Hierarchical Structures Composed of Confined Carbon Nanotubes inCocontinuous Ternary Polymer BlendsEyal Cohen,*,†,§ Lior Zonder,†,§ Amos Ophir,‡ Samuel Kenig,‡ Stephen McCarthy,† Carol Barry,†
and Joey Mead†
†Department of Plastics Engineering, University of Massachusetts Lowell, 1 University Ave., Lowell, Massachusetts 01854, UnitedStates‡Department of Plastics Engineering, Shenkar College of Engineering and Design, 12 Anna Frank St., Ramat Gan 52526, Israel
*S Supporting Information
ABSTRACT: An hierarchical structure, composed of aternary cocontinuous polymer blend, where carbon nanotubesare mostly localized in one of the phases through π−πinteractions, is fabricated by direct melt mixing of polyamide12 and polypropylene, as the two major components of theternary blend, together with pyridine-modified poly(ethylene-co-methacrylic acid) as the minor component that can formstrong interactions with the CNTs via π−π interactions andconfined the percolated network at the polyamide/polypro-pylene interface. The hierarchical structure was designed bymeans of surface energies, and the obtained morphology was verified using electron microscopy. This ternary structure has lowerelectrical resistivity as compared to cocontinuous binary composites. Different polymer viscosities were used in this study inorder to emphasize the importance of kinetics during cocontinuous morphology formation.
1. INTRODUCTION
Melt mixing is currently the most convenient and cost-effective method for incorporating conductive fillers, such ascarbon nanotubes (CNTs), into a polymer matrix in order tofabricate conductive polymer composites.1 Conductivity isachieved through the formation of a three-dimensional networkof conductive particles which enables electron transport acrossthe polymer matrix.2
A promising method to reduce the percolation threshold isby melt mixing an immiscible polymer blend together with aconductive particle and generating a cocontinuous morphologywhere the conductive particles are preferentially localized insideone of the phases or at the interface (this is known as doublepercolation).3−11 Advanced structures are based on thelocalization of the conductive particle at the interface of abinary polymer blend. Several studies12,13 have shown thatcarbon black particles can be located at the interface of a binaryblend during melt mixing by tuning either the thermodynamicconditions or the kinetics of the mixing process. Interfacelocalization of CNTs in polymer blends was addressed by someresearch groups. Potschke and co-workers14,15 used reactivemodifiers in different polymer blends to control the CNTlocalization and change the blend kinetics toward interfaciallocalization. Yet, confining the CNTs at the interface itselfremained difficult to achieve and the CNTs tended to migrateto one of the phases eventually. From a thermodynamic pointof view it is unlikely for high aspect ratio particles, such asCNTs, to be located at the interface, as was shown by Goldel etal.16−18 and by Krasovitski and Marmur.19 For example, in
polycarbonate/poly(styrene−acrylonitrile) blends, where theCNTs were premixed in the polystyrene, the CNTs werelocated in the polycarbonate phase even though only minordifferences in surface energies favor the polycarbonate phase. Itwas claimed that high aspect ratio particles form an unstablecurvature at the polymers/particle intersection, which acts toincrease wetting angle of the preferred phase. In order forequilibrium to be restored, molecules of the preferred phaseadvance along the surface and cover the particle. An exceptionis the work by Baudouin et al.20 which showed an interfaciallocalization of CNTs in polyamide/poly(ethylene−methylacrylate) melt blends under certain mixing conditions.A possible route to fabricate a stable microstructure where
CNTs are localized at the interface of an immiscible blend is bythe utilization of a ternary blend. In this ternary blend, the twomajor phases are cocontinuous; the third minor componentselectively contains the CNTs and is localized at the interface ina continuous manner. Morphologies, thermodynamic character-istics, and kinetics aspects of ternary polymer blends composedof two major phases and one minor phase are welldemonstrated in the literature.21−24 The suggested hierarchicalstructure is similar to the cellular structure suggested by Wineyet al.,25,26 who prepared a percolated CNT framework either bypressing coated polymer pellets or by infiltration of polymerresin onto a CNTs framework. The ternary blend strategy
Received: September 11, 2012Revised: January 24, 2013Published: February 18, 2013
could lead to the same CNT structures, by applyingconventional melt mixing techniques, and without using priorsteps to produce the CNT framework.In this work we demonstrate that hierarchical structures
composed of a cocontinuous ternary polymer blend, whereCNTs are mostly confined in the continuous minor phase, arefeasible and can overcome the low tendency of high aspect ratioparticles to be located at the interface of two polymers. Thesehierarchical structures can be exploited as a route to loweredCNT percolation thresholds. Instead of trying to locate theCNTs at the interface, the CNTs were simply confined into aninterphase of a minor third polymeric component, which wascontinuous throughout the blend. To meet this goal, wedesigned and fabricated a cocontinuous ternary polymer blendcomposed of polyamide 12 (PA) and polypropylene (PP) asthe major phases while a poly(ethylene-co-methacrylic acid)copolymer (EMAA) with two different viscosities was used asthe minor phase that has the potential to be located at the PA/PP interface. In order to confine the CNTs in the minor phase,not only were the CNTs premixed with the minor phase butalso the EMAA minor phase was chemically modified with 4-aminomethylpyridine (AMP) in such a way that π−πinteractions could be formed between the modified EMAAand the CNTs, as we had already shown in our previousstudy.27
2. EXPERIMENTAL SECTIONMaterials. Two polymers were used as the major components in
the ternary blends. Homopolymer grade PP was purchased fromCarmel Olefins (Capilene E50E) and PA from EMS-Grivory(Grilamid L25); the PA was vacuum-dried overnight before use. Forthe minor phase two commercial grades of EMAA with an 11.5 wt %concentration of methacrylic acid comonomer were used. The first wasa high-viscosity EMAA (DuPont Nucrel1202HC, melt flow index of1.5 g/10 min), and the second was a low-viscosity EMAA (DuPontNucrel699, melt flow index of 95 g/10 min). The high- and low-viscosity minor phase polymers were designated respectively asEMAA1 and EMAA95. The CNTs used in this study were multiwalledcarbon nanotubes purchased from Nanocyl (NC7000). The CNTswere used as received without any purification but were reported tohave 90% purity and an average surface area of 250−300 m2/g. AnXPS survey indicated that the CNTs were composed of 98.86%carbons atoms (63.4% sp2; 14.8% sp3) and 1.14% oxygen atoms. The4-aminomethylpyridine (AMP), tetrahydrofuran (THF), m-cresol,formic acid (80%), and methanol were obtained from Sigma-AldrichIsrael and were used as received.Modification of Poly(ethylene-co-methacrylic acid) with
Aminopyridine. Both the EMAA1 and EMAA95 were modifiedwith AMP according to the procedure described elsewhere.27 EMAAs(50 g) were dissolved in THF (850 mL) at 67 °C for 24 h in a 1 Lreaction flask. A stoichiometric amount of AMP (4:1 molar ratio withrespect to the methacrylic acid content) was added, and the reactioncontinued for 24 h. The modified polymers (mEMAA1 andmEMAA95) were precipitated with cold methanol, filtered, washedextensively with methanol again, and vacuum-dried at 60 °C for 24 h.The modified polymers were then pressed at 180 °C and cut intopellets.Blend Preparation. The four different minor components
(EMAA1, mEMAA1, EMAA95, and mEMAA95) were melt blendedwith 14.3 wt % CNTs in a Brabender mixer at 80 rpm and 200 °C for8 min to form four different master batches. Ternary polymer blendswere prepared using a 16 mm corotating twin-screw extruder (Prism,Eurolab) operating at 250 rpm with a barrel temperature of 230 °C.The PA and PP were dry-mixed with the different EMAA masterbatches in order to produce 47/47/6 v/v/v ternary blends. Each dryblend was fed directly into the feed throat of the twin-screw extruder.The CNT loading was kept constant at 1 wt %. This CNT loading was
chosen because (1) we found, in a separated experiment, that thepercolation threshold of the PA is about 2 wt % and (2) the PEAAmatrix could not be effectively premixed with more than 14 wt % CNTdue to limitations with our equipment. The nanocomposite pelletswere compression molded into 2 mm thick rectangular plates at 215°C for 7 min. In addition to the CNT composites, neat ternary blendsprepared without any CNTs were used as control samples.
Characterization. FTIR spectra of EMAA and mEMAA wererecorded in order to verify the chemical modification. The analysis(Bruker Alpha) was performed between 375 and 2000 cm−1, theresolution was 2 cm−1, and the results were based on an average of 22scans. FTIR samples were prepared by casting 1% w/v polymer/THFsolutions on KBr windows.
The polar and disperse components of the surface tension for allblend constituents were obtained by measuring the contact anglebetween a solid surface sample of the polymers and three testingliquids: deionized, ultrafiltered water (0.2 μm filter), diiodomethane(DIM), and ethylene glycol (EG). Measurements were performedaccording to the sessile drop method using a video-based, software-controlled, CA analyzer (OCA 20, Dataphysics Instruments,Germany). 5 μL drops from the three liquids were used to calculatethe surface tensions according to the Owens−Wendt geometric meanequation. Standard surface tensions (mN/m) for these liquids wereobtained from the OCA 20 database. For water, γ = 72.1, γd = 19.9,and γp = 52.2; for DIM γ = 50.8, γd = 49.5, and γp = 1.3; and for EG γ= 48, γd = 29, and γp = 19 (where γ is the total surface energy and γdand γp are the disperse and polar components, respectively). Themeasured surface energies, along with their dispersive and polarcomponents for the polymers used in this study, are presented asSupporting Information. Since the temperature coefficients for CNTsand for some of the polymers are not available in the literature, thedata presented are for ambient temperature, and no temperaturecorrections were performed. Interfacial tensions between the differentcomponents were calculated according to the harmonic meanequation.
A solvent extraction method was used to characterize the existenceof a cocontinuous morphology of the different composites. Three 100mg specimens from each nanocomposite were dissolved in m-cresolfor 48 h, then washed in formic acid and in ethanol, and finallyvacuum-dried in order to selectively dissolve the PA phase. Weight lossafter extraction was used to determine the degree of continuity of thePA within the composites. The EMAA proportion and continuityalong the PA/PP interphase were obtained using the same selectiveextraction technique. The same 100 mg samples were used after theextraction of the PA phase. The samples were immersed in THF for 24h, which dissolved the EMAA phase without dissolving the PP phase.The proportion of the EMAA along the PA/PP interface wascalculated by comparing the lost weight of the EMAA to the weight ofthe EMAA according to the composite composition. It was presumedthat not all of the EMAA will be located at the PA/PP interface due tothermodynamic and kinetic of the blend formation.
Morphology analysis was performed using high-resolution SEM(Zeiss Gemini Ultra-55). SEM image samples were cryo-fractured inliquid nitrogen, chrome sputtered, and mounted on SEM sampleholders. SEM images were scanned at 5 kV. TEM (FEI, Tecnai G2 12Twin) images were performed on CNT composites at an acceleratingvoltage of 150 kV on 110 nm thick samples. The contrast between thedifferent phases was obtained by exploiting the different tendencies ofCNTs to be localized at the different phases and therefore “stained”the phases differently.
Electrical volume resistivity was recorded using Keithley electro-meter (model 6517B) equipped with 8009 high-resistance testapparatus on 2 mm thick samples. The voltage was set to 100 V forall samples. At least three specimens were measured from eachcompound.
Viscoelastic behavior of the different composites was characterizedin the melt state by a dynamic oscillatory shear rheometer (TAInstruments’ ARES ex2000) in parallel plate geometry with a diameterof 25 mm and plate gap of 1.9−2.0 mm. The measurements werecarried out at 230 °C, using a constant strain of 1% and with angular
frequency range of 0.01−100 Hz; these conditions were confirmed tobe within the linear viscoelastic region of the materials.
3. RESULTS AND DISCUSSION
3.1. EMAA Modification. The chemical reaction, in whichthe amine end of the AMP reacts with the carboxylic moietiesof the EMAA, is shown in Scheme 1. The reaction includesboth the formation of the salt complex at low temperature andthe condensation of the salt complex into an amide bond atelevated processing temperatures.The FTIR spectra for the modified EMAA1 and EMAA95
are presented in Figures 1a and 1b, respectively. Both EMAA1and EMAA95 undergo an acid−base reaction where theaminopyridine reacts with the carboxylic acid to form anammonium salt as evident by the disappearance of thecarboxylic acid peak (1700 cm−1) and by the formation ofthe carboxylate peak (1540 cm−1). As the pyridine-modifiedEMAAs were exposed to elevated temperatures, the carboxylatepeak disappeared and a typical amide bond peak was formed at1697 cm−1. The characteristic aminopyridine peak is noticeableat 1590 cm−1 for both the salt complex and the amide bondmodification. Both the EMAA1 and the EMAA95 exhibited
similar FTIR spectra after modification which indicates that thesame degree of modification was achieved regardless of theEMAA viscosity.
3.2. Morphology of Ternary Blends. Four morphologiesare available for a ternary polymer blend composed of twomajor phases and one minor phase. In addition tomorphologies where the minor phase is dispersed as dropletsinside one of the two major phases, two other uniquemorphologies are also possible. The first is a state where theminor phase is spread at the interface of the two major phases,whereas in the second state, none of the phases is completelyspread on the interface of the other two; the second state is alsoknown as partial wetting where droplets of the minor phase arelocalized at the interface. The spread state and the partialwetting state are well characterized in the literature.21−24
The morphology of a ternary blend system can be predictedby a set of three values which consider the tendency of eachcomponent to spread at the interface of the two othercomponents. Three spreading coefficients (λikj), which indicatethe tendency of component k to spread at the interface ij, canbe predicted from the measured interfacial tensions (γ),following the procedure described in ref 24, according to eq 1:
Scheme 1. Reaction Scheme for the EMAA Modification with Aminopyridinea
aAt low temperature a salt is formed. As the salt complex is exposed to high temperature by hot press the carboxylate condensed into the amidebond.
Figure 1. FTIR spectra for (a) EMAA1 modification and (b) EMAA95 modification. After the reaction the aminopyridine forms a carboxylatecomplex. As the salt is heated (hot press) the carboxylate condensed into an amide bond.
In order for k to spontaneously spread on the ij interface, thespreading coefficient must be positive. For a ternary blend totake the desired morphology, where the minor phase is spreadbetween the two major phases, the spreading coefficient for theEMAA need to have a positive value (spread), while thecoefficients of the two other component needs to have negativevalues, in order not to form a spread interface between the two
major phases. Other combinations of the three calculatedcoefficients would give rise to other undesirable morphologies,such as droplets of the minor phase in one of the two majorphases and a partial wetting morphology.24
Table 1 lists the spreading coefficients, continuity values, andproportion of EMAA at the interface in the ternary blends. Thespreading coefficients for the different polymer compositionsindicated that, for the four ternary blends studied in thisresearch, the desired morphologywhere the two major
Table 1. Spreading Coefficients Values and Selective Extraction Results
composite components spreading coeff [mN/m]a continuity of PA [%]b proportion of EMAA at the interface [%]b
aErrors were calculated based on a set of five drops from each liquid and includes error propagation throughout the surface energies, surface tensionsand spreading coefficients. bErrors are based on three samples measurements.
Figure 2. SEM images for the different CNT ternary blend nanocomposites, taken at low magnification in order to characterize the blendmorphology: (a) PA/PP blend; (b) PA/PP/EMAA1; (c) PA/PP/mEMAA1; (d) PA/PP/EMAA95; (e) PA/PP/mEMAA95.
phases are continuous and the minor phase spreads at theinterface to form a continuous pathis possible from athermodynamic point of view. By considering the errorpropagation thorough out the calculations, it was predictedthat some of the ternary blends (such as the mEMAA95-basedblend) could exhibit a partial wetting morphology instead ofthe desired interfacial spreading of the minor component. Themorphology of the blends was verified by the selectiveextraction technique. By measuring the weight differencebefore and after the exposure of the different samples to m-cresol, the degree of continuity of the PA was calculated. Asshown in Table 1, the PA phase is 94.7−99.2% continuous in allof the nanocomposites. The solvent extraction technique wasalso used to calculate the proportion of the EMAA phasebetween the PA/PP phases (Table 1). For four of the five CNTcomposites, EMAA was mostly (about 80%) located at the PA/PP interface. The only exception was for the EMAA1-basedcomposite where the high-viscosity EMAA1 existed in a smallerfraction (40%) at the interface. This difference was probablydue to kinetic factors which changed the blend morphology;the higher viscosity EMAA1 failed to fully spread on the PA/PPinterface. It has been already shown that kinetics effects, mostlyrelated to viscosity and elasticity of the materials, cansometimes lead to discrepancies from the spreading coefficientsprediction.28
Further evidence for the ternary blend morphology wasdemonstrated in the SEM images for the different blends. Asshown in Figure 2a, the binary blend exhibited a cocontinuousmorphology. For the blend based on EMAA1 (Figure 2b), thecocontinuous morphology was disrupted and turned intodroplets. In contrast, the use of pyridine-modified mEMAA1(Figure 2c) not only preserved the cocontinuous morphologybut also changed the morphology from a coarse cocontinuousstructure into much finer cocontinuous structure. The changein the PA/PP structure in the presence of the mEMAA1 couldonly be attributed to the localization of the EMAA between thePA and the PP phases. As the EMAA is spread at the PA/PPinterface the coalescence and coarsening behaviors are reduced.Those SEM images (Figures 2b and 2c) were consistent withthe selective extraction analysis. The changes in themorphology between EMAA1 and mEMAA1 blends wasrelated to the pyridine modification of the EMAA whichresults in a viscosity reduction (which will be discussed later)due to the replacement of the hydrogen bonds with pyridinebulky group.27 Interfacial localization and spreading, indicatedby finer cocontinuous morphology, also was observed withEMAA95 and pyridine-modified mEMAA95 (Figures 2d and2e). Thus, low-viscosity minor component (i.e., EMAA95compared to EMAA1) favors the possibility of the EMAAforming a continuous interphase at the PA/PP interface duringmelt mixing, even though both EMAA grades had the samethermodynamic tendencies to do so.3.3. CNT Selective Localization. The specific localization
of particle in a polymer blend can be predicted from thewetting coefficient which describes the tendency of a particletoward a specific polymer phase, as described elsewhere,16
according to eq 2:
ωγ γ
γ=
−−
− −AB CNT
CNT B CNT A
AB (2)
where γCNT‑i is the interfacial tension between the CNTs andthe different polymers. When ω > 1, the CNTs are located in
phase A, and when ω < −1, the CNTs are located in phase B.With −1 < ω < 1, the particle should be located at the ABinterface, but as was mentioned earlier, high aspect ratioparticles such as CNTs are unlikely to remain at the interface.In such cases, the CNT localization was determined accordingto the CNT−polymer interfacial tension (i.e., whether ω ispositive or negative).The wetting coefficients presented in Table 2 indicated that
CNTs have a tendency to be located in the PA phase rather
than the PP phase. The differences in interfacial tensionbetween the CNT−EMAA and the CNT−PA were too small topredict in which phase the CNTs would be confined. On theother hand, when the copolymer was modified with 4-aminopyridine (mEMAA), the CNTs demonstrated a strongthermodynamic preference of toward the mEMAA phase,which was the purpose of introducing π-electron moieties ontothe EMAA chain.Based on the spreading and wetting coefficients, a blend
composed of PA/PP and pyridine-modified mEMAA as theminor phase has the potential (neglecting kinetic consid-erations, as in the case of the high viscosity EMAA1) to form anhierarchical structure, where the mEMAA is spread at the PA/PP interface, due to spreading coefficients, and the CNTs areconfined in that mEMAA phase due to wetting coefficients.The specific localization of the CNTs within the different
blends is shown by the SEM images in Figure 3. On the basis ofour experience, we could differ between the PA phase, which isthe ductile fracture phase, and the PP phase. The CNTs in thebinary PA/PP blend were exclusively localized in the PA phase(Figure 3a), as was predicted form the wetting coefficientsanalysis. The addition of unmodified high-viscosity EMAA1caused droplet morphology (Figure 3b); the causes of thismorphology were discussed earlier. A high concentration ofCNTs was confined at the interphase when pyridine-modifiedmEMAA1 was used (Figure 3c). SEM images of the low-viscosity EMAA95 and modified mEMAA95 (Figures 3d and3e, respectively) showed a concentration of CNTs in a thinlayer located between the PA and the PP main phases. This thinlayer was the methacrylic acid copolymer.The impressive confirmation of the claimed hierarchical
structure is illustrated in a series of TEM images. Figure 4ashows that CNTs are localized in the PA phase in a binary PA/PP blend. When a pyridine-modified mEMAA1 was introducedto the blend, most of the CNTs were confined within the
Table 2. Wetting Coefficients for the CNTs with theDifferent Polymers
polymers ωAB−CNTa interpretation
PP/PA −1.0 CNTs located in the PA phaseEMAA1/PP 3.2 CNTs located in the PA phase or EMAA1 phase
or at the PA/EMAA1 interfaceEMAA1/PA −0.1mEMAA1/PP 1.9 CNTs located in the mEMAA1 phasemEMAA1/PA 1.7EMAA95/PP 4.0 CNTs located in the PA phase or EMAA95
phase or at the PA/EMAA95 interfaceEMAA95/PA −0.2mEMAA95/PP
mEMAA1 phase (Figures 4b, 4c, and 4d). The mEMAA1 phaseby itself was cocontinuous and spread along the PA/PPinterface. Because of density differences between the mEMAA1,PA, and PP phases, a high-magnification image (Figure 4c)clearly showed a clear boundary, confirming that the mEMAA1was spread at the PA/PP interface and that the CNTs preferredto be localized in the mEMAA1 phase or close to it. Theseresults indicate that such hierarchical structures are, in fact,feasible using conventional melt mixing techniques.3.4. Electrical and Rheological Properties. The
existence of an effective CNT pathway along the interfacewas reflected in the electrical resistivity of the differentcomposites (Figure 5). Three important results should benoticed. First, the blend composed of the high-viscosity EMAA(EMAA1) exhibited very high resistivity compared to thebinary blend. This higher resistivity was attributed to poorerpercolation network formation with the EMAA1-based blend’sdroplet morphology. Second, confining the CNTs into aspecific thin layer of EMAA caused a significant reduction inthe volume resistivity of the composite from 1.6 × 1011 to 3.5 ×109 Ω·cm. This behavior could be attributed to the formation ofa confined and efficient CNT percolation network along theEMAA pathway, as compared to a percolation network throughthe volume of the PA phase. Third, the modified EMAA (suchas mEMAA95)-based composites exhibited lower volumeresistivity than composites based on unmodified EMAA (4 ×106 compared to 3.5 × 109 Ω·cm). Thus, the efficientconfinement of the CNTs, induced by π−π interactions with
the pyridine pendent group, improved the percolation networkstructure and enabled lower volume resistivity.The storage modulus of the molten polymer, in the low-
frequency region, is very sensitive to changes in themicrostructure of the material, and as such, may serve as anindication for changes in blend morphology as well as the stateof dispersion of the CNTs. Figures 6a and 6c present thestorage modulus and complex viscosity of different blendcompositions without CNTs and with 1 wt % CNT loadings.First, changes in the behavior of the G′ and |η*| curves in thelow frequency range were observed for the EMAA1-basedcomposite and, to a greater extent, for the mEMAA1-basedcomposite. The increase in the storage modulus forcocontinuous blends is usually attributed to the refinement ofthe cocontinuous structure induced by the third component.30
The EMAA1, however, seemed to bring about a morphologicaltransformation from fully cocontinuous to partially disperseddomains (as seen in the SEM images and selective extractionresults); this transformation also may contribute to theincreased elasticity.31 In contrast, the mEMAA1 polymer keptthe cocontinuous structure intact and further refined it byspreading at the interface. This change was expressed by theappearance of a plateau in the storage modulus curve. The sametrend was observed for the 1 wt % CNT-filled ternary blends,although the elasticity plateau produced by the mEMAA1composite was significantly higher compared to the EMAA1composite and the binary PA/PP composite. The viscoelasticbehavior of the mEMAA1 composite correlated well with its
Figure 3. SEM images for the different CNTs ternary blends nanocomposites, taken at high magnification for the characterization of the CNTlocalization: (a) PA/PP blends, (b) PA/PP/EMAA1, (c) PA/PP/mEMAA1, (d) PA/PP/EMAA95, and (e) PA/PP/mEMAA95.
low electrical resistivity and serves to support our hypothesisthat, in this case, well-dispersed CNTs are localized in themEMAA1 interphase, which is spread at the interface betweenthe PA and PP major phases.The addition of EMAA95 (Figure 6b) resulted in minor
changes to the G′ and |η*| curves. Conversely, the addition ofmEMAA95 (Figure 6d) yielded a general reduction in theelasticity and viscosity of the melt. It also produced a slightflattening of the storage modulus curve in the terminalfrequency range. When 1 wt % CNTs were added, the storage
modulus in the low-frequency region was magnified and lessfrequency dependent for all the blends. The low-frequency G′and complex viscosity curves of the mEMAA95 thirdcomponent in the case of the composite ternary blends aredistinct in comparison to the CNT-filled EMAA95 and PA/PPblends; this blend exhibits higher viscosity, higher elasticity, andmore of a solidlike behavior. As discussed earlier, the AMPmodification decreases the total viscosity of the blend;therefore, the behavior observed for the mEMAA95 must berelated to the state of dispersion and localization of the CNT atthe blends interface.The effect of the pyridine modification on the rheological
response is summarized in Figure 7. The addition of EMAAonly slightly increased the storage modulus of the compositecompared to a simple binary PA/PP CNT composite. Incontrast, when a pyridine-modified EMAA minor phase wasintroduced, there was a dramatic increase in the storagemodulus values as well as a flatting of the curve at the terminalregion (low-frequency region).
4. CONCLUSIONS
An hierarchical structure based on a ternary polymer blendcomposed of PA and PP as the two major continuous phasesand EMMA as a third minor component which was alsocontinuous along the PA/PP interface was fabricated by directmelt mixing. In order to confine CNTs in the EMAA phase andform a low percolation threshold network along the blend
Figure 4. TEM images of (a) PA/PP composite where CNTs are located only in the PA phase with no preference toward the interface and (b−d)PA/PP/mEMMA1 composite where CNTs are located in the PA phase, with a high concentration in the mEMAA1 interphase due to the spreadingof the pyridine-modified mEMAA1 and its high attraction toward CNTs.
Figure 5. Electrical volume resistivity for the different compositesstudied in this work. All samples were loaded with 1 wt % CNTs, anddiffer by the polymer composition (either 50/50 v/v PA/PP or 47/47/6 v/v/v PA/PP/EMAA).
interfaces, the CNTs were premixed with the EMAA and theEMAA was modified with amino(methylpyridine), which formsnoncovalent π−π interactions with the CNTs. The hierarchicalternary polymer blend morphology, where EMAA is spread onthe PA/PP phase, was predicted using spreading coefficienttheory, while wetting theory predicted 2.6 mN/m lowerinterfacial tensions between CNTs and pyridine-modifiedEMAA compared with pristine EMAA. It was found thatviscosity variations affected the resultant morphology. Higherviscosity of the minor EMAA component depressed spreadingat the PA/PP interface, and low-viscosity EMAA encouragedinterfacial spreading; thus, kinetics considerations affect thetheoretically obtained morphology as predicted by thermody-
namic theories such as the spreading coefficient. Pyridinemodification of the higher viscosity EMAA enabled spreadingof EMAA at the interface because the modification reduced theviscosity of the copolymer. This work set the concept andguidelines toward the fabrication of other hierarchical structureswith unique physical properties, composed of ternary polymerblend and nanoparticles.
■ ASSOCIATED CONTENT*S Supporting InformationTable S1. This material is available free of charge via theInternet at http://pubs.acs.org.
■ ACKNOWLEDGMENTSThe authors thank Mr. Mark Schnider and Mr. Guy Goldbergfrom the Weizmann Institute for their help with SEM imaging.The authors thank Yona Lichtenfeld and Dr. Einat Nativ-Rothfrom the Ilse Katz Institute (IKI - Ben Gurion University) forthe TEM imaging.
■ REFERENCES(1) Kim, J. Y.; Park, H. S.; Kim, S. H. J. Appl. Polym. Sci. 2007, 103,1450−1457.
Figure 6. Storage modulus and complex viscosity for the different ternary blends, without CNTs (open symbols) and with 1 wt % CNTs (filledsymbols).
Figure 7. Effect of the third polymeric component on the storagemodulus of the different ternary blend composites containing 1 wt %CNTs.
