Composites of poly(ethylene terephthalate) and multi-walled carbon nanotubes
The unique and extraordinary mechanical, electrical and thermal conductivity properties of multi-walled carbon nanotubes (MWCNTs) make them ideal candidates as functional fillers for polymeric materials. Poly(ethylene terephthalate) (PET) is an important engineering plastic finding widespread application in automotive, electronic and packaging technologies. The key challenge to achieving enhanced polymer properties is efficient dispersion and distribution of MWCNTs in the polymer matrix. In this chapter, we describe the preparation of composites of a PET with MWCNTs via melt mixing and, employing many of the characterisation techniques described elsewhere in this book, we investigate the extent of MWCNT dispersion by correlating microscopic observations with electrical and rheological percolation measurements. We also report the effect of MWCNT addition on PET thermal stability using thermal gravimetry analysis, and on PET crystallisation behaviour using a combination of Fourier transform infrared and Raman spectroscopy, differential scanning calorimetry and wide angle X-ray diffraction.
Iijima reported the synthesis of MWCNTs in 1991 with the preparation of coaxial tubes of graphitic sheets (Iijima, 1991), and clearly identified the concentric arrangement of multi-walled carbon nanotubes using high resolution transmission electron microscopy. However, it is generally accepted that such structures had existed prior to this, but had not been characterised fully (see McClory et al., 2009 and references therein). Subsequently, the unique and extraordinary mechanical properties of carbon nanotubes (CNTs) were first demonstrated by Ajayan et al. during the cutting of thin slices of an epoxy–CNT composite (Ajayan et al., 1994). The CNTs retained structural integrity and were aligned normal to the cutting surface, and thus established the prospective use of CNTs as a mechanism of mechanical reinforcement for polymeric materials. Since this account by Ajayan et al. in 1994, there has been an exponential growth in research into polymer–CNT composites for a multitude of applications, including, but not limited to, aerospace, automotive, electronic, biomedical and technical textiles. The high aspect ratio and exceptional mechanical (tensile strength 63 GPa (Yu et al., 2000)), Young’s modulus 1TPa (Treacy et al., 1996), electrical (current density > 107 A/cm2 (Frank et al., 1998)) and thermal (thermal conductivity > 2000 W/mK Φ 9.8 nm (Fujii et al., 2005)) properties of MWCNTs have fuelled the desire to design and characterise a new class of advanced polymeric materials with tailored physical properties. A homogeneous composite nano- and/or microstructure achievable through excellent CNT dispersion and distribution, coupled with the efficient transfer of applied load across this large interfacial region are critical factors for the successful translation of CNT mechanical and other properties to a polymer matrix. Interfacial shear strength (ISS), a measure of polymer–CNT interactions, is dependent on the topography of the nanotube surface and chemical (e.g. hydrogen bonding) or van der Waals interactions between CNTs and polymer matrix (Schadler et al., 1998). Many indirect experimental measurements of polymer–CNT interfacial adhesion have been attempted but have employed other mechanical test methods incorporating mathematical models (Wagner et al., 1998) or computer simulations (Liao and Li, 2001) to generate relevant results. The first direct measurement of the ISS between a MWCNT and a polymer (polyethylene-butene) was performed using atomic force microscopy (AFM) (Barber et al., 2003). An individual MWCNT attached to an AFM tip was pulled out of the polymer and using the measured pull-out force, an average ISS was determined. This study determined an average ISS of 47 MPa, associated with covalent bonding existing between defect sites on the MWCNT and the polyethylene–butene matrix. Ideally, the addition of small quantities of CNTs will provide the plastics industries with solutions for material enhancement by utilising existing conventional polymer processing technologies, at a relatively low cost.
This chapter provides a comprehensive account of the preparation and characterisation of composites of an important engineering plastic, poly(ethylene terephthalate) (PET) with pristine MWCNTs, using many of the characterisation techniques described elsewhere in this book. The structure of the composites is examined using a combination of microscopic methods: polarised optical microscopy (POM), field emission scanning electron microscopy (FESEM) and high resolution transmission electron microscopy (HRTEM) across the length scales. The extent of CNT dispersion and distribution throughout the PET matrix is correlated with composite properties. In particular, the effect of MWCNT addition on PET crystallisation behaviour and kinetics is studied using a range of techniques, including Fourier transform infrared spectroscopy (FTIR), Raman spectroscopy and differential scanning calorimetry (DSC). The thermal stability of PET after MWCNT addition is assessed using thermo-gravimetric analysis. In the final section, the formation of both electrical and rheological percolated networks of CNTs in the PET matrix is investigated using volume resistivity and oscillatory rheology measurements, respectively.
Poly(ethylene terephthalate) (PET) is a thermoplastic polyester resin which can be synthesised by the condensation reaction between terephthalic acid (TPA) and ethylene glycol or ethanediol. It is an important industrial engineering polymer as a result of its relatively low cost/high performance comparison. Due to the polarity of the ester groups located in the polymer main chain, PET is a semi-crystalline polymer which exists as an imperfect two-phase system of interconnected crystalline and amorphous domains (Vaia, 2000). It is an important engineering plastic used widely in the automotive, electronic, packaging and textile fibre industries due to its high strength, high chemical and dimensional stability and dielectric properties. Closely monitoring processing parameters, such as temperature, screw speed, residence time, and in particular cooling profiles can effectively control PET morphology, chain orientation and subsequently the physical properties of PET. Despite significant research effort, the successful incorporation of CNTs into polymer matrices for property enhancement and reinforcement has been limited predominantly by inadequate CNT dispersion and distribution and poor polymer–CNT interfacial adhesion. The problems arise from the tendency of neighbouring CNTs to agglomerate due to the π–π stacking effect of the hexagonal carbon arrangement of the tube structure. PET–MWCNT composite production methodologies have been studied extensively to overcome these problems using various approaches involving solution mixing (Chen et al., 2009), in-situ polymerisation (Lee et al., 2005), and melt-mixing techniques (Kim et al., 2007). Numerous CNT pre-treatments are currently under investigation with the goal of achieving enhanced CNT dispersion and increased interfacial interaction.
Modification or treatment of the CNT surface prior to PET–CNT composite preparation has included nanotube purification by acid (Jin et al., 2008), inert argon plasma treatments (Ahn et al., 2003), chemical functionalisation by acid (Tzavalas et al., 2006), organic molecules (Yoo et al., 2008), and fluorine and hydrogen plasma treatments (Yu et al., 2003; Luo et al., 2009). A coagulation method described by Hu et al. involved the initial purification of MWCNTs by oxidation in air, followed by acid treatment in HCL, prior to ultrasonication in ODCB–phenol, followed by PET addition (Hu et al., 2006). The dispersion of MWCNTs in ODCB–phenol remained stable for up to several weeks. Pre-treatment of the tubes did not affect their mean length (5–15 μm) although opening of tube end-caps was observed (from TEM images) and the addition of carboxyl groups (− COOH) was confirmed by FTIR analysis. The low electrical and rheological percolation thresholds, 0.9 wt% and 0.6 wt% respectively, were attributed to the homogeneous dispersion of high aspect ratio COOH–MWCNTs and augmentation of tube–matrix interactions achieved by the design of the solution process and nanotube pre-treatment. In particular, the formation of a uniformly dispersed and distributed network of COOH–MWCNTs in PET in this study was due to the already stable dispersion of CNTs in ODCB–phenol. The CNTs resisted re-agglomerating on addition of PET during solution mixing, due to the enhanced interaction between PET and COOH–MWCNTs by the π–π conjugation of common aromatic groups and esterification of hydroxyl (PET) and carboxyl (MWCNT) groups. Evidence supporting improved interfacial interaction between PET and pre-treated MWCNTs was provided by Jin et al., by the preparation of acid and diamine functionalised MWCNTs prior to in-situ polymerisation (Jin et al., 2007). The improved dispersion of carboxyl and diamine functionalised CNTs in PET compared with PET–pristine MWCNTs could be explained both by the presence of functional groups and the ease of processing of shorter tubes, the latter present due to tube scission during acid pre-treatment. The promotion of interfacial interactions between PET and carboxyl and diamine groups on the MWCNT surface was evident from the absence of nanotube pull-out and increased tube surface wetting with PET. Raman spectroscopy, an effective tool for interfacial analysis of polymer–CNT composites, recorded an up-shift of the G-band position of diamine functionalised MWCNTs to a higher frequency (+ 10 cm−1) which was interpreted as the insertion of PET polymer chains between tubes, separated and dispersed to a greater extent than the pristine or acid-functionalised tubes. The diamine-functionalised MWCNTs yielded the largest improvement in mechanical properties (+ 350%, + 290%), in strength (for a 0.5 wt% CNT loading) and modulus (for a 2 wt% CNT loading), respectively. In contrast, the mechanical properties of the untreated CNTs dropped drastically, mainly due to large CNT agglomerates acting as stress concentration points within the bulk material. In a similar study by the same authors, MWCNTs first treated with nitric acid, then acetic anhydride, before the same in-situ polymerisation process achieved mechanical reinforcement similar to that obtained for diamine-functionalised tubes despite the reduction in CNT aspect ratio and associated tube damage due to the functionalisation process (Jin et al., 2008). Extensive damage to the structure of MWCNTs after treatment with concentrated nitric and sulphuric acid was evident from SEM imaging, although addition of 0.05 wt% COOH–MWCNTs to PET resulted in a 23% increase in storage modulus and accelerated the rate of crystallisation by acting as an efficient nucleating agent at very low loadings (Wang et al., 2007). During in-situ polymerisation of PET–MWCNT composites, ethylene-glycol (EG) is usually ultrasonicated with either pristine or pre-treated tubes before polymerisation. In order to produce homogeneous dispersions of functionalised CNTs in EG, it is important to select and attach structurally similar chemical moieties to the surface of CNTs that will form stable dispersions in EG prior to polymerisation. Wang et al. showed digital photographs of treated and pristine MWCNTs ultrasonicated in a range of solvents for 10 mins and allowed to stand for 15 days. Pristine and nitric acid pre-treated MWCNTs had completely or partially formed a precipitate after the 15 days had passed. Only COOH–MWCNTs added to H2O and EG formed stable dispersions and remained in solution, due to hydrogen bonding (Wang et al., 2007). These stable dispersions promoted the formation of nanotube networks and interfacial interactions of CNTs in PET matrices after esterification and polycondensation following TPA addition. Similarly, Lee et al. formed stable dispersions of EG and modified MWCNTs by grafting 4-methoxybenzoic acid and 4-ethoxybenzoic acid to pristine MWCNTs via a Friedel-Crafts acylation reaction in polyphosphoric acid, yielding MWCNTs with methoxybenzoyl (− MeO) and ethoxybenzoyl (− EtO) functional groups attached (Lee et al., 2005). Dispersions of EtO–MWCNTs in EG were significantly more homogeneous than pristine MWCNTs or MeO–MWCNTs in EG due to the chemical similarity of ethoxy groups on EtO–MWCNTs and EG molecules (OCH2CH2O) (Lee et al., 2005). SEM images indicated the alkoxybenzoyl functional groups uniformly coated the MWCNT surface, the tube diameter increased from 18 nm to 40 nm and 45 nm after the grafting of MeO and EtO groups, respectively. After nanocomposite formation, the fractured surface of the PET–EtO–MWCNT nanocomposite revealed a highly adhered layer of polymer to the surface of the EtO–MWCNTs, to the extent that the boundary between the matrix and CNT was indiscernible. Compared with pristine and MeO-functionalised MWCNTs, the compatibility of EtO–MWCNTs in EG prior to polymerisation is the governing factor in the production of homogeneously dispersed MWCNT in PET by in-situ polymerisation. However, from an industrial scale perspective, both solution and in-situ polymerisation processes are limited due to environmental issues regarding the extensive use of harmful organic solvents and also the high cost and low yield of materials. In contrast, melt-mixing of polymer–CNT composites is favoured by industry due to the convenient adaptation of conventional polymer processing technologies to the nanocomposite production process. Ideally, the addition of small quantities of CNTs could provide the plastics industry with advanced materials by utilising already existing manufacturing equipment and technologies at a relatively low cost.
Previous studies on PET–MWCNT composites prepared by melt mixing have focussed mainly on the effect of the extrusion process and CNT addition on PET chain conformations, crystallinity and crystallisation behaviour of PET (Tzavalas et al., 2006; Kim et al., 2007; Tzavalas et al., 2008). It is generally accepted there are three known PET chain conformations: a crystalline phase involving only trans crystalline segments (TC); an intermediate phase involving trans noncrystalline segments, i.e. not belonging to the crystalline phase (TX); and finally a phase lacking in general order consisting of gauche conformations (G) (Tzavalas and Gregoriou, 2008). Tzavalas et al. studied the trans and gauche conformations of PET chains in the presence of COOH–MWCNTs produced using a batch melt-mixing process, and reported the change in the relative distribution of these units throughout the PET matrix, thus altering its final crystalline content (Tzavalas et al., 2006). DSC analysis indicated the crystallisation temperature (Tc) increased and the full width at half maximum (FWHM) decreased, demonstrating the addition of CNTs’ increased PET crystallinity by acting as a nucleating agent and promoting the growth of more perfect crystallites. This effect was observed up to 2 wt% CNT loading, however, higher CNT loadings were difficult to disperse and resulted in reduced PET crystalline content. The addition of modified MWCNTs up to 2 wt% increased the proportion of trans conformers in the crystalline region which would originally have resided in non-crystalline regions. The newly promoted crystals formed were of a higher order and hence more perfect with a narrower size distribution (Tzavalas et al., 2006). FT-IR, Raman spectroscopy and DSC can readily monitor crystal transformations due to thermal treatment and changes in crystallinity due to the presence of CNTs. In other studies, Tzavalas et al. uniaxially stretched PET and PET–MWCNT composites to investigate the effect of MWCNTs on the chain conformation of PET using FTIR spectroscopy (Tzavalas and Gregoriou, 2008). After uniaxial stretching, low (1.5 wt%) CNT content had a minor effect on the crystallisation of PET as a small transition of gauche to trans crystalline conformations occurred. At higher CNT concentrations, increased PET crystal conformation transitions from G to TX and TX to TC occurred with a sharp increase at 3 wt% MWCNT loading. This behaviour was not observed in unstretched PET–MWCNT composites in a similar study by the same authors (Tzavalas et al., 2008). In contrast, a sharp decrease in crystalline content was observed for a 1 wt% CNT loading, before crystallinity increased again with increased CNT concentration. CNT reagglomeration and poor dispersion, observed by TEM imaging, at 1 wt% CNT are thought to account for the drop in crystallinity.
Chain orientation and crystallisation effects as a consequence of melt processing PET are of extreme importance, and of particular interest due to the dependence of material properties on these factors. The majority of studies on melt-processed PET–MWCNT composites to date have focussed on PET crystallinity and crystal structure (Tzavalas et al., 2008; Tzavalas and Gregoriou, 2008), chain conformations (Chen et al., 2009), effect of CNT addition and nucleation effects (Tzavalas et al., 2006), matrix–CNT interfacial interactions (Gao et al., 2008), and mechanical properties (Kobayashi et al., 2007). A small number of studies on melt mixed PET–CNT composites have focused on bulk material properties such as electrical conductivity, viscoelastic properties, thermal and dynamic mechanical properties. These properties are altered as a result of the melt processing regime employed and the extent of formation of continuous CNT networks throughout the PET matrix (Anand et al., 2007; Kim et al., 2007). Anand et al. melt processed PET and pristine SWCNTs in a lab-scale Haake Kneader at 270 °C. Improvements in modulus (50%) and strength (25%) were achieved on addition of 1 wt% SWCNTs, although the composites were embrittled due to the lack of interfacial interactions, evident from a 30% reduction in strain at break. The electrical conductivity of PET increased from 1017 S/cm to 107 S/cm and an electrical percolation threshold of just above 2 wt% SWCNTs was recorded. In a further study by Kim et al., PET–pristine-MWCNT composites were melt mixed using a Haake rheometer equipped with a twin-screw. Improvements in modulus (9%) and tensile strength (10%) were recorded on addition of 1 wt% MWCNTs. At higher CNT loadings, the formation of interconnected or network-like structures of MWCNT in the PET–MWCNT composites was detected, from rheological measurements (Kim et al., 2007).
Research into the melt-processing of PET–pristine-MWCNT composites by twin screw extrusion has not been widely explored. This chapter gives a comprehensive account of the characterisation of PET–pristine-MWCNT composites, focusing on composite morphology, CNT dispersion and orientation, crystal structure, crystallinity, non-isothermal crystallisation kinetics, viscoelastic and electrical properties and thermal degradation.
The PET used in this study was TS5 Laser + Melinar, IV = 0.82–0.84 dl/g, melt viscosity of 655 Pa.s at 295 °C, Tm = 250 °C, supplied by DuPont Ltd. The MWCNTs used in this study were produced using chemical vapour deposition (> 90% purity) and supplied by Sun Nanotech Co Ltd, from People’s Republic of China. These tubes had diameters between 10 nm and 30 nm and varied in length from 1 μm to 10 μm, giving an average aspect ratio of approximately 500.
All materials were dried for 12 hours at 80 °C prior to processing. The PET–MWCNT composites (4.5 g) were melt mixed using a Haake twin screw microcompounder for MWCNT loadings of 0, 0.1, 0.5, 1.0, 3.0, 7.0 and 10% by weight. The extruder barrel temperature was set at 255 °C and a constant screw speed of 25 rev/min and recycle time of 2 minutes was employed. Under constant screw speed, the torque measured at the motor increased from 8 N/cm for neat PET to 11.2 N/cm on addition of 10 wt% MWCNTs, see Fig. 18.1.
The viscosity of the composites increased with increasing MWCNT concentration due to the high aspect ratio of the CNTs. A similar increase in melt viscosity has also been reported by Pötschke et al. on addition of 15 wt% CNTs to polycarbonate (Pötschke et al., 2002).
Figures 18.2 (a) and 18.2 (b) show polarised optical photographs of virgin PET and the PET–1 wt% MWCNT composite. The samples were heated on a hot stage accessory operating between 250 and 270 °C and cooled at 4 K/min to room temperature. The images shown in Fig. 18.2 (a) and 18.2 (b) were captured immediately after crystallisation and spherulite impingement was complete. As can be observed, the addition of CNTs to PET significantly altered the crystalline texture of the polymer when compared to the neat polymer. The CNTs provide a nucleating site for PET crystal growth and initiate heterogeneous nucleation, resulting in the formation of smaller crystallites. The extent of CNT dispersion and distribution throughout the PET matrix was studied across the length scales using a combination of optical microscopy, FESEM (Joel 6400) and HRTEM (FEI Tecnai F20). For low MWCNT loadings (< 1 wt%), few large MWCNT agglomerates were observed, see Fig. 18.2 (b). However, at higher loadings, agglomerates of MWCNTs were readily seen in the PET matrix, particularly when the MWCNT loading was 10 wt%, see Fig. 18.2 (c). Typically, agglomerates having diameters between 1 μm and 5 μm were distributed in the PET matrix, although larger agglomerates up to 10 μm were also evident. Examination of the composites by SEM also revealed the CNTs, for lower loadings only, were highly dispersed in the PET matrix. By way of example, Fig. 18.2 (d) shows a representative SEM image of the PET–10 wt% MWCNT composite. At higher magnification (× 400 k), images were obtained using HRTEM, and individual MWCNTs were observed dispersed in the polymer matrix, see Fig.18.2 (e). In Fig. 18.2 (f), a crosssectional view of a MWCNT embedded in the PET matrix was obtained, where the concentric arrangement of CNT walls is identifiable. The CNT wall thickness was estimated to be up to 20 layers thick, giving a CNT outer diameter of 30 nm.
Fourier transform infrared spectroscopy (FTIR) has been employed to characterise and confirm chemically modified surfaces of CNTs’ post-organic treatment (Jin et al., 2007; Ahn et al., 2008). Normally, treatment with concentrated acids such as sulphuric or nitric, carboxyl groups attach to the CNT surface and new peaks evolve in the FTIR spectra of the acid-treated CNTs associated with carboxyl groups. Peaks appearing at 1721–1730 cm−1 are indicative of C-O stretching, at 1176 cm−1C-O stretching and a peak at 1400 cm−1 indicates O-H bending, all attributed to the addition of carboxyl functionality to CNTs (Jin et al., 2007; Ahn et al., 2008). In an FTIR study of acid and acetic anhydride functionalised CNTs, peaks attributed to carboxyl groups described in (Jin et al., 2007) and (Ahn et al., 2008) were obtained, but with an additional peak at 3000–2800 cm−1 corresponding to sp3 C-H stretching, indicating the presence of –CH3 groups on the surface of the MWCNTs used (Jin et al., 2008). Figures 18.3 (a) and 18.3 (b) show infrared spectra for neat PET, pristine MWCNTs and PET–MWCNT composites prepared in this study between 800 cm−1 and 4000 cm−1 and an expanded view of the region between 1000 cm−1 and 2000 cm−1, respectively (Perkin Elmer Spectrum 1000).
The peaks observed between 2800 cm−1and 4000 cm−1associated with inter- and intra-molecular hydrogen bonding decrease in intensity on addition of MWCNTs to PET, suggesting that the MWCNTs readily infiltrate between PET chains, see Table 18.1. Peaks displayed between wavelengths of 1900 cm−1 and 2200 cm−1 associated with carbon stretching vibrations and aromatic summation bands were also less intense for the composites. Studies on the correlation of PET chain conformations with the addition of MWCNTs has been investigated by Tzavalas et al. (Tzavalas et al., 2008), trans segments (1340 cm−1), gauche segments (1370 cm−1) and in-plane vibration (G structure 1018 cm−1) of ethylene glycol (EG) were also identified in previous work by Cole et al. (Cole et al., 2002a, 2002b). These assignments were used by Tzavalas to map the relative population of gauche and trans crystal conformations as CNT content was increased (Tzavalas et al., 2008). Other peaks assigned to trans segments of PET occur at 1098 cm−1 (carbonyl stretching) and 1259 cm−1 (in plane ester bond vibration). Figure 18.3 (b) shows the evolution of 2 peaks, one at 1342 cm−1 assigned to trans segments of EG and another at 1379 cm−1 assigned to gauche segments.
In work similar to that of Tzavalas et al. (Tzavalas et al., 2008), a study on the change in the proportion of PET chains from amorphous to crystalline regions was performed. In Figs 18.4 (a) and 18.4 (b) the ratio of normalised transmission intensities of a range of trans to gauche segments demonstrated a clear change in crystalline structure.
As CNT content was increased, the proportion of polymer chains residing in the trans crystalline phase increased relative to those residing in the gauche phase which indicated a clear transition to a more crystalline structure with the addition of MWCNTs. The CNTs initiate heterogeneous nucleation, inducing ordered crystal growth and alter composite morphology.
Raman spectroscopy has been widely used in the analyses of carbon nanotube materials (Han et al., 1989) and polymers, including PET (Du et al., 2004). This technique was used to evaluate the dispersion of MWCNTs in PET and the changes in the crystallinity of PET on addition of the CNTs. An Avalon Instruments Raman Station spectrometer was used to obtain Raman spectra of the PET–MWCNT composites using an excitation wavelength of 785 nm (1.58 eV) and a maximum power of 100 mW. However, in order to avoid laser damage to the samples, this power was reduced. A very long working distance (18 mm) lens was used to provide a laser spot size of over 200 μm as it was believed that a large spot size would give a more representative indication of the bulk material than a surface confocal spot. The spectral range collected was 250–3200 cm−1 in all cases as the advanced Echelle spectrograph employed in this system provided the ability to give this large spectral range in a single scan. The detector system used in this instrument was an Andor CCD camera, air-cooled to a temperature of − 50 K. The laser system used was unpolarised, polarisation optics were not used in the excitation or collection optical paths, avoiding possible complications due to directional effects along the axis of the thin cylindrical specimen (an extruded rod) used to collect spectra. Samples were held vertically in a standard cuvette holder accessory, placed in the electro-mechanical 3-axis stage and user-positioned using an on-board alignment camera. Care was taken to avoid laser-induced thermal damage of the samples. Samples were assumed to be photo-chemically stable and therefore any change in a spectrum with respect to time was attributed to thermal modification of the sample. Laser power was decreased until the form of the spectrum did not change with a change in integration/scan time. Furthermore, samples were examined (after the Raman spectrum had been obtained) under an optical microscope to examine for visible evidence of laser-induced changes. Figure 18.5 shows the Raman spectra for PET, MWCNTs and composites of PET loaded with 0.1, 0.5, 1, 3, 7 and 10 % MWCNTs by weight.
The evolution of the characteristic carbon nanotube D, G and D* bands around 1300, 1600 and 2600 cm−1 are clearly evident (Han et al., 1989; Du et al., 2004). It is also apparent that the spectra of the composites is a summation of the PET and MWCNT spectra, as may be expected. The optical limiting properties of MWCNTs and polymer–MWCNT composites has been previously documented (Fakirov et al., 1975; Tongyin et al., 1983). The Raman spectra from the composites will be relatively under-represented in the polymer spectrum due to the nanotube absorption of both the excitation and scattered radiation, hence the weak contribution from the PET in the composite samples having in excess of 3 wt% MWCNT. Two dominant peaks exist in the PET spectrum, one at 1613 cm−1, which is reported as varying little with change in crystallinity (Kiflie et al., 2002), and the other at around 1730 cm−1. From Fig. 18.5, it is clear that the peak at 1613 cm−1 is in the same locality as the G band of the MWCNTs and so, although the two peaks may be independent, they overlap. The proximity of these peaks is not surprising since the 1613 cm−1 band in PET is attributed to C-C ring stretch, and the nanotube structure is composed of graphene-like rings. The nanotube peaks neither interfere nor overlap with the peak near 1730 cm−1 and this peak is definable in all spectra, although it is very weak, and therefore noisy, in the 7 and 10 wt% MWCNT traces. The peak at the nominal relative wavenumber of 1730 cm−1 is associated with a carbonyl stretching mode within the PET macromolecule (Kiflie et al., 2002). In this study, the position of this peak was found to be at 1727 cm−1 for the amorphous PET sample with no MWCNTs added. Previously Yeh et al. had correlated the full width of this peak at half height (FWHH) with the degree of crystallinity in a stressed PET fiber sample (Yeh et al., 1998). This width decreased by almost 50% between amorphous and semi-crystalline fibers indicating the narrower the band width, the greater the crystalline content. Considering the spectra of Fig. 18.6, which have been normalised for peak height, it is evident that the shape of this peak changes to a more triangular form with increasing MWCNT content.
This type of change of peak shape results in a decrease in the FWHH. In order to measure the FWHH, the standard method in the post-processor used (Thermogalactic-Grams™) is to fit a Gaussian, Lorentzian or mixed Gaussian/Lorentzian and then take the FWHH of the mathematical curve. It was apparent that for many of the spectral lines under consideration a single fitted curve (even a mixed Gaussian/Lorentzian) did not give an adequate fit. The inadequacy of single mathematic curves to fit to PET peaks has been reported previously (Kiflie et al., 2002; Antoniadis et al., 2009). Although it would be possible to fit multiple mathematical curves to the spectral peaks, this would result in a complication when comparing peaks between spectra as a single FWHH figure would not exist. It was decided that a more accurate way to proceed was to measure the FWHH without presuming anything about the nature of the peak. Therefore, a FORTRAN program was written to calculate the peak FWHH numerically from the data points acquired in the spectrum. The results from the numerical analyses are given in Table 18.2 and plotted graphically in Figs 18.7 (a) and 18.7 (b). Both the peak FWHH and the weighted peak position change with nanotube content. Initially, the FWHH decreased with increasing MWCNT content to 3 wt% MWCNTs and thereafter appears to rise. The narrowest FWHH peak measured showed a 25.7% reduction over the amorphous which is significant, and is likely to be caused by the formation of a more crystalline phase in PET. The position of the weighted peak centre also upshifts with increasing MWCNT content to 3 wt% MWCNT, and thereafter appears to level off. It should be noted that the accuracy of the results for the 7 and 10 wt% MWCNT may be compromised due to the very weak nature of the signal of these peaks.
|Wt% CNT||FWHH (cm−1) (calculated)||Weighted peak position (cm−1)|
The XRD traces for neat PET, pristine MWCNTs and PET–MWCNT composites are shown in Fig. 18.8 (PANalytical X’Pert PRO diffractometer with Cu-Kα radiation (λ = 1.5406 Å) at a scanning rate of 0.2 °/min over the range 1–50 °(2θ). As expected, the diffractogram for neat PET showed a broad amorphous peak, however, as the proportion of MWCNTs added was increased the evolution of four clearly defined peaks for the PET–MWCNT composites, particularly at higher MWCNT loading, was obvious. These peaks were at 2θ = 16°, 17.5°, 21.5° and 22.7°, confirmation that addition of MWCNTs to PET initiated polymer crystallisation.
The diffractogram for neat MWCNTs showed two reflections at 2θ = 26° and 43° corresponding to the 002 and 100 crystal planes, respectively. The main 002 reflection is associated with the concentric arrangement of the CNTs and corresponds to a d-spacing of 3.41 Å. This peak was absent from the composite materials, suggesting the tearing and shearing forces applied during melt extrusion were effective at disentangling the MWCNTs, although there may be some overlap between 100 reflection for PET and the 002 MWCNT reflection. The PET crystal faces have been assigned to the following diffraction peaks at 2θ angles: = 16.3°; (010) = 17.5°; = 21.5°; = 22.7°; (100) = 26.1°; = 27.65° (Fakirov et al., 1975). They are labelled in Fig.18.8 as: 1 = ; 2 = (010); 3 = ; 4 = ; 5 = (100); 6 = . The diffraction peaks obtained for the PET–MWCNT composites appear at the same 2θ angles assigned to crystalline PET crystal faces, but appear narrower and more distinct with increasing MWCNT loading. This may be due in part to the gradual perfection of the crystallite although this rapid crystallization did not allow the polymer to reach a highly regular semicrystalline framework (Tongyin et al., 1983). By way of example, the d-spacings for the composites of PET with 1 wt% and 10 wt% MWCNT are listed in Table 18.3. For each peak, the d-spacing decreased with increasing MWCNT loading, again supporting the hypothesis that MWCNTs act as nucleating agents for PET.
The physical and mechanical properties of PET post processing are governed by the associated crystallisation processes (Kiflie et al., 2002). For this reason, the crystallisation and melt behaviour of semi-crystalline polymers such as PET are of extreme interest to researchers and industrialists alike. The introduction of MWCNTs to PET can both enhance the original properties of the PET while simultaneously altering the intended morphology of the polymer significantly. Previous studies have shown the addition of only small amounts of MWCNTs, typically 0.01 wt% can induce heterogeneous nucleation of crystallite growth and increase the temperature at which crystallisation begins (Wang et al., 2007). The nucleating effect of MWCNTs is largely dependent on the initial state of dispersion of the nanotubes and the interaction between CNTs and polymer chains. In this study, the iso-thermal crystallisation behaviour of PET–MWCNT composites was investigated. Samples ranging in mass from ~ 5–10 mg were sealed in an aluminium pan, heated, then cooled and reheated at a constant rate of 10 K/min. The glass transition (Tg) regions of the DSC curves obtained are shown in Fig. 18.9. On addition of MWCNTs to PET a slight reduction in Tg from about 75 °C for virgin PET to 72 °C for the composite with 1 wt% MWCNT loading. However, this minor reduction in Tg was concomitant with a significant decrease in specific heat capacity (ΔCp) such that no Tg was detected for the composites with higher CNT content. At higher MWCNT loading, PET chain dynamics is hindered and the degree of hydrogen bonding is reduced, in agreement with our FTIR observations.
The thermograms obtained for the reheat and cooling cycles are shown in Figs 18.10 (a) and 18.10 (b), and the corresponding calorimetric data; temperature of melting (Tm), crystallisation temperature (Tc), enthalpy of fusion (ΔHm), enthalpy of crystallisation (ΔHc), percentage crystallinity (Xc) and degree of supercooling (ΔT) in Table 18.4. Xc was calculated using, Xc = ΔHm/ΔHt, where ΔHm is the enthalpy of melting for the sample of interest and ΔHt is the enthalpy of melting for a theoretically 100% crystalline PET, taken as 117.6 J/g (Quintanilla et al., 1996).
There was little change, possibly 2 K, in Tm of PET on addition of up to 10 wt% MWCNTs, although this may be within instrument error. Similarly, the PET crystalline content changed little when the MWCNTs were added. However, the shape of the PET melting peak obtained on addition of MWCNTs implies a change in crystallite size and order. This observation was probed further by examining the crystallisation exotherms, see Fig.18.10 (b). Tc of neat PET was 183 °C which increased to 209 °C, 211 °C and 218 °C on the addition of 0.1 wt%, 1 wt% and 10 wt% MWCNTs, respectively. Consequently, the degree of supercooling (ΔT) decreased from 61 °C for PET alone to 28 °C for the composite with 10 wt% MWCNTs added. The 35 °C increase in Tc and 33 °C decrease in ΔT is further confirmation that the MWCNTs act as nucleating agents in PET, accelerating the rate of crystallisation and initiate crystal growth at the CNT surface. The crystallisation curve narrowed on MWCNT addition, the full width at half height (FWHH) decreased from 16 °C to 7 °C on the addition of only 0.1 wt% MWCNTs. This behaviour is typical of the growth of uniform, discrete crystalline regions and a reduction in crystal size distribution, as a consequence of the good dispersion of MWCNTs throughout the PET matrix, particularly at MWCNT loadings below 1 wt%. As CNT loading increased, the curves became gradually broader again as the variation in size and number of PET crystal growth regions increased, due to the increase in the number and size of CNT agglomerates. This effect on crystallisation behaviour was clearly evident from polarised optical microscopy experiments (see Figs 18.2(a) and 18.2 (b)), where the addition of MWCNTs to PET induced the growth of smaller, more imperfect crystals by accelerating crystal growth due to the presence and abundance of random points for crystal nucleation, i.e. on the CNT surface. The average crystal size decreased due to the nucleating effect of the CNTs, resulting in an increase in Tc and a change in shape of the crystallisation curves. Similar crystallisation behaviour was observed in separate studies by Wang (Wang et al., 2007) and Tzavalas (Tzavalas et al., 2008) where 0.01 wt% and 1 wt% MWCNTs induced a decrease in Tc by 6 °C and 16 °C, respectively. Wang et al. also observed a double melting peak, where the lower temperature peak became more intense as CNT loading increased. This is due to the presence of less perfect crystallites forced to initiate by the presence of high CNT loading (Wang et al., 2007). Multiple melting endotherms are common in polyesters, such as PET, due to the formation of crystals with different imperfections and a range of melt histories (Tzavalas et al., 2008).
Numerous mathematical models can be used to describe the iso-thermal and non-isothermal crystallisation kinetics of polymeric materials. A method commonly used to describe the complex processes involved in non-isothermal crystallisation kinetics was suggested by Ozawa (Ozawa, 1971) and was based on the same theory as predicted by Avrami (Avrami, 1939, 1941) and Evans (Evans, 1945). The Ozawa approach, described later in the text, has been applied in previous studies for the description of the non-isothermal crystallisation of PET (Ozawa, 1971; Sajkiewicz et al., 2001) and indeed many other semi-crystalline polymers, including Nylon 12 (McFerran et al., 2008), poly(propylene) (Mubarak et al., 2001), and poly(p-phenylene sulphide) (López and Wilkes, 1989).
Non-isothermal crystallisation behaviour of unfilled PET and the PET 10 wt% MWCNT composite was investigated using a range of cooling rates. The samples were heated from 40 °C to 280 °C at a rate of 10 K/min, held at 280 °C for 10 minutes, to ensure complete crystallite melting, then cooled to room temperature at cooling rates of 5 K, 10 K, 20 K, 30 K and 40 K/min and heat flow as a function of time and temperature recorded. The relative degree of crystallinity (Xt) can be determined by integrating the area under the peak during the non-isothermal scan as a function of time and temperature:
where A(T) is the partial area of the peak at temperature (T) and A(Total) is the total peak area. The Ozawa model was applied to describe the crystallisation kinetics of unfilled PET and the PET–10 wt% MWCNT composite under nonisothermal conditions. The activation energies were also obtained using the Kissinger method (Boswell, 1980). The non-isothermal crystallisation curves for neat PET and the PET–10 wt% MWCNT composite for the various cooling rates are shown in Figs 18.11 (a) and 18.11 (b) respectively and the values of Tc obtained as a function of cooling rate listed in Table 18.5. It is clear from Figs 18.11 (a) and 18.11 (b) that as the cooling rate increased, the crystallisation onset temperature (Tcon), peak crystallisation temperature (Tc) and crystallisation offset temperature (Tcoff) shifted to lower temperatures.
Moreover, as cooling rate decreased, the crystallisation peaks became more intense. This occurred as a slower cooling rate resulted in the formation of more perfect crystallites and as such crystal packing and density was improved. From Table 18.5, it can be seen that Tc increased from 182 °C to 209 °C and from 212 °C to 229 °C with decreasing cooling rate for neat PET and the PET–10 wt% MWCNT composite, respectively, again confirming the ability of MWCNTs to nucleate PET crystallisation.
For non-isothermal experiments, the recorded temperature of the sample must be corrected to account for the thermal lag between a point in the sample and the calorimeter furnace at any given time. The recorded temperatures were corrected using Equation 18.2 (Mubarak et al., 2001):
where, Tdisp is the display temperature and (K/min) is the cooling rate. Then, the degree of relative crystallinity can be plotted against temperature or time by converting temperature into time using Equation 18.3:
where, t is the crystallisation time, T is the crystallisation temperature, To is initial temperature when crystallisation begins (t = 0) and () is the cooling rate. Figures 18.12 (a) and 8.12 (b) show plots of Xt against temperature for neat PET and the PET–10 wt% MWCNT composite, respectively.
A series of S-shaped curves are obtained which show that as cooling rate increased crystallisation occurred more rapidly. Both neat PET and the PET–MWCNT composites crystallise at lower temperatures with increasing cooling rate. The temperature range in which crystallisation occurred increased with increased cooling rate, e.g. for neat PET at cooling rate of 5 K/min the crystallisation range is 36 K whereas at 40 K/min the range is 60 K. Furthermore, the gradients of the curves are steeper for neat PET compared to those obtained for the PET–10 wt% MWCNT composite as addition of MWCNTs to PET has extended the crystallisation process but yields the growth of more perfect crystals.
A series of sigmoidal shaped curves were also obtained for a plot of Xt versus time, see Figs 18.13 (a) and 18.13 (b). Again, it was evident that as cooling rate decreases the time required for complete crystallisation was longer, therefore the crystallites have more time to form. Addition of MWCNTs to PET retards polymer crystallisation and extends the time for PET crystallisation to complete. For example, for a cooling rate of 5 K/min it took neat PET 7.2 mins to crystallise fully, whereas it took the PET–10 wt% MWCNT composite 11.2 mins to crystallise. In the primary crystallisation region, i.e. below Xt = 0.8, the rate of crystallisation is high, as shown by the steepness of the gradient of the curve in this region. Above Xt = 0.8, the rate of crystallisation slows and the curve plateaus, perhaps indicative of the onset of secondary crystallisation. For neat PET, the rate of crystallisation slows with decreasing cooling rate which can be attributed to spherulite impingement. The rate of crystallisation also slows on addition of MWCNTs to PET, particularly at slower cooling rates.
Several methods have been employed to describe the non-isothermal crystallisation kinetics of semi-crystalline polymers (Mubarak et al., 2001). Ozawa derived a non-isothermal kinetic model to describe the process of nucleation and crystallite growth by modifying the Avrami model to allow for all non-isothermal processes (Ozawa, 1971). This method is used when crystallisation occurs at a constant cooling rate:
where X(t) is the relative degree of crystallisation at a temperature (T), K*(T), is the cooling function which is related to the crystallisation rate and indicates how fast crystallisation proceeds, (φ) is the cooling rate and (m) is the Ozawa exponent which is dependant on the dimension of crystal growth. By transforming Equation 18.4 into a double-logarithmic form:
a plot of ln[− ln(1–X(t))] versus ln(φ) at a given temperature will produce a straight line and the kinetic parameters (m) and (K*(T)) can be derived from the slope and the intercept, respectively. Such graphs were constructed for neat PET and the PET–10 wt% MWCNT composite from data points taken at different temperatures in the range 440 K to 510 K, see Figs 18.14 (a) and 18.14 (b), respectively. Ozawa reported previously that this treatment was suitable for modeling the crystalisation behaviour of PET, although he ignored secondary crystallisation (Ozawa, 1971).
The data presented in Figs 18.14 (a) and 18.14 (b) could be fitted with straight lines with reasonable correlation coefficients (> 0.9) for all the crystallisation temperatures examined, confirming that Equation 18.5 could be used to describe the non-isothermal crystallisation kinetics of PET and PET–MWCNT composites. The Ozawa exponents (m) were calculated from the slope of each trendline, and the values obtained for m are listed in Table 18.6 and the variation in m with temperature for neat PET and the PET–10 wt% MWCNT composites is shown in Fig. 18.15.
|Sample||Temperature (°K)||Ozawa exponent (m)|
|PET–10 wt% MWCNT|
There was some scatter in m for neat PET, understandable as non-isothermal crystallisation is a dynamic process and a function of time and cooling rate. For any given crystallisation temperature, the PET–10 wt% MWCNT composite had a lower Ozawa exponent compared to PET alone, see Table 18.6. This suggested that the addition of MWCNTs to PET significantly affected the crystal growth mechanisms and again demonstrated crystallite nucleation of PET by MWCNTs. The influence of MWCNTs and varying cooling rates on the activation energy (ΔE) required for crystallisation under non-isothermal conditions has been described by Kissinger using Equation 18.6 (Kissinger 1956):
where Tp is the peak temperature of crystallisation, φ is the cooling rate and R is the gas constant. A plot of ln(φ/Tp2) against 1/Tp yields a straight line with slope = Δ E/R (Liu et al., 1998), see Fig. 18.16.
Activation energies of 144 kJ/mol and 251 kJ/mol were determined for PET and the PET–10 wt% MWCNT composite, respectively. An increase in activation energy on the addition of MWCNTs suggests that more energy is required for PET crystallisation to take place. Presumably, addition of MWCNTs to PET reduces polymer chain conformation and hinders chain dynamics. This corroborates the results shown in Figs 18.11 (a) and 18.11 (b). A combination of higher activation energy, together with a shift of the crystallisation curves to higher temperatures, confirmed that MWCNTs act as nucleating agents for PET.
The thermal stability of PET on addition of MWCNTs was investigated using thermo-gravimetric analysis (Hi-Res TGA 2950, 10 K/min from RT to 500 °C). Figure 18.17 (a) shows the percentage mass loss for neat PET and PET–MWCNT composites with increasing temperature. From the weight loss plots, the onset of thermal degradation (Tond) for neat PET was at 441 °C. Tond decreased by about 10 to 12 K on addition of up to 10 wt% MWCNTs. The residual yield for neat PET was 13.5%, this decreased slightly to 11% for the composites with 0.1 wt% and 1 wt% MWCNTs, before increasing significantly to 25% on addition of 10 wt% MWCNTs to this PET.
This behaviour would suggest that at low MWCNT loadings (< 1 wt%), the CNTs accelerate the thermal degradation of PET. However, for a loading of 10 wt% MWCNTs, the onset of thermal degradation decreased by 13 °C relative to unfilled PET, yet the residual yield for this composite was 25%. This behaviour may be associated with the high thermal conductivity of MWCNTs and more effective dissipation of thermal energy for the 10% mass fraction. As the MWCNT content was only 10 wt% for this composite, the composition of the residue must contain not only MWCNTs which are known to be thermally stable well above 500 °C (714 °C for the MWCNTs used in this study), residual metal catalysts, metal oxides, amorphous carbon, but also un- or partially degraded PET. It has been proposed that CNTs create a ‘barrier effect’ by preventing the release of volatiles and decomposed products from the composite material, resulting in the retardation of the thermal decomposition of the composites (Kim et al., 2007). A critical concentration of MWCNTs is required to enhance the thermal stability of PET. The thermal stability of PET is better demonstrated by examining the derivative weight as a function of temperature, Fig. 18.17 (b). Here, the peak maxima (Tmax) is clearly seen to decrease with increasing MWCNT content, suggesting the onset of thermal degradation of PET is shifted to lower temperatures. Some variability in the thermal stability of melt-processed PET–MWCNT composites has been reported previously. Kim et al. obtained an increase of 4 K in the thermal degradation temperature of PET on addition of 2 wt% MWCNTs (Kim et al., 2007). Ahn et al. also achieved an increase in the degradation temperature of PET by 10 K on addition of 3 wt% MWCNTs (Ahn et al., 2008). The authors proposed that the introduction of van der Waals or hydrogen bonding between carboxyl functionalised MWCNTs and PET retarded the onset of thermal degradation (Ahn et al., 2008). In another study, welldispersed SWCNTs had no appreciable effect on the thermal stability of PET (Anand et al., 2007). The unpredictability of the thermal stability of PET–CNT composites implies that this property is controlled by a combination of factors, including CNT dispersion and distribution, CNT orientation – hence barrier properties, thermal conductivity of CNTs, polymer–CNT interactions and polymer crystallinity.
Oscillatory or dynamic shear rheology is a sensitive method for detecting the formation of possible CNT networks and in qualifying the distribution of CNTs within a polymeric material without the need for laborious microscopy and imaging techniques. By way of example, the rheological properties of virgin PET and composites of PET with 0.5 wt%, 3 wt% and 7 wt% MWCNTs were measured using an Advanced Rheometric Expansion System (ARES, TA) strain controlled rheometer with 25 mm parallel plate geometry. Samples of ~ 1 mm thick were equilibrated at 270 °C for 2 minutes using N2 as the gas source for the oven to prevent any oxidative degradation. A strain sweep at 10 rad/s was first performed to establish the linear viscoelastic region and appropriate strains selected for frequency sweeps. The rheological properties, including tan delta, storage modulus (G'), loss modulus (G") and complex viscosity (η*) of neat PET and PET–MWCNT composites were measured at a constant temperature of 270 °C within the viscoelastic limit of each material. Unfilled PET and the PET–MWCNT 0.5 wt% composite exhibited near-Newtonian behaviour while the composites with 3 wt% and 7 wt% MWCNTs exhibited a clear transition to shear-thinning behaviour. The viscoelastic properties of neat PET were largely unchanged on addition of 0.5 wt% MWCNTs. However, addition of 3 wt% MWCNTs and above to PET resulted in an increase in η* and G', especially at low frequencies where the instrument is most sensitive in the detection of pseudosolid like behaviour, see Figs 18.18 (a) and 18.18 (b). For example, at 0.3 rad/s, the η* increased from 6 Pa.s for (0.5 wt% MWCNT) to 148 Pa.s (3 wt%), then to 856 Pa.s (7 wt%). G' for neat PET increased by 2 orders of magnitude from about 0.31 to 35 Pa with addition of 3 wt% MWCNTs, and by 3 orders of magnitude from 0.31 to 200 Pa on addition of 7 wt% MWCNTs.
18.18 Variation in: (a) complex viscosity (η *) as a function of ω; (b) storage modulus (G’) as a function of ω; (c) Cole–Cole plots (G’ versus G”); and (d) inverse loss tangent (tan δ)−1 as a function of ω, for neat PET and PET–MWCNT composites.
The addition of the high modulus CNTs, as expected, increased the storage modulus of the polymer. A rheological percolation threshold is evident between 0.5 and 3 wt% manifest by a large step increases in η* and G' at low frequencies. Kim et al. observed similar behaviour for melt processed PET–MWCNT composites, although a rheological percolation effect was noticeable on addition of 0.5 wt% MWCNTs (Kim et al., 2007). This behaviour is indicative of the influence of polymer–CNT–polymer interactions and the onset of formation of an interconnected network of CNTs and CNT agglomerates which restrict longrange polymer chain motion.
Further evidence for the formation of a rheological percolated network can be extracted by constucting a Cole–Cole plot, i.e. log G' versus log G", see Fig. 18.18 (c). This is analogous to the Cole–Cole plots used in dielectric spectroscopy and described by Han et al. to probe temperature induced changes in the microstructure of homo-polymers, block copolymers and blends (Han et al., 1989). Any positive deviation from a linear relationship between G' and G" is evidence for the formation of a percolated network. From Fig. 18.18 (c), it is obvious that the curves obtained for the 3 wt% and 7 wt% MWCNT composites have noticeably deviated from that for neat PET. The slopes of the Cole–Cole plots decreased with increasing MWCNT content, again indicative of the transition from liquidlike to solid-like behaviour as a consequence of well-dispersed, homogeneous network of high aspect ratio MWCNTs restricting polymer flow. Further evidence for network formation was apparent from a plot of inverse tan δ (i.e. (tan δ)−1 = G'/G") versus frequency, as an increase in (tan δ)−1 is a measure of the increase in ‘solidity’ of the composite. It can be seen, across the frequency range studied, there is a significant increase in tan δ−1, and clearly a percolated network is formed below 3 wt% MWCNTs. The formation of a rheological percolated network is governed by a combination of factors including the state of CNT dispersion, CNT alignment and polymer–CNT interactions (chemical and physical).
The formation of an electrically conducting percolated network can be studied by investigating the effect of MWCNT loading on the volume resistivity of a polymer. Volume resistivity is defined as the ratio of the potential gradient parallel to the current in the material to the current density (Shah, 1998). Volume resistivity measurements were made using high accuracy meters, a Keithley 6517A electrometer and a shielded test enclosure (Keithley 8009 resistivity test fixture). High resistivity testing was performed in accordance with ASTM-D257. For samples of low resistivity, two-point probe measurements were used using a Keithley 6517A electrometer in combination with two-point probes and the assistance of silver epoxy contact paint in order to minimise contact resistance effects. Volume resistivity measurements are made by using a strip of composite film of known thickness and geometry from point to point and measurements were made on multiple samples. The change in volume resistivity of neat PET and the PET–MWCNT composites is shown in Fig. 18.19.
The volume resistivity of neat PET was reduced by 12 orders of magnitude (i.e. electrical conductivity increased by 1012 S/cm), from 1017 to 105 .cm on the addition of 3 wt% MWCNTs. An electrical percolation threshold (ρc) was estimated at approximately 2 wt% MWCNTs. This value is comparable to that reported by Anand et al. for PET–SWCNT composites, although a more simple lab-scale kneading process was used to prepare these composites. However, these values are larger than those previously reported for Logakis et al. who recorded a ρc = 0.1–0.2 wt% for PET–MWCNT composites prepared using a combination of dry-mixing and twin-screw extrusion (Logakis et al., 2010). In the study by Logakis et al., the authors compared network formation and subsequent electrical conductivity of PET–MWCNT composites prepared by 3 methods, in-situ polymerisation, masterbatch dilution by melt mixing and dry blending prior to melt mixing. Electrical percolation values/ranges of 0.06 wt% (0.05–0.1 wt%) and (0.1–0.2 wt%) MWCNTs were achieved for each of the three preparation methodologies, respectively. In contrast, for composites of PET with SWCNTs, a lower electrical percolation of 0.024 wt% was obtained for composites prepared by direct mixing. Interestingly, in-situ polymerisation of a PET with the same SWCNTs yielded a higher electrical percolation, ρc = 0.048 wt%. The authors suggested that SWCNTs must retain some degree of agglomeration to form an electrically conducting network (Hernández et al., 2009). There should be no expectation that an electrical and rheological percolation can be achieved for the same CNT loading. For example, for composites of PET with MWCNTs prepared by solution mixing, rheological and electrical percolation thresholds of 0.6 wt% and 0.9 wt% were obtained, respectively (Hu et al., 2006). The discrepancy arises as the distance required to bridge polymer chains by CNTs and thus constrain polymer chain dynamics is different to the CNT loading required to form an electrically conducting pathway throughout the PET matrix to allow electrons to flow.
Composites of MWCNTs and PET were readily prepared using conventional twin-screw extrusion. Examination of these composite materials across the length scales using a combination of POM, FESEM and HRTEM revealed the CNTs were well-dispersed and distributed throughout the PET matrix. However, the number of MWCNT agglomerates increased with increasing MWCNT loading. MWCNTs act as nucleating agents for PET, initiating heterogeneous nucleation and alter PET crystallisation kinetics. From a FTIR study of PET chain conformations, the proportion of polymer chains residing in the trans crystalline phase increased relative to those residing in the gauche phase, indicating that addition of MWCNTs induced a more crystalline PET structure. Further evidence for increased PET crystalline content on MWCNT addition was also confirmed by Raman spectroscopy. The isothermal and non-isothermal crystallisation kinetics of these composites was investigated using DSC. Both the rate of crystallisation and temperature of crystallisation (Tc) of PET increased significantly on addition of MWCNTs. A critical concentration of MWCNTs (> 1 wt%) was required to enhance the thermal stability of PET. MWCNTs provide a barrier to the release of volatiles and decomposed products, retarding the thermal degradation process. An electrical and rheological percolation of the MWCNTs used in this study was achieved at approximately < 2 wt% and < 3 wt%, respectively.
It is conclusive that MWCNTs alter the crystalline structure and electrical, thermal and viscoelastic properties of PET. Exploitation of such composite materials will only be fully realised if they can be prepared using conventional polymer processing methods, such as extrusion and injection moulding. However, there is little published literature which attempts to correlate the relationships between melt processing parameters (e.g. screw speed, screw profile, residence time, temperature) and polymer–CNT composite morphology and properties. Although there has been some focus on this topic recently, see Chapter 4 and (Villmow et al., 2008), more systematic and detailed studies are required in the future. Furthermore, for engineering polymers, such as PET, these materials in most instances undergo secondary processing which typically includes uniaxial and/or biaxial deformation, e.g. stretch blow moulding of a PET bottle. Therefore, the final properties of the composite are only part governed by the polymer–CNT morphology achieved post melt mixing. Again, future work must focus on the effect of quasi-solid and solid state deformation on polymer–CNT composite micro- and nano-structure.
The authors acknowledge the support of The Royal Society (574006/G503/24135) and The Nuffield Foundation (NAL/00696/G). We thank Dr Andrew Dennis (Avalon Instruments) for helpful discussions and BM thanks DEL and QUB for funding a studentship. The authors also acknowledge the formal exchange agreement between UQ and QUB that facilitates ongoing collaborations.
Antoniadis, G., Paraskevopoulos, K.M., Bikiaris, D., Chrissafis, K. Melt crystallisation mechanism of poly(ethylene terephthalate)/multi-walled carbon nanotubes prepared by in situ polymerisation. Journal of Polymer Science:Part B: Polymer Physics. 2009; 47(15):1452–1466.
Cole, K.C., Ajji, A., Pellerin, E. New insights into the development of ordered structure in poly(ethylene terephthalate). 1. Results from external reflection infrared spectroscopy. Macromolecules. 2002; 35(3):770–784.
Cole, K.C., Ajji, A., Pellerin, E. New insights into the development of ordered structure in poly(ethylene terephthalate), 2. Results from transmission infrared spectroscopy of thin films. Macromolecular Symposia. 2002; 184:1–18.
Hernández, J.J., García-Gutiérrez, M.C., Nogales, A., Rueda, D.R., Kwiatkowska, M., Szymczyk, A., Roslaniec, Z., Concheso, A., Guinea, I., Ezquerra, T.A. Influence of preparation procedure on the conductivity and transparency of SWCNT-polymer nanocomposites. Composites Science and Technology. 2009; 69(11–12):1867–1872.
Hu, G., Zhao, C., Zhang, S., Yang, M., Wang, Z. Low percolation thresholds of electrical conductivity and rheology in poly(ethylene terephthalate) through the networks of multi-walled carbon nanotubes. Polymer. 2006; 47:480–488.
Jin, S.H., Yoon, K.H., Park, Y.-B., Bang, D.S. Properties of surface-modified multiwalled carbon nanotube filled poly(ethylene terephthalate) composite films. Journal of Applied Polymer Science. 2008; 107(2):1163–1168.
Kobayashi, H., Shioya, M., Tanaka, T., Irisawa, T., Sakurai, S., Yamamoto, K. A comparative study of fracture behaviour between carbon black/poly(ethylene terephthalate) and multiwalled carbon nanotube/poly(ethylene terephthalate) composite films. Journal of Applied Polymer Science. 2007; 106(1):152–160.
Lee, H.-J., Oh, S.-J., Choi, J.-Y., Kim, J.W., Han, J., Tan, L.-S., Baek, J.-B. In situ synthesis of poly(ethylene terephthalate) (PET) in ethylene glycol containing terephthalic acid and functionalised multiwalled carbon nanotubes (MWNTs) as an approach to MWNT/PET nanocomposites. Chem. Mater. 2005; 17(20):5057–5064.
Logakis, E., Pissis, P., Pospiech, D., Korwitz, A., Krause, B., Reuter, U., Pötschke, P. Low electrical percolation threshold in poly(ethylene terephthalate)/multi-walled carbon nanotube nanocomposites. European Polymer Journal. 2010; 46(5):928–936.
Luo, P., Jiang, K., Wang, K., Yang, J., Xu, T., Chen, L., Fu, Q. Plasma treatment-induced fluorine-functionalised multi-walled carbon nanotubes to modify poly(ethylene terephthalate) obtained via in situ polymerisation. Polymer International. 2009; 59(2):198–203.
Quintanilla, L., Alonso, M., Rodríguez-Cabello, J.C., Pastor, J.M. Structural analysis of poly(ethylene terephthalate) reinforced with glass fiber: thermal behaviour and correlation between PA-FTIR and DSC measurements. Journal of Applied Polymer Science. 1996; 59(5):769–774.
Tzavalas, S., Drakonakis, V., Mouzakis, D.E., Fischer, D., Gregoriou, V.G. Effect of carboxy-functionalised multiwall nanotubes (MWNT-COOH) on the crystallisation and chain conformations of poly(ethylene terephthalate) PET in PET-MWNT nanocomposites. Macromolecules. 2006; 39:9150–9156.
Tzavalas, S., Mouzakis, D.E., Drakonakis, V., Gregoriou, V.G. Poly(ethylene terephthalate)-multiwall nanotube nanocomposites: effect of nanotubes on the conformations, crystallinity and crystallisation behaviour of PET. Journal Polymer Science: Pt. B:Polymer Physics. 2008; 46(7):668–676.
Villmow, T., Pegel, S., Pötschke, P., Wagenknecht, U. Influence of injection molding parameters on the electrical resistivity of polycarbonate filled with multi-walled carbon nanotubes. Composites Science & Technology. 2008; 68(3–4):777–789.
Wang, Y., Deng, J., Wang, K., Zhang, Q., Fu, Q. Morphology, crystallisation and mechanical properties of poly(ethylene terephthalate)/multiwall carbon nanotube nanocomposites via in situ polymerisation with very low content of multiwall carbon nanotubes. Journal of Applied Polymer Science. 2007; 104:3695–3701.
Yoo, H.J., Jung, Y.C., Cho, J.W. Effect of interaction between poly(ethylene terephthalate) and carbon nanotubes on the morphology and properties of their nanocomposites. Journal of Polymer Science: Part B: Polymer Physics. 2008; 46:900–910.