Chapter 6: Injection moulding of polymer–carbon nanotube composites – Polymer-Carbon Nanotube Composites

6

Injection moulding of polymer–carbon nanotube composites

C.Y. Lew, C. Dewaghe and M. Claes,     Nanocyl S.A., Belgium

Abstract:

The electrical resistivity of a polymer–carbon nanotube (CNT) composite is sensitive to injection moulding parameters and, therefore, imparts flexible property design to meet different application requirements. In this chapter, the authors attempt to identify via a 2i factorial design of experiment the interactions between injection parameters, volume resistivity, surface resistivity and surface finish, using a low viscosity polycarbonate (melt flow index, MFI, 38 g/10 min) and then verify the interactions in more detail using a higher viscosity polycarbonate (MFI 20 g/10 min). Results indicated that the volume resistivity was most sensitive to injection melt temperature, while injection speed has the greatest influence on surface conductivity. Surface resistivity was found to increase with injection speed while volume resistivity decreases with injection speed. The effect of polymer molecular weight (Mw) on electrical resistivity was also investigated. Volume resistivity decreases with increasing Mw, but the reverse trend was observed for surface resistivity.

Key words

carbon nanotube

electrical resistivity

injection moulding

polycarbonate

rheology

6.1 Introduction

Collectively, the global market for electro-conductive polymer compounds is forecast to reach US$2.78 billion by 2014 and the filler-based composites segment currently accounts for 84% of the segment, according to a recent analysis by Market and Markets Consulting.1 This is driven by increasing application in electrostatic discharge (ESD), electro-conductive (EC), electromagnetic interference (EMI) and radio frequency interference (RFI) applications. By economy scale, both in dollar and volume, the fabrication of conductive compounds in manufactured products is dominated by injection moulding mobilised by fast growing demand for electronics goods and automotives components. In addition, given the ease of processing and improved heat dissipation properties, polymer–carbon nanotube (CNT) composites become attractive in the manufacture of automotive under-the-hood metals replacement in close proximity to a heat source such as an engine or in parts that would generate a high level of heat (e.g. the power transmission system and the fuel rails) or in power computing.

In the past few years, multi-walled CNTs have progressively captured wider industrial acceptance, as well as being a cleaner composite solution (i.e. lower out-gassing material) to carbon black, delivering new design opportunities inconceivable for other traditional conductive fillers, at least as measured by products’ performance and cost efficiency. Out-gassing materials, when subjected to heat under low pressure, emit volatiles that would contaminate the application, therefore requiring a very high vacuum environment. Today, CNT composites have entered the mass production shop floor as, for example, polyoxymethylene–CNT and polyamide 12–CNT in fuel lines application for automotives, as polycarbonate–CNT in integrated-circuit tray and hard-disk drive spindle and high density polyethylene–CNT in extrusion-blown moulded drums.

Because of its relatively low loading level, typically at a fraction of conventional conductive fillers (e.g. 2 wt% CNT versus 10 wt% conductive carbon black), the polymers can retain a higher degree of mechanical properties and an aesthetically more appealing surface finish. For an application requiring very high conductivity, the polymer–CNT composite could transform the design economy of connectors, switches, sensors, actuators, convertors, control modules, supercapacitors, batteries and transistors.

Electrically conductive plastics have been commercially used for many years already and the application of CNT has grown at an exponential rate, from US $290 million in 2006 to an estimated US$5 billion by 2012 (Barron and Khan, 2008). Despite the glitch of the current economic climate, CNT-based polymer products are well positioned to enter mainstream commercial production in this decade.

6.2 Background

Injection moulding is characterised by flow at high shear rate and the shear field is generally non-uniform. Predicting the physical properties of an injection-moulded polymer can be a formidable task because even a minor alteration of an injection parameter or mould design can influence the polymer crystallisation behaviour, chain orientation, skin layer formation and residual stress. Furthermore, the effect of an injection parameter can differ from one polymer to another and the variation is non-linear. For example, increasing the injection speed in general enhances the orientation of polymeric-chain lengthwise which, in turn, increases the tensile strength of polypropylene (PP) (Aarøe et al., 2009). However, the reverse trend was found for acrylonitrile butadiene styrene (ABS), while the tensile strength of PC was less sensitive to injection speed (Aarøe et al., 2009; Mekhilef et al., 1991). In a polymer system filled with a conductive particle such as CNT, a higher injection speed can lead to more non-uniform shear across the flow direction and, as a result, the electrical conductivity may vary across the moulding dimension.

For a pressure-driven incompressible viscous flow through a channel, shear stress and shear rate are greatest at the boundary and lowest near the centreline. This difference in shear field gives rise to shear gradient across the flow stream. The gradient in shear rate causes a gradient in the elastic energy for a viscoleastic polymer melt and on the basis of shear-field fractionation, the larger and more elastic chain (i.e. higher molecular weight, Mw) will diffuse away from the wall and towards the centreline to reduce their potential energy (Shelby and Caflish, 2004). Similarly, in a particle-concentrated flow, the larger and more elastic particles will migrate away from the wall. This phenomenon is termed shear-induced migration theory and is widely studied in the domain of physics of fluids (Leighton and Acrivos, 1987; Phillips et al., 1992).

Not only can the change of injection moulding parameter affect the shear flow and in turn particle distribution of CNT in a polymer, the presence of CNT can also affect the shear flow and solidification behaviour of the polymer, both of which cause change in the final properties. For example, the increase in mould temperature will reduce the shear gradient and therefore more uniform properties across the moulding. Similarly, if CNT can act as a viscous thinning agent (Kharchenko et al., 2004; Ma et al., 2008; Lew et al., 2009b), the shear gradient may also become narrower. Or if the presence of CNT causes the crystal size and size distribution to decrease (Zhang and Zhang, 2007), mechanical toughness could be improved.

To date, there are two general agreements on the effect of injection moulding parameters on the pattern of electrical conductivity of a polymer–CNT composite. First is that by increasing the injection melt temperature, the electrical conductivity increases (Lellinger et al., 2008). For example, the electrical surface resistivity of an injection-moulded PP–CNT decreases from 5 × 108 ohm.sq−1 at 165 °C to 1.8 × 106 ohm.sq−1 at 180 °C and 6 × 103 ohm.sq−1 at 205 °C (Lew et al., 2009a). This is attributed to the fact that higher injection temperature (a) reduces the shear (and hence orientation of CNT), thus enhancing the build-up of conductive network; (b) decreases the orienting skin effect; and (c) possibly reduces the shear gradient, therefore reducing shear-induced migration of CNT away from the wall. Second, the electrical resistivity generally decreases at lower injection speed due to lesser degree of CNT orientation in the core and skin layer (Lellinger et al., 2008; Villmow et al., 2008). A short overview of recent data on the relationship between injection moulding parameters and electrical resistivity is given below.

Park et al. (2009) performed an injection moulding study of PC–CNT (2 wt%) using different injection pressure, hold pressure, injection speed and mould temperature and analysed the surface resistivity across the specimen length. Independent of injection parameter, surface resistivity was always higher at near-end region and lowest at near-gate region. They attributed the non-homogeneous distribution of surface resistivity to difference in flowability between CNT and melt, affected by injection parameters. When the relative velocity of the melt was greater than the CNT (higher injection speed), surface resistivity became higher and less uniform. This was corroborated via SEM that showed, for the specimen produced with a combination of a low mould temperature and a high injection speed (60 °C/22 mm · s−1), CNT distribution was non-homogeneous. In the near-gate region, the CNT concentration was higher as compared to the end region. When an experiment was performed using a high mould temperature and a low injection speed (120 °C/13 mm s−1), distribution of CNT was more or less uniform across the length. To test if difference in flowability was the main reason for non-homogeneity, measurement was conducted across the lateral thickness of the specimen and found the surface resistivity decreases when moving from the surface towards the core region. Difference in hold pressure, however, has no effect on the surface resistivity.

Chandra et al. (2007) studied the effect of injection moulding parameters on volume resistivity of PC–CNT (1.8 wt%). Volume resistivity was measured across the width and length of specimen by cutting the gauge section of the specimen. At a higher injection temperature, because of reduced melt viscosity and higher diffusivity which promoted polymer penetration into the CNT bundles, volume resistivity was lower. Non-homogeneous distribution of volume resistivity was observed across both the length and width of the specimen, independent of injection parameters. At a low injection temperature (255 °C), (a) across the width, volume resistivity was higher at the near-mid section and lower at the near-edge section when injection speed was high (9 cm3 · s−1), but the distribution was narrower when the injection speed was low (5 cm3 · s−1); (b) while across the length, volume resistivity was always higher at the near-gate section and lower at the near-end section, but the difference became greater at a high injection speed. At a high injection temperature (285 °C), the effect of injection speed on volume resistivity across the width was reversed, but the effect of injection speed on volume resistivity across the length was the same.

Hong et al. (2004) investigated the effect of shear flow on injection moulding of PP carbon black and found that the conductivity of the composite changed drastically due to non-homogeneous pressure/shear flow caused by the concentration gradient across the flow direction. Because the shear rate and the shear stress were maximal near the wall and minimum at the centre of flow, a shear gradient would develop and cause the filler to fractionalise away from the wall toward the centre. Consequently, the region near the surface became filler-depleted while the region near the centre became filler-enriched. A loss of surface and volume conductivity, however, does not automatically point to particle migration, given other factors such as annealing, crystallisation (quasi-static or shear-induced), and orientation of the skin and shear layer can have significant effect. To verify the shear-induced migration kinetics could be extended to the filler with a larger dimension, capillary flow experiments were performed with a PP compound containing 30 wt% of glass bead at shear rates of 7.2 and 4865 s−1. Because the shear gradient increases proportional with the shear rate, the strand produced at the lower shear exhibited a rough spherical texture (less migration) while the strand produced at a higher shear exhibited a smoother surface (more migration). Additionally, surface conductivity measurements were performed across the lateral thickness of compression- and injection-moulded carbon black polypropylene and polystyrene specimens. For the compression-moulded specimen, surface conductivity was more or less constant across the lateral depth. For the injection-moulded specimen, conductivity increases with lateral depth. This would indicate that the spatial conductivity gradient increases with the injection velocity. However, the volume conductivity measured across the total length was roughly the same for both the compression-moulded and the injection-moulded specimens. In this case, a compensation effect occurs where the depletion of filler near to the surface enriches the bulk.

Villmow et al. (2008) recently carried out one of the most thorough studies on the influence of injection moulding parameters on the electrical resistivity of PC–CNT composites using a design of experiment method. Correlation between electrical resistivity and morphology of the composites was examined via TEM analysis of the skin layer. Villmow found the uniformity of through-plane volume resistivity of the composites was sensitive to the concentration of CNT. The uniformity was poorer at the lower CNT concentration (up to five orders of magnitude at 2 wt% and two orders at 5 wt%). TEM analysis of the skin layer revealed a skin layer with highly oriented CNT in the case of a higher injection speed and a low melt temperature (measuring higher resistivity), but a network-like structure at a low injection speed and a high melt temperature (measuring lower resistivity).

The complexity of injection moulding behaviour on the CNT distribution and orientation in a polymer is so great that many features of its electrical resistivity response in the processing parameters regime have not yet been fully captured. Hence there is an urgent need for more comprehensive experimental data to further develop an understanding in a predictive relationship for polymer processors and injection moulders. The in-depth study by Villmow et al. (2008) has contributed significantly to the understanding of injection moulding parameters affecting the through-plane volume resistivity of PC–CNT and could be extended to other polymer CNT composites. The work carried out in this chapter was intended to understand more thoroughly the relationship between surface resistivity, in-plane volume resistivity and injection parameters.

6.3 Experiment design and materials

6.3.1 Materials and methods

Materials

The multi-walled carbon nanotube used in this study was a 90% purity commercial grade CNT with the trade name NC7000, manufactured by Nanocyl S.A. Belgium. It has a mean diameter and length of 9.5 nm and 1.5 μm respectively, according to the manufacturer’s technical datasheet. Five different grades of polycarbonates produced by Bayer Materials Science A.G., with different melt flow indices (MFI) and different molecular weights (Mw) were used in the study (see Table 6.1). Mw values were measured via gel permeation chromatography (GPC).

Table 6.1

List of polycarbonates used in the study

Makrolon 2205 (MFI 38) was used in the 2 × factorial statistical process optimisation study. The subsequent individual parametric study was performed using Makrolon 2405 (MFI 20). For the study of the effect of Mw on the electrical resistivity of injection-moulded PC–CNT, three additional polycarbonates were compared, Makrolon 0D2015 (MFI 63), Makrolon 2805 (MFI 10) and Makrolon 3105 (MFI 6.5).

Processing

The PC–CNT compounds used in the study were manufactured via twin-screw extrusion processing using an industrial Leistritz ZSK-27 MAXX 48 L/D ratio co-rotating extruder, adapted with an automated CNT replenishment system. Extrusion was carried out at 300 rpm screw speed using proprietary screw configuration optimised for PC–CNT performance. The extruder consisted of 11 heated barrel zones where the temperature profile was ramped from: zone 1: 250 °C, zone 2: 270 °C, zones 3–10: 280 °C, zone 11: 290 °C and the melt temperature in the die was controlled at 300 ± 1.5 °C. CNT was gravimetrically fed via zone 4 by means of a twin-screw side feeder system. Vent ports were fitted at appropriate zones for effective moisture venting.

Injection moulding was performed using an Engel Victor 80 tons injection moulding machine. Specimens were ASTM D256 Izod bar (for resistivity, Izod impact and flexural testing) and ASTM D638 tensile bar (for tensile testing). The Makrolon 2205 PC used in the 2 × factorial study contained 3.0 wt% CNT while the Makrolon 2405 PC used in the parametric study contained 2.5 wt% CNT. All compounds were dried at 120 °C in an air circulating oven for 4 hours prior to injection moulding.

Characterisation

Resistivity measurement was carried out using a Keithley 2405 DC multimeter. Referring to Fig. 6.1, in-plane volume resistance was measured by means of electrodes contact at opposite edges of an Izod test bar. The edges were cryogenically fractured to preserve the original composite morphology, and painted lightly with silver paint to enhance electrode contact. Surface resistance measurement was performed by means of electrodes contact with two thin silver lines painted on an Izod bar. Volume resistivity, ρv and surface resistivity, ρs were calculated according to Equations 6.1 and 6.2. All resistivity results were an average of eight specimens’ measurements.

6.1 Schema of injection-moulded Izod test bar used for electrical resistivity measurement.

[6.1]

[6.2]

The state of agglomerate dispersion was analysed using an Axio-Zeiss Imager-M1 light optical microscopy, equipped with an AxioCam MRc5 CCD camera in transmission mode. Specimens of about 100 micron thick were obtained from microtome. The specimens were pre-melted between two thin glass slides at 310 °C using a Mettler-Toledo FP2HT hot-stage, then gently pressed and allowed to solidify. The surface finish (defects) of injection-moulded Izod test bars were analysed by means of the following procedure. Bright light of a suitable intensity was projected at the test specimen and a digital camera fixed vertically above the specimen and focused to obtain a photo of the specimen’s surface. Digital grids were printed on the photos using graphical software. The indices representing the degree of surface defects were computed by calculating the numbers and size of defects per square grid.

Impact testing was carried out using a Tinius Olsen model IT 503 impact tester. The notch was prepared in accordance with ASTM D256 standard, using an automated Tinius Olsen model 899 notcher. Tensile and flexural testing was performed using a Tinius Olsen H25KS extensometer in accordance with ASTM D638 and D790 standards respectively.

6.3.2 Design of experiments

In this section, a 7 factor × 2 level fractional design of experiment was carried out to evaluate the influence of injection parameters on volume resistivity, surface resistivity and surface finish of PC composites (see Table 6.2). The PC used here was a Makrolon 2205 compound containing 3 wt% CNT. Constant injection parameters were; 28 cm3 fill volume, 25 seconds mould cooling time, 3.7 cm3 volume decompression after plasticising, and 10 cm3.s−1 volume decompression speed.

Table 6.2

Parameters and setpoint values used in the 27 DOE study

The aim of a design of experiment (DOE) activity is to obtain the maximum information with the minimum number of tests to identify the main factors that affect the responses. In this case, the factors were injection parameters and responses were the volume resistivity, surface resistivity and surface finish. A full factorial experiment (which consists of varying one factor at a time and performs experiments for all levels of all factors) is most effective, but it runs up a very large number of trials and often untenable cost. Taking a simple example of an experiment consisting of 7 factors, k and 2 levels i, the total number of experiments, n, that need to be performed in a full factorial design before considering replication would already be, n = ik = 27 = 128. Therefore, fractional factorial design was used to reduce the number of experiments and yet obtain adequate representation of the relationship between the experiment responses and variation of the factors.

The experiment was structured in form of 27−h fractional factorial treatment where h describes the fraction of the full factorial used or, formally, the number of generators assigned to which effects or interactions are confounded, (i.e. difficult to estimate independently of each other). Experiments were generated from the full factorial treatment by choosing an alias structure, where the alias structure determined the confounding effects. Referring to Table 6.3, e, f and g were chosen as confounding factors with interactions (thus, k = 3 and n = 16) generated by e = a · b · c, f = a · c · d and g = b · c · d. The effect of factor e, therefore, was a combination of the main effect of factor e and three factors’ interaction involving a, b and c. Another important feature of a fractional design is the defining relation, say, denoted by Z, which gave the set of interaction columns equal in the design matrix to a column of plus (+) sign. Since e = a · b · c, f = a · c · d and g = b · c · d, then a · b · c · e, a · c · d · f, and b · c · d · g were three additional columns of (+) sign, and consequently so was c · e · f · g. This defining relation allowed the combination of parameter set points of the DOE to be determined in Table 6.4.

Table 6.3

27−k fractional factorial DOE consisted of 16 runs, where e, f, g were generators

Table 6.4

Lower and upper level values listing for the 27−kfractional factorial DOE

6.4 Analysis

6.4.1 Main effects of injection moulding parameters

The injection moulding process was carried out using the combination of DOE parameters listed in Table 6.4. Figure 6.2 compares the response of volume resistivity with respect to the various injection factors (parameters), termed as ‘main effects’ plots. The response’s values were average measurement from six injected specimens and given in Table 6.5. All specimens were conditioned at room temperature for at least 24 hours before measurement. In a main effect plot, a parameter is more influential if the absolute gradient of the slope between the lower and upper data point is greater, and vice versa. The data points in each of the plots were calculated by taking the sum average of the volume resistivity of the parameter, corresponding to its lower or upper level set point. For example, in Fig. 6.2 (a), the lower data point was the sum average of ρv for samples A–H in Table 6.5. Samples A–H correspond to the lower level set point of injection speed (8 cm3 · s−1) listed in Table 6.4. As such, its upper data point was the sum average of ρv for samples I–P, with an upper level set point of (30 cm3 · s−1).

Table 6.5

Volume and surface resistivity values of Makrolon 2205, 3 wt% PC–CNT

6.2 Main effect plots of volume resistivity versus injection moulding parameters

By examining the gradient of slopes in the plots in Fig. 6.2, injection temperature, back pressure and hold pressure time first appeared to have greatest effect on the volume resistivity of the PC composites. However, this could be misleading since the slope’s gradient was affected by the chosen lower and upper level set points, for example, if the upper set point of injection speed in Fig. 6.2 (a) was chosen as 60 cm3 · s−1 instead of 30 cm3 · s−1, the gradient of its slope could become greater than that of the main effect plot for injection temperature in Fig. 6.2 (c). But this does not mean that the injection speed had a greater influence on the volume resistivity than the injection temperature. Therefore, the slope should first be normalised with respect to a coefficient of tolerance, r, where r is the magnitude (upper limit minus lower limit) of tolerance of a parameter (see Table 6.6). Outside the tolerance, a sample could not be suitably injection moulded, e.g. injecting at 250 bar back pressure would cause problematic specimen ejection from mould. The values of normalised slopes are given in Table 6.7. Injection temperature and back pressure appeared to have greatest influence on the volume resistivity and within the processing tolerance, the influence of hold pressure least significant.

Table 6.6

Injection moulding tolerance used for calculation of normalised main effect slopes

Table 6.7

Effect of injection moulding parameters affecting the volume and surface resistivities by order of ranking. The greater the slope’s value, the greater the effect

Figure 6.3 shows the main effect plots of surface resistivity versus the various injection parameters. Within the set points studied, surface resistivity appeared to be sensitive to only two major parameters: injection speed and injection temperature. The increase of surface resistivity with injection speed was most likely to be attributed to the formation of an orienting skin layer which in turn disrupts the conductive network path of the CNT. This result is consistent with similar findings by Villmow on through-plane volume resistivity (Villmow et al., 2008).

6.3 Main effect plots of surface resistivity versus injection moulding parameters.

Figures 6.4 and 6.5 show the main effect plots for two types of surface defects presented on the injection-moulded test specimens: pinholes and spots. Their corresponding intensity indices, p and q are given in Table 6.8, where 0 = lowest (no defect) and 1 = greatest. The p and q indices were obtained through methods described in the experimental section by calculating the numbers and size of pinholes and spots presented on the grids of the specimen photos. The normalised slopes for the main effect plots of surface defects were calculated in Table 6.9. Interpretation based on the slope’s gradient showed pinholes defect and spots formation to be most sensitive to injection speed and mould temperature. Increasing injection speed or reducing mould temperature typically improves the surface finish of the PC composites.

Table 6.8

Index of surface defects

Samples code Index, p pinholes Index, q spots
A 1 1
B 1 1
C 0.5 0.5
D 1 1
E 1 1
F 1 1
G 0.9 1
H 0.4 1
I 0 0
J 0.3 1
K 1 0.1
L 0 0
M 0 0
N 0 1
O 0 0
P 0.2 0.8

Note: 0 = lowest intensity, 1 = greatest intensity.

Table 6.9

Effect of injection moulding parameters affecting the formation of pinholes and spots surface defects by order of ranking. The greater the slope’s value, the greater the effect

6.4 Main effect plots of surface defect, pinholes index, p versus injection moulding parameters.

6.5 Main effect plots of surface defect, spots index, p versus injection moulding parameters.

Based on the obtained main effect results of volume resistivity, surface resistivity, pinhole index and spots formation index, a regression model using statistical software analysis was applied and the equation optimisation was performed by means of partial derivations and solving the system of equations (the optimisation procedure can be found in Joiner et al., 2004). The analytically optimised equation returned with parameters similar to run 12 (sample L) trial in Table 6.5, except Ph = 390 bar. The following results (within margin of errors) were compared, both parameters yielded the same resistivities but run 12 gave much better surface finish.

(a) Optimised: ρV = 8.2 ohm · cm, ρS = 9.2 ohm · sq−1, no pinholes, minor visible spots.

(b) Run 12, L: ρV = 6.7 ohm · cm, ρS = 8.6 ohm · sq−1, no pinholes, no visible spots.

In conclusion, the DOE optimisation analysis gave a result very close to the ideal processing condition despite not being as satisfactory compared to run 12 trial. This could be attributed to the lack of data points for a more accurate regression model fitting and can be improved by either reducing the number of generator h from 3 to 2 or by running a central composite design augmented with centre points that allow estimation of curvature.

6.4.2 Steady-state rheological analysis

Capillary rheology is an experimental technique whereby a melt sample experiences die shear to mimic realistic viscosity and high shear rates (of up to 10,000 s−1) relevant to most extrusions and injection moulding processes. By applying this rheological method in steady-state conditions, information with regard to the power-law regime of a composite melt that is not observed at lower shear rates (< 100 s−1) can be studied. Figure 6.6 shows the plots of steady-state apparent shear viscosity, η versus shear rate of PC composites and their virgin melt based on Makrolon 2205. Experiments were performed using a dual capillary rheometer and results were Bagley corrected for die effect. The virgin PC exhibited, at gamma dot < 1000 s−1, Newtonian-like flow behaviour (i.e. constant shear viscosity versus shear rate). In contrast, the PC composite exhibited intense shear thinning across the range of shear rate studied. The data plot demonstrated that η of the PC composite converged with η of the virgin PC at a shear rate K · ≈ 1500 s−1 and at K · ≥ 2000 s−1, η of the PC composite appeared to cross over below the η of the virgin PC.

6.6 Plots of shear viscosity versus shear rate of virgin PC and PC–CNT, 3 wt% and 15 wt%, based on Makrolon 2205.

Lew et al. (2009b) had investigated the steady-state rheological behaviour of PC composites (with 3 wt% CNT) using five different molecular weight PCs (between 13,000 and 23,500 g/mol) and observed the same phenomenon for all PC composites. The viscosity of all the PC composites melt would converge and then cross over to below their virgin melt. The convergence shear rate was dependent on the molecular weight, lower for higher Mw PC composite. Lew attributed this behaviour to flow-induced orientation of CNT that in turn led to interface chain slippage. This is given by the anisotropy structure realignment of the CNT particles and when exfoliated, its semi-graphitic wall could further act as plasticising agent.

Figure 6.7 shows the shear viscosity versus shear rate plots of different Mw PC (as-supplied pellets) and their corresponding compounded PC composites with 3 wt% CNT loading. All composites shear thinned intensely irrespective of Mw and more remarkably, the PC composites exhibited unique flow modification where the melt exhibited power law flow behaviour. This could, in part, explain the reduced die swellability found previously: (a) shear thinnability of virgin PC melt increases with increasing Mw; and (b) shear thinnability of PC composites roughly followed a power law described by η = K · n−1 where the exponential n was almost independent of Mw.

6.7 Plots of shear viscosity versus shear rate of various virgin PCs and their corresponding 3 wt% PC–CNTs. In the order from highest Mw to lowest Mw are Makrolon 3105, Makrolon 2805, Makrolon 2405, Makrolon 2205 and Makrolon OD2015.

As early as 2004, researchers from the National Institute of Standards and Technology (NIST) (Kharchenko et al., 2004) had observed the shear thinning behaviour of PP melt imparted by CNT. They also found the presence of CNT would eliminate die swell in polymers, an effect crucial to processing. Vega et al. (2009) had compared the extrusion die swell ratio of a virgin HDPE and HDPE–CNT and they too observed that HDPE–CNT exhibited lower die swell compared to the virgin HDPE and the increase in die swell versus apparent shear rate was lower for the HDPE–CNT. Several authors ascribed the decrease in the polymer–CNTs viscosity relative to their respective virgin polymer to an increase in molecular mobility as a result of the increase in free volume (Jin et al., 2008; Ma et al., 2008; Rahmatpour and Aalaie, 2008).

The Mw data measured for all materials in Fig. 6.7, obtained via gel permeation chromatography (GPC) analysis, showed slightly lower Mw for the composites vis-à-vis their virgin polymers, which could be an indication of degradation. The Mw data for the virgin PCs were measured on the as-supplied pellets without processing history. Pötschke et al. (2002) reported a similar reduction in Mw of melt compounded PC–CNT which was discussed to be due to high shear forces during composite preparation. However, from the GPC data of virgin PCs (for example, Makrolon 2205) compounded under the same processing history as its PC composites, it was shown that its Mw was slightly lower than that of the Makrolon 2205 PC composites (Mw of virgin as-supplied PC = 15,000 g/mol, Mw of PC composite = 14,500 g/mol, Mw of virgin PC undergone same processing history = 14,000 g/mol) (Lew et al., 2009a). This would indicate the Mw degradation found in the composites was most likely attributed to degradation of the virgin PC alone and not caused by the presence of CNT.

Figure 6.8 shows the plots of virgin PC and PC composite melts at 1.8, 2.2, 2.6, 3.0 and 4.0 wt% CNT loading, based on Makrolon 2405 (MFI 20 g/10 min). The entire composite melts demonstrated shear thinning effect within the range of shear rates investigated. The flow behaviour of the virgin PC was Newtonian-like at K · < 500 whereas all composites had already exhibited intense shear thinning at K · = 50 s−1. At lower CNT loading (1.8 wt%), three distinct melt flow regimes could be observed: (a) power law flow at K · < 500 s−1; (b) Newtonian plateau between 900 and 2000 s−1; and then (c) power law flow at K· > 4000 s−1. These three regimes appeared to conflate into a single power law regime as the CNT loading increases. In addition, the power law exponent n decreases (i.e. the slope’s gradient increases) in proportion with CNT loading and this had concurred with study by Teng et al. (2008) on polypropylene–CNT composites. The same phenomenon was seen formerly in Fig. 6.6 where the curvature in the viscosity versus shear rate plot at 3 wt% CNT changed into a linear at 15 wt% of CNT loading.

6.8 Plots of shear viscosity versus shear rate of virgin PC and PC–CNT containing different CNT loading, based on Makrolon 2405.

The results showed that at 1.8 wt% CNT, the PC composite already exhibited lower viscosity than its virgin melt at K · = 100 s−1 and electrical resistivity data for this composite (shown later in Section 6.4.4, Fig. 6.13) recorded favourably values of ρV < 102 ohm-cm and ρS < 103 ohm · sq−1. Another unique behaviour was observed in Fig. 6.8 where the rate of shear thinning decreases with increasing CNT percentage, or in terms of power law, the exponential n increases with increasing loading. This could be ascribed to an increase in the chain plasticising effect discussed previously. Furthermore, there appeared an inverse intersecting point at K · 1000 s−1 at which the proportional relationship between CNT loading and viscosity was reversed.

Most of the rheological experiments found in the literature for polymer CNT composite were conducted in dynamic oscillatory mode and typically the composite registered a higher shear viscosity compared to its virgin melt, although the difference became smaller as the rate of deformation increases (McNally et al., 2005). Discrepancies in flow mechanics arise between steady-state shear and oscillatory experiments because: (a) in a steady-state flow the polymer experiences linear mode deformation vis-à-vis nonlinear deformation in dynamic mode; and (b) deformation strain and rate experienced in steady-state experiment (K · = 100–10,000 s−1) exceed the critical elongational for the polymer chain to change from coil (entanglement) behaviour into stretched conformation (laminar) behaviour. Typical deformation rates in a dynamic oscillatory experiment are in the range of 0.001–100 s−1 or rads−1. In addition, the shear viscosity in dynamic mode is more complex (η*) because it is very sensitive to the elastic (G′) and viscous (G″) components. Finally, it should be highlighted that flow behaviour in the majority of polymer processing is more closely related to steady state kinematics in both deformation mechanics and shear rate, > 300 s−1 for single-screw extrusion, > 2000 s−1 for twin-screw extrusion and > 5000 s−1 for injection moulding.

6.4.3 Analysis of nanotube agglomerates in composites

The structure of agglomerates derived from dispersion in a compounding process typically does not change during the injection moulding process, in the absence of mixing kinematics and because of short residence time. However, change in local distribution of CNT agglomerates can be induced during injection moulding, especially during the mould-filling process arising from the difference in flow behaviour and shear gradient experienced in different sections of the mould. Therefore, it needs to be pointed out that the section from which an injection-moulded specimen was obtained could have an effect on the outcome of the analysis. All specimens used for agglomerates inspection in this work were obtained from the mid-plane section of injection-moulded Izod test bars.

Figures 6.96.12 show the optical microscopy inspection of the patterns of CNT agglomerates presented in Makrolon 2405 PC composites. For composites containing different levels of CNT loading, from 1.4 to 4 wt%, the agglomerates were visible at two levels of dimension: (a) well-distributed sub-5 micron particles; and (b) 10–20 micron particles. However, at a CNT loading of 3 wt% or higher, agglomerate domains of larger than 20 micron became visible.

6.9 Optical microscopy inspection of agglomerates in Makrolon 2205 PC–CNT containing different CNT loading: (a) 1.4; (b) 1.8; (c) 2.2; (d) 2.6; (e) 3.0; (f) 4.0 wt%.

6.10 Optical microscopy inspection of agglomerates in Makrolon 2205, 3 wt% PC–CNT moulded at different injection speeds: (a) 10; (b) 20; (c) 60; (d) 100 cm3.s−1.

6.11 Optical microscopy inspection of agglomerates in Makrolon 2205, 3 wt% PC–CNT moulded at different injection melt temperatures: (a) 245; (b) 255; (c) 265; (d) 275; (e) 285; (f) 295 °C.

6.12 Optical microscopy inspection of agglomerates in Makrolon 2205, 3 wt% PC–CNT moulded at different injection back pressures: (a) 25; (b) 50; (c) 100; (d) 200 bar.

The influence of injection fill speed on the patterns of agglomerates is illustrated in Fig. 6.10. No significant difference in the pattern of agglomerate distribution and size was observed within the range of injection speed applied. The influence of the injection moulding temperature (from 245 to 295 °C) on the pattern of agglomerate distribution and size is shown in Fig. 6.11. Two patterns were found in the observation: (a) between 245 and 265 °C, larger and poorly distributed agglomerates; and (b) between 275 and 295 °C, smaller and better distributed agglomerates. This could indicate the presence of a critical injection temperature between 265 and 275 °C. More coincidently, the same critical temperature between 265 and 275 °C was observed later in the volume resistivity versus injection temperature plots in Fig. 6.15, at which point the resistivity began to level off. In the twin-screw extrusion investigation of PC–CNT by Lew et al. (2009a), he had shown, through scanning electron microscopy and transmission electron microscopy, the effect of the extrusion temperature (230–290 °C) on the morphologies of CNT aggregates. Despite the fact that aggregates of PC–CNT extruded at higher temperature were slightly larger than aggregates extruded at lower temperature, they exhibited a looser structure due to more improved wetting with the polymer. The reduction in larger agglomerates size observed in this work could therefore be attributed to a critical decrease in surface tension occurring at between 265 and 275 °C, which caused the agglomerates to rupture. However, more detailed transmission electron microscopy and scanning electron microscopy work are required to determine this phenomenon. Figure 6.12 shows the effect of injection back pressure on the size and distribution pattern of the CNT agglomerates. Generally, a more increase in back pressure will contribute to a more improved mixing in the nozzle section and increase in the plasticising time which might improve the rupturing of agglomerates. However, the photomicrographs in Fig. 6.12 did not detect significant change of the agglomerate size within the range of back pressure studied (25–200 bar).

6.4.4 Analysis of electrical resistivity

This section presents the parametric correlation between injection moulding parameters and behavioural change in electrical resistivity. The wider range of injection moulding parameters and parameter values than available hitherto in the literature were investigated (see Table 6.10). The PC grade used in this study was Makrolon 2405 (MFI 20 g/10 min) and contained 2.5 wt% CNT, unless stated otherwise. For each of the investigated parameters, values of all other parameters were maintained constant (see Table 6.11). The parameters investigated were: (a) injection temperature; (b) injection speed; (c) back pressure; (d) hold pressure; (e) hold pressure time; (f) mould temperature; (g) plasticising speed; and (h) volume decompression after plasticising.

Table 6.10

List of injection moulding conditions used for parametric study in Section 6.4.4

Parameter Unit Values
Injection speed cm3 · s−1 10–20–30–40–50–60–80–100
Plasticising speed cm · s−1 5–10–20–30–40–50–62
Injection temperature ºC 235–245–255–265–275–285–295–305–315–325–335
Mould temperature ºC 40–50–60–70–80–90–100–110–120–130–140–150
Back pressure bar 25–50–100–150–200–300
Hold pressure bar 50–100–200–300–400–500–600–750–900–1050
Hold pressure time s 1.5–3.5–6.5–10–15–21–28
Decompression volume % 0.37–11.9–22.2–44.4–59.3–74.1

Table 6.11

Standard injection moulding conditions for Makrolon 2405, 2.5 wt% PC–CNT ASTM D256 Izod test bar

Parameter Unit Values
Injection speed cm3 · s−1 30
Plasticising speed cm · s−1 40
Injection temperature °C 295
Mould temperature °C 120
Back pressure bar 50
Hold pressure bar 450
Hold pressure time s 10
Mould cycle time s 24
Decompression volume % 11.9

Injection-moulded polymer is usually characterised by tri-layer morphology. The outer layer is a highly oriented skin, formed as a result of extensional flow that took place at the flow front. The mid-zone is a shear layer formed as a result of shear flow, where chains are highly oriented along the flow direction. The third layer is the core which was formed after cessation of flow. The thickness of the skin and shear layers decreases while the core increases with the increase of the mould temperature or injection speed.

Figure 6.13 shows the electrical percolation plots of volume resistivity and surface resistivity versus wt% CNT loading. The lowest and highest test points were 1.4 and 5.0 wt%. The corresponding volume and surface resistivity at the lowest test point were 3.4 × 103 ohm · cm and 1.0 × 1012 ohm · sq−1. In order to estimate the percolation threshold, curve fitting analysis was performed according to Equations 6.3 and 6.4 for volume and surface resistivity respectively.

6.13 Plots of volume and surface resistivities of Makrolon 2405 PC–CNT versus wt% CNT loading.

[6.3]

[6.4]

where in Equation 6.3, a = 8.574, b = 101.187, c = − 179.926, in Equation 6.4, a = − 1.737, b = − 0.884, c = − 3.45 and r2 = 1 for both equations. The estimated percolation threshold was remarkably low, ranging between 1.05–1.15 wt% (6 × 109 −1.5 × 107 ohm · cm) for volume resistivity and 1.5 – 1.6 wt% (1.6 × 109 – 4.0 × 107 ohm sq−1) for surface resistivity. In addition to favourable injection moulding parameters, this was largely attributed to optimisation of twin-screw processing condition during the compounds production.

Figure 6.14 shows the relationship between electrical resistivity and polymer Mw. All compounds contained 3 wt% CNT and were produced under the same extrusion and injection moulding condition. The polycarbonates were Makrolon OD2015, 2205, 2405, 2805 and 3105. A unique trend was observed where ρV 1/Mw, ρS Mw and at a critical Mw the resistivity levels off and then increases after critical Mw. In a steady state shear flow, shear gradient increases with Mw since a higher Mw species is more shear sensitive. This might possibly induce greater CNT migration towards the core and depletes CNT near to the wall which, in turn, increases surface resistivity. Another possible explanation was as Mw increases, the wall section experiences greater shear stress and, as a result, increases CNT orientation near to the surface.

6.14 Plots of volume and surface resistivities of 3 wt% PC–CNT injection moulded from PCs with different molecular weights.

Figure 6.15 shows the influence of injection temperature on electrical resistivity. Both surface and volume resistivity were observed to decrease with injection temperature. This was due to alteration of flow behaviour at higher temperature where:

6.15 Plots of volume and surface resistivities of Makrolon 2405, 2.5 wt% PC–CNT versus injection melt temperature.

• polymer viscosity decreases (and hence shear stress), and as a result decreases CN T orientation which, in turn, increases network connectivity between the CNT particles, as reported by Villmow et al. (2008);

• relaxation of the polymer chain improves, causing short-range reflocculation of the CNT particles;

• reduced shear gradient which caused less migration away from the wall. As volume conductivity acts in parallel with surface conductivity, a decrease in surface resistivity also reduces the volume resistivity. However, because the effect has a greater influence on the surface, therefore the rate of conductivity decrease is greater for the surface compared to the volume.

Figure 6.16 illustrates the relationship between injection speed and electrical resistivity. The opposing trend was observed for the volume resistivity (decreases with speed) and surface resistivity (increases with speed). The increase in volume resistivity with injection speed could be attributed to the migration of CNT from the wall towards the core because of increased shear gradient which, in turn, resulted in higher concentration of CNT in the sample’s bulk. On the contrary, Villmow et al. (2008) had shown that the volume resistivity of PC–CNT composites rises with increasing injection speed, attributed to more formation of skin layer with more highly oriented CNT. Contradiction between results in this study and that of Villmow et al. was due to different types of volume resistivity measurement used; in-plane volume resistivity versus through-plane volume resistivity by Villmow. The through-plane volume resistivity measurement applied by Villmow was sensitive to the morphology of the skin layer as in surface resistivity measurement. Figure 6.16 shows that the surface resistivity (sensitive to skin layer) increases as the injection speed increases and is consistent with the through-plane volume resistivity results by Villmow et al. (2008). As the injection speed decreases, orientation of CNT was less intense in the skin layer and therefore gave rise to lower surface resistivity.

6.16 Plots of volume and surface resistivities of Makrolon 2405, 2.5 wt% PC–CNT versus injection speed.

Figure 6.17 shows that both the volume and surface resistivity increase with the back pressure. The trend is subtle but clear. To explain the trend, it is important for the reader to understand the effect of back pressure on the flow kinematics. Back pressure is related to the amount of force (hydrostatic pressure) exerted on melt in the nozzle section during the plasticising stage. Increasing the back pressure will (a) increase the nozzle melt filling time, which means longer plasticising time; and (b) increase the flow pressure. Therefore, an increase in back pressure can be associated with an increase of mixing in the nozzle melt and injection speed. This would lead to increased orientation of CNT on the skin and hence higher resistivity. However, the flow behaviour derived from the increase of back pressure was different from that of the increase of injection speed, in which the flow experienced a surge when exiting the nozzle and therefore reducing the shear gradient. Because of the reduction in shear gradient but the increase in skin orientation, the volume resistivity also increases.

6.17 Plots of volume and surface resistivities of Makrolon 2405, 2.5 wt% PC–CNT versus injection back pressure.

Figure 6.18 shows the relationship between electrical resistivity and injection hold pressure. No significant difference was observed for the volume resistivity on alteration of the injection hold pressure. This may be attributed to: (a) insensitivity of the CNT-enriched core (bulk) region to an increase in hold pressure or (b) insensitivity of the core to an increase in compaction force because the specimen’s core, gates or sprue have already been solidified before the application of hold pressure. The increase in hold pressure (compaction force), however, led to an increase in the surface resistivity. Figure 6.19 shows the relationship between electrical resistivity and injection hold pressure time. Both the volume and surface resistivities appeared to increase with increasing hold pressure time up to a point and then level off (15 seconds for volume resistivity and 10 seconds for surface resistivity).

6.18 Plots of volume and surface resistivities of Makrolon 2405, 2.5 wt% PC–CNT versus injection hold pressure.

6.19 Plots of volume and surface resistivities of Makrolon 2405, 2.5 wt% PC–CNT versus injection hold pressure time.

The effect of mould temperature on electrical resistivity is shown in Fig. 6.20. Surface resistivity increases proportionally with the mould temperature. This is an obvious effect of thicker skin layer formation and greater skin layer orientation, as had been demonstrated by Villmow et al. (2008) through TEM analysis. No notable effect was detected on the volume resistivity for mould temperature between 40 and 100 °C. This was consistent with work by Lellinger et al. (2008) that showed the volume resistivity was consistent across the mould temperature range from 70–100 °C.

6.20 Plots of volume and surface resistivities of Makrolon 2405, 2.5 wt% PC–CNT versus injection mould temperature.

The effect of screw plasticising speed on the electrical resistivity is shown in Fig. 6.21. There was a clear increase in volume resistivity with increasing plasticising speed, but the magnitude was very small and almost negligible. Nonetheless, plasticising speed was found to have more effect on the surface resistivity. Surface resistivity was lowest at the very low plasticising speed of 5–10 cms−1, followed by an abrupt rise at between 10–20 cms−1 and, then, to decrease gradually with a further increase in speed. When the plasticising speed was too low, the time required for melting and nozzle filling increased and the flow pressure profile in the nozzle became more homogeneous. Because of increased residence time, melt temperature was also higher and, thus, contributed to a lower surface resistivity. When the plasticising speed was too high, not only the melt residence time in the nozzle was reduced, but the melt could be non-uniform (in this case, back pressure would play a more important role). Figure 6.22 shows the influence of volume decompression after plasticising on the electrical resistivity. The unique pattern was seen in which the volume resistivity was reciprocal to the surface resistivity. A reversal in the curve gradient was observed at a percentage decompression between 40 and 50%. The kinematics of influence was not clear but may be associated with the reduction in back pressure, increased melt relaxation and modification of flow behaviour at higher volume decompression.

6.21 Plots of volume and surface resistivities of Makrolon 2405, 2.5 wt% PC–CNT versus injection plasticising speed.

6.22 Plots of volume and surface resistivities of Makrolon 2405, 2.5 wt% PC–CNT versus volume decompression after plasticising.

6.5 Conclusion

The work carried out in this chapter is intended to provide a clearer insight into the relationship between injection moulding parameters and the electrical resistivity of polymer CNT composites. Polycarbonates were used in this study, but the trend is expected to be applicable for larger families of polymers. To meet this objective, our work features more comprehensive experimental data points and sets of injection parameters than hitherto available in the literature. Resistivity in both volume (in-plane) and surface was tested and explained in parallel since their relationship is not always proportional and the mechanism governing their evolution is different. The relationship between injection parameters and surface finish was also investigated given its ramifications in industrial products development, where surface aesthetic value is an important issue. In the first section, an experiment design was conducted to find the optimum process–properties relationship between volume resistivity, surface resistivity, and surface finish. The optimum set of found parameters was adjusted for the parametric study in the subsequent sections on the effect of individual injection condition. The effect of molecular weight (viscosity) on modification of flow behaviour and, consequently, electrical resistivity was investigated using five different grades of polycarbonates.

Results indicated the two most influential parameters on the modification of electrical resistivity (both volume and surface) were the injection melt temperature and injection speed. On the other hand, the surface finish was most affected by the injection speed and mould temperature. In general, both the volume and surface resistivities followed a proportional increase relationship with the injection melt temperature. Increasing injection speed decreases the volume resistivity but increases the surface resistivity. Higher mould temperature and greater injection speed improve the surface finish (reduce the formation of spots and pinholes).

In addition to the classical explanation relating the modification of electrical resistivity (in particular surface resistivity) to the alteration of the chain and nanotube orientation, a complementary explanation was given based on the influence of shear-induced migration. Shear-induced migration relates reciprocally to shear field fractionation of lower molecular weight or smaller species away from the point of lowest shear rate (centreline). Following the shear-induced migration kinetics, it may be possible to decrease the surface resistivity by means of blunting the flow profile or reducing the flow gradient of a shear stream during mould filling.

The steady-state rheological analysis showed that incorporation of CNT at a lower percentage (< 5 wt%) has not affected the melt viscosity in the region of processing windows corresponding to industrial plastic fabrication. On the contrary, the presence of CNT has demonstrated a shear thinning effect even at very low shear rate (< 100 s−1), where the flow of the virgin polymer is governed by a Newtonian flow behaviour. The shear thinning effect imparted by CNT would have significant implications in terms of melt processability. Shear thinning of a polymer–CNT composite decreases its melt viscosity to become comparable or lower than its virgin form, and therefore, improved processability.

6.6 References

Aarøe, E.R., Blaimschein, K.S., Decker, L., Stentoft-Christensen, H.C., Examining the influence of injection speed and mould temperature on the tensile strength of polypropylene and ABS Report no. 71738, Experimental Plastic Technology, DTU Mekanik. Institut for Mekanisk Teknologi, June 2009.

Barron, A.R., Khan, M.R. Carbon nanotubes: opportunities and challenges. Adv. Mater. Processes. 2008; 166:41–43.

Chandra, A., Kramschuster, A.J., Hu, X., Turng, L.-S., Effect of injection molding parameters on the electrical conductivity of polycarbonate/carbon nanotube nanocomposites. Proc. Annual Technical Conference of the Society of Plastics Engineers, Cincinnati, US. 2007:2184–2188. [3].

Hong, C.-M., Kim, J., Jana, S.C. Shear-induced migration of conductive fillers in injection molding. Polym. Eng. Sci. 2004; 44:2101–2109.

Jana, S.C., Doni, S., On the effects of migration of filler particles on conductivity and mechanical properties. Proc. Annual Technical Conference of the Society of Plastics Engineers, Dallas, US. 2001:2022–2026. [59].

Jin, S.H., Choi, D.K., Lee, D.S. Electrical and rheological properties of polycarbonate/multiwalled carbon nanotube nanocomposites. Coll. Surf. A: Physicochem. Eng. Asp. 2008; 313–314:242–245.

Joiner, B., Cryer, J., Ryan, B.F. Minitab B, 5th edn. New York: Brooks/Cole; 2004.

Kharchenko, S.B., Douglas, J.F., Obrzut, J., Grulke, E.A., Migler, K.B. Flow-induced properties of nanotube-filled polymer materials. Nature. 2004; 3:564–568.

Leighton, D., Acrivos, A. The shear-induced self-diffusion in concentrated suspension. J. Fluid Mech. 1987; 181:415–439.

Lellinger, D., Xu, D., Ohneiser, A., Skipa, T., Alig, I. Influence of the injection moulding conditions on the in-line measured electrical conductivity of polymer-carbon nanotube composites. Phys. Stat. Sol. (B). 2008; 245:2268–2271.

Lew, C.Y., Claes, M., Luizi, F., The influence of processing conditions on the electrical properties of polypropylene nanocomposites incorporating multiwall carbon nanotube. Proc. 6th European Conference on Additives and Colors, Antwerp, Belgium. 2009:1–7.

Lew, C.Y., Xia, H., McNally, T., Fei, G., Vargas, J., Millar, B., Douglas, P., Claes, M., Luizi, F., A unified strategy to incorporating nanotube in twin-screw extrusion processing. Proc. Polymer Processing Society Annual Regional Conference, Larnaca, Cyprus. 2009:1–10.

Ma, A.W.K., Chinesta, F., Mackley, M.R., Ammar, A. The rheological modelling of carbon nanotube (CNT) suspensions in steady shear flows. Int. J. Mater. Forming. 2008; 1:83–88.

McNally, T., Pötschke, P., Halley, P., Murphy, M., Martin, D., Bell, S.E.J., Brennan, G.P., Bein, D., Lemoine, P., Quinn, J.P. Polyethylene multiwalled carbon nanotube composites. Polymer. 2005; 46:8222–8232.

Mekhilef, N., Ait-Kadi, A., Ajji, A., Effect of processing conditions on mechanical properties of polycarbonate-polyethylene blends. Proc. Annual Technical Conference of the Society of Plastics Engineers, Montreal, Canada. 1991:1118–1121.

Park, D.H., Yoon, K.H., Park, Y.-B., Lee, Y.S., Lee, Y.J., Kim, S.W. Electrical resistivity of polycarbonate/multiwalled carbon nanotube composites under varying injection molding conditions. J. Appl. Polym. Sci. 2009; 113:450–455.

Phillips, R.J., Armstrong, R.C., Brown, R.A., Graham, A.L., Abbott, J.R. A constitutive model for concentrated suspensions that accounts for shear-induced particle migration. Phys. Fluids A. 1992; 4:30–40.

Pötschke, P., Fornes, T.D., Paul, D.R. Rheological behaviour of multiwalled carbon nanotube/polycarbonate composites. Polymer. 2002; 43:3247–3255.

Rahmatpour, A., Aalaie, J. Steady shear rheological behaviour, mechanical properties, and morphology of polypropylene/carbon nanotube nanocomposites. J. Macromol. Sci., Part B. 2008; 47:929–941.

Shelby, M.D., Caflisch, G.B. Shear field induced diffusion and molecular weight fractionation during polymer processing. Polym. Eng. Sci. 2004; 44:1283–1294.

Teng, C.C., Ma, C.C.M., Huang, T.W., Yuen, S.M., Weng, C.C., Chen, C.H., Su, S.F. Effect of MWCNT content on rheological and dynamic mechanical properties of multiwalled carbon nanotube/polypropylene composites. Composites: Part A. 2008; 39:1869–1875.

Vega, J.F., Martinez-Salazar, J., Trujillo, M., Arnal, M.L., Muller, A.J., Bredeau, S., Dubois, Ph. Rheology, processing, tensile properties, and crystallization of polyethylene/carbon nanotube nanocomposites. Macromolecules. 2009; 42:4719–4727.

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. Comp. Sci. Tech. 2008; 68:777–789.

Zhang, H., Zhang, Z. Impact behaviour of polypropylene filled with multi-walled carbon nanotubes. Eur. Polym. J. 2007; 43:3197–3207.

6.7 Appendix: list of units

Nomenclature Unit
ρV Volume resistivity Ω · cm
ρS Surface resistivity Ω · sq−1
Tinj Injection temperature* °C
Tmo Mould temperature °C
Pba Back pressure bar
Ph Hold pressure bar
Vinj Injection speed cm3 · s−1
Vplas Plasticising speed cm · s−1
Vde Volume decompression % cm3
th Hold pressure time s

*Melt temperature measured at the injection nozzle.


1Markets and Markets Consulting, ‘Global electroactive polymers market 2009–2014’, Report CH1055, February 2010.