06239554

Embed Size (px)

Citation preview

  • 7/28/2019 06239554

    1/5

    IEEE SENSORS JOURNAL, VOL. 12, NO. 10, OCTOBER 2012 3031

    High-Sensitivity Miniature Smoke DetectorEzzat G. Bakhoum, Senior Member, IEEE

    Abstract This paper introduces a novel new smoke detector

    characterized by its very small size and very high sensitivity. Thenew detector is fundamentally based on an -particle radiationsource like the well-known -particle smoke detector. However,instead of utilizing the principle of ionization of the air, the particles are made to strike the gate of an n-channel MOSFET .This results in a net positive charge on the gate of the transistor.The current through the MOSFET will be proportional to thecharge on the gate and, hence, to the intensity of the particles.If particles of smoke enter the detector and screen the particles,the positive charge on the gate drops, which leads to a reducedcurrent or total cut-off of the MOSFET .

    Index Terms Alpha-particle detector, ionization chamber, N-channel MOSFET, smoke detector, smoke detector sensitivity.

    I. INTRODUCTION

    T HE UBIQUITOUS alpha-particle smoke detector hasbeen in wide use for the past four decades. Fig. 1(a)shows the basic structure of that popular smoke detector: an -particle emitter (such as Americium 241) emits particles intoan ionization chamber that consists of two metal electrodesseparated by air [1], [2]. The particles ionize the airmolecules, which results in the release of a large number of free electrons inside the chamber. Accordingly, the resistivityof the air inside the chamber decreases, and a small currentcan be circulated between the two metal electrodes. Whensmoke particles (which contain a large number of carbonatoms), however, enter the chamber, they quickly attach tothe particles and hence reduce the degree of ionization of the air. The momentary increase in resistivity results in a lowercurrent between the electrodes, which can be detected with theexternal circuit.

    Unfortunately, the change in the resistivity of the air inthe traditional smoke detector is quite small in practice;which necessitates the use of a large ionization chamber andan elaborate external circuit to detect the small changes inresistivity that occur when smoke particles enter the detector.Accordingly, the smoke detector has been a traditionally largedevice. In recent years, however, there has been an interestin a very small smoke detector that can be hidden close to apassenger seat in an aircraft, train, or bus [3]. Such a detectorcan be useful not only for detecting cigarette smokers, butalso for detecting potential terrorists (who have attempted

    Manuscript received November 9, 2011; revised July 4, 2012; acceptedJuly 10, 2012. Date of publication July 13, 2012; date of current versionAugust 7, 2012. The associate editor coordinating the review of this paperand approving it for publication was Prof. Michael J. Vellekoop.

    The author is with the Department of Electrical and ComputerEngineering, University of West Florida, Pensacola, FL 32514 USA(e-mail: [email protected]).

    Color versions of one or more of the gures in this paper are availableonline at http://ieeexplore.ieee.org.

    Digital Object Identier 10.1109/JSEN.2012.2208741

    (a)

    (b)

    I

    VDS++++

    +++

    Fig. 1. (a) Principle of operation of the traditional -particle smoke detector.(b) Principle of operation of the new detector.

    in recent years to use explosive/combustible materials insidepublic transportation systems). It is the objective of this paperto introduce such a miniature smoke detector. The fundamentalprinciple of the new detector is shown in Fig. 1(b). Instead of relying on the principle of ionization of the air, the particlesemitted from the -particle source are made to strike directlythe gate of an n-channel MOSFET transistor. The particles,being positively charged helium nuclei, will deposit theirpositive charges on the gate. The MOSFET is known to behighly sensitive to a charge on the gate, and will immediatelystart conducting when the particles reach the gate. Thecurrent I through the MOSFET will be proportional to thecharge on the gate, and hence to the number of particles

    reaching the gate. If smoke particles enter the detector andget attached to the particles, a lower positive charge will bepresent on the gate, which will lead to a lower current or totalcut-off of the MOSFET.

    Theoretical and experimental investigations will show thatthis principle is highly sensitive by comparison with the oldtechnique of detecting smoke; and hence this new device canbe made substantially smaller than the traditional ionization-based smoke detector (in fact, the new detector is much smallerthan the audio buzzer that is typically connected to the detectoras an alarming mechanism). It is interesting to point out that

    1530437X/$31.00 2012 IEEE

  • 7/28/2019 06239554

    2/5

    3032 IEEE SENSORS JOURNAL, VOL. 12, NO. 10, OCTOBER 2012

    +++

    +++

    + Vcc

    +

    + Vcc

    V10 M

    10 k

    2N2222

    MMBF170

    Fig. 2. Circuit used in the present prototype. A DC buzzer can be connectedto the output of the driving transistor to obtain an audible alarm.

    optical smoke detectors have been the subject of much researchduring the past decade [4][8], while radiation based detectors

    have not received similar attention (except perhaps for onepaper by B. Liu et al. [9] in which the -particle sourcewas replaced by a Beta emitter). Of course, the advantage of radiation based detectors by comparison with optical detectorsis their capability to detect very minute (invisible) amounts of smoke.

    II . QUALITATIVE DESCRIPTION OF THE DETECTOR

    Fig. 2 shows a schematic of the actual circuit used in thepresent prototype.

    As shown, an -particle source is arranged such thatthe emitted particles strike a small plate that is directlyconnected to the gate of an n-channel, enhancement modeMOSFET. The -particle source is a commercially availablethin foil containing the 241 Am isotope. 1 An optional 10 Mresistor can be connected to the gate, as shown, to drain thepositive charge from the gate when the particles are notpresent. The voltage on the drain terminal of the MOSFETis sensed with a voltage-follower circuit consisting of a high-impedance opamp and an NPN transistor. The purpose of theNPN transistor is to act as a load driver (for driving a DCbuzzer or other means for indicating the presence of voltage).When the particles are striking the gate of the MOSFET and

    a positive charge is present, the MOSFET conducts and hencethe output voltage will be low. If however, the particles arescreened by the presence of smoke and no positive chargeis present on the gate, the MOSFET shuts off. But since theinput to the opamp is connected to a pull-up resistor, the outputvoltage in that case will be high (indicating an alarm).

    Fig. 3 shows a photograph of the new detector next to atraditional -particle smoke detector (for size comparison).The prototype shown is complete except for the optional DCbuzzer, which is not included in the photograph.

    1The half-life of 241 Am is 432 years.

    Fig. 3. Photograph of the new detector (small circuit board on right), nextto a traditional -particle smoke detector.

    n-channel (Si)

    GateMetal SiO 2

    niar Decr uoS

    p-substrate (Si)

    Fig. 4. Simplied structure of the n-channel MOSFET .

    III. T HEORY OF OPERATION

    A. Determination of the Gate Voltage V GS

    The radioactive foil used in the present detector con-tains approximately 0.1 Ci of the -particle emitter 241Am(one-tenth of the quantity that is typically used in householdsmoke detectors). 0.1 Ci is equivalent to 3700 emissions/s.Since the helium nucleus contains two protons, then theequivalent of 7400 positive electron charges will be strikingthe gate of the MOSFET each second. To determine the gate-source voltage V GS , consider the simplied cross-sectionaldrawing of an n-channel MOSFET shown in Fig. 4. 2 Thereare two capacitances of interest in this application. The rstis the gate-source capacitance, which can be measured andis usually given in the datasheet of the device. The secondcapacitance is the gate-channel capacitance [10], which is asubstantially smaller and much harder to determine capaci-tance (that capacitance must be calculated from an elaboratemathematical model of the device). Due to the bombardment

    2Please note: Fig. 4 is a simplied schematic. Additional details, such asn+ regions, are not shown.

  • 7/28/2019 06239554

    3/5

    BAKHOUM: HIGH-SENSITIVITY MINIATURE SMOKE DETECTOR 3033

    of the gate with particles and the deposition of posi-tive charges on the gate, electrons will be attracted to thechannel by induction. Since there is no well-dened voltagesource between the gate and the source terminals, the onlycapacitance that must be accounted for when attempting tocalculate the voltage rise on the gate will be the gate-channelcapacitance. The normal approach would be to use the well-known equation Q = CV [11], where Q is the charge onthe gate and C is the capacitance, to determine the unknownvoltage. Unfortunately, such a calculation cannot be performedsince the gate-channel capacitance cannot be determined withreasonable accuracy.

    An alternative approach for determining the steady-stategate-source voltage V GS will be the following: the steady-state current that will be owing into the gate terminal will begiven by

    I =d Qdt

    =7400 1.6 10 19

    1= 1.18 10 15 A. (1)

    (The optional high-value resistor that is shown in Fig. 2 isnot assumed to be present in this analysis). Under steady-stateconditions, this very small current of about 1 fA will constitutea leakage current that will ow to the source terminal, throughthe insulating Silicon Dioxide (SiO 2) layer that isolates thegate terminal. The leakage current density J through theSiO2 layer will be related to the electric eld intensity E between the gate and the source terminals by the well-knownrelationship [11]

    J = E (2)

    where is the conductivity of SiO 2 (this value is approx-imately 10 16 1m 1 [12]). The above equation can bewritten as

    I A

    = V GS x

    (3)

    where A is the surface area of the gate electrode and x is thedistance, in general, between the gate and the source (whichin practice can be very non-uniform). From Eq. (3), V GS canbe calculated by using the following expression:

    V GS = I x A

    . (4)

    The ratio x / A can be now determined by knowledge of the gate-source capacitance. According to the datasheet of theMMBF170 transistor used in the present prototype, C 24 pF[13]. From the well-known equation [11]

    C = 0 r A x

    (5)

    where 0 is the permittivity of free space, r is the relativepermittivity of the insulating material ( r = 3.9 for SiO 2), theratio x / A can be calculated to be

    x A

    =0 r

    C 1.438 m 1. (6)

    By substituting with the quantities determined above intoEq. (4), the steady-state V GS is found to be approximatelyequal to 17 V. Direct measurement of the gate voltage witha Keithley high-impedance electrometer have conrmed this

    1 2 3 4 5 6 7 80

    V DS (V)

    I D (mA)

    V cc

    R = 0.6 mA

    Vcc (6V)

    VGS = 17 V

    Load LineA

    B

    Fig. 5. Load line for the 10 k pull-up resistor, and the two operating curvesof the MOSFET . The drain current I D is plotted versus the drain-source voltageV DS , with an arbitrary scale for I D .

    estimated voltage. 3 It is to be pointed out that the steady-stategate voltage is typically reached within a fraction of a secondin the present application.

    B. Determination of the ON-OFF Operating Points of the Detector

    Fig. 5 shows the load line for the circuit in Fig. 2 and thetwo MOSFET characteristic curves that are of interest in this

    application. Because the load attached to the MOSFET (the10 k pull-up resistor) is large, the MOSFET will be operatingin the ohmic region as Fig. 5 shows. When V GS = 0 (nocharge on the gate), the MOSFET will be cut-off and V DS V cc , which is 6 V in the present prototype. This is the loweroperating point in Fig. 5. The determine the upper operatingpoint (where V GS = 17 V), we must solve the following twosimultaneous equations:

    I D =V cc R

    1 R

    V DS

    I D =V DS Ron

    (7)

    where R is the value of the load resistance (10 k ) and Ronis the ON resistance of the MOSFET (the inverse of the slopeof the characteristic curve shown in Fig. 5). The rst of theabove equations is the load-line equation, and the second isthe equation representing the characteristic curve. The solutionof the two simultaneous equations is

    V DS =V cc

    1 + R/ Ron(8)

    3The electrometer (Keithley model 6517B) showed a reading of about 8 V,which is to be expected since the impedance of the electrometer is comparableto the impedance of the MOSFET described here.

  • 7/28/2019 06239554

    4/5

    3034 IEEE SENSORS JOURNAL, VOL. 12, NO. 10, OCTOBER 2012

    5 10 15 20 25 30 350

    1

    2

    3

    4

    5

    6

    Obscuration (%)

    Sensor Output, V DS (V)

    Fig. 6. MOSFET drain-source voltage (output of the circuit in Fig. 2) as afunction of the obscuration.

    5 10 15 20 25 30 350

    15

    30

    45

    60

    75

    90

    Obscuration (%)

    Conventional Smoke Detector Output (% of Full Scale)

    Fig. 7. Output of a conventional household ionization-based smoke detector(as percentage of full-scale output), as a function of obscuration.

    with R = 10 k , and the MOSFET resistance Ron beingtypically equal to 1 , it is clear that V DS 0 when theMOSFET is fully ON.

    IV. EXPERIMENTAL RESULTS A. Sensor Output as a Function of Obscuration

    When smoke enters the detector with various densities andscreens the particles, the operating point of the sensorwill move from point A in Fig. 5 to point B. Accordingly,the output voltage ( V DS ) will increase from 0 to approxi-mately V cc (which is 6 V in the present prototype). Fig. 6shows the measured output voltage as a function of the

    10 20 30 40 50 60 700

    1

    2

    3

    4

    5

    6

    Source-Gate Distance (mm)

    Sensor Output, V DS (V)

    Fig. 8. Output of the sensor as a function of the distance between the -particle source and the gate, at a xed obscuration of 30%.

    Fig. 9. MOSFET drain-source voltage (output of the circuit in Fig. 2) as afunction of the obscuration, at temperatures of + 80 C, + 25 C, and 40 C.

    obscuration (in percentage). The level of obscuration wasdetermined at each point in the graph of Fig. 6 in accordancewith standard NFPA-270 [14], where a laser detector 4 wasused in a conical radiant source conguration as describedin [14] and [15] to measure the smoke obscuration. Theresults in Fig. 6 were obtained by testing the sensor instill air. The tests were also conducted by using moving

    4Detector model Analaser, from Fenwall Protection Systems,Minneapolis, MN.

  • 7/28/2019 06239554

    5/5

    BAKHOUM: HIGH-SENSITIVITY MINIATURE SMOKE DETECTOR 3035

    air, and a very small improvement in the sensitivity wasobserved. Since the improvement is incremental, it will not bereported here.

    Fig. 7 shows the measured output of a conventional house-hold smoke detector (as percentage of full-scale output) as afunction of obscuration. By comparing gures 6 and 7, it isclear that the new sensor has substantially higher sensitivityat low levels of obscuration. The sensitivity is, however, thesame for levels of obscuration higher than 35%.

    B. Sensor Output as a Function of the Distance Between the -Particle Source and the MOSFET Gate

    The output of the sensor was measured as a function of the distance between the -particle source and the gate of theMOSFET, at a xed obscuration of 30%. The result is shownin Fig. 8.

    For the present prototype, the distance between the -particle source and the gate is 5 mm.

    C. Effect of the Ambient Temperature

    The MOSFET is known to be slightly sensitive to tem-perature. More specically, the drain current I D decreasesat temperatures higher than room temperature and increasesat lower temperatures. For the circuit in Fig. 2, the resultis that the output voltage V DS drifts higher and lower withtemperature. To test the effect of temperature variation, thesensor was placed inside a variable temperature chamberand the tests shown in Fig. 6 were repeated. Fig. 9 shows theresults for temperatures of + 80 C and 40 C. Clearly, theeffect of the ambient temperature on the performance of thesensor is minimal.

    V. CONCLUSIONThe new -particle smoke detector described in this paper

    is substantially smaller and substantially more sensitive thanconventional smoke detectors that depend on the principleof air ionization. The principle used in the new detector,namely, the transfer of the charge of the particles to thegate of a MOSFET, allows the construction of a detector that ischaracterized by small size and high sensitivity by comparisonwith the ionization chamber that is used in conventional -particle based detectors. As Fig. 6 shows, the sensitivity of the new detector is substantially high at low levels of smokeobscuration. The sensor will be very useful for detecting

    smoke in tight places, such as next to a passengers seatin a bus or airplane. As indicated earlier in this paper, theadvantage of radiation based detectors by comparison withoptical detectors is their capability of detecting very minuteamounts of smoke.

    REFERENCES[1] W. Boyes, Instrumentation Reference Book . Burlington, MA: Elsevier,

    2010.[2] S. A. Dyer, Survey of Instrumentation and Measurement . New York:

    Wiley, 2001.[3] C. Hipsher and D. Ferguson, Fire protection, AERO Mag. , vol. 2, pp.

    1119, Jun. 2011.[4] J. Cheon, J. Lee, I. Lee, Y. Chae, Y. Yoo, and G. Han, A single-chip

    CMOS smoke and temperature sensor for an intelligent re detector, IEEE Sensors J. , vol. 9, no. 8, pp. 914921, Aug. 2009.

    [5] Z. J. Aleksic, The analysis of the transmission-type optical smokedetector threshold sensitivity to the high rate temperature variations, IEEE Trans. Instrum. Meas. , vol. 53, no. 1, pp. 8085, Feb. 2004.

    [6] E. D. Lester and A. Ponce, An anthrax smoke detector, IEEE Eng. Med. Biol. Mag. , vol. 21, no. 5, pp. 3842, Sep.Oct. 2002.

    [7] Z. J. Aleksic, Evaluation of the design requirements for the electricalpart of transmission-type optical smoke detector to improve its thresholdstability to slowly varying inuences, IEEE Trans. Instrum. Meas. ,vol. 49, no. 5, pp. 10571062, Oct. 2000.

    [8] Z. J. Aleksic, Minimization of the optical smoke detector false alarmprobability by optimizing its frequency characteristic, IEEE Trans. Instrum. Meas. , vol. 49, no. 1, pp. 3742, Feb. 2000.

    [9] B. Liu, D. Alvarez-Ossa, N. P. Kherani, S. Zukotynski, and K. P. Chen,Gamma-free smoke and particle detector using tritiated foils, IEEE Sensors J. , vol. 7, no. 6, pp. 917918, Jun. 2007.

    [10] D. Neamen, An Introduction to Semiconductor Devices . New York:McGraw-Hill, 2007.

    [11] W. H. Hayt and J. A. Buck, Engineering Electromagnetics . New York:McGraw-Hill, 2006.

    [12] James F. Shackelford, and W. Alexander, CRC Materials Science and Engineering Handbook . Boca Raton, FL: CRC Press, 2000.

    [13] Fairchild Semiconductor. (2010). BS170/MMBF170 N-Channel Enhancement Mode Field Effect Transistor , San Jose, CA [Online].Available: http://www.fairchildsemi.com

    [14] National Fire Protection Association. (2002). NFPA 270 - Standard Test Method for Measurement of Smoke Obscuration using a Conical Radiant Source in a Single Closed Chamber , Quincy, MA [Online]. Available:http://www.nfpa.org

    [15] P. E. Patty. (2010). A Scientic Approach to Characterize Smoke from Flaming and Smoldering Fires . Underwriters Laboratories Inc.,Northbrook, IL [Online]. Available: http://www.ul.com

    Ezzat G. Bakhoum (SM08) received the B.S.degree from Ain Shams University, Cairo, Egypt,in 1986, and the M.S. and Ph.D. degrees fromDuke University, Durham, NC, in 1989 and 1994,respectively, all in electrical engineering.

    He was a Senior Engineer and Managing Partnerwith ESD Research, Inc., Durham, from 1994 to

    1996. He was a Senior Engineer with LockheedMartin/L3 Communications, Inc., Camden, NJ, from1996 to 2000. He was a Lecturer with the ElectricalEngineering Department, New Jersey Institute of

    Technology, Newark, NJ, from 2000 to 2005. He is currently an AssociateProfessor with the University of West Florida, Pensacola.