Composites Science and Technology 73 (2012) 27–33

Contents lists available at SciVerse ScienceDirect

Composites Science and Technology

journal homepage: www.elsevier .com/ locate/compsci tech

Characterization of non-covalently, non-specifically functionalized multi-wallcarbon nanotubes and their melt compounded composites with an ethylene–octenecopolymer

Osayuki Osazuwa a, Kyle Petrie a, Marianna Kontopoulou a,⇑, Peng Xiang b, Zhibin Ye b, Aristides Docoslis a

a Department of Chemical Engineering, Queen’s University, Kingston, Ontario, Canada K7L 3N6b School of Engineering, Laurentian University, Sudbury, Ontario, Canada P3E 2C6

a r t i c l e i n f o

Article history:Received 17 April 2012Received in revised form 17 July 2012Accepted 23 August 2012Available online 1 September 2012

Keywords:A. Carbon nanotubesA. NanocompositesB. Electrical propertiesB. Mechanical propertiesE. Melt compounding

0266-3538/$ - see front matter � 2012 Elsevier Ltd. Ahttp://dx.doi.org/10.1016/j.compscitech.2012.08.015

⇑ Corresponding author. Tel.: +1 613 533 3079; faxE-mail address: marianna.kontopoulou@chee.quee

a b s t r a c t

Multi-wall carbon nanotubes (MWCNTs) were functionalized with a hyperbranched polyethylene (HBPE)using a non-covalent, non-specific functionalization approach. The adsorption behavior of HBPE onMWCNTs was characterized by means of an adsorption isotherm. HBPE adsorption reached a plateauvalue of 0.3:1.0 (w/w) HBPE:MWCNT, corresponding to a surface coverage of approximately 30%. Thefunctionalized MWCNTs were better dispersed in tetrahydrofuran (THF), forming smaller aggregatescompared to their unmodified counterparts. Pristine and HBPE-functionalized MWCNTs were melt com-pounded with a low-viscosity ethylene–octene copolymer (EOC) matrix. Electrical and rheological perco-lation thresholds were observed at nanotube loadings of less than 1 wt%. Functionalization did not affectsignificantly the electrical conductivity and rheological properties. The improved dispersion of the func-tionalized MWCNTs within the EOC matrix resulted in improved ductility over the non-functionalizedcounterpart. This study demonstrates that this method of functionalization results in partial surface cov-erage of the nanotubes, therefore providing an efficient means for achieving good nanotube dispersion,without compromising their surface properties.

� 2012 Elsevier Ltd. All rights reserved.

1. Introduction

Polymer/carbon nanotube (CNT) composites have been widelystudied, because of their superior mechanical, thermal and electri-cal properties [1–7]. The preparation of polymer/carbon nanotubecomposites has been an area of active scientific research, since theexceptional properties of CNTs may extend the end-use applica-tions of polymeric materials. Owing to their low density, improvedmechanical properties and electrical conductivity, these compos-ites can find applications in structural materials, antistatic films,electromagnetic interference (EMI) shielding, and coatings for elec-trostatic painting, hydrogen storage media and nanometer-sizedsemiconductor devices, probes and interconnects [8], automotiveparts, reinforced aerospace materials, and sporting goods [9].

The use of CNTs in polymer composites has been greatly hin-dered due to their incompatibility with polymers and conse-quently, their poor dispersion in polymer matrices, especially inmelt compounding operations. CNTs tend to exist as bundles ofsingle wall carbon nanotubes (SWCNTs), or aggregates of multi-wall carbon nanotubes (MWCNTs) held together by van der Waals

ll rights reserved.

: +1 613 533 6637.nsu.ca (M. Kontopoulou).

forces and p–p interactions, which make their separation and suc-cessful dispersion into polymers a very difficult task. Other factorsinfluencing the dispersion of CNT include the viscosity of the poly-mer matrix, the melt compounding procedure, etc. [10,11].

Melt compounded thermoplastic composites containingMWCNTs have the potential of producing composite systems, withelectrical conductivity higher than 10�4 S/m [12], at relatively lownanotube loadings, suitable for applications such as electrostaticpainting, electromagnetic interference (EMI) shielding, electro-static discharge and conductive coatings, and other injectionmolded parts. However, in order to broaden the field of applicationof these composites to include large structural components, themechanical properties must also remain at an acceptable level,and not be compromised by excessive aggregation.

A number of surface functionalization methods have beendeveloped in order to either render MWCNTs dispersible in sol-vents, or improve their dispersibility in polymers. Local strain incarbon nanotubes, arising from pyramidalization and misalign-ment of the p-orbitals of the sp2-hybridized carbon atoms, makesnanotubes more reactive than a flat graphene sheet, thus moreamenable to chemical functionalization [13]. Covalent functionali-zation of nanotubes can improve the properties of the resultingcomposites through better nanotube dispersion in the matrix andenhanced nanotube/polymer interfacial binding [14]. Covalent

28 O. Osazuwa et al. / Composites Science and Technology 73 (2012) 27–33

functionalization can, however, disrupt the extended p conjuga-tion in CNTs, which may introduce structural defects resulting inprofound decline in the electrical properties of the nanotubes. Analternate way of tuning the surface properties of nanotubes with-out introducing irreversible changes is through non-covalent func-tionalization [15–17]. Recently non-covalent, non-specificfunctionalization using a hyperbranched polyethylene (HBPE) thatestablishes only CH–p interactions with the nanotubes was em-ployed successfully to improve the dispersion of MWCNTs in or-ganic solvents [18] and in polyolefin-based composites preparedby solution blending [19].

While our previous work [19] has demonstrated the efficacy ofthis approach in composites prepared by solution blending, the fo-cus of the present research is to develop melt compounded poly-olefin/MWCNT composites with low electrical percolationthresholds while at the same time maintaining good mechanicalproperties by avoiding excessive aggregation through non-covalent, non-specific nanotube functionalization using HBPE.Composites of MWCNTs with an ethylene–octene copolymer wereprepared by melt compounding. The morphology, surface proper-ties, electrical conductivity, rheological and mechanical propertiesof the composites containing pristine nanotubes and HBPE-modi-fied nanotubes are studied in order to understand the effects ofnon-covalent, non-specific functionalization of the MWCNTs withHBPE on the composite properties.

2. Materials and methods

2.1. Materials and characterization

MWCNTs (purity > 95%, diameter 30 ± 15 nm and length 1–5 lm) from Nanolab Inc. (Massachusetts, USA) and (purity > 90%,diameter 10–15 nm and length 0.1–10 lm) from Sigma Aldrich(Missouri, USA) were used as received. The specific surface area(SSA) of the nanotubes was 300 m2/g and 224 m2/g, respectively,determined by Brunauer–Emmett–Teller (BET) characterization.MWCNT samples weighing 0.5–1.0 g were first degassed at110 �C for 24 h and then subjected to a multipoint BET physisorp-tion analysis (Autosorb-1 Quantachrome, USA) for nitrogen relativevapor pressures in the range 0.1–0.3 at 77 K. Based on the shape ofthe N2 adsorption–desorption isotherms (type II according to IU-PAC classification [20,21]) in the whole relative pressure range,the estimated SSA corresponds to external surface area. The aver-age pore size distribution of the MWCNTs as determined by theN2 sorption experiments was 1.48 nm, whereas the average poresize determined through the BJH method was 2.3 nm.

HBPE was synthesized from ethylene using a chain walking Pd–diimine catalyst, as described in detail by Ye and Li [22]. HBPE hasa complex and irregular dendritic structure, containing a largenumber of branches of various lengths (from methyl to hexyl andhigher) and, more importantly, abundant branch-on-branch struc-tures [19]. The polyolefin matrix was a poly(ethylene-co-octene)(EOC), trade name Engage 8130, density 0.864 g/cm3, MFI 13 g/10 min at 190 �C, with copolymer content of 42 wt%, obtained fromDow Chemical (Michigan, USA). The melting and crystallizationtemperatures of this polymer, as determined by differential scan-ning calorimetry (DSC) are 56 �C and 38 �C, respectively and thedegree of crystallinity is 13%.

2.2. Functionalization of MWCNTs

Mixtures of HBPE and MWCNTs in tetrahydrofuran (THF) withmass ratio between 0.0 and 3.0 were prepared by adding HBPE intodispersions containing 2.0 mg MWCNT/mL of THF. The resultingmixtures were sonicated for 1 h, and then stirred overnight. The

supernatant solutions were vacuum filtered drop-wise throughTeflon membrane with pore size of 0.22 lm. After being washedtwice with equal volumes of THF (6 mL) the filters were dried ina vacuum oven overnight at room temperature [19]. Thermo gravi-metric analyses were then carried out on dried samples using aQ500 TGA by TA Instruments (Delaware, USA). Data from TGAanalysis was used to construct an adsorption isotherm [19].

2.3. Melt compounding

EOC composites containing pristine and HBPE-functionalizedMWCNTs at contents ranging from 0.1 wt% to 5.0 wt% were com-pounded using a DSM Research 5 mL Micro-Compounder (DSM Re-solve, Geleen, Netherlands) at a temperature of 150 �C, screwspeed of 90 rpm and mixing time of 10 min.

2.4. Morphological characterization

HBPE/MWCNT suspensions in THF (2 mg/mL) were depositedonto carbon and Cu-Formvar/Carbon film and washed with THFto ensure monolayer coating. Following solvent evaporation, thesamples were placed into a Hitachi-7000 Transmission ElectronMicroscope (TEM) (Hitachi HTA, Schaumburg, Illinois, USA) andimages were captured at a magnification of 40,000�. TEM imageswere also obtained for ultra-thin films of the polymer compositesprepared using a Leica ultra-microtome. Images of the compositeswere obtained using a FEI Tecnai 20 instrument (FEI Co., Eindho-ven, Netherlands). Composite melts were observed using an Olym-pus BX 51 optical microscope (Tokyo, Japan). Composite films wereloaded on a Linkam SCC 450 Hot Stage (Surrey, UK) at 150 �C andpressed to a thickness of 20 lm at 150 �C. Images were recordedusing transmitted light.

2.5. Dynamic light scattering (DLS)

The size distribution of MWCNTs (pristine and functionalized)suspended in THF was analyzed by means of dynamic light scatter-ing (DLS) (Zetasizer Nano ZS, Malvern Instruments Ltd., Worcester-shire, UK) equipped with a monochromatic coherent 4 mW HeliumNeon laser (k = 633 nm) as the light source and Non-Invasive Back-Scattering (NIBS) technology (ALV GmbH, Germany). Very dilutesuspensions containing 5.0 � 10�5 wt% MWCNT in THF were pre-pared by sonicating the suspension for 2 min and were allowedto settle overnight. The supernatant solution was used for the sizedistribution analysis at 25 �C. All measurements were performed ina glass cuvette with square aperture. Data analysis was done usingthe DTS (Dispersion Technology Software) 5.10, which includescumulant analysis (in accordance to ISO 13321) and the multi-modal size distribution, non-negative least square (NNLS) algo-rithm [21] to obtain the mean particle diameter andpolydispersity of MWCNT in THF.

2.6. Electrical conductivity

Volume resistivity was measured under DC current at roomtemperature. Samples were prepared by compression moldingthe melt-compounded composites in a Carver press at 140 �C anda force of 1400 N to get a thin film of 0.6 mm. The thin compositefilm with a diameter of 6 cm was put into the measuring chamber(Keithley 8009 Resistivity Test Fixture) of the Keithley 6517B Elec-trometer/High Resistance Meter (Keithley Instruments, Inc., Cleve-land, Ohio, USA) for an electrification time of 1 min.

O. Osazuwa et al. / Composites Science and Technology 73 (2012) 27–33 29

2.7. Rheological measurements

Rheological characterization was carried out using a ViscoTechoscillatory rheometer by Reologica equipped with 20 mm parallelplate fixtures using a gap of 1.2 mm. Compression molded diskswith a diameter of 20 mm were prepared using the Carver pressas described above. Stress sweep experiments were carried outfrom 1 to 103 Pa at a frequency of 0.1 Hz and temperature of180 �C to identify the limits of linear viscoelasticity. Given the ex-treme strain sensitivity of the samples, strain-controlled experi-ments were carried out at a low strain of 0.2%, within the linearviscoelastic region using a frequency range of 0.01–25.1 Hz at180 �C.

2.8. Mechanical properties

Tensile tests were conducted in accordance with ASTM D 638using test specimens with the dimensions: 7.62 mm � 3.50 mm �1.45 mm. The tests were conducted on an Instron 3369 electrome-chanical test instrument (Instron, Norwood, Massachusetts, USA)with a crosshead speed of 200 mm/min. The average values andstandard deviation from testing six specimens are reported.

3. Results and discussion

3.1. HBPE adsorption on MWCNTs

Fig. 1a shows the adsorption isotherm depicting the actualamount of adsorbed HBPE as a function of the originalHBPE:MWCNT ratio used for functionalization. From Fig. 1a itcan be seen that, after washing with THF to remove all looselybound polymer, only a fraction of the HBPE is adsorbed on theMWCNTs. A plateau appears around an applied HBPE:MWCNTmass ratio of 1, which translates into an adsorbed amount of0.3–0.32 g of HBPE/g of MWCNT. Beyond this mass ratio, no in-crease is observed in the adsorbed HBPE amount. As a result, theratio of 1.0 is used in the preparation of EOC/MWCNT composites.A comparison between the data obtained in the present work usingthe Nanolab (NL) MWCNTs with those obtained in our previouswork [19] using nanotubes obtained from Sigma–Aldrich (SA) re-veals an adsorbed amount between 0.2 and 0.23 g of HBPE/g ofMWCNT for the latter. Fig. 1b shows the adsorption isotherms nor-malized by the specific surface area (SSA) of pristine MWCNTs ac-quired from BET measurements (300 m2/g and 224 m2/g for NL andSA MWCNTs, respectively). It can be noticed that following the nor-malization procedure, a very good agreement between the plateau

Fig. 1. (a) Adsorption isotherms of HBPE on Nanolab (NL) and Sigma–Aldrich (SA) MWCNand (b) adsorption isotherms with retained mass ratio normalized with respect to the spare presented for each mass ratio.

values of the adsorption isotherms exists, at 0.3 g of HBPE/g ofMWCNT. This method of reporting carbon nanotube surface cover-age appears to be particularly useful as it renders the results inde-pendent of nanotube dimensions and source of supply.

To obtain further insight into the degree of carbon nanotubesurface coverage by HBPE, results from the adsorption isothermand BET method were combined with DLS measurements of theHPBE radius. Knowing that HBPE has a highly compact sphericalchain architecture [22] and a ‘‘hydrodynamic’’ diameter in THF of9.5 ± 1.4 nm [19], which is comparable in magnitude to the exter-nal diameter of MWCNTs (30 nm mean value), wrapping, clipping,or entanglement of the polymer around the nanotubes is not prob-able. Furthermore the average pore size of the MWCNT as deter-mined by our BET measurements is 1.47 nm, which is much lessthan the hydrodynamic diameter of HBPE, thus excluding the pos-sibility of entrance and diffusion of HBPE within the pores of theMWCNTs. It is more likely that all of the globular-shaped HBPE isnon-covalently adsorbed on the external MWCNT sidewalls. There-fore, by taking into account experimentally measured quantities,namely adsorbed amount and molecular dimensions of HBPE(see above) along with specific surface area of the carbon nano-tubes (300 m2/g), we can estimate the fractional surface coverageof the nanotubes by HBPE to be approximately 30% (plateau value).In other words, at the plateau phase of adsorption less than onethird of the carbon nanotube surface is masked by HBPE.

3.2. MWCNT dispersion

The effectiveness of the functionalization process on the disper-sion of nanotubes in a solvent is clearly evidenced from the TEMimages of Fig. 2. The modified nanotubes are less aggregated andbetter dispersed in THF than the unmodified samples. This is be-cause the filler–filler interactions, which mainly consist of van derWaals forces between nanotubes, weaken in the presence of theHBPE coating. This results in the reduction of the size of MWCNTaggregates, as also revealed by DLS data, which provide a more rep-resentative picture of the quality of dispersion that has beenachieved. Specifically, the mean particle diameters of nanotubessuspended in THF were found to be equal to 333.4 ± 41.2 nm forthe pristine and 232.9 ± 12.3 nm for the HBPE-modified nanotubes(Fig. 3a). An example of the measured size distributions is shown inFig. 3b. It should be noted that the above values correspond to thepopulation of carbon nanotubes that remained stably suspendedin THF overnight and illustrate the effect of HBPE on the generationof significantly smaller aggregates. DLS studies of the wholeMWCNT suspensions were not possible, as the latter are thermody-

T at room temperature, constructed based on nanotube concentration of 2.0 mg/mLecific surface area of MWCNT, as obtained from BET measurements. Two replicates

Fig. 2. TEM images showing the effect of HBPE on nanotube disentanglement: (a) pristine MWCNT and (b) HBPE-functionalized MWCNT; both samples were prepared fromnanotube dispersions (2 mg/mL) in THF.

Fig. 3. Dynamic light scattering data showing: (a) mean particle size and standard deviation and (b) an example of the observed size distribution of pristine and HBPE-functionalized MWCNT suspensions in THF.

Fig. 4. Optical micrographs of MWCNT/EOC composite containing 1 wt% MWCNT, scale bar is 100 lm: (a) pristine MWCNT/EOC and (b) HBPE-functionalized MWCNT/EOC.

30 O. Osazuwa et al. / Composites Science and Technology 73 (2012) 27–33

namically unstable and tend to aggregate and settle during mea-surements. Fig. 4 shows images of the MWCNT/EOC compositescontaining 1 wt% MWCNTs with sizable aggregates (up to 50 lm),which disappear when HBPE-modified MWCNTs are used. TEMimaging of the composites further confirmed that functionalizedMWCNTs are better dispersed within the EOC matrix. Fig. 5 showsclearly that the composites with modified nanotubes appear lessaggregated and better dispersed.

The improved dispersion can be explained by the more thermo-dynamically favored interactions between the HBPE-coatedMWCNTs and the polyolefin matrix, resulting in stronger interfa-

cial interactions and better stress transfer to the MWCNTs duringcompounding, thus breaking up more efficiently the filler aggre-gates. HBPE would also act to prevent secondary aggregation, giventhat the nanotube interactions become weaker.

3.3. Electrical conductivity

Introduction of carbon nanotubes into a polymer matrix in-creases significantly the conductivity of the composites, as shownin Fig. 6. Addition of carbon nanotubes at concentrations in therange between 1 and 2 wt% resulted in an increase in conductivity

Fig. 5. TEM images of: (a) 3 wt% pristine MWCNT/EOC composite, (b) 3 wt% HBPE-functionalized MWCNT/EOC composite; scale bar is 100 nm.

Fig. 6. Plot of volume resistivity vs. MWCNT loading for pristine and HBPE-functionalized MWCNT composites showing fit using the power law equation.

O. Osazuwa et al. / Composites Science and Technology 73 (2012) 27–33 31

that is higher by almost five orders of magnitude compared to theunfilled polymer. The electrical response of polymer composites isgenerally described by the power law equation shown in Eq. (1)[23,24], which fits data points above the percolation threshold:

r ¼ ð/� /cÞt ð1Þ

where r is the electrical conductivity, / is the nanotube composi-tion, /c is the electrical percolation threshold and t is the criticalexponent. The electrical percolation threshold is defined as the va-lue of the filler content above which the electrical properties in-crease in an exponential manner. As shown in Fig. 6, the electricalconductivity data can be fitted well by Eq. (1). The best fit for theexperimental data yielded /c of 0.91 wt% and t of 2.4 and 2.8 forthe pristine and modified MWCNT composites, respectively. The va-lue of the critical exponent for both composite systems indicatesthe formation of a three-dimensional percolating network at0.91 wt%. The percolation threshold for this polymer is generallylower than other polyolefins in the literature, which are typicallyin the range of 6.0–8.0 wt% [25]. This difference is presumablydue to the relatively low melt viscosity of the EOC matrix used in

this work. Theoretical predictions for the critical exponent rangefrom 1.6 to 2.0 for three-dimensional percolating systems, althoughexperimental values from 0.7 to 3.1 have been reported [26–29].

The maximum conductivity achieved at 5.0 wt% nanotube load-ing is 1.45 � 10�4 S/m for the pristine MWCNT/EOC composite and6.60 � 10�5 S/m for the HBPE-modified composite, making thesecomposites static dissipative materials (10�2–10�9 S/m) [30]. It isvery interesting to note that nanotube functionalization did notimpact negatively the electrical properties of the system. Thiscan be explained on the basis of the estimated fractional carbonnanotube surface coverage by HBPE. Specifically, as mentioned ear-lier, surface coverage calculations show that only one-third of thenanotube surface is covered by HBPE, leaving approximatelytwo-thirds of the nanotube surface available for filler–filler con-tact. This is very important because it shows that better nanotubedispersion in polyolefin matrix is achievable without a concomi-tant compromise in the electrical properties of the resultingcomposites.

3.4. Rheological properties

The presence of a percolation threshold is also clearly shownfrom the rheological characterization of the MWCNT/EOC compos-ites (Fig. 7a–d), which depicts a loss of the Newtonian plateau andpronounced deviation from the terminal flow behavior G0 /x2

above 1 wt% MWCNT and the appearance of a plateau at high con-centrations. The loss tangent, tan d, where d is the phase angle;(Fig. 7c) clearly shows the transition from viscoelastic liquid- to so-lid-like behavior. These observations are indicative of the fillerinteractions leading to the formation of a 3D percolation network,which restrains the long-range motion of the polymer chains [31].The large aggregates present at the higher MWCNT contents areapparently responsible for the higher viscosities compared to theHBPE-modified counterparts, due to an increased hydrodynamiceffect. The transition from liquid- to solid-like behavior is bettervisualized through Fig. 7d, which shows the composition depen-dence of the reduced viscosity, g�r defined as:

g�r ¼g�

g�0ð2Þ

Fig. 7. Frequency dependence of: (a) storage modulus, G0 , (b) complex viscosity, g⁄, and (c) tand at different MWCNT contents for MWCNT/EOC composites, and (d) Plot of (g�r )vs. (/ � /g) and fitting with the Guth model. Inset shows g�r at lower MWCNT content.

Fig. 8. Elongation at break (bars) and secant modulus at 10% strain (line) showingthe effect of nanotube loading and functionalization on the mechanical propertiesof the composites.

32 O. Osazuwa et al. / Composites Science and Technology 73 (2012) 27–33

where g�0 is the zero shear viscosity of the pure polymer and g� isevaluated at a fixed frequency (in this case 0.1 rad s�1). At lownanotube loadings, the MWCNTs act as isolated objects, and the vis-cosity can be modeled following Guth’s, model [32,33]. Above thegeometrical percolation of the nanotubes, the MWCNT 3D networkstructure dominates the viscoelastic response following a typicalpower-law like behavior associated with composites near the per-colation threshold [33,34]. The composition dependence of the re-duced viscosity at low and intermediate nanotube concentrationsmay, therefore, be modeled as:

g�r ¼ 1þ 0:67ða/Þ þ 1:62ða/Þ2 þmð/� /gÞt ð3Þ

where / is the nanotube volume fraction, /g is the geometrical per-colation threshold, a is the aspect ratio of nanotubes and t is thescaling exponent. The linear and quadratic terms result from Guth’smodification of Einstein’s relationship for anisotropic fillers in di-lute solution, while the power-law term models the behavior nearthe percolation threshold [33]. In Fig. 7d the data is representedin terms of the mass fraction, because of the uncertainty in the va-lue of the density of the nanotubes (between 1.4 and 1.9 g/cm3). Afit of the experimental data with Eq. (3) delivered a geometrical per-colation threshold, /g of 0.54 wt% and 0.62 wt% and a scaling expo-nent, t of 2.2 and 2 for pristine and modified MWCNT composites,respectively. These values are lower than the value of 0.91 wt% forthe electrical percolation, but overall functionalization of the nano-tubes shows no significant effect on the rheological percolationthreshold, in agreement with the findings of electrical conductivity.

3.5. Mechanical properties

As shown above, modification of MWCNT with HBPE does notaffect negatively the electrical properties and the percolation

behavior of the composites. In terms of mechanical properties, itis well known that the inclusion of fillers in a polymer matrixhas a reinforcing and stiffening effect, which is counteracted by asignificant loss in ductility. Fig. 8 shows an increase in the secantmodulus (line) from 1.3 MPa for the EOC matrix to 2.0 and2.2 MPa for 3 wt% modified and unmodified MWCNT/EOC compos-ites, respectively, at 10% strain. Fig. 8 also shows a drop in the elon-gation at break (bar) of the EOC matrix upon addition of MWCNT.However it is worth noting that at the same filler loading, the

O. Osazuwa et al. / Composites Science and Technology 73 (2012) 27–33 33

HBPE–MWCNT/EOC composite has higher elongation at break thanits counterpart containing the pristine MWCNTs, as a result of im-proved interfacial adhesion and better dispersion of the modifiednanotubes in the polymer matrix. These results suggest that func-tionalization of the nanofillers reduces the negative impact on duc-tility typically experienced by these composites (the elongation atbreak dropped by only 19% with respect to the neat matrix for themodified, compared to 31% for the unmodified MWCNT compos-ites). This finding may be of significance when other polyolefinmatrices are used, which may suffer from significant loss of ductil-ity in the presence of MWCNTs. We plan to test this hypothesis infuture studies by using this functionalization approach with a vari-ety of polyolefin matrices.

4. Conclusions

In the present work, non-covalent, non-specific functionaliza-tion of MWCNTs with HBPE was used as a means of improvingthe state of dispersion of MWCNTs in an ethylene–octene copoly-mer matrix. DLS characterization and optical microscopy revealedthat the size of MWCNT aggregates is reduced upon functionaliza-tion. TEM images show that functionalized nanotubes are betterdispersed in suspension and also in the polymer matrix. When alow-viscosity polyolefin matrix is used, this functionalization pro-cedure reduces the size and number of MWCNT aggregates in thecomposites, while maintaining a low percolation threshold andnot affecting adversely the electrical conductivity of the composite.In addition to the expected modulus increase, the functionalizedMWCNT/EOC composites maintained better ductility, because ofthe improved dispersion, suggesting that this approach can be uti-lized to obtain composites with good conductivity, without com-promising mechanical properties. The results of this studydemonstrate that physically adsorbed compatibilizing moleculescan provide an easy means for achieving good nanotube dispersion.Moreover, this method of functionalization has a small impact onthe filler’s surface properties by virtue of the fact that physicaladsorption can result only in partial surface coverage.

Acknowledgements

This research was supported with funding from the Natural Sci-ences and Engineering Research Council of Canada (NSERC), E.I.DuPont and the Queen’s Graduate Scholarship program.

References

[1] Pötschke P, Fornes T, Paul D. Rheological behavior of multiwalled carbonnanotube/polycarbonate composites. Polymer 2002;43(11):3247–55.

[2] Potschke P, Abdel-Goad M, Alig I, Dudkin S, Lellinger D. Rheological anddielectrical characterization of melt mixed polycarbonate-multiwalled carbonnanotube composites. Polymer 2004;45(26):8863–70.

[3] Abdel-Goad M, Potschke P. Rheological characterization of melt processedpolycarbonate-multiwalled carbon nanotube composites. J Non Newton FluidMech 2005;128(1):2–6.

[4] Sung Y, Kum C, Lee H, Byon N, Yoon H, Kim W. Dynamic mechanical andmorphological properties of polycarbonate/multi-walled carbon nanotubecomposites. Polymer 2005;46(15):5656–61.

[5] Shaffer MSP. WA. fabrication and characterization of carbon nanotube/poly(vinyl alcohol) composites. Adv Mater 1999;11(11):937–41.

[6] Kumar S, Dang T, Arnold F, Bhattacharyya A, Min B, Zhang X, et al. Synthesis,structure, and properties of PBO/SWNT composites. Macromolecules2002;35(24):9039–43.

[7] Iijima S. Helical microtubules of graphitic carbon. 1991; 354(56).[8] Lau A, Hui D. The revolutionary creation of new advanced materials—carbon

nanotube composites. Compos Part B: Eng 2002;33(4):263–77.[9] Moniruzzaman M, Winey K. Polymer nanocomposites containing carbon

nanotubes. Macromolecules 2006;39(16):5194–205.[10] Socher R, Krause B, Müller M, Boldt R, Pötschke P. The influence of matrix

viscosity on MWCNT dispersion and electrical properties in differentthermoplastic nanocomposites. Polymer 2012;53(2):495–504.

[11] Alig I, Pötschke P, Lellinger D, Skipa T, Pegel S, Kasaliwal G, et al. Establishment,morphology and properties of carbon nanotube networks in polymer melts.Polymer 2012;53(1):4–28.

[12] Rutkofsky M, Banash M, Rajagopal R, Chen J. Using a carbon nanotube additiveto make electrically conductive commercial polymer composites. ZyvexApplication Note 9709 2006.

[13] Niyogi S, Hamon M, Hu H, Zhao B, Bhowmik P, Sen R, et al. Chemistry of single-walled carbon nanotubes. Acc Chem Res 2002;35(12):1105–13.

[14] Koval’chuk A, Shevchenko V, Shchegolikhin A, Nedorezova P, Klyamkina A,Aladyshev A. Effect of carbon nanotube functionalization on the structural andmechanical properties of polypropylene/MWCNT composites. Macromolecules2008;41(20):7536–42.

[15] O’Connell M, Bachilo S, Huffman C, Moore V, Strano MS, Haroz E, et al. Bandgap fluorescence from individual single-walled carbon nanotubes. Science2002;297:593.

[16] Jiang L, Gao L, Sun J. Production of aqueous colloidal dispersions of carbonnanotubes. J Colloid Interface Sci 2003;260(1):89–94.

[17] Tan S, Goak J, Lee N, Kim J, Hong S. Functionalization of multi-walled carbonnanotubes with poly(2-ethyl-2-oxazoline). Macromol Symp 2007;249–250(1):270–5.

[18] Xu L, Ye Z, Cui Q, Gu Z. Noncovalent nonspecific functionalization andsolubilization of multi-walled carbon nanotubes at high concentrations with ahyperbranched polyethylene. Macromol Chem Phys 2009;210(24):2194–202.

[19] Petrie K, Docoslis A, Vasic S, Kontopoulou M, Morgan S, Ye Z. Non-covalent/non-specific functionalization of multi-walled carbon nanotubes with ahyperbranched polyethylene and characterization of their dispersion in apolyolefin matrix. Carbon 2011;49(10):3378–82.

[20] Sing K, Everett D, Haul R, Moscou L, Pierotti R, Rouquerol J, et al. Reportingphysisorption data for gas–solid systems. Pure Appl Chem 1985;57:603–19.

[21] Li Z, Pan Z, Dai S. Nitrogen adsorption characterization of aligned multiwalledcarbon nanotubes and their acid modification. J Colloid Interface Sci2004;277(1):35–42.

[22] Ye Z, Li S. Hyperbranched polyethylenes and functionalized polymers by chainwalking polymerization with Pd-diimine catalysis. Macromol React Eng2010;4(5):319.

[23] Antonucci V, Faiella G, Giordano M, Nicolais L, Pepe G. Electrical properties ofsingle walled carbon nanotube reinforced polystyrene composites. MacromolSymp 2007;247(1):172–81.

[24] Du F, Scogna R, Zhou W, Br S, Fischer J, Winey K. Nanotube networks in polymernanocomposites: rheology and electrical conductivity. Macromolecules2004;37(24):9048–55.

[25] McNally T, Pötschke P, Halley P, Murphy M, Martin D, Bell SJ, et al. Polyethylenemultiwalled carbon nanotube composites. Polymer 2005;46(19):8222–32.

[26] Hu G, Zhao C, Zhang S, Yang M, Wang Z. Low percolation thresholds ofelectrical conductivity and rheology in poly(ethylene terephthalate) throughthe networks of multi walled carbon nanotubes. 2006; 47: 480–88.

[27] Martin C, Sandler J, Shaffer M, Schwarz M, Bauhofer W, Schulte K, et al.Formation of percolating networks in multi-wall carbon-nanotube–epoxycomposites. Compos Sci Technol 2004;64(15):2309–16.

[28] Sandler J, Kirk J, Kinloch I, Shaffer M, Windle A. Ultra-low electrical percolationthreshold in carbon-nanotube-epoxy composites. Polymer 2003;44(19):5893–9.

[29] Regev O, ElKati P, Loos J, Koning C. Preparation of conductive nanotube–polymer composites using latex technology. Adv Mater 2004;16(3):248–51.

[30] Al-Saleh M, Sundararaj U. An innovative method to reduce percolationthreshold of carbon black filled immiscible polymer blends. Composites PartA 2008;39:284–93.

[31] Abbasi S, Carreau P, Derdouri A, Moan M. Rheological properties andpercolation in suspensions of multiwalled carbon nanotubes inpolycarbonate. Rheol Acta 2009;48(9):943–59.

[32] Guth E. Theory of filler reinforcement. J Appl Phys 1945;16(1):29.[33] Chatterjee T, Yurekli K, Hadjiev V, Krishnamoorti R. Single-walled carbon

nanotube dispersions in poly(ethylene oxide). Adv Funct Mater 2005;15(11):1832–8.

[34] Schaefer D, Chen C. Structure optimization in colloidal reinforcing fillers:precipitated silica. Rubber Chem Technol 2002;75(5):773–94.