Sensors and Actuators B 135 (2009) 656–663

Contents lists available at ScienceDirect

Sensors and Actuators B: Chemical

journa l homepage: www.e lsev ier .com/ locate /snb

Porous single-wall carbon nanotube films formed by in Situ arc-dischargedeposition for gas sensors application

Nguyen Duc Hoa, Nguyen Van Quy, Yousuk Cho, Dojin Kim ∗

Department of Materials Science and Engineering, Chungnam National University, Daejeon, Republic of Korea

a r t i c l e i n f o

Article history:Received 2 July 2008Received in revised form 16 October 2008Accepted 21 October 2008Available online 5 November 2008

Keywords:

a b s t r a c t

A gas sensor structure for high performance is developed based on a porous film structure of single-wallcarbon nanotubes (SWNTs). The SWNTs are directly deposited on the substrate in situ in the arc-dischargechamber, and the steric hindrance during the deposition permits high porosity in the carbon nanotubethin film. The SWNTs are characterized and their capacities of sensing NO and NH3 gases are investigateddown to 2 ppm-level concentration at room and elevated temperatures. The effect of sensor porosity isexamined and compared with that of a nonporous sensor structure. The response and recovery behaviorsare examined with temperature variations.

SWNTsIn-situ depositionGas sensorPM

© 2008 Elsevier B.V. All rights reserved.

1

ooarieohciems

vmwfptt

pplcd

aTtooipfibcivtc

0d

orousethanol treatment

. Introduction

Many of the gas sensors that are made of semiconductingxide materials such as SnO2, TiO2, In2O3, and ZnO are basedn the changes in electrical resistance of the materials upon gasdsorption [1,2]. In the sensor structures, the principle naturallyequires a larger surface-area-to-volume ratio for high sensitiv-ty; for this matter, thin films and porous thick films have beenxtensively studied. Recently, nanostructures, such as carbon nan-tubes (CNTs), with extremely high surface-area-to-volume ratios,ave begun attracting wide attention in the study of their appli-ation to various sensors. So far, CNT-based gas sensors have beennvestigated for the detection of H2, N2, NO2, and NH3, among oth-rs [3–13]. Both single-wall carbon nanotubes (SWNTs) [3–7] andultiwall carbon nanotubes (MWNTs) [8–13] have been tested as

ensing probes.In principle, a single wire of CNT can form a sensor and pro-

ide important intrinsic sensing properties when used as a sensingaterial. However, until the technology to regularly array the singleires is developed, it is an inappropriate unit of engineering device

or mass production. Therefore, the realistic engineering sensor atresent should be made of a thin film form of CNTs. In the filmype CNT sensors, not only the intrinsic CNT properties, but alsohe extrinsic conditions can affect final sensor performance. Exam-

∗ Corresponding author. Tel.: +82 42 821 6639; fax: +82 42 823 4224.E-mail address: dojin@cnu.ac.kr (D. Kim).

OdstmC

o

925-4005/$ – see front matter © 2008 Elsevier B.V. All rights reserved.oi:10.1016/j.snb.2008.10.041

les of these external conditions are film thickness, density, andorosity in relation to gas permeability, the CNT alignment (paral-

el and vertical), and entanglement in relation to the CNT-to-CNTontacts that determine the current paths, impurity materials andistribution, among others.

If we review the thin film type CNT gas sensor structures, theyre mostly either pasted CNTs [5,6] or CVD-grown CNTs [11,13–15].he method of pasting CNT is a direct and easy way of film forma-ion, but requires sequential processing of synthesis and collectionf CNTs, as well as dispersion in solutions or polymer, and printingn substrates [4,16]. In addition to the above-mentioned complex-ty of the process, controlling the thinness of the CNTs for bettererformance of the sensor device is a great challenge. When thelms are thick, the sensitivity of the sensor decreases because theuried part in the film will not act as the probe. This will also be thease for the composite blended with polymer. On the other hand,n the self-assembled CNT film case, most of the MWNTs are grownertically by the CVD method using catalyst metals [11,13,14] andhe electrodes are patterned parallel to the plane. Therefore, theurrent paths are formed through contacts among the vertical CNTs.ne can easily recognize that this geometry does not efficientlyeliver the signal changes to the electrodes. The efficiency and theensitivity will be improved if the CNTs are grown laterally between

he electrodes [15]. For efficiency in mass production, however, this

ethod should be developed to control the areal density of theNTs.

Therefore, in this study, the authors report the developmentf a CNT sensor structure with a potential for ultimate sensitiv-

Actuators B 135 (2009) 656–663 657

iSirsttg

2

sdgFccTa1dtiblbHtr

oWTtfaehCottiom

eeetessFttcfitgtp

N.D. Hoa et al. / Sensors and

ty. The key feature of the thin film sensor is the porosity in theWNT film. It is performed by direct in situ deposition of SWNTsn the arc-discharge chamber on the patterned electrodes. The fab-ication method is very simple and skips the complex processingteps of manipulating CNTs. Furthermore, it is a method extendableo mass production techniques. The basic gas sensing proper-ies of the fabricated sensor devices are tested on NH3 and NOases.

. Experimental

A silver thin film was deposited on thermally grown SiO2ubstrate using an AC sputter and patterned into a comb-type inter-igitated electrode. The measurements of the finger width and theap between the fingers were all at 100 �m, as shown in the inset inig. 1(a). The substrate was mounted on the wall of an arc-dischargehamber for in situ SWNT deposition. A graphite rod inserted withatalyst wires made of Ni, Fe, and Mo was used as the carbon source.he CNT synthesis was carried out in an optimized condition ofrcing current density 85 A/cm2 in 400 Torr of H2 atmosphere formin. The synthesized SWNTs directly sit on the substrate whilerifting in air to the wall. Once thick deposits of SWNTs developed,he morphology displayed a thicket of entangled tubes with poros-ty. The high porosity of the CNT films deposited in situ is causedy the steric hindrance created while the CNT bundles seat layer byayer on the substrate. The response of these films is not expected toe sensitive to film thickness as long as the porosity is maintained.owever, if the film condenses to compaction, only the top-laid

ubes exposed outside will react with the gases and the electricalesponse of the device is expected to decrease.

Two types of SWNT sensor probes were tested. One is made upf as-deposited SWNTs and the other, of methanol-treated SWNTs.e will refer to the sensors as ‘AD’ and ‘MT’-sensor, respectively.

he latter was made by dipping the AD-sensor in methanol solu-ion for 2 min, after which it was dried up by hot plating at 100 ◦Cor 20 min. This treatment was originally intended to enhance thedhesion between the SWNTs and SiO2 substrate, and was meant tostablish a mechanically stable device structure. At the same time,owever, the pores are removed by the methanol treatment and theNTs pack densely. As a consequence, only the CNT tubes exposedn the surface of the film will react to the gases flowing upon it. Inhe morphological condition of Fig. 1, the SWNT films were madehin so that the gases can freely react to them even after the pack-ng brought about by the methanol treatment. Therefore, the effectf the possible chemical modification of the tube surfaces by theethanol treatment can be detected.The morphology of the SWNTs was examined using field-

mission scanning electron microscopy (FE-SEM) and transmissionlectron microscopy (TEM), while their chemical structures werexamined using thermogravimetric analysis (TGA), X-ray pho-oelectron spectroscopy (XPS), and Raman spectroscopy. Thelectrical and gas-sensing properties of the sensors were mea-ured in a stainless steel chamber with a volume of 1500 cm3. Thechematic diagram of the sensor measurement system is shown inig. 2. The system is connected with several gas cylinders, but onlyhe results for NO and NH3 sensing are reported in this study. Theest gases were NH3 and NO diluted with N2 to 1000 ppm in moleoncentration. High-purity nitrogen was used as the carrier gas forurther control of the gas concentrations. The regulated test gases

n the range of 2–60 ppm concentrations were made to flow into theest chamber using mass flow controllers. The total flow rate of theases was kept constant at 500 sccm. The change in sensor resis-ances as response and recovery behavior was measured using arogrammable electrometer (Keithley 2400). The sensor response,

Fig. 1. FE-SEM images of (a) the as-synthesized SWNTs and (b) methanol-treatedSWNTs, (c) HR-TEM image of SWNTs.

658 N.D. Hoa et al. / Sensors and Actuators B 135 (2009) 656–663

S

S

wt

3

3

sssspbgctso[trStSanϕMpt

mT(wot

Ft

cmobafdctts

ausTSbaTgvTtp

Fig. 2. The schematic diagram of the sensing measurement system.

, is defined as:

(%) = 100(R − Ro)Ro

for NH3 and S (%) = 100(Ro − R)R

for NO

here Ro is the resistance of the sensor with N2 and R is the resis-ance of the sensor under the test gases (NH3 or NO).

. Results and discussion

.1. The SWNT sensor structure

The SEM and TEM images of the fabricated SWNT-based sen-ors, both the AD- and MT-sensors, are illustrated in Fig. 1. It can beeen that the SWNT bundles are entangled with each other, and theubstrate surface can be seen by indicating the thinness of the depo-ition (Fig. 1(a)). The defocusing in the photo for the AD-sensor isartly due to the floating of the bundles above the substrate causedy the steric hindrance. The essential consequence of the entan-lement is the formation of a network of SWNTs or provision ofonducting channels. One can see in Fig. 1(b) that the methanolreatment makes the floating tubes collapse and stick on the sub-trate. We showed elsewhere that the collapse of porous SWNTsriginates from the hydrophobic nature of the tubes in methanol17]. The net consequence is more frequent contacts among theubes, which provide more conducting paths within SWNTs for cur-ent flow. The surface density of CNTs roughly estimated from theEM image of the MT-sensor was ∼1011 bundles/cm2. The apparenthickness of the AD- and MT- sensors measured by the cross-sectionEM images was ∼2 �m and ∼100 nm, respectively. Therefore, thes-deposited porous CNT films contracted to 1/20 of the origi-al volume via methanol treatment. If one defined the porosity(%) = 100(to − t)/to with to and t as the thickness of the AD- andT-sensors, respectively, and if the CNT film in the MT-sensor is

erfectly compact, the porosity of the AD-sensor can be estimatedo be ∼95%.

In the as-synthesized SWNTs, amorphous carbon and catalystetal particles are found to be the major impurities as examined by

EM (Fig. 1(c)), and as confirmed by TGA and Raman spectroscopyFig. 3). The TGA was carried out in dry air (flowing rate of 50 sccm)ith increasing temperature to 900 ◦C (temperature increment rate

f 5 ◦C/min). The amorphous carbon covering the SWNTs as well ashe crystalline defects in SWNTs can act as gas adsorption sites, and

aibtr

ig. 3. (a) TGA graph of the powder of as-synthesized SWNTs and (b) Raman spec-roscopy of AD- and MT-sensors.

an consequently affect the response of the sensors [18]. In the TGAeasurement shown in Fig. 3(a), the first weight lose of ∼4 wt.%

ccurring at ∼295 ◦C is ascribed to the evaporation of water. Theurning out of amorphous carbon then occurred, reaching ∼405 ◦Cnd the amount was ∼10 wt.%. When the temperature increasedurther, the carbon nanotubes began to burn out and the weightecrease stopped at ∼700 ◦C. The remaining weight of ∼39 wt.%orresponds to the metallic oxides from the catalyst particles, andhe amount reached 6–7 vol. % in the synthesis body. This high con-ent of catalyst and water is the typical characteristic of the SWNTsynthesized in hydrogen ambient [17].

Raman spectroscopy is a simple and useful technique to char-cterize the quality of the carbon nanotubes. The results examinedsing an Ar-ion laser with a wavelength of 514.532 nm for the as-ynthesized and methanol-treated SWNTs are shown in Fig. 3(b).here were no remarkable differences in the crystal structure of theWNTs after the methanol modification. The Raman spectra of car-on nanotubes typically consist of two major peaks at ∼1340 cm−1

nd ∼1590 cm−1, or the so-called D- and G-bands, respectively [19].he D-band is known to be associated with defective, disorderedraphite or glassy carbon, while the G-band is related to the sp2

ibration of a two-dimensional hexagonal lattice in the graphite.he very high G-band with respect to the D-band is a typical fea-ure of the SWNTs having high crystalline quality. The small D-bandeak indicates that the film contains small but finite amount of

morphous carbon (and defects) as confirmed also by the HRTEMmage in Fig. 1(c). The SWNTs show the vibration peaks of the radialreath mode (RBM) in the range of 100–300 cm−1, which are relatedo the diameters of the tubes. The RBM peaks shown in Fig. 3(b-left)eveal the tube diameters to be at ∼1.4 nm.

N.D. Hoa et al. / Sensors and Actuators B 135 (2009) 656–663 659

Fh

mtCMSoCbt2oarcoa

3

sAwdgtro

Fai

kfS∼tomo

tai(bbs[iAtaito

ig. 4. (a) XPS survey scan of as-synthesized and methanol-treated SWNTs and (b)igh resolution scan of C1s peaks with Gaussian fits.

The XPS investigation demonstrates the state of bonding in theaterial. In the XPS spectra of the as-synthesized and methanol-

reated SWNTs shown in Fig. 4(a), the major peaks correspond toand O, in addition to the small peaks for catalyst particles of Fe,o, and Ni. The C1s peak comes from the carbon networks in the

WNTs, but the high oxygen peak suggests a strong tendency forxygen adsorption on the SWNTs. The high magnification of the1s peak shown in Fig. 4(b) reveals shoulder peaks derived frominding with oxygen. The major peak at 284.5 eV is assigned tohe sp2 carbon–carbon bonding. The other small peaks at about85.5, 287.6, and 288.6 eV are assigned to carbons attached toxygen, or C–OH, C O, and O–C O groups, respectively [20]. Themount of carbon atoms chemically bound to oxygen containingadicals is small, the majority of which is molecular oxygen physi-ally adsorbed on the SWNT surfaces. We thus found that adsorbedxygen determines the apparent conductance of the SWNT films inir as discussed in the following section.

.2. The electrical and gas sensing properties

The electrical resistances of the two types of sensors were mea-ured with temperature variation at a high vacuum of ∼10−6 Torr.

sensor was mounted on a chuck in vacuum and its resistanceas monitored while it was heated up to 400 ◦C and then cooled

own to room temperature. The results are shown in Fig. 5; theeneral trends were the same in both types of sensors except forhe huge difference in the absolute resistance values. One needs toecall that many CNTs reveal p-type semiconducting behavior withxygen adsorption [21]. The SWNT film sensors are those usually

(atSb

ig. 5. Dependence of the sensor resistance on temperature for (a) as-synthesizednd (b) methanol-treated SWNTs. The dependence was measured firstly withncreasing temperature and successively with decreasing temperature in vacuum.

ept in air so that nitrogen and oxygen are adsorbed on the sur-ace. Jhi et al. [22] calculated that oxygen is adsorbed on the wall ofWNT with a binding energy of ∼0.25 eV and a binding distance of0.27 nm. About 0.1 electron transfers from the carbon nanotube

o an oxygen molecule adsorbed on the tube wall, or the adsorbedxygen molecules supply hole carriers to the carbon nanotube. Thisay be the origin of the p-type semiconducting behaviors generally

bserved in SWNTs.Both AD- and MT-sensors displayed ‘V’-shaped changes in resis-

ance when measured with temperature increase (closed symbols)s shown in Fig. 5. However, they exhibited typical semiconduct-ng behaviors when remeasured while cooling down from 400 ◦Copen symbols). Such difference in the temperature dependenceefore and after annealing was also observed by Hone et al. [23],ut with lower transition temperature. Annealing in air also showedimilar V-shaped changes with transition temperature of ∼100 ◦C5], where the V shape was implicitly interpreted by a metal-nsulator transition [23]. However, it is not the case in Fig. 5.ctually when the SWNTs are heated up to 400 ◦C in vacuum,he majority of the physisorbed molecules, particularly of water,re desorbed and the intrinsic properties of the SWNTs could bellustrated by the semiconducting temperature-dependent resis-ivity curve (open symbols). The behavior can be attributed to thexygen-chemisorbed SWNTs as determined by XPS measurementalmost identical to Fig. 4(b)). Raman spectroscopy measurement

lso revealed no remarkable difference before and after the heatreatments. Therefore, the V-shaped R-T curves in the non-annealedWNT structures do not come from the intrinsic change in the car-on nanotube properties.

6 Actuators B 135 (2009) 656–663

bcodtoamittabdratbr

tarmtTtnothtmtitisriwhoi

eogCdicprttN

ftirei

Fg

hfptdecsN2dttT

t1Ketoi�t

60 N.D. Hoa et al. / Sensors and

Therefore, the apparent V shape of the changes (closed sym-ols) could be observed from the competition between the twoounteracting effects; one is the increase in resistance with des-rption of the physisorbed water molecules, and the other is theecrease in resistance from the semiconducting behavior as theemperature increases. The SWNT sensor film initially mountedn the chuck contains moisture on the surfaces, thus displayingn enhanced conductance. As the sensor is heated up, the waterolecules desorb and the sensor resistance increases to that of

ntrinsic SWNTs. On the other hand, the temperature increase tendso reduce the resistance owing to the semiconducting nature ofhe SWNTs. If the latter effect is dominant, a decreasing R willppear at <200 ◦C. However, as the molecular desorption effectecomes dominant and/or the decrease of R from the semicon-ucting behavior is small at >200 ◦C, it shows a slight increase inesistance. Note that the change in R of the CNTs is small at >200 ◦Cs illustrated by the open triangles. This observation suggests thathe surface condition of the CNT as the sensing probe needs toe carefully specified for proper interpretation of the gas-sensingesults.

Another remarkable feature is the huge difference in the resis-ance between the two sensors. At 50 ◦C, the resistance of the AD-nd MT-sensors measured in vacuum is ∼25,000 and ∼310 ohm,espectively, as denoted in Fig. 5. The sample-to-sample unifor-ity in the as-deposited state was good since the resistance of

he MT-sensor before the methanol treatment was ∼26,000 ohm.herefore, the dramatic decrease in resistance to ∼310 ohm afterhe methanol treatment in the MT-sensor (with nearly the sameumbers of CNTs) can be attributed to (i) the surface modificationf the SWNTs by methanol, and/or (ii) the enhanced contacts amonghe tubes. The XPS peaks for carbon (C1s) and Raman spectroscopy,owever, did not show noticeable differences after the methanolreatment. Furthermore, heating up to 400 ◦C would cause the

olecules on the surface of the SWNTs to evaporate. Therefore,he decrease in resistance in the MT-sensor is mainly due to thencreased number of the contacts among the SWNTs. The resis-ance difference between the sensors at 50 ◦C after the annealings more dramatic (∼600,000 and 800 ohm, respectively, in Fig. 5),howing about 750 times difference. This suggests that as many cur-ent paths were newly established through the methanol treatmentn the MT-sensor. These paths were created by newly formed net-

orks or contacts when the nanotubes collapsed. Conversely, thisuge difference in resistance is indicative of the very high porosityf the as-deposited SWNTs when the in situ arc-discharge techniques used.

The ratio in electrical resistance R/Ro in the AD-sensor withxposure to 40 ppm of NH3 and NO gases as the reducing andxidizing species at room temperature is shown in Fig. 6(a). Nitro-en was used as the reference gas. It is a well-known fact that theNT resistance of the sensor increases with exposure to NH3 andecreases with exposure to NO [3,5,6,13]. The polar NH3 molecule

s known as an active electron dopant and transfers electrons toarbon nanotubes [24]. The electrons neutralize the holes in the-type SWNTs resulting in an increase in resistivity. Therefore, theesistance of the SWNT sensor increases upon exposure to NH3. Onhe other hand, NO is an acceptor dopant to the p-type SWNTs, andhus the resistance of the SWNT sensor decreases upon exposure toO gas.

The sensor response estimated from Fig. 6(a) was 17% and 46%or the same concentration of 40 ppm of NH3 and NO, respec-

ively, at room temperature. The sensor responses with variationsn the gas concentrations at a range of 2–40 ppm are summa-ized in Fig. 6(b). For the discussion, we consider the dynamicquilibrium of the gas adsorption and desorption during the sens-ng measurements. When a homogeneously diluted gas (say, NH3

cEetU

ig. 6. Sensing behaviors of AD-sensor. (a) The responses to 40 ppm NO and NH3

ases and (b) sensor response change with gas concentration.

omogeneously diluted in N2) flows steadily to hit the CNT sur-ace, the hitting probability of NH3 and N2 on the CNT surface isroportional to the respective partial pressure of NH3 and N2 inhe flowing gas. When the number of sites for NH3-adsorption isetermined only by the statistical hits on the CNT surface, the lin-ar sensitivity change with NH3 concentration as shown in Fig. 6(b)an be explained. In the investigated concentration range, the sen-or response for NO was about three times higher than that forH3. However, the sensor response for NO begins to saturate above0 ppm while it still increases linearly up to 40 ppm for NH3. Suchifferences in response behaviors between the gases can be relatedo the differences in (i) the adsorption efficiency, (ii) the electronransfer efficiency to the CNTs, and/or (iii) the binding energies.his needs to be explored in a separate study.

Note also that the recovery behaviors are very poor at roomemperature for both gases in Fig. 6(a). Recovery time longer than2 h for the NOx sensors based on CNTs has been reported byong et al. [3]. Such a long recovery is due to a strong bondingnergy between NOx molecules and CNTs sites [7]. The recoveryime of gas desorption can be expressed by the Arrhenius formf � = �0

−1exp(−EB/kBT) [7], where EB is the adsorption energy, Ts the absolute temperature, kB is the Boltzmann’s constant, and0 is the attempt frequency. The theoretical calculation of adsorp-ion energy between NO2 and SWNTs is about 0.6 eV and does notlearly depend on the diameter and chirality of SWNTs [25]. With

B, kB, and �0 as the constants for a given gas specie, the recov-ry time will decrease at high temperatures. Recently, the recoveryime of the NOx sensor based on CNTs could also be decreased byV illumination [26–29].

N.D. Hoa et al. / Sensors and Actuators B 135 (2009) 656–663 661

Fsd3

tFtrtpoomtt

Fgc

os

thaaf

ig. 7. Sensing behaviors of AD-sensor. (a) The change in electrical resistance ratio ofensor upon exposure of 60 ppm NO at different temperature, (b) the temperatureependence of sensor response, and (c) two cycle recovery of sensor operated at00 ◦C.

In order to investigate the recovery behavior of the sensor, theemperature of the AD-sensor was varied from 30 ◦C to 350 ◦C.ig. 7(a) shows the change in the ratio of R/Ro upon exposureo 60 ppm NO at different temperatures. One observes that theesistance at recovery approaches the stand-by value at increasedemperatures. At 300 ◦C, the sensor resistance fully recovered byurging the chamber of N2. Thus, the temperature dependence

f recovery time is roughly shown. However, we simultaneouslybserved the sensor response decreasing with temperature as sum-arized in Fig. 7(b). For example, the sensor response at room

emperature was 90%, but it decreased to 15% at 350 ◦C. Therefore,he response and recovery are the trade-off terms for carbon nan-

gc(ci

ig. 8. Sensing behaviors of MT-sensor. (a) The responses to 40 ppm NO and NH3

ases, (b) the sensor response change with temperature to 40 ppm NO and (c) threeycle’s recovery of sensor operated at 300 ◦C.

tube sensors. A good reproducible response and recovery of theensor observed at 300 ◦C is illustrated in Fig. 7(c).

Fig. 8(a) shows the R/Ro ratio for the MT-sensor upon exposureo NO and NH3 at room temperature. The MT-sensor also showedigher sensor response for NO than for NH3 as the AD-sensor. Theyre about 5% and 0.4%, respectively, at 40 ppm concentration of NOnd NH3. Note that the absolute sensor responses are lowered byactors of about 10–40 with respect to the AD-sensor. This sug-ests that the sensor of high resistance (AD-sensor of ∼25,000 ohm)

an exhibit higher sensor response than that of lower resistanceMT-sensor of ∼310 ohm). It is because the relative effect of �Raused by the gas adsorption and consequent electrons transfers higher for the larger Ro sensor. The comparison clearly indi-

6 Actua

chiww

wtttoibfmwTMgiwr

cCcomvlmnat

4

sddtflatt(pi

A

N

R

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

B

NHMed

62 N.D. Hoa et al. / Sensors and

ates that CNT sensors of high resistance of porous structure canave high sensor response and it can be a figure-of-the-merit

n the resistor-type gas sensors. Therefore, semiconducting CNTsould show better sensing performance than metallic ones asell.

The sensor response change with temperature in the MT-sensoras also measured for 40 ppm NO gas as shown in Fig. 8(b), where

he sensor response decreased with temperature. The trend is dueo a reduced adsorption rate of the gas molecules with tempera-ure as was also shown in the AD-sensor. A three-cycle recoveryf the MT-sensor measured at 300 ◦C with 40 ppm NO is shownn Fig. 8(c). The response and recovery times (�) were estimatedy fitting the transitional resistance curves with the exponentialunction R ∼ exp(−t/�). The response and recovery of the AD-sensor

easured with NO at 300 ◦C were 0.76 and 7.5 min, respectively,hile they were 0.3 and 1.39 min, respectively, for the MT-sensor.

herefore the response and recovery behaviors are better with theT-sensor. The comparison suggests that the reaction with the

as is limited to the superficial CNTs in the MT-sensor resultingn fast response and recovery, while the longer characteristic times

ith the AD-sensor is due to diffusion through the porous CNTs foreaction.

Overall, the porous SWNT film sensor is a practical sensor fabri-ation technology based on present technologies. It offers a simpleNT synthesis method that does not require any further compli-ated synthesis technology development like those in single-CNTr aligned-CNT sensors. It also well approaches the low cost ofass production technology. It is a many-wire structure that pro-

ides high reliability and high enough currents detectable at aow cost in the sensor electronics. As far as the sensing perfor-

ance is concerned, the single-wire type will present the intrinsicature of the material as a sensor, but the film type may approachsimilar performance through the high porosity of the struc-

ure.

. Conclusions

We introduce a simple method to fabricate carbon nanotube gasensors with high performance. The key feature of the method is theirect deposit of SWNTs on patterned electrodes in situ in an arc-ischarge chamber. The steric hindrance among the tubes duringhe deposition guarantees high porosity in the CNT film so that freeow of, and contact to, the gases are possible. The SWNT sensorsre expected to detect NO and NH3 below ppm-level concentra-ion in atmospheric pressure. The SWNTs show higher sensitivityo NO than to NH3. Although the response time is relatively fastwithin a few minutes) at room temperature, the recoveries are veryoor. Temperature increase greatly improves the recovery behav-

ors.

cknowledgment

This research was conduced with the financial support of theational Research Laboratory program, Korea.

eferences

[1] I.D. Kim, A. Rothschild, B.H. Lee, D.Y. Kim, S.M. Jo, H.L. Tuller, Ultrasensitivechemiresistors based on electrospun TiO2 nanofibers, Nano Lett. 6 (2006)2009–2013.

[2] E. Comini, Metal oxide nano-crystals for gas sensing, Anal. Chim. Acta 568

(2006) 28–40.

[3] J. Kong, N.R. Franklin, C. Zhou, M.G. Chapline, S. Peng, K. Cho, H. Dai,Nanotube molecular wires as chemical sensors, Science 287 (2000) 622–625.

[4] I. Sayago, E. Terrado, M. Aleixandre, M.C. Horrillo, M.J. Fernández, J. Lozano, E.Lafuente, W.K. Maser, A.M. Benito, M.T. Martinez, J.J. Gutiérrez, E. Munoz, Novel

Narie

tors B 135 (2009) 656–663

selective sensors based on carbon nanotube films for hydrogen detection, Sens.Actuators B 122 (2007) 75–80.

[5] N.H. Quang, M.V. Trinh, B.H. Lee, J.S. Huh, Effect of NH3 gas on the electricalproperties of single-walled carbon nanotube bundles, Sens. Actuators B 113(2006) 341–346.

[6] E. Bekyarova, M. Davis, T. Burch, M.E. Itkis, B. Zhao, S. Sunshine, R.C. Haddon,Chemically functionalized single-walled carbon nanotubes for ammonia sen-sors, J. Phys. Chem. B 108 (2004) 19717–19720.

[7] S. Peng, K. Cho, P. Qi, H. Dai, A-b initio study of CNT NO2 gas sensor, Chem. Phys.Lett. 387 (2004) 271–276.

[8] M.K. Kumar, S. Ramaprabhu, Nanostructured Pt functionlized multiwalled car-bon nanotube based hydrogen sensor, J. Phys. Chem. B 110 (2006) 11291–11298.

[9] C.S. Huang, B.R. Huang, Y.H. Jang, M.S. Tsai, C.Y. Yeh, Three-terminal CNTs gasfor N2 detection, Diamond Relat. Mater. 14 (2005) 1872–1875.

10] J. Suehiro, H. Imakiire, S. Hidaka, W. Ding, G. Zhou, K. Imasaka, M. Hara,Schottky-type response of carbon nanotube NO2 gas sensor fabricated ontoaluminum electrodes by dielectrophoresis, Sens. Actuators B 114 (2006) 943–949.

11] W.S. Cho, S.I. Moon, Y.D. Lee, Y.H. Lee, J.H. Park, B.K. Ju, Multiwall carbon nan-otube gas sensor fabricated using thermomechanical structure, IEEE Elect. Dev.Lett. 7 (2005) 450–498.

12] Y.H. Li, Y.M. Zhao, Y.Q. Zhu, J. Rodriguez, J.R. Morante, E. Mendoza, C.H.P. Poa,S.R.P. Silva, Mechanical and NH3 sensing properties of long multi-walled carbonnanotube ropes, Carbon 44 (2006) 1821–1825.

13] S.G. Wang, Q. Zhang, D.J. Yang, P.J. Sellin, G.F. Zhong, Multi-walled carbonnanotube-based gas sensors for NH3 detection, Diamond Relat. Mater. 13 (2004)1327–1332.

14] L. Valentini, C. Cantalini, I. Armentano, J.M. Kenny, L. Lozzi, S. Santucci, Highlysensitive and selective sensors based on carbon nanotubes thin films for molec-ular detection, Diamond Relat. Mater. 13 (2004) 1301–1305.

15] Y.T. Jang, S.I. Moon, J.H. Ahn, Y.H. Lee, B.K. Ju, A simple approach in fabricatingchemical sensor using laterally grown multi-walled carbon nanotubes, Sens.Actuators B 99 (2004) 118–122.

16] S.I. Moon, K.K. Park, Y.H. Lee, J.K. Kim, S.W. Kim, B.K. Ju, Electrochem. Sol. St.Lett. 9 (2006) H78–H80.

17] H.J. Song, Y.S. Cho, M.C. An, Y.J. Kang, D.J. Kim, A simple approach to the fabri-cation of transparent single-wall carbon nanotubes films of high electrical andoptical performance, J. Kor. Phys. Soc. 53 (2008) 2111–2114.

18] J. Andzelm, N. Govind, A. Maiti, Nanotube-based gas sensors–role of structuraldefects, Chem. Phys. Lett. 421 (2006) 58–62.

19] S. Lefrant, I. Baltog, M. Baibarac, Surface-enhanced Raman scattering studies onchemically transformed carbon nanotube thin films, J. Raman Spect. 36 (2005)676–698.

20] T.I.T. Okpalugo, P. Papakonstantinou, H. Murphy, J. McLacughlin, N.M.D. Brown,High resolution XPS characterization of chemical functionalised MWCNTs andSWCNTs, Carbon 43 (2005) 153–161.

21] N. Park, S. Hong, Electronic structure calculations of metal-nanotube contactswith or without oxygen adsorption, Phys. Rev. B 72 (2005), 045408(1-5).

22] S.H. Jhi, S.G. Louie, M.L. Cohen, Electronic properties of oxidized carbon nan-otubes, Phys. Rev. Lett. 85 (2000) 1710–1713.

23] J. Hone, M.C. Llaguno, N.M. Nemes, A.T. Johnson, J.E. Fischer, D.A. Walters, M.J.Casavant, J. Schmidt, R.E. Smalley, Electrical and thermal transport propertiesof magnetically aligned single wall carbon nanotube films, Appl. Phys. Lett. 77(2000) 666–668.

24] H. Chang, J.D. Lee, S.M. Lee, Y.H. Lee, Adsorption of NH3 and NO2 molecules oncarbon nanotubes, Appl. Phys. Lett. 79 (2001) 3863–3865.

25] Jijun Zhao, Alper Buldum, Jie Han, Jian Ping Lu, Gas molecule adsorption incarbon nanotubes and nanotube bundles, Nanotech. 13 (2002) 195–200.

26] J. Robert, Chen, R. Nathan, Franklin, Jing Kong, Jien Cao, W. Thomas, Tombler,Yuegang Zhang, Hongjie Dai, Molecular photodesorption from single-walledcarbon nanotubes, Appl. Phys. Lett. 79 (2001) 2258–2260.

27] Jing Li, Yijiang Lu, Qi Ye, Martin Cinke, M. Jie Han, Meyyappan, Carbon nanotubesensors for gas and organic vapor detection, Nano Lett. 3 (2003) 929–933.

28] J.H. Lee, J.D. Kim, H.W. Seo, J.W. Song, E.S. Lee, M. Won, C.S. Han, Bias modulatedhighly sensitive NO2 gas detection using carbon nanotubes, Sens. Actuators B129 (2008) 628–631.

29] T. Helbling, R. Pohle, L. Durrer, C. Stampfer, C. Roman, A. Jungen, M. Fleischer,C. Hierold, Sensing NO2 with individual suspended single-walled carbon nan-otubes, Sens. Actuators B: Chem. 132 (2008) 491–497.

iographies

guyen Duc Hoa received the degree of Master of Science in Materials Science atanoi University of Technology (HUT), Vietnam in 2003. He is now in PhD course inaterials Science and Engineering, Chungnam Nat. Univ., Korea. His research inter-

sts are nano-materials fabrication, characterization and application to electronicevice, gas sensor and biosensor. Contact: hoa273@yahoo.com.

guyen Van Quy received the degree of Master of Materials Science and Engineeringt Chungnam Nat. Univ., Korea in 2005. He is now in PhD Scholar course in Mate-ials Science and Engineering, Chungnam Nat. Univ., Korea. His current interestsnclude fabrication, characterization of carbon nanotube and its application to nanolectronic devices of field emitter and sensor. Contact: quynguyenh@yahoo.com.

Actua

YNLa

DS

Department of Materials Science and Engineering at Chungnam Nat. Univ., Korea.

N.D. Hoa et al. / Sensors and

ousuk Cho received his PhD in Materials Science and Engineering at Chungnam

at. Univ., Korea in 2007. He is now at a post doctor in Nano Materials and Applicationab. at Chungnam Nat. Univ. His current interests are functionalizing of CNTs and itspplication to field emitter, sensor and solar cell devices. Contact: yousuk@cnu.ac.kr.

ojin Kim received his PhD in Materials Science and Engineering at University ofouthern California, USA in 1989. He is a professor of School of Nanotechnology and

HeId

tors B 135 (2009) 656–663 663

is current research interests are carbon nanotube synthesis and applications tolectronic devices of field emitters, solar cells, gas sensors, etc. Also interested inII–V magnetic semiconductors growth and electrical characterizations. Contact:ojin@cnu.ac.kr.