Smart Gas Sensor for Miniaturization of E- nose

Rajesh Kumar S. A. Imam

Dep’t. Of ECE, AKG Engineering College, Dep’t. Of ECE, Jamia Millia Islamia,
Ghaziabad, UP, INDIA & PhD scholar NEW DELHI, INDIA
(JMI, New Delhi, India) Email: [email protected]
Email: [email protected]
any communication media. It is also connected to the processor
Abstract- Smart gas sensor for miniaturization of E-nose is which can react it self. An example, if gas sensor array detects
the need of the fast developing decade. Some researchers any toxic gasses up to dangerous level of lethal concentration
have developed very sensitive integrated sensing elements then processor will switch on the exhaust fan in the laboratory.
for gas detection. The polymer coated micro bridges in high
density arrays can detect ppm level concentrations of vapors
using on chip integrated read out techniques. The II. GAS SENSOR ARRAY
demonstrated technology is very suitable for miniaturization In this section we discussed the toxic gasses, materials
of electronic nose devices which provide low power
consumption (<1µw/bridge) and small form factor. In this used for gas sensors and technology used for gas sensing
paper we present a smart gas sensor which consist a set of technology.
gas sensor array followed by interfacing circuits and a
processor. In this processor we have used GPD (Gate Plated
Device) which provide us very high speed and low power
consumptions. First of all in this paper we have gone
through concentration of all the gasses which is immediately
dangerous to life and health. Many gas sensors are available
now days which can detect the various level of concentration
of different gasses. In this paper we present the new setup
which provides us alarming system with automatic control FIGURE – 1 (Block diagram of toxic gas detector)
to the protection unit. We assembled these four blocks
together to provide solution in critical situation like A. TOXIC GASSES
detection of toxic gasses in the mines, where some times
response time is very less. In that situation a human being Before going for design part, a table is provided with list of all
can’t respond within required time. Designing an intelligent the toxic gasses with their lethal concentration (in ppm) at room
situated agent is a difficult task because the designer must temperature and pressure (760 mm of Hg). The listed range of
see the problem from the agent's viewpoint, considering all LC (Lethal Concentration) of their respective gasses is
its sensors, actuators, and computation systems. dangerous to life [1].
NAME OF GASSES MW lethal NAME OF GASSES MW lethal
concentrati concentrati
Index Terms— Toxic gasses, concentration, controller, nano (760m on (760m on
gas sensor. m Hg m Hg
pressur (LC in pressur (LC in
e) PPM)* e) PPM)*
I. INTRODUCTION
1,3- 162 1334 HEXAFLUOROACET 166 422.67
The role of adaptation, learning and optimization are becoming HEXAFLUOROBUTA ONE
DIENE HEXAFLUOROPROP 150 19.8mg/l/h
increasingly essential and intertwined. The capability of a ENE
system to adapt either through modification of its physiological ACETYL FLUORIDE 62 2.50 mg/l/h
HYDROGEN 80.92 2860
structure or via some revalidation process of internal BROMIDE
mechanisms that directly dictate the response or behavior is AMMONIA GAS 17.03 4000 HYDROGEN 36.46 2810
CHLORIDE
crucial in many real world applications. In the past decade AMMONIA-15N 18.02 4000
HYDROGEN 20.01 1300
researcher are taken more interest in the application of multi FLUORIDE GAS
sensor arrays to gas classification and quantification. Most of AMMONIA-D3 20.06 4000

this work has been focused on systems using microelectronic HYDROGEN IODIDE, 127. 9 2860
ANTIMONY 124.8 20 ANHYDROUS
gas sensors featuring small size and low-cost fabrication, HYDRIDE
HYDROGEN 80.98 2
making them attractive for consumer applications. A number of ARSENIC 169.9 20 SELENIDE GAS
PENTAFLUORIDE
interesting applications have also emerged in the last decade, HYDROGEN 34.08 712
ARSINE 77.9 20
whether related to hazard detection, poisonous and dangerous SULFIDE

gases or to quality and environmental applications such as air BIS(TRIFLUOROMET 170 10 KETENE 42.04 16.1
HYL)PEROXIDE
quality control. We illustrate our approach by implementing
BORON ---------- 380
several designs taken from literature. The main goal of new TRIBROMIDE -- METHANETHIOL 48.11 1350
setup is to prevent lives in the situation where reaction time is
very less. The response of signal will send to control room via
BORON 117.2 2541 METHYL BROMIDE ----- 1007 sensor array is connected with encoder circuit and controller
TRICHLORIDE
unit. It is observed that metal–oxide gas sensors often
METHYL 80. 6 600
CHLOROSILANE respond to a wide range of gas species and are therefore
BORONTRIFLUORID 67.81 806
E NITROGEN 46 115
typically only partially selective. However, grain size in
BROMINE 115.4 290
DIOXIDE, the metal–oxide film does have an effect on the relative
(LIQUEFIED)
CHLORIDE resistivity of the film to different gas species. Grain size
NITROGEN DIOXIDE 46 0.22
BROMOMETHANE 94.94 811.14
mg/l/h
of a particular metal–oxide film is affected by the growth
temperature. Individual micro hotplate heaters can be
CARBON 28.01 3760
MONOXIDE NITROGEN DIOXIDE 46 116.93 used to control this temperature during the film growth
NITROGEN 30.01 115 process. An array of these micro-gas-sensors with
CARBON 28.01 3614
MONOXIDE,
MONOXIDE OR different film structures (e.g., different grain sizes) can be
NITRIC
LIQUEFIED used to achieve different response signatures for different
NITROGEN 92.02 115
TETROXIDE,
gases.
CARBON 30.03 3614 (LIQUEFIED)
MONOXIDE
Material Concentration range Temperature
NITROGEN 92.02 27.9
CARBON-13C 29.02 3760
TETROXIDE, GAS (degree
MONOXIDE centigrade)
NITROSYL 65.46 35
CARBONYL 82.46 ------ CHLORIDE
CHLOROFLUORIDE SnO2/ Pt, Au/Pt 1000-8000 ppm H2 200
NITROUS ACID, 75.07 320
CARBONYL 66.01 360 ETHYL
FLUORIDE
OXYGEN 54 2.6 SnO2/ZnO, WO3 12.5- 100 ppm 200
CARBONYL 60.07 2140 DIFLUORIDE chloroform, benzene
CHLORINE DIOXIDE 67.45 250 PERCHLORYL 102.5 770 In2O3, Pt, Pd 300
FLUORIDE

CHLORINE 67.45 ----------- PERFLUOROISOBUT ----- 1.2

MONOXIDE YLENE SnO2/ Pt, WO3 /Au, Up to 2000 ppm CH4 300
CHLORINE 92.45 299 ZnO
PHOSGENE, GAS 98.92 5
TRIFLUORIDE
CHLORINE, 70.91 293
PHOSPHINE 34 20
SnO2, SnO2/Pt, Nb2O5 100- 10000 ppm CO, NH3 450
(LIQUEFIED)

CHLORINE, GAS 70.91 293 PHOSPHOROUS 96 SnO2, SnO2,Al2O3 10- 20000 PPM CH4 350
OXYCHLORIDE
CHLOROTRIFLUORO 116. 5 2000 SnO2, SnO2/Pt, Pd 750 ppm C2H5OH, 1% 400
PHOSPHORUS 126 260
ETHYLENE
PENTAFLUORIDE CH4, 1% CO2
CYANOGEN 52.04 350
PHOSPHOROUS 208
TRICHLORIDE SnO2 1-4000 ppm CO 450
CYANOGEN 61.48 1.2 mg/l/h
CHLORIDE
SELENIUM 193 50
HEXAFLUORIDE 5-100 ppm NO 300
DEUTERIUM 81.93 300
BROMIDE
SILICON 104.1 450
DEUTERIUM 37.47 3120 TETRAFLUORIDE SnO2/Pt, SnO2/Cr, 50 – 200 ppm CO 250
CHLORIDE
STIBINE 20
DIAZOMETHANE 42.04 20 0.5 – 2 ppm NO2
SULFUR DIOXIDE 64.06 2520
DIBORANE 27.67 80
SULFUR 254. 1 10
DICHLOROACETYLE 94.92 45.6
PENTAFLUORIDE
NE
SULFUR 108.1 40 It is desirable to combine an array of unique gas sensor
DICHLOROSILANE 101 314
TETRAFLUORIDE elements (different material types or microstructures)
ETHYL NITRITE 75.07 320 SULFURYL 102. 1 3020 with integrated electronics that can address individual
FLUORIDE
elements and output a signal that can be analyzed to
ETHYLSILANE 60.17 ------ TELLURIUM 241. 6 25 determine gas classification. An efficient interface circuit
HEXAFLUORIDE
FLUORINE 38 185
is needed to measure the response of each array element.
TUNGSTEN 297. 9 217
HEXAFLUORIDE
To keep the power consumption of the gas sensor system
FORMALDEHYDE 30.03 0.66 low, an efficient heater power-switching scheme is also
GERMANIUM 76. 6 622 needed. Section III-A describes such an interface system
HYDRIDE or
GERMANE
developed in this work.

C. LATEST TECHNIQUES

B. MATERIAL USED IN GAS SENSORS A monolithic CMOS microhotplate-based conductance-type gas

Setup of gas sensor array with asynchronous controller is sensor system is described. A bulk micromachining technique is
shown in figure 1. It consist a set of different gas sensors which used to create suspended microhotplate structures that serve as
can detect different toxic gasses (shown in part II). Output of sensing film platforms. An array of microhotplate gas sensors
with different sensing film properties is fabricated by using a yield, refresh time and wafer-level reliability. This Cu wiring is
different temperature for each microhotplate during the LPCVD also going to improve RC Delay Constraints, resulting in the
film growth process. improvement of an enhancement in the performance of
Memory Transaction in MMU (Memory Management Unit) by
Sensors Sensitivity an sub45nm DRAM cell with reduced leakage current and SCE
(Short Channel Effect). In Cell fabrication, Hf (Hafnium) is
Integrated gas sensor array size Carbon mono oxide
proposed to use as High-K replacing conventional SiO2
2x2 Au/SnO2 200mW
(Silicon-di-oxide) gate dielectric and Also the poly-silicon gate
Low power micro-machined Hydrogen and ammonia 90 electrode is replaced with a metal gate electrode to enable
MOSFET gas sensor mW sufficient current to be driven through the transistor. Using these
technologies we have realized structural and electrical
Micro sensor array with microhot Butane (3000ppm) 100 advantage of our GPD (Gate Plated Device). To increase the
plate on Si mW speed of device we are using Cu (Copper) as interconnects in
DRAM cell which will improve the yield, refresh time and
SOI CMOS smart Gas sensor Carbon mono oxide and
wafer-level reliability. This Cu wiring is also going to improve
Methane 10 m W
RC Delay Constraints of the design. Also using this device
Ammonia SAW gas Sensor Ammonia (80 ppm) structure we can achieve Low Power Memory transaction at a
very High speed. Thus our design will also permit the lower
SIC based field effect gas sensor Hydrogen and granularity transaction at full bandwidth. The design of a Novel
hydrocarbons DRAM cell for the next generation High Speed Processors.
Since according to Aldahm’s prediction that Speed of
schottky diode sensor Hydrocarbons and oxygen
processors increases with an drastic ratio in comparison to their
decrease in die size or chip area, Since The speed of execution
Intelligent gas sensor (array of All organic compounds can
eight Semiconductor gas sensor) be identified of the processors are increasing Drastically , So for these high
speed processors, need of very high speed DRAM which are
having adequate Charge storing capacity also came into picture.
In addition to the Novel DRAM cell for the DRAM Operation
III. INTERFACE CIRCUITS Speeds. As when we are increasing the operation speed of our
design and also working on sub 45nm technology, then the
The minimum requirement for interface circuits is shown below major constraint which dominate the whole design is the Static
and Dynamic power consumption of our design. So for the
improvement of Power Consumption we are using a very novel
Technique at the Device level to design a transistor which will
be having a very low power consumption. For designing this
cell, which will going to give excellent performance in
improving Power Consumption along with suppressing the soft
error in P-well. Moving forward from the device level, to the
circuit level also we can apply the multi-VDD and multi-Vth
technique, which will not only reduce the leakage current
problem but also helps in improving the total power loss in the
DRAM design. To provide a multi-level power supply for VDD
and Vbias, we have included an independent level shifter circuit
in the design chip without effecting too much to the chip area.
Fig 2. Interfacing circuit Along with that for the full functioning of MMU we have
introduce an memory controller circuit which will monitor all
In this section, an interface circuit for automatic sensor
the handshaking of memory signals between DRAM and
selection and conductance measurement is described. The processing elements.
interface design has both analog and digital circuitry, as
shown in Fig.2. The analog circuit is comprised of an V. RESULTS
operational amplifier, and the digital circuitry has two
In this Smart Gas sensor for miniaturization of E- nose
decoders/multiplexers.
GPD processor is an important section. So we are showing here
some important result of GPD Processor. The leakage current
IV. PROCESSOR USING GPD
between trench to trench is expected to be 3.162X10 -17 Amps
The details of processor is discussed in this section. The design compare to the conventional cell which has 1.58X10 -11 Amps,
of Novel High Speed Transaction, we are using Cu (Copper) which gives us 99% leakage current Improvement in cell, which
wiring as interconnects in DRAM cell which will improve the will automatically improves the Power Consumption of our
Designed cell. Also we can achieve the Drain saturation current
of the order of 440µA at Vds of 1.8v which will be very good for
data access and the data stored in a memory can be read out
rapidly. Also we expect our sub-threshold current to be around
10-4A which can be small enough for transfer gate to keep
Charge stored cell capacitor from leaking.

The multi-Vdd multi-Vth technique which we have

applied in our design is expected to give us at least of around
55% improvement in the power Consumption since we have
decreasing the threshold voltage of the transistor.

The Copper wiring which we are using in our design,

can obtain the improved M1 line sheet resistance (~45%) and
via resistance (~60%) compared with Al-based wiring. Both the
reductions in k-value using FSG instead of SiO2 and the
decrease in metal height using low-resistivity Cu instead of Al
Fig 4. MOS HEIP CU wiring and Al one
alloy significantly decreased the intra-line capacitance. Intra-
line capacitances are expected to decrease by ~ 50 % and ~ 43
% for M1 and M2, respectively as compared with that of
conventional Al-based interconnection scheme. The electrical
performances like, TRCD and TRP are also assume to decreased
by ~ 3% and ~3.2 %, respectively, where TRCD denotes the
delay between row address strobe signal and column address
strobe signal and TRP, how fast bit line level reaches its
predetermined value. Further improvement for operation speeds
is expected by optimization of circuit design.

Fig 5. Que Depth Efficiency for Single Thread and Dual

Thread Examples

Dual-threading has the same total number of column

accesses as single-threading, but only one-half as many devices
are accessed for each row transaction. This reduces the total
power to 75% to 87.5% that of a single-threaded module,
similarly In quad-threading has the same number of total column
accesses as single-threading, but only one-quarter as many row
accesses. This reduces the total power to 62.5% to 81.3% that of
a single-threaded module.
Fig 3. Comparison of Qbd between Cu and Al
V. CONCLUSION
By the use of multi-threading technique in our
design we are expecting the multi threading reduces the total
Based on the general technique available, Both the
power for each transaction. Typically, the module power
Low power Consumption and the High Speed is very hard to
required for row accesses (the ACTIVE command) accounts for
even think about at submicron technology, But Here we have
25% to 50% of the total power, with the rest consumed by the
presented our novel techniques to reduce Power consumption
column operation (the READ command).
and to achieve Very high data transfer speed, all from the
Device level to circuit level to system level. Also after applying
all these techniques, a rough estimation of results are done
which shows a very excellent and positive results.

So we conclude that with the help of these novel

techniques we can achieve the total power reduction of around
35- 40% of the general design and an improvement in data
retention time is expected to be around 5-7% of the general
design. The future work of this design can be think of the use of
transistors made up of carbon nanotubes which will have its Syed A Imam was born in
own advantages at different levels. Bihar, India on Jan-4, 1961. He
received the M. Sc. Engg
REFERENCES degree from AMU, Aligarh
and PhD. degree in Electronics
& Comm. Engg
[1] Stanford TGO Data sheet, Environment health and safety, 723-
0448.
[2] D. Gajski, W. Dutt, A. Wu and S.Lin, “High level synthesis:
introduction to chip and system design”, Kluwer Academic from Jamia Millia Islamia Central University, New
Publisher, 1992 Delhi, in 1998, and 2008, respectively.
[3] K. T. Lau, S. Edwards, and D. Diamond, “Solid-state ammonia
sensor based on Berthelot’s reaction,” Sens. Actuators, B, Chem.,
vol. 98, no. 1, 12–17, 2004 Since 1990, he has been part of Jamia Millia Islamia
[4] L. Byrne, K. T. Lau, and D. Diamond, “Monitoring of headspace University, where he is Assistant Professor in the
total volatile basic nitrogen from selected fish species using Department of Electronics and Communication
reflectance spec-troscopic measurements of pH sensitive films,” Engineering.
Analyst, vol. 127, no. 10, 1338–1341, Oct. 2002.
[4] S. Tan, M. Devoret, H. Dai, A. Thess, R. Smalley, L. Geerligs, and His current research interests are in the field of
C.Dekker, “Individual single-wall carbon nanotubes as quantum
sensing technologies, electronic and biosensors, signal
wires,” Nature, vol. 386, pp. 474–477, 1997.
[5] Z. Yao, H. Postma, L. Balents, and C. Dekker, “Carbon nanotube processing and digital circuits.
intramolecular junctions,” Nature, vol. 402, pp. 273–276, 1999.
[6] R. H. Baughman, C. Cui, A. A. Zakhidov, Z. Iqbal, J. N. Barasci,
G.M. Spinks, G. G. Wallace, A. Mazzoldi, D. de Rossi, A. G.
Rinzler, O. Jaschinski, S. Roth, and M. Kertesz, “Carbon nanotube
actuators,”Science, vol. 284, pp. 1340–1342, 1999
[7] C. Marliere, P. Poncharal, L. Vaccarini, and A. Zahab, “Effect of
gas adsorption on the electrical properties of single-walled carbon
nanotubes mats,” in Mat. Res. Soc. Symp. Proc., vol. 539, 2000,
pp. 173–178.
[8] K. Lau, S. Baldwin, R. Shepherd, P. Dietz, W. Yerazunis, and D.
Diamond, “Novel fused-LEDs devices as optical sensors for
colorimetric analysis,” Talanta, vol. 63, no. 1, pp. 167–173, May
2004.
[9] Fukasawa et al., “BEOL process integration with Cu/SiCOH (k=2.8)
low-k interconnects at 65 nm groundrules” Proc. IEEE IITC, pp. 9-
11 (2005).
*10+J. Chen et al., “The Enhancement of Gate-Induced-Drain-Leakage
(GIDL) Current in Short Channel SOI MOSFET and its Application in
Measuring LateralBipolar Current Gain b”, IEEE Electron Device
Lett., Vol. 13, No. 11, pp. 572-574, November 1992.
*11+M. Lemme, et al., “Influence of Channel Width on nand p-type
Nano-Wire-MOSFETs on Silicon on Insulator Substrate”,
Microelectron. Eng., in press.

Bibliography:

Rajesh Kumar received BE in

Electronics & Communication
Engineering from Bangalore
University in 2001 and M.Sc.
(ENGG) in Control System form
National Institute Of

Technology Patna, Bihar (India). He taught at RPS

Institute Of Technology, Patna, Bihar (India) From Dec
2002 To Feb 2007. Since March 2007, He Is With The
Department Of ECE, Ajay Kumar Garg Engineering
College, Ghaziabad, Up (India). His research interest is
in the area of Sensing Technology. He is Currently
Pursuing towards PhD Degree under Faculty of
Engineering and Technology, Deptt. Of Electronics
And Communication, Jamia Millia Islamia, New Delhi,
India.