Polymer–carbon nanotube conductive nanocomposites for sensing
In this chapter, recent studies on carbon nanotube conductive polymer nanocomposites (CPC) for sensing are presented. Starting from basic concepts related to CPC transducers, such as percolation, conduction mechanisms and sensing principles, this chapter successively addresses synthesis, fabrication, characterization and structure/sensing properties of these advanced materials issues. Special attention is given to conductive pathways that structure and provide an understanding of sensing behaviour. Finally, examples are given for the main fields of applications, i.e. temperature, strain, chemical sensing.
Carbon nanotube (CNT) smart materials (Kang et al., 2006a) are being used as actuators (Baughman et al., 1999), able to convert current into mechanical work and as sensors, able to convert external solicitations such as strain, temperature, chemicals, into interpretable electrical signals (Kauffman and Star, 2008). In this latter case, conductive polymer nanocomposites (CPCs) are very attractive for designing effective transducers due to their high versatility. Since their discovery (Radushkevich and Lukyanovich, 1952), carbon nanotubes (CNTs) have attracted a lot of attention (see Fig. 25.8 on p. 769) and more recently their implementation in CPCs has shown their high potential for smart applications such as chemical sensing (Castro et al., 2009), temperature sensing (Lisunova et al., 2007) and strain sensing (Thostenson and Chou, 2006a). After some basic concepts related to conduction in C P C, this chapter presents the different steps of CNT-filled CPCs’ transducers’ development: formulation, processing, characterization and properties. Guidelines are given to optimize both materials’ selection and CNT dispersion in the polymer matrix for the desired sensing application. Finally, the potential of CNT-filled CPCs is shown by presenting different examples.
Percolation is a statistical concept, which can describe the interconnection of isolated objects. In this theory, long distance communication is possible once the number of objects becomes larger than a threshold value. On both sides of this transition considerable changes are expected which can result from only a small variation in the number of objects. Initially developed to describe crystals’ growth (Broadbent and Hammersley, 1957), this scaling law has found many different applications in elasticity of gels (De Gennes, 1976), forest fires (Stauffer and Aharony, 1994), economics (Tartarin and Pajot, 1996), powder compaction (Imbert et al., 1997), fluid displacement in porous media (Sahimi et al, 1998), including electrical conduction in polymer composites (Feller et al., 2002a, 2004) and particularly CNT-filled composites (Pötschke et al., 2004a; Du et al, 2004).
The statistical percolation theory is especially effective in describing electrical properties of heterogeneous media like conductive polymer composites. The dependency of conductivity on filler concentration takes the simple form of a scaling law according to Equation 25.1.
Usually, experimental results are fitted by plotting log(ρ/ρ0 versus log(Φ – Φc) and incrementally varying Φc until the best linear fit is obtained (Carmona, 1989). The critical exponent t is the slope of the linearized curve in the percolation zone expected theoretically to depend on the system dimensionality: t = 1.33 and 2 corresponding to 2- and 3-dimensional networks (Stauffer and Aharony, 1994) respectively. But as, experimentally, t values up to 6 have been reported for short carbon fibres (Carmona and Mouney, 1992; Feller et al., 2002a), there is no evidence of any clear relationship between t and the conducting network morphology. Nevertheless the suddenness of the transition is inversely proportional to the distribution of the interfiller junction’s nature. On the other hand, Φc more obviously decreases with increasing filler L/D shape factor (Celzard et al., 2008) and at constant L/D is increasing with filler curvature or waviness (Dalmas et al., 2006). Additionally to these geometrical considerations, Φc depends on dispersion level, physicochemical interactions between fillers and matrices (Bauhofer and Kovacs, 2009) and exclusion volumes (Grunlan et al., 2001, 2004) as represented in Fig. 25.1. The greater the interactions between macromolecules and nanofillers, the higher the level of dispersion, and consequently the lower the conductivity due to insulation of the filler by the polymer. Moreover, when the nanofiller has different levels of organization, like multi-walled carbon nanotubes (MWNTs) in Fig. 25.2 at both the nano and micro scale, the influence of the shape factor and dispersion on conductivity is more complicated to establish. Preferably, an intermediate level of dispersion will be targeted so that an optimum between interconnection and disaggregation of fillers will be found. In fact, this raises the question of the nature of conduction in CPCs.
Electronic conduction in conductive polymer nanocomposites is achieved through various processes, of which the most important definitely are ohmic conduction, due to direct contact between nanofillers, and tunnelling conduction, taking place when electrons can circulate through a small insulating barrier. The first step in the characterization of a conducting network architecture is the determination of current transport mechanisms operating in CPC by plotting the current/voltage (I/V) curves. Fitting curves with Equation 25.2 determines the deviation of conduction from linearity and thus evaluates the proportion of ohmic and tunnel contribution, as shown in Fig. 25.3. Thus a value of n close to 1 will mean that conduction is mainly ohmic, whereas larger values correspond to a tunnelling-dominated phenomenon.
Nevertheless, Celzard et al. (1998) suggest that tunnel resistivity ρT expressed by Equation 25.3 is the major component in CPCs as there is almost always an insulating polymer film between two conducting fillers. This equation well expresses that by increasing voltage, electrons gain energy to cross the insulating barrier and add an increasing contribution to conduction.
On the other hand, for sensing applications, it is preferable to keep the transducer conductivity in ohmic mode so that any perturbation of electrons circulation will generate a detectable tunnel flow.
Before going deeper into the illustration of the intelligent properties of C P C, it is worth defining some words like transducer, sensor or e-nose that will be used extensively in the following. According to an online technical encyclopedia (McGraw-Hill, 2010), a ‘transducer’ is ‘an electronic device that changes one form of energy into electrical signals’ whereas the definition for ‘sensor’ is ‘a device that responds to a physical stimulus (as heat, light, sound, pressure, magnetism, or a particular motion) and transmits a resulting impulse (as for measurement or operating a control)’. In this chapter we will consider that a ‘CPC transducer’ stands for a CPC material able to transduce any chemical of physical solicitation into recordable variations of electrons motion under tension. The CPC transducer is electrically interfaced by wires or electrodes to constitute a ‘CPC sensor’ ready to be connected to an acquisition chain for signal processing and analysis. Additionally, an electronic nose (e-nose) or tongue (e-tongue) will be considered as being a device associating several vapour or liquid sensors (respectively) in parallel. In this case, electrical signals are analysed by pattern recognition algorithms to enable identification and quantification of chemicals. Looking now at the past, it seems that CPCs’ smart properties history starts with pioneer works that early evidenced the potential of CPCs for respectively temperature (Meyer, 1973, 1974), strain (Kost et al, 1983, 1984), and chemical sensing (Freund and Lewis, 1995; Lonergan et al., 1996), and many CPC transducers have found various applications. But it is only recently that CNTs have been introduced into CPC formulations to develop transducers (Wei et al., 2006; Du et al, 2007; Castro et al., 2009; Lu et al., 2009a; Kumar et al., 2010; Kobashi et al, 2008, 2009; Pötschke et al, 2009). Although the sensing principle is almost the same whatever the variable to be measured, i.e. an increase in interfiller gap, there are important differences to consider depending on the nature of the solicitation that the CPC must transduce.
Temperature sensing with CPCs results from their original thermo-electrical behaviour upon heating/cooling. Typically, the important interfiller gap increase resulting from polymer phase volume expansion during melting will induce a resistivity rise of several decades. As described in Fig. 25.4, the commutation temperature can be determined at the beginning of the resistivity jump, known as the positive temperature coefficient (PTC) effect. This temperature coincides with the beginning of polymer crystalline phase melting (Feller et al., 2002b) or the amorphous phase softening (Pillin et al., 2002), although in the latter case a lower PTC amplitude is generally obtained. With single polymer phase CPCs, an undesirable phenomenon upon heating, related to reaggregation of fillers in the melt is often observed, and is called negative temperature coefficient (NTC). Cross-linking the matrix or using co-continuous polymer phases can prevent this effect, which must not be mismatched with the reversible NTC effect observed upon cooling that corresponds to normal and reversible crystallization of the polymer.
Strain sensing is based on the ability of CPCs to transduce any elongations that will modify the interfiller gap and as a result generate tunnelling conduction which depends exponentially on the inter-CNT gap Z, according to
Figure 25.5 illustrates the kind of signal that can be obtained during strain sensing tests. Possible drawbacks can be non-linearity of signal due to the fact that the polymer matrix has been moved out of its linear range, which can result in permanent drift of the sensor. Thus, depending on the amplitude of the deformation to be followed, it is necessary to select a polymer matrix so that it will be always more elastic than the sample to monitor.
The chemo-electrical behaviour described in Fig. 25.6 is certainly the most complicated effect to understand. It can be used to design sensors for chemicals in both gas or liquid state but obviously the requirements will differ depending on the target. Both positive and negative vapour coefficients can be interpreted depending on the electronic conduction mechanism (Lu et al., 2010). For example, the sensing mode of PaniNP transducers can be switched from the negative vapour coefficient (NVC) to positive vapour coefficient (PVC) effect by simply tuning initial conductivity through the addition of a small amount of CNTs. CPCs’ chemo-electrical response to organic molecules can be interpreted and quantified from the analysis of changes in electrons’ motion within the CNTs’ percolated network. It is assumed that during their diffusion through the composite, vapour molecules can disconnect CNT–CNT junctions by increasing the gap between nanotubes directly by adsorbing on carbon or indirectly by relaxing macromolecules in the vicinity of the junction. Consequently, quantum-tunnelling conduction (less effective) will develop to the detriment of ohmic conduction, resulting in an important resistance increase even for a small amount of solvent molecules. The amplitude of this phenomenon is classically evaluated by following the evolution of Ar the relative resistance defined by Equation 25.5.
Ar depends on many parameters, of which at least the following must be controlled: sample thickness, temperature, initial resistance, amount of molecules in the sensor surrounding and specific interactions of analytes with macromolecules from the matrix. The Flory-Huggins intermolecular interaction parameter, calculated from Equation 25.6, is found to well predict the sensitivity of CPC transducers to volatile organic compounds (Kumar et al., 2010) and organic liquids (Kobashi et al., 2009):
Nevertheless, there is generally no univocal relation between one CPC formulation and one vapour based on a ‘key–lock’ principle that prevents direct identification. To overcome this difficulty, electronic noses (e-nose) analyse a combination of electrical responses from a set of selected CPCs by mimicking the mammalian sense of olfaction (Lonergan et al., 1996; Shevade et al., 2003).
Additionally, dielectric permittivity and even the size of solvent molecules are also able to interpret chemical sensors’ selectivity (Lu et al., 2009b). Interestingly chemo-electrical responses upon volatile organic components (VOC) exposure are proportional to the amount of molecules in the vicinity of the CPC transducer, according to the Langmuir–Henry-clustering (LHC) model derived from classical sorption formalism. This model describes quite well which diffusion regime takes place in the CPC transducer: simple adsorption, diffusion, clustering, corresponding respectively to the three terms of Equation 25.7:
where bL is the Langmuir affinity constant, f′ is the vapour fraction over which Langmuir’s diffusion is replaced by Henry’s diffusion, f is the solvent fraction, kH is Henry’s solubility coefficient, n′ the number of vapour molecules associated in clusters.
On the one hand, this model identifies which kind of diffusion mode is more suitable for sensing (Fig. 25.7) whereas, on the other, it makes possible the determination of organic molecules’ concentration in the atmosphere, from relative amplitude measurement.
It is thanks to these remarkable properties that CPCs have attracted attention for smart materials development (Fig. 25.8), but their great sensitivity can also be a drawback if not properly exploited through the right choice of components during the formulation step.
25.8 Histogram detailing the number of CNT publications (N) per year between 1991 and 2007 (data obtained from ISI Web of Knowledge according to Kauffman and Star, 2008).
As with any other application, the choice of the different components of the composite formulation is crucial. This first step consists in selecting and assembling the different pieces of what will become a sensitive material. Although there is a wide range of materials potentially available, hundreds of different matrices and tens of conducting fillers, there will in the end only be a few suitable combinations.
First of all, what are the reasons why CNT should be chosen among all conducting fillers used in CPC formulations for sensing? Metal nanofillers like gold nanoparticles (AuNP) can be self-assembled with dendrimers of poly(propyleneimine) to design vapour sensors with good sensitivity to volatile organic compounds (VOC) like toluene, 1-propanol and water (Krasteva et al., 2002, 2003). The resistivity can be adjusted by the number of generations of the dendrimer that will vary the interparticle gap and thus the tunnelling capability through the conducting network. Another strategy has been used with poly(electrolyte)-coated AuNP and more simply AuNP functionalized with citrates prior to dispersion in poly(styrene) (Bouvrée et al., 2007). In both cases, CPC thin layers sensitive to VOC were obtained. Examples of use of iron particles in diphasic poly(ethylene)/poly(oxyethylene) systems (Mamunya et al., 2007) or silver and copper particles (Boiteux et al., 1999, 2006) dispersed in epoxy resin are reported to be very sensitive to temperature. Nevertheless the use of spherical fillers leads to high percolation thresholds and metal particles have also a strong tendency to aggregate since their surface energy is high, which imposes a treatment of their surface. Moreover their high density (d#20 for gold), high cost, and in some cases their sensitivity to oxidization are a penalty that promotes the use of carbon nanofillers in many cases. Carbon nanoparticles (CNPs), also known by the name ‘carbon black’, have long been used by the Chinese in inks for their colour since 2850 BC and more recently by Goodyear in rubber formulations for their reinforcement effect since the end of nineteenth century, and have been introduced in different polymer matrices for the design of e-noses (Freund and Lewis, 1995). However, one of the main advantages of CNTs among CNPs is their huge shape factor (L/D > 1000) resulting in percolation thresholds lower than 0.02 wt% MWCNT when dispersed in epoxy by heat shearing (Martin et al., 2004), less than 0.1 wt% SWCNT when dispersed in poly(ε-caprolactone) in solution under sonication (Mitchell et al., 2002), or less than 0.04 wt% MWCNT in PA 6.6 matrix (Krause et al., 2010). Moreover, even at the same effective content (normalized towards content at percolation threshold), CNT-filled chitosan transducers were found to be more sensitive to vapours than their CNP-filled homologues, especially at low analyte concentrations (Kumar et al., 2010). Despite all these advantages and although they have been successfully used to design sensors (Kauffman and Star, 2008), carbon nanotubes used on their own suffer from a lack of adjustable selectivity, due to the fact that their response mechanism is only based on the affinity of target molecules for their surface. To cope with this problem, it is necessary either to graft them with functional molecules or to associate them with a polymer matrix.
One of the major interests in CPC sensors is in the large diversity of the polymer matrices that can be associated with the conducting fillers. Depending on sensing and application targets, the choice of the polymer nature will provide a powerful adjustable capability. For temperature sensing, the driving parameter that will determine the commutation temperature, as expressed in Fig. 25.4, is the melting temperature of the polymer matrix used to disperse the conducting nanofiller. Once the temperature increases above this peculiar value, the CPC resistivity is dramatically increased which can trigger an action like an alarm. Some examples of commutation temperatures covering the range 35–160 °C are given in Table 25.1, showing that it is rather easy to select the right polymer for the suitable target temperature range to sense. For strain sensing the polymer matrix will be chosen so that its practical extension will be in its linear elastic domain. Depending on the strain range to sense, some percentages or some tens of percentages, an amorphous matrix in its glassy or elastomeric state will be chosen respectively. An appropriate selection by the proper mechanical characterization will prevent non-reversible plastic deformation of the sensing often responsible for initial resistance drifts illustrated by Fig. 25.5. Concerning vapour sensing, the choice of matrices is wide, depending on chemical functions present on macromolecules, expected to interact with the target molecules. As there is generally no univocal relationship between a macromolecule and a solvent, it is necessary to select a set of polymers whose combination will give a unique recognition pattern. Equation 25.6 can guide the selection in a first step although some other important parameters must be taken into account such as transducer degradation or non-linearity due to extensive swelling.
The choice of the matrix is not only based on functional criteria but also on ease of processing. Thermoplastics-based transducers in the form of micronic films or fibres can be prepared via melt mixing processes, such as twin-screw extrusion (Kobashi et al., 2008), internal batch mixing (Lee et al, 2006) or compression moulding (Vidhate et al., 2009) or melt spinning (Pötschke et al., 2009). On the other hand, the preparation of nanometric layers requires the use of solution mixing. An effective dispersion of carbon nanotubes in the polymer matrix can be achieved by sonication and ultracentrifugation (Bonnet et al., 2007a), different film preparation techniques like casting (He et al, 2005; Pioggia et al., 2007; Liu et al., 2007), spin coating (Gau et al., 2009b), or spray deposition (Castro et al., 2009) will produce ultra thin transducers. Additionally, the use of the layer-by-layer technique initially developed by Mamedov et al. (2002) with polyelectrolytes was used to build hierarchical 3D architectures by Lu et al. (2009b) to increase specific surface and enhance vapour transducers’ sensitivity. Solution mixing has also been widely used for CNT dispersion in epoxy resins for structural composites’ health monitoring strain. Filler dispersion done under stirring was improved by the use of surfactants like sodium dodecyl sulphate and controlled sonication (Park et al., 2003, 2007), and high shear mixing (Zhang et al., 2007). However, as sonication is suspected to alter the carbon nanotubes’ structures, it has been combined with a three rolls mill calender, taking the benefit from the very high shear rate of this technique (Thostenson and Chou, 2006b; Böger et al., 2008; Nofar et al., 2009).
Further to filler dispersion in the matrix, CPCs are prepared via common thermoplastics or thermosets processing techniques. No doubt that this fast step is one of the most important as it will fix the morphologies at the different scales from nano to macro. Consequently, it is necessary to find strategies to control the multiscale architecture and provide the CPCs with stable and reproducible properties for sensing.
There are many ways to use exclusion volumes to concentrate and localize the conducting fillers into confined areas by phase segregation as very well illustrated in Fig. 25.10. To prevent the complete mixing of the filler and the matrix it is possible to act on the viscosity of the matrix, or on mutual interactions. A good illustration of this method has been given by Mierczynska et al. (2004) and Lisunova et al. (2007) who associated CNT with UHMWPE power by hot compaction (Fig. 25.9), whereas Yu et al. (2008) confined CNT between latex microparticles from aqueous emulsions for thermoelectrical applications. Additionally, Mu et al. (2008) obtained conductive composite via coating of polystyrene beads with 0.5 wt% SWCNT and found higher conductivity in comparison to simply mixed composite. Besides the considerable decrease of percolation associated with this process, stable morphologies are obtained, provided that materials are not submitted to high temperature cycles. If this is the case, as in temperature sensors, the use of co-continuous structures can secure morphologies even when the conducting phase is in the liquid state. Lu et al. (2009a) have developed diphasic blends of polypropylene and CNT-filled poly(ε-caprolactone) 50PP/50(PCL-3%wCNT) for temperature sensing which were found to both stabilize sensing signal and ensure a good reproducibility upon cycling (Fig. 25.10). Although for both strain and chemical sensing there is no strong need for co-continuous structures use, there is some reason to use the insulating polymer phase to provide the transducers with additional mechanical strength or barrier capability. Volume exclusion can also be generated by the introduction of a second filler such as clay nanoparticles (Feller et al., 2004; Etika et al., 2009), rubber microparticles (Zribi et al., 2006) or ceramic microparticles (Droval et al., 2008) that will provide additional functionality to the CPC such as stabilization of CNP dispersion by mutual adsorption, increase of electrical conductivity and increase of thermal conductivity respectively. Figure 25.11 shows an example of conducting architecture in which triple percolation, the percolation of each filler in one polymer phase associated with phase percolation of the two polymers leading to co-continuity, has been achieved to uncouple electrical and thermal conductivity. Nevertheless, any addition of any filler will also strongly modify the CPCs’ rheological properties that must be addressed in the final optimization step.
25.9 Segregated network of 0.004 vol% of MWNT in UHMWPE matrix by hot compaction (Lisunova, et al., 2007)
25.11 Triple percolated CPC 50(sPS-28vol%Al2O3)/50(HDPE-23vol%CNP) observed by SEM (Droval et al., 2008).
Layer by layer is a powerful protocol to develop progressively aggregated structures in 2D (Decher, 1997; Schneider and Decher, 2004). Nevertheless, Bouvrée et al. (2007) found that the use of polyelectrolyte to generate self-association of conducting fillers led to insufficiently conductive percolated networks for sensing. This drawback was overcome by spraying layer by layer an aPS–AuNP solution which was more effective in structuring vapour transducers in 2D as represented in Fig. 25.12. Bouvrée et al. (2007) have extended this technique to hierarchical double-percolated PC-2wt%CNT conductive architecture of PC–CNT (Fig. 25.13) and showed that this morphology was compatible with a good dispersion of CNT in the polymer matrix at the nanoscale, evidenced by A F M, whereas, at the microscale, optical microscopy shows that polymer–CNT microdroplets formed during spraying can coalesce together upon drying inside the same layer but also between different layers in three dimensions. This process finally resulted in a hierarchically structured vapour transducer whose thickness and composition can be adjusted (by the number of layers) to tailor the chemo-electrical properties. Mamedov et al. (2002) and Loh et al. (2007) have also demonstrated the validity of this technique by dip coating to fabricate strain sensors from PVA–CNT solutions.
25.12 2D layer-by-layer sprayed aPS-AuNP transducer observed by AFM (Bouvrée et al., 2007).
25.13 Hierarchical double-percolated PC-2wt%CNT structure observed by optical microscopy (Lu et al., 2009b).
Whatever the processing technique used for transducer fabrication, before evaluating sensing properties, the first round of characterization will be electrical and morphological observations at different scales, i.e. from nano to macro, of CNT-based composite sensor.
Electrical measurements in quiescent conditions obviously represent the most relevant technique for CPC transducers’ initial characterization. Mainly two kind of information can be derived from such properties. First, the percolation curve expresses the quality of the conducting network in terms of interconnections’ effectiveness. Ideally an optimum between CNT debundling and connectivity must be found. Figure 25.14 is a good illustration of the influence of elaboration route, polymerization filling or simple solution mixing, on CNT dispersion and finally electrical properties (Zhang et al., 2005). This shift in percolation curves is expected to result from partial CNT bundles exfoliation during in-situ polymerization and better wetting and penetration of CNT aggregates by lower molar mass chains. Another illustration of what can be learnt from percolation curves is given in Fig. 25.15 where it is clearly seen that increasing the shear rate during CNT dispersion in epoxy resin too much leads to more insulating C P C, evidencing the existence of an optimum to reach during dispersion. There are many other outputs of percolation curves’ interpretation, such as demonstrating the influence of conductive filler aspect ratio, conductive filler size, alignment of nanofillers, presence of dispersive agent, or nanofiller surface modification.
25.15 Influence of stirring speed (50, 500, 2000 rpm) on percolation curve of epoxy–CNT (Bauhofer and Kovacs, 2009).
Additionally, Fig. 25.16 shows the potential of alternative current experiments to investigate percolated architecture. Cutting frequency sweep curves at low frequency gives back the percolation curve, whereas below 1%wt CNT there is a critical frequency (around 2000 Hz), above which the current is increasing with frequency. This is another appearance of quantum tunnelling, according to Equation 25.3, i.e. only when electrons have a higher energy than the barrier value can they add their contribution to ohmic conduction.
It is finally the so-called current/voltage curves that will complete the transducers’ electrical properties’ characterization. In Fig. 25.17, the curves fitted with Equation 25.2 well illustrate that the divergence from linearity over 1 V induced by quantum tunnelling contribution is emphasized as filler content approaches the percolation threshold. Fitting parameters in Table 25.2 allow us to quantify this additional conduction.
Fitting parameters of PMMA–CNT transducers from Equation 25.2
Electrical characterizations globalize the different phenomena responsible for conduction in CPCs. To go deeper into understanding the relationships between structure and electrical behaviour, imaging morphologies at the different scales from nano to macro are helpful and meaningful. At the microscale, optical microscopy (OM) is a pertinent technique to observe sprayed microdroplets welding (Lu et al., 2009b), as presented in Fig. 25.13, or microphase co-continuous morphologies as in Fig. 25.10 (Lu et al., 2009b; Liu et al., 2007); CNT aggregates remaining in CPCs depending on processing conditions can also be shown by OM (Villmow et al., 2008), nanofiller dispersion homogeneity can also be checked as, even if individual CNTs cannot be seen by this technique, CNT micronic aggregates are perfectly visible (Quercia et al., 2005). Laser scanning confocal microscopy (LSCM) has also been successfully used to study the dispersion of carbon nanotubes in a poly(styrene) matrix by (Bellayer et al., 2005).
Scanning electron microscopy (SEM), transmission electron microscopy (TEM) and also atomic force microscopy (AFM) will allow much higher magnification from submicron to nanometer, imaging details of nanofillers such as grafting and quality of dispersion at the meso scale. SEM has been extensively used to investigate the CPC network structure at the microscale (Zhang et al., 2005; Li et al., 2007), CNT agglomerates (Thostenson and Chou, 2006b), transducers’ thickness (Song et al., 2009), or composites’ fractured surface (Thostenson and Chou, 2008). Using high accelerating voltage provides complementary information on the CNTs’ networks, which tend to charge, leading to highly contrasted images due to enriched secondary electrons emission. (Zhang et al., 2005) gave a good illustration of this technique, showing that CNT networks in a polyurethane-urea matrix looked like entangled coils rather than individual tubes. To be able to image and quantify fillers dispersion at the nanoscale, it is necessary to use the higher magnification achieved with T E M, as showed by (He et al., 2005; Quercia et al., 2005; Zhang et al., 2005; Ferrara et al., 2006; Wang et al., 2007; Thostenson and Chou., 2006b; Kobashi et al., 2008; Pegel et al., 2009; Liu et al., 2009). However, this technique requires a heavy sample preparation step and only permits visualizing nanofillers in a 2D plane which makes statistical treatment a necessity to determine not only the local but also the global state of dispersion. Since its discovery in the late 1980s, AFM has demonstrated increasing capabilities and ease of use to characterize CPCs’ structure and morphology at the nanoscale. In tapping mode, forces act on the tip due to the tip–sample interactions that result in deflection or torsion of the cantilever, which are detected by a position-sensitive optical sensor. By detecting movements of the cantilever, the height differences of the surface can be resolved on a nanometric scale. Thus it is possible to measure CNTs’ dimensions and dispersion level after proper conditioning of the sample by depositing it onto freshly cleaved mica substrate, as illustrated in Fig. 25.18 (Lu et al., 2009b), or straight onto fractured strand surfaces (Pötschke et al., 2004). Castro et al. (2009) also used AFM to check the efficiency of grafting by in-situ polymerization, or to investigate the CNT network connectivity through the thickness by using current sensing (CS) mode (Kumar et al., 2009). As most CPC transducer sensing properties rely on its conducting network structure, it is particularly interesting to use the latter technique to probe its nanomorphology. In Fig. 25.19 (Kumar et al., 2009) investigated the nano-electrical properties of a MWCNT-filled PMMA-based sensor material by CS–AFM in parallel with classical AFM mode distinguished from the insulating polymer matrix. The current image shows carbon nanotubes connected to the electrode, which is 15 μm far from AFM tip, through the 3D-built CNT network within the whole volume of the sample.
25.18 AFM images (amplitude) of PC–CNT composite on mica: (a) PC-1% CNT (5 μm × 5 μm); (b) higher resolution of PC-1% CNT (w/w) (2 μm × 2 μm); (c) PC-2%NT (w/w), (5 μm × 5 μm); (d) higher resolution of PC-2% CNT (w/w) (2 μm × 2 μm); and (e) 3D image of PC-1% CNT (w/w) (height image) (Lu et al., 2009).
Once transducers’ electrical properties and structures have been characterized in a quiescent state, the second step consists in determining their sensing performances. The transduction mechanism responsible for the sensing ability of CNT-filled CPCs is almost always based on the disconnection of the CNTs’ conducting architecture. As only some nanometers of interfiller gap fluctuations will generate large resistance variations, due to quantum tunnelling, this will make the CPCs very sensitive to their environment. Nevertheless the origin of this disconnection can be several, so that it is necessary to precisely control experimental conditions and possible sources of interference to make sure to really characterize the desired phenomena. Depending on which parameter has to be monitored, electrical measurements will be coupled with chemical, thermal or mechanical solicitations generally applied in loops to evaluate the sensor dynamics, reproducibility, sensitivity and durability.
Resistivity measurements are usually done by two or four points probes on the surface of samples, thanks to silver paste ensuring a smooth transition between electrodes and CPC transducer. In some cases, silver or copper threads embedded in sample bulk or copper foils glued on its surface, are used to monitor resistance changes. Depending on the targeted applications, the experimental set-up will present some particularities.
For temperature sensing, a thermal cycle is imposed on samples in a computer-controlled oven where the resistivity changes induced by temperature variations are recorded to analyse both PTC and NTC effects. The heating/cooling speeds must be adjusted to prevent artefacts related to samples’ inertia. Generally the first cycle is different from the others due to the relaxation of stresses resulting from previous processing and then all the following cycles are quite superimposable, provided that no dynamic percolation has increased the fillers’ aggregation and consequently decreased resistivity (Zribi et al., 2006).
For chemical sensing, several devices can be used depending on the physical state of molecules to analyse, either vapour or liquid, dynamic or static mode. Static mode is the simplest protocol, which consists in injecting a defined amount of target molecules (from several ppm to saturated conditions) into a cell of known volume. The dynamic mode is more complex to handle, as the sensor is placed into a gaseous stream of molecules, the content of which is adjusted by blending a stream of pure nitrogen (or air) with a second stream saturated with the vapour to analyse (Lu et al., 2009b; Castro et al., 2009). For liquid sensing characterization, immersion/drying cycles of the CPC are performed either manually or with the help of an automated device; solvent drops remaining on samples are wiped off at the beginning of drying. Different samples’ geometry can be tested with suitable clamps, U-shaped films or filaments (Kobashi et al., 2008, 2009).
Characterization of strain sensing properties is done classically by stretching transducers previously glued onto a substrate (metal, epoxy, etc.) with a mechanical testing machine coupled with electrical measurement. In the case of structural composites (Thostenson and Chou, 2006a) or films (Zhang et al., 2007b), samples are directly tested in isolated clamps. Mechanical solicitations can be monotonic tension, three/four bending, compression, tension-compression cyclic loading, fatigue, creep. Complementary devices, like acoustic emission, have also been combined with electrical measurements to investigate damage mechanism in thermoset-based composites.
The smart characteristics of CNT-based CPC sensors towards temperature are illustrated by a large positive temperature coefficient (PTC) effect of electrical resistivity. This effect can be followed by a negative temperature coefficient (NTC) effect. Mierczynska et al. (2004) and Lisunova et al. (2007) achieved a very low percolation threshold (~ 0.05 vol%) with CPCs based on the segregated network concept. They investigated the thermo-electrical behaviour of UHMWPE-2%CNT above its commutation temperature around Tcom = 112 °C (Table 25.1) and evidenced a PTC effect of two decades amplitude; this value is comparable to that obtained by Mironi-Harpaz and Narkis (2001) with co-continuous 70PVDF/30(UHMWPE-6%CNP) but two times less than the four decades obtained by He et al. (2005) with HDPE-5.4%CNT. The same authors report PTC amplitudes of seven decades with HDPE-16%CNP but interestingly using three times less the amount of CNT prevents the appearance of an NTC effect in the liquid state. This was explained by a higher stability of the CNT network than the carbon nanoparticles’ (CNP) percolated structure. Correlatively (Deng et al., 2009) carefully studied the influence of polypropylene matrix melting and re-crystallization in carbon-filled CPCs on the formation and destruction of the conductive network. They conclude on the higher strength of CNT network compared to CNP. Moreover, Lee et al. (2006) found many positive effects of the addition of only 0.5 wt% of CNT to HDPE-25%CNP CPC. The first effect was to decrease the CPC percolation threshold by about 20%, the second to increase CPC stability in the heating/cooling cycling, the third to enhance PTC intensity by 1.5 decade, and finally to increase HDPE crystallinity. (Gao et al., 2009) confirmed the strong effect of processing conditions of segregated UHMWPE–CNT CPCs, mainly related to crystallization, on PTC characteristics like amplitude and Tcom’ which could be increased from 1.37 to 2.05 by decreasing moulding temperature from 200 to 160 °C, or increasing melting enthalpy from 133.6 J/g to 151.2 J/g by increasing isothermal treatment at 100 °C duration from 12 to 84 hrs. In complement, (Jiang et al., 2006) studied the combined effects of pressure and temperature on methyl vinyl silicone rubber (VMQ) filled with γ-aminopropyl triethoxy silane (APS) modified CNT. They found that it was possible to dramatically modify VMQ-2.5%CNTs’ sensing behaviour, i.e. increasing APS content from 1 to 5% enhanced sensitivity to pressure and shift Tcom from 50 to 20 °C. Many works on temperature sensing or temperature self-regulation concern the poly(ethylene) matrix CPC due to the fact that very early, Meyer (1973) had identified its capability to generate high amplitude PTC effect, presumably thanks to its low glass transition temperature Tg. Nevertheless, high PTC amplitude is mostly interesting for self-regulation whereas temperature monitoring only needs a measurable change of relative resistance Ar. From this point of view, it is much more interesting to change Tcom of the CPC transducer to vary the measurement range. Thus, Lu et al. (2009a) developed a transducer based on a CNT-filled poly(caprolactone) PCL conducting phase, with low Tcom (Fig. 25.20) able to sense temperature rise over 50 °C, making it possible to design sensors that can trigger an alarm when the human pain threshold is reached. This can be particularly valuable for firemen’s protective clothing (Fig. 25.21). Figure 25.20 shows the addition of an external matrix like PP or poly(amide12) PA12 to obtain a co-continuous morphology and secure CPC thermo-electrical behaviour over PCL melting, is followed by a decrease in PTC amplitude. Although this decrease is acceptable for P P, with PA12, no significant resistance change is observed upon heating which prevents temperature sensing. A similar behaviour previously encountered with poly(butylene terephthalate)/poly(ethylene-co-ethyl acrylate)-carbon nanoparticle 60PBT/40(EEA-CNP) had been explained by the migration of CNP density at the interface between the two polymers (Feller et al., 2004). Finally, PP–PCL–CNT co-continuous CPCs have been developed to design temperature transducers in either film or fibre form able to resist temperature overshot up to 150 °C (INTELTEX, 2010). There are also possible applications in electronics. Applications for PTC composites include thermistor, circuit protection devices, and self-regulating heaters. Coupled with a wireless communication system to a computer, temperature sensors can be integrated into structures.
Since early works on health monitoring of structural composites reported by (Wang and Chung, 1997) who used the conductivity of reinforcing carbon fibres to follow dynamic strain and damage, this topic has attracted a lot of attention. (Park et al., 2003) demonstrated that introducing only a small amount of carbon nanotubes (less than 0.5 vol.%) could provide an alternative way to monitor carbon fibres (CF) reinforced epoxy (EP) composites damage in a non-destructive way, just following the evolution of CPC resistance. In fact, carbon nanotubes are three orders of magnitude smaller than usual reinforcing fibres in structural composites, they can be used to create a conductive network in the polymer, penetrating the matrix-rich region surrounding the fibres, and thus allowing the direct monitoring of deformation-induced resistance changes. Moreover, Fig. 25.22 shows that the electrical signal will alert slightly in advance carbon fibres’ breakage measured mechanically, which is very interesting in applications to anticipate structural composites’ failure. A few years later, Thostenson and Chou (2006b) and Thostenson et al. (2007) reported that building carbon nanotubes conducting networks in epoxy/glass fibre (UP/GF) composites was effective to monitor damage in advanced polymer-based composites. The inherent multi-functionality of carbon nanotubes was successfully illustrated in this work to develop a ‘smart materials’, both charge-induced deformations and in-situ damage monitoring were investigated. Additionally, it was shown that the percolating carbon nanotubes network was very sensitive to initial stages of matrix-dominated failure, confirming carbon nanotubes’ potential for health monitoring (Fig. 25.23). Zhang et al. (2007b) have also demonstrated that thanks to the CNT network, delamination defects could be detected via electrical resistance measurement in laminated composite structures and that the use of MWNT additives induces fast heating of crack interfaces allowing self-healing, i.e. reparation of damage after detection and recovery of up to 70% of the strength. Another contribution to the topic has been added by Wichmann et al. (2008) who developed a direction-sensitive bending strain sensor by generating a gradient in electrical conductivity throughout a single block of epoxy-CNT composite. Electrical resistance becomes positive or negative, depending on the direction of bending deflection.
25.22 Damage sensitivity of fiber fracture of EP/CF composites with 0.1 vol% CNT (Park et al., 2003).
25.23 Demonstration of crack healing in (a) scheme of fractured and healed samples. (b) Load vs displacement curves of virgin (epoxy-0.5 wt % CNT) and healed (virgin + 30% uncured heat-curing additive) demonstrating 55% to 70% recovery of the ultimate failure load (Zhang et al., 2007).
Instead of using the composite itself to monitor strain or stress, another approach has been recently proposed by Alexopoulos et al. (2009) who embedded CNT fibres (prepared by coagulation of CNT dispersion in a PVA solution) in glass-fibres reinforced composites. Parabolic and exponential correlations were found between mechanical stresses and the electrical responses, and it was shown that the constants thus obtained were linearly connected to the composites’ previous loading history. The same kind of strategy was used by Kang et al. (2006b) who developed thin layers of poly(methyl methacrylate) PMMA-impregnated carbon nanotubes bucky papers to form a piezoresistive strain sensor material for health monitoring applications. Strain sensors were glued under vacuum on the surface of a composite cantilever beam (to overcome slippage issues and sensors’ distorted responses) before impedance spectroscopy and strain testing. A linear symmetric strain response in both compressive and tensile cases was observed for the SWNT–10%PMMA sensors, even though sensitivity was lower than bucky papers sensors. Moreover, a biomimetic artificial neuron was developed by extending the length of the sensor. This long continuous strain sensor is low cost, simple to install and is lightweight, so that it may allow different target applications like crack propagation prediction in composites’ structures (buildings, bridges, aircraft) or tactile sense application in artificial skin development. Zhang et al. (2006b) used poly(carbonate)–CNT CPC for strain sensing application and reported strain gauge factor (defined by the ratio of relative amplitude and deformation, (ΔR/R)/ε) of 3.5 times higher than traditional ones, whereas Zhang et al. (2007b) have demonstrated a universal resistivity–strain dependency of poly(urethane)–urea elastomer matrix associated with amino-functionalized CNT. Knite et al. (2007) compared piezo-electrical properties of CPC composed of a poly(isoprene) matrix (PI) filled with either MWNTs or CNPs. They report a lower amplitude and a lack of reversibility of PI–CNT samples at large strains (40%) compared to their homologues PI–C N P, and suggest using CNT-based material only for small strain sensing applications, and preferably PI–CNP composites for large strain sensing applications. In a more recent work, Knite et al. (2009) show that PI–CNT can also be sensitive to pressure and chemicals. Another interesting illustration of strain sensing is given by Liu et al. (2007) who investigated the use of poly(L-lactic acid) PLLA (biocompatible and biodegradable) in association with CNTs for biomedical applications such as an implantable and wearable strain sensor (Fig. 25.24). They report gauge factors up to 30 soliciting sensors on a 4 points bending mode, depending on the initial resistance. Targeting long-term and large-scale infrastructure health monitoring, Loh et al. (2005) prepared thin CPC films, through an original layer-by-layer (LbL) assembly method. From 35 to 200 bi-layers of carbon nanotubes dispersed in either poly(vinyl alcohol) (PVA) or poly(sodium styrene-4-sulfonate) (PSS), were assembled in thin sensor films and tested upon monotonic deformation. A linear relation between resistance and strain was observed and they achieved a gauge factor (Ar/ε) of 4.52. On the other hand, these transducers exhibited interesting pH sensitivity, assuming promising application in corrosion monitoring.
25.24 Experimental set-up for remote measurement of strain inside the human body with PLLA-5wt% CNT (Liu et al., 2007).
CPC-based sensors should find many applications in manufacturing industries such as civil engineering, the automotive or aircraft industry as strain or pressure sensors for damage detection and load history. Aircraft parts are subjected to strain and pressure while in use. CPC-based sensors could impart information about surface strain state, appearance of microcracks and the evolution of straininduced damage. Fibre-reinforced composites for structural applications are particularly concerned since microscale damages have implications on the durability and performance of the composites. Structure health monitoring is one concern but sensors can also be used upstream for in-situ processing monitoring. For instance, during the cooling stage of thermoset-based composites, residual stress or strain can be monitored to control the prepared parts and to prevent defects and damage propagation. CPC-based sensors could be used in civil structures like bridges, tunnels, roads or buildings to follow the ageing process and detect defects induced by fatigue, impact or corrosion. Embedded sensors could help civil engineers in structure inspections and long-term monitoring.
Since early findings on electrical noses (Gardner and Barlett, 1992), lots of combinations of nanofillers and polymer matrices have been experimented to design new chemical transducers with increasing sensitivity and discriminating ability. Philip et al. (2003) and Abraham et al. (2004) first showed that CNTs dispersed in a PMMA matrix could efficiently sense a set of organic vapours (dichloromethane, chloroform and acetone) and that CNT oxidation by potassium permanganate with the help of a phase transfer catalyst was effective in considerably increasing transducers’ sensitivity. Since then, there has been great interest in developing carbon nanotubes-based transducers, which, compared to the more classical carbon nanoparticles, have a much larger aspect ratio and now comparable cost. Thus chemoresistive CNT-based CPCs have been used to design sensors for humidity (Chen et al., 2005; Su and Huang, 2006; Su and Wang, 2007; Yu et al., 2006; Kumar et al., 2010), acid vapours (Bavastrello et al., 2004), ammoniac (Zhang et al., 2006b; Woo et al., 2007), amines (Ma et al., 2006; Lee et al., 2008), warfare agents (Wang et al., 2008; Chang and Yuan, 2009), volatile anaesthetic agent sevoflurane (Chavali et al., 2008), and VOC (Zhang et al., 2005; Quercia et al, 2005; Sathanam et al., 2005; Wang et al., 2007; Wei et al., 2006). Thanks to the tremendous increase in sensitivity resulting from CNT functionalization by Philip et al. (2003), Wang et al. (2007) have experimented with CNT grafting with poly(styrene), poly(4-vinylpyridine), poly(styrene-b-4-vinylpyridine) and poly(styrene-co-4-vinylpyridine) by nitroxide mediated ‘living’ free radical polymerization which proved to lead to effective transducers for methanol, chloroform and tetrahydrofuran. Electrical responses are synthesized by plotting sensors’ sensitivity as a function of vapour content and nature, which gives a good overview of CPCs’ transducer performances. More recently, Castro et al. (2009) used the same concept by developing in-situ ring opening polymerization of poly (ε-caprolactone) to graft CNT according to the scheme in Fig. 25.25. This CNT surface treatment was found to improve both dispersion and sensitivity of transducers leading to stable, reproducible and quantitative electrical signals upon dynamic chemical solicitations as can be seen in Fig. 25.26. Interestingly, Fig. 25.27, that was obtained by taking the maximum value of Ar (the relative amplitude of the electrical response) when a stable value is reached, shows that the evolution of Ar with vapour volume fraction f (v/v) fits well with the LHC model (Equation 25.7). However, chloroform exhibits a totally different chemo-electrical behaviour. The origin of this peculiarity is well evidenced by observing in Fig. 25.28 the evolution of Ar with vapour nature, i.e. as a function of inverse of Flory-Huggins parameter χ12. Experimental data fit well with Equation 25.8 that predicts an exponential evolution of Ar with the inverse of χ12 (Equation 25.6). Nevertheless, if the model succeeds in predicting the increase in PCL–CNT transducers’ sensitivity with water, methanol, toluene and chloroform, it fails with tetrahydrofuran, suggesting the need to also take into account additional parameters like saturating pressure or size of vapour molecules to fully describe the phenomenon.
25.25 Scheme of CNT grafting by in-situ polymerization of poly (ε-caprolactone) from Castro et al. (2009).
25.26 Typical response of PCL-g-1%CNT to different concentrations of tetrahydrofurane (THF) in nitrogen (Castro et al., 2009).
25.27 Relative amplitude Ar of PCL-g-1%CNT transducers exposed to different volume fractions of water, methanol, toluene, THF and chloroform vapours in dynamic mode. Curves were fitted using LHC model without Langmuir contribution (HC) model (Equation 25.7). The right y axis is used for chloroform which has much larger amplitude than other vapours.
25.28 Evolution of PCL-g-CNT response with 1/χ12 (experimental data fit with Equation 25.8).
A more radical way to identify vapours is to use a statistical tool like principal component analysis (PCA) treatment developed for e-noses, which will combine all the responses of different transducers submitted to the same chemical vapours. The results presented in Fig. 25.28 clearly show the high potential of such data processing in providing recognition patterns for vapour identification. The common method of sensing investigations is to study each vapour separately but (Zhang et al., 2008) have started to address an important issue, sensing of mixed vapours. Using a single poly(styrene)-CNT transducer they have found that unexpectedly after exposure to vapour mixtures, resistance does not decrease as usual (PVC effect) but sharply increases (NVC effect). To explain this phenomenon, the formation of a parallel conductive water layer able to decrease sample resistivity has been assumed. Another case of the NVC effect has been reported by Li et al. (2007) with a PMMA–CNT transducer in the presence of methanol vapours and has been interpreted by a reduction reaction at the CNT surface which could take place due to the fact that they were chemically modified by NH3OH and HCl. Apart from the chemical nature and composition of CPC constituents that will determine transducers’ chemo-electrical properties, Lu et al. (2009b) have reported the influence of processing conditions that can be adjusted to tailor the sensors’ characteristics. Particularly, the spray layer-by-layer (LbL) deposition process hierarchically structures the sensitive film in 3D and provides an effective way to control initial resistance and thus the transducers’ sensitivity. Inkjet printing technique is another technique that has been used by Mabrook et al. (2008) which could be promising to fabricate thin CNT-filled CPC transducers.
To complement vapour sensing, CN-filled CPCs have also proved their ability to sense liquid chemicals. Narkis et al. (2000) and Srivastava et al. (2000) initiated interest in using co-continuous CPC filaments of PP/(PA6-CNP) and HIPS/(EVA–CNP) for liquid sensing. Later on, Segal et al. (2005) used co-continuous filaments of PP/TPU–CNP and PS/PANI-DBAS to sense methanol, ethanol and 1-propanol with good selectivity and sensitivity. They interpreted the sensing mechanism of these CPCs in terms of interphase debonding because the short response time of these CPCs suggested a preferential permeation through the interphase regions rather than a diffusion-controlled process through the bulk phases. Applying this principle to CNT-filled co-continuous CPC like PP/(PCL–CNT) gives the records of Fig. 25.29. Electrical responses are found to be reproducible and stable upon exposure to several organic solvents. The reason for the asymmetry of chemical cycles as compared to vapour sensing is that sorption is much quicker than desorption as molecules are invading the composite deeper and in much greater amounts than in the case of vapour, so that the clustering phase is reached, generating the matrix swelling (at least for good solvents). Nevertheless, the same kind of features as for vapour sensing have been observed by Kobashi et al. (2008, 2009), who recently investigated the liquid sensing properties of poly(lactic acid)/multi-walled carbon nanotube (PLA–CNT) CPC films and fibres (Fig. 25.30). Moreover, they have demonstrated the influence of the polymer matrix crystallinity on transducers’ resistance changes occurring during immersion. Possible target applications for such fibres would be leakage detection in tanks or containers (Pötschke et al., 2009). Another interesting issue of liquid sensing is the detection of biomolecules like glucose. (Teh and Lin, 2005) by associating dodecylbenzenesulfonate-doped poly(pyrrole) (DBS-PPy) with CNT found that transducers were responsive to oxidants, like hydrogen peroxide (H2O2), indirectly allowing detection of glucose (C6H12O6) which in the presence of oxygen, glucose is enzymatically oxidized at the anode to yield gluconolactone (C6H10O6) and hydrogen peroxide (H2O2).
Finally, Pioggia et al. (2007a, 2007b, 2008) report the development of so-called electronic tongues. Associating conductive polymer composite transducers in arrays, they were able to associate five compounds (including hydrophilic hydrogels consisting of blends of poly(vinyl alcohol) PVA and poly(allylamine) PAA loaded with CNT) with different chemical characteristics and gustative perceptions (sodium chloride, citric acid, glucose, glutamic acid and sodium dehydrocholate for salty, sour, sweet, umami and bitter, respectively). Impedance measurements were done at 150 Hz, during dipping in solution/drying in air cycles, and use of principal component analysis allowed a fairly good degree of discrimination.
Although many applications already use the e-noses and e-tongues principle to analyse qualitatively and quantitatively the composition of complex environments, there is still a need to develop combinations of new materials to widen the range of detectable substances. For example, in the foodstuffs industry, the sensors developed can be used for automatic quality control. Raw components are likely to undergo spoilage because of bacteria and humidity or temperatures issues and this technology offers the opportunity to follow food products during processing steps. This strategy could be used for quality control during production but also to monitor the food during storage, or even during the ageing process. Ageing in some cases is coupled with specific gas/vapour emissions that could be analysed and quantified to allow final product quality requirements. Thus, meat or seafood freshness, milk spoilage, fruit and vegetables’ quality monitoring appear promising areas for carbon nanotubes-based CPC sensors. In industrial plants, e-noses can be used to monitor pollutants, toxic substances, and detection of leaks in chemicals pipes or containers. Sensors can be used to analyse exhaust gases in industrial smokestacks. In collective housing, sensors could detect gas/vapours in air conditioning distribution main pipes to prevent air contamination throughout the building. Other applications concern safety areas for detection of biological and chemical weapons, detection of drugs or explosives. Security issues regarding terrorist actions are of special concerns and nanoscale sensors’ development could fulfil some military applications. Recently the study of nanostructured polymer sensor materials in the detection of potential chemical warfare agents and industrial chemical reagents like D M MP, D C M, T H F, CHCL3, MEK and xylene was reported (Wang et al., 2008; Chang and Yuan, 2009). They noticed the lack of adaptive sensors for selectivity among existing sensors, and proposed the use of MWNT-based CPC as fast-acting, inexpensive, simple to operate, sensitivity and selective sensor materials. High potential risks facilities, like airports, train stations or sport/cultural events that attract masses of people, could be equipped with such devices. Recent research has also focused on medical applications for non-invasive diagnostics (breath, urine, sweat). The objective is to detect the diseases in their early stage so that medical care can be quickly conducted. Great efforts are provided, for instance, in early stage detection of cancer markers since this disease will become one of the main causes of fatalities in the near future. One permanent challenging issue regarding these applications lies in decreasing the concentration range of the targeted molecules, i.e. from ppm to ppb.
The aim of this chapter was to give a quick overview of the different sensing applications of CNT–polymer composites (temperature sensors, thermistances, vapour sensors, e-noses, e-tongues, strain sensors, durability sensors). Many applications have been and are still being discovered in this very active field of research as attested by recent reviews on sensors’ materials (Huang and Choi, 2007; Kauffman and Star, 2008; Hatchett and Josowicz, 2008; Hierlemann and Gutierrez-Osuna, 2008; Bondavalli et al., 2009; Qureshi et al, 2009). Considering also the increasing demands related to smart functions for industrial but also domestic applications, there is still a lot of room for creativity and new developments. A good example of integration of advanced functionalities is illustrated by intelligent textiles (INTELTEX, 2010), capable of monitoring strain, temperature or chemicals, thus finding applications such as anomalous heart rate detection, real-time monitoring of body temperature or blood pressure measurement. Such smart textiles can also be used for the clothing of sportsmen operating in extreme conditions at low temperatures (mountain climbing or diving) where lightweight wearable and self-heating fabrics can be helpful. Multifunctional textiles are robust and can be used in the protective clothing of firemen or soldiers, enabling near skin temperature monitoring, detection of mechanical strain due to impact or penetration and detection of toxic volatiles in rescue or military operations conditions.
The authors would like to thank Przeyslaw Michalak for alternative current measurements and the European Commission for financial support (INTELTEX, 2010).
Alexopoulos, N.D., Bartholome, C., Poulin, P., Marioli-Riga, Z. Structural health monitoring of glass fiber reinforced composites using embedded carbon nanotube (CNT) fibers. Composites Science & Technology. 2009; 70(12):1733–1741.
Baughman, R.H., Cui, C.X., Zakhidov, A.A., Iqbal, Z., Barisci, J.N., Spinks, G.M., Wallace, G.G., Mazzoldi, A., De Rossi, D., Rinzler, A.G., Jaschinski, O., Roth, S., Kertesz, M. Carbon nanotube actuators. Science. 1999; 284:1340–1344.
Bavastrello, V., Stura, E., Carrara, S., Erokhin, V., Nicolini, C. Poly(2,5-dimethylaniline)–CNT nanocomposite: a new material for conductimetric acid vapours sensor. Sensors & Actuators B. 2004; 98:247–253.
Bellayer, S., Gilman, J.W., Eidelman, N., Bourbigot, S., Flambard, X., Fox, D.M., De Long, H.C., Trulove, P.C. Preparation of homogeneously dispersed trialkyl imidazolium compatibilized multiwalled carbon nanotube/polystyrene nanocomposites via melt extrusion. Advanced Functional Materials. 2005; 15(5):910–913.
Böger, L., Wichmann, M.H.G., Meyer, L.O., Schulte, K. Load and health monitoring in glass fibre reinforced composites with an electrically conductive nanocomposite epoxy matrix. Composites Science & Technology. 2008; 68:1886–1894.
Bonnet, P., Albertini, D., Bizot, H., Bernard, A., Chauvet, O. Amylose/SWNT composites: From solution to film: synthesis, characterization and properties. Composites Science & Technology. 2007; 67:817–821.
Bouvrée, A., Feller, J.F., Balnois, E., Salaun, A.C., Grohens, Y. Conductive polymer composites (CPC) transducers: Structuring of carbon and gold nanoparticles in thin layers for vapour sensing. In: Transducers Proceeding. France: Lyon; 2007:10–14. [June].
Celzard, A., McRae, E., Marêché, J.F., Furdin, G., Sundqvist, B. Conduction mechanisms in some graphite–polymer composites: effects of temperature and hydrostatic pressure. Journal of Applied Physics. 1998; 83(3):1410–1419.
Celzard, A., Pizzi, A., Fierro, V. Physical gelation of water-borne thermosetting resins by percolation theory: urea-formaldehyde, melamine-urea-formaldehyde, and melamine-formaldehyde resins. Journal of Polymer Science: Part B: Polymer Physics. 2008; 46:971–978.
Chavali, M., Lin, T., Wu, R., Luk, H., Hung, S. Active 433 MHz-W UHF RF-powered chip integrated with a nanocomposite m-MWCNT/polypyrrole sensor for wireless monitoring of volatile anesthetic agent sevoflurane. Sensors & Actuators A. 2008; 141:109–119.
Dalmas, F., Dendievel, R., Chazeau, L., Cavaillé, J.Y., Gauthier, C. Carbon nanotube filled polymer composites: numerical simulation of electrical conductivity in three-dimensional entangled networks. Acta Materialia. 2006; 54(11):2923–2931.
Deng, H., Skipa, T., Zhang, R., Lellinger, D., Bilotti, E., Alig, I., Peijs, T. Effect of melting and crystallization on the conductive network in conductive polymer composites. Polymer. 2009; 50:3747–3754.
Droval, G., Feller, J.F., Salagnac, P., Glouannec, P. Conductive polymer composites (CPC) with double percolated architecture of carbon nanoparticles and ceramics microparticles for high heat dissipation and sharp PTC switching. Smart Materials & Structures. 2008; 17(025011):1–10.
Du, D., Huang, X., Cai, J., Zhang, A. Amperometric detection of triazophos pesticide using acetylcholinesterase biosensor based on multiwall carbon nanotube–chitosan matrix. Sensors & Actuators B: Chemical. 2007; 127(2):531–535.
Etika, K.C., Liu, L., Hess, L.A., Grunlan, J.C. The influence of synergistic stabilization of carbon black and clay on the electrical and mechanical properties of epoxy composites. Carbon. 2009; 47:3128–3136.
Feller, J.F. Conductive polymer composites: influence of extrusion conditions on positive temperature coefficient effect (PTC) of poly(butylene terephthalate)/poly(olefin)-carbon black blends. Journal of Applied Polymer Science. 2004; 91(4):2151–2157.
Feller, J.F., Bruzaud, S., Grohens, Y. Influence of clay nanofiller incorporation on electrical and rheological properties of conductive polymer composite (CPC). Materials Letters. 2004; 58(5):739–745.
Feller, J.F., Linossier, I., Grohens, Y. Conductive polymer composites (CPC): comparative study of poly(ester)-short carbon fibres and poly(epoxy)-short carbon fibres mechanical and electrical properties. Materials Letters. 2002; 57(1):64–71.
Feller, J.F., Linossier, I., Levesque, G. Conductive polymer composites (CPC): comparative study of poly(ethylene-co-ethyl acrylate)-carbon black and poly(butylene terephthalate)/poly(ethylene-co-ethyl acrylate)-carbon black electrical properties. Polymers for Advanced Technologies. 2002; 13(10–12):714–725.
Ferrara, M., Neitzert, H.C., Sarno, M., Gorrasi, G., Sannino, D., Vittoria, V., Ciambelli, P. Influence of the electrical field applied during thermal cycling on the conductivity of LLDPE/CNT composites. Physica E: Low-dimensional Systems & Nanostructures. 2007; 37(1–2):66–71.
Gao, J.F., Yan, D.X., Huang, H.D., Dai, K., Li, Z.M. Positive temperature coefficient and time-dependent resistivity of carbon nanotubes (CNT)/ultrahigh molecular weight polyethylene (UHMWPE) composite. Journal of Applied Polymer Science. 2009; 114:1002–1010.
Grunlan, J.C., Gerberich, W.W., Francis, L.F. Lowering the percolation threshold of conductive composites using particulate polymer microstructure. Journal of Applied Polymer Science. 2001; 80(4):692–705.
Grunlan, J.C., Mehrabi, A.R., Bannon, M.V., Bahr, J.L. Water-based single-walled-nanotube-filled polymer composite with an exceptionally low percolation threshold. Advanced Materials. 2004; 16(2):150–153.
Imbert, C., Tchoreloff, P., Leclerc, B., Couarraze, G. Indices of tableting performance and application of percolation theory to powder compaction. European Journal of Pharmaceutics & Biopharmaceutics. 1997; 44(3):273–282.
INTELTEX, Development of intelligent multi-reactive textiles integrating nano-filler based CPC-fibres. European Integrated Project supported through the 6th Framework Programme for Research and Technological Development of European Commission (NMP2-CT-2006–026626), 2006–2010 available at: www.inteltex.eu
Jiang, M.J., Dang, Z.M., Xu, H.P. Significant temperature and pressure sensitivities of electrical properties in chemically modified multiwall carbon nanotube/methylvinyl silicone rubber nanocomposites. Applied Physics Letters. 2006; 89(182902):1–3.
Kang, I., Heung, Y.Y., Kim, J.H., Lee, J.W., Gollapudi, R., Subramaniam, S., Narasimhadevara, S., Hurd, D. Introduction to carbon nanotube and nanofiber smart materials. Composites: Part B: Engineering. 2006; 37:382–394.
Knite, M., Ozols, K., Zavickis, J., Tupureina, V., Klemenoks, I., Orlovs, R. Elastomer-carbon nanotube composites as prospective multifunctional sensing materials. Journal of Nanoscience & Nanotechnology. 2009; 9(6):3587–3592.
Kobashi, K., Tobias, V., Timo, A., Häußler, L., Pötschke, P. Investigation of liquid sensing mechanism of poly(lactic acid)/multi-walled carbon nanotube composite films. Smart Materials & Structures. 2009; 18(035008):1–15.
Krasteva, N., Guse, B., Besnard, I., Yasuda, A., Vossmeyer, T.S. Gold nanoparticle/PPI-dendrimer based chemiresistors’ vapor-sensing properties as a function of the dendrimer size. Sensors & Actuators B: Chemical. 2003; 92:137–143.
Krause, B., Ritschel, M., Täschner, C., Oswald, S., Gruner, W., Leonhardt, A., Pötschke, P. Comparison of nanotubes produced by fixed bed and aerosol CVD methods and their electrical percolation behaviour in melt mixed polyamide 6.6 composites. Composites Science & Technology. 2010; 70:151–160.
Kumar, B., Castro, M., Lu, J., Feller, J.F., Conducting polymer nanocomposites (CPC): nanocharacterisation of layer by layer sprayed PMMA-CNT vapour sensors by atomic force microscopy in current sensing mode (CS-AFM). Materials Research Society Symposium Proceeding, 2009. [1143-KK02–06.].
Kumar, B., Feller, J.F., Castro, M., Lu, J. Conductive bio-polymer nano-composites (CPC): chitosan-carbon nanotube transducers assembled via spray layer by layer for volatile organic compound sensing. Talanta. 2010; 81:908–915.
Lee, J.H., Kim, S.K., Kim, N.H. Effects of the addition of multi-walled carbon nanotubes on the positive temperature coefficient characteristics of carbon-black-filled high-density polyethylene nanocomposites. Scripta Materialia. 2006; 55:1119–1122.
Lisunova, M.O., Mamunya, Y.P., Lebovka, N.I., Melezhyk, A.V. Percolation behaviour of ultrahigh molecular weight polyethylene/multi-walled carbon nanotubes composites. European Polymer Journal. 2007; 43:949–958.
Liu, F., Zhang, X., Li, W., Cheng, J., Tao, X., Li, Y., Sheng, L. Investigation of the electrical conductivity of HDPE composites filled with bundle-like MWNTs. Composites A: Applied Sciences & Manufacturing. 2009; 40:1717–1721.
Liu, Y., Chakrabartty, S., Gkinosatis, D.S., Mohanty, A.K., Lajnef, N., Multi-walled carbon nanotubes/Poly(L-lactide) nanocomposite strain sensor for biomechanical implants. IEEE Biomedical Circuits & Systems Conference, Montreal, Canada. 2007:119–122.
Loh, K.J., Kim, J., Lynch, J.P., Wong, N., Kam, S., Kotov, N.A. Multifunctional layer-by-layer carbon nanotube–polyelectrolyte thin films for strain and corrosion sensing. Smart Materials and Structures. 2007; 16:429–438.
Loh, K.J., Lynch, J.P., Kotov, N.A., Conformable single-walled carbon nanotube thin film strain sensors for structural monitoring. Proceedings of the 5th International Workshop on Structural Health Monitoring, Stanford, C A, U S A. 2005:12–14. [September].
Lonergan, M.C., Freund, M.S., Severin, E.J., Doleman, B.J., Grubbs, R.H., Lewis, N.S. Array-based vapour sensing using chemically sensitive polymer composite resistors. Chemistry of Materials. 1996; 8(9):2298–2312.
Lu, J., Castro, M., Kumar, B., Feller, J.F. Thermo- and chemo-electrical behaviour of carbon nanotube filled co-continuous conductive polymer nanocomposites (CPC) to develop amperometric sensors. Materials Research Society Symposium Proceeding. 2009; 1143:KK05–KK14.
Lu, J., Kumar, B., Castro, M., Feller, J.F. Vapour sensing with conductive polymer nanocomposites (CPC): polycarbonate-carbon nanotubes transducers with hierarchical structure processed by spray layer by layer. Sensors & Actuators B: Chemical. 2009; 140:451–460.
Lu, J., Park, B.J., Kumar, B., Castro, M., Choi, H.J., Feller, J.F. Hybrid vapour sensor: polyaniline nanobead-carbon nanotube architecture with tuneable chemo-electrical behaviour. Nanotechnology. 2010; 21(255501):1–10.
Ma, X., Zhang, X., Li, Y., Li, G., Wang, M., Chen, H., Mi, Y. Preparation of nano-structured polyaniline composite film via “carbon nanotubes seeding” approach and its gas-response studies. Macromolecular Materials & Engineering. 2006; 291:75–82.
Mabrook, M.F., Pearson, C., Jombert, A.S., Zeze, D.A., Petty, M.C. The morphology, electrical conductivity and vapour sensing ability of inkjet printed thin films of single-wall carbon nanotubes. Carbon. 2008; 47(3):752–757.
Mamunya, Y.P., Muzychenko, Y.V., Lebedev, E.V., Boiteux, G., Seytre, G., Boulanger, C., Pissis, P. PTC effect and structure of polymer composites based on polyethylene/polyoxymethylene blend filled with dispersed iron. Polymer Engineering & Science. 2007; 47:34–42.
Martin, C.A., Sandler, J.K.W., Shaffer, M.S.P., Schwarz, M.K., Bauhofer, W., Schulte, K., Windle, A.H. Formation of percolating networks in multi-wall carbon nanotube-epoxy composites. Composites Science & Technology. 2004; 64(15):1236–2309.
McGraw-Hill, Encyclopedia of Science & Technology, 5th edn, 2010. published online: http://www.answers.com/library/Sci%252DTech+Encyclopedia-cid-85017
Mu, M., Walker, A.M., Torkelson, J.M., Winey, K.I. Cellular structures of carbon nanotubes in a polymer matrix improve properties relative to composites with dispersed nanotubes. Polymer. 2008; 49:1332–1337.
Nofar, M., Hoa, S.V., Pugh, M.D. Failure detection and monitoring in polymer matrix composites subjected to static and dynamic loads using carbon nanotube networks. Composites Science & Technology. 2009; 69:1599–1606.
Park, J.M., Kim, D.S., Lee, J.R., Kim, T.W. Non destructive damage sensitivity and reinforcing effect of carbon nanotube/epoxy composites using electro-micromechanical technique. Materials Science & Engineering C. 2003; 23:971–975.
Park, J.M., Kim, D.S., Kim, S.J., Kim, P.G., Yoon, D.J., DeVries, K.L. Inherent sensing and interfacial evaluation of carbon nanofiber and nanotube/epoxy composites using electrical resistance measurement and micromechanical technique. Composites: Part B. 2007; 38:847–861.
Pillin, I., Feller, J.F., Pimbert, S., Levesque, G. Conductive poly(ester)/poly(olefin)/carbon black composites: influence of glass transition temperature and crystallinity of the poly(ester) matrix on the electrical properties. Plastics Rubbers & Composites. 2002; 31(78):300–306.
Pioggia, G., Di Francesco, F., Marchetti, A., Ferro, M., Ahluwalia, A. A composite sensor array impedentiometric electronic tongue: Part I. Characterization. Biosensors & Bioelectronics. 2007; 22:2618–2623.
Pioggia, G., Di Francesco, F., Ferro, M., Sorrentino, F., Salvo, P., Ahluwalia, A. Characterization of a carbon nanotube polymer composite sensor for an impedimetric electronic tongue. Microchimca Acta. 2008; 163:57–62.
Pötschke, P., Abdel-Goad, M., Alig, I., Dudkin, S., Lellinger, D. Rheological and dielectrical characterization of melt mixed polycarbonate-multiwalled carbon nanotube composites. Polymer. 2004; 45:8863–8870.
Pötschke, P., Andres, T., Villmow, T., Pegel, S., Brünig, H. Liquid sensing properties of fibres prepared by melt spinning from poly(lactic acid) containing multiwalled carbon nanotubes. Composites Science & Technology. 2009; 70:343–349.
Pötschke, P., Bhattacharyya, A.R., Janke, A. Melt mixing of polycarbonate with multiwalled carbon nanotubes: microscopic studies on the state of dispersion. European Polymer Journal. 2004; 40(1):137–148.
Santhanam, K.S.V., Sangoi, R., Fuller, L. ‘A chemical sensor for chloromethanes using a nanocomposite of multiwalled carbon nanotubes with poly(3-methylthiophene). Sensors & Actuators B: Chemical. 2005; 106:766–771.
Segal, E., Tchoudakov, R., Mironi-Harpaz, I., Narkis, M., Siegmann, A. Chemical sensing materials based on electrically conductive immiscible polymer blends. Polymer International. 2005; 54:1065–1075.
Shevade, A.V., Ryan, M.A., Homer, M.L., Manfreda, A.M., Zhou, H., Manatt, K.S. Molecular modelling of polymer composite–analyte interactions in electronic nose sensors. Sensors & Actuators B: Chemical. 2003; 93:84–91.
Tartarin, R., Pajot, S. A percolation model of the breakdown of a Soviet-type economy: repression, disequilibria and market diffusion. In: The Transformation of Economic Systems: Discontinuous Jumps and Continuous Adaptations, Second Budapest EACES Studies Workshop. Budapest University of Economic Sciences; 1996:16–17. [December.].
Villmow, T., Pötschke, P., Pegel, S., Häussler, L., Kretzschmar, B. Influence of twin-screw extrusion conditions on the dispersion of multi-walled carbon nanotubes in a poly(lactic acid) matrix. Polymer. 2008; 49:3500–3509.
Wang, X.J., Chung, D.D.L. Real time monitoring of fatigue damage and dynamic strain in carbon fibres polymer matrix composite by electrical resistance measurement. Smart Materials & Structures. 1997; 6:504–508.
Wichmann, M.H.G., Buschhorn, S.T., Böger, L., Adelung, R., Schulte, K. Direction sensitive bending sensors based on multi-wall carbon nanotube/epoxy nanocomposites. Nanotechnology. 2008; 19(475503):1–5.
Woo, C.S., Lim, C.H., Cho, C.W., Park, B., Ju, H., Min, D.H., Lee, C.J., Lee, S.B. Fabrication of flexible and transparent single-wall carbon nanotube gas sensors by vacuum filtration and poly(dimethyl siloxane) mold transfer. Microelectronic Engineering. 2007; 84:1610–1613.
Yu, H., Cao, T., Zhou, L., Gu, E., Yu, D., Jiang, D. Layer by layer assembly and humidity sensitive behaviour of poly(ethyleneimine)/multiwall carbon nanotube composite films. Sensors & Actuators B. 2006; 119:512–515.
Zhang, B., Fu, R.W., Zhang, M.Q., Dong, X.M., Lana, P.L., Qiu, J.S. Preparation and characterization of gas sensitive composites from multiwalled carbon nanotubes/polystyrene. Sensors & Actuators B. 2005; 109:323–328.
Zhang, B., Dong, X., Fu, R., Zhao, B., Zhang, M. The sensibility of the composites fabricated from polystyrene filling multi-walled carbon nanotubes for mixed vapors. Composites Science & Technology. 2008; 68:1357–1362.
Zribi, K., Feller, J.F., Elleuch, K., Bourmaud, A., Elleuch, B. Conductive polymer composites obtained from recycled poly(carbonate) and rubber blends for heating and sensing applications. Polymers for Advanced Technology. 2006; 17(9–10):727–731.