Carbon Nanotube Hybrids and Their Polymer Nanocomposites

Carbon Nanotube Hybrids and Their Polymer Nanocomposites

CHAPTER 2 Carbon Nanotube Hybrids and Their Polymer Nanocomposites Raja Nor Othman*,†, Arthur Norman Wilkinson‡ * Department of Mechanical Engineeri...

3MB Sizes 0 Downloads 34 Views

CHAPTER 2

Carbon Nanotube Hybrids and Their Polymer Nanocomposites Raja Nor Othman*,†, Arthur Norman Wilkinson‡ *

Department of Mechanical Engineering, Faculty of Engineering, Universiti Pertahanan Nasional Malaysia, Kuala Lumpur, Malaysia † Centre for Defence Research and Technology, Universiti Pertahanan Nasional Malaysia, Kuala Lumpur, Malaysia ‡ North West Composites Centre, School of Materials, The University of Manchester, Manchester, United Kingdom

2.1 INTRODUCTION In 1993, TEM images of single-walled carbon nanotubes (SWNTs) were published by Iijima and Ichihashi [1] and Bethune et al. [2], both in Nature. The structure of SWNTs was described as a single graphene sheet, rolled onto a single tube. The SWNTs were reported to exist in bundles, to have fullerene-like hemispheric caps formed at the end of the tubes [1], and to have diameters ranging from 0.7 to 1.6 nm. Compared with carbon nanotubes (CNTs) having multiple walls, SWNTs were predicted to exhibit superior properties due to their strong one-dimensionality and the crystalline perfection of their structure. Based on their crystallographic configurations, SWNTs can be further divided into three types, which are zigzag, armchair, and chiral. In the case of zigzag and armchair, the two opposite CdC bonds are parallel and perpendicular to the tube axis, respectively [3]. Chiral conformation of SWNTs refers to the atomic arrangement when the opposite CdC bond is located at an angle with the tube axis [3]. Hamada et al. reported that all armchair and one-third of zigzag SWNTs are metallic, while the remaining configurations exhibit semiconducting behavior [4]. These descriptions are depicted in Fig. 2.1 [6, 7]. CNTs are known to demonstrate outstanding mechanical and electric properties. Early studies reported in the literature on the estimation of Young’s modulus of CNTs, using techniques such as TEM [8, 9] and atomic force microscope (AFM) [10], reported values of 1.8  1.4 [8] and 1.25 TPa [9]. Using AFM, a Young’s modulus value of 1.28  0.59 TPa was reported by [10]. The CNT was pinned at one end, while the other end was exposed to a force applied by an AFM tip. Several groups performed theoretical calculations and obtained values of Young’s modulus in the region of 1 TPa [11, 12]. In a study performed by Dai et al. [13], the measured electric resistivity of SWNTs ranged from 7.8  1.0 to 117  19 Ω m, depending on the diameter and the structure (straight or curved) of the measured CNTs. The curved CNTs recorded higher values in resistivity, which could be due to structural defects in the outer wall. However, Ebbesen Synthesis, Technology and Applications of Carbon Nanomaterials https://doi.org/10.1016/B978-0-12-815757-2.00002-4

© 2019 Elsevier Inc. All rights reserved.

29

30

Synthesis, Technology and Applications of Carbon Nanomaterials

Fig. 2.1 Schematic diagram showing the different crystallographic configurations of single-walled CNTs, SWNTs ([5], and references therein). (Reproduced with permission from P.-C. Ma, et al., Dispersion and functionalization of carbon nanotubes for polymer-based nanocomposites: a review, Compos. A: Appl. Sci. Manuf. 41 (10) (2010) 1345–1367.)

et al. [14] obtained values of resistivity ranging from 5.1  106 to 5.8 Ω cm, depending on the diameter of the tubes. Consequently, the range of values of electric conductivity reported for SWNTs, and CNTs range quite widely from 102–106 to 103–105 S/cm, respectively ([5] and references therein). In the case of thermal conductivity, the values reported for SWNTs and CNTs are 6000 and 2000 W/m K, respectively ([5] and references therein). Due to these excellent mechanical, electric, and thermal properties and their high-aspect-ratio structure, both CNTs and SWNTs are often used in the areas of field emission, energy storage, and biological applications, among many others. They are also often incorporated into polymer matrices as reinforcements in efforts to produce stiff, strong, multifunctional nanocomposites [5, 15, 16]. In spite of the exciting possibilities that CNTs offer in enhancing various properties of polymer nanocomposites, there are also issues that arise, particularly associated with the processing of these nanocomposites. While CNTs offer high aspect ratios (> 1000) and extremely large surface area, which is beneficial for a wide range of applications, it is very difficult to disperse them effectively within polymer matrices. They easily form agglomerates due to strong van der Waals forces that exist between the tubes and in some cases have tube-tube physical entanglements originating from their growth process [17]. In order to fully exploit the outstanding properties of CNTs, it is essential to have the tubes separated individually. In the simulation shown in Fig. 2.2 [5], homogeneous dispersion is more easily achieved for micron-scale fillers (both spherical and fibrous) in contrast to using nanoscale fillers such as graphite nanoplatelets (GNPs) and CNTs where the number of particles in the unit volume and the particle-particle contact area are much greater. In addition, in real-life situation where van der Waals forces operate, the particle-particle interactions will be much stronger [5].

Carbon Nanotube Hybrids and Their Polymer Nanocomposites

1

0.5 mm

1

1

1

0.5 mm

0.5 mm

0.5 mm

1 0 0

(A)

0.5 mm

0.5 mm 1 0

1 0 0

(B)

0.5 mm

0.5 mm 1 0

1 0 0

(C)

0.5 mm

0.5 mm 1 0

1 0 0

(D)

0.5 mm

0.5 mm 1 0

Fig. 2.2 Distribution of fillers of different dimensions at the same 0.1 vol% in a 1 mm3 cube. The particles are as follows: (A) Al2O3 spherical particles, 100 μm in diameter; (B) carbon fibers of 5 μm diameter and 45 μm in length; (C) GNPs of 45 μm in length and 7.5 nm in thickness; (D) CNTs of 12 nm in diameter and 20 μm in length. (Reproduced with permission from P.-C. Ma, et al., Dispersion and functionalization of carbon nanotubes for polymer-based nanocomposites: a review, Compos. A: Appl. Sci. Manuf. 41 (10) (2010) 1345–1367.)

Many studies focused on the effort to improve the dispersion of CNTs in polymers with an objective to break down their bundle structure by applying high shear forces [18], intensive tip sonication [19], and/or mixing the CNTs with surfactants [20]. These involve many additional processes, which can be both time-consuming and inefficient for industrial processing [21]. Another drawback of dispersing agglomerated CNT within matrices is that it will result in a significant increase of viscosity of the mixture [21–23], which could adversely affect the processing of the nanocomposite. For example, it was reported [22] that CNTs can only be added at loadings less than 5 wt.% into thermosetting composites as higher loadings caused an unacceptable viscosity increase of the mixture. In an effort to minimize the problems of dispersion, chemical functionalization of the CNTs is often performed to promote good dispersion within polymer matrices. These chemical functionalizations involve CNTs undergoing purification, disentanglement, and activation treatments before grafting them to the selected polymer [24]. Besides these tedious chemical preparations, this treatment is also rather selective, which means that only specific functional groups attached to the CNTs would react with the functional groups of a particular polymer. Moreover, these processes fail to reduce the viscosity increase upon CNT addition [25–27].

2.2 CARBON NANOTUBE HYBRIDS Growing CNTs on substrates of various configurations has been performed via chemical vapor deposition (CVD). In general, growth involves heating a furnace to a high temperature (550–1200°C) and placing the substrate in the heating zone of the furnace then allowing the carbon source (typically hydrocarbon, CO, or alcohol) to flow through the furnace. The catalyst can be placed within the furnace before reaction starts (nonfloating type) or can be introduced together with the carbon source (floating type). According to the growth mechanism postulated by Baker [28], a hydrocarbon gas must decompose on the surface of catalyst particles to release hydrogen and carbon. The catalyst particles

31

32

Synthesis, Technology and Applications of Carbon Nanomaterials

provide sites for CNT growth where their size determines the diameter of the CNTs produced. Common catalysts include transition metals such as Fe, Ni, or Co. The catalyst needs to be deposited on the substrate (nonfloating type) before CNT growth can take place. The type of substrate is also important as its interaction with the catalyst determines the quantity and quality of the CNTs produced. Common substrates used to grow CNTs include silica oxide (SiO2), alumina oxide (Al2O3), and magnesium oxide (MgO). At the end of the reaction, the CNTs would normally be detached from the substrate. In these cases, the substrates need not be of a particular geometry, as the CNTs will be removed for their final applications. In the case of producing CNT hybrids to minimize inhomogeneous dispersion, the substrates will remain attached to the CNTs, and the whole hybrid structure will be incorporated into polymer matrices. As such, the choice of substrate (type and geometry) plays a very important role in producing CNT hybrids, which would later affect the final properties of the resultant composites. For example [29], the use of a spherical-shaped and micron-sized substrate played a major role in minimizing any viscosity increase of epoxy resin suspensions after adding the CNT hybrids [23]. Besides providing the sites for CNT growth, these substrates act as a carrier to disperse CNT within the matrices. Strong bonding between the CNTs and the substrates enabled the CNTs to adhere to the substrate even after undergoing intense shear mixing to produce epoxy resin composites, as evidenced in [30]. The following sections discuss the growth of CNTs on carbon or noncarbon substrates, mainly via CVD. The enhanced properties of the corresponding composites containing these CNT hybrids will be presented in Sections 2.3 onward.

2.2.1 CNT Growth on Carbon Substrates One of the earliest works on the growth of CNT on carbon substrates was performed to improve the mechanical properties of epoxy-carbon fiber (CF) composites. In 2007, Bekyarova et al. reported the deposition of CNTs and SWNTs on the surface of CF via electrophoresis [31]. The process was described as follows: CF fabric was first positioned in the middle of a stainless steel frame, with two stainless steel plates placed on either side of the fabric to act as counter electrodes. This setup was then placed in an aqueous dispersion of CNTs, before a positive potential of 10 V/cm was applied to obtain CNTs positioned on the surface of the CF fabric. Composites were prepared by infiltrating this hybrid structure with epoxy resin using vacuum-assisted resin transfer molding (VARTM). The authors reported that this process took about 2 h to complete for a relatively small area (10  15 cm) fabric layup but the addition of sodium hydroxide (<1%) to the dispersion reduced the deposition time to 1 h while maintaining the homogeneity of the CNT deposited. No information was provided on the morphology and the density of the CNTs deposited.

Carbon Nanotube Hybrids and Their Polymer Nanocomposites

Qian et al. further demonstrated the growth of CNT on PAN-based CFs of  7.5 μm in diameter via CVD [32]. Prior to the CNT growth, the CFs underwent acid oxidation followed by base washing to improve their surface chemistry. Iron was used as a catalyst for CNT growth under a flow of acetylene (as the hydrocarbon source) and both hydrogen and argon (carrier gas). The reaction temperature was 750°C, and the growth process took place in an hour. SEM images showed that the CNTs had grown homogeneously on the surface of the CFs, with their diameter measured to be in the range of 21–53 nm, consistent with the diameter of the iron particles. Fig. 2.3 shows the progress of the CNT growth [32]. The growth of CNTs on carbon fiber via CVD was also demonstrated by Wang et al., utilizing a xylene/ferrocene combination [33]. The process was conducted at 800°C at atmospheric pressure for 30 min. As a result, a structure of “hairy” MWNT-grafted CFs was observed under FESEM. In a separate work, carbon nanofibers (CNF) were also reported to grow around CFs via plasma-enhanced CVD [34]. Next, the hybrid structure was coated with ZrC preceramic polymer powders. Further pyrolysis yielded a hybrid of CNF-CF/ZrC ceramic layer. The whole process is depicted in Fig. 2.4 [34].

2.2.2 CNT Growth on Noncarbon Substrates Noncarbon substrates have also been used to produce CNT hybrids, before incorporated them into polymer matrices. The growth of CNTs on woven alumina cloth via CVD was reported by Yamamoto et al. [35] where the iron catalyst was predeposited on the fiber surface prior to the growth. A “forest” of CNTs was obtained at the reaction temperature of 800°C. The authors varied a wide range of reaction parameters, where the addition of hydrogen for 2 min for pretreatment was shown to give the most uniform growth coverage of CNT forests, for all the catalyst concentrations and sample locations tested. The growth of CNTs on stainless steel nanopowders by CVD was reported over a range of temperature between 600 and 800°C [36]. The stainless steel powders acted as both substrate and catalyst. After 3 min reaction time, a layer of densely packed CNTs was observed to grow on the surface of the powders under a flow of ethylene and hydrogen. Qian et al. [37] also demonstrated the growth of CNTs and nitrogen-doped CNTs on silica fibers, where the growth time was reported to control the thickness of the layer of grafted CNTs. Further statistical analysis showed that longer reaction time resulted in an increase of the both the diameter and the length of the CNTs, with the nitrogen-doped type showing smaller diameters and lengths at the same reaction time. Micron-sized spherical substrates have also been used to create CNT hybrid structures. He et al. synthesized CNTs on the surface of spherical alumina microparticles via CVD over a range of temperatures from 450 to 900°C [38]. Ferrocene and xylene were used as the catalyst and carbon source, respectively, and they were injected into the reactor at a rate of 0.2 mL/min. The authors demonstrated the effects of introducing

33

34

Synthesis, Technology and Applications of Carbon Nanomaterials

Fig. 2.3 SEM images showing the progress of the CNT growth on carbon fibers from (A) carbon fiber surface oxidation, (B) iron catalyst deposition, and (C) CNT growth. (Reproduced with permission from H. Qian, et al., Hierarchical composites reinforced with carbon nanotube grafted fibers: the potential assessed at the single fiber level, Chem. Mater. 20 (5) (2008) 1862–1869.)

Carbon Nanotube Hybrids and Their Polymer Nanocomposites

Fig. 2.4 Schematic showing the fabrication process of CNF-CF, coated with ZrC. (Reproduced with permission from L. Yan, et al., Carbon nanofiber arrays grown on three-dimensional carbon fiber architecture substrate and enhanced interface performance of carbon fiber and zirconium carbide coating, ACS Appl. Mater. Interfaces 9 (20) (2017) 17337–17346.)

hydrogen and varying its flow rate during a 15 min reaction time. As shown in Fig. 2.5 [38], the flow rate of hydrogen played a critical role in changing the structure of the hybrid from “six branch” (when only argon was supplied) to “short-dense homogeneous” (when the hydrogen/argon ratio was kept at 40%). The aspect ratio was reported to be highest when no hydrogen was supplied, as hydrogen was shown to depress decomposition of both the catalyst precursor and hydrocarbon source. Zakaria et al. also synthesized CNTs on the surface of alumina powder, aimed at improving properties of epoxy composites such as tensile and thermal [39], flexural and dielectric [40], and compressive properties [41]. Growth was performed using CVD at 800°C, with nickel as the catalyst predeposited on the substrate. The same group also performed CNT growth on muscovite, utilizing nickel as the catalyst [42]. Similar structures were observed with alumina powder [39–41], as similar reaction conditions were used. Full coverage of CNTs was seen coating the substrates (for both alumina powder and muscovite), although the CNTs were observed to agglomerate and form bundles. With a few exceptions [37, 38], the studies presented so far did not discuss details of the morphology of the CNTs obtained or the growth process of the CNTs on the particular substrate. The substrates reported so far were also nonporous, so the CNTs grew only from the surface. As such, the CNTs could be separated from the substrate due to the stresses generated upon incorporating them into polymer matrices. The growth of CNTs on porous, micron-sized spherical silica gel was reported by Othman et al. [43], using CVD at 760°C and injecting ferrocene and toluene during the reaction. Two types of silica gel substrates with different pore diameters and specific surface areas were used. The values of pore diameter and specific surface area of the first

35

Synthesis, Technology and Applications of Carbon Nanomaterials

2mm

1mm

1mm

1mm

1mm

1mm

h 1200

12 10 8

800

6 400

4

CNT length (mm)

CNT aspect ratio (L/D)

36

2 0

0 0

10

20

30

40

Hydrogen ratio (vol.%)

Fig. 2.5 SEM images showing the change of the structure of CNTs grown on micron-sized alumina oxide due to the influence of hydrogen flow rate. The CNT aspect ratio (L/D) and hydrogen ratio were as follows: (A) L/D ¼ 100 when 40% H2, (B) L/D ¼ 120 when 40% H2, (C) L/D ¼ 150 when 20% H2, (D) L/D ¼ 350 when 10% H2, (E) L/D ¼ 450 when 5% H2, and (F) L/D ¼ 1000 when no H2. (G) HRTEM image showing CNTs when hydrogen ratio was 10%. (H) The relation between of CNT length and L/D, with hydrogen ratio. (A) shows the six-branch structure, while (F) shows the shortdense homogeneous structure. (Reproduced with permission from D.L. He, et al., Diameter- and length-dependent self-organizations of multi-walled carbon nanotubes on spherical alumina microparticles, Carbon 48 (4) (2010) 1159–1170.)

Carbon Nanotube Hybrids and Their Polymer Nanocomposites

Fig. 2.6 SEM images showing CNT growth on various substrates (rows) as a function of growth time (columns). The middle row (D, E, and F) refers to a substrate with 6–8 nm pore size, while the bottom row (G, H, and I) refers to a substrate with 26–34 nm pore size. The top row shows the flat, nonporous wafer (A, B, and C) used as a comparison. (Reproduced with permission from R.N. Othman, I.A. Kinloch, A.N. Wilkinson, Synthesis and characterisation of silica-carbon nanotube hybrid microparticles and their effect on the electrical properties of poly(vinyl alcohol) composites, Carbon 60 (0) (2013) 461–470.)

substrate were 6–8 nm and 430–530 m2/g, respectively. For the second substrate, the values of pore diameter and specific surface area were 26–34 nm and 70–170 m2/g, respectively. The growth progress for both substrates was shown in Fig. 2.6 [43] where full coverage could only be seen after 3 h growth. To get a view of the substrate’s core, the hybrid structure produced after 1 h was crushed with pestle and mortar. As shown in Fig. 2.7 [43], a higher density of CNTs was observed within the substrate, compared with on its surface. As such, it was proposed that most of the catalyst was deposited deep inside the silica pores. Consequently, more CNTs grew within the substrates during the 1 h growth period, while relatively few CNTs were formed on the surface. Correlations between the mean inner and outer diameters of the CNTs grown and the pore size of the substrate were also established [43], using TEM to measure the diameters of at least 100 tubes per sample. The results indicated that most of the CNTs grown on the substrate with 6–8 nm pore size followed the growth mechanism suggested by Oberlin et al. [44], in that the carbon atoms appear to have diffused onto the surface of the catalyst, as the inner diameter of the tubes formed was equivalent to the size as

37

38

Synthesis, Technology and Applications of Carbon Nanomaterials

Fig. 2.7 SEM images showing the interior structure of the crushed particle after 1 h growth time (left). The enlarged image (right) shows dense CNT coating of the interior, while only a few CNTs are seen on the surface. (Reproduced with permission from R.N. Othman, I.A. Kinloch, A.N. Wilkinson, Synthesis and characterisation of silica-carbon nanotube hybrid microparticles and their effect on the electrical properties of poly(vinyl alcohol) composites, Carbon 60 (0) (2013) 461–470.)

the catalyst particles formed in the pores and remained below 8 nm for all reaction times. In contrast, there was no correlation between inner diameters of the CNTs grown on the substrate of 26–28 nm pore size. The values of outer diameter for the CNTs grown on the porous substrates remained almost constant as the reaction time was increased. Therefore, as more carbon was supplied, tube elongation occurred instead of tube thickening. The mean outer diameter of the CNTs grown on the substrate of 26–28 nm pore size was below 34 nm (maximum size of the pores) for all reaction times, indicating that most CNTs grown follow the growth mechanism suggested by Baker [28]. The dissolved carbon diffused through the particle and is precipitated at the trailing end to form CNTs. This way, CNTs grow with the same outer diameter as the width of the catalyst particle. In discussing the growth of catalyst nanoparticles, the Ostwald ripening phenomenon [45] should be considered; this is a spontaneous processes where larger particles grow at the expense of smaller particles as they dissolve and redeposit into the bigger particles. Eventually, the catalyst particles grow bigger, increasing the diameter of the CNTs. Therefore, various methods have been tried to inhibit this phenomenon during CNT growth, including introducing water during CVD [46], applying lower hydrocarbon pressure during growth [47], and controlling the heat treatment time and temperature of the catalyst [48]. In this work [43], as the catalyst was deposited within nanosized pores, catalyst coarsening was restricted by the pore size. As such, the Ostwald ripening was restricted and could be controlled, at least up to the 3 h reaction time.

2.3 COMPOSITES CONTAINING CNT HYBRIDS Dispersing nanoscale fillers remains a challenging issue as the viscosity of the mixture typically increases significantly. As a result, the overall composite processing cycle would be affected due to the change in viscosity profile. Most importantly, the structural properties

Carbon Nanotube Hybrids and Their Polymer Nanocomposites

of the resultant composite may deteriorate due to inhomogeneous dispersion of the fillers. This section first discusses the rheology characterization of suspensions containing these carbon hybrid nanomaterials, followed by the final mechanical and thermal properties. The electric properties of the resultant composites will be discussed in detail with special emphasis on equivalent circuits and percolation thresholds.

2.3.1 Rheological Characterization of Epoxy Suspensions Containing CNT Hybrids In composites, processing the subject of rheology, particularly viscosity, should not be ignored as it significantly affects the processing procedure and subsequently composite manufacturing. This subsection starts by reviewing some of the studies performed to assess the viscosity of conventional CNTs in various suspensions. Characterization of suspensions containing CNT hybrids from [43] follows, where CNTs are grown on spherical, porous silica substrates. The rheological behavior of dispersions containing these hybrid silica-CNT particles will be very different to that of dispersions containing conventional CNT, due to the difference in aspect ratio and structure morphology. The earliest work that studied the interactions of CNTs in suspension was performed by Shaffer et al. [26], in which chemically treated CNTs (prepared following a procedure from [49]) suspended in water were analyzed. Although the treated CNTs could be dispersed to give individual particles [49], a dramatic increase in apparent viscosity was observed even as low as 0.7 vol% loading [26]. Further increasing the CNT loading up to 5 vol% caused the formation of a viscoelastic gel [27]. Kinloch et al. [25] further studied CNT interactions in water by increasing the loading up to 11 vol%. The CNTs used were treated similarly to those in [46]. The authors performed extensive rheometer measurements both in dynamic and steady-shear modes at 25°C. A dynamic strain sweep was first performed to determine the linear viscoelastic region (LVR) where storage modulus (G0 ) and loss modulus (G00 ) did not vary with strain amplitude (γ). This was followed by a dynamic frequency sweep within the LVR to examine the dispersion structure [25]. Steady-shear mode experiments were also performed to study the flow properties of the suspensions. In dynamic strain measurements [25], the values of G0 exceeded G00 at the beginning of the strain sweep even at 0.4 vol% (the lowest loading used in this study), before crossover occurred at γ  12% and subsequently G00 > G0 [25]. The modulus crossover occurred at lower strains as the filler loading was increased. This indicated that even at low loading (0.4 vol%), “solid-like” characteristics dominate in these suspensions, which is typical for concentrated and reversibly flocculated networks [25]. Subsequently, dynamic frequency sweeps within the LVR revealed that G0 and G00 values were independent of applied frequency (from 0.1 to 100 rad/s). At a fixed frequency, G0 and G00 exhibited power law relations as the particle loading increased. The results from steady-shear mode testing showed that for all loadings, the suspension shear thinned as shear rate increased [25]. In these suspensions,

39

40

Synthesis, Technology and Applications of Carbon Nanomaterials

the CNTs interacted with each other, as shown by the fact they did not obey the CoxMerz rule [25]. Rahatekar et al. [50] studied the rheology of aligned CNTs in epoxy resin with CNT loadings ranging from 0.0035 to 0.5 wt.%. The CNTs used were the same type used previously that gave a percolation threshold of 0.0025 wt.% [18]. Although it was stated [18] that “the aligned CNT used did not lead to such a strong increase in resin viscosity,” the shear viscosity increased by 100 times for loadings less than 1 wt.% [50]. Similar to CNT suspensions in water [25], shear thinning was observed as shear rate was increased for sample loadings >0.15 wt.%. The authors stated that at high shear rates, the agglomerate size started to decrease and therefore the interconnection between the agglomerates also decreased, resulted in a uniform dispersion [50]. This implied that the agglomerates of CNTs could be ruptured upon application of high shear, hence destroying the conductive network formed previously. CNTs are often chemically treated in an attempt to improve their dispersion [24, 49]. Kim et al. [51] studied the rheological behavior of epoxy resin suspensions containing untreated, amine-, acid-, and plasma-treated CNTs. In their study, the complex viscosity exhibited by treated CNTs at a fixed loading of 1 wt.% and 0.1 Hz was six orders of magnitude higher than neat epoxy, while the increase was three order of magnitude higher for untreated CNT [51]. The authors regarded this as an “improvement” in rheological properties although it is known that increased viscosity often presents a serious hurdle in composite processing. In contrast, Song et al. [52] reported that poorly dispersed CNTs in an epoxy resin generated higher viscosity increase than solvent-assisted well-dispersed CNTs for all loadings (0.5, 1, and 1.5 wt.%). The values of G0 and G00 were also higher for the poorly dispersed CNT compared with the well-dispersed for all the ranges of frequency tested. In a separate study, Seyhan et al. [53] observed that amine-treated CNTs generated lower viscosity (from steady-shear sweep) compared with nontreated CNTs in epoxy resin (for loadings of 0.05, 0.1, and 0.3 wt.%). The increases in shear viscosity [53] were attributed to strong particle-particle interactions. Shear thinning was also observed as the shear rate was increased, similar to results reported previously [25, 50]. A comprehensive rheological characterization of CNT-silica gel hybrids was reported by Wilkinson et al. [23]. The preparation of these CNT hybrids was described previously [43], and the particle chosen for rheology studies was referred to as SG6_3 [23]. Rheological results are also shown for suspensions containing the equivalent loadings of conventional CNTs (referred as NC in the text). Rheology investigations were performed on epoxy resin suspensions without hardener addition. Fig. 2.8A [23] shows the dynamic shear modulus data for suspensions containing CNTs (NC in the figure) under a strain sweep. The storage and loss shear moduli, G0 and G00 , relate to the elastic and viscous responses of the sample, respectively. In the case of neat epoxy resin, G0 remains constant at about 0.06 Pa before decreasing, starting at γ ¼ 30% to an order of magnitude lower at the highest strain limit (γ ¼ 500%).

Carbon Nanotube Hybrids and Their Polymer Nanocomposites

Fig. 2.8 Oscillatory shear measurement for suspensions containing both CNTs (NC) and CNT-silica hybrid particles (SG6_3), conducted at room temperature and a fixed frequency of 1 rad/s. The dependency of G0 and G00 with strain amplitude, γ (%) for CNT-epoxy (A) and CNT hybrid-epoxy (B), is shown. (Reproduced with permission from A.N. Wilkinson, I.A. Kinloch, R.N. Othman, Low viscosity processing using hybrid CNT-coated silica particles to form electrically conductive epoxy resin composites, Polymer 98 (2016) 32–38.)

The corresponding value of G00 (10 Pa) remains unchanged throughout the whole range of γ studied. G00 dominates G0 in the neat resin for all of the γ range investigated. However, the addition of only 0.17 wt.% CNTs into the resin causes the modulus profile to change dramatically. Other studies have also reported significant changes in dynamic modulus under a strain sweep as a result of dispersing CNTs, for loadings as low as 0.4 [25] and 0.004 vol% [54] in water and 0.25 vol% in a polyester resin [55]. At low strain, the values of G0 are higher than those of G00 for all CNT loadings. However, a

41

42

Synthesis, Technology and Applications of Carbon Nanomaterials

crossover between G0 and G00 occurs at a higher strain, known as the crossover strain, γ c. After γ c, G00 dominates G0 indicating predominantly viscous behavior of the dispersion from this point. Hough et al. [54] also observed the crossover strain in their mixture of SWNTs-water-NaDDBS surfactant, where the onset of fluidization (G00 > G0 ) occurred for all nanotube loadings studied (0.04–0.03 vol%). The authors reasoned that this was due to network destruction as the nanotubes rotated through angles large enough to separate themselves from the rest. Similar behavior was observed in [29] for CNT-epoxy resin suspensions due to the same reasons proposed by Hough et al. [54]. The effects of growing CNTs onto the surface of micron-sized spherical silica gel particles are highlighted in the moduli-strain curves in Fig. 2.8B. Unlike those for the NC CNTs, the modulus profiles of the CNT hybrid suspensions are similar to those of the neat resin. There are only slight increases in the values of G0 as particle loading is increased while the corresponding G00 values remain almost unchanged with addition of the particles. For each loading, G0 decreases slowly at higher γ, whereas the values of G00 remain constant for all loadings and over the entire range of γ. For all of the loadings investigated, G00 dominates G0 at all values of γ. The suspensions remain viscous even up to the highest loading employed, which is 1.65 wt.% of CNTs (pCNT) equating to 5 wt.% of the hybrid particles. A modulus crossover is not observed for the CNT hybrid resin suspensions, which indicates that under all the loadings studied, the CNTs on the particles do not form an elastic network in the resin suspension. The results in Fig. 2.8 [23] emphasize the very significant rheological differences obtained as a result of incorporating CNT hybrid particles compared with conventional CNTs. For both suspensions, a short plateau region is observed at low strains, with NC CNT suspensions (Fig. 2.8A) showing a shorter region compared with the CNT hybrid region (Fig. 2.8B). This plateau region is known as the linear viscoelastic (LVR) region where the strain varies linearly with stress [56], hence causing G0 and G00 to be independent of γ. The length of this LVR indicates the stability of the “structure,” with a shorter range indicates a greater complexity of the fluid structure, as demonstrated in [55]. Similarly, in this work [29], the LVR region was also restricted to a much narrower strain range for suspensions containing 1.65 wt.% of conventional CNTs compared with those of lower loadings (Fig. 2.8A). However, in the case of suspensions containing CNT hybrids (Fig. 2.8B), this region becomes longer for G0 compared with G0 of conventional CNT-epoxy suspensions with the same CNT loadings, which demonstrates that grafting promotes network stability. Unlike the NC CNT-epoxy resin suspensions, the G00 values of the CNT hybrid suspensions remain constant for all loadings throughout the whole range of γ studied. Such behavior has never before been observed in CNT suspensions due to strong interactions between the tubes in a suspension unless the CNTs have been treated with acid [25] or stabilized with the addition of surfactant [54]. This demonstrates the effectiveness of incorporating CNT-silica hybrids where the network formed remains stable despite the CNT loading reaches 1.65 wt.% and is subjected to strains up to 500%.

Carbon Nanotube Hybrids and Their Polymer Nanocomposites

Fig. 2.9 Frequency-sweep measurements conducted within the LVR, showing G0 and G00 versus frequency for NC CNT suspensions (A) and CNT hybrid (SG6_3) suspensions (B). (Reproduced with permission from A.N. Wilkinson, I.A. Kinloch, R.N. Othman, Low viscosity processing using hybrid CNTcoated silica particles to form electrically conductive epoxy resin composites, Polymer 98 (2016) 32–38.)

Analysis of G0 and G00 within LVR can provide information regarding the structure of a dispersion [25, 55, 57, 58]. This was performed by comparing G0 and G00 values of the suspension during frequency sweep experiments, which were conducted at a fixed strain specified within LVR, γ 0. The modulus behavior obtained from a frequency sweep is shown in Fig. 2.9 [23]; these were conducted at γ 0 ¼ 1% for the NC (NC) CNT suspensions and γ 0 ¼ 2% for the hybrid particle suspensions. In the case of the NC CNT suspensions (Fig. 2.9A), the values of both G0 and G00 are almost constant throughout the frequency range studied. Formation of a plateau at low frequency is observed for all the

43

Synthesis, Technology and Applications of Carbon Nanomaterials

CNT loadings, and this plateau region stretches to wider frequency ranges as the CNT loading increases. Ultimately, for the highest CNT loading of 1.65 wt.%, G0 becomes constant throughout the frequency range studied (0.1–100 rad/s). The change of modulus behavior from frequency-dependent to frequency-independent was also observed in other studies of CNT dispersions such as poly(ether ether ketone) (PEEK) melt composites [59] and CNT-epoxy resins [60], where this transition was interpreted as a rheological percolation phenomena [60]. It is also observed in Fig. 2.9 that G0 > G00 at all frequencies, except for the lowest loading of 0.17 wt.%, for which a crossover occurs at 20 rad/s and subsequently G00 > G0 . Similar modulus behavior, where G0 > G00 , was observed for other CNT dispersions [25, 59, 60], which is typical of a strong gel [25]. Interestingly, the G0 and G00 profiles for the CNT hybrid suspensions show that the hybrid particles did not form a mechanically strong network; therefore, G00 always dominates G0 , with the values increasing in line with frequency (Fig. 2.9B). These spectra show that the suspension exhibit viscous behavior despite the high loading of pCNT ¼ 1.65 wt.%. This indicates that the grafting process prevents the nanotubes from interacting sufficiently to increase the elastic properties and hence the viscosity of the resin suspensions. A Van Gurp-Palmen (VGP) plot is a plot of phase angle (δ) as a function of complex modulus (G*) within the LVR [61]. These plots may be used to assess the “elasticity” of a suspension; therefore, several authors have interpreted the rheology of their CNTpolymer melt systems by plotting various loadings on a VGP plot [29, 60, 62]. Fig. 2.10 shows the VGP plot for both the NC and hybrid particle (SG6_3) suspensions at various loadings. It is obvious that all the phase angle data for suspensions containing hybrid particles are essentially constant at  90° and overlap with those of the epoxy resin, over a range of G* stretching from 1 to 1000 Pa, indicating viscous characteristics as the particles are not interacting. In contrast, the data for the NC CNT suspensions 90 75

CNT (wt.%) 0 0.33 0.17; 1.65 0.66;

60

d (0)

44

45 30

Hybrid CNT (wt.%)

15 0 100

0.17; 0.66; 101

0.33 1.65 102

103

104

105

106

G* (Pa) Fig. 2.10 Van Gurp-Palmen plots of phase angle (δ) versus complex modulus (G*) for all particles.

Carbon Nanotube Hybrids and Their Polymer Nanocomposites

show steep curves that become increasingly insensitive to G* at higher loading. In the case of 1.65 wt.% CNTs, the curve is a vertical line for δ ranging from 0° to 9.5°, at a fixed G* of 0.6 MPa. The starting points on the G* axis increase (0.3, 2.3, 6.2, and 600 kPa) for the four CNT loadings and the range of δ drops and becomes much narrower at a higher loading. The G* increasing and the drop in the range of δ values indicate that the suspension behaves very elastically, similar to trends observed in the literature. For example, very low concentrations of CNTs have shown “horizontal” curves (δ  90°) for the whole range of G* in various melt systems such as polycarbonate [63], poly(butylene succinate) [64], and poly(butylene terephthalate) [65], whereas higher loadings generated increasingly elastic behavior, marked by an increase in the starting value of G* and a decrease in δ [63–65]. It has been discussed by Barnes et al. [66] (and references therein) that incorporation of spheres in water resulted in a minimum increase in viscosity compared with other particle shapes. It was shown that, at the same loading, rod-shaped particles caused the viscosity to increase significantly and that the higher the aspect ratio of the rods, the higher increase in the viscosity at a given loading. The fact that CNT-silica hybrid particles are essentially spherical and conventional CNTs have a high aspect ratio is the main reason for the differences in viscosity data observed in this work. Therefore, regardless of the particle loading applied, the rheological behavior of suspensions containing these spherical CNT hybrids is similar to that of the neat resin, which is in stark contrast to the viscosities of the suspensions of high-aspect-ratio CNTs.

2.3.2 Mechanical and Thermal Properties of Composites Containing Carbon-Hybrid Nanomaterials CNT hybrids are used to assist the dispersion of CNTs within polymer matrices. Due to the improved dispersion of CNTs, it is expected that the mechanical and thermal properties of the resultant composites would also improve compared with composites containing conventional CNTs/substrate under the same processing conditions. Table 2.1 summarizes (chronologically) the improvements in mechanical properties of the resultant composites upon incorporating CNT hybrids, listed down. The results listed in Table 2.1 show that growing CNTs on to substrates provided an alternative approach in improving the degree of dispersion within polymer matrices. These dual-filler systems also provided additional benefits to the composites. As the CNTs remain attached to the substrates, the improvements in properties are often due to the improvement in dispersion degree and strong interfacial interactions between fillers and matrix [41].

2.3.3 Electrical Properties of Composites containing Hybrid CNT-Silica This subsection discusses in detail the electric properties of the composites containing the CNT-silica hybrid particles synthesized by Othman et al. [43], where CNT was grown

45

46

Epoxy

• Enhancement in interlaminar strength of the composite

Carbon fiber

Electrophoresis

Carbon fiber

CVD

Epoxy



Stainless steel particle

CVD

Polyurethane



Alumina spherical particle

CVD

Epoxy



Alumina oxide particle

CVD

Epoxy

• •

Muscovite filler

CVD

Epoxy

• • •

by 27% by attaching 0.25% CNTs, compared with using carbon fiber alone In single fiber pullout tests, the apparent interfacial shear strength increased from 75 (carbon fiber alone) to 118 MPa (CNTs on carbon fiber) Tensile strength increases up to 19% for composites containing CNTs grown on stainless steel (SS) particles, compared with the composites containing mixtures of SS and CNTs (physical mix) At all loading up to 0.15 wt.%, the values of thermal conductivity of the composites were higher for hybrid particles, compared with composites containing CNT alone The flexural strength of composites containing 5 wt.% CNT-Al2O3 hybrid particle was 132.62 MPa, about 30% increment from neat epoxy Compared with 123.31 MPA for composites containing 5 wt.% CNT-Al2O3 (physical mixture), about 21% increment from neat epoxy The tensile properties of the composites containing hybrid CNT-MUS at 5 wt.% loading increased  86% compared with neat CNT Tensile modulus increased  1345% for the composites containing hybrid CNT-MUS at 5 wt.% loading compared with neat epoxy The microhardness properties increased 10% for composites containing hybrid CNT-MUS at 5 wt.% loading compared with the neat epoxy

References

2007

[31]

2008

[32]

2010

[36]

2010

[67]

2015

[40]

2016

[42]

Synthesis, Technology and Applications of Carbon Nanomaterials

Table 2.1 Mechanical and thermal properties enhancement upon incorporating CNT hybrids within polymer matrices Synthesis method of the Polymer Substrate hybrid CNT matrix Improvements Year

Alumina oxide particle Glass fiber

CVD

CVD

Epoxy

Epoxy

• Compressive strength increased  73% for composites •

• • Alumina spherical particle

CVD

Epoxy



Carbon fiber

CVD

Epoxy



2016

[41]

2017

[68]

2017

[69]

2017

[33]

Carbon Nanotube Hybrids and Their Polymer Nanocomposites



containing 5 wt.% CNT-Al2O3 hybrid particle, compared with neat epoxy The flexural strength of the composites increased from 658 (composites containing glass fiber only) to 707 MPa when hybrid CNT-glass fiber was applied at 1.5 wt.% CNTs Young’s modulus increased from 64 (single glass fiber) to 76.8 GPa (0.2 wt.% CNT-single glass fiber) Strength at break increased from 1085 (single glass fiber) to 1538 MPa (7 wt.% CNT-single glass fiber) Thermal diffusivity at room temperature increased from 0.147 (neat epoxy) to 0.283 mm2/s when CNT-Al2O3 loading was increased to 10%, corresponding to a 92% enhancement. Young’s modulus increases from 2.12 (neat epoxy) to 2.7 GPa (0.1 wt.% CNT-Al2O3 loading), which corresponds to a 26% enhancement The interlaminar shear strength of the control composite (carbon fiber/epoxy) increased from 23.45  1.85 to 31.62  1.10 MPa, upon using CNT-CF fillers, increasing by 34.8  11.6%.

47

Synthesis, Technology and Applications of Carbon Nanomaterials

on the surface of spherical silica gel particles. The polymer matrices used are poly(vinyl alcohol) (PVOH) and epoxy resin. Poly(vinyl alcohol) has become a common medium in which to study dispersion behavior as a wide range of carbon nanoparticles can be readily dispersed in a PVOH matrix including CNTs [70–72], carbon black (CB) [73], carbon nanodisks [74], and graphene [75]. Due to this, the dispersion of CNT-silica hybrid within a PVOH matrix will be presented before the results of incorporating the particles into epoxy resin, a more complicated and widely used polymer. The electric properties of a CNT-polymer nanocomposite can be analyzed from values of impedance (Z*) response measured by applying an alternating current to a specimen. The real part of the impedance (Z0 ) represents a material’s resistance, while the imaginary part of the impedance (Z00 ) represents the reactance, which normally consists of capacitance that stores energy in an electric field. Values of Z0 can be plotted against Z00 for every frequency taken, where the shapes of the curves characterize the “circuit equivalence” of the specimen [76]. Such a plot is known as a Nyquist plot, as shown in Fig. 2.11A and B [43]. This technique has proved useful and is widely used in the literature to analyze the Z0 and Z00 components [72, 77–79].

–0.4

Z⬙ (kW)

–0.3 –0.2

20 wt.% 10 wt.%

Increasing frequency

–0.1 0.0 0.0

0.2

0.4

(A)

0.6

0.8

Z⬘ (kW)

–9

C1

C2

R1

R2

Increasing frequency –6 Z⬙ (kW)

48

Rc

–3

0

(B)

5 wt. % 2.5 wt. % 1.25 wt. % 0

3

6

9 Z⬘ (kW)

12

15

18

(C)

Cluster

Interface

Fig. 2.11 Nyquist plots for CNT-silica hybrids in PVOH for loadings of 20 and 10 wt.% (A) and 5, 2.5, and 1.25 wt.% (B) and the proposed equivalent circuit model (C) [43]. (Reproduced with permission from R.N. Othman, I.A. Kinloch, A.N. Wilkinson, Synthesis and characterisation of silica-carbon nanotube hybrid microparticles and their effect on the electrical properties of poly(vinyl alcohol) composites, Carbon 60 (0) (2013) 461–470.)

Carbon Nanotube Hybrids and Their Polymer Nanocomposites

To begin with, the circuit model shown in Fig. 2.11C [43] is proposed, which consists of Rc, cluster, and interface components. A similar model that consists of cluster and interface components was suggested earlier by Garret et al. [77] and Battisti et al. [79] for CNT-based composites and was found to fit the experimental data well. The cluster component comes from the dispersed conductive particles in PVOH. The remainder of the specimen without agglomerates can be referred to as interface. The impedance data (from the Nyquist plot) intercepts the x-axis (Z0 ) at high frequency (Fig. 2.11A and B). The Z0 value at this intercept denotes the high frequency resistance and is referred to as Rc. This high frequency resistance was also observed by Ahn et al. [80] and interpreted as the electric contact resistance. In a separate study, Peng et al. [81] reported that the value of high frequency resistance decreases for the higher conductivity electrodes. As two terminal measurements were used in this work [43], Rc must have occurred due to contact resistance, in agreement with [80]. It is further assumed that each cluster and interface component is constructed from a resistor and a capacitor connected in parallel. These components, which composed of cluster, interface, and Rc, are proposed to be connected in series (Fig. 2.11C). The impedance (Z*) equations that relate Z0 and Z00 with the electric components shown in Fig. 2.11C are as follows [43]: 00

Z ∗ ðωÞ ¼ Z 0 ðωÞ  jZ ðωÞ Z 0 ðωÞ ¼ Rc + 00

Z ðωÞ ¼

(2.1)

R1 R2 2 + 1 + ðωC1 R1 Þ 1 + ðωC2 R2 Þ2

(2.1a)

ωC1 R12 ωC2 R22 2 + 1+ 1 + ðωC1 R1 Þ

(2.1b)

where Z* is the complex impedance, Z0 is the real part of the impedance, Z00 is the imaginary part of the impedance, and ω (2πf) is the angular frequency (where f is the frequency in Hz). R1 and C1 are the resistor and capacitor that form the cluster circuit, R2 and C2 are the resistor and capacitor that form the interface circuit, and Rc is the resistance at high frequency. The plots of Z0 and Z00 are shown in Fig. 2.11A and B. They are semicircular in shape, where the perimeter becomes bigger as the sample becomes less conductive, as also observed in other studies [72, 77]. The data fit is also shown on the same graph for each loading. As the fit intercepts the experimental data at almost every point, the proposed model is valid in representing the circuit equivalence of the samples for all loadings between 20 and 1.25 wt.%. For loadings less than 1.25 wt.%, the plots do not exhibit the semicircular shape, due to high resistance. It is also apparent that the plot intercepts the x-axis at high frequency, emphasizing the emergence of a high frequency resistance, Rc. The fitting parameters for the SG6_3 hybrid samples are shown in Table 2.2 [43]. Generally, the values of R1 are less than the values of R2 for all loadings. This is expected,

49

50

Synthesis, Technology and Applications of Carbon Nanomaterials

Table 2.2 Fitting parameters for the circuit model proposed in Fig. 2.11 for the SG6_3 hybrid samples Reinforcement Interface p Quality R1 (Ω) C1 (F) R2 (Ω) C2 (F) (wt.%) Rc (Ω) of fit, χ 2

20 10 5 2.5 1.25

6.5  0.3

1.1  0.2

(2.1  0.9)  107 9.8  2.4 19.7  2.3 (6.6  1.6)  109 130.6  51.7 199.0  48.6 (6.9  3.6)  109 190.0  66.0 228.2  9.7 (6.0  0.2)  109 418.3  151.0 382.4  17.9 (7.6  0.1)  109

195.1  1.2

(2.9  0.1)  106 740.5  4.5 (7.5  0.1)  107 8026.0  65.8 (7.4  0.1)  108 9270.0  256.0 (6.4  0.2)  108 16130.0  374.2 (3.5  0.1)  108

0.001 0.001 0.001 0.012 0.008

Reproduced with permission from R.N. Othman, I.A. Kinloch, A.N. Wilkinson, Synthesis and characterisation of silicacarbon nanotube hybrid microparticles and their effect on the electrical properties of poly(vinyl alcohol) composites, Carbon 60 (0) (2013) 461–470.

as R1 is the resistive element from cluster components. The plots indicate the existence of the high frequency resistance Rc, which implies that the contact resistance from the current collector contributes to the emergence of Rc. This is expected as the data were collected from a two terminal sensing set up in this impedance experiment. It also becomes apparent that the Rc values varied from 6.5 (20 wt.%) to 418.3 Ω (1.25 wt.%). It is thought that the silica within the PVOH is mainly responsible for the shift of Rc to higher values at lower loadings. It is proposed that the study of PVOH containing the silica substrate without CNTs might provide clues in addressing this observation. Conductive fillers can enhance the electric properties of insulative polymers. Generally, the presence of low loading of fillers causes little change in a composite’s electric properties; however, there exists a critical filler loading at which a sample’s conductivity will increase by several orders of magnitude. This critical loading is known as the critical percolation threshold loading, pc. Addition of more filler beyond pc typically only gives moderate further increases in conductivity. As such, the region where this extreme change in electric properties occurs is important and is a common subject of investigation [82]. It would be an added advantage to achieve pc at low loadings, as high CNT content would tend to reduce the mechanical properties of a composite [22]. Percolation theory describes the random distribution of a noninteracting group of particles (spheres) in a medium [83], that is, how these particles are connected into clusters and relating their connectivity with the resulting properties of the whole system. The theory has been widely used in explaining changes in electric properties of initially insulating polymers as a result of adding electrically conductive fillers, particularly CNTs [18, 71, 72]. Depending on the distribution of these fillers, an insulative polymer may become electrically conductive. At low filler loadings, the density of clusters is low; as

Carbon Nanotube Hybrids and Their Polymer Nanocomposites

a result, the clusters are disconnected, and a conductive network is not formed, resulting in an insulative composite. However, continuous addition of conductive filler increases the density of clusters to a point at which a connected network starts to form. Accordingly, this connected network, better known as percolated network, allows electrons to travel within the composite. After reaching this stage, further addition of conductive particles will give only a moderate increase in the system’s conductivity. Percolation theory describes the critical loading (pc) at which this drastic change occurs, the dimensionality of the formed percolated network, and the conductivity of the composites above the threshold. The equation derived from percolation theory that describes the connectivity of the conductive cluster-CNT composite within this critical region is [71, 84] σ ðpÞ ¼ σ c ðp  pc Þt ; for p > pc

(2.2)

where σ is the specific conductivity of the composite, σ c is the proportionality constant, pc is the critical loading (wt.%), and p is the particle loading (wt.%). In theory, the power constant, t, describes the dimensionality of the clusters where t values of 1.33 and 2.0 represents two-dimensional and three-dimensional clusters, respectively [83]. The t obtained for the SG6_3 hybrid composites [43] was 1.987 indicating that a three-dimensional conductive network was formed. Zhang et al. [72] reported a t value of 1.73  0.04 CNT-PVOH composites and concluded the presence of a threedimensional conductive network. Kilbride et al. [71] measured a t value of 1.36, suggesting the formation of a two-dimensional network of CNTs within PVOH matrices. On the other hand, Hernandez et al. [74] reported a value of t > 2 for their CNT-PVOH composites of 2.7  0.5, which was due to a polymer coating separating the CNT clusters. Based on this [71, 74], it may be implied that only a relatively narrow polymer thickness exists in between the SG6_3 CNT hybrid-PVOH composites [43]. Due to this, the fitting gives a t value close to that predicted by theory for clusters connected in a threedimensional network. The following paragraphs describe an analysis of electric conductivity properties of CNT-silica hybrid particles within an epoxy resin matrix. Impedance plots for SG6_3 CNT-silica-epoxy composites are shown in Fig. 2.12 [30]. The data exhibit a semicircular shape for all loadings, which implies that the impedance response is generated by both cluster and interface components. The equivalent circuit model shown in Fig. 2.12D is proposed to represent the impedance data, and the fitting parameters are shown in Table 2.3 [30]. The data fit intercepts the experimental data at almost every point, which confirmed the validity of the proposed model in describing the electric conductivity of the composites. The values of R1 are less than the values of R2 for all loadings, which is expected, as the R1 is the resistive element from cluster components. The difference between R1 and R2 becomes more apparent at higher loadings, which demonstrates the effectiveness of the CNT-silica hybrids in forming a conductive network within the epoxy matrix.

51

Synthesis, Technology and Applications of Carbon Nanomaterials

–140

–6

Z⬙ (kW)

Z⬙ (kW)

Increasing frequency –3

–70 0.66 wt.% SG6_3 experimental fit

1.65 wt.% SG6_3 experimental fit

0

0 0

0

12

6 Z⬘ (kW)

(A)

140 Z⬘ (kW)

(B)

280

–2

C1 Z⬙ (MW)

52

C2

–1

0.33 wt.% SG6_3 experimental fit

0 0

(C)

2

4

R1

R2

Cluster

Interface

(D)

Z⬘ (MW)

Fig. 2.12 Nyquist plots for CNT-silica hybrids in an epoxy resin for loadings of 1.65 (A), 0.66 (B), and 0.33 wt.% (С). The square symbols are the experimental data, while the dashed line is the fitted model. The direction of the increasing frequency is shown in (A). The circuit equivalent model representing the electric properties of these composites (D). (Reproduced with permission from R.N. Othman, A.N. Wilkinson, The impedance characterization of hybrid CNT-silica epoxy nanocomposites, IJAME 10 (1) (2014) 1832–1840.) Table 2.3 Fitting parameters for the circuit model proposed in Fig. 2.12D for the SG6_3 hybrid samples Coefficient of Clusters Interface determination, R2 p (wt.%) R1 (Ω)

C1 (F)

R2 (Ω)

C2 (F)

Z0

Z00

1.65 0.66 0.33

2.9  107 9  1010 1  108

1.1  104 2.6  105 3.6  106

5  1011 5.4  1011 5  106

0.999 (7.5  0.1)  10 7 (7.4  0.1)  108

0.999 0.001 0.001

40 1.4  103 7.1  104

Reproduced with permission from R.N. Othman, A.N. Wilkinson, The impedance characterization of hybrid CNTsilica epoxy nanocomposites, IJAME 10 (1) (2014) 1832–1840.

Fig. 2.13 [30] shows the plot of specific conductivity obtained at 100 Hz as a function of filler loading. Fitting yielded a pc of 0.16 wt.% with R2 ¼ 0.9946. Given the same matrices and processing conditions, the percolation threshold of the CNT-silica hybrid was an order of magnitude higher than the conventional CNTs [30]. The following paragraphs discuss reasons for this.

Carbon Nanotube Hybrids and Their Polymer Nanocomposites

Fig. 2.13 Semilog plot of specific conductivity as a function of p. The inset showing a fit to the equation is derived from percolation theory. (Reproduced with permission from R.N. Othman, A.N. Wilkinson, The impedance characterization of hybrid CNT-silica epoxy nanocomposites, IJAME 10 (1) (2014) 1832–1840.)

The orientation of conductive fillers within a matrix greatly affects the formation of a percolated network. To form SG6_3, CNTs were grown within the pores of the silica and on the substrate’s spherical surface. As such, the CNTs are uniformly distributed along the substrate’s surface, radially. This geometry possesses an advantage such that any properties imparted by the hybrid particles to the matrices are isotropic in nature. This particle arrangement is preferred if an improvement in the mechanical properties of a composite by adding conductive reinforcement is required, whereas conventional CNTs may freely rotate within a matrix depending on the shear force applied during processing. Du et al. [85] showed that for SWNTs in PMMA, the highest conductivity was achieved for slightly aligned SWNTs compared with isotropic SWNTs. In a separate study, Behnam et al. [86] performed Monte Carlo simulations to calculate the electric resistivity resulting from CNT alignment and reached the same conclusion as [85]. Moreover, aggregated clusters of CNTs with random CNT orientations have been shown to be crucial in achieving a conductive network in both epoxy resins [18, 85, 87, 88] and polymer melts [88, 89]. By manipulating the shear forces during processing, CNTs may form aggregated clusters that allow percolation to take place at loadings as low as 0.0025 wt.% [18]. In the case of CNT-silica hybrids, the strong interaction between the silica and CNTs gives a stable structure, which will prevent aggregation of CNTs to form clusters. Based on this argument, it is the stable isotropic structure of the CNT-silica hybrids that results in the order of magnitude higher percolation threshold shown in the cured epoxy in this work compared with the nongrafted CNTs.

53

Synthesis, Technology and Applications of Carbon Nanomaterials

The aspect ratio of a particle is defined as the ratio of its length to its diameter. CNTsilica hybrids consist of CNTs grown radially over its surface. Macroscopically, therefore, the aspect ratio of CNT-silica hybrids may be approximated to that of a sphere, which is 1. It is known that particle geometry or more accurately the aspect ratio greatly influences the value of pc. Higher aspect ratio results in a percolated network forming at lower loading and vice versa. For example, Hernandez et al. [74] measured the pc values of PVOHbased nanocomposites, upon incorporating carbon nanodisks (CNDs) of lower aspect ratio and CNTs of higher aspect ratio, to be 2.1 and 0.4 vol%, respectively, a factor of 5 difference. Similarly, Sandler et al. showed that the value of pc obtained from an essentially spherical carbon black was at least two orders of magnitude higher than for CNTs [18]. Martin et al. [87] discussed relations between the aspect ratio of CNTs and their percolation threshold, with the particle’s excluded volume serving as the upper limit for the determination of the threshold, as illustrated in Fig. 2.14. The shaded area is based on the calculation of excluded volume for high-aspect-ratio particles, as calculated by Celzard et al. [90], the black squares are experimental data from [29], and the experimental results for CNT-silica hybrid particles obtained from [30] are shown as red circles. It becomes apparent that the percolation thresholds obtained in [30] and also by Martin et al. [87] are a few orders of magnitude lower than those predicted by excluded volume theory. Martin et al. [87] reasoned that this was due to the fact that the original statistical 102

Percolation threshold (wt%)

54

Theoretical prediction (excluded volume approach)

101

100

10-1

10-2 Experimental percolation thresholds 10-3 1

10

100 Aspect ratio

1000

10000

Fig. 2.14 The relationship between percolation threshold (wt.%) and a filler’s aspect ratio determined from the excluded volume approach, adapted from Martin et al. [87]. The shaded area represents the limit of percolation threshold for high-aspect-ratio fillers, previously calculated by Celzard et al. [90]. The back squares are the experimental pc obtained by Martin et al. [87], while the red circles are the pc of CNT-silica hybrid. (Reproduced with permission from C.A. Martin, et al., Formation of percolating networks in multi-wall carbon-nanotube-epoxy composites, Compos. Sci. Technol. 64 (15) (2004) 2309–2316.)

Carbon Nanotube Hybrids and Their Polymer Nanocomposites

percolation ignored interparticle or matrix-particle interactions. Since the van der Waals forces that exist within CNT bundles are not taken into account, the resulting pc is rather low compared with those estimated by excluded volume theory. This is further supported by other studies that revealed extremely low values of percolation thresholds in their CNT composites [18, 59, 91]. It is proposed that addition of hardener may also cause the CNT-silica hybrid particles to form clusters, which eventually lead to the formation of a percolated network at low pc compared with the predictions of excluded volume theory. Without hardener, network formation does not occur. According to percolation theory, the t values of 1.33 and 2.0 reflect the dimensionality of a network formed in two dimensions and three dimensions, respectively. However, the values of t obtained in [30] was 2.94. Although percolation theory has been widely used to describe the critical loading that causes an insulator-conductor transition in composites, a wide range of t values have been reported in the literature [18, 71, 82, 87]. For example, Sandler et al. [18] reported a t value of 1.2 for their aligned CNTepoxy composites with a percolation threshold of 0.0025 wt.%. The authors argued that the lower values recorded in their work did not imply a reduction in the network’s dimensionality. Percolation theory was initially derived from a random distribution of particles; however, in their case, the CNTs reaggregate after the addition of hardener. Similarly, Kilbride et al. [71] reported a t value of 1.36, which may suggest the formation of a two-dimensional network of CNTs within PVOH matrices. However, as the charges must have traveled through the film, not just on the surface, it seems implausible that a two-dimensional network had formed. According to Bauhofer and Kovacs [82], there was also no correlation between the t value and the maximum conductivity and percolation threshold and concluded that the dimensionality of a percolated CNT network cannot be determined directly by evaluating the t value. These are also thought to be the reasons for the deviation for t values determined experimentally [30] as opposed to those stated theoretically.

2.4 CONCLUSIONS This chapter highlights a method of addressing the dispersion problem for CNTs, mainly in the composite manufacturing process. Growing CNTs on the surface of a substrate is suggested as a method that could allow a homogeneous dispersion of CNTs within matrices if the substrate is of micron scale. As such, problems related to agglomeration due to van der Waals interaction could be minimized. Various research groups have reported enhancement in composite properties such as mechanical, thermal, and electric properties, by using hybrid CNT. It would therefore be interesting to explore the effects of incorporating these hybrid particles on other properties. An extensive discussion has been provided on the rheological analysis of epoxy suspensions containing CNTs grown on a spherical silica substrate. For the first time, it was

55

56

Synthesis, Technology and Applications of Carbon Nanomaterials

reported that CNTs could be added to low viscosity resins, like epoxy, at high loadings without causing an increase in the suspension viscosity [23]. Further study on the curing kinetics and curing behavior of epoxy resins containing these hybrid particles could provide intriguing findings. To date, the CNT hybrids were prepared mainly via batch CVD processes, which means that the rate of production is rather low. This is due to the extremely confined space for the substrate (approximately one layer of silica gel) within the heating zone [43]. As such, this presents a serious hurdle if these particles are to be produced on a large scale. Because of these issues, the synthesis of CNT hybrids is proposed to be performed using fluidized-bed reactors. It has been reported that various CNT manufacturers, such as Arkema and SouthWest NanoTechnologies, have synthesized CNTs and SWNTs produced on a large scale using fluidized-bed reactors. If this can be realized, more applications can benefit from using these hybrid particles.

ACKNOWLEDGMENTS We would like to thank the Ministry of Higher Education, Malaysia, for funding under research grants FRGS/2/2013/TK04/UPNM/03/1 and RACE/F2/TK/UPNM/1 and Prof. Ian Kinloch from The University of Manchester for useful discussions.

REFERENCES [1] S. Iijima, T. Ichihashi, Single-shell carbon nanotubes of 1-nm diameter, Nature 363 (6430) (1993) 603–605. [2] D.S. Bethune, et al., Cobalt-catalyzed growth of carbon nanotubes with single atomic-layer walls, Nature 363 (6430) (1993) 605–607. [3] N. Grobert, Carbon nanotubes-becoming clean, Mater. Today 10 (1-2) (2007) 28–35. [4] N. Hamada, S. Sawada, A. Oshiyama, New one-dimensional conductors-graphitic microtubules, Phys. Rev. Lett. 68 (10) (1992) 1579–1581. [5] P.-C. Ma, et al., Dispersion and functionalization of carbon nanotubes for polymer-based nanocomposites: a review, Compos. A: Appl. Sci. Manuf. 41 (10) (2010) 1345–1367. [6] M.S. Dresselhaus, G. Dresselhaus, R. Saito, Physics of carbon nanotubes, Carbon 33 (7) (1995) 883–891. [7] E.T. Thostenson, Z. Ren, T.-W. Chou, Advances in the science and technology of carbon nanotubes and their composites: a review, Compos. Sci. Technol. 61 (13) (2001) 1899–1912. [8] M.M.J. Treacy, T.W. Ebbesen, J.M. Gibson, Exceptionally high Young’s modulus observed for individual carbon nanotubes, Nature 381 (6584) (1996) 678–680. [9] A. Krishnan, et al., Young’s modulus of single-walled nanotubes, Phys. Rev. B 58 (20) (1998) 14013–14019. [10] E.W. Wong, P.E. Sheehan, C.M. Lieber, Nanobeam mechanics: elasticity, strength, and toughness of nanorods and nanotubes, Science 277 (5334) (1997) 1971–1975. [11] E. Hernandez, et al., Elastic properties of C and BxCyNz composite nanotubes, Phys. Rev. Lett. 80 (20) (1998) 4502–4505. [12] N. Yao, V. Lordi, Young’s modulus of single-walled carbon nanotubes, J. Appl. Phys. 84 (4) (1998) 1939–1943. [13] H.J. Dai, et al., Single-wall nanotubes produced by metal-catalyzed disproportionation of carbon monoxide, Chem. Phys. Lett. 260 (3-4) (1996) 471–475.

Carbon Nanotube Hybrids and Their Polymer Nanocomposites

[14] T.W. Ebbesen, et al., Electrical conductivity of individual carbon nanotubes, Nature 382 (1996) 54–56. [15] S. Liu, et al., A review of extending performance of epoxy resins using carbon nanomaterials, Compos. Part B 136 (2018) 197–214. [16] J.N. Coleman, et al., Small but strong: a review of the mechanical properties of carbon nanotubepolymer composites, Carbon 44 (9) (2006) 1624–1652. [17] Y. Liu, A. Wilkinson, Rheological percolation behaviour and fracture properties of nanocomposites of MWCNTs and a highly crosslinked aerospace-grade epoxy resin system, Compos. A: Appl. Sci. Manuf. 105 (2018) 97–107. [18] J.K.W. Sandler, et al., Ultra-low electrical percolation threshold in carbon-nanotube-epoxy composites, Polymer 44 (19) (2003) 5893–5899. [19] K.P. Ryan, et al., Carbon nanotubes for reinforcement of plastics? A case study with poly(vinyl alcohol), Compos. Sci. Technol. 67 (7) (2007) 1640–1649. [20] M.F. Islam, et al., High weight fraction surfactant solubilization of single-wall carbon nanotubes in water, Nano Lett. 3 (2) (2003) 269–273. [21] A. Godara, T. Vaugien, Carbon nanotubes in polymer matrix composite structures, JEC Compos. 46 (2009) 64–68. [22] M.S.P. Shaffer, J.K.W. Sandler, in: S.G. Advani (Ed.), Carbon Nanotube/Nanofibre Polymer Composites. Processing and Properties of Nanocomposites, World Scientific, New Jersey, 2007. [23] A.N. Wilkinson, I.A. Kinloch, R.N. Othman, Low viscosity processing using hybrid CNT-coated silica particles to form electrically conductive epoxy resin composites, Polymer 98 (2016) 32–38. [24] X.L. Xie, Y.W. Mai, X.P. Zhou, Dispersion and alignment of carbon nanotubes in polymer matrix: a review, Mater. Sci. Eng. R. Rep. 49 (4) (2005) 89–112. [25] I.A. Kinloch, S.A. Roberts, A.H. Windle, A rheological study of concentrated aqueous nanotube dispersions, Polymer 43 (26) (2002) 7483–7491. [26] M.S.P. Shaffer, X. Fan, A.H. Windle, Dispersion and packing of carbon nanotubes, Carbon 36 (11) (1998) 1603–1612. [27] M.S.P. Shaffer, A.H. Windle, Analogies between polymer solutions and carbon nanotube dispersions, Macromolecules 32 (20) (1999) 6864–6866. [28] R.T.K. Baker, Catalytic growth of carbon filaments, Carbon 27 (3) (1989) 315–323. [29] S. Abbasi, et al., Rheological properties and percolation in suspensions of multiwalled carbon nanotubes in polycarbonate, Rheol. Acta 48 (9) (2009) 943–959. [30] R.N. Othman, A.N. Wilkinson, The impedance characterization of hybrid CNT-silica epoxy nanocomposites, IJAME 10 (1) (2014) 1832–1840. [31] E. Bekyarova, et al., Multiscale carbon nanotube-carbon fiber reinforcement for advanced epoxy composites, Langmuir 23 (7) (2007) 3970–3974. [32] H. Qian, et al., Hierarchical composites reinforced with carbon nanotube grafted fibers: the potential assessed at the single fiber level, Chem. Mater. 20 (5) (2008) 1862–1869. [33] Y. Wang, et al., High interlaminar shear strength enhancement of carbon fiber/epoxy composite through fiber- and matrix-anchored carbon nanotube networks, ACS Appl. Mater. Interfaces 9 (10) (2017) 8960–8966. [34] L. Yan, et al., Carbon nanofiber arrays grown on three-dimensional carbon fiber architecture substrate and enhanced interface performance of carbon fiber and zirconium carbide coating, ACS Appl. Mater. Interfaces 9 (20) (2017) 17337–17346. [35] N. Yamamoto, et al., High-yield growth and morphology control of aligned carbon nanotubes on ceramic fibers for multifunctional enhancement of structural composites, Carbon 47 (3) (2009) 551–560. [36] X.H. Nguyen, et al., Synthesis of sea urchin-like particles of carbon nanotubes directly grown on stainless steel cores and their effect on the mechanical properties of polymer composites, Carbon 48 (10) (2010) 2910–2916. [37] H. Qian, et al., Synthesis and characterisation of carbon nanotubes grown on silica fibres by injection CVD, Carbon 48 (1) (2010) 277–286. [38] D.L. He, et al., Diameter- and length-dependent self-organizations of multi-walled carbon nanotubes on spherical alumina microparticles, Carbon 48 (4) (2010) 1159–1170.

57

58

Synthesis, Technology and Applications of Carbon Nanomaterials

[39] M.R. Zakaria, et al., Enhancement of tensile and thermal properties of epoxy nanocomposites through chemical hybridization of carbon nanotubes and alumina, Compos. A: Appl. Sci. Manuf. 66 (2014) 109–116. [40] M.R. Zakaria, et al., Improving flexural and dielectric properties of MWCNT/epoxy nanocomposites by introducing advanced hybrid filler system, Compos. Struct. 132 (2015) 50–64. [41] M.R. Zakaria, et al., Compressive properties and thermal stability of hybrid carbon nanotube-alumina filled epoxy nanocomposites, Compos. Part B 91 (2016) 235–242. [42] N.S.A.A. Bakhtiar, et al., New generation of hybrid filler for producing epoxy nanocomposites with improved mechanical properties, Mater. Des. 91 (2016) 46–52. [43] R.N. Othman, I.A. Kinloch, A.N. Wilkinson, Synthesis and characterisation of silica-carbon nanotube hybrid microparticles and their effect on the electrical properties of poly(vinyl alcohol) composites, Carbon 60 (0) (2013) 461–470. [44] A. Oberlin, M. Endo, T. Koyama, Filamentous growth of carbon through benzene decomposition, J. Cryst. Growth 32 (3) (1976) 335–349. [45] R. Viswanatha, D.D. Sarma, Growth of nanocrystals in solution, in: C.N.R. Rao, A. M€ uller, A. K. Cheetham (Eds.), Nanomaterials Chemistry: Recent Developments and New Directions, WileyVCH, Germany, 2007, pp. 139–170. [46] P.B. Amama, et al., Role of water in super growth of single-walled carbon nanotube carpets, Nano Lett. 9 (1) (2009) 44–49. [47] K. Hasegawa, S. Noda, Moderating carbon supply and suppressing Ostwald ripening of catalyst particles to produce 4.5-mm-tall single-walled carbon nanotube forests, Carbon 49 (13) (2011) 4497–4504. [48] Z.Y. Zhan, et al., The effects of catalyst treatment on fast growth of millimeter-long multi-walled carbon nanotube arrays, Appl. Surf. Sci. 257 (17) (2011) 7704–7708. [49] K. Esumi, et al., Chemical treatment of carbon nanotubes, Carbon 34 (2) (1996) 279–281. [50] S.S. Rahatekar, et al., Optical microstructure and viscosity enhancement for an epoxy resin matrix containing multiwall carbon nanotubes, J. Rheol. 50 (5) (2006) 599–610. [51] J.A. Kim, et al., Effects of surface modification on rheological and mechanical properties of CNT/epoxy composites, Carbon 44 (10) (2006) 1898–1905. [52] Y.S. Song, J.R. Youn, Influence of dispersion states of carbon nanotubes on physical properties of epoxy nanocomposites, Carbon 43 (7) (2005) 1378–1385. [53] A.T. Seyhan, et al., Rheological and dynamic-mechanical behavior of carbon nanotube/vinyl esterpolyester suspensions and their nanocomposites, Eur. Polym. J. 43 (7) (2007) 2836–2847. [54] L.A. Hough, et al., Viscoelasticity of single wall carbon nanotube suspensions, Phys. Rev. Lett. 93 (16) (2004) 168102. [55] M.J. Kayatin, V.A. Davis, Viscoelasticity and shear stability of single-walled carbon nanotube/unsaturated polyester resin dispersions, Macromolecules 42 (17) (2009) 6624–6632. [56] C.W. Macosko, Rheology Principles, Measurements, and Applications, Wiley-VCH, New Jersey, 1994. [57] Z.H. Fan, S.G. Advani, Rheology of multiwall carbon nanotube suspensions, J. Rheol. 51 (4) (2007) 585–604. [58] J. Sumfleth, S.T. Buschhorn, K. Schulte, Comparison of rheological and electrical percolation phenomena in carbon black and carbon nanotube filled epoxy polymers, J. Mater. Sci. 46 (3) (2011) 659–669. [59] D.S. Bangarusampath, et al., Rheological and electrical percolation in melt-processed poly(ether ether ketone)/multi-wall carbon nanotube composites, Chem. Phys. Lett. 482 (1-3) (2009) 105–109. [60] M. Chapartegui, et al., Specific rheological and electrical features of carbon nanotube dispersions in an epoxy matrix, Compos. Sci. Technol. 70 (5) (2010) 879–884. [61] M. van Gurp, J. Palmen, in: Time-temperature superposition for polymeric blends, Xiith International Congress on Rheology, Proceedings, Univ Laval, Dept Mech Engn, Quebec City, 1996. [62] E.K. Hobbie, D.J. Fry, Rheology of concentrated carbon nanotube suspensions, J. Chem. Phys. 126 (12) (2007) 124907. [63] P. P€ otschke, et al., Rheological and dielectrical characterization of melt mixed polycarbonatemultiwalled carbon nanotube composites, Polymer 45 (26) (2004) 8863–8870.

Carbon Nanotube Hybrids and Their Polymer Nanocomposites

[64] G. Wang, et al., Rheology, crystallization behaviors, and thermal stabilities of poly(butylene succinate)/ pristine multiwalled carbon nanotube composites obtained by melt compounding, J. Appl. Polym. Sci. 121 (1) (2011) 59–67. [65] D. Wu, L. Wu, M. Zhang, Rheology of multi-walled carbon nanotube/poly(butylene terephthalate) composites, J. Polym. Sci. B Polym. Phys. 45 (16) (2007) 2239–2251. [66] H.A. Barnes, J.F. Hutton, K. Walters, An Introduction to Rheology, Elsevier BV, Oxford, 1989. [67] M. Bozlar, et al., Carbon nanotube microarchitectures for enhanced thermal conduction at ultra low mass fraction in polymer composites, Adv. Mater. 22 (14) (2010) 1654. [68] D. He, et al., Design of electrically conductive structural composites by modulating aligned CVD-grown carbon nanotube length on glass fibers, ACS Appl. Mater. Interfaces 9 (3) (2017) 2948–2958. [69] D. He, et al., Multifunctional polymer composites reinforced by carbon nanotubes-alumina hybrids with urchin-like structure, Mater. Today Commun. 11 (2017) 94–102. [70] M.S.P. Shaffer, A.H. Windle, Fabrication and characterization of carbon nanotube/poly(vinyl alcohol) composites, Adv. Mater. 11 (11) (1999) 937–941. [71] B.E. Kilbride, et al., Experimental observation of scaling laws for alternating current and direct current conductivity in polymer-carbon nanotube composite thin films, J. Appl. Phys. 92 (7) (2002) 4024–4030. [72] J. Zhang, et al., Electrical and dielectric behaviors and their origins in the three-dimensional polyvinyl alcohol/MWCNT composites with low percolation threshold, Carbon 47 (5) (2009) 1311–1320. [73] J. Barkauskas, A. Vinslovaite, Investigation of electroconductive films composed of polyvinyl alcohol and graphitized carbon black, Mater. Res. Bull. 38 (8) (2003) 1437–1447. [74] Y.R. Hernandez, et al., Comparison of carbon nanotubes and nanodisks as percolative fillers in electrically conductive composites, Scr. Mater. 58 (1) (2008) 69–72. [75] X. Zhao, et al., Enhanced mechanical properties of graphene-based poly(vinyl alcohol) composites, Macromolecules 43 (5) (2010) 2357–2363. [76] R.A. Gerhardt, Impedance spectroscopy and mobility spectra, in: Encyclopedia of Condensed Matter Physics, Elsevier, London, 2005, pp. 350–363. [77] M.P. Garrett, et al., Separation of junction and bundle resistance in single wall carbon nanotube percolation networks by impedance spectroscopy, Appl. Phys. Lett. 97 (16) (2010) 163105. [78] R.Q. Ou, et al., Fabrication and electrical conductivity of poly(methyl methacrylate) (PMMA)/carbon black (CB) composites: comparison between an ordered carbon black nanowire-like segregated structure and a randomly dispersed carbon black nanostructure, J. Phys. Chem. B 110 (45) (2006) 22365–22373. [79] A. Battisti, A.A. Skordos, I.K. Partridge, Dielectric monitoring of carbon nanotube network formation in curing thermosetting nanocomposites, J. Phys. D. Appl. Phys. 42 (15) (2009) 155402. [80] D. Ahn, et al., Applying functionalized carbon nanotubes to enhance electrochemical performances of tin oxide composite electrodes for Li-ion battery, J. Power Sources 212 (0) (2012) 66–72. [81] Z. Peng, et al., High performance ordered mesoporous carbon/carbon nanotube composite electrodes for capacitive deionization, J. Mater. Chem. 22 (14) (2012) 6603–6612. [82] W. Bauhofer, J.Z. Kovacs, A review and analysis of electrical percolation in carbon nanotube polymer composites, Compos. Sci. Technol. 69 (10) (2009) 1486–1498. [83] D. Stauffer, A. Aharony, Introduction to Percolation Theory, Taylor and Francis, London, 1994. [84] I. Jurewicz, et al., Locking carbon nanotubes in confined lattice geometries—a route to low percolation in conducting composites, J. Phys. Chem. B 115 (20) (2011) 6395–6400. [85] F.M. Du, J.E. Fischer, K.I. Winey, Effect of nanotube alignment on percolation conductivity in carbon nanotube/polymer composites, Phys. Rev. B 72 (12) (2005) 121404. [86] A. Behnam, J. Guo, A. Ural, Effects of nanotube alignment and measurement direction on percolation resistivity in single-walled carbon nanotube films, J. Appl. Phys. 102 (4) (2007) 044313. [87] C.A. Martin, et al., Formation of percolating networks in multi-wall carbon-nanotube-epoxy composites, Compos. Sci. Technol. 64 (15) (2004) 2309–2316. [88] W. Bauhofer, et al., Shear-controlled electrical conductivity of carbon nanotubes networks suspended in low and high molecular weight liquids, Polymer 51 (22) (2010) 5024–5027.

59

60

Synthesis, Technology and Applications of Carbon Nanomaterials

[89] T. Skipa, et al., Shear-stimulated formation of multi-wall carbon nanotube networks in polymer melts, Phys. Status Solidi B Basic Solid State Phys. 246 (11-12) (2009) 2453–2456. [90] A. Celzard, et al., Critical concentration in percolating systems containing a high-aspect-ratio filler, Phys. Rev. B 53 (10) (1996) 6209–6214. [91] J. Sandler, et al., Development of a dispersion process for carbon nanotubes in an epoxy matrix and the resulting electrical properties, Polymer 40 (21) (1999) 5967–5971.