power System Protection and Control, 37, 15, 2009, pp. 48-52. [7]. Cao T. J., Dai C. Z., Two methods of data synchronization in optical fiber differential: improved interpolation and clock relay, in Proceedings of the 2nd International Conference on Electricity Distribution, 2010, pp. 1-6. [8]. Li, Q. Y., Wang, N. C., Yi D. Y., Numerical Analysis, Tsinghua University Publishing, Beijing, 2008. [9]. Hu H. L., Li Q., Lu S. F., Comparison of two electronic transformer error measuring methods, High Voltage Engineering, 37, 12, 2011, pp. 3022-3027. [10]. Li W. Z, Li B. W., Ni C. K., Study on synchronization method for optical differential protection in smart substation, Power System Protection and Control, 40, 16, 2012, pp. 136-140. [11]. Liu K., Zhou Y. Q., Wang H. T., Research and Design on data sampling system of electronic transducer, High Voltage Engineering, 33, 1, 2007, pp. 111-115. [12]. Li C., Yuan Y. B., Luo Q., Research on interfacing technology for digital protection based on ECT/EVT, Power System Technology, 31, 9, 2007, pp. 84-87 . [13]. Ma C., Li L. J., Li C. S., Study of data synchronization of Sagnac fiber optic current transformer, Power System Protection and Control, 40, 8, 2012, pp. 38-44 . [14]. Li G. H., Digital substation networking technologies, Electric Power Automation Equipment, 33, 2, 2013, pp. 142-146. [15]. Guo L., Pan J. M., Lu J. L., Application of interpolation algorithms in smart substation, Electric Power Automation Equipment, 30, 10, 2010, pp. 103-106 . [16]. Liu Y. Q, Gao H. L., Gao W. C., A novel data processing approach to relay protection in digital substation, Automation of Electric Power Systems, 35, 15, 2011, pp. 68-72. [17]. Luo Y., Duan X. Y., Zhang Mingzhi, Research on Time Synchronization for IEC 61850-9-2 Process Bus, Power System Technology, 36, 11, 2012, pp. 229-234. ___________________ 2015 Copyright ©, International Frequency Sensor Association (IFSA) Publishing, S. L. All rights reserved. (http://www.sensorsportal.com) 89 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 90-95 Sensors & Transducers © 2015 by IFSA Publishing, S. L. http://www.sensorsportal.com Performance Characteristics of GaAs/Al0.32Ga0.68As Quantum-Well Lasers HADJAJ Fatima, BELGHACHI Abderrahmane, and HELMAOUI Abderrachid Laboratory physics and semiconductor devices, Bechar University Po. Box No. 417, 08000 Bechar, Algeria Tel.: +213-792-904250; fax: +213-49-220960 E-mail: [email protected] Received: 1 September 2015 /Accepted: 28 September 2015 /Published: 30 September 2015 Abstract: Simulating electrical characteristics of quantum well laser diodes helps understanding their behavior and provides an insight comprehension of the influence of technological parameters, such as number of wells, cavity length and effect of temperature on their performance. In this paper we present a study of electrical characteristics of GaAs/Al0.32Ga0.68As quantum well laser diodes emitting at 0.8 µm. Our results indicate that better output performance and lower threshold current could be obtained for a single quantum well and losses are reduced, and we observe also a gradual and nonlinear decrease in output optical power with the increase of temperature. Copyright © 2015 IFSA Publishing, S. L. Keywords: Multi-quantum well lasers, electrical characteristics, GaAs/AlGaAs. 1. Introduction Quantum-well (QW) semiconductor lasers offer the advantages of low threshold current density and high-power capability with good efficiency, [1]. The application of quantum well structures to semiconductor laser diodes has received considerable attention because of its physical interest and as well as its superior laser characteristics. By controlling the width of the quantum wells one can modify the electron and hole wave function which leads to modification of laser characteristics as well as introduction of new concepts to optical devices [2]. The loss level plays an important role in determining the relative advantages and disadvantages of the SQW and MQW structures. Under low loss conditions the SQW has the advantage, since it has a lower total transparency current density J0 and the total internal loss . At high loss, however, the MQW has the advantage [3]. Because the phenomena of gain 90 saturation can be avoided by increasing the number of QW's although the injected current to achieve this maximum gain also increases by the increase in the number of wells [4]. The saturated gain of the SQW may not be large enough to attain the threshold gain and one must therefore turn to MQW [3]. In this work, we investigate the effect of wells number, cavity length and temperature on the output power, threshold current, and quantum efficiency. Maximum output power and lower threshold current were obtained for the case of a single quantum well (SQW) due to the low losses. Effective change in the output power and threshold current was observed with the variations in temperature and cavity length. 2. Optical Power The most common characteristics of a laser diode is the power vs. current curve (P-I). It plots the drive http://www.sensorsportal.com/HTML/DIGEST/P_2728.htm Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 90-95 = − , (1) where ηi is the internal differential quantum efficiency, αm is mirror losses, αi is the internal optical loss, q is the elementary charge, hv is the photon energy, and Ith is the threshold current of a semiconductor laser [5]. To obtain a laser action in a semiconductor, the medium should be prepared in a form a p-n junction diode with highly degenerate ptype and n-type regions, in this way the inverse is produced in the junction region. This can be achieved by forward biasing the junction .When the junction is forward biased with a voltage that is nearly equal to the energy gap voltage, electrons and holes are injected across the junction in sufficient number to create a population inversion in a narrow zone called the active region. The amount of population inversion, and hence the gain is determined by the current flowing in the laser diode. At low current values the losses offset lasing action. In this case the radiation exists due to spontaneous emission which increases linearly with the drive current. Beyond a critical value of the current (the threshold value), the lasing commences and the output power increases rapidly with increasing current. Moreover, we investigated the effect of the number of quantum wells on the performance of our structure. We varied the number of quantum wells from one to five and investigated the output power versus the injection current for various numbers of wells as shown in Fig. 1. 80 70 1QW 2QW 3QW 4QW 5QW Power (mW) 60 50 40 30 20 10 0 0 20 40 60 80 100 120 Current (mA) Fig. 1. Output power versus injection current for a 60 Å GaAs/Al0.32Ga0.68As SQW and MQW laser 100 µm wide by 300 µm long. With the increase of well number, more injection current is required to obtain the output power. This is because threshold current increases with the increase of the number of wells. In this study with the data we used, we obtained that, as far as the value of the number of wells decreased you can see the increased output power with a less threshold current, for instance a single well, the output power is 31.04 mW and the threshold current is of 8.2 mA for an injection current of 50 mA, while for five wells, the power is 9.84 mW for the same injection current, in this case, the threshold current is about 36.8 mA. As it can be seen in Fig. 2, the threshold of the laser is strongly affected by the laser’s temperature. Typically, laser threshold will increase exponentially with temperature as Ith α exp (T/To), where T is the laser temperature in degrees Kelvin (typically 100 to 400 K). And T0 is the “characteristic temperature” of the laser. 45 40 T = 100 K T = 200 K T = 300 K T = 400 K 35 30 Power (mW) current applied to the laser against the output light intensity. This curve is used to determine the laser’s operating point (drive current at the rated optical power) and threshold current (current at which lasing begins). The P-I characteristics on each facet of the laser device are given by: 25 20 15 10 5 0 0 10 20 30 40 50 60 70 Current (mA) Fig. 2. Output power versus injection current at different temperatures for a 60 Å GaAs/Al0.32Ga0.68As SQW laser 100 µm wide by 300 µm long. This increasing threshold current can be explained by lower gain of the quantum wells at higher temperature, and lower characteristic temperature T0 represents a high increase of threshold current with increasing temperature. The threshold current is predominated by the gain that is needed to compensate the internal losses and the mirror losses, respectively. Thus the temperature dependence of the gain is the main reason for increasing threshold current with increasing temperature [6]. In our structure, the output power has been estimated to be 31.04 mW at room temperature with an injection current of 50 mA. The output power versus the injection current for various cavity length is shown in Fig. 3. With increasing cavity length the output power decreases, it can be explained as follows; the laser with longer cavity length has higher current density and higher electron density, therefore the radiative recombination, and consequently output power is decreased. In the case of 300 µm length cavity, the threshold current is about 8.2 mA, and a power of 31.04 mW for an injection current of 50 mA, while for a length of 500 µm, the current threshold is of the order of 13.2 mA, and a power of about 22.23 mW for the same injection current. 91 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 90-95 of the quasi-Fermi levels into the conduction band or valence band takes place at high current injection, the gain saturates when the sub-band of electrons and holes are completely reversed. 100 90 L L L L L 80 Power (mW) 70 60 = = = = = 100 200 300 400 500 µm µm µm µm µm 50 14 40 12 30 Threshold current (mA) 20 10 0 0 20 40 60 80 100 120 Current (mA) Fig. 3. Output power versus injection current at different cavity length for a 60 Å GaAs/Al0.32Ga0.68As SQW laser 100 µm wide by 300 µm long. 10 = = 240 8 6 4 2 100 150 200 3. Threshold Current 400 33 (2) 2 −1 , where NQW and Γ are the number of QWs and the optical confinement factor of a single QW, respectively, where R = 0.32 is the reflectivity of naturally cleaved mirrors of the laser cavity, L is the cavity length [8], J0 is the transparent current density and G0 is the gain coefficient to describe the quantum ⁄ [9]. The threshold well gain G as = current is calculated by the usual formula: 32 31 30 29 28 27 26 25 0,00 0,01 0,02 0,03 0,04 0,05 0,06 0,07 Cavity length (cm) (3) Fig. 5. Threshold current density versus the cavity length for SQW laser, the transparent current density and modal gain are estimated to be 48 A/cm2 and 204.7 cm-1, respectively. 140 2 Threshold current density (A/cm ) The threshold current density Jth that corresponds to the modal gain value that satisfies the oscillation condition can be obtained from the modal gain-current density plots. Then the threshold current calculations can be performed using desired structure cavity width W, length L, and mirror reflectors [10]. Fig. 4 shows dependence of threshold current Ith on temperature for a cavity length of 300 μm. It was observed from our analysis that Ith increases with the increase of temperature. This is due to the increase of cavity losses with the increase of temperature; hence more current is needed to achieve population inversion to overcome cavity losses. However, The threshold current density decreases with the increase in cavity length as shown in Fig. 5 which is plotted for T = 300 K. We can see from Fig. 6 a significant increase of the threshold current density to the shorter lengths of the cavity, which is due to the high gain saturation at high injection current density for quantum well laser. This gain saturation is due to the step-like shape of the density of state functions and the fact that penetration 92 350 34 Threshold current density (A/cm ) + = 300 Fig. 4. Threshold current versus the temperature for SQW laser. The characteristic temperature is 240 K in the range of 100 K to 400 K. The threshold current density of a MQWs laser with identical QWs is then obtained from the threshold condition, where the gain equals the cavity losses. [7], it is given by the expression: = 250 Temperature (K) 120 1QW 2QW 3QW 4QW 5QW 100 80 60 40 20 0,00 0,01 0,02 0,03 0,04 0,05 0,06 0,07 Cavity length (cm) Fig. 6. Threshold current density versus the cavity length for SQW and MQW laser. Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 90-95 5,0 Inverse external differential quantum efficiency If the gain thus obtained is insufficient to compensate the losses, oscillation threshold is not reached. The maximum modal gain available with NQW is thus N times larger than modal gain with 1QW since each well can now provide its saturation gain, which is equal to that of a SQW laser. We can consequently avoid the saturation effect by increasing the number of QW’s although the injected current to achieve this maximum gain also increases by N times. Owing to this gain saturation effect there exists an optimum number of QW’s for minimizing the threshold current for a given total loss, on the other hand, the increase in the threshold current is substantially shifted to shorter length cavity by increasing the number of wells ( see figure (6)), such that for a structure with a single well and a length of 300 µm, the threshold current density is about 27.47 A/cm2 while for a structure with five wells, it is about 122.62 A/cm2 for the same length. Theoretically, the lower threshold current is obtained with the minimum number of wells, Therefore we can conclude that the multi-quantum well laser can be operated with lower threshold current density when designed with long cavities. ηi = 0.50 ηi = 0.74 ηi = 0.87 4,5 4,0 3,5 3,0 2,5 1/ηi = 2 1/ηi = 1.4 2,0 1,5 1/ηi = 1.1 1,0 0,00 0,01 0,02 0,03 The inverse external quantum efficiency (1/ηext) as a function of the cavity length is plotted in Fig. 7. The external differential quantum efficiency decreases linearly with increasing cavity length; we can see also that the shorter cavities could provide higher external differential efficiency. The main source of deterioration of differential efficiency in short cavity lasers is an increase of internal loss αi caused free carrier absorption which rises proportionally to the carrier concentration. The mirror loss coefficient rises with shorter cavity lengths, requiring higher gain and consequently more carriers in the quantum well. The inverse external quantum efficiency is given by: 0,05 0,06 0,07 Fig. 7. Inverse external differential quantum efficiency versus cavity length at different values of internal efficiency for SQW laser. 5. Current and Total Loss Factor Total optical loss, αtot, in a semiconductor laser includes two principal terms: = 4. External Quantum Efficiency and Internal Quantum Efficiency 0,04 Cavity length (cm) + (5) The external (mirrors) optical loss is αm = (1/L)ln(1/R) [12]. The internal loss αi includes the diffraction loss, the scattering and the free carrier absorption losses in the active region and mirrors[13]. If losses in a laser are high, it is necessary to have a high gain and therefore multiple quantum well lasers are preferable. From Fig. 8 we see that, for low loss, the injected threshold current is minimum in the case of 1QW. On the other hand, if the αtot = 140 cm-1, the threshold current with 1QW is larger than that of 2QW. We can see also that a five-well structure (5QW) will have the lowest threshold current if we increase the total loss higher than 140 cm-1. 250 1⁄ = 1⁄ 1+ ⁄ 1⁄ (4) Lw = 60 Å Threshold current (mA) 200 The internal quantum efficiency (ηi) and internal optical loss (αi) are extracted to be 74 % and 10 cm-1 respectively. The value of αi is lower than the typical values of 10-20 cm-1 reported for double heterostructure lasers [11]. However, in the case of QW laser, the maximum total loss (internal plus mirror) is limited due to the existence of a maximum (saturated) optical modal gain. The internal quantum efficiency is then determined by plotting the curve of inverse external differential quantum efficiency versus cavity length, it is of the order of 0.74 (1/ηi = 1.4) in our structure. The low internal loss and the high internal quantum efficiency provide high external differential quantum efficiency. αtot αtot 150 -1 = 29 cm -1 = 140 cm 100 50 0 1 2 3 4 5 Number of quantum wells Fig. 8. Threshold current versus number of quantum wells for various values of the total loss of the cavity. In this case, the quantum well thickness Lw is assumed to be 60 Å. 93 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 90-95 At higher values of αtot, which call for large laser modal gain gmod, a larger number of wells are needed. Furthermore, we can notice also that the multiquantum well with narrow thickness present low threshold currents for all numbers of wells from 1 to 5, the optimal number of wells for better performance threshold will be increased with higher losses. Another important parameter that can affect the performance of multiple quantum well is its thickness. Fig. 9 shows the threshold current as a function of the number of QW for various values of the total loss of the cavity for QW thickness of 100 Å. In this case, the number of QW with each QW thickness is optimized so that the threshold current is minimal. We can see that the threshold current of 60 Å thick well is much lower than that of 100 Å and this current is minimized with thinner QW’s when total loss is low. 400 350 Lw = 100 Å Threshold current (mA) 300 αtot= αtot= 250 -1 29 cm -1 140 cm 200 150 100 50 0 1 2 3 4 5 Number of quantum wells Fig. 9. Threshold current versus number of quantum wells for various values of the total loss of the cavity. In this case, the quantum well thickness Lw is assumed to be 100 Å. We can conclude from these figures that the optimum number of quantum wells (leading to lower threshold current) is 3 when the total loss is around 140 cm-1. If one wants to get a very low threshold current, a diode short (60 Å) and low loss (≈ 29 cm-1) is preferable in this case, the optimal number is obviously 1.This is mainly due to the fact that the current for transparency (gain equals to zero) is minimized at the thickness of Lw = 60 Å in the case of NQW = 1 and also due to the fact that the optimum number in QW lasers with each thickness is 1 in the case of low loss for our structure. 6. Conclusion We have numerically investigated the characteristics of GaAs/Al0.32Ga0.68As QW laser and the effect of quantum well number, cavity length and temperature on its performance. The threshold current and current-power characteristics as a function of numbers of wells, cavity length and temperature were obtained for QW structure. It is shown that the 94 threshold current increases with decreasing cavity length and increasing temperature and its variation with the number wells depends on loss. Whether the SQW or the MQW is the better structure depends on the loss level. At low loss, the SQW laser is always better because of its lower transparency current density and lower internal loss. At high loss, the MQW is always better because the phenomena of gain saturation can be avoided by increasing the number of QW although the injected current to achieve this maximum gain also increases. Owing to this gain saturation effect, there exists an optimum number of QW for minimizing the threshold current for a given total losses. The temperature increase in the quantum well is an important parameter which limits the maximum output power of the laser. This temperature increase is effective for high-power operation; we conclude that the cavity length, temperature and its quantum well number play important roles in determining the laser performance. Smaller values of threshold current density, Jth indicate superior performance. Jth can be minimized by maximizing the internal quantum efficiency, minimizing the loss coefficient and minimizing the quantum well thickness. The device performance was significantly improved due to the reduced total loss of ≈29 cm-1 without degradation of electrical properties, resulting in a high output power of 31.04 mW and a low threshold current of 8.2 mA and with an internal efficiency of 74 % for one well at room temperature for the 300 µm cavity and quantum well thickness of 60 Å. References [1]. P. S. Zory, Quantum Well Lasers, Academic Press, Inc, 1993. [2]. S. R. Selmic, T. M. Chou, J. Sih, J. B. Kirk, A. Mantle, J. K. Butler, D. Bour, G. A. Evans, Design and characterization of 1.3-μm AlGaInAs-InP multiplequantum-well lasers, IEEE Journal of selected topics in Quantum Electronics, Vol. 7, No. 2, 2001, pp. 340-349. [3]. M. Balkanski and R. F. Wallis, Semiconductor Physics and Applications, Oxford University Press, Oxford, 2000, pp. 1- 69. [4]. Y. Arkawa and A. Yariv, Quantum well lasers-gain, spectra, dynamics, IEEE J. Quantum Electron., 22, 1986, 1887. [5]. S. L. Chuang, Book, Physics of Photonic Devices, 2nd ed., Wiley, New York, 2009. [6]. I. Vurgaftman, J. R. Meyer and L. R. Ram-Mohan, Band Parameters for III-V Compound Semiconductors and Their Alloys, J. Appl. Phys., Vol. 89, No. 11, 2001, pp. 5815-5875. [7]. J. Z. Wilcox, G. L. Peterson, S. Ou, J. J. Yang, M. Jansen and D. Schechter, Gain and threshold current dependence for multiple-quantum well lasers, J. Appl. Phys., 64, 1988, 6564. [8]. M. Razeghi, Technology of Quantum Devices, Springer, 2009. [9]. A. K. Dutta, N. K. Dutta and M. Fujiwara, WDM Technologies: Active Optical Components, Academic Press, 2002. Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 90-95 [10]. M. F. Khodr, Effects of Non Parabolic Bands on Nanostructure Laser Devices, Proceedings of SPIE, 7039, 2008, 70390T. [11]. F. Lelarge, B. Dagens, J. Renaudier, R. Brenot, A. Accard, F. van Dijk, D. Make, O. LeGouezigou, J. -G. Provost, F. Poingt, J. Landreau, O. Drisse, E. Derouin, B. Rousseau, F. Pommereau, and G. -H. Duan, Recent Advances on InAs/InP Quantum Dash Based, Semiconductor Lasers and Optical Amplifiers Operating at 1.55 μm, IEEE J. Sel. Topics Quantum Electron., Vol. 13, 2007, pp. 111–124. [12]. N. A. Pikhtin, S. O. Slipchenko, Z. N. Sokolova, and I. S. Tarasov, Internal Optical Loss in Semiconductor lasers, Semiconductors, Vol. 38, No. 3, 2004, pp. 360-367. [13]. K. H. Ha and Y. H. Lee, Determination of Cavity Loss in Proton Implanted Vertical-Cavity Surface Emitting Lasers, Jpn. J. Appl. Phys., 37, 1998, p. 372. ___________________ 2015 Copyright ©, International Frequency Sensor Association (IFSA) Publishing, S. L. All rights reserved. (http://www.sensorsportal.com) 95 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 96-103 Sensors & Transducers © 2015 by IFSA Publishing, S. L. http://www.sensorsportal.com Multi-Model Adaptive Fuzzy Controller for a CSTR Process * Shubham Gogoria, Tanvir Parhar, Jaganatha Pandian B. Electronics and Instrumentation, VIT University, 632014, India * Tel.: +91-8681910995 * E-mail: [email protected] Received: 19 August 2015 /Accepted: 20 September 2015 /Published: 30 September 2015 Abstract: Continuous Stirred Tank Reactors are intensively used to control exothermic reactions in chemical industries. It is a very complex multi-variable system with non-linear characteristics. This paper deals with linearization of the mathematical model of a CSTR Process. Multi model adaptive fuzzy controller has been designed to control the reactor concentration and temperature of CSTR process. This method combines the output of multiple Fuzzy controllers, which are operated at various operating points. The proposed solution is a straightforward implementation of Fuzzy controller with gain scheduler to control the linearly inseparable parameters of a highly non-linear process. Copyright © 2015 IFSA Publishing, S. L. Keywords: CSTR, Linearization, State-space, Fuzzy, Weight distributor, Adaptive. 1. Introduction CSTR processes have been rigorously used in chemical industries for a long time. ProportionalIntegral-Derivative (PID) controllers have been used in process control most extensively. But these mainstream algorithms are incapable of controlling complex non-linear complex systems with accuracy. Earlier control systems were constricted by lack of computational power. But, with the increase of computing technology, it has become feasible to implement computationally expensive algorithms, like Fuzzy logic controllers. Attempts have been made to implement a PID controller over CSTR process, using cascaded control algorithms [1-3] Multi-loop PID controllers and Neuro-PID controllers have also been devised for an optimal concentration and temperature control [4-6]. But problems with most of the approaches is one of the two process variables (either process temperature or concentration) have been taken 96 as a constant. Where in practical conditions, it is not the case. Both the variables depend on each other, through a differential relation [7]. Hence, this paper focuses on controlling both the output variables, using a novel Multi-model Fuzzy controller approach. The conventional mathematical model of the CSTR process is used [8]. The non-linear equations are linearized at various operating points and cast into state space models [9-10] Using the state-space model at different operational points, transfer function matrices have been formed [11]. A Fuzzy controller maps inputs to outputs, using a set of rules that might be linear or non-linear relation. The transfer function matrix is used to tune Fuzzy Logic controller parameters. A weighing algorithm has been designed, to select the appropriate controller and state-space model, corresponding to the input signal. Since the Fuzzy controller has an oscillatory manipulative variable, so an integrator has been implemented to eliminate fluctuations. http://www.sensorsportal.com/HTML/DIGEST/P_RP_0207.htm Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 96-103 2. System Modelling The modelling of the CSTR process has been done, keeping the following assumptions in mind: 1. Volume of the reactor remains constant, during the process. 2. The chemicals are mixed perfectly inside the vessel. 3. The reaction follows the following dynamics A B + Heat (Q). Linearizing (1) and (2), transfer function matrices are obtained for the CSTR model. The idea behind computing the matrices is to develop a state space model for the system. 2.1. Mass Balance Equation In as CSTR process, two state variables are controlled, namely reactor temperature (T) and reactor concentration (C). The following differential equations symbolize the process in time domain. ( ) = Fig. 1. The Continuously Stirred Tank Reactor. ( ) ( )− ( ) ( ) exp − − ( ) (1) Where matrix A and B are Jacobian matrices of state and input variables respectively and C is output matrix. and ( ) = ( ) ( )− ( ) − exp − + ( ) ( ) ( ) ( ) = = (2) ) ∗( ( ) − ( )) (1−exp where: F = Feed flow rate; V = Volume of reactor; Caf = Feed concentration; Ca = Reactor concentration; K0 = Reaction rate constant; E = Activation energy; R = Ideal gas constant; T = Reactor temperature in K; H = Heat of reaction; hA = Heat transfer coefficient; Tf = Feed temperature; Tc = Coolant temperature; ; = Liquid densities; Cp; Cpc = Specific heats. = ( )=[ ∶ ′ (8) = = − (9) + + ∗ exp( + (10) ) here = exp − (11) The Jacobean matrix B is given by State input variables are given by ( )=[ (7) = ∶ ] (3) ] (4) = = , where 2.2. Linearization = The nonlinear equations are formed into state space variables as follows: = + y=Cu (5) (6) , = 0, = (12) (13) , (14) 97 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 96-103 = −ℎ [ − exp 2.3. Transfer Function Matrix ℎ (15) −ℎ − exp −ℎ ] At operation point 1 0.092s + 0.4123 0.0448 s + 23.891s + 23.337 s + 23.891s + 23.337 −0.947s − 35.452 −0.9413s − 12.576 s + 23.891s + 23.337 s + 23.891s + 23.337 The output matrix C is given by 1 0 C= 0 1 At operating point 2 Table 1 shows steady state parameters for the process. Table 1. Steady State Operating Data. S. No. 1 2 3 4 5 6 7 8 9 10 11 12 13 Parameters T F V hA E/R ΔH , , 14 Values 0.0882 mol/l 441.2 K 100 l/min 100 l/min 1mol/l 350 K 350 K 100 l 7×10 cal/min K 10000 K 2×10 cal/min K 1000 g/l 1cal/(gK) 72×10 Table 2 shows the all the operating points around which system has been linearized. The system is inherently stable at all five operating points, as their Eigen values are negative [12]. Table 3 shows Eigen values. 1 2 3 4 5 Feed Flow (LPM) 102 100 100 97 98 Coolant Flow (LPM) 97 100 103 103 109 0.0091 + 0.1371 + 21.681 + 20.69 −0.912 − 21.154 + 21.681 + 20.69 0.0424 + 21.681 + 20.69 −0.9053 − 10.252 + 21.681 + 20.69 At operating point 3 0.009 + 0.135 + 20.503 + 19.879 −0.8877 − 25.4 + 20.503 + 19.879 0.0413 + 20.503 + 19.879 −0.8823 − 8.9347 + 20.503 + 19.879 At operating point 4 0.0089 19.344 −0.868 19.344 + 0.1298 + 18.3697 − 22.717 + 18.3697 0.0392 19.344 + 18.3697 −0.8628 − 7.969 19.344 + 18.3697 At operating point 5 0.080087 + 0.1266 + 17.903 + 17.225 −0.83 − 18.153 + 17.903 + 17.225 0.0.377 + 17.903 + 17.225 −0.825 − 6.3863 + 17.903 + 17.225 3. Controller Design 3.1. Fuzzy Logic Table 2. Operating Points. S. No. In this section five transfer function matrices are obtained at five different operating points. They are – Conc. (mol/l) Temp (K) 0.0762 0.0882 0.0989 0.1055 0.1275 444.7 441.2 438.77 436.8 433 A fuzzy logic controller is widely used in machine control approaches that require computing based on "degrees of truth" rather than the usual "true or false" Boolean logic. The mapping between the input variables and the outputs is done through a set of predefined functions called membership function, which are also known as "Fuzzy Set" [13]. The architecture for a fuzzy controller is as shown in Fig. 2. Table 3. Eigen Values. Operating Point 1 2 3 4 5 98 Eigen Value -22.8703, -1.0204 -20.6802, -1.0005 -19.4563, -1.0006 -18.2740, -0.9705 -16.8432, -0.9808 Fig. 2. Fuzzy Controller. Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 96-103 3.2. Fuzzy Logic Controller The fuzzy controller is designed to control the temperature and concentration of the chemical product. It uses four input variables which are error and differential error of concentration and temperature of the reactor. The manipulated variables are feed flow rate and coolant flow rate. Separate controllers are used to control the manipulative variables. If a single fuzzy controller is used for all operating for all five points, it will have ten outputs (F, points). So the total input membership functions will be 7 and the total output membership functions will be7 . Hence it becomes very difficult to make all the rules for fuzzy controller manually. So, instead of one fuzzy controller, five pairs of fuzzy controllers are used, for each operating point. Each controller has a similar set of rules (Table 4 and 5 shows the rules). So, it becomes less tiresome to create the rules manually [14]. assigns different weights to the outputs of the fuzzy controllers [15], corresponding to the present input, in accordance with the following algorithm. Fig. 3. Input Membership function. Table 4. Fuzzy Rules for Feed Flow rate. E HN MN LN Z LP MP HP HN MN LN Z LP MP HP HP HP HP HP HP HP HP HP HP HP MP MP LP LP MP MP LP LP Z Z Z LP LP LP Z LN LN LN Z Z Z LN LN MN MN LN LN MN MN HN HN HN HN HN HN HN HN HN HN Table 5. Fuzzy Rules for Coolant Flow rate. E HN MN LN Z LP MP HP HP MP LP Z LN MN HN HP HP HP HP HP HP HP HP HP HP MP MP LP LP MP MP LP LP Z Z Z LP LP LP Z LN LN LN Z Z Z LN LN MN MN LN LN MN MN HN HN HN HN HN HN HN HN HN HN The fuzzy logic controller uses triangular membership functions. Fig. 3 and Fig. 4 depict the membership function of input and output variable respectively. The system has been linearized at five operating points. Since for each operating point a pair of fuzzy controllers needed, hence a total of ten controllers or five pairs have been designed for the process. The transfer function matrices were used to tune each pair of fuzzy controllers simultaneously. 3.3. Weighing Algorithm Since, each pair of fuzzy controller has been tuned around a particular operating point, so a scheduler Fig. 4. Output Membership function. If there are n operating points and y is the input at that instant, 1. Initialize a variable i to 0. 2. Iterate steps 3 and 4 till i is less than n. and ( + 1) 3. Check if y lies between operating point. • if yes, then update the weight of fuzzy output as [ [ ]− ] [ ]=1− [ [ ] + [ + 1]] And the output of ( + 1) Fuzzy As [ + 1] = 1 − [ ] And increase i to i+1. • else, update the weight of Fuzzy to 0 4. Update = + 1. The system has a predefined set-point of temperature and concentration. The output of the controller is F and which goes into actuator. The actuator output serves as the input for the process. The error and differential error, with respect to the set point are given to the controller. The weight scheduler assigns weight to the corresponding controllers as per the set-point at that instant. So, the process is adapts the best controller needed to track the set-point. The SIMULINK model for the whole process is shown in Fig. 5. 99 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 96-103 Fig. 5. CSTR Model. 4. Results 5. Conclusions 4.1. Simulation In this paper the authors have proposed an Adaptive Fuzzy controller for set-point tracking of a CSTR process. The objective was to control both the reactor concentration and temperature with minimal error. CSTR processes have wide industrial applications. These processes need an accurate temperature and concentration to yield desirable products. The Adaptive-Fuzzy controller has an almost negligible steady state error for various operational points. Hence, it makes Adaptive-Fuzzy controller an apt controller for industrial processes. By linearizing the system around different operating points, a better approximation of the system can be made and a weighed combination of Fuzzy controllers can be used to take a control action around the given set-point. MATLAB provides with an comprehensive environment for algorithm development and simulation. SIMULINK is an add-on that is used to model the process in a block diagram format. The system has been linearized and simulated in MATLAB 2013. In order to test the set-point tracking capability of the control algorithm, various set-points have been tested at continuous intervals of time. The initial conditions of the system were: = 0.762 = 444.7 = 102 / = 97 / / From the responses it can be said that the controller designed for the CSTR process is able to maintain the desired set-points for dynamic change in concentration and temperature. The variation in controller output is presented in Fig. 6 and Fig. 7. Fig. 8 and Fig. 9 represents the set point tracking of reactor concentration and temperature. Fig. 6. Feed Flow. 100 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 96-103 Fig. 7. Coolant Flow. Fig. 8. Reactor Concentration. Fig. 9. Reactor Temperature. References [1]. Aslam Farhad, Gagandeep Kaur, Comparative analysis of conventional, P, PI, PID and fuzzy logic controllers for the efficient control of concentration in CSTR, International Journal of Computer Applications, 17, 6, 2011, pp. 12-16. [2]. D. Krishna, K. Suryanarayana, G. Aparna, R. Padma Sree, Tuning of PID Controllers for Unstable Continuous Stirred Tank Reactors, International Journal of Applied Science and Engineering, 10, 1, 2012, pp. 1-18. [3]. Kumar Sandeep, Analysis of Temperature Control of CSTR Using S Function, International Journal of 101 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 96-103 [4]. [5]. [6]. [7]. [8]. [9]. Advanced Research in Computer Science and Software Engineering, Vol. 2, Issue 5, 2012. Yazdani A. M., M. A. Movahed, S. Mahmoudzadeh, Applying a novel fuzzy-PI controller on the model of continuous stirred tank reactor, International Journal of Machine Learning and Computing, Vol. 1, No. 2, 2011, pp. 218-223. Upadhyay Rahul, Rajesh Singla, Analysis of CSTR temperature control with adaptive and PID controller (a comparative study), International Journal of Engineering and Technology, Vol. 2, No. 5, 2010, pp. 453-458. Vinodha R., S. Abraham Lincoln, J. Prakash, Multiple model and neural based adaptive multi-loop PID controller for a CSTR process, World Academy of Science, Engineering and Technology, Vol. 68, 2010, pp. 505-510. Shyamalagowri M., R. Rajeswari, Modeling and Simulation of Non Linear Process Control Reactor – Continuous Stirred Tank Reactor, International Journal of Advances in Engineering Technology, Vol. 6, Issue 4, 2013, pp. 1813-1818. Stephanopoulos George, Chemical process control: An introduction to theory and practice, Prentice Hall, New Jersey, 1984. Boobalan S., K. Prabhu, V. Murali Bhaskaran, Fuzzy Based Temperature Controller For Continuous Stirred Tank Reactor, International Journal of Advanced [10]. [11]. [12]. [13]. [14]. [15]. Research in Electrical, Electronics and Instrumentation Engineering, Vol. 2, Issue 12, 2013, pp. 5835-5842. Luyben William L., Process modeling, simulation and control for chemical engineers, McGraw-Hill Higher Education, 1989. Om Prakash Verma, Sonu Kumar, Gaurav Manik, Analysis of Hybrid Temperature Control for Nonlinear 4th International Conference on Soft Computing for Problem Solving (SocProS’14), Vol. 2, 2015, pp. 103-121. Debeljkovi Dragutin Lj., Mia B. Jovanovi, Ljubomir A. Jacic, Transfer function matrix and fundamental matrix of linear singular-descriptive systems, Scientific-Technical Review, Vol. LIV, No. 1, 2004, pp. 77-90. Mendel Jerry M., Type-2 fuzzy sets and systems: an overview, IEEE Computational Intelligence Magazine, Vol. 2, No. 1, 2007, pp. 20-29. Tani Tetsuji, Shunji Murakoshi, Motohide Umano, Neuro-fuzzy hybrid control system of tank level in petroleum plant, IEEE Transactions on Fuzzy Systems, Vol. 4, No. 3, 1996, pp. 360-368. Dougherty Danielle, Doug Cooper, A practical multiple model adaptive strategy for multivariable model predictive control, Control Engineering Practice, Vol. 11, No. 6, 2003, pp. 649-664. ___________________ 2015 Copyright ©, International Frequency Sensor Association (IFSA) Publishing, S. L. All rights reserved. (http://www.sensorsportal.com) 102 Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 104-105 Sensors & Transducers © 2015 by IFSA Publishing, S. L. http://www.sensorsportal.com SENSOR TECHNOLOGY LTD Ten Top Torque Tips (White Paper) 1 Bob Dobson, 2 * Tony Ingham 1 BDL Ltd, High Bank, River, Petworth, West Sussex, GU28 9AX, UK Tel.: 01798 861677 2 Sensor Technology Ltd., Balscott Mill, Balscote, Banbury, Oxon, OX15 6EY, UK * Tel.: 01295 730746 E-mail: [email protected] Abstract: Driveshafts deliver power as a rotary force, and in most applications there is a need to know the amount of power in the system. But getting measurements from a turning shaft requires some engineering ingenuity, so here Tony Ingham from Sensor Technology in Banbury runs through the basics. Keywords: Torque sensors; Turning force; Strain gauges; Wheatstone bridge; Slip rings; Rotary transformer; Surface acoustic waves; Piezoelectric strain gauges; Torque transducer 1. Torque is a turning force and, because every material from which a driveshaft can be made has a degree of elasticity, it is also a twisting force that deforms the shaft in the direction of rotation. This deformation is usually only very slight, but it is 104 massively important when it comes to measuring torque. 2. Probably the most common way to measure torque in a turning shaft is to glue strain gauges onto it. A strain gauge is effectively a zig-zag of wire encapsulated in a flexible substrate for protection. When the shaft twists under torque the zig-zag will be stretched, which will change its electrical impedance. Therefore, the resistance of the wire is proportional to the torque in the shaft. However, there is a problem in that electrical leads connected to the strain gauges would wrap around the shaft as it turned and eventually snap. Fortunately, there are solutions to this, which we will look at later. 3. In fact, it is normal practice to use not one, but four strain gauges spaced along the shaft. All are at 45 deg to the direction of rotation, two to the left and two to the right. These are then connected into a Wheatstone bridge configuration, which produces an electrical output that is linearly proportional to the torque in the shaft as it rotates. http://www.sensorsportal.com/HTML/DIGEST/P_W_5.htm Sensors & Transducers, Vol. 192, Issue 9, September 2015, pp. 104-105 The circuitry is completed with a power supply, amplifier and a display, recorder or computer. These are usually mounted somewhere close to the driveshaft in a secure, static location. 4. We now come onto the elephant in the room. How do we connect the stationary parts of the circuit to the spinning Wheatstone bridge? One answer is slip rings – collars mounted onto and standing proud of the shaft, which contacts static brushes. Unfortunately, the slip rings are rather delicate; so much care must be exercised in their use. Also, they need to be set up with some precision so that a constant and even contact is maintained in operation. Because of brush wear, slip rings need regular attention and they are not really suitable for longterm use, nor for deployment in harsh working environments. It is also notable that the contact between the stationary brushes and rotating collars will create a degree of electrical noise which, particularly at higher speeds, will interfere with signal transmission. A final shortcoming of slip rings is that they create a drag force, which must be accounted for in signal measurements and frequently checked to make sure that it has not changed in value. 5. An alternative to the slip ring is the rotary transformer, sometimes called an inductive loop. This consists of two adjacent electrical coils – one static the other rotating with the driveshaft – the relative motion of which induces a current in the transducer. There is no physical contact between the coils, nor between shaft and transducer, yet power and signals are passed between them. This overcomes many of the drawbacks of slip ring systems; set up is easier, operation is more robust, there is no friction and higher speed operations can be accommodated. However, every rotary transformer will have a maximum operating speed due to its inertia. They are also susceptible to noise and errors, especially if the coils become misaligned. 6. The next development on a torque sensor is the type that is based on radio telemetry. These operate on the internationally license-free 2.4 GHz frequency band and are based on the concept of mounting a receiver pickup so that it can communicate with the strain gauges. The pickup can actually be many meters from the driveshaft, so long as communication is maintained. Installation and maintenance are straightforward because there is no slip ring to adjust, coils to align or cabling to accommodate. However, a battery is required to power the signals, which although small and long lasting, does require consideration. 7. Perhaps the ultimate type of torque sensor is that based on the detection of Rayleigh waves or surface acoustic waves (SAWs). These are noncontact and use piezoelectric strain gauges. In a SAW sensor, the surface waves are produced by passing an alternating voltage across the terminals of two interleaved comb-shaped arrays, laid onto one end of a piezoelectric substrate. A receiving array at the other end of the transducer converts the wave into an electric signal. The wave frequency is dependent upon the spacing of the teeth in the array and the direction of wave propagation is at right angles to the teeth. Therefore, any change in its length, caused by the dynamic forces of the shaft's rotation, alters the spacing of the teeth and hence the operating frequency. To measure the torque in a rotating shaft, two SAW sensors are bonded to a shaft at 45° to the axis of rotation. When the shaft is subjected to torque, a signal is produced, which is transmitted to the adjacent stationary pick-up via the RF couple. Interestingly, SAWs were first detected by 19th century gentleman–scientist and Nobel Laureate Lord Rayleigh when he was investigating the cause and effects of earthquakes. 8. Selection of the type of torque transducer will be based on many considerations including: the working environment, the expected length of operation, the rotational speed of the driveshaft, mechanical connection options and costs. There is no overall ‘best’, but an optimum choice for each individual situation. 9. Positioning a torque sensor can be a complicated decision if a true reading is to be obtained. Inaccuracies can creep in due to the effects of adjoining elements in the drive train, the damping effect of the driveshaft’s own end couplings and drag caused by contact-type sensors. 10. The final piece of advice is that expert help is usually available through the company supplying your torque sensor. Availing yourself of this service will probably save time, money and frustration. References [1]. Tony Ingham, Measuring torque is fundamental to many machines, test rigs and other engineering installation. [2]. TorqSense represents the ultimate in wireless torque monitoring and is available in a range of sizes. [3]. Heavy-duty materials handling environments can be too harsh for slip ring-based measuring equipment, making wireless the preferred option. [4]. http://www.sensors.co.uk/media-centre/ten-toptorque-tips/ Sensor Technology Ltd. Apollo Park, Ironstone Lane, Wroxton, Banbury OX15 6AY Tel: +44(0)1869 238400, fax: +44(0)1869 238401, http://www.sensors.co.uk/ 105 Aims and Scope Sensors & Transducers is established, international, peer-reviewed, open access journal (print and electronic). It provides the best platform for the researchers and scientist worldwide to exchange their latest findings and results in science and technology of physical, chemical sensors and biosensors. The journal publishes original results of scientific and research works related to strategic and applied studies in all aspects of sensors: reviews, regular research and application specific papers, and short notes. In comparison with other sensors related journals, which are mainly focused on technological aspects and sensing principles, the Sensors & Transducers significantly contributes in areas, which are not adequately addressed in other journals, namely: frequency (period), duty-cycle, time-interval, PWM, phase-shift, pulse number output sensors and transducers; sensor systems; digital, smart, intelligent sensors and systems designs; signal processing and ADC in sensor systems; advanced sensor fusion; sensor networks; applications, etc. By this way the journal significantly enriches the appropriate databases of knowledge. Sensors & Transducers journal has a very high publicity. It is indexed and abstracted very quickly by Chemical Abstracts, EBSCO Publishing, IndexCopernicus Journals Master List (ICV=6.13 in 2011), Open J-Gate, ProQuest Science Journals, CAS Source Index (CASSI), Ulrich's Periodicals Directory, Scirus, Google Scholar, etc. Since 2011 to 2014 Sensors & Transducers journal was covered and indexed by EI Compendex (CPX) index (including a Scopus, Embase, Engineering Village and Reaxys) in Elsevier products. The journal is included in the IFSA List of Recommended Journals (up-dated 9.12.2014), which contains only the best, established sensors related journals. The global Impact Factor of the journal is 0.705 (2013). Physical, chemical and biosensors Digital, frequency, period, duty-cycle, time interval, PWM, pulse number output sensors and transducers Theory, principles, effects, design, standardization and modelling Smart sensors and systems Intelligent sensors and systems Sensor instrumentation Virtual instruments Sensors interfaces, buses and networks Signal processing, sensor fusion Remote sensing Frequency (period, duty-cycle)-to-code converters, ADC Technologies and materials Nanosensors and NEMS Microsystems Applications Further information on this journal is available from the Publisher's web site: http://www.sensorsportal.com/HTML/DIGEST/Submission.htm Subscriptions An annual subscription includes 12 regular issues and some special issues. Annual subscription rates for 2015 are the following: Electronic version (in printable pdf format): 590.00 EUR; Printed with b/w illustrations: 950.00 EUR; 25 % discount is available for IFSA Members. Prices include shipping costs by mail. Further information about subscription is available through IFSA Publishing's web site: http://www.sensorsportal.com/HTML/DIGEST/Journal_Subscription.htm Advertising Information If you are interested in advertising or other commercial opportunities please e-mail [email protected] and your enquiry will be passed to the correct person who will respond to you within 24 hours. Please download also our Media Planner 2015: http://www.sensorsportal.com/DOWNLOADS/Media_Planner_2015.pdf Instructions for Authors Please visit the journal web page http://www.sensorsportal.com/HTML/DIGEST/Submission.htm Authors must follow the instructions very carefully when submitting their manuscripts. Manuscript must be send electronically in both: MS Word for Windows and Acrobat (pdf) formats by e-mail: [email protected]
