Improved through-plane electrical conductivity in a carbon-filled thermoplastic via foaming.

INTRODUCTION

Electrically conductive thermoplastic composites are growing in demand within the energy, electronic, and medical sectors for applications such as thermal switches, EMI shielding, blood sensors, and fuel cell bipolar separator plates. The advantages often sought by selecting these conductive polymers, over more traditional materials such as metals or ceramics, are their low density, good chemical resistance, and barrier properties, as well as being more readily formed into complex shapes without expensive secondary processing steps. The conductivity of these materials is dependent upon the volume content, inherent conductivity, surface and interfacial properties, and aspect ratio of the fillers incorporated [1-6]. Polymer composites containing low aspect ratio powder type fillers, such as carbon black (CB), exhibit a relatively uniform volumetric conductivity throughout a molded part; however, they are not easily processed by extrusion or injection molding due to their extremely high viscosities. Low aspect ratio fillers tend to have a higher percolation threshold characteristic, as compared to high aspect ratio fillers at equivalent surface area. Therefore a large amount of the additive must be added to the polymer to achieve significantly higher conductivity, compared to the inherently insulating base material, and as a result the molten material exhibits shear viscosities too high for traditional processing machinery [5, 6]. In place of powders, high aspect ratio conductive fillers (such as carbon fibers) have been used as the polymer composites which exhibit a lower viscosity due to their relatively low percolation threshold and preferential fiber alignment within a flow field [6]. The preferential orientation of high aspect ratio fillers in the flow direction within molded parts leads to anisotropic electrical properties favoring electron transport predominantly in the in-plane direction [6-8]. This fact is not necessarily a negative depending on the application; however, for uses in which conductivity through one of the opposing planes is more highly desired, materials with near exclusive in-plane conductivity are not deemed suitable. In addition, a unidirectional fiber composite exhibits the highest percolation threshold while a randomly oriented fiber composite has the lowest percolation threshold due to more fiber-fiber contacts [9]. The present work examines a method of disrupting the preferential orientation of high aspect ratio fillers during the processing of electrically conductive composites through foaming. A previous article by the authors [6] studied the conductivity-rheology relationship to formulate hybrid filler composites suited to injection molding processes.

Wilson [10] discussed the possibility of reorientation of glass fibers due to bubble growth in composite foams, as shown in Fig. 1, though Wilson never reported experiments to prove his theory. He speculated that bubble growth in close proximity to a fiber would result in out-of-plane reorientation; though such circumstances are not easily achieved since it is known that the glass fibers in composite do not act as nucleating sites during foaming [11, 12]. Yang et al. [13] prepared carbon nanofiber-polystyrene composites by solution casting and found that the foam composite had a similar conductivity and percolation threshold to its solid composite counterpart. None of the studies mentioned above demonstrated the feasibility of using foaming as a mechanism for controlling fiber orientation. Therefore, this work examines the effect of foaming at different processing conditions on the fiber orientation and consequent electrical conductivity of injection molded polymer composite plates. Matrix viscosity, fiber length, and the kinetics of bubble growth versus solidification are all expected to influence the mechanics of fiber reorientation phenomenon. The principle outcome sought in this work is a strategy for maximizing the through-plane to in-plane conductivity ratio within a molded sheet.

[FIGURE 1 OMITTED]

EXPERIMENTAL

Materials

Cyclic olefin copolymer (COC), (Topas 6013S-04 supplied by Ticona, MFR 14 g/10 min at 260[degrees]C, 2.16 kg) was chosen as the matrix polymer as it represents a suitable candidate for bipolar plate manufacturing. The amorphous character of this polymer was also an important consideration for this study to minimize concerns regarding crystallization affecting the fiber orientation. The carboneous fillers used in this study included 5-mm long, 7.5 [micro]m diameter PAN-derived chopped carbon fibers (CF) (AGM-94 from Asbury Carbons) and a high surface area CB (Ketjenblack EC-600JD from Akzo Nobel with aggregate size of 30-100 nm and surface area of 1250 [m.sup.2]/g). Electrical conductivities of the used COC, CF, and CB as reported by the suppliers were <[10.sup.-16], 625, and 10-100 S/cm, respectively. Further properties of the fillers have been reported in a previous article [6].

A masterbatch exothermic chemical blowing agent (CBA) was used for the trials, IM 2240 supplied by Dempsey Corporation. The CBA masterbatch contained 20 wt% 5-Phenyl tetrazol which was the active component involved in the evolution of nitrogen gas for foaming. The carrier resin of the masterbatch was polycarbonate. The CBA was selected for its high decomposition temperature which suits the conditions necessary for processing the COC polymer in an injection molding machine. The onset of decomposition for CBA was determined by TGA analysis to be 225[degrees]C with the majority of gas evolved at 240[degrees]C.

Procedure

A Leistritz ZSE 27-40 corotating twin screw extruder with side feeder was used to produce the composite material used in the study. A temperature profile of 260[degrees]C for barrel zones 2-9 and 250[degrees]C for zone 1 and the die was used and a constant screw speed of 100 rpm was selected. COC pellets and CB were fed to the first block of the TSE by a weight-loss feeder, while a CF masterbatch (prepared on a Haake Rheocord batch mixer) was introduced into the melt by a side feeder half way down the length of the machine. The screw design was configured to maximize distributive mixing by use of comb elements and minimize shear through the limited use of 30[degrees] and 60[degrees] kneading blockings after the melting zone. The compounded material (prior to injection molding and foaming) contained 10 vol% CF and 2 vol% CB.

An Arburg 55 ton injection-molding machine with a 30 mm 20 L/D plasticating unit was used to prepare rectangular plaques with face dimensions of 14.5 cm X 3 cm and a thickness of 3 mm. A [2.sup.4-1] fractional factorial design was used to evaluate several different processing factors as shown in Table 1, which produced both foamed and nonfoamed samples. The factors examined were blowing agent content, injection flow rate, melt temperature, and mold temperature. All other processing factors were kept constant for the trials except packing pressure, which was set at 500 bar for all nonfoamed runs and not used for the foamed runs. To produce the foamed samples, IM2240 pellets were dry blended with the composite pellets and fed into the hopper. The temperatures of the first three zones for the plasticating unit were always set below 220[degrees]C to prevent premature decomposition of CBA and gas loss, while the temperatures of the last two zones on the plasticating unit and nozzle were set above 240[degrees]C to obtain full decomposition. The injection nozzle was also equipped with a shut-off nozzle to prevent gas loss during injection.

[FIGURE 2 OMITTED]

Analysis

As shown in Fig. 2, the injection molded plaques were cut using a high pressure water jet into 3.75 mm X 27 mm rectangular samples for which in-plane electrical conductivity was measured along the long (L) and short (W) axes using a 4-probe method. Disks of 20-mm diameter were cut by the same method for determination of through-plane (T) electrical conductivity using a 2-probe method. These measurement techniques are described in a previous article [6]. A minimum of 7 and 3 samples for the in-plane and through plane measurements, respectively, were used for each reported value. In most cases, the applied voltage was kept around 0.5-1 V to minimize joule heating; however, due to the broad range of the conductivities this was not always possible.

The density of the produced samples was determined by the buoyancy method. To measure the fiber length distribution (FLD), samples were ashed in an electrical furnace at 550[degrees]C for 10 min to remove the polymer. Initial examination showed that fibers collected in this manner were consistent with the solvent digestion technique. To obtain well spaced fibers for viewing under the microscope, the collected fibers from the furnace were suspended in xylene and then spread over microscope slides. Digital images were taken of the slide under an optical microscope after the solvent had evaporated, and these micrographs were analyzed using Sigma Scan Pro 3.0 image analysis software from Jandel Scientific to quantify the lengths of at least 1000 fibers from each sample. To characterize CF orientation and foam structure, composites were mounted in epoxy resin and metallographically polished with an automatic Struer polisher. Digital images taken of the polished surface by reflective optical microscopy were quantitatively analyzed using Northern Eclipse 6.0 software. By adjusting the grey scale and contrast for the images, we were able to distinguish fibers or cells to measure either fiber orientation or cell size. The distributions were determined from the analysis of approximately 1500 fibers and 1000 foam cells. To quantitatively express the orientation state of the fibers, 3D orientation factors were calculated according to the following expressions:

[f.sub.X] = [1/n] [summation] [cos.sup.2] [alpha] [cos.sup.2] [beta] (1)

[f.sub.Y] = [1/n] [summation] [cos.sup.2] [alpha] [cos.sup.2] [beta] (2)

[f.sub.Z] = [1/n] [summation] [sin.sup.2] [beta] (3)

where, [alpha] and [beta] are the in-plane and out-of-plane orientation angles, respectively as defined in Fig. 3. The parameter [alpha] was directly determined by the image analyzer while [beta] was calculated from the dimensions of the major (b) and minor (a) axes of the elliptical fiber shape shown in Fig. 3 and given in Eq. 4:

[beta] = [sin.sup.-1](b/a) (4)

Depending on the characterized polished plane, the X and Y subscripts in Eqs. 1 and 2 were arbitrarily specified for the polished plane axes while Z in Eq. 3 was the axis perpendicular to the polished plane. For example, when analyzing the LW polished plane, [f.sub.X] = [f.sub.L], [f.sub.Y] = [f.sub.W], and [f.sub.Z] = [f.sub.T]. The fiber orientation calculations were constrained such that each orientation factor could only vary from 0 to 1 and [f.sub.L] + [f.sub.W] + [f.sub.T] = 1. For a composite with random oriented fibers [f.sub.L] = [f.sub.W] = [f.sub.T] = 1/3 and for a composite with unidirectional orientation in the L direction [f.sub.L] = 1 and [f.sub.W] = [f.sub.T] = 0. For the core region of the foamed samples, orientation factors [f.sub.L], [f.sub.W], and [f.sub.T] were determined in all three planes (LW, LT, and WT) and the arithmetic average was reported as a single value to demonstrate the order of the orientation in L, W, and T directions. For the skin region, orientation factors were only obtained from the WT plane and compared with the orientation state in the core WT plane.

[FIGURE 3 OMITTED]

RESULTS

The electrical conductivity values reported in this article were intentionally low in favor of using standard fillers, which are relatively simple to compound into the material. For higher conductivities, the foaming strategy could also be used with metal-coated CFs and carbon nanotubes. The filler concentrations used in this work were selected based on compression molding trials previously reported [6]. In that earlier work, radial flow was used to orient the fibers to achieve reasonably representative samples as would be produced on an injection-molding machine, albeit at low shear.

Microstructure of the Foamed Composite

The nominal, maximum, and minimum cell sizes observed among the different foamed composites are tabulated in Table 2. According to the measured apparent densities included in the table, the resulting void fraction in the foamed composites was relatively consistent for the different samples at 18%. The nominal cell size varied between 20 and 45 [micro]m, where the smallest and largest cell size was observed for Runs 6 and 7, respectively. The relatively small cell size found in all these foams was largely attributed to the high viscosity of the composite [6]; though, the CB may have also partially contributed as nucleating sites during foaming. The cell size distributions of foamed composites are presented in Fig. 4 indicating a nonuniform cell structure with a broad range of sizes. As seen in the figure, the samples from Run 6 featured less than 3% of the cells present were larger than 100 [micro]m, the approximate length of the fibers, while Run 7 had 20% of its cells larger than 100 [micro]m.

The flow field within the mold cavity can significantly elongate the shape of a bubble depending on the extent of shear present during foam growth [14]. Similar to cell size, the shape of the foam cells was considered to have a possible influence on the path of electrons passing through the formed composite part. A set of micrographs are presented in Fig. 5 showing the LT plane of the foamed samples for Runs 6 and 7 which correspondingly demonstrated the highest and lowest conductivities, respectively. It can be seen for the Run 7 sample that the bubbles present were highly elongated with their major axis more closely orientated in the L direction. The orientation of their major axes was measured and it was determined that the bubbles were oriented with an average angle of 20[degrees] and 31[degrees] out of the LW plane for Runs 7 and 6, respectively. Cumulative distribution of the bubble elongation in the LT plane, defined as the ratio of the major axis to the minor axis of a bubble, is shown in Fig. 6. As seen it appeared that Run 7 had more elongated bubbles than the other foam runs. Therefore, the samples from Run 7 had the largest cell size in the LW plane and these bubbles were sheared along the flow and less bubble expansion took place toward the T direction resulting in lower fiber reorientation.

Fiber Characterization

The dispersion of fibers within a polymer melt is often a complex task, as sufficient mixing must be achieved to separate the individual fibers from their as-supplied bundles and then homogeneously distributed throughout the polymer matrix. As summarized in Table 2, the nominal fiber length was around 100 [micro]m after processing with little variance in the fiber dimension noted between the different samples, regardless of whether they were processed with the blowing agent. From an original fiber length of 5 mm within the bundles, the resulting separated fibers were found to exhibit substantial breakage due to processing; fracturing was attributable to the original size and mechanical strength of the fibers, the concentration of the fibers, the viscosity of the matrix, and the level of shear applied [15, 16]. These smaller fibers found in the compounded samples would have increased the needed filler concentration to establish a percolating network structure within the molded sample [3], but they were intentionally produced in this manner under the assumption that fibers with similar dimensions to the foam bubbles would result in more noticeable changes in the final orientation of the filler. Future studies will test the validity of the correlation between fiber length and bubble size on the resulting degree of reorientation.

[FIGURE 4 OMITTED]

[FIGURE 5 OMITTED]

[FIGURE 6 OMITTED]

The orientation of the CFs in the core and skin regions of the molded specimens was quantified by the orientation factors ([f.sub.L], [f.sub.W], and [f.sub.T]) in Table 3. The core measurements were taken from 0 to 0.5 mm away from the mid-plane (through-plane axis) while the skin measurements were made at distance of 1.5 mm from the mid-plane of the specimen. Differences between the skin and core in terms of fiber morphology are commonly reported in the literature for injection molded parts [17, 18] with the majority of fibers highly aligned in the longitudinal (L) direction within the skin, and randomly within the LW plane for the core. According to the table, the resulting orientation factors within the skin and core regions of the nonfoamed composites were found to be in good agreement with the findings of those previous works. Such preferential alignment of fibers in the skin resulted from the streamlines of an advancing flow front (i.e. fountain flow) as well as the higher shear stress applied to the fibers close to the mold wall [19, 20]. Orientation of fibers for the W direction in the core was attributed to the expanding flow leaving the gate to fill the width of the mold cavity. The flat velocity profile, resulting from the non-Newtonian rheology of the melt, preserved that transverse fiber orientation as the melt advanced to fill the whole mold [18]. In addition, the greater orientation of fibers in the W direction could have occurred during the packing stage [21]. Of particular interest to this work were the results of fiber orientation in the T direction (through-plane). As indicated in Table 3, generally both in the core and the skin, [f.sub.T] was much smaller than [f.sub.L] and [f.sub.W] for the nonfoamed samples, confirming our expectation that the fibers become preferentially oriented in the flow direction for both injection flow rates used in these trials. This can be visually observed in Fig. 7 where most of the fibers were seen lying in the LW plane. The ratio of [f.sub.T]:[f.sub.L]:[f.sub.W] is consistent with the injection molded samples reported in the literature [22, 23].

[FIGURE 7 OMITTED]

Comparing [f.sub.T] values of the foamed and nonfoamed composites in the core from Table 3, it can be seen that bubble growth influenced out-of-plane rotation of the fibers for all tested processing conditions. The most notable increase for [f.sub.T] was by threefold corresponding to the conditions for Run 6, which resulted in a value that was now comparable to [f.sub.W] and [f.sub.L]. In the skin, the foam composites still had significantly larger [f.sub.T] values than the nonfoamed composites; though this difference was much smaller than the difference found in the core. To examine these findings in more detail, the cumulative distribution of [f.sub.T] was presented in Fig. 8 for all eight runs. According to the graph, the majority of all fibers (~80%) in the nonfoamed samples were oriented in-plane and [f.sub.T] never exceeded 0.5. Conversely, the majority of the foamed samples showed an incremental improvement in the quantity of fibers now oriented out-of-plane. For the foamed sample produced from Run 6 as little as 30% of the fibers remained oriented in-plane with the vast majority oriented out-of-plane with [f.sub.T] values as great as 0.9. Finally as seen in Table 3, contrary to the nonfoamed composites, [f.sub.T] was higher in value than [f.sub.W] for the foamed composites in the core indicating preferential orientation in the L direction rather than W direction. The lack of a packing stage or bubble growth may have created the observed preferred fiber orientation.

Electrical Conductivity

The conductivity values for both foamed and nonfoamed COC composites are reported in Table 4. The magnitude of the deviations stated was reasonable for polymer composites. Because of the low aspect ratio CFs used in this work, the electrical conductivities of the samples were relatively low, in the order of [10.sup.-5] S/cm. First examining the through-plane values, higher measured conductivity was generally found for the foamed composites compared to the nonfoamed counterparts when produced at low injection flow rate. The greatest improvement in through-plane conductivity was found for Run 6, which previously noted corresponds with the highest degree of fiber reorientation found (i.e. [f.sub.T] = 0.21 in Table 3) and the finest cell morphology (referring to Table 2 and Fig. 4). For the other foam samples, Runs 5 and 8 yielded conductivities similar in magnitude to their nonfoamed counterparts. Considering the volume exclusion effect of the gas phase in the foamed samples, lowering the volume fraction of carbon filler in the molded sample, the through-plane electrical conductivity values determined for these three samples indicate a marked improvement in electron transport through the composite material can be achieved with the right foaming conditions. On the other hand, Run 7 demonstrated the lowest conductivity among the foamed samples, which was somewhat unexpected since according to Fig. 8 it possessed a larger fraction of fibers orientated out-of-plane compared to Runs 5 and 8. However, as indicated earlier, the samples from Run 7 were found to have larger bubbles than the other foams and those bubbles observed in the LT plane were more elongated. Since all of the foams in these trials exhibited the same degree of porosity (based on part density), the notion of a volume exclusion effect does not appear to adequately explain the low conductive value for Run 7. It would appear from these results that the foam morphology plays a significant role in achieving high conductivity values, as opposed to our previous belief that maximizing out-of-plane fiber orientation was the only important factor.

[FIGURE 8 OMITTED]

The last two columns in Table 4 indicate the anisotropy in the conductivity for the composites. In general, the conductivity through the materials was higher in the main direction of the flow (L) followed by the transverse direction of the flow (W) and finally perpendicular to the main flow direction (T). This is in agreement with the fiber orientation results in Table 3. Looking at the individual [[sigma].sub.L]/[[sigma].sub.W] and [[sigma].sub.L]/[[sigma].sub.T] average values for the foam and nonfoam runs revealed that the L/W anisotropy was relatively low and comparable between the two material systems, and yet the L/T anisotropy was found to be much larger and significantly more for the foamed samples. On a basis of similar average conductivity as displayed in Table 4, Runs 4 and 6 can be compared to show the differences in anisotropy between the nonfoam and foam composites--the results indicate that the foam composite has less anisotropy, with lower [[sigma].sub.L]/[[sigma].sub.T] and [[sigma].sub.L]/[[sigma].sub.W] ratios.

Analysis of the DOE Factors

The results of the factorial analysis including the effects of CBA and processing conditions on fiber orientation and electrical conductivity are shown in Table 5. Since a half fraction factorial design was used in this study, both main effects and two-way interactions can be studied. Considering the responses elicited by the factors on the through-plane (T) properties as our main interest, it was seen that among the main effects that CBA had the largest influence on through-plane fiber orientation ([f.sub.T]) while melt temperature was the dominant factor on through-plane conductivity. On the other hand, an increase in CBA decreased the through-plane conductivity, which can be attributed as a volume exclusion effect. Therefore, improvement of the through-plane fiber orientation by foaming was not necessarily accompanied by an improvement in through-plane conductivity. Morphological properties such as cell size and shape were also important in improving the through-plane conductivity by foaming, as previously stated. Melt temperature was another important main factor in fiber orientation. An increase in melt temperature resulted in an increase in [f.sub.T]. Conversely, increasing injection flow rate or mold wall temperature reduced [f.sub.T].

For L and W directions, analysis of the main effects showed that an increase in CBA content and processing conditions corresponded to an increase in [[sigma].sub.L] and [[sigma].sub.W] and that melt temperature was the dominant factor in changing conductivity along these directions.

Generally seen in the table, the magnitude of influence for two-way interactions were not dramatically lower than main effects, meaning that they played an important role on fiber orientation and conductivity. The interaction between injection speed and CBA content confounded with melt and mold temperature interaction had the largest (negative) effect on the electrical conductivity for all directions, meaning that the high injection flow rate in the presence of CBA (foaming) diminished the electrical conductivity the most, which is in agreement with our observations made in the previous sections.

Examining the Blowing Agent

It was considered early in this work that the blowing agent residuals could contribute to the conduction of electrons through the material. To evaluate this concern, the same compounds used for injection molding, both with and without IM2240, were prepared in a batch mixer and allowed to fully degas before compression molding into test plaques. Geometrical averaged conductivities of 0.14 [+ or -] 0.02 and 0.22 [+ or -] 0.04 S/cm were obtained for the compound without and with IM2240, respectively. Based on these values, it was concluded that the organic residuals of the blowing agent had no significant impact on the observed conductivities in this work.

In the trials discussed so far, the degree of foaming has not been properly examined. Using the findings from the DOE trials to identify a suitable processing condition, the effects of the blowing agent were further investigated. One nonfoam (A) and two foam composites with different levels of foaming (B and C) were produced as stated in Table 6. For these samples, a PHT powder (containing 99% active ingredient) at 2 wt% had been used instead of the masterbatch to act as the blowing agent. The content of powder CBA used had the equivalent concentration of active ingredient to 10 wt% PHT masterbatch. The pure PHT allowed the addition of higher CBA content with much less volumetric effect on the composite compared to the IM2240 concentrate, since no carrier resin and much less inorganic residue were being incorporated; lower volumetric disturbance was important to maximize the volumetric content of the carbon fillers in the matrix. As seen in Table 7, through-plane and LWT averaged electrical conductivity of the foam composites (B and C) were higher than the nonfoam composite. With a higher degree of foaming (i.e. comparing the densities of B and C composites) it was found that electrical conductivity improved in the T and W directions but became lower in the L direction. Conductivity in the T direction improved by more than one order of magnitude from sample A to sample C similar to improvement observed in the factorial design experiments where Run 6 had through-plane conductivity by more than one order of magnitude higher than the average through-plane conductivity of the nonfoamed composites. The value of [[sigma].sub.L]/[[sigma].sub.T] was now significantly decreased (by about three orders of magnitude for foam composite (C)) compared to nonfoam composite (A) without diminishing the average electrical conductivity. The orientation factor of fibers for these foam composites are presented in Table 8. As seen, there was an improvement of through-plane conductivity by foaming denoted by the increase of [f.sub.T]. From optical microscopy it was found that composite (C) had a thicker core than composite (B), namely 1.9 and 1.4 mm, respectively. It would appear that the negative effect of reducing the volume fraction of the conductive material within the molded part by foaming was minor in comparison to the gain in electron pathways obtained through fiber reorientation, in regards to the property of through-plane electrical conductivity. Micrographs in Fig. 9 demonstrate that many more fibers were oriented in the T direction for the foamed composites compared with the nonfoam composite. Despite introducing nonconductive bubbles to the composite, the average conductivity of the foamed composites was higher than the nonfoamed composite. This could be due to the fact that bubble growth causes the conductive fillers to become more concentrated in the remaining polymer. Finally, it is worth mentioning that the magnitude of conductivity values for sample C are close to those of Run 6 from the factorial experiments (Table 1) which had been produced under similar processing conditions. This suggests repeatability of the results.

DISCUSSION

Improvement of fiber orientation in the through-plane direction is of great importance in improving through-plane electrical conductivity. As observed in our results here and also as reported by others [7, 8], through-plane conductivity of injection molded composites is 2-3 orders of magnitude lower than in-plane conductivity. This major drawback hinders use of conductive polymer composites in applications where conductivity is required to be greater in the through-plane direction. Based on the optimal conditions from the DOE trials, we surmise that the mechanism for fiber reorientation required bubble growth in the presence of shear. A low injection flow rate condition likely resulted in foam flow during mold filling (i.e. bubbles were nucleated and grew as the material filled the cavity) [12]. It is thought that the driver for fiber reorientation was the shear forces acting upon a fiber once being pushed out-of-plane by adjacent bubble growth. Future studies are needed to show bubble growth in relation to the flow field. The benefit of higher melt temperature, as indicated by the DOE analysis, was the reduction in the melt viscosity which lowered the resistance on the fiber to rotate in the flow field. The CF concentration in this study was held in the vicinity of the percolation threshold (i.e. 10 vol% for an average length of ~100 [micro]m); however, had the CF concentrations been substantially higher then fiber reorientation by bubble growth may not have significantly occurred. Higher fiber content would lead to greater viscosity of the melt and increased fiber-fiber interactions, both foreseeably acting as obstacles to fiber rotation. Using fiber-particulate hybrid systems has the advantage of avoiding the necessity for a fully percolated network of fibers.

[FIGURE 9 OMITTED]

The electrical conductivity property of a foamed sample was found to be dependent not only on fiber orientation, but also on porosity, cell size, and cell shape. Smaller, more uniformly shaped bubbles corresponded with higher conductivity. The potential negative influence resulting from the volume exclusion effect of the gas phase was never observed in this work, rather there appeared to be a benefit in concentrating the carbon filler in the polymer pathways between bubbles which led to higher conductivities. The presence of larger and more elongated bubbles along the L direction may have created a torturous path for the electron transport in the through-plane direction.

Besides improving through-plane conductivity, foaming offer many advantages and disadvantages which must be considered by the user [11, 24]. For filled composites like our material, the plasticizing effect of dissolved gases generated by blowing agents will certainly be viewed as a positive during processing, though the reduction of mechanical properties will need to be considered in material selection [12].

CONCLUSIONS

The main goal of this research was to investigate whether foaming could sufficiently disrupt the preferential flow induced alignment of fibers during mold filling and increase the frequency of their orientation in a more favorable direction to improve through-plane conductivity. It was found that foaming at low injection speed and high melt temperature could enhance the through-plane fiber orientation factor by more than two folds, and consequently through-plane electrical conductivity by more than one order of magnitude. The concurrence of the melt flow and bubble growth is considered to be the key mechanism for fiber reorientation while the cell size and shape play an important role so as to not disrupt the conductive paths spanning the bulk of the composite.

ACKNOWLEDGMENTS

The authors wish to thank Asbury Carbons, Dempsey International, Lydall, and Ticona for their donation of materials and helpful comments. In addition, thanks to the assistance of Branden Job and Brandon Shiplo for their assistance in some of the measurements of the samples.

REFERENCES

1. J.-C. Huang, Adv. Poiym. Technol., 21, 299 (2002).

2. W. Di, G. Zhang, Z.-D. Zhao, and Y. Peng, Polym. Int., 53, 449 (2004).

3. D.M. Bigg, Polym. Eng. Sci., 19, 1188 (1979).

4. V.S. Mironov, M. Park, C. Choe, J. Kim, S. Lim, and H. Ko, J. Appl. Polym. Sci., 84, 2040 (2002).

5. F. Mighri, M.A. Huneault, and M.F. Champagne, Polym. Eng. Sci., 44, 1755 (2004).

6. G.H. Motlagh, A.N. Hrymak, and M.R. Thompson, J. Polym. Sci.: Part B, 45, 1808 (2007).

7. M. Weber and M.R. Kamal, Polym. Compos., 18, 726 (1997).

8. A. Dani and A.A. Ogale, Compos. Sci. Technol., 56, 911 (1996).

9. M.T. Kortschot and T.R. Woodhams, Polym. Compos., 9, 60 (1988).

10. M.G. Wilson, SPE J., 27(6), 35 (1971).

11. J.L. Throne, J. Cell. Plast., 12, 161 (1976).

12. M.R. Thompson, X. Qin, G. Zhang, and A.N. Hrymak. J. Appl. Polym. Sci., 102, 4696 (2006).

13. Y. Yang, M.C. Gupta, K.L. Dudley, and R.W. Lawrence, Adv. Mater., 17, 1999 (2005).

14. C.A. Villamizar and C.D. Han, Polym. Eng, Sci., 18, 699 (1978).

15. R. von Turvovich and L. Erwin, Polym. Eng. Sci., 23B, 743 (1983).

16. O.L. Forgacs and S.G. Mason, J. Colloid Sci., 14, 457 (1959).

17. A.M. Brito, A.M. Cunha, A.S. Pouzada, and R.J. Crawford, J. Int. Polym. Proc., 6, 370 (1991).

18. T.D. Papathanasiou and D.C. Guell. Eds., Flow Induced Alignment in Composite Materials, Chapter 4, Woodhead Publishing Limited, Cambridge, England (1997).

19. R.S. Bay and C.L. Tucker III, Polym. Compos., 13, 317 (1992).

20. R.S. Bay and C.L. Tucker III, Polym. Compos., 13, 332 (1992).

21. J.C. Malzahn and J.M. Schultz, Compos. Sci. Technol., 25, 187 (1986).

22. J.S. Cintra and C.L. Tucker III, J. Rheol., 39, 1095 (1995).

23. A. Larsen, Polym. Eng. Sci., 21, 51 (2000).

24. S.T. Lee, C.B. Park, N.S. Ramesh, Polymeric Foams: Science and Technology, CRC Press LLC, Boca Raton, Florida (2007).

G.H. Motlagh, A.N. Hrymak, M.R. Thompson

MMRI/CAPPA-D, Department of Chemical Engineering, McMaster University, Hamilton, Ontario, L8S 4L7, Canada

Correspondence to: M.R. Thompson; e-mail: [e-mail]

Contract grant sponsors: Ontario Centres of Excellence Emerging Materials and Knowledge Program, Iranian Ministry of Science, Research and Technology.

TABLE 1. Fractional factorial design to produce injection molded
samples.

     CBA, IM2240  Injection         Melt temperature  Mold temperature
Run  (wt%)        flow rate (cc/s)  ([degrees]C)      ([degrees]C)

1     0            10               260                80
2     0            10               300               120
3     0           100               260               120
4     0           100               300                80
5     5            10               260               120
6     5            10               300                80
7     5           100               260                80
8     5           100               300               120

TABLE 2. Properties of the foamed composites prepared according to the
fractional factorial design.

                                              Cell size, LW plane
                Part density    Fiber length  ([micro]m)
           Run  (g/[cm.sup.3])  ([micro]m)    Nominal  Max.  Min.

Nonfoamed  1    1.11            101
           2    1.11             98
           3    1.11            102
           4    1.11            102
Foamed     5    0.90             94            33      527    6.5
           6    0.92             98            23      144    6.5
           7    0.91            106            44      441   11.8
           8    0.91            105            22      312    6.5

Average standard deviation for reported density, fiber length, and cell
size values were 1.5, and 11%, respectively.

TABLE 3. Orientation factors in three principal directions (L, W, and T)
for the factorial design injection molded samples.

                Fiber length              Core
           Run  ([micro]m)    [f.sub.T]  [f.sub.W]  [f.sub.L]

Nonfoamed  1    101           0.079      0.41       0.51
           2     98           0.076      0.43       0.49
           3    102           0.052      0.49       0.46
           4    102           0.065      0.51       0.42
Foamed     5     94           0.099      0.39       0.51
           6     98           0.210      0.36       0.43
           7    106           0.109      0.36       0.53
           8    105           0.101      0.32       0.58

                             Skin
           Run  [f.sub.T]  [f.sub.W]  [f.sub.L]

Nonfoamed  1    0.059      0.33       0.61
           2    0.052      0.33       0.62
           3
           4
Foamed     5
           6    0.077      0.34       0.58
           7
           8    0.074      0.30       0.62

Average standard deviation for reported [f.sub.L], [f.sub.W], and
[f.sub.T] values were 11, 13, and 13%, respectively.

TABLE 4. Electrical conductivity results (a) for factorial design
injection molded composites (10 vol% CF and 2 vol% CB) in the L, W, and
T directions.

                            S/cm x [10.sup.5]
         Run  [[sigma].sub.T]  [[sigma].sub.W]  [[sigma].sub.L]

Nonfoam  1     0.12                3.0              6.7
         2     1.27               12.0             43
         3     1.46            1,300              822
         4     4.9             1,020            2,680
Foam     5     0.90              913            1,300
         6    31.4             2,030            2,280
         7     0.01               14.4             43
         8     0.66              447            1,970

              S/cm x [10.sup.5]                [[sigma].sub.L]/
         Run  [[sigma].sub.average (LWT)] (b)  [[sigma].sub.W]

Nonfoam  1      1.32                           2.2
         2      8.65                           3.5
         3    116                              0.7
         4    238                              3.0
Foam     5    102                              2.1
         6    525                              1.1
         7      2.02                           3.0
         8     83.4                            6.4

         Run  [[sigma].sub.L]/[[sigma].sub.T]

Nonfoam  1       55
         2      350
         3      273
         4      343
Foam     5    2,108
         6       87
         7    1,655
         8    6,876

Reported conductivity values have been multiplied by [10.sup.5].
(a) Average standard deviations for the data shown were 33, 22, and 36%
for L, W, and T directions, respectively.
(b) All average values are geometric average.

TABLE 5. Results of the factorial analysis showing the effects of
factors on fiber orientation factor and electrical conductivity.

Factor [down arrow]\                     Effect
Variable [right arrow]  [f.sub.T]  [f.sub.W]  [f.sub.L]  [[sigma].sub.T]

CBA (IM2240), A          0.062     -0.103      0.043     -0.25
Injection flow rate, B  -0.034      0.023      0.013     -0.43
Melt temperature, C      0.028     -0.008     -0.023      1.17
Mold temperature, D     -0.034     -0.003      0.038      0.25
AB + CD                 -0.015     -0.058      0.073     -1.39
AC + BD                  0.023     -0.028      0.008      0.52
AD + BC                 -0.026     -0.003      0.028     -0.12

Factor [down arrow]\                 Effect
Variable [right arrow]  [[sigma].sub.W]  [[sigma].sub.L]

CBA (IM2240), A          0.6              0.65
Injection flow rate, B   0.52             0.58
Melt temperature, C      0.59             0.81
Mold temperature, D      0.46             0.43
AB + CD                 -1.76            -1.36
AC + BD                  0.34             0.15
AD + BC                  0.11             0.29

TABLE 6. Injection molding conditions for injection molded composites
(10 vol% CF and 2 vol% CB) with CBA powder.

     Injection  Melt          Mold          CBA     Shot  Injection
     flow rate  temperature   temperature   powder  size  pressure
Run  (cc/s)     ([degrees]C)  ([degrees]C)  (wt%)   (%)   (MPa)

A    4          300           80            0       100   370
B                                           2        86   285
C                                           2        78   280

TABLE 7. Electrical conductivity of the composites (10 vol% CF and 2
vol% CB) with CBA powder.

     Density                  S/cm x [10.sup.5]
Run  (g/cc)   [[sigma].sub.T]  [[sigma].sub.W]  [[sigma].sub.L]

A    1.12      1.60            2,370            2,230
B    1.01     26.0             2,270            3,820
C    0.91     86.1             2,660            1,270

     S/cm x [10.sup.5]      [[sigma].sub.L]/  [[sigma].sub.L]/

Run  [[sigma].sub.average]  [[sigma].sub.W]   [[sigma].sub.T]

A    203                    0.94              1393
B    609                    1.68               147
C    662                    0.48                15

Reported conductivity values have been multiplied by [10.sup.5].
The average standard deviations for the reported conductivity data shown
were 24, 10, and 10% for L, W, and T directions, respectively.

TABLE 8. Orientation factors for composites with CBA powder.

                          Core
           Run  [f.sub.T]  [f.sub.W]  [f.sub.L]

Nonfoamed  A    0.06       0.51       0.43
Foamed     B    0.127      0.4        0.47
           C    0.127      0.34       0.54

                         Skin
           Run  [f.sub.T]  [f.sub.L]  [f.sub.W]

Nonfoamed  A
Foamed     B    0.081      0.22       0.7
           C    0.083      0.21       0.7