Polyaniline nanoparticle–carbon nanotube hybrid network vapour sensors with switchable

chemo-electrical polarity

This article has been downloaded from IOPscience. Please scroll down to see the full text article.

2010 Nanotechnology 21 255501

(http://iopscience.iop.org/0957-4484/21/25/255501)

Download details:

IP Address: 75.111.128.142

The article was downloaded on 29/05/2010 at 03:00

Please note that terms and conditions apply.

View the table of contents for this issue, or go to the journal homepage for more

Home Search Collections Journals About Contact us My IOPscience

IOP PUBLISHING NANOTECHNOLOGY

Nanotechnology 21 (2010) 255501 (10pp) doi:10.1088/0957-4484/21/25/255501

Polyaniline nanoparticle–carbon nanotubehybrid network vapour sensors withswitchable chemo-electrical polarityJianbo Lu1, Bong Jun Park2, Bijandra Kumar1, Mickael Castro1,Hyoung Jin Choi2 and Jean-Francois Feller1

1 Smart Plastics Group, European University of Brittany (UEB), LIMATB-UBS,Lorient 56321, France2 Department of Polymer Science and Engineering, Inha University, Incheon 402-751, Korea

E-mail: jean-francois.feller@univ-ubs.fr

Received 30 November 2009, in final form 29 April 2010Published 28 May 2010Online at stacks.iop.org/Nano/21/255501

AbstractChemo-resistive sensors were prepared from monodisperse poly(aniline) nanoparticles(PaniNP) synthesized via oxidative dispersion polymerization. Poly(styrene sulfonic acid)(PSSA) was used as the stabilizer and dopant agent. PaniNP transducers were assembled byspraying layer by layer a solution containing different concentrations of PaniNP and multi-wallcarbon nanotubes (MWNT) onto interdigitated electrodes. This process led to stable sensorswith reproducible responses upon chemical cycling. Chemo-electrical properties of thesesensors have been investigated in sequential flows of pure nitrogen and nitrogen saturated with aset of volatile organic compounds (VOC). Interestingly the sensing mode of PaniNP transducers(the NVC or PVC effect) can be switched simply by increasing PaniNP content or by theaddition of only 0.5% of MWNT to reach a resistance lower than 150 �. Due to their originalconducting architecture well imaged by atomic force microscopy (AFM), i.e. a doublepercolated conductive network, PaniNP–MWNT hybrids present both higher sensitivity andselectivity than other formulations, demonstrating a positive synergy. Mechanisms are proposedto describe the original chemo-electrical behaviours of PaniNP-based sensors and explain theorigin of their selectivity and sensing principle. These features make them attractive to beintegrated in e-noses.

S Online supplementary data available from stacks.iop.org/Nano/21/255501/mmedia

(Some figures in this article are in colour only in the electronic version)

1. Introduction

Over the last decade, the development and design of smartmaterials from conductive polymer composites (CPC) havedrawn increasing interest, especially for sensing applications.CPC sensitivity to elongation, temperature or organic vapourshas made them very attractive in designing transducers tomonitor accidental events like air pollution, overheatingor overloading. Compared to conventional insulatingpolymers, intrinsically conducting polymers have physicaland chemical properties that can be tailored over a widerange of characteristics [1, 2]. Conducting polymerscontain a π -electron backbone responsible for their unusual

electronic properties such as electrical conductivity, lowenergy optical transitions, low ionization potential and highelectron affinity. This extended π -conjugated system of theconducting polymers has single and double bonds alternatingalong the polymer chain. The higher values of the electricalconductivity obtained in such organic polymers have ledto the name ‘synthetic metals’ [3]. Due to their uniqueelectrical properties, interaction of conjugated polymers withgas or vapour molecules (electron acceptors or donors) causeschanges in both carrier density and mobility, leading tosignificant changes in conductivity. Besides, the conformationof the conducting polymer chain can be modified due tostrong interaction with certain organic solvents, which is

0957-4484/10/255501+10$30.00 © 2010 IOP Publishing Ltd Printed in the UK & the USA1

Nanotechnology 21 (2010) 255501 J Lu et al

called a ‘secondary doping’ process. Consequently thereoccurs a change in conductivity [4, 5]. The p-type andn-type sensing response, i.e. an increase or decrease inresistivity, can both be observed for conducting polymerstowards different organic solvents. The vapour responsedepends on the nature of the dopant anion in the polymerwhereas the same sensor has p-type and n-type sensingresponses to different vapours [6]. Thus by simply followingtheir electrical resistivity variation, environmental monitoringcan be done, which makes conducting polymers promisingmaterials for sensing applications [7–11]. Poly(aniline)(Pani) is one of the conducting polymers having the highestmechanical flexibility, environmental stability and controllableconductivity with acid/base (doping/undoping) [12, 13]. Thesynthesis of poly(aniline) can be carried out by dispersionpolymerization in the presence of a polymeric stabilizer, whichprevents PaniNP from aggregating during polymerization; atthe end of the process uniform and spherical monodispersePaniNP are obtained [14–16]. A lot of applications canbe found for poly(aniline)-based composites such as foodquality control [10]. Pani can be used to sense alcohols [17],since n-type responses are obtained due to interactions withmethanol and short-chain alcohols that generate a ‘secondarydoping’ effect. Conversely long-chain alcohols can hardlydiffuse into poly(aniline), which causes strong barrier effects,affecting charge transfer and leading to a p-type sensingresponse [17–19]. Nevertheless, increasing the poly(aniline)specific surface by processing it in the form of nanofibres isfound to enhance its sensitivity to acid vapours due to bettervapour diffusion [20].

Nevertheless, although poly(aniline)-based smart materi-als have demonstrated a good sensitivity and reproducibility,their lack of stability to humidity and irradiation still affectstheir development for applications. To this extent, carbonnanotubes (MWNT), which have recently attracted a lot ofattention due to their exceptional properties [21–24], couldprovide Pani with a high level of stability [25]. Some examplesof the association of MWNT with conducting polymers hasalready led to a positive synergy [26–28]. Therefore it isexpected that the addition of only a small amount of MWNT toPani will significantly improve the mechanical, electronic andchemo-electrical properties of poly(aniline)-based hybrids, asshown by some authors in the literature [29–32]. Moreover,PaniNP–MWNT hybrids have already been used successfullyfor some smart applications like electro-rheological fluids(ER) whose viscosity can be tuned with DC electric fieldstrength [33, 34], electro-chemical biosensing [35] or gassensing [36, 37]. However, reviewing the literature, the effectof poly(aniline) concentration on vapour sensing behaviour isstill not fully elucidated. Thus, developing a chemo-resistivehybrid material for vapour sensing by building a conductingarchitecture through the association of PaniNP with carbonnanotubes has been a promising perspective.

In this paper we have investigated PaniNP–MWNT hybridchemo-electrical properties and compared their characteristicsas vapour sensors with neat PaniNP or MWNT alone. A spraylayer by layer (LbL) assembly has been chosen to structurePaniNP–MWNT hybrid films straight onto interdigitated

microelectrodes in order to control finely the conductivearchitecture building. The transducer’s structure has beenimaged by atomic force microscopy (AFM) to determinethe nature of PaniNP/MWNT association at the nanoscale.The validation of sensor performance by sequential exposureto a set of standard VOC showed that chemo-electricalproperties could be tuned by varying PaniNP concentrationand that, interestingly, only a small amount of MWNTadded to poly(aniline) nanoparticles could enhance not onlythe transducer’s conductivity, but also switches the vapourresponse from negative to positive.

2. Experimental details

2.1. Materials

Poly(styrene sulfonic acid) (PSSA) (Mw = 80 000 g mol−1)obtained from Sigma-Aldrich (USA) is used as the polymericsteric stabilizer and the pendant aromatic ring of PSSA actsas a dopant agent due to its acidity; PSSA is formed by ionexchange of poly(sodium 4-styrene sulfonate) (PSSS) VERSATL-502 from Alco Chemical (USA) with HCl. Multi-wallcarbon nanotubes modified on their surface by carboxylicacid (MWNT–COOH) NC 3101 kindly provided by Nanocyl(Belgium) had a carbon purity higher than 95%. This gradewas chosen to increase MWNT dispersibility in water and thusimprove interactions with PaniNP. Methanol and toluene werereceived from Aldrich (France) and chloroform was obtainedfrom Acros (France). All solvents were used without anyfurther purification.

2.1.1. Synthesis of Pani nanoparticles. Poly(aniline)nanoparticles (PAPSSA) were synthesized by oxidativedispersion polymerization using the PSSA long chain as astabilizer with acidity and as a dopant agent. 37.5 g ofPSSA was slowly dissolved in 350 ml of deionized water (D-water) and then used for nanoparticle preparation. 41.67 gof aqueous hydrochloric acid (HCl, 35.5 wt% in H2O) and0.2 mol of aniline were added to the solution and thereaction mixture was continuously stirred at 0 ◦C. Then150 ml of aqueous solution containing 0.1 mol of an oxidant,ammonium peroxydisulfate (NH4)2S2O8 (APS), was addedthrough dropping by 5 ml min−1 and stirred for 24 h at 0 ◦C.The light yellow solution gradually darkened and acquired anemerald colour over a period of 40 min and eventually turneddark green, which is a characteristic of the doped poly(aniline).The resulting dark green dispersions were purified three timesby centrifugation, filtered and washed in order to removeoligomers and monomers in excess.

2.1.2. PaniNP solution preparation. The poly(aniline)nanoparticle suspension concentration was varied through adilution process; the required amount of PaniNP is achievedby controlling the ratio of solute and solvent. D-water wasused as the solvent to lower the concentration of the PaniNPmother suspension (MS) (0.05 g cm−3). PaniNP dispersionwas ensured by sonication at 25 ◦C for 2 h prior to filmdeposition.

2

Nanotechnology 21 (2010) 255501 J Lu et al

Table 1. Dilution process and electrical properties of PaniNP composite-based thin film.

Sample Total mass (g) PaniNP MS/D-water (g/g) [PaniNP] (wt%MS) [PaniNP] (wt%)a Layer nb Resistance (�)

1%PaniNP 20 4/16 20 1 10 70001%PaniNP–0.005%MWNT 20 4/16 20 1 10 1102%PaniNP 20 8/12 40 2 10 40003%PaniNP 20 12/8 60 3 10 7004%PaniNP 20 16/4 80 4 10 1105%PaniNP 20 20/0 100 5 10 80

a For volume % use dPani = 1.40 g cm−3 [46] and dMWNT# 1.75 g cm−3 [50].

2.1.3. PaniNP–MWNT hybrid sensor preparation. PaniNPtransducers were then processed by spray layer by layer (LbL)assembly onto interdigitated electrodes (figure 1) obtained bycapacitor cleavage [38]. A home-made device allows theprecise control of spraying conditions like nozzle flow rate(index 2), air pressure (ps = 0.20 MPa), sweep speed (ps =0.20 MPa) and target to nozzle distance (dtn = 8 cm) [39].In order to obtain reproducible transducers with relativelylow resistance the LbL process was monitored by followingthe evolution of electrical resistance as a function of layerdeposition. A two-step method was used to prepare PaniNP–MWNT hybrids. Firstly 1 mg of MWNT was added to16 g of D-water to obtain a homogeneous solution (6.25 ×10−5 g cm−3) by 6 h sonication at 25 ◦C. In the second step4 g of 100% PaniNP MS (0.05 g cm−3) was introduced intothe MWNT solution. A uniform mixture of MWNT (5 ×10−5 g cm−3) and PaniNP (0.01 g cm−3) was obtained after 2 hof sonication at 25 ◦C. PaniNP–MWNT (1/0.005) transducerswere then processed by spray LbL in the same conditions asfor PaniNP alone to lead to a final MWNT/PaniNP ratio in thesolid state of 0.5%. After spray deposition, all samples werevacuum dried at 25 ◦C for one day prior to chemical sensinginvestigation. Sample characteristics are listed in table 1.

2.2. Techniques

2.2.1. FT-IR and UV–vis characterization. Successfulpolymerization of PaniNP–PSSA was confirmed by Fouriertransform infrared (FT-IR) using KBr pellets. In addition, UV–vis absorption spectra were tested to get the optical absorbanceof aqueous dispersions containing 1 wt% of PaniNP in thewavelength range of 200–800 nm. Additional figures showingspectra and the attribution of peaks can be found in the‘supporting information (available at stacks.iop.org/Nano/21/255501/mmedia)’ document.

2.2.2. Transmittance electron microscopy (TEM) characteri-zation. PaniNP dispersion was diluted in order to observesingle particles by TEM. The diluted dispersion was droppedonto the TEM grid. The mesomorphology of PaniNP wasobserved using a transmittance electron microscope (TEM,Philips CM200) operated at an acceleration voltage of 120 kV.

2.2.3. Atomic force microscopy characterization (AFM).AFM images were obtained under ambient conditions usinglight tapping mode AFM (TM-AFM) on a multimode scanningprobe microscope (Nanoscope IIIa, Veeco). The ratio of the

Figure 1. SEM picture of the interdigitated electrode obtained from acleaved capacitance.

set point amplitude to the free amplitude was maintained atapproximately 0.9. RTESP. AFM tips (Veeco), with typicalresonance frequency between 300 and 400 kHz and with tipradii between 5 and 15 nm, were used. Diameters of PaniNPwere measured using the section analysis software of themicroscope (V6.13r1 by Digital instruments).

2.2.4. Dynamical vapour sensing measurement. Chemo-electrical properties of PaniNP-based sensors were investigatedby recording their electrical signals when exposed to 10 minperiodic cycles of dry nitrogen and vapour streams. Thedynamic system consists in mass flow controllers, solventbubblers and electrical valves controlled by LabView software.Bubbling dry nitrogen in liquid solvent provides a saturatedvapour stream, which was in turn diluted by nitrogen flow tothe desired concentration at room temperature. The sensingdevice is presented in figure 2 where samples are placed ina 9 cm × 3 cm × 3.5 cm chamber. The design of thedevice allows us to keep constant the total flow rate at Qv =100 cm3 min−1. Electrical characteristics of the transducerswere recorded with a Keithley 6517A multimeter [40]. In thispaper all experiments were done in nitrogen flows saturatedwith the vapour to be analysed.

3. Results

3.1. Structure and morphology of materials

3.1.1. Characterization of PaniNP by TEM. Paninanoparticle size distribution was determined by TEM. In

3

Nanotechnology 21 (2010) 255501 J Lu et al

Figure 2. Schematic of vapour sensing device.

Figure 3. TEM images of PaniNP. (a) Necklace structure of nanoparticles at low magnification, scale bar 0.8 μm. (b) Individual nanoparticleat high magnification, scale bar 100 nm.

figure 3(a), PaniNP shows a spherical and uniform shaperesulting from PSSA stabilization and a well-controlledpolymerization rate [14]. However, an aggregation of PaniNPwas observed, which is induced by strong interactions betweenPaniNP. From figure 3(b), it is possible to estimate the diameterof a single poly(aniline) nanoparticle as about 120 nm with adistribution of 20 nm.

3.1.2. Characterization of Pani–MWNT hybrids. Pani–MWNT solutions were sprayed onto a freshly cleavedmica substrate to observe the surface morphology. ThePaniNP–MWNT microstructure was characterized by TM-AFM. Figure 4(a) reveals at first sight a uniform dispersion ofspherical PaniNP. It is also interesting to discover how MWNTcan interconnect PaniNP to form a hybrid conductive network.Figure 4(b) represents the PaniNP–MWNT architecture athigher magnification. Unlike what is classically observedwhen polymers homogeneously coat a nanotube’s surface,spherical PaniNP are decorating carbon nanotubes featuringa necklace-like microstructure leading to original hybridarchitecture.

3.2. Electrical properties of PaniNP and PaniNP–MWNThybrids

3.2.1. Electrical properties of PaniNP and PaniNP–MWNThybrids in air. Figures 5 and 7 shows that the initialresistance of sprayed LbL PaniNP transducers (measured atroom temperature in air) decreases regularly with increasingPaniNP concentration in the suspension. Since PaniNP contentis easily adjusted, dilution is also a very convenient way toadjust the sample’s conductivity. Interestingly only adding0.005 wt% of MWNT to the 1 wt% PaniNP suspension allowsreducing the transducer’s initial resistance by about two ordersof magnitude. This behaviour can be understood by observingagain in figure 4 how MWNT can very well interconnectPaniNP aggregates to build the hybrid network.

3.2.2. Sensitivity of PaniNP and PaniNP–MWNT hybrid toorganic vapour. Chemo-electrical properties of PaniNP andPaniNP–MWNT sensors were compared when exposed tosequential pure nitrogen/saturated vapour flows. Signals wererecorded and normalized by expressing the relative amplitude

4

Nanotechnology 21 (2010) 255501 J Lu et al

Figure 4. AFM images of PaniNP–MWNT hybrids showing theMWNT network built within PaniNP: (a) 2 μm × 2 μm and(b) 1 μm × 1 μm.

Figure 5. Initial resistance of 1, 2, 3, 4 and 5% PaniNP and 1%PaniNP–0.005% MWNT dry films.

of responses Ar, defined in equation (1), in order to allow theirdirect comparison:

Ar = Rv − Rinit

Rinit(1)

where Rv is the resistance of the sensor when exposed to thevapour and Rinit is the initial resistance in nitrogen.

Figure 6. Evolution of relative amplitudes Ar with PaniNP content(1, 2, 3, 4 and 5%) exposed to saturated vapours of (a) chloroformand (b) water.

Chemical sensing experiments were done with twodifferent categories of solvents: polar solvents, like waterand methanol, and more dispersive solvents, like toluene andchloroform, to investigate PaniNP selectivity. Figure 6(a)summarizes the electrical responses of different PaniNPtransducers (made from suspension concentrations varyingfrom 1 to 5 wt% PaniNP) exposed to a water vapour stream.Surprisingly Ar is found to be negative for 1, 2 and 3%PAniNP while for 4% PaniNP it becomes positive. Similarlyin figure 6(b), upon exposure to chloroform vapour, exactly thesame phenomenon is observed. It seems that at low PaniNPcontent (1, 2 and 3% PaniNP) sensors exhibit a negativevapour coefficient effect (NVC) while at high concentrations(4 and 5% PaniNP) a positive vapour coefficient effect(PVC) is obtained. In figure 7 all PaniNP sensor responses(from 1 to 5% PaniNP) towards methanol, water, toluene andchloroform vapour are summarized and compared to that of100% MWNT and 1%PaniNP–0.005%MWNT. It is clearfrom this graph that, whatever the vapour, an NVC/PVCswitching transition occurs when PaniNP content increasesfrom 3 to 4% PaniNP. Therefore PaniNP content appearsto be a very effective way of tuning the electrical responsepolarity. Moreover, comparing more precisely the dynamic

5

Nanotechnology 21 (2010) 255501 J Lu et al

Figure 7. Selectivity of CPC (MWNT, 1% PaniNP–0.005% MWNTand 1, 2, 3, 4 and 5% PaniNP) towards methanol, water, toluene andchloroform vapour, influence of MWNT presence and Pani content.

Figure 8. Vapour response of 1%PaniNP exposed to methanol,water, toluene and chloroform.

chemo-electrical behaviour of 1%PaniNP and 1%PaniNP–0.005%MWNT responses in figures 8 and 9, respectively,shows that just adding a small amount of SWNT can alsoswitch the sensor’s response polarity. Additionally almostmirror responses are obtained, i.e. selectivity is quite the same,whatever the sensing mode (PVC or NVC). Consequently, theaddition of a small amount of MWNT to PaniNP formulationsgives a second route to switch the sensor’s chemo-electricalresponse polarity and also allows enhancing the responseamplitude. Moreover, in the presence of vapour molecules theresistance increase is immediate and, after removal of vapourmolecules in a pure nitrogen flow, the initial resistance isquickly recovered within seconds. Additionally all electricalsignals are reversible and reproducible. In almost all casessignals reached saturation within the cycle time, but formethanol in figure 9 Ar is still slightly increasing after 600 s,suggesting that equilibrium conditions are not completelyreached. This phenomenon is often observed when theaffinity of solvent molecules for the transducer is too high orwhen the number of diffusing molecules is important. Thusthe diffusing mode tends to be ‘clustering’, as discussed inprevious works [24, 25].

Figure 9. Vapour response of 1%PaniNP–0.005%MWNT exposed tomethanol, water, toluene and chloroform.

4. Discussion

4.1. NVC/PVC switching mechanism

On the basis of previous observations of NVC/PVC polarityswitching it is interesting to propose a mechanism able todescribe this original effect. At first if can be stated that,whatever the route used to generate this phenomenon, it mustbe intimately related to the initial resistance of transducersand thus to the conductive network architecture. In fact,figure 5 shows that all sensors with an initial resistancehigher than 150 � will lead to an NVC effect, whereasbelow this value all sensors give rise to a PVC effect.Nevertheless, It is likely that the origin of NVC/PCV switchingwill be different whether conductivity increase has beenobtained through PaniNP or MWNT addition. But beforetrying to explain the origin of this chemo-electrical transitionit is necessary to better understand independently whichmechanisms can be responsible for NVC and PVC. Classically,the sensing principle of most conductive polymer compositetransducers (obtained by dispersing conducting nanofillersinto an insulating polymer matrix) finds its origin in thePVC effect (p-type sensing). When subjected to vapourmolecules, the dense percolated network (carbon or metal) willbe sensitive to the weakest degradation of its connectivity byconversion of close internanofiller contacts into less conductivetunnel junctions. Thus, adsorption of solvent moleculesonto polymer chains and conductive nanofillers will tend toincrease the internanofiller gap and decrease the transducer’sconductivity [42–45]. This behaviour is characteristic of anetwork built from ‘hard’ fillers in close contact.

4.1.1. Case 1: PaniNP PVC response. In the case of aPaniNP conducting network, as schematized in figure 10(a)Case 1, a single polymer phase is responsible for bothelectronic conduction and interaction with analytes. Thus thePVC effect must result from PaniNP disconnection due tosolvent molecule adsorption on their surface and interactionwith the PASSA coating, increasing the gap between them.

6

Nanotechnology 21 (2010) 255501 J Lu et al

Figure 10. Different cases explaining (a) Case 1: PVC in PaniNP,(b) Case 2: NVC in PaniNP and (c) Case 3: PVC in PaniNP–MWNThybrids.

This is only possible assuming that VOC molecules will notpenetrate too deeply in PaniNP and generate plasticizationand relaxation of macromolecules but only degrade interfacialconduction. In ‘Case 1 sensing’ PaniNP are considered as‘hard’ spheres only swollen in their PSSA shell.

4.1.2. Case 2: PaniNP NVC response. Conversely, ifVOC molecules can swell PaniNP as depicted in figure 10(b)Case 2, this will result in volume expansion and interdiffusionof poly(aniline) macromolecules at the interface betweennanoparticles, leading to an increase in transducer conductivity,i.e. negative vapour coefficient effect. The NVC effect(also called n-type sensing) of ‘Case 2’ is in agreement withthe literature where it has been observed that conducting

polymer transducer films subjected to polar organic vapourmolecules could be partially swollen on their surface (due tomolecular interactions like hydrogen bonding between solventand dopant (PSSA) or poly(aniline) macromolecules) and leadto an increase in conductivity [6, 17, 41]. Moreover, accordingto percolation theory, in the vicinity of the threshold, theconnectivity level is so weak that the PaniNP network is onlymade of a couple of ohmic conductive pathways, the majorityof PaniNP not being in close contact. In such conditions onlysome of the electrons can have enough energy to jump throughthis gap by tunnelling. Thus, volume expansion resulting fromPaniNP surface swelling can decrease average interparticledistance, increasing transducer conductivity and resulting in anNVC effect. In ‘Case 2’ PaniNP are considered as ‘smooth’spheres able to coalesce under expansion; not only their PSSAshell has been swollen but part of their poly(aniline) core,allowing interdiffusion of conducting chains. But in fact howcan the swelling contribution be either positive or negative?It depends on the swelling depth of PaniNP, which in turnis a function of the amount of VOC molecules relative tothe number of macromolecules available for swelling on thePaniNP surface. In other words, for a given amount of analytemolecules, swelling depth will be inversely proportional tothe number of PaniNP. Thus in transducers with a highconcentration of PaniNP (4 and 5%) VOC molecules willsimply adsorb on the surface, whereas for lower amounts ofPaniNP (1–3%) molecular relaxation or complete swelling willtake place [23, 24, 40]. Finally the NVC/PVC switchingeffect can be understood as a transition in the sensing mode,controlled by swelling thickness of PaniNP.

4.1.3. Case 3: PaniNP–MWNT PVC response. Therepresentation of figure 10(c) Case 3 inspired from themorphology revealed by AFM in figure 4 describes thesensing mode assumed for the PaniNP–MWNT hybrid. Theconducting architecture corresponds to a double percolatednetwork of PaniNP and MWNT. The light PaniNP networkinterconnected by a few MWNT is being expanded uponanalyte diffusion, disconnecting the MWNT network andthus resulting in a PVC effect. Such a PVC effectresulting from poly(aniline) expansion was clearly shown withother transducers of the same chemical nature but differentarchitecture, i.e. MWNT coated by poly(aniline) [47].

4.2. A route to design tuneable chemo-resistive sensors

PaniNP and PaniNP–MWNT transducers have demonstratedtheir versatility in designing tuneable chemo-resistive sensors.

4.2.1. Tailoring sensing mode. Depending on initial re-sistance, transducer composition and analyte uconcentration/

PaniNP-specific surface ratio, it is possible to obtain bothNVC/PVC effect sensing modes. Chemo-electrical propertiesof sensors can be tailored by varying PaniNP concentrationor by adding a small amount of MWNT to the formulations.In transducers with resistivities higher than 150 �, noeffective conductive network is obtained due to insufficientinterconnection of aggregates; the conduction mechanism is

7

Nanotechnology 21 (2010) 255501 J Lu et al

Table 2. Solubility parameters of solvent vapours and PaniNP.

δT (J cm−3)1/2 δd (J cm−3)1/2 δp (J cm−3)1/2 δH (J cm−3)1/2 Vmolar volume (cm3 mol−1)

Toluene 18.16 18.00 1.40 2.00 106.30MeOH 29.61 15.10 12.30 22.30 40.70Water 47.90 15.50 16.00 42.40 18.10Chloroform 18.95 17.80 3.10 5.70 79.70PaniNPa 21.97 17.40 8.10 10.70

a From [48, 49].

mainly governed by a decrease in the tunnelling effect and thesensing mode will be the NVC effect. In transducers withresistivities lower than 150 �, electronic conduction in thepercolated network being initially ohmic will be degraded byincreasing tunnelling junction number and consequently thesensing mode will be the PVC effect. Interestingly the bestcombination of sensing effects is achieved with an appropriatedouble percolated architecture of PaniNP–MWNT hybrids thatmaximizes sensitivity, reproducibility and selectivity leadingto a positive synergetic effect.

4.2.2. Tailoring selectivity. Whatever the sensing modechosen, NVC or PVC, PaniNP-based sensors are selectivetowards the nature of the vapour, as can be noticed fromfigure 7. As expected from previous works on MWNT-filled polymer nanocomposite sensors [23, 24], the selectivityof PaniNP–MWNT hybrids will be closely related to therespective interactions of the two constituents with solventmolecules. Classically this correlation is well expressed by theFlory–Huggins parameter χ12 which can be calculated fromequations (2) and (3):

χ12 = V

RT(δT pol − δT sol)

2 (2)

with V the molar volume of the solvent (cm3 mol−1); T isthe temperature (K), R = 8.314 J mol−1; δT sol is the solventglobal solubility parameter (J1/2 cm−3/2); δT pol is the polymerglobal solubility parameter (J 1/2 cm−3/2); δT is derived fromequation (3) and table 2:

δ2T = δ2

d + δ2p + δ2

H (3)

δT is the global solubility parameter from dispersion bondsbetween molecules (J1/2 cm−3/2); δd is the solubility parameterfrom dispersion bonds between molecules (J 1/2 cm−3/2); δp isthe solubility parameter from polar bonds between molecules(J1/2 cm−3/2); δH is the solubility parameter from hydrogenbonds between molecules (J1/2 cm−3/2).

Normally the selectivity brought by the polymer followsquite well equation (4), expressing the fact that the relativeamplitude will increase if χ12 tends to zero, i.e. similarsolubility parameters of both solvent and polymer (seeequation (2)):

Ar = aeb

χ12 (4)

with Ar the relative amplitude, a and b constants, and χ12 theinteraction parameter.

Thus the smaller χ12, the larger the swelling of theinterphase, inducing an increase of interfiller gap, and

Figure 11. Correlation between Ar and χ12 for all PaniNPtransducers plotted from table 3 data.

consequently (according to equation (5)) the higher theresistivity increase, i.e. relative amplitude of Ar:

ρ = aebZ (5)

where ρ is the tunnel resistivity, a and b are positive constants,and Z is the gap between two vicinal CNTs [51].

Nevertheless surprisingly as shown in figure 11, withPaniNP transducers, experimental values of Ar are better fittedwith a logarithmic law like equation (6):

Ar = C log

[1

χ12

]+ d (6)

where Ar is the relative amplitude, c and d are positiveconstants, and χ12 is the interaction parameter.

All data (extracted from table 3) follow this evolutionexcept for some points of PaniNP 1% corresponding tointeractions with methanol vapour which give an overestimatedvalue for Ar, and PaniNP 5%, which does not show anypeculiar selectivity.

The fact that PaniNP transducers behave totally differentlytowards selectivity than classical conductive compositetransducers is a strong indication of the competition betweenadsorption/disconnection and swelling/reconnection takingplace during organic vapour diffusion.

Finally these results demonstrate that PaniNP-basedtransducer selectivity can be tailored by adjusting the initialconductivity and addition of MWNT.

8

Nanotechnology 21 (2010) 255501 J Lu et al

Table 3. χ12 Flory–Huggins interaction parameter and Ar, the relative amplitude of PaniNP-and MWNT-based transducers.

χ12[PaniNP] Ar[PaniNP−CNT] Ar[CNT] Ar[PaniNP−1%] Ar[PaniNP−2%] Ar[PaniNP−3%] Ar[PaniNP−4%] Ar[PaniNP−5%]

Water 4.906 0.63 0.05 −0.5 −0.4 −0.2 0.38 0.4Methanol 0.957 0.9 0.1 −0.5 −0.2 −0.1 0.3 0.5Toluene 0.622 0.28 0.05 −0.2 −0.18 −0.18 0.05 0.25Chloroform 0.295 0.1 0.08 −0.2 −0.1 −0.09 0.1 0.45

5. Conclusion

PaniNP-based chemo-resistive transducers have demonstrateda high potential for vapour sensing applications. Tailoring themain chemo-electrical properties, i.e. sensing mode (PVC orNVC), sensitivity and selectivity toward VOC, can be achievedby modifying the PaniNP conducting architecture using aspray LbL process. Simply adjusting PaniNP and MWNTcontent in the solution to spray allows us to obtain transducerswith very different characteristics. The most striking featuresof PaniNP–MWNT hybrids are their NVC/PVC switchingcapability, resulting from the intimate association of theirpercolated networks well imaged by AFM and their non-traditional selectivity due to the ability of PaniNP to bothcoalesce/expand and conduct electrons. Different mechanismshave been proposed to explain these original behaviours. Thispeculiarity makes PaniNP–MWNT hybrid transducers goodcandidates to be integrated into e-noses, whose principle isto associate components of very different responsiveness andselectivity.

Acknowledgments

The authors are grateful to Herve Bellegou for his contributionto this work. This research was financed by a STAR Franco-Korean collaboration program.

References

[1] Rajesh, Ahuja T and Kumar D 2009 Recent progress in thedevelopment of nano-structured conductingpolymers/nanocomposites for sensor applications SensorsActuators B 136 275–86

[2] Adhikari B and Majumdar S 2004 Polymers in sensorapplications Prog. Polym. Sci. 29 699–766

[3] Dai L 2004 Intelligent Macromolecules for Smart Devices(London: Springer)

[4] McDiarmid A G and Epstein A J 1995 Secondary doping inpolyaniline Synth. Met. 69 85–92

[5] Lee K H, Park B J, Song D H, Chin I J and Choi H J 2009 Therole of acidic m-cresol in polyaniline doped bycamphorsulfonic acid Polymer 50 4372–7

[6] McDiarmid A G 1997 Polyaniline and polypyrrole: where arewe headed? Synth. Met. 84 27–34

[7] Unde S, Ganu J and Radhakrishnan S 1996 Conductingpolymer-based chemical sensor: characteristics andevaluation of polyaniline composite films Adv. Mater. Opt.Electron. 6 151–7

[8] Kukla A L, Pavluchenko A S, Shirshov Y M, Konoshchuk N Vand Posudievsky O Y 2009 Application of sensor arraysbased on thin films of conducting polymers for chemicalrecognition of volatile organic solvents Sensors Actuators B135 541–51

[9] Hatchett D W and Josowicz M 2008 Composites of intrinsicallyconducting polymers as sensing nanomaterials Chem. Rev.108 746–69

[10] Guadarrama A, Rodrıguez-Mendez M L, de Saja J A, Rıos J Land Olıas J M 2000 Array of sensors based on conductingpolymers for the quality control of the aroma of the virginolive oil Sensors Actuators B 69 276–82

[11] Janata J and Josowicz M 2002 Conducting polymers inelectronic chemical sensors Nat. Mater. 2 19–24

[12] Huang X J and Choi Y K 2007 Chemical sensors based onnanostructured materials Sensors Actuators B 122 659–71

[13] Jin Z, Su Y and Duan Y 2000 An improved optical pH sensorbased on polyaniline Sensors Actuators B 71 118–22

[14] Park S Y, Cho M S and Choi H J 2004 Synthesis and electricalcharacteristics of polyaniline nanoparticles and theirpolymeric composite Curr. Appl. Phys. 4 581–3

[15] Hwang J Y, Chin I and Choi H J 2003 Effect of poly (sodium4-styrenesulfonate) stabilizer on the synthesis andcharacterization of polyaniline nanoparticles Mol. Cryst. Liq.Cryst. 407 7–13

[16] Cho M S, Park S Y, Hwang J Y and Choi H J 2004 Synthesisand electrical properties of polymer composites withpolyaniline nanoparticles Mater. Sci. Eng. C 24 15–8

[17] Athawale A A and Kulkarni M V 2000 Polyaniline and itssubstituted derivatives as sensor for aliphatic alcoholsSensors Actuators B 67 173–7

[18] Blackwood D and Josowicz M 1991 Work function andspectroscopic studies of interactions between conductingpolymers and organic vapours J. Phys. Chem. 95 493–502

[19] Topart P and Josowicz M 1992 Transient effects in theinteraction between polypyrrole and methanol vapourJ. Phys. Chem. 21 8662–6

[20] Huang J, Virji S, Weiller B H and Kaner R B 2004Nanostructured polyaniline sensors Chem.—Eur. J.10 1314–9

[21] Treacy M M J, Ebbesen T W and Gibson J M 1996Exceptionally high Young’s modulus observed for individualcarbon nanotubes Nature 381 678–80

[22] Calvert P 1999 Nanotube composites a recipe for strengthNature 399 210–1

[23] Castro M, Lu J, Bruzaud S, Kumar B and Feller J F 2009Carbon nanotube/poly(ε-caprolactone) composite vapoursensors Carbon 47 1930–42

[24] Lu J, Kumar B, Castro M and Feller J F 2009 Vapour sensingwith conductive polymer nanocomposites (CPC):polycarbonate-carbon nanotubes transducers withhierarchical structure processed by spray layer by layerSensors Actuators B 140 451–60

[25] Kumar B, Feller J F, Castro M and Lu J 2010 Conductivebio-polymer nano-composites (CPC): Chitosan-carbonnanotube transducers assembled via spray layer by layer forvolatile organic compound sensing Talanta 81 908–15

[26] Douns C, Nugent J, Ajayan P M, Duguette D J andSanthanam K S V 1999 Efficient polymerization of aniline atcarbon nanotube electrodes Adv. Mater. 11 1028–31

[27] Chen G Z, Shaffer M S P, Coleby D, Dixon G, Zhou W,Fray D J and Windel A H 2000 Carbon nanotube andpolypyrrole composites: coating and doping Adv. Mater.12 522–6

9

Nanotechnology 21 (2010) 255501 J Lu et al

[28] Maity M and Biswas M 2008 Preparation and properties ofpoly(N-vinylcarbazole) and MWCNT nanocompositesJ. Nanosci. Nanotechnol. 8 1751–6

[29] Feng W, Bai X D, Lian Y Q, Liang J, Wang X G andYoshino K 2003 Well-aligned polyaniline/carbon-nanotubecomposite films grown by in situ aniline polymerizationCarbon 41 1551–7

[30] Blanchet G B, Fincher C R and Gao F 2002 Polyanilinenanotube composites: a high-resolution printable conductorAppl. Phys. Lett. 82 1290–2

[31] Long Y and Chen Z 2004 Synthesis and electrical properties ofcarbon nanotube polyaniline composites Appl. Phys. Lett.85 1796–8

[32] Mottaghitalab V, Xi B, Spinks G M and Wallace G G 2006Polyaniline fibres containing single walled carbonnanotubes: enhanced performance artificial muscles Synth.Met. 156 796–803

[33] Park S J, Park S Y, Cho M S, Choi H J and Joo J 2004 synthesisand characterization of nanocomposites of multi-walledcarbon nanotube and polyaniline nanoparticle Mol. Cryst.Liq. Cryst. 425 21–6

[34] Choi H J, Park S J, Kim S T and Jhon M S 2005Electrorheological application of polyaniline/multi-walledcarbon nanotube composites Diamond Relat. Mater.14 766–9

[35] Chang H, Yuan Y, Shi N and Guan Y 2007 ElectrochemicalDNA biosensor based on conducting polyaniline nanotubearray Anal. Chem. 79 5111–5

[36] Li Y, Wang H, Cao X, Yuan M and Yang M 2008 A compositeof polyelectrolyte-grafted multi-walled carbon nanotubesand in situ polymerized polyaniline for the detection of lowconcentration triethylamine vapour Nanotechnology 19 1–5

[37] Zhang T, Nix M B, Yoo B Y, Deshusses M A and Myung N V2006 Electrochemically functionalized single-walled carbonnanotube gas sensor Electroanalysis 18 1153–8

[38] Freund M S and Lewis N S 1995 A chemically diverseconducting polymer-based ‘electronic nose’ Proc. NatlAcad. Sci. 92 2652–6

[39] Feller J F and Grohens Y 2005 Electrical response ofpoly(styrene)/carbon black conductive polymer composites(CPC) to methanol, toluene, chloroform and styrene vaporsas a function of filler nature and matrix tacticity Synth. Met.154 193–6

[40] Bouvree A, Feller J F, Castro M, Grohens Y andRinaudo M 2009 Conductive polymer nano-biocomposites

(CPC): Chitosan-carbon nanoparticle a good candidate todesign polar vapour sensors Sensors Actuators B138 138–47

[41] Li Y, Wang H and Yang M 2007 n-type gas sensingcharacteristics of chemically modified multi-walled carbonnanotubes and PMMA composite Sensors Actuators B121 496–500

[42] Hu J W, Cheng G S, Zhang M Q, Li M W, Xiao D S, Chen S G,Rong M Z and Zheng Q 2005 Electrical resistance responseof poly(ethylene oxide)-based conductive composites toorganic vapours: effect of filler content, vapour species, andtemperature J. Appl. Polym. Sci. 98 1517–23

[43] Chen S G, Hu J W, Zhang M Q, Li M W and Rong M Z 2004Gas sensitivity of carbon black/waterborne polyurethanecomposites Carbon 42 645–51

[44] Feller J F, Langevin D and Marais S 2004 Influence ofprocessing conditions on sensitivity of conductive polymercomposites to organic solvent vapours Synth. Met. 144 81–8

[45] Feller J F, Guezenoc H, Bellegou H and Grohens Y 2005 Smartpoly(styrene)/carbon black conductive polymer compositesfilms for styrene vapour sensing Macromol. Symp.222 273–80

[46] Barthet C, Armes S P, Lascelles S F, Luk S Y andStanley H M E 1998 Synthesis and characterization ofmicrometer-sized, polyaniline-coated polystyrene latexesLangmuir 14 2032–41

[47] Chang C P and Yuan C L 2009 The fabrication of aMWNTs–polymer composite chemoresistivesensor array todiscriminate between chemical toxic agents J. Mater. Sci.44 5485–93

[48] Yong K C, Foot P J S, Morgan H, Cook S and Tinker A J 2006Conductive poly(butadiene-co-acrylonitrile)-polyanilinedodecylbenzenesulfonate [NBR-PAni.DBSA] blendsprepared in solution Eur. Polym. J. 42 1716–27

[49] Jain R and Gregory R V 1995 Solubility and rheologicalcharacterization of polyaniline base inN-methyl-2-pyrrolidinone and N,N ′-dimethylpropyleneurea Synth. Met. 74 263–6

[50] Potschke P, Abdel-Goad M, Alig I, Dudkin S andLellinger D 2004 Rheological and dielectricalcharacterization of melt mixed polycarbonate-multiwalledcarbon nanotube composites Polymer 45 8863–70

[51] Gau C, Kuo C Y and Ko H S 2009 Electron tunnelling incarbon nanotube composites Nanotechnology 20 395705

10