Вы находитесь на странице: 1из 35

Transworld Research Network 37/661 (2), Fort P.O.

, Trivandrum-695 023, Kerala, India

Biopolymers and Nanocomposites as Studied by Dielectric Spectroscopy, 2009 : ISBN: 978-81-7895-399-1 Editor: Galder Kortaberria

Nanocomposites based on polymer matrix and carbon nanotubes as studied by dielectric spectroscopy
M. Blanco, G. Kortaberria, A. Jimeno, P. Arruti, A. Tercjak and I. Mondragon Materials + Technologies Group, Chemical Engineering Department Universidad del Pas Vasco/Euskal Herriko Unibertsitatea, Plaza Europa 1 20018 Donostia, Spain

Abstract
In this review work, the use of dielectric spectroscopy as a tool for analyzing electrical and dielectric properties of different nanocomposites based on both thermosetting or thermoplastic polymer matrices and carbon nanotubes (CNTs) is reviewed. The insulator-to-conductor transition (percolation threshold) of the nanocomposites, as found by several authors, is presented for a wide variety of nanocomposites. The application of the percolation theory
Correspondence/Reprint request: Dr. G. Kortaberria, Materials + Technologies Group, Chemical Engineering Department, Universidad del Pas Vasco/Euskal Herriko Unibertsitatea, Plaza Europa 1, 20018 Donostia, Spain. E-mail: galder.cortaberria@ehu.es

M. Blanco et al.

and its power laws to those nanocomposites is presented as carried out by several authors, together with the theoretical parameters obtained for different types of nanocomposites. Meaningful explanations of the obtained parameter values are given for each system. The effect of filler amount and/or chemical modification on the properties and theoretical parameters is presented. The effect of CNTs on the molecular dynamics of the polymer matrix is also reviewed as reported by several authors.

Introduction
Carbon nanotubes (CNTs) were discovered in 1991 by Sumio Iijima [1]. From that time the CNTs have generated great interest for several applications based on their electrical and mechanical properties. They posses one hundred times the tensile strength of steel, thermal conductivity better than the purest diamond and electrical conductivity similar to copper, but with the ability to carry much higher currents [2]. CNTs include two variants, singlewalled and multiwalled structures as can be seen in Figure 1. A single-wall carbon nanotube (SWNT) is a rolled-up tubular shell of graphene sheet which is made up of benzene-type hexagonal rings of carbon atoms. SWNT have diameters typically between 0.4-4 nm [3]. The chirality (the chiral angle between hexagons and the tube axis) of the nanotubes is a key parameter. Depending on the chirality, SWNTs can exhibit distinct physical properties being either metals or semiconductors nanotubes (Fig. 1BD). These cylindrical sheets are capped by hemispherical ends. The closure of the cylinder is result of pentagon inclusion in the hexagonal carbon network of the nanotube walls during the growth process.

Figure 1. SWCNTs with different chiralities: armchair (B), zigzag (C) and chiral (D) tubes, respectibely. MWCNT formed by graphene sheets rolled up into concentric cylinders (E). (Reproduced with permission of [2]. Copyright 2006, Springer)

Polymer/CNT nanocomposites by dielectric spectroscopy

The multiwall carbon nanotubes (MWNT) (Fig. 1E) consist on a stack of graphene sheets rolled up into concentric cylinders with a constant layer spacing of 0.34 nm [3-5]. MWNTs tend to have diameters in the range 10100 nm [3]. The walls of each layer of the MWNT, i.e. the graphite basal planes, are parallel to the central axis. MWNTs are reported always electrically conductive and posses conductivities similar to copper [6]. The addition of carbon nanotubes to both thermosetting and thermoplastic polymers improved not only the mechanical properties but also the thermal stability and the electrical conductivity of the resulting composites [7-11]. CNT containing nanocomposites present percolation threshold behaviour. That is, at low concentration, conducting fillers are dispersed within polymeric matrix as insolated clusters and the electrical properties of the composites are dominated by the matrix. Beyond a critical concentration of conducting filler, known as percolation threshold, filler clusters begin to connect each other to form a filler network throughout the entire composite, resulting in several orders of magnitude increase in the conductivity and dielectric properties of the composite. The transition from isolated cluster to connected network of conducting filler is referred to as the percolation transition [12]. The inclusion of conductive fillers can greatly alter the dielectric characteristics of the polymer medium in which they are dispersed. Then, the addition of CNTs will influence the dielectric properties of the obtaining nanocomposites. The dielectrical response of the composite is strongly dependent on the volume fraction, the size, and the shape of the reinforced conductive phase as well as the intereactions between polymer and nanotubes [13]. In this regard, dielectric spectroscopy has been demonstrated a reliable method to study the state of dispersion of CNT in different matrices [14-29]. Dielectric properties and dynamics of nanocomposites based on polymeric matrices and CNTs as studied by dielectric spectroscopy are presented in the present review. In the first section a review of the theoretical background is presented. In the second one main findings of several authors regarding dielectric properties of nanocomposites based on thermoplastic matrices and CNTs are presented, ending with the third section in which dielectric properties of thermoplastic matrix-based nanocomposites are presented.

1. Theoretical background

Complex conductivity *()=()i() and complex permittivity *()= ()i() (being angular frequency) are one of the most important dielectric parameters reported by several authors in the dielectric study of CNT-based nanocomposites. Two types of models have been developed for the description of the frequency dependence of *() and *() in percolating systems: (i) the equivalent circuit model, which treats a percolation system as

M. Blanco et al.

a random mixture of resistors and capacitors (alternatively: resistorsisolators, resistorssuperconductors, conductorinsulator, etc.) [30-38]; and (ii) models based on charge carrier diffusion on percolation clusters [36, 39-49]. For twocomponent systems- percolation clusters of high conductivity (1) embedded in a matrix of considerable lower conductivity (2)- macroscopic effective conductivity near the percolation threshold pc can be written in the scaling form [23, 30, 34-37]:

( p, ) 1 ( p pc )t [( 2 / 1 )( p pc ) ( t + s ) ]

(1)

where p is the concentration of conducting filler, + and denote scaling functions for p>pc and p<pc, respectively, s and t are critical exponents. When treating a percolation system as a random mixture of resistors (R) and capacitors (C) having a characteristic relaxation time = 1/0 = RC, scaling (1) for the complex conductivity near the percolation threshold takes the form [35-37]:

* ( p, ) 1/ R( p pc )t [(i / 0 )( p pc ) (t + s ) ]

(2)

One of the main physical content of this key scaling law is the existence of a critical time

c ( p)

1 0 ( p pc ) (t + s ) c ( p)

(3)

that diverges as the percolation threshold is approached from both sides. In the static or DC (i.e. zero-frequency) limit the capacitors become insulators and the RC model becomes the conductorinsulator model [35]. Dependence of DC conductivity (DC) on filler content (p) near the percolation threshold pc can be described [30-32, 36] by power laws:

DC ( p ) ( p pc ) s
for concentrations of filler below percolation threshold (p < pc) and:

(4)

DC ( p) ( pc p )t

(5)

above the percolation threshold (p > pc). For the frequency dependence of AC conductivity AC and the real part of the permittivity at the percolation threshold Bergman and Imry [37] derived the following laws:

Polymer/CNT nanocomposites by dielectric spectroscopy

AC ( ) ( t / t + s ) ' ( ) ( )( s / t + s )

(6) (7)

The composition dependence of the dielectric constant was found to be [37, 38]:

' ( p ) ( p pc ) s

(8)

for both p > pc and p < pc. A wide variety of approaches have been applied for the determination of these exponents. Straley [50] calculated for twodimensional (2D) resistor lattice s = t = 1.10 0.05 and for three-dimensional (3D) resistor lattice s =0.70 0.05, t = 1.70 0.05. Fish and Harris [51] applied the low-density series expansions for analysing of random resistor network and obtained s = t = 1.43 0.02 for 2D and t = 1.95 0.03 for 3D. Adler [52, 53] reanalysed the data of Fish and Harris and calculated with series expansions method t =1.31 0.10 for 2D and t = 2.02 0.05 for 3D. AlexanderOrbach conjecture [54] gives for 2D t = 4/3 1.264. Gingold and Lobb [55] employed finite-size scaling to analyse the critical behaviour of large three-dimensional random resistor lattices and reported for this case t = 2.003 0.047. Derrida et al. [56, 57] and Herrmann et al. [58] used transfermatrix calculations for analysing of random resistor network and obtained s = t = 1.30 0.01 for 2D and s = 0.75 0.04, t = 1.9 0.1 for 3D. Zabolitzky [59] used large-scale Monte-Carlo simulations and obtained results, which are very close to those from Derrida. As mentioned above, an alternative to the equivalent circuit model are models based on charge carrier motions on (fractal) percolation structures. This type of models do not directly include capacitances (e.g. to express gaps between clusters), but polarization effects are included by the finite time needed for the charge carriers to path through the percolation clusters. The frequency dependence of the conductivity is described by diffusion of conducting sites on percolation clusters. The correlation time , which a charge carrier needs to traverse a cluster of correlation length , is given [35, 40-44] by:

d ( p pc ) d

(9)

The exponent dw is the effective fractal dimensionality of the random walk (diffusion exponent) and is the exponent of the concentration dependence

M. Blanco et al.

of the correlation length. The numerical value of the correlation length exponent for three-dimensional percolation was found [45, 46, 55] 0.88. Earlier for 2D-model it was found [60, 61] = 4/3. The value of increases with concentration for p < pc and decreases for p > pc. For frequencies < ( = 1/) the length scale of the diffusive movements of charge carrier is considerably larger than the average cluster size and the charge carriers can explore different clusters within one period, i.e. the diffusion is normal and the conductivity becomes constant. For frequencies above the charge carriers visit only parts of the percolation cluster within one period and anomalous diffusion at the fractal percolation clusters takes place. The composition dependence of the static permittivity near the percolation threshold can be described [41-44] by:

s ( p pc ) 2 +

(10)

for both, p < pc and p > pc, where is the percolation exponent, which characterises the probability that a site belongs to the infinite cluster (for p > pc). It is important to note, that (10) has the same form as (8) with exponent s = 2. The numerical value for this critical exponent for 3D was found to be s = 1.33 0.01 [53-56].

2. Nanocomposites based on CNTs and thermoplastic matrices


Ptschke et al. [23] studied the influence of the CNT content on the complex permitivity spectra and the frequency dependent conductivity for MWNT/polycarbonate (PC) composites in the frequency range between 104 and 107 Hz. They found significant different behaviour in complex permittivity and conductivity spectra for matrices with MWNT contents lower than 1.5 wt % and matrices with higher amounts, as shown in Figure 2. In addition, a change in DC of more than ten decades between 1 and 1.5 wt % MWNT was reported, as it can be seen in Figure 3. That is, below this concentration the composites were very resistant to electrical flow, whereas above this value they were conductive. The value of 1.44 for pc and the value of 2.1 for the percolation critical exponent t obtained from fitting the composition dependence of the DC were found to be in accordance with the percolation theory for a three dimensional percolation structure. The static permitivity increased with MWNT content below the percolation threshold whereas its value did not decrease above pc. They addressed it to remaining (micro-) capacitors formed at the contacts between CNT aggregates or bundles. In addition, they analyzed the effect of the melt

Polymer/CNT nanocomposites by dielectric spectroscopy

Figure 2. The effect of frequency on the (a) real part , (b) imaginary part of the complex permittivity, and (c) AC (expressed also as ) for PC composites of varying MWNT content. (Reproduced with permission of [23]. Copyright 2003, Elsevier Ltd)

Figure 3. DC versus MWNT concentration p in PC for the samples shown in Figure 2. The insert shows the values of DC above the pc versus (p pc). (Reproduced with permission of [23]. Copyright 2003, Elsevier Ltd)

M. Blanco et al.

processing conditions (mixing time and screw speed) on the distribution of MWNT in the matrix. They found that the increase of the mixing time could improve the distribution of MWNT in the polymer matrix cosiderably. At concentration values near pc, the increase of the mixing time could transform the system from a non percolated into a percolated one. The proposed explanation was that longer mixing time promoted polymer diffusion between MWNT aggregates or the bundles of the masterbatch. DC was found to be sensitive to the changes of the critical nanotube content at pc with mixing time. The screw speed acted differently for compositions below and above the percolation threshold pc. Below pc, the increase in screw speed induced better dispersion, whereas above pc the efficient transfer of mechanical energy due to nanotube network formation could lead to the enhanced breakage of the MWNT and the decrease of DC. They concluded remarking the fact that dielectric spectroscopy is a viable method for the characterization of the state dispersion of MWNT in nanocomposites. Wang et al. [19], analyzed the dielectric properties of nanocomposites based on untreated MWNT and poly(vinylidene fluoride) (PVDF). By measuring AC they found a conductor-insulator translation at a content of MWNT between 0.016 and 0.02. They applied percolation theory to their systems to describe the conductivity of the nanocomposite near the conductorinsulator transition, applying an equation analogous to (4) for AC. The best fits gave pc = 0.0161 and a value of 0.85 for the critical exponent. They found that pc was 10 times smaller than that of common two-phase random composites, while the critical exponent exhibited the same value with the universal ones (0.8-1). They pointed out that the Swiss-cheese model [12] based on a medium made of conducting fillers embedded in an insulating matrix, could be applied for their nanocomposites with p pc, where the conduction was controlled by interfiller tunneling. They pointed out that the percolation threshold was quite low compared with that for polymer/conductive filler composites due to a high aspect ratio and the unique physical properties of MWNT. The low percolation threshold rendered the nanocomposites highly flexible. The also found a large enhancement of the dielectric constant near pc. The attained value for the percolation nanocomposite at room temperature was around 200, about 20 times higher than that of PVDF. The result suggested that the dielectric constant might be improved without the need to chemically functionalize the CNTs. They fitted the variation of with filler content by the power law of equation (8). They obtained a dielectric critical exponent of 0.31, in accordance with the universality of the percolation theory and a percolation thereshold of pc = 0.016. Both critical exponents were lower than the universal 3D lattice value. They pointed out that the increase of the dielectric constant could be due to the formation of minicapacitor networks in the nanocomposites when the concentration of MWNT increased. They also noted that the of

Polymer/CNT nanocomposites by dielectric spectroscopy

the percolation nanocomposite was as high as 0.8 at room temperature and low frequency. The value was still mantained when the percolation nanocomposite with a high dielectric constant was employed as a dielectric in some application fields. They pointed out that for the percolation composites, used to increase rapidly, with both features associated with the approach of percolation [12, 62, 63]. In comparison to the polymer matrix, the nanocomposites also exhibited relatively high , increasing with MWNT concentration due to their electrically conductive character. Furthermore, towards higher frequency, for the nanocomposite was lower than 0.4 with an increase in frequency, concluding that this fact was very attractive for them because the nanocomposites might be used to create some devices such as high charge-storage capacitors with various shapes. Nam et al. [18] studied dielectric and other physical properties of nanocomposites based on semicrystalline PVDF and MWNT. The evolution of with the MWNT content showed an experimental pc between 2 and 2.5 wt%. Similar tendency was also found in thermal and electrical conductivity. The electrical conductivity increased from 10-14 to 1 S/cm with increasing the content from 0 to 5 wt% but did not increase with a further raise of filler content. This indicated that there was conductivity saturation from a critical content because of the formation of an infinite cluster [64]. They also found that MWNT acted as nucleating agent. The melt crystallization temperature of PVDF increased with MWNT content because the MWNT phases dispersed finely within the PVDF matrix promoted heterogeneous nucleation. Supercooling required for PVDF crystallization and the size of spherulites in PVDF decreased with the increase in MWNT content. Park et al. [17] analyzed the electrical and dielectric properties of MWNT/poly(methyl methacrylate) (PMMA) nanocomposites prepared by melt mixing. They employed MWNT with two different pretreatments, concentrated acid treatment and thermal treatment. They found that the electrical and dielectrical properties of the composites were greatly affected by the pretreatment of the nanotube. They first measured , , and DC in the frequency range from 10-2 to 106 Hz. For acid pretreated MWNT/PMMA nanocomposites, there was no significant change of and , with MWNT content and frequency. Nanocomposites containing more than 3 wt% of thermally treated MWNT showed decreases of and , with increase of frequency. A significant difference between two kinds of MWNT was also observed in DC. As shown in Figure 4, the pc clearly existed around 3 wt% of thermally treated MWNT. At this concentration, conductivity increased sharply by ten orders of magnitude. They fitted DC data to (5) in order to estimate t and pc. They obtained the following values for the thermally treated MWNT: pc = 3.26 wt% and t = 1.18. For acid pretreated ones, there was almost

10

M. Blanco et al.

Figure 4. DC versus weight fraction of MWCNT in the nanocomposites. The straight line is a linear regression fit to the data points. (Reproduced with permission of [17]. Copyright 2005, Polymer Society of Korea)

no difference in conductivity respect to pure PMMA due to a poor dispersion of these nanotubes as a result of the excessive association of functional groups generated during the treatment, causing aggregated clusters in the matrix. They concluded pointing out that the obtained percolation threshold was not comparable to the lower values reported in the literature for other polymers. They related this fact to the preparation method of the nanocomposites. Most polymer/MWNT nanocomposites were prepared by solvent-assisted processes. In terms of mixing efficiency, they concluded that solvent-assisted process was superior to melt mixing, where viscosity of polymer melt was high enough to prevent sufficient mixing and there is lack of interfacial mediator between two phases. Clayton et al [21] studied dielectric properties and molecular dynamics of nanocomposites based on PMMA and 0.26 wt% of SWNT by dielectric spectroscopy. Nanocomposites were prepared via in situ polymerization induced by heat, ultraviolet light or ionizing radiation. By dielectric spectroscopy they measured , and tan in the frequency range from 1 to 105 Hz at temperatures ranging from -150 C to 200 C. They found two relaxations in PMMA, the and relaxations. The former was related to the segmental motion of the polymer main chain, whereas the second one was related to the hindered rotation of the ester side group attached to the main chain. PMMA also presented an relaxation, as result of main-chain slippage with the secondary relaxation [65]. In the nanocomposite with CNT they found the same relaxations. From Arrhenius plots they obtained activation energies for the relaxation in pure PMMA and in the nanocomposites synthetized by

Polymer/CNT nanocomposites by dielectric spectroscopy

11

different polymerization methods. Activation energies were very similar in all the cases, in agreement with activation energies found previously in the literature for PMMA and PMMA/CNT nanocomposites [65-69]. They pointed out that side-group rotation of the samples obtained with different polymerization methods was not hindered by the addition of the CNTs. They pointed out that the extent of alignment of the dipoles in an electric field could be determinated from the dielectric constant data. Dielectric constant data obtained for nanocomposites were higher than those for pure PMMA synthetized by the same methods. The UV-initiated polymer samples exhibited the largest increase, followed by the heat-induced and radiation-induced. They correlated the dielectric constant values with the refractive index, in order to understand both the electronic nature of the polymer and the effect of CNTs. According to Krevelan [70], if the sample is a non-polar insulator, the dielectric constant for low frequencies can be expressed by Maxwells equation = n2, where n is the refractive index. A large different between the squared refractive index and the dielectric constant is a result of permanent dipoles and the semi-conductive character of the sample [70]. They found the largest difference in the UV-initiated nanocomposite. The CNTs, with onethird metallic character and two-thirds semi-conductive character, contributed to the conductive nature of the nanocomposite. The refractive indexes of nanocomposites were found to be slightly lower than their neat counterparts. They concluded pointing out that while dielectric properties were significantly affected by the CNTs, optical properties were less sensitive to the filler. Zhu et al [26] prepared nanocomposites based on polyimide (PI) and MWNTs by mixing of poly(amic acid) (PAA) solution and MWNT/DMAc suspension followed by mixture casting, evaporation and thermal imidization. To increase the chemical compatibility between matrix and filler, MWNTs were modified with mixed strong acid. They measured electrical conductivity and dielectric constant of the nanocomposites at 10 kHz. By measuring the electrical conductivity they found that the increase was not obvious when the MWNT content was lower than 7 wt%. The conductivity showed a sharp increase in the range 7-10 wt%, illustrating that the percolation threshold in the conductivity of the nanocomposites was in this range, as it can be seen in Figure 5. They pointed out that compared with other polymer/MWNT nanocomposites, their nanocomposites showed lower electrical conductivity [71-73]. They gave two possible reasons for this behavior: the fact that MWNTs were well dispersed in the matrix, with a higher amount of MWNTs needed to form conductive path; and the shorter length of acid-treated MWNTs, with the need of more MWNTs for them to connect each other and reach percolation threshold. They also measured of the nanocomposites. It

12

M. Blanco et al.

Figure 5. Variation of electrical conductivity of the PI/MWCNT nanocomposites with the content of MWCNTs, at 10 kHz. (Reproduced with permission of [26]. Copyright 2006 Elsevier Ltd)

reached the value 60 when the MWNT content was 10 wt%, 17 times of the pure PI. Moreover, imilar to the change in conductivity, increased sharply at 7 wt% content. The enhancement was explained according to the percolation theory, applying (8). For the nanocomposites studied, they found that pc was around 0.07 and the critical exponent around 1.14. The percolation threshold was found to be lower than that of the two-phase random composite, with spherical fillers [12]. They pointed out that their MWCNTs were tube-shaped, which led to the formation of conductive path at lower content. In the other hand, their pc value was much lower than that of PI/CNT nanocomposites found by Ounaies et al [74] (0.05 vol%), who measured electrical properties of nanocomposites based on pure, non-modified SWCNTs. They explained this fact in terms of interactions between matrix and nanotubes due to their chemical modification. They pointed out that the stronger the interactions, the higher the percolation threshold. ee et al [25] studied dynamics of SWNT/PI nanocomposites by dielectric relaxation spectroscopy (DRS) and dynamic mechanical spectroscopy (DMS), by using both pristine (SWNT-p) and chemically-modified nanotubes (SWNTf) in order to study the effect of modification on dynamics. Nanocomposites were prepared by mixing desired amounts of PI and SWNTs in toluene at room temperature. They pointed out that modification of nanotubes was very important to dynamics, as evidenced by the difference in the normal mode relaxation in nanocomposites with pristine (PISP) and functionalized (PISF) SWNTs. The difference can be seen in Figure 6.

Polymer/CNT nanocomposites by dielectric spectroscopy

13

Figure 6. Dielectric loss in the frequency domain at 253 K for (a) PISP and (b) PISF nanocomposites with SWCNT loading as a parameter. (Reproduced with permission of [25]. Copyright 2007, Elsevier Ltd)

The incorporation of SWNT-p led to an increase in the intensity of over the entire frequency window by as much as four orders of magnitude at 2 wt% loading, as shown in Figure 6a. The normal mode process (N) in PISP appeared masked by conductivity but remained visible as a shoulder. After a careful examination of the spectra they revealed that the time scale of the N process decreased (shifted to higher frequency) with increasing filler loading. They advanced a scenario whereby the PI chains in PISP nanocomposites were partitioned between bulk and confined domains. The observed decrease in the time scale was envisioned as a consequence of the strong effect on dynamics of the PI chains that were confined within the SWNT-p agglomerates. PISP nanocmoposites were characterized by a difference in the average relaxation time for the normal mode process as measured by DRS and DMS. The discrepancy was attributed to the fact that the DRS spectra were affected by the presence of the confined PI phase whereas the DMS measurement captured the bulk response. A very different picture emerged from the spectra of PISF nanocomposites as it is seen in Figure 6b. The increase in loss intensity was much less pronounced and the normal mode process remained clearly visible at lower loadings. Moreover, they observed an increase in the time scale of the normal mode process in the nanocomposite with 0.37 wt% SWNT-f, indicating a decrease in chain mobility. PISF nanocomposites were characterized by better dispersion of SWNT-f in the PI matrix and stronger interactions between these two components due to enhanced compatibility [75-77]. Those interactions promoted the elongation of PI chains on or near the nanotube surface, The resulting increase in the time scale of the normal mode process

14

M. Blanco et al.

was reflected in the shift of the spectra to lower frequency. They found an excellent agreement in the average relaxation time for the normal mode process calculated from DRS and DMS data, due to the good dispersion of SWNT-f in the matrix. They also measured the conductivity of the nanocomposites in the frequency region from 10-1 to 106 Hz. They pointed out that the frequency dependence of conductivity could be expressed by a sum of the DC component and the frequency power law terms according to [78, 79]:

( ) = DC + Ai si

(11)

where Ai is temperature dependent parameter in ith frequency domain and si is the frequency exponent in ith frequency domain. Figure 7 shows conductivity in the frequency domain at several temperatures for PISP 0.37 wt% nanocomposite. Conductivity spectra were characterized by frequency independent DC ionic region at low frquency and a frequency-dependent region (above some critical fc frequency [80-82]). The DC conductivity domain became progressively smaller with decreasing temperature and vanished at about 273 K. Conversely, the transition frequency fc shifted to higher frequency with increasing temperature. They pointed out that Ishii et al [83] reported that the exponent s was typically greater than 1 at lower frequencies and less than 1 at higher frequencies. For Lee et al, in their PISP nanocomposite, s varied between 0.1 and 0.6 at higher frequencies and increased to about 1.0 at lower frequencies. But in the case of PISF nanocomposite, s remained constant at approximately

Figure 7. Conductivity in the frequency domain for PISF (0.37 wt%) nanocomposite with temperature as a variable. (Reproduced with permission of [25]. Copyright 2007, Elsevier Ltd)

Polymer/CNT nanocomposites by dielectric spectroscopy

15

1.0 over the entire frequency range, independent of SWNT-f loading. They also analyzed the effect of SWNT-p and SWNT-f loading on conductivity at 253 K. They pointed out that conductivity increased with increasing loading in all nanocomposites, but it was much more pronounced in PISP ones. They explained this fact with the building-up of an electrical charge pathway through network of agglomerated SWNT-p in the PI matrix [84, 85]. The increase in conductivity was found to be less pronounced in PISF nanocomposites, pointing out the role of covalent functionalization in reducing the electrical conductivity of isolated nanotubes because the initial sp2 hybridiezed electronic state of SWNT was converted to the more resistive sp3 configuration due to the ligand attachment on the side walls [86]. Liang et al [28] analyzed the dielectric properties of nanocomposites based on low density polyethylene (LDPE) and MWNTs prepared by melt compounding, in the frequency range from 40 to 107 Hz. They also introduced a SEBS-g-MA elastomer, a compatibilizer commonly used to improve the dispersion of fillers in polyolefins, and studied its effect on dispersion and dielectric properties. They found that of LDPE/SEBS-g.MA/MWNT nanocomposites was much lower than that of the LDPE/MWNT nanocomposites at higher MWNT content regime (> 2 vol%). They explained this fact by the effect of SEBS-g-MA, which favored the dispersion of MWNT in LDPE matrix, thereby preventing MWNT from agglomeration and from forming a pathway network. They also analyzed the dependence of and AC on frequency of LDPE/MWNT nanocomposites, as it can be seen in Figure 8. The of pure LDPE and its nanocomposites with low MWNT content was independent of frequency. However, as the MWNT content reached 3.6 vol%, a pronounced dependence on frequency was observed. They applied (7) for

Figure 8. (a) Dielectric constant and (b) AC conductivity versus frequency for LDPE/MWNT nanocomposites. (Reproduced with permission of [28]. Copyright 2006, Elsevier Ltd.)

16

M. Blanco et al.

taking into account the frequency dependence of , obtaining a critical exponent value of 0.26, relatively close to the theoretically predicted exponent value of 0.27 for materials with percolation threshold [37]. Respecting the frequency dependence of AC, they found that for pure LDPE and its nanocomposites with less than 3.6 vol% MWNT increased almost linearly with frequency. However, AC of the nanocomposite with 3.6 vol% content was frequency independent below a critical value (i.e. < 103 Hz), which was the characteristic of a nondielectric material with a pronounced direct current. They used (6) for describing the frequency dependence of AC and found an exponent of 0.66 for the nanocomposite with 3.6 vol% MWNT content. They associated this low value with the polarization effect between MWNT clusters [87]. Moreover, the sum of the critical exponents describing the power laws for and AC should be near 1, condition that was almost obeyed for the nanocomposite with 3.6 vol%, indicating that it was confined in the percolation regime, facilitating the formation of MWNT network within the matrix. Finally, they measured AC as a function of temperature for the 3.6 vol% nanocomposite, in order to investigate the nature of the charge transfer. They concluded pointing out that conductivity decreased with increasing temperature, which is characteristic of ohm resistors. Nogales et al [88] studied dielectric properties of low percolation threshold nanocomposites based on oxidized SWNTs and poly(butylene terephtalate) (PBT). Nanocomposites were prepared by in situ polycondensation reaction process, after ultrasonication of nanotubes in 1,4 butanediol. They measured and AC as a function of frequency for samples with different SWNT concentrations. They found that for SWNT concentrations below 0.2 wt%, a AC (f) fs dependence was observed, the exponent s being close to 1, which was the expected behavior for insulating materials. However, for the sample with 0.2 wt% SWNT, conductivity was frequency independent below a critical frequency, fc, above which it followed a AC (f) fs dependence with s = 0.75. They pointed out that frequency independent AC (f) was characteristic of a nondielectric behavior. Regarding the frequency dependence of , they found that at low SWNT concentration it remained nearly frequency independent. However, a decrease of was observed as the SWNT concentration increased. In particular, for the sample with 0.2 wt% SWNT content, was frequency independent below a critical frequency above which it followed a (f) f-y dependence with y = 0.22. They pointed out that the frequency dependence followed by both and AC was similar to that reported for different CNTpolymer nanocomposites [23, 89, 90]. They noted that their value s = 0.75 was compatible with power laws derived for percolating random mixtures in which polarization effects between clusters were considered [87]. Moreover, the sum of critical exponents describing the power laws for and AC should be 1 near

Polymer/CNT nanocomposites by dielectric spectroscopy

17

the percolation threshold. This condition was obeyed in their case for the sample with 0.2 wt% of SWNT. The behavior exhibited by both and AC at lower frequencies as a function of SWNT content indicated that the sample with 0.2 wt% was in the percolating regime; i. e., a charge transfer across a conducting SWNT network started to occur. To further support this argument and to evaluate the nature of this charge transfer, conductivity was measured versus frequency at several temperatures for the sample with 0.2 wt% SWNT, as it can be seen in Figure 9. They noted that the limiting constant value conductivity values at low frequencies, DC, exhibited a temperature behavior. In the inset of Figure 9 values of DC are represented as a function of reciprocal temperature. It exhibited a non linear behavior with reciprocal temperature. This dependence of the DC was attributed to the thermal fluctuation-induced tunneling mechanism [91]. They pointed out that assuming that a tunneling, activated by local temperature fluctuations of charge carriers from the conducting nanotubes through a potential barrier of varying height was formed in the polymer matrix, the following temperature dependence of DC was expected:

DC exp[

T1 ] (T + T0 )

(12)

where T1 is related to the energy required for a charge carrier to overcome the insulator gap and T0 is the temperature around which thermal fluctuations are

Figure 9. AC as a function of frequency for the nanocomposite with 0.2 wt% of SWNT at different temperatures. The inset shows the DC obtained from the plateau at lower frequencies as a function of the reciprocal temperature. (Reproduced with permission of [88]. Copyright 2004, American Chemical Society)

18

M. Blanco et al.

significant. They obtained a reasonable description of the data for the 0.2 wt% nanocomposite, as shown by the continuous line in the inset in Figure 9, using the values of T0 = 225 K and T1 = 1545 K. They concluded underlaying that dielectric measurements indicated that for the system PBT/SWNT a relatively low percolation threshold in the proximity of 0.2 wt% of SWNT was confirmed, being this value lower than the values previously reported in CNT nanocomposites with thermoplastic matrices and approaching those observed with thermosets. Alig et al [24] carried out dielectric spectroscopy studies on melt processed poly(propylene) (PP)/MWNT nanocomposites. They measured conductivity and the related complex permittivity * in the frequency range from 20 Hz to 106 Hz, for nanocomposites containing 2, 3.5, and 5 wt% of MWNT, after stopping the extruder (recovery after shearing) and during cooling (non-isothermal crystallization). For the sample containing 2 wt% MWNT, DC in the sheared melt at 220 C was found to be less than 10-7 S/m, whereas the value in the annealed melt (20 min after stopping the extruder) increased to about 1.4 x 10-3 S/m. This effect was less pronounced for the samples with 3.5 and 5 wt% MWNT, which were clearly above the percolation threshold. They explained the differences in conductivity values by a strong dependence of the effective number of conducting bonds on the arrangement of the nanotubes and clusters in the matrix as well as on the arrangement of the polymer chains in the gaps between the nanotubes. DC and s ( when frequency tends to zero) were found to increase with annealing of the melt sample with 2 wt% MWNT. This behavior was consistently expressed in the AC and permittivity spectra, Figure 10, which were analyzed in the frame of percolation theory, applying equations (4), (5) and (8) presented in the background. They also applied also the power law that predicted AC near the percolation threshold, equation (6). The currently accepted values for the critical exponent have been 0.5 and 0.72 for 2D and 3D in the equivalent circuit model, and 0.34 and 0.6 for anomalous charge carrier diffusion [35, 36, 43, 44]. In the charge carrier diffusion model [39-42, 47, 49], for frequencies f < fc (where fc is the critical or cross-over frequency) the charge carriers could explore different clusters within one period, i.e. the diffusion was normal. For frequencies above fc the charge carriers visited only parts of the percolation cluster within one period and anomalous diffusion at the fractal percolation clusters took place. This critical frequency, fc (1/), followed the power law of equation (9). For the experimental curves shown in Figure 10, DC, fc, and n were stimated, where n was the critical exponent, named as t/(t + s) in equation (6) They analyzed the evolution of those parameters versus annealing time, as it can be seen in Figure 11.

Polymer/CNT nanocomposites by dielectric spectroscopy

19

Figure 10. Frequency dependence of (a) and AC (b) for the nanocomposite with 2 wt% MWNT during pseudo-isothermal annealing after stopping the extruder. As an example, the fitted curve for AC is shown for t = 280 s. (Reproduced with permission of [24]. Copyright 2007, Elsevier Ltd.)

They described the features for three quantities in a percolation picture by a gradual change of the difference between the effective concentration of electrical active filler particles (or bonds) and the percolation threshold, assuming that all conductive sites or bond contributed. The reduction of conductivity with shear was explained by a reduction of the effective amount of conductive filler by rupture of the CNT contacts due to mechanical deformation. Therefore, the time development of DC, fc and n was a representation of the partial recovery of the idealized percolation structure for the given concentration. The simultaneous increase in DC and fc was an indication that the system was above percolation threshold. The decrease in n was related to an increasing distance to the threshold. and s showed a similar recovery during the rest time. The assumption of a combined nanotubepolymer-nanotube network (highly conductive nanotubes separated in their local contact regions by polymer chains) was supported by the high permittivity values (about 103, see Figure 10a) for the almost recovered percolation network above the percolation threshold. In a similar way, they studied the

20

M. Blanco et al.

Figure 11. Changes in DC (a), the cross-over frequency fc (b), and the power n of the frequency dependence of AC (c) for the nanocomposite with 2 wt% MWNT duging annealing. (Reproduced with permission of [24]. Copyright 2007, Elsevier Ltd.)

cooling and crystallization behavior. They obtained the evolution of DC, fc and n with temperature during cooling from 180 C to 120 C, extracted from the evolution of AC and versus frequency at different temperatures during cooling, as it can be seen in Figure 12. While DC decreased continuously with decreasing temperature, n and fc passed through a minimum. They pointed out that such behavior was typical for a system passing a critical point. At pc, the cross-over frequency fc was expected to approach zero (the charge carriers had to explore the infinite cluster which needed infinite time), whereas n was supposed to aproach its critical value at pc correspoding to the dimension of the lattice. The minimum in fc at about 138 C, seemed to indicate the percolation threshold for the effective concentration of fillers (or contacts). Assuming that n was related to a situation close to the threshold, the exponent of about 0.4 was in the order of the accepted values for the 3D anomalous charge carrier diffusion (n 0.34-0.6 for 2D and 3D). They pointed out that although the estimation of fc and n was somewhat arbitrary, they extracted the following

Polymer/CNT nanocomposites by dielectric spectroscopy

21

conclusions in the frame of the percolation theory: (i) the system changed during crystallization from a situation above the threshold to a conductivity below pc and (ii) the time dependent effective percolation concentration was due to morphological changes in the semi-crystalline structure of the nanocomposite, since the MWNT concentration did not change in the process of crystallization. The second conclusion supported the assumption that the electrical percolation network had to be considered as a combination of nanotubes and polymer in the contact regions between nanotubes. The crystallization of the polymer matrix alone, would only decrease the total conductivity, but could not explain a passing through of the percolation threshold for a constant CNT content. In other words, the change of state of polymeric synapses acted like a change of the effective content of conducting sites, by decresing the conductive bonds between nanotubes and/or clusters. Following the given discussion for the breaking of conductive bonds by shear, they assumed a reduction of conductive bonds by structural changes (crystallization) of the chains in the contact region.

Figure 12. Changes in DC conductivity DC (a), the cross-over frequency fc (b), and the power n (c) for the nanocomposite with 2 wt% MWNT during cooling from 180 C to 120 C. (Reproduced with permission of [24]. Copyright 2007, Elsevier Ltd.)

22

M. Blanco et al.

Wanjale et al [13] studied dielectric behavior of nanocomposites based on poly(1-butene) (PB) and MWNT, prepared by melt processing, in the frequency range from 10-2 to 106 Hz. They measured the electrical conductivity versus frequency for pure PB and nanocomposites with 3, 5 and 7 wt% MWNT. They found that for the nanocomposite with 7 wt% MWNT the conductivity was independent of the frequency, while for the rest of the nanocomposites and for pure PB it was dependent at the higher frequencies. They also measured the at a frequency of 1 MHz for all the nanocomposites and represented it versus the MWNT content. They found that increased moderately till 5 wt% MWNT and then exhibited a sudden increase as the content increased to 7 wt%. They concluded underlaying that the remarkable increase in conductivity and for 7 wt% MWNT confirmed the efficient dispersion and interconnectivity of MWNT in PB matrix, which led to the electrical percolation threshold. Valentini et al [27] characterized nanocomposites based on poly(3octylthiophene) (P3OT) and SWNT following a dielectric route. They measured the real (Z) and imaginary (Z) parts of complex impedance (Z*) in the frequency range from 101 to 106 Hz for pure P3OT and nanocomposite after isothermal crystallization processes carried out at 166 C, 164 C, and 160 C. For both systems, at low frequency, the value of Z tended to decrease when increasing crystallization temperature. At high frequency, only for the nanocomposite Z tended to zero with an onset frequency that increased when increasing crystallization temperature. In Z versus frequency plots, both systems showed a relaxation process with a maximum, but it shifted to higher frequencies when increasing isothermal crystallization temperature only in the case of nanocomposite, suggesting a significant alteration of chain conformation due to nanotube interaction. The observed changes were certainly a result of microstructural changes induced by the incorporation of SWNTs. In particular, the shift of the peak at higher frequencies in the nanocomposite indicated the diluting effect of SWNTs when blended with the matrix. Those results, suggested that the polymer was intercalated between nanotubes, decreasing the interaction between individual nanotubes. They concluded remarking that the observed influence of the crystallization temperatures of the relaxation process could correspond to a progressive redistribution of the energy over the internal degrees of freedom, either related to the particular molecular structure, i.e. segmental and side chain motions, or to localized diffusive motions. San et al [22] studied dielectric properties of nanocomposites based on polythiopene (PT) conducting polymer and SWNTs prepared by in situ polymerization of PT in the presence of SWNTs, in the frequency range from 1 kHz to 10 MHz. They first measured the temperature dependence of at different frequencies for both pure PT and nanocomposite, as it can be seen in Figure 13.

Polymer/CNT nanocomposites by dielectric spectroscopy

23

Figure 13. 3D plot of the frequency and temperature dependence in the for (a) PT and (b) PT/SWNT. (Reproduced with permission of [22]. Copyright 2007, Elsevier Ltd.)

The value of for PT increased from 3 to 8 and was independent on temperature as well as frequency. For the nanocomposite, varied from 4 to 35 with increasing temperature in different frequencies. It appeared considerable higher than that of PT over all temperatures and frequencies. This increase was related to the fact that the complex permittivity * of the material depended on the polarizability of the molecules; the higher the polarizability, the higher the permittivity. They pointed out that for analyzed materials orientation and interfacial polarization played an importan role among all the polarizations. Interfacial polarization arised for electrically heterogeneous

24

M. Blanco et al.

materials in which the conductivity of the matrix and that of the inclusion were different, constraining charge carriers at phase boundaries. Interfacial polarization resulted in an increase in the permittivity. They measured temperature and frequency dependence of for both PT and nanocomposite, concluding that the last presented fairly big values as compared with pure PT. They also measured the temperature and frequency dependence of conductivity values for pure PT and nanocomposite. They noted that the conductivity of PT was very small till 100 C for all frequencies and almost became closer to zero. After this temperature, AC increased with temperature at all frequencies. For the nanocomposite, in the frequency region from 1 kHz to 32 kHz, AC was almost constant and very small for all the temperatures and frequencies. After 32 kHz, however, it increased linearly with temperature and frequency, unlike the corresponding character of PT. The conductivity value increased at around 4 times with compared to that of PT. The increase of conductivity with temperature revealed the semiconductor behavior of the nanocomposite. Park et al [29] studied dielectric properties of nanocomposites based on silicone elastomer and MWNTs, together with mechanical and magnetic properties. They measured versus temperature values for pure silicone rubber and for nanocomposites with 0.2, 0.5 and 0.7 wt% of MWNT at 50 Hz and 100 kHz. It showed nearly the same range, from 3 to 5, in both low and high frequencies, and it increased with increasing MWNT content. Comparing the dielectric results with dynamic mechanical analysis (DMA) results, they certified the relation of permittivity and the modulus:

= ( 0 r ) / E (V / Z )2

(13)

where , 0, r, E, V and Z are the strain, free-space permittivity, relative permittivity, modulus, voltage, and thickness, respectively. The strain was increased with increasing permittivity and decreasing the modulus. The equation (13) could be changed to stress term, , as expressed below:

= 0 r (V / Z ) 2

(14)

Comparing both equations to get an optimum characterized elastomer, the modulus and dielectric permittivity should be able to control their range in consideration of their applications. Authors pointed out that in their study, the pristine silicone elastomer showed a higher modulus than the nanocomposites, but the permittivity was the lowest. On the contrary, the 0.7 wt% MWNT nanocomposite showed a low storage modulus and a high permittivity-high strain property. They concluded underlaying that the 0.7 wt% MWNT nanocomposite was a good candidate material for actuator applications.

Polymer/CNT nanocomposites by dielectric spectroscopy

25

3. Nanocomposites based on CNTS and thermosetting matrices


Sandler et al [73] characterized nanocomposites based on epoxy matrix and untreated catalytically-grown MWNTs (weight percentages ranging from 0.0225 to 0.15 wt%) in terms of AC impedance spectroscopy performed at room temperature. They measured the real (Z) and imaginary (Z) parts of the complex impedance (Z*) as a function of frequency (1-106 Hz). The complex admittance (Y* = 1/Z* = Y + jY) of the nanocomposites was also calculated and modelled as a parallel resistor (R) and capacitor (C). The specific admittance was calculated from the modulus of the complex admittance and was used as a means of comparing the frequency behavior. From the real part of the complex admittance they obtained AC values:

AC = Y ' d / A

(15)

where d is the sample length or distance between the electrodes and A is the contact area. They found that the nanocomposites at higher filler contents showed a purely ohmic behavior. In contrast, the pure epoxy was an ideal dielectric material and displayed an increase in the capacitive component with increasing frequency. The sample with 0.0225 wt% CNT content displayed similar behavior. The specific admittance obtained was larger than that of the pure epoxy due to the presence of conductive aggregates within the specimen. However, those aggregates did not percolate. Below the percolation threshold the admittance of the nanocomposite was dominated by capacitive effects. Above the percolation threshold the CNT aggregates provided a conductive three-dimensional path, and all samples containing more than 0.04 wt% showed a purely ohmic behavior, indicated by the frequency-independence of the specific admittance. Since Z did not contribute significantly to the admittance of the nanocomposites containing more than 0.04 wt%, the percolation threshold was therefore taken to be between 0.0225 and 0.04 wt%. In later papers [72, 92] they found an ultra-low electrical percolation threshold in MWNT/epoxy nanocomposites. They prepared nanocomposites based on epoxy and aligned MWNTs with weight fractions ranging from as low as 0.001 up to 1 wt% and analyzed electrical properties by AC impedance spectroscopy. They presented plots of AC obtained from equation (15) versus frequency. They found that the sample containing 0.001 wt% of nanotubes presented an increase in the capacitive component with increasing frequency, similarly to the pure epoxy matrix. The sample containing 0.0025 wt% showed a frequency independent conductivity at frequencies below 10 Hz, which was then followed by a region of increasing conductivity. The increase was again

26

M. Blanco et al.

similar to that of the pure matrix. With further increase of filler content, the conductivity became frequency independent over the range investigated, indicating non-dielectric behavior. By plotting AC versus MWNT content (underlaying that plots of DC gave very similar results) they found an increase in conductivity of about two orders when increasing the loading from 0.001 to 0.0025 wt%. A further increase was found for loadings of more than 0.005 wt%. Thus indicated that an infinite network of percolated nanotubes started forming above 0.001 wt%. The conductivity for higher loading fractions was analyzed by an equation analogous to (5) for AC. This analysis revealed a percolation threshold of about 0.0025 wt%, with a critical exponent t of 1.2. This exponent did not reflect a reduction in system dimensionality in their case but rather the aggregation process of the nanotubes during sample preparation. They concluded underlaying that the formation of these conducting networks was not a true statistical percolation process based on the random distribution of individual high-aspect ratio fillers, but rahter was attributed to the mutual attraction of the nanotubes after the addition of the hardener. In addition, thermally activated hopping transfers between disconnected or only weakly connected parts of the filler networks through the matrix was observed previously to reduce the conductivity exponent of an experimental component system [93]. Some of the individual nanotubes or nanotube clusters in their samples were essentially isolated by a polymer coating, which prevented direct contact. The resulting hopping transfer between the clusters also explained the relatively low maximum conductivity observed. The value of about 2 S/m for a loading fraction of 1 wt%, was orders of magnitude lower than the expected intrinsic nanotube conductivity. Allaoui et al [10] fabricated MWNT/epoxy resin nanocomposites (with nanotube content ranging from 0.5 to 4 wt%) and characterized them in terms of AC impedance spectroscopy. They presented measurements of both AC and DC as a function of the weight percentage of MWNTs. They observed nine orders of magnitude change in the conductivity over the range of MWNTs addition up to 4 wt%, corresponding to a percolation phenomenon. The conduction threshold was revealed by both conductivities between 0.5 and 1 wt%. From AC versus frequency (10-1-106) measurements shown in Figure 14, they observed two behaviors: the insulators with conductivity inversely proportional to the frequency and the conductors with conductivity independent of the frequency. The 0.5 wt% nanocomposite behavior was very similar to that of the matrix, except that its conductivity was an order of magnitude higher. The nanocomposite was still insulator. The nanocomposites with higher content were conductor. They pointed out that the threshold of a conducting or interconnected network of the CNTs was reached at 1 wt%.

Polymer/CNT nanocomposites by dielectric spectroscopy

27

Figure 14. AC as a function of the frequency for nanocomposites with several MWNT content. (Reproduced with permission of [10]. Copyright 2002, Elsevier Ltd.)

In a later paper [15] Allaoui et al studied the dielectric properties of MWNT/epoxy nanocomposites, measuring both AC and complex permittivity * . They prepared two types of samples: the first was made using as-prepared MWNTs and was called sample A; the other one based on shortened MWNTs (shortened to mean length about 10 m, i. e. divided by 10) was called sample B. They first presented AC conductivity plot at 100 Hz as a function of the weight percentage of MWNTs (from 0 to 4 wt%) for the two types of samples. Over the range of wt% studied, they found that increased by five orders of magnitude. The insulator-to-conductor transition was achieved at low wt%. This transition was very sharp in the case of sample A while it was very progressive for sample B. They fitted experimental results by using percolation theory (an equation analogous to (5) for AC, and found critical t exponents of 1.0 for sample A and 4.9 for sample B. They pointed out that none of their experimental data was in agreement with the universal exponent of 3D systems predicted by classical percolation theory, t = 1.94. The very high value of the critical exponent for sample B could be explained by the presence of tunneling conduction, favoured by the short length of the MWNTs. The percolation threshold calculated from equation (5) varied little with the aspect ratio (pc = 0.8 for sample A and pc = 0.7 for sample B). The insulator-to-conductor transition changed from sharp to gradual when the aspect ratio was divided by ten. The aspect ratio seemed to control the type of conduction. In sample B, cluster-cluster conduction was favoured rather than MWNT-MWNT conduction. They also presented AC versus frequency plots, which where fitted to equation (6) obtaining the following exponent values: t/(t + s) = 0.79 for sample A and 0.90 for sample B. They also presented permittivity spectra of both samples, by plotting and versus frequency. They observed two types

28

M. Blanco et al.

of behaviors: the insulators with low frequency-independent permittivity and the conductors with high permittivity decreasing with frequency. Low fraction nanocomposites (0.4 and 0.5 wt%) behave very similarly to the matrix, except that was one order of magnitude higher. The permittivity was not much affected by the MWNTs treatment. The permittivity near the percolation threshold was fitted to equation (7), obtaining values of t/(t + s) = 0.06 (sample A) and 0.04 (sample B). They pointed out that the general scaling law (the sum of critical exponents describing the power laws for and AC should be 1 near the percolation threshold) was relatively well satisfied. The critical exponents were found to be very close to those corresponding to the intercluster polarization, indicating that this might be the main phenomenon governing the frequency dependence of the dielectric constants. Kim et al [14] measured electrical and dielectric properties of nanocomposites based on epoxy resin and SWNT concentrations of 0.01-0.21 wt%. They measured DC, complex permittivity * and dielectric modulus M*. They first presented DC versus mass fraction of the nanocomposites at room temperature and found that they followed typical percolative behavior. They fitted experimental data to equation (5) yielding the threshold pc = 0.074 wt% and a critical exponent t = 1.3. They pointed out that the percolation threshold was anomalously small compared with known values from the continuum percolation theories. Such small threshold was ascribed primarily to the large aspect ratio of SWNTs. They noted that percolation threshold was proportional to the inverse of the average excluded volume of filler particles [94-96]. The excluded volume was found to be proportional to the aspect ratio when this was large. They pointed out that the threshold was inversely proportional to the aspect ratio when the length of the filler particles was much larger than the diameter. Assuming the aspect ratio of SWNT to be roughly 1000, they got a percolation threshold of around 25 vol%, a reasonable estimation. They also measured dielectric properties in the frequency range 101-106 Hz. They found that for the nanocomposites with various mass fractions the dielectric properties (* and M*) could be classified into three categories. Samples of low mass fraction (p << pc, i.e. 0.01 wt%) followed the tendency of pure epoxy: permittivity data were only slightly elevated from those of pure epoxy and the dielectric relaxation intrinsic to the epoxy (related to the motion of molecular dipoles) appeared without frequency shift. In the mid-range of mass fraction (p pc, i.e. 0.05 wt%) a new relaxation was revealed in the M* graphs. Inside the midrange composition the increase of mass fraction seemed to make the peak position to move to higher frequencies. This relaxation was related to the polarization at the interfaces between nanotubes or small clusters of nanotubes. Higher mass fractions (p > pc, i.e. 0.1-0.2 wt%) induced conduction throughout the sample and increased at low frequencies. They found bends in the versus frequency graphs, which induced sharp peaks in

Polymer/CNT nanocomposites by dielectric spectroscopy

29

the tan versus frequency graphs. The size of the peaks increased as the mass fraction increased and simultaneously the peak moved to higher frequencies. However, there was no noticiable temperature dependence. From normalized tan versus relative frequency plots they found that the curves of the various mass fractions were of universal shape after scaling and that peaks fitted well with a ubiquitous Debye relaxation curve regardless of temperature and mass fraction. In the same sense, they concluded underlaying that of their samples was inversely proportional to frequency, which meant that conductivity was frequency independent. Kim et al [81] measured the electrical conductivity of oxidized MWNT/epoxy nanocomposites with respect to the chemical treatment of the MWNT, by means of a dielectric analyzer in the frequency range of 10-1-106 Hz. They first presented the frequency dependence of the AC of the nanocomposites of oxidized MWNT/epoxy with up to 3 vol% of CNTs at 25 C, as it can be seen in Figure 15.

Figure 15. Frequency dependence of AC conductivity AC of epoxy based nanocomposites with different contents of (a) A0-MWNT, treated with HNO3 at room temperature for 4 hours, (b) A1-MWNT, treated with HNO3 at 100 C for 1 hour, (c) A4-MWNT, treated with HNO3 at 100 C for four hours, and (d) B0-MWNT, treated with H2O2/NH4OH at room temperature for 4 hours. (Reproduced with permission of [81]. Copyright 2005, Elsevier Ltd.)

30

M. Blanco et al.

Below the threshold, AC conductivity increased linearly with frequency and the nanocomposites behave as a dielectric. In the percolation transition the AC was equal to DC up to the characteristic frequency, above which the conductivity increased again generally with a lower slope than that of the dielectrics. Above the percolation threshold conductivity remained constant at a given frequency range. AC varied in the same way independently of the type of oxidation of MWNTs. However, the threshold occurred at a different MWNT content depending on the oxidation condition and solution. It increased in the order A0, A1 and A4 for the A-series. The percolation of B0MWNT nanocomposite occurred at around 0.02 vol%, similar to that of the A1-MWNT. In addition, B0-MWNT/epoxy nanocomposites showed higher conductivity for the same MWNT content at a given frequency. The values of DC were deduced from the AC versus frequency plot when frequency tended to zero. They fitted the dependence of DC on the filler concentration by equation (5), obtaining pc and t values for each type of nanocomposites. Regarding pc, they found an explicit trend in which the stronger the oxidation conditions the higher the percolation threshold was observed, regardless of the solution used. In this way, A4-MWNT nanocomposite exhibited an exceptionally high pc. They attributed this to the fact that pc lied in the partially damaged crystalline structure of the MWNT at the given MWNT content. They concluded that in order to achive a low percolation in the MWNT nanocomposites, the purification of the MWNT had to be carried out in relatively mild acidic or basic condition. In order to improve the electrical properties of the nanocomposites, the wet process of the MWNT had to be carried out under conditions in which no damage was caused to the MWNT, by controlling the oxidative conditions, including the types of solution, its concentration, treatment time and temperature. Moisala et al [97] studied electrical conductivity behavior of epoxy nanocomposites containing 0.005-0.5 wt% of SWNTs or MWNTs in terms of AC impedance spectroscopy. They presented AC versus frequency plots for the nanocomposites with different filler type and amount, as it can be seen in Figure 16. They fitted data according the percolation theory to an equation analogous to (5) for AC. They found that all the MWNT samples were electrically conductive with a percolation threshold as low as 0.0025 wt%. For the SWNT samples percolation occurred at 0.05 wt% for the chemically treated ones and 0.23 for the ball-milled fillers. They pointed out that results were interesting in that the concentrations of nanotubes required for electrical percolation were much lower than would be predicted for randomly oriented and positioned nanotubes for the case of the MWNT nanocomposites. The discrepancy was explained in terms of the nanotubes starting to aggregate in a highly local scale before they were trapped by the setting resin [92]. Furthermore,

Polymer/CNT nanocomposites by dielectric spectroscopy

31

Figure 16. Impedance spectroscopy plots of the nanocomposite samples containing different types of nanotubes: (a) MWNTs, (b) chemically treated SWNTs; (c) ballmilled SWNTs; (d) electrical conductivity at 1 Hz for the nanocomposite materials as a function of nanotube content.

the application of pre-dispersed SWNTs as the filler resulted in electrical perocolation at a concentration of only 0.005 wt%, which was a lower value than previously reported [8]. Barrau et al [90] investigated AC and DC conductivities of MWNT/polyepoxy nanocomposites from 20 to 110 C in the frequency range 10-2-106 Hz as a function of nanotube weight fraction ranging from 0.04 to 2.5 wt%. They first presented the frequency dependence of the real part of the complex conductivity, () for various filler contents at room temperature. From 0.3 wt% they observed that () became independent of the frequency at the lower frequencies of measurements and was identified as DC. Above a characteristic onset frequency c, the conductivity increased with increasing frequency. The frequency region of constant conductivity extended to higher frequencies with increasing weight fraction of CNTs. They represented () by the following equation:

' ( ) = (0) + AC ( ) = DC + A s

(16)

32

M. Blanco et al.

where A is a temperature-dependent constant and s is an exponent depending both on frequency and temperature with values in the range 0-1. This type of behavior was noted by Jonscher, who called it the universal dynamic response (UDR) [98, 99]. The value of DC could be estimated from the plateau values in the () versus frequency plot. They pointed out the existence of a critical frequency c beyond which a power law was followed. From their experimental data they obtained DC and c for different CNT weight fractions. Both increased with CNT content at room temperature. Then they presented DC values as a function of the weight fraction of nanotubes. A dramatic increase by 10 orders of magnitude was observed between 0.2 and 0.4 wt%. They fitted experimental data to equation (5) obtaining pc = 0.3 wt% and t = 1.44 as fitting parameters. They pointed out that the behavior was similar to that found by other authors for nanocomposites based on other polymer matrices [100, 101]. The high aspect of the nanotubes made the percolation possible with a very small content of nanotubes. They pointed out that critical exponents value was smaller than the universal value to 3D percolating systems (t = 1.94) [36]. This lower value could mean that the percolation took place in a network displaying more dead arms than a classical random network. They also observed that the threshold was temperature-independent in the investigated temperature range. They found t = 1.61 and 1.71 for T = 50 and 100 C, respectively. The critical exponent increased with increasing temperature and tended toward the value predicted by 3D percolation theory of randomly distributed objects. That increase was associated with the reduction in the number of dead arms present when the temperature approached the Tg.

References
Iijima, S. 1991, Nature, 354, 56 Merkoi, A. 2006, Microchim. Acta, 152, 157 Daenen, M., de Fouw, R. D., Hamers, B., Janssen, P. G., Schouteden, K., Veld, M. A., Reynhout, X. E., and Reijenga, J. C. 2003, The wonderous world of carbon nanotubes, Eindhoven University of Technology, Eindhoven 4. Saito, R., Dresselhaus, G., and Dresselhaus, M. S. 2003, Physical Properties of Carbon Nanotubes, Imperial College Press, London 5. Harris, P. J. F. 2003, Carbon Nanotubes and Related Structures, Cambridge University Press, Cambridge 6. Kaiser, A. B., Dsberg, G., and Roth, S. 1998, Phys. Rev. B, 57, 1418 7. Coleman, J. N., Curran, S., Dalton, A. B., Davery, A. P., McCarthy, B., Blau, W., and Barklie, R. C. 1998, Phys. Rev. B, 58, 7492 8. Biercuk, M .J., Llaguno, M. C., Radosavljevic, M., Hyun, J. K., Johnson, A. T., and Fischer, J. E. 2002, Appl. Phys. Lett., 80, 2767 9. Kymakis, E., Alexandou, I., and Amaratunga, G. A. 2002, Synth. Met., 127, 59 10. Allaoui, A., Bai, S., Cheng, H. M., and Bai, J. B. 2002, Comp. Sci. Tech., 62, 15 1. 2. 3.

Polymer/CNT nanocomposites by dielectric spectroscopy

33

11. Grimes, C. A., Mungle, C., Kouzoudis, D., Fang, S., and Eklund, P. C. 2000, Chem. Phys. Lett., 319, 460 12. Nan, C. W. 1993, Prog. Mater. Sci., 37, 1 13. Wanjale, S. D., and Jog, J. P. 2006, J. Macrom. Sci., Part B: Physics, 45, 1053 14. Kim, B., Lee, J., and Yu, I. 2003, J. Appl. Phys., 94, 6724 15. Allaoui, A., Bai, J. B., and Rieux, N. 2003, Polym. & Polym. Comp., 11, 171 16. Valentini, L., Puglia, D., Frulloni, E., Armentano, I., Kenny, J., and Santucci, S. 2003, Comp. Sci. Tech., 64, 23 17. Park, W. K., Kim, J. H., Lee, S., Kim, J., Lee, G. W., and Park, M. 2005, Macromol. Res., 13, 206 18. Nam, Y. W., Kim, W. N., Cho, Y. H., Chae, D. W., Kim, G. H., Hong, S. P., Hwang, S. S., and Hong, S. M. 2007, Macromol. Symp., 249-250, 478 19. Wang, L., and Dang Z. M. 2005, Appl. Phys. Lett., 87, 042903 20. Zhang, S., Zhang, N., Huang, C., Ren, K., and Zhang, Q. 2005, Adv. Mater., 17, 1897 21. Clayton, L. M., Sikder, A. K., Kumar, A., Cinke, M., Meyyappan, M., Gerasimov, T. G., and Harmon, J. P. 2005, Adv. Funct. Mater., 15, 101 22. San, S. E., Yerli, Y., Okutan, M., Yilmaz, F., Gunaydin, O., and Hames, Y. 2007, Mater. Sci. Eng. B, 138, 284 23. Potschke, P., Dudkin, S., and Alig, I. 2003, Polymer, 44, 5023 24. Alig, I., Lellinger, D., Dudkin, S., and Potschke, P. 2007, Polymer, 48, 1020 25. Lee, H., Pejanovic, S., Mondragon, I., and Mijovic, J. 2007, Polymer, 48, 7345 26. Zhu, B. K., Xie, S. H., Xu, Z. K., and Xu, Y. Y. 2006, Comp. Sci. Tech., 66, 548 27. Valentini, L., Armentano, I., Biagiotti, J., Marigo, A., Santucci, S., and Kenny, J. 2004, Diamond Related Mater., 13, 250 28. Liang, G. D., and Tjong, S. C. 2006, Mater. Chem. Phys., 100, 132 29. Park, I. S., Kim, K. J., Nam, J. D., Lee, J., and Yim, W. 2007, Polym. Eng. Sci., 47, 1396 30. Efros, A. L., and Shklovskii, B. I. 1976, Phys. Status Solidi B, 76, 475 31. Kirkpatrick, S. 1973, Rev. Mod. Phys., 45, 574 32. Kirkpatrick, S. 1976, Phys. Rev. Lett., 36, 69 33. Stephen, M. J. 1978, Phys. Rev. B, 17, 44444453 34. Straley, J. P. 1976, J. Phys. C: Solid State Phys., 9, 783795 35. Clerc, J. P., Giraud, G., Laugier, G., and Luck, J. M. 1990, Adv. Phys., 39, 191 36. Stauffer, D., and Aharony, A. 1994, Introduction to percolation theory, Taylor and Francis, London 37. Bergman, D. J., and Imry, Y. 1977, Phys. Rev. Lett., 39, 1222 38. Stroud, D., and Bergman, D. J. 1982, Phys. Rev. B, 25, 2061 39. Stauffer, D. 1979, Phys. Rep., 54, 1 40. Straley, J. P. 1980, J. Phys. C: Solid State Phys., 13, 2991 41. Gefen, Y., Aharony, A., Mandelbrot, B. B., and Kirkpatrick, S. 1981, Phys. Rev. Lett., 47, 1771 42. Gefen, Y., Aharony, A., and Alexander, S. 1983, Phys. Rev. Lett., 50, 77 43. Bunde, A., and Havlin, S. 1996, Fractals and disordered systems, Springer, Berlin 44. Sahimi, M. 1994, Applications of percolation theory, Taylor and Francis, London 45. Herrmann, D. W., and Stauffer, D. 1981, Z Phys. B, 44, 339

34

M. Blanco et al.

46. Ziff, R.M., and Stell, G.1988, LaSC University of Michigan, Report No. 88, Footnote 26 47. Scher, H., and Lax, M. 1973, Phys. Rev. B, 7, 4491 48. Maass, P., Meyer, M., and Bunde, A. 1995, Phys. Rev. B, 51, 8164 49. Hong, D. C., Halvin, S., Herrmann, H. J., and Stanley, H. E. 1984, Phys. Rev. B, 30, 4083 50. Straley, J. P. 1977, Phys. Rev. B, 15, 5733 51. Fish, R., and Harris, A. B. 1978, Phys. Rev. B, 18, 416 52. Adler, J., Meir, Y., Aharony, A., Harris, A. B., and Klein, L. J. 1990, J. Stat. Phys., 58, 511 53. Adler, J. 1985, J. Phys. A: Math. Gen., 18, 307 54. Alexander, S., and Orbach, R. 1982, J. Phys. Lett., 43, 625 55. Gingold, D. B., and Lobb, C. J. 1990, Phys. Rev. B, 42, 8220 56. Derrida, B., and Vannimenus, J. 1982, J. Phys. A: Math. Gen., 15, 557 57. Derrida, B., Stauffer, D., Herrman, H. J., and Vannimenus, J. 1983, J. Phys. Lett., 44, 701 58. Herrmann, H. J., Derrida, B., and Vannimenus, J. 1984, Phys. Rev. B, 30, 4080 59. Zabolitzky, J. G. 1984, Phys. Rev. B, 30, 4077 60. Den Nijs, M. P. 1979, J. Phys. A: Math. Gen., 12, 1857 61. Nienhuis, B. 1982, J. Phys. A: Math. Gen., 15, 199 62. Dang, Z. M., Lin, Y. H., and Nan, C. W. 2003, Adv. Mater., 15, 1625 63. Dang, Z. M., Nan, C. W., Xie, D., Zhang, Y. H., and Tjong, S. C. 2004, Appl. Phys. Lett., 85, 97 64. Kotsilkova, R., Nesheva, D., Nedkov, I., Krusteva, E., and Stavrev, S. 2004, J. Appl. Polym. Sci., 92, 2220 65. McCrum, N. G., Read, B. E., amd Williams, G. 1967, Anelastic and Dielectric Effects in Polymeric Solids, Wiley, New York 66. Harmon, J. P., Muisener, P. A., Clayton, L., and DAngelo, J. 2001, Mater. Res. Soc. Symp. Proc., 697, 425 67. Muisener, P. A., Clayton, L., DAngelo, J., Harmon, J., Sidker, A. K., Kumar, A., Meyyappan, M., and Cassell, A. M. 2002, J. Mater. Res., 17, 2507 68. Tatro, S. 2002, Ph.D. Thesis, University of South Florida 69. Tatro, S., Clayton, L., Muisener, P. A., Rao, A. M., and Harmon, J. P. 2004, Polymer, 45, 1971 70. van Krevelan, D. W. 1990, Properties of Polymers: Their Correlation with Chemical Structure, Their Numerical Estimation and Prediction from Additive Group Contributions, Elsevier, New York 71. Shaffer, M. S., and Windle, A. H. 1999, Adv. Mater., 11, 937 72. Sandler, J. K., Kirk, J. E., Kinloch, I. A., Shaffer, M. S., and Windle, A. H. 2003, Polymer, 44, 5893 73. Sandler, J. K., Shaffer, M. S., Bauhofer, W., Schulte, K., and Windle, A. H. 1999, Polymer, 40, 5967 74. Ounaies, Z., Park, C., Wiseb, K. E., Siochic, E. J., and Harrison, J. S. 2003, Comp. Sci. Tech., 63, 1637 75. Feijoo, J., Cabedo, L., Gimenez, E., Lagaron, J., and Saura, J. 2005, J. Mater. Sci., 40, 1785

Polymer/CNT nanocomposites by dielectric spectroscopy

35

76. Hu, N., Zhour, H., Dang, G., Rao, X., Chen, C., and Zhang, W. 2007, Polym. Int., 56, 655 77. Hadjev, V. G., Mitchell, C. A., Arepall, S., Bahr, J. L., Tour, J. M., and Krishnamoorti, R. 2005, J. Chem. Phys., 122, 124708 78. Bisquert, J., Garcia-Belmonte, G., and Ruus, J. 2004, J. Electrochem., 40, 352 79. Narula, A. K., Singh, R., and Chandra, S. 2000, Bull. Mater. Sci., 23, 227 80. Hinderman-Bischoff, M., and Ehrburger-Dolle, F. 2001, Carbon, 39, 375 81. Kim, Y. J., Shin, T. S., Choi, H. D., Kwon, J. H., Chung, Y. C., and Yoon, H. G. 2005, Carbon, 43, 23 82. Rimska, Z., Kresalek, V., and Spacek, J. 2002, Polym. Comp., 23, 95 83. Ishii, T., Abe, T., and Shirai, H. 2003, Solid State Commun., 127, 737 84. Tamburri, E., Orlanducci, S., Terranova, M. L., Valentini, F., Palleschi, G., and Curulli, A. 2005, Carbon, 43, 1213 85. Potschke, P., Fornes, T. D., and Paul, D. R. 2002, Polymer, 43, 3247 86. Farmer, D. B., and Gordon, R. G. 2005, Electrochem. Solid State Lett., 8, G89 87. Song, Y., Noh, T. W., Lee, S. I., and Gaines, J. R. 1986, Phys. Rev. B, 33, 904 88. Nogales, A., Broza, G., Roslaniec, Z., Schulte, K., Sics, I., Hsiao, B. S., Sanz, A., Garcia-Gutierrez, M. C., Rueda, D. R., Domingo, C., and Ezquerra, T. A. 2004, Macromolecules, 37, 7669 89. Kilbribe, B. E., Coleman, J. N., Fraysse, J., Fournet, P., Cadek, M., Drury, A., Hutzler, S., Roth, S., and Blau, W. J. 2002, J. Appl. Phys., 92, 4024 90. Barrau, S., Demont, P., Peigney, A., Laurent, C., and Lacabanne, C. 2003, Macromolecules, 36, 5187 91. Sheng, P. 1980, Phys. Rev. B, 21, 2180 92. Martin, C. A., Sandler, J. K., Shaffer, M. S., Schwarz, M. K., Bauhofer, W., Schulte, K., and Windle, A. H. 2004, Comp. Sci. Tech., 64, 2309 93. Reghu, M., Yoon, C. O., Yang, C. Y., Smith, D. M., Heeger, A. J., and Cao, Y. 1994, Phys. Rev. B, 50, 13931 94. Balberg, I., Anderson, C. H., Alexander, S., and Wagner, N. 1984, Phys. Rev. B, 30, 3933 95. Balberg, I., Binenbaum, A., and Webman, I. 1984, Phys. Rev. Lett., 54, 1465 96. Bug, A. L., Safran, S. A., and Webman, I. 1985, Phys. Rev. Lett., 54, 1412 97. Moisala, A., Li, Q., Kinloch, I. A., and Windle, A. H. 2006, Comp. Sci. Tech., 66, 1285 98. Jonscher, A. K. 1977, Nature, 267, 673 99. Dyre, J. C., and Schroeder, T. B. 2000, Rev. Mod. Phys., 72, 873 100. Kilbride, B. E., Coleman, J. N., Fraysse, J., Fournet, P., Cadek, M., Drury, A., Hutzler, S., Roth, S., and Blau. W. J. 2002, Appl. Phys., 92, 4024 101. Stphan, C., Thien Phap, N., Lahr, B., Blau, W. J., Lefrant, S., and Chauvet, O. 2002, J. Mater. Res., 17, 396

Вам также может понравиться