Effect of multiwall carbon nanotubes on the phase separation of concentrated blends of...

European Polymer Journal 53 (2014) 253–269

Contents lists available at ScienceDirect

European Polymer Journal

journal homepage: www.elsevier .com/locate /europol j

Macromolecular Nanotechnology

Effect of multiwall carbon nanotubes on the phase separationof concentrated blends of poly[(a-methyl styrene)-co-acrylonitrile] and poly(methyl methacrylate) as studiedby melt rheology and conductivity spectroscopy

⇑ Corresponding author. Tel.: +32 16 32 23 59; fax: +32 16 32 29 91.E-mail address: paula.moldenaers@cit.kuleuven.be (P. Moldenaers).

1 Present address: Department of Materials Engineering, Indian Instituteof Science, Bangalore 560012, India.

Suryasarathi Bose a,1, Ruth Cardinaels a, Ceren Özdilek a, Jan Leys b, Jin Won Seo c,Michael Wübbenhorst b, Paula Moldenaers a,⇑a Department of Chemical Engineering and Leuven Materials Research Center, KU Leuven, Willem de Croylaan 46 – Box 2423, B-3001 Leuven, Belgiumb Department of Physics and Astronomy, KU Leuven, Celestijnenlaan 200D – Box 2416, B-3001 Leuven, Belgiumc Department of Metallurgy and Materials Engineering, KU Leuven, Kasteelpark Arenberg 44 – Box 2450, B-3001 Leuven, Belgium

a r t i c l e i n f o

Article history:Received 22 June 2013Received in revised form 7 January 2014Accepted 27 January 2014Available online 2 February 2014

Keywords:BlendsCarbon nanotubesPhase separationPercolationRheologyElectrical conductivity

a b s t r a c t

Thermally induced phase separation in concentrated lower critical solution temperature(LCST) blends of poly[(a-methyl styrene)-co-acrylonitrile]/poly(methyl methacrylate)(PaMSAN/PMMA) in presence of multiwall carbon nanotubes (MWNTs) with different sur-face characteristics (�NH2 functionalized and polyethylene modified) was monitored bymodulated differential scanning calorimetry, melt rheology, conductivity spectroscopyand microscopic techniques. At a concentration of 2 wt%, the MWNTs clearly reduce themacromolecular mobility of the blend components and induce phase separation at lowertemperatures as compared to the neat blends. Electron microscopic images revealed thatphase separation resulted in a selective thermodynamically driven localization of bothtypes of MWNTs in the PaMSAN phase whereas they were randomly distributed in themono-phasic materials. Hence, a percolative MWNT network was formed at much lowerconcentrations as compared to those needed for percolation in the blend components.However, the significant changes in viscosity and elasticity of the blend components,brought about by the MWNTs, can also affect morphology development. This way, effectivepercolation in the blends can be hindered, as was observed for the 60/40 blend with poly-ethylene coated MWNTs (PE-MWNTs). Finally, a dramatic transition from an insulatingone-phasic material at room temperature to a highly conductive material in the meltwas induced by the phase separation.

1. Introduction

Phase separation of polymer blends involves a complexinterplay between thermodynamics and kinetics. This

complexity can be further amplified by the addition ofnanoparticles as they can alter the free energy of the sys-tem and can lead to a redistribution of polymer moleculesbetween the particle surface and the bulk, depending onthe affinity of the polymers for the particles [1–4]. Thesephenomena can affect the miscibility window of thephases and the kinetics of phase separation. In addition,nanoparticles can impart intriguing properties tothe blends such as morphological stability, electrical

254 S. Bose et al. / European Polymer Journal 53 (2014) 253–269

conductivity, and structural integrity. Hence, the effect ofnanoparticles on the compatibility and the phase separa-tion process has been the subject of several experimentalstudies and simulations [1–14]. For example, Balazs andco-workers [9] found numerically that hard mobile nano-particles with a preferential affinity for one of the compo-nents, slow down the coarsening of the domains in the latestages of spinodal decomposition whereas particles withno preferential adsorption do not affect the phase separa-tion kinetics. In the latter scenario i.e. in absence of adsorp-tion of polymer chains depletion forces become dominantand could lead to particle–particle aggregation and henceinduce phase separation. In addition, the effect of nanopar-ticles on phase separation characteristics such as criticaltemperatures, kinetics and morphology development wasreported to depend on the particle size, concentrationand the interaction between the particle and the host ma-trix [2–4,12,15]. Hence, assessing the interaction betweenthe nanoparticles and the components is essential to gaina thorough understanding of the thermodynamics andkinetics of phase separation.

In the recent past, phase separation has been exploredwith several techniques including differential scanning cal-orimetry (DSC), microscopy, light transmission, light scat-tering, electrical conductivity and rheology [11,16–18]. InDSC, phase separation can often be detected by means ofan endothermic peak or kink in the heat flow curves[11,17,19]. In addition, if the enthalpic effects of demixingare sufficiently large, the kinetics of demixing can beprobed with modulated DSC [17]. Rheological measure-ments have been widely employed to study the phasebehavior of blends through the changes in linear viscoelas-tic properties as the system evolves from a homogeneousthrough a metastable to the phase-separated regime [20].In the latter regime, polymer blends exhibit pronouncedelastic properties, long relaxation times and a failure ofthe time–temperature superposition [20–22]. In this con-text, the point of phase transition can be inferred bothqualitatively and quantitatively from rheological measure-ments using one of the many fingerprints. The most widelyused one is the observation of a shoulder in the storagemodulus (G0) – frequency curve that appears at low fre-quencies upon phase separation [20–22]. Similarly, Cole–Cole plots (g0 as function of g00 or G0 as function of G00) orrelaxation spectra show an additional relaxation at low fre-quencies in phase separated systems [16]. Other represen-tations such as isochronal plots of G0 with temperature thatshow either a clear change in slope or an inflection(depending on the dynamic asymmetry of the compo-nents) in the vicinity of the transition point are also usedto determine the phase separation temperature [20,23].In addition, the time evolution of the dynamic moduli un-der isothermal conditions has been extensively investi-gated to correlate the viscoelastic response of the blendswith the resulting morphology development after bringingthe blend into the unstable region of the phase diagram[20,24–26]. For example, Kapnistos [20] and Polios [25]showed that the storage modulus increases with time dur-ing the early stage of spinodal decomposition and de-creases in the later stages due to the breakup of theinterconnected domains. Kim et al. [26] observed that both

the storage and loss modulus increase initially, followed bya decrease at longer time scales due to the coarsening ofthe cocontinuous structures. Particularly in the case of par-ticle-filled blends for which optical techniques can oftennot be used, rheological techniques can provide essentialinformation about the phase separation characteristics[8,10–14]. However, in the case of dynamically symmetricblends, rheology can only pick up signals when the corre-lation length of the concentration fluctuations is large en-ough [10,13].

The potential of carbon nanotubes (CNTs) as filler inpolymeric nanocomposites has widely been explored ow-ing to their extraordinary combination of properties (forrecent reviews see Refs. [27,28]). As a result of their largeaspect ratio, they can form an interconnected network atrelatively low concentrations. The abundance of p-electronclouds and the large specific surface area of CNTs facilitateinteractions with the matrix polymers. Recently, it wasshown that the adsorption of macromolecular chains onthe surface of CNTs in a binary blend can be partially irre-versible due to the high desorption energy [29]. Hence, it israther interesting to investigate phase separation in pres-ence of CNTs as the co-existence of macromolecular chainson the surface of the CNTs could alter the thermodynamicsas well as the kinetics of phase separation. Moreover, thedispersion state of CNTs can be manipulated using phaseseparation as a tool leading to percolation at low CNT con-centrations [30]. Although there are ample reportsaddressing structure–property relations in polymericblends/composites containing CNTs ([27,28] and refer-ences therein), to our knowledge, the effect of CNTs onthermodynamics and kinetics of phase separation has onlyreceived limited attention [8,10].

Phase separation of the neat lower critical solution tem-perature (LCST) blend of poly[(a-methyl styrene)-co-acry-lonitrile]/poly(methyl methacrylate) (PaMSAN/PMMA)was extensively studied in the past [24,31,32]. In our ear-lier investigations [30,33], we studied the percolation ofvarious types of MWNTs in PaMSAN/PMMA blends and as-sessed the potential of MWNTs for coalescence suppres-sion in 85/15 PaMSAN/PMMA blends [30]. In the presentstudy, we investigate in detail the phase separation in con-centrated 40/60 and 60/40 PaMSAN/PMMA blends in pres-ence of different types of MWNTs as the system transitsfrom a homogeneous mono-phasic state via a metastableto the phase-separated state. First, the state of dispersionof the MWNTs in the blends before and after phase separa-tion and the phase-separated morphology will be dis-cussed. Subsequently, the phase separation temperatureand kinetics will be determined. Finally, the resulting elec-trical and rheological percolation of the MWNTs will beaddressed.

2. Experimental part

2.1. Materials

Both poly-a-methylstyrene-co-acrylonitrile (PaMSAN,Luran KR2556) and polymethylmethacrylate (PMMA, Luc-ryl G77) were obtained from BASF. The blend PaMSAN–

S. Bose et al. / European Polymer Journal 53 (2014) 253–269 255

PMMA is miscible at low temperatures and the transitionto a phase separated state is reported to occur at around165 �C for blends containing 40 wt% PaMSAN and 175 �Cfor blends containing 60 wt% PaMSAN [31]. The character-istics of these polymers and the phase behavior of the neatblends have been extensively discussed in literature [31].Two types of modified MWNTs were used in this studyand were kindly provided by Nanocyl (SA, Belgium) [34];the first type has an amine functionalisation (NH2-MWNTs,NC3152, length: 600–700 nm, diameter: 9–10 nm,NH2-functionalisation less than 0.5%); the second type(PE-MWNTs, NC9000, length: 1.2–1.5 lm, diameter:9–10 nm) is coated with an in situ polymerized polyethyl-ene (PE) layer and predispersed in high density polyethyl-ene (HDPE). For functionalised MWNTs the surface freeenergy is of the order of 45 mN/m [35,36]. The surface freeenergy at 220 �C is estimated to be 25.9 mN/m for PMMAand in the range of 27.4–34.5 mN/m for PaMSAN [30].Hence, the NH2-MWNTs are expected to be localized inthe PaMSAN phase. Due to the nonpolar nature of PE[37], the PE-MWNTs might have a lower surface free en-ergy and become more compatible with PMMA as com-pared to the NH2-MWNTs. However, even for this type ofMWNTs Pötschke et al. observed migration from the PEphase towards the PA phase with larger surface free energy[38] and Özdilek et al. observed selective localization ofthese MWNTs in the PaMSAN phase of PMMA–PaMSANblends [33].

2.2. Preparation of the blends

Two sets of blends consisting of either 40 wt% or 60 wt%of PaMSAN in PMMA were prepared with or withoutMWNTs by melt-mixing using a 15 cc DSM microcom-pounder at 200 �C (±2 �C) with a rotational speed of60 rpm for 20 min. All preparations were performed undernitrogen atmosphere in order to prevent oxidative degra-dation. It is important to note that the processing temper-ature of 200 �C is well above the cloud point of 165 �C forthe 40/60 and 175 �C for the 60/40 PaMSAN/PMMA blends[31]. It was not possible to process them below their cloudpoint temperature due to limitations in the torque in themini-extruder. The extrudate strands were melt-pressedat 160 �C (below the cloud point temperature) by meansof a laboratory press (Collin). The melt-pressed samplesof the neat blends were transparent and hence one-phasic.

2.3. Characterizations

Dynamic mechanical thermal analysis (DMTA) was per-formed using a Q800 (TA Instruments). Samples were pre-pared by compression molding disks of 0.85 mm thicknessand cutting these to obtain the appropriate dimensions of17 � 10 mm2. A tensile force of 0.1 N preload was appliedto the test specimen using a film tension clamp and dy-namic oscillatory loading at a frequency of 10 Hz and withan amplitude of 0.05% strain was applied. Storage modulus(E0) and tand values were obtained during a temperatureramp of 3 �C/min. The change in slope in E0 and the onsetof the peak in tand versus temperature curves were usedto determine the Tg of the blends.

Modulated differential scanning calorimetry (MDSC)was carried out with a Q2000 (TA Instruments), usingstandard Al TZero pans containing 10 mg of sample. Beforeperforming the experiments, the TZero, temperature, heatflow and heat capacity calibrations were performed withsapphire and indium standards. Temperature ramps fromthe homogeneous (120 �C) to the phase separated regime(260 �C) were performed to determine the phase separa-tion temperature. During the tests, the sample chamberwas purged with nitrogen (50 ml/min). In DSC, the heatflow and hence the sensitivity can be increased by increas-ing the heating rate whereas lower heating rates result inan improved resolution of the heat flow peaks. ModulatedDSC is an extension of standard DSC in which the appliedtemperature is modulated with a sine wave. In additionto the total heat flow signal, modulated DSC allows forthe separation of the in-phase (reversing) and out-of-phase(non-reversing) parts of the heat flow signal, which canoffer a combination of high sensitivity and high resolution[17]. In addition, quasi-isothermal MDSC measurements(with an average heating rate of 0 �C) can be used to studythe kinetics of phase separation [17]. Due to the absence ofstrong interactions such as hydrogen bonds, betweenPaMSAN and PMMA, their phase separation is accompa-nied by rather weak heat flow effects. Hence, a relativelyhigh heating rate of 3 �C/min was found to be necessaryto clearly pick up the phase separation. In addition to theconstant average heating rate, the temperature was modu-lated with an amplitude of 2 �C and a period of 80 s.

The viscoelastic properties of the blends were studiedusing an AR2000ex stress-controlled rheometer (TA Instru-ments) with parallel plate geometry (25 mm diameter and1 mm gap) under N2 atmosphere. Isochronal dynamic tem-perature ramp measurements at a frequency of 0.1 rad/s,which is low enough to be in the terminal region [33], wereperformed from the homogeneous (120 �C) to the phaseseparated regime (220 �C) to determine the phase separa-tion temperatures. Since low heating rates do not deterio-rate sensitivity in rheology, a low heating rate of 0.3 �C/min was selected to enable an accurate determination ofthe phase separation temperatures. The strain amplitudewas within the linear viscoelastic region as determined apriori.

To investigate the kinetics of phase separation, 5 h longtime sweep experiments at an elevated temperature(220 �C) were performed at 1% strain and a frequency of0.1 rad/s. G0 showed almost time independent values after5 h, reflecting completion of the phase separation process.The resulting blend morphology and MWNT dispersionwas characterized by small amplitude oscillatory measure-ments as a function of frequency.

Additional morphological analysis was performed bymeans of either field-emission gun (FEG) scanning electronmicroscopy (Philips XL30) or transmission electron micros-copy (TEM, Philips CM 200 FEG). To investigate the disper-sion of the MWNTs in the one-phasic blends, compressionmolded samples were cryofractured in liquid nitrogen andsubsequently mounted on the SEM sample holder. Thesamples were gold sputtered to avoid charging. The phaseseparated samples were visualized with TEM since SEMdid not provide sufficient contrast to distinguish the

separate phases of PaMSAN and PMMA. For TEM, thecompression molded samples (pressed at 160 �C in theone-phasic region) were first annealed at 220 �C for 5 h(to allow complete phase separation) and subsequentlyquenched to freeze the morphology. The samples werethen sectioned by ultramicrotomy to thin lamellae(100 nm) and mounted on a Cu-TEM grid with holeycarbon film.

Conductivity spectroscopy measurements were per-formed both at room temperature and at an elevated tem-perature of 220 �C on compression molded samples(16.5 mm diameter and 1.75 mm thickness) in the fre-quency range of 10�2–107 Hz using a Novocontrol Alphahigh resolution dielectric analyzer. The instrument mea-sures the complex impedance spectra of the sample. Com-plex conductivities were calculated from the compleximpedance values by taking the sample geometry intoaccount. In this work, the DC conductivity will be used tocharacterize the sample response. The latter is obtainedfrom the real part of the complex conductivity at low fre-quencies, as described by Leys et al. [39]. For conductivityspectroscopy measurements at 220 �C, the samples wereplaced between two brass plates, separated by a circularTeflon spacer to maintain a fixed sample geometry in themelt state. The same spacer was used for all measurementsin order to eliminate systematic errors in sample dimen-sions. A Novocontrol Quatro temperature controller, whichuses nitrogen gas flow, was used to control the sampletemperature with an accuracy of 0.1 �C. Room temperatureconductivities were measured for the miscible blends withboth types of MWNTs. In addition, frequency scans wereperformed at a fixed temperature of 220 �C during phaseseparation for 5 h. As each frequency scan took about3 min, 100 such scans were performed.

3. Results and discussion

3.1. State of dispersion of MWNTs in the mono-phasic systemsas probed by DMTA and FEG-SEM

To evaluate the effect of phase separation on the local-ization of MWNTs in the blends one should know their dis-tribution in the mono-phasic blends, which are the startingmaterials for phase separation. As mentioned above, theblends were melt-mixed at 200 �C, which is well abovethe cloud point for both the 40/60 and 60/40 blends, butthey were melt-pressed at a temperature in the one-phasicregion (160 �C) prior to further characterizations. It isenvisaged that this methodology (i.e. squeeze flow defor-mation) allows the phase separated domains (generatedduring melt mixing) to mix again or at least transform intomicroheterogeneities [31]. The fact that the melt-pressedsamples of the neat blends were transparent, suggestshomogeneity. However, the nanocomposites were darkthus requiring more quantitative methods to assess homo-geneity. Hence, the Tg of the melt-pressed samples wasdetermined. It has been reported that DMTA can resolveseparate Tgs of matrix and dispersed phase for domainsizes on the segmental level (order of 5–10 nm) [40]. InFig. 1 the storage modulus and tan delta as a function of

temperature are shown for neat blends and blends withPE-MWNTs or NH2-MWNTs. It is evident from Fig. 1 thatthe melt-pressed blends indeed show a single Tg (one peakin tand versus temperature curves) confirming their homo-geneity. A change in slope in the E’ versus temperaturecurves (as indicated in Fig. 1) is evident at �118 �C for40/60 and �120 �C for 60/40 neat PaMSAN/PMMA blends.The onset of the peak in the tand versus temperaturecurves also coincides with these values. As expected, theTgs of the blends are between those of the components(Tg of PaMSAN is �124 �C and Tg of PMMA is �109 �C)[31]. Furthermore, it can be seen from Fig. 1 that 1 wt%of MWNTs has a limited effect on the blend Tg whereaswith 2 wt% NH2-MWNTs the rise in Tg is substantial in bothblends, essentially indicating reduced macromolecularchain mobility in presence of MWNTs in the composites.The reinforcement effects of MWNTs (especially in theglassy regions) can be judged from the values of the stor-age modulus (E’) of the blends. All the blends investigatedhere, except for the 60/40 PaMSAN/PMMA blend with PE-MWNTs, displayed a substantial increase in the E’ values(especially in the glassy region) in presence of MWNTs.For the 40/60 blends, PE-MWNTs at 1 wt% result in aslightly larger increase of E’ as compared to NH2-MWNTsat 1 wt%. This behavior is in line with the expectationsbased on the longer length of the PE-MWNTs. The increasein the dynamic mechanical properties indicates a good dis-persion and strong interactions between the MWNTs andthe components of the blends. However, the nature ofthe interactions with the components can be stronglydependent upon the surface functionality of the MWNTs.The reduction of E’ of the 60/40 blend with 1 wt%PE-MWNTs points to less good interactions between theMWNTs and the blend components in case of this system.The PE-MWNTs are coated with an in situ polymerized PElayer and predispersed in HDPE. Presumably, the morepolar nature of the NH2-MWNTs can lead to specific/non-specific interactions with the major blend componentPaMSAN (such as p–p interactions) in the 60/40 blends.The latter could possibly lead to better reinforcement ofthe blends due to a good dispersion.

Electron microscopy can provide further information onthe state of dispersion of the MWNTs in the mono-phasicmaterials. The insets of Fig. 1 are FEG-SEM images of thecryofractured melt-pressed blends with 2 wt% NH2-MWNTs. One can clearly observe a random distributionof the MWNTs, appearing as white dots, representative ofthe conductive material in the bulk.

3.2. State of dispersion of MWNTs and morphology of thephase-separated systems as probed by TEM

The blend morphology after phase separation and thecorresponding dispersion of the MWNTs was studied bycollecting TEM images of samples that were annealed for5 h at 220 �C, thus allowing phase separation to proceeduntil full completion. TEM images of PaMSAN/PMMAblends with different MWNTs at 2 wt% are shown inFig. 2. Although the samples were not stained, the twocomponents had sufficient contrast. The PaMSAN compo-nent appears dark while the PMMA appears bright in the

40 80 120 160 2000

117.6 oC

Temperature (oC)

40 80 120 160 200Temperature (oC)

40/60 40/60+1wt% NH

2-MWNTs

40/60+1wt% PE-MWNTs 40/60+2wt% NH

2-MWNTs

125.6 oC

60/40 60/40+1wt% NH

2-MWNTs

60/40+1wt% PE-MWNTs 60/40+2wt% NH

2-MWNTsSt

120 oC

δ 126 oC

Fig. 1. Storage modulus and tand as a function of temperature for PaMSAN/PMMA blends with NH2-MWNTs or PE-MWNTs (a) 40/60; and (b) 60/40 (insetsshow the FEG-SEM images of the composites containing 2 wt% NH2-MWNTs).

bright-field TEM images. The 40/60 PaMSAN/PMMAblends in presence of MWNTs (at 2 wt%) show a cocontin-uous morphology (see Fig. 2a and b). A similar morphologytype was observed by other authors for the neat PaMSAN/PMMA blend with this composition, after annealing at220 �C for either 10 min or 4 h [31,41]. On the contrary,the 60/40 PaMSAN/PMMA blends containing either NH2-MWNTs or PE-MWNTs exhibit a droplet–matrix type ofmorphology (see Fig. 2d and e). This morphology is com-pletely different from that observed in literature for theneat 60/40 PaMSAN/PMMA blends for which cocontinuousmorphologies were reported after 10 and 100 min ofannealing [24,31]. This might indicate that the presenceof MWNTs affects the morphology type obtained afterphase separation. It is envisaged that selective localizationof nanoparticles in the blends can alter the viscosity (orelasticity) ratio which can further significantly affect theresulting morphology. An estimation of these parameterscan also shed light on the phase inversion concentrationand will be discussed in more detail in Section 3.5. How-ever, as morphology studies for neat 60/40 PaMSAN/PMMA blends in literature are limited to 100 min ofannealing, the neat 60/40 blend was subjected here tothe same annealing protocol as that used in the presentwork for the MWNT-containing blends. The neat blendsalso showed a droplet-matrix morphology after 5 h ofannealing (see Fig. 2c). This indicates the breakup of the

initially formed cocontinuous structure into a droplet-ma-trix morphology, irrespective of the presence of MWNTs.As can be seen in Fig. 2c–e, this formation of a droplet–ma-trix morphology by breakup rather than nucleation andgrowth of small droplets, can result in morphologies witha broad distribution of droplet sizes. When comparingFig. 2d and e, clear differences in droplet size and dropletsize distribution can be seen for blends with either 2 wt%NH2-MWNTs or 2 wt% PE-MWNTs. This suggests the occur-rence of breakup at different stages of the coarsening pro-cess, leading to largely different final morphologies. Thisaspect will further be investigated by studying the kineticsof phase separation (see Section 3.4.).

From the high resolution TEM images in the insets ofFig. 2 it is evident that both types of MWNTs are selectivelylocalized in the PaMSAN phase of the blends, irrespectiveof the final morphology. It is worth pointing out at thisstage that the localization of MWNTs in the PAMSAN phaseis energetically favorable from a thermodynamic point ofview (Section 2.1). The observed localization is thus a clearmandate to the fact that MWNTs migrate (or re-distribute)to the more preferred phase during phase separation. Thisthermodynamically driven migration of MWNTs leads toan increase in their local concentration in a specific phasewhich facilitates their formation of network-like structuresin the blends. In this context, dynamic frequency sweepsand electrical conductivity spectroscopy can be used as a

Fig. 2. TEM images of phase separated PaMSAN/PMMA blends (a) 40/60 with 2 wt% NH2-MWNTs; (b) 40/60 with 2 wt% PE-MWNTs; (c) 60/40 neat blends;(d) 60/40 with 2 wt% NH2-MWNTs; and (e) 60/40 with 2 wt% PE-MWNTs.

probe to obtain further insight in the percolative network-like structure formation of MWNTs in the PaMSAN/PMMAblends (see Sections 3.5 and 3.6).

3.3. Effect of MWNTs on phase separation temperatures

The effect of nanoparticles on the phase separation tem-perature of LCST-type blends was extensively discussed byGinzburg [4,6]. According to this model the total free en-ergy of the system (blend with nanoparticles) depends onthe polymer–polymer, particle–particle and polymer–par-ticle interactions. Though this model describes the stability

(or the free energy) of the homogeneous phase, it neglectsthe effect of particle ordering or segregation on the poly-mer–particle interactions. In the case of MWNTs, it is wellunderstood that they are dispersed as an exfoliated net-work-like structure or in smaller bundles in a polymer ma-trix, where particle ordering is a common phenomenoneven at very low concentrations. This is due to their highaspect ratio which facilitates the formation of a percolativenetwork-like structure even at very low concentrations[27]. Interestingly, in the PaMSAN/PMMA blends theMWNTs were observed to be randomly distributed in thebulk in case of one-phasic materials (Fig. 1) whereas during

phase separation the thermodynamic forces drive theMWNTs to migrate to their preferred phase, resulting innetwork-like structures (Fig. 2). It is envisaged that thispreference of the MWNTs for one of the blend componentsand their redistribution in the blend, might affect theoccurrence of phase separation. These effects are howevernot yet described by the available models. Hence, the effectof the presence of MWNTs in the blends on the phaseseparation temperatures is studied here in more detail.

3.3.1. As probed by MDSCTo determine the phase separation temperatures of

neat PaMSAN/PMMA blends and blends filled withMWNTs, MDSC scans with a heating rate of 3 �C/min wereperformed. For the present blend, separation of the heatflow signal in the reversing and non-reversing contribu-tions did not improve the sensitivity or resolution. Hence,the total heat flow is used to characterize phase separation.This total heat flow is similar to the heat flow that wouldbe obtained in a standard DSC experiment. Heat flowcurves for 60/40 PaMSAN/PMMA blends are shown inFig. 3a. It can be seen from this figure that at the tempera-ture at which phase separation is expected (175 �C), anendothermic peak starts in the heat flow signal. Similarendothermic peaks or kinks at the phase separation tem-perature were observed by others [11,17,19]. The relation

Fig. 3. (a) DSC curves obtained with a heating rate of 3 �C/min. Curves are shibaselines; (b) smoothed extra heat flow due to phase separation for 60/40 blenextra heat flow due to phase separation for 40/60 blends with different concent

between this endothermic peak and phase separationwas confirmed by the absence of any peak in the secondheating ramp of a phase-separated sample of which themorphology was frozen by quenching. At first glance, theneat 60/40 blend and the corresponding blends filled withPE-MWNTs or NH2-MWNTs in Fig. 3a show a heat flowpeak in the same temperature interval. To characterizeand compare this in more detail, the extra heat flow origi-nating from phase separation was calculated by subtract-ing the baseline (straight dashed lines in Fig. 3a). Thethus obtained extra heat flow is plotted in Fig. 3b and cfor 60/40 PaMSAN/PMMA and 40/60 PaMSAN/PMMAblends respectively. It can be seen that for all blends phaseseparation starts at about 160 �C and is completed at220 �C, with a maximum at 187 �C and 183 �C for neat60/40 and 40/60 blends respectively. The maxima (de-noted as TDSC) of all the blends investigated here are listedin Table 1. TDSC is higher than the phase separation temper-atures obtained by Laun et al. [31] from the optical appear-ance of the blends after 32 h of annealing under isothermalconditions. This difference is caused by the fast heatingrate in the present experiments [19]. The major conclusionfrom Fig. 3b and c is that the TDSC temperatures at whichthe extra heat flow is maximum for 60/40 and 40/60 PaM-SAN/PMMA blends are not affected by the presence of1 wt% or 2 wt% PE-MWNTs or NH2-MWNTs. Although the

fted vertically to enable comparisons. Straight dashed lines indicate theds with different concentrations and types of MWNTs; and (c) smoothedrations and types of MWNTs.

Table 1Comparing the critical phase separation temperature obtained fromrheology (Trheo and Ts) and differential scanning calorimetry (TDSC).

PaMSAN/PMMA blends TDSC

(±1 �C)Trheo from tand

(±1 �C)Ts from FL(±1 �C)

40/60 183 173 21540/60 + 1 wt% PE-MWNTs 183 171 21640/60 + 1 wt% NH2-MWNTs 183 173 21540/60 + 2 wt% PE-MWNTs 184 – –40/60 + 2 wt% NH2-MWNTs 181 169 21060/40 187 176 21560/40 + 1 wt% PE-MWNTs 184 173 21460/40 + 1 wt% NH2-MWNTs 188 171 21560/40 + 2 wt% PE-MWNTs 186 – –60/40 + 2 wt% NH2-MWNTs 186 167 210

extra heat flow curves as shown in Fig. 3b and c could beobtained in a reproducible way, it appeared that the reso-lution was too low to accurately determine the onset tem-perature of phase separation. Hence, to determine thelatter with a higher resolution, rheology experiments wereperformed at a lower heating rate (see next section).

3.3.2. As probed by melt rheologyThe effect of MWNTs on the phase separation tempera-

ture was also investigated rheologically by performing iso-chronal temperature ramp measurements with a heatingrate of 0.3 �C/min at a fixed frequency of 0.1 rad/s and astrain amplitude in the linear viscoelastic region (deter-mined a priori). This technique is widely used to determinethe phase separation temperatures in complex systems(e.g. polymer blends with nanoparticles) as sample turbid-ity makes it rather difficult to determine the cloud point insuch materials. In the blends investigated here only a lim-ited interval of temperatures was studied in the homoge-neous region due to the proximity of the LCST to Tg. Asthe system approaches the phase separation temperaturethe viscoelastic properties of the blend change substan-tially. A typical example illustrating the effects of phaseseparation is shown in Fig. 4a for the neat 60/40 PaM-SAN/PMMA blend. At low temperatures the dynamic mod-uli (G0 and G00) decrease with increasing temperature due toenhanced chain mobility. As the phase boundary is ap-proached, competing forces (chain mobility and thermody-namics) result in either a distinct slope change or anupturn in G0 depending on the dynamic asymmetry of theblends (contrast in Tg) [20]. G00 exhibits a similar behaviorhowever, with a phase lag (see Fig. 4a). It should be notedthough that apart from the dynamic asymmetry, thechange in slope in G0 is also very sensitive to the experi-mental conditions e.g. heating rate and frequency em-ployed during the measurements and more importantlythe interfacial tension in the generated two-phasic system[20]. Since the blends investigated here represent weak dy-namic asymmetry, phase separation shows up as a slopechange rather than a distinct peak in the curves of the dy-namic moduli versus temperature. In this case, deducingthe phase separation temperatures from the plots of tandas a function of temperature appears to be more accuratesince this variable shows a linear behavior (on log scale)in the pretransitional region and deviates from linearityupon phase separation. This deviation is quite prominent

(Fig. 4d and e) in contrast to the slope change in G0 versustemperature (Fig. 4b and c) and its occurrence will be re-ferred to as the rheological phase separation temperature(Trheo) in this paper.

Table 1 lists the phase separation temperatures ob-tained from temperature dependent tand plots for the var-ious blends investigated here. For the blends with 1 wt%MWNTs, the rheological data show limited effects of theMWNTs on the phase separation temperatures, althoughthere is a tendency for a decrease of the phase separationtemperature. However, for the blends with 2 wt% NH2-MWNTs, a clear reduction of the phase separation temper-ature can be observed. This decrease is more prominent inthe case of the 60/40 PaMSAN/PMMA blends. It is note-worthy that for the blends with 2 wt% NH2-MWNTS a dis-tinct effect of the MWNTs on the Tg of the blends was alsopresent (Fig. 1), confirming the effect of the nanoparticleson the chain dynamics of the polymer components forthese systems. A similar reduction of the phase separationtemperatures was also observed for PaMSAN/PMMAblends containing thermally reduced graphene sheets[14]. Also in this case, the graphene sheets showed a pref-erential localization in the PaMSAN phase after phase sep-aration [14]. In that case, effects were however alreadyclear at concentrations of 1 wt%, which might be causedby the larger surface to volume ratio of platelets as com-pared to tubes. It is worth pointing out that the phase sep-aration temperatures of the neat blends determined fromthe temperature ramp experiments, are still slightly higherthan the cloud point temperatures, as found by Laun et al.[31]. This is most probably caused by the fact that the elas-tic contribution of concentration fluctuations in dynami-cally weakly asymmetric blends is small. Hence, asufficient amount of interface has to be formed beforethe elasticity originating from the interface rather thanconcentration fluctuations can be picked up in the rheolog-ical signal.

In addition to the onset temperature of phase separa-tion, the dynamic temperature measurements also allowto obtain the spinodal phase separation temperature (Ts)by using the theoretical approach of Fredrickson and Lar-son (FL) [23,42]. The theory yields the following expres-sions for the dynamic storage (G0) and loss moduli (G00),respectively:

G0 ¼ kBTx2

1920p13

ØN1þ

ð1� ØÞN2

( )" #1=2

� 1Øa2

1W1þ 1ð1� ØÞa2

� �2

½2ðvs � vÞ��5=2 ð1Þ

G00 ¼ kBTx2

240p13

ØN1þ

ð1� ØÞN2

( )" #�12

� 1Øa2

1W1þ 1ð1� ØÞa2

� �½2ðvs � vÞ��

12 ð2Þ

where v and vs designate the interaction parameter andthe interaction parameter at the spinodal point respec-tively and Ø the volume fraction. Rgi denotes the radiusof gyration of species i, Ni is the number of segments,

145 155 165 175 185 195 205 215 225102

Temperature (oC)

PαMSAN G'

60/40 PαMSAN/PMMA G'

(a)155 165 175 185 195 205 215 225

PαMSAN/PMMA Blends 40/60 40/60 + 1 wt% NH

2-MWNTs

40/60 + 1 wt% PE-MWNTs 40/60 + 2 wt% NH

2-MWNTs

145 160 175 190 205 220

140 160 180 200 220

4 60/40

Temp. (C)

2-MWNTs

60/40 + 1 wt% PE-MWNTs 60/40 + 2 wt% NH

2-MWNTs

176 oC

(d)103

2-MWNTs

60/40 + 1 wt% PE-MWNTs 60/40 + 2 wt% NH

2-MWNTs

140 150 160 170 180 190 200 210 220

10PαMSAN/PMMA Blends

40/60 40/60 + 1 wt% NH

2-MWNTs

40/60 + 1 wt% PE-MWNTs 40/60 + 2 wt% NH

2-MWNTs

(e)0.0020 0.0021 0.0022 0.0023 0.00240

(G''2 /G

1/T (K)

Ts= 212 oC Trheo= 176 oC

Temperature (oC)

155 165 175 185 195 205 215 225Temperature (oC) Temperature (oC)

Temperature (oC)

Fig. 4. (a) Dynamic moduli as a function of temperature at 0.1 rad/s and 0.3 K/min for PaMSAN and neat 60/40 PaMSAN/PMMA blends; G0 as a function oftemperature at 0.1 rad/s and 0.3 K/min for (b) 40/60; (c) 60/40 PaMSAN/PMMA blends with different MWNTs; tand as a function of temperature for (d) 60/40; (e) 40/60 PaMSAN/PMMA blends with different MWNTs; and (f) Fredrickson–Larson plots for the neat 60/40 blend.

Wi is the rate of motion of the subunit of length ai (is of theorder 109–1011 s�1 for polymer melts). Assuming the fol-lowing expression for the interaction parameter v:

v ¼ Aþ BT

the following relation can be obtained:

1Ts� 1

� �ð4Þ

where C is given by:

C ¼ 45pkB

� �2=3 a21

Øþ a2

1� Ø

� �ð5Þ

A linear relationship can be obtained by plotting G002

� �2=3

versus 1T and the intercept with the horizontal axis yields

the spinodal temperature Ts. A typical example illustratingthe technique of obtaining the Ts from the FL plots is shownin Fig. 4f for a 60/40 PaMSAN/PMMA blend. The Ts for the

different blends investigated here are listed in Table 1.Similar to the results of the other techniques, the Ts is onlylowered as compared to that of the neat blends in the caseof blends with 2 wt% MWNTs. In summary, MWNTs(at 2 wt%) were observed to induce phase separation inboth 40/60 and 60/40 PaMSAN/PMMA blends at lowertemperatures as compared to the neat blends.

3.4. Phase separation kinetics as probed by timesweepmeasurements

First, it was attempted to follow the phase separationkinetics at 220 �C by means of MDSC, similar to the workof Dreezen et al. [17]. However, the change in heat capacityduring phase separation was too limited for this system toobtain reliable information. Alternatively the kinetics wasstudied by means of rheology, more specifically timesweep experiments were performed with a fixed frequency(0.1 rad/s) and strain (1%) after a stepwise increase intemperature from room temperature to 220 �C. It has beenreported that at this temperature, both the neat blends(40/60 and 60/40 PaMSAN/PMMA) phase separate by spin-odal decomposition [31]. It is envisaged that during theearly stages of spinodal decomposition, G0 rises due to anincrease in both concentration fluctuations and specificinterfacial area. However, in the late stages of spinodaldecomposition, G0 decreases due to interfacial tension dri-ven coarsening or breakup of the cocontinuous structure toa droplet–matrix structure [20,24–26] until steady stateconditions are reached [24]. Fig. 5a shows the time evolu-tion of G0 for neat 40/60 and 60/40 PaMSAN/PMMA blends.It is evident from Fig. 5a that for both blends, only the sig-nals from the late stages of spinodal decomposition couldbe picked up i.e. the interfacial tension driven coarseningof co-continuous structures. Similar observations were alsoreported by Vinckier and Laun [24]. The curves of thedynamic moduli versus time do not allow for a differenti-ation between a cocontinuous morphology, as observedfor the 40/60 blend or a droplet–matrix type of morphol-ogy, as observed for the 60/40 blend. This indicates thatbreakup of the cocontinuous structure occurs only afterrather long annealing times, at which the structure hasalready substantially coarsened and the interfacial area isnot too much reduced by the transition to spherical drop-lets. The difference in the absolute value of the dynamicmodulus of 60/40 versus 40/60 PaMSAN/PMMA blendsshould not be interpreted as a manifestation of differencesin phase structure, as it is known that PaMSAN is moreelastic than PMMA [31].

The evolution of the dynamic modulus during phaseseparation changes dramatically in the blends in presenceof MWNTs, as can be seen in Fig. 5b–e. For the 40/60 blendswith a cocontinuous structure, the reduction in dynamicmodulus as a function of time can only be noticed for theblend with a concentration of NH2-MWNTs as low as0.5 wt% (see Fig. 5g). Interestingly, even for this blend,the dynamic modulus starts to increase as a function oftime after about 10 min. For blends with larger concentra-tions of MWNTs, G0 increases as a function of time duringthe complete annealing experiment of 5 h (see Fig. 5b).With increasing concentration of MWNTs, a corresponding

increase of the dynamic modulus is found. For the highestconcentration of MWNTs, moduli that are about 500 timeslarger than that of the neat blend are obtained. Based onthe morphology of the blends containing 2 wt% MWNTs(Fig. 2a and b), this large increase in storage modulus can-not originate from the slightly increased interfacial areadue to the finer morphology as compared to that of theneat blend. In addition, it is not expected that MWNTscan increase the interfacial tension of the blend to such alarge extent that it could explain this large storage modu-lus. However, it is known that adding fillers of nanoscopicdimensions to polymers can increase the elasticity of thematerial significantly, particularly above the percolationthreshold of the filler. The percolation thresholds ofNH2-MWNTs and PE-MWNTs in the PaMSAN phase are�2–3 wt% [30] and �0.5–1 wt% [33] respectively. Hence,the thermodynamically driven migration of MWNTs tothe PaMSAN phase during phase separation is expectedto increase their effective concentration. This effective risein the concentration is by a factor of �2.5 in the 40/60blends, which results in concentrations well above the per-colation threshold of both MWNTs in PaMSAN. Conse-quently, the network of MWNTs is also expected toincrease the melt elasticity of the blends as compared tothat of the neat blends and can presumably mask themicrostructural developments which otherwise can bejudged from the time evolution of G0. Moreover, at an ele-vated temperature the dynamic percolation of MWNTsplays a dominant role in the formation of a ‘network-like’structure in the composites. To gain insight, G0 as afunction of time was monitored for the filled PaMSANhomopolymer with 2 wt% NH2-MWNTs (see Fig. 5e).Interestingly, the PaMSAN/PMMA blends with only0.5 wt% of MWNTs exhibit significantly higher G0 valuesas compared to PaMSAN with 2 wt% NH2-MWNTsalthough the effective concentration of NH2-MWNTs inthe PaMSAN phase of the blend is expected to be only1.25 wt%. This clearly shows that the effects on the blendstorage modulus of phase separation induced localizationof NH2-MWNTs in one of the blend phases exceeds themere concentration effect. This could partially be causedby a reduction of the morphological coarsening, which al-lows for a larger amount of interface in the MWNT-filledblend as compared to the neat blend. However, these ef-fects are not sufficient to explain the significant increasein G0. Based on the TEM images in Fig. 2b, it can be con-cluded that the presence of the PMMA domains, well dis-persed throughout the polymer blend, drives the MWNTstowards the center of the PaMSAN domains (energeticallyfavored phase), thereby further facilitating the formationof a percolative MWNT network. It is worth noting thatPE-MWNTs percolate better in PaMSAN (�0.5–1 wt%) ascompared to NH2-MWNTs. This early onset of percolationcan be related to the filler dimensions. The length of PE-MWNTs is higher than that of NH2-MWNTs by a factortwo. Hence, a larger storage modulus is also obtained forthe blend with 2 wt% PE-MWNTs as compared to that with2 wt% of NH2-MWNTs (Fig. 5b).

Finally, Fig. 5 shows that despite the fast increase of themodulus at short time scales, it takes much longer time toreach a steady state in the blends with MWNTs; in the neat

0 100 200 3000

60/40 PαMSAN/PMMA

40/60 PαMSAN/PMMA

Time (min)

0 100 200 3000

40/60 PαMSAN/PMMA+ 2wt% NH2-MWNTs

40/60 PαMSAN/PMMA + 1wt% NH2-MWNTs

40/60 PαMSAN/PMMA + 2wt% PE-MWNTs

Time (min)

Time (min) Time (min)

0 100 200 3000

60/40 PαMSAN/PMMA + 1 wt% NH2-MWNTs

(c) 60/40 PαMSAN/PMMA with 2 wt% NH2-MWNTs

0 100 200 300150

35060/40 PαMSAN/PMMA + 2wt% PE-MWNTs(d)

0 100 200 300

PMSAN+2 wt% NH2-MWNTs

40/60 + 0.5 wt% NH2-MWNTs

Time (min)

(e) 60/40 + 0.5 wt% NH2-MWNTs

Fig. 5. Time evolution of G0 at 0.1 rad/s and with 1% strain for (a) neat 40/60 and 60/40 blends; (b) 40/60 blends with MWNTs; (c) 60/40 blends with NH2-MWNTs; (d) 60/40 blends with PE-MWNTs; and (e) filled homopolymers versus phase separated systems.

blends, it takes less than 150 min for G0 to reach its satu-rated stage (Fig. 5a) whereas it takes more than 200 minfor blends with 2 wt% MWNTs (Fig. 5b). Two different sce-narios have been identified that can affect the process ofphase separation in presence of nanoparticles [5,6]. Inone case in which nanoparticles move randomly through-out the system it is envisaged that the domain growth isdisturbed and this phenomenon is often termed as ‘weakcoupling’. In contrast, if the nanoparticles are thermody-namically driven to a specific phase, often termed as‘strong coupling’, the domain growth is believed to be slo-wed down instead of arrested. In our case the latter sce-

nario i.e. ‘strong coupling’ is believed to be the dominantmechanism affecting phase separation. However, the slowdynamics of MWNTs in these high viscous polymer melts,will also lead to slow structural reorganizations of theMWNT network, which could cause a gradual but veryslow increase of the storage modulus with time, asobserved in Fig. 5.

In the 60/40 blends, that are characterized by spinodaldecomposition into a cocontinuous structure, followed bybreakup of this structure into a droplet–matrix type ofmorphology, a slightly different evolution of the storagemodulus is obtained in some cases. For the blends with

NH2-MWNTs (Fig. 5d), an increase of the storage moduluswith time can be observed for the two concentrations ofMWNTs. Moreover, the values of G0 are of the same orderof magnitude as those of 40/60 PaMSAN/PMMA blends(Fig. 5b). Similar to those blends, the large values of G0

point to the formation of percolated MWNT networks.However, for the 60/40 blend with 2 wt% of PE-MWNTs,a clear decrease of the storage modulus at short annealingtimes is observed and a reversal of the evolution of G0 onlyoccurs after about 25 min. This blend also shows a remark-ably lower value of the storage modulus as compared tothe other blends. It can be noted here that this blend is alsothe only one for which the elastic modulus E’ of the one-phasic blend in the glassy region showed lower values ascompared to that of the neat blend (Fig. 1), indicating weakinteractions and/or less good dispersion of the PE-MWNTsin this blend. Based on an effective concentration of PE-MWNTs of 3.3 wt% in the 60/40 PaMSAN/PMMA blend,percolation of the nanotubes and large values of G0 wouldbe expected. However, the absence thereof suggests thatMWNT network formation is clearly prohibited, whichallows coarsening to proceed much further, resulting inlarger droplets, as is confirmed by the TEM images inFig. 2c–e. The kinetics of fibril breakup processes is closelyrelated with the interfacial tension, but also matrix viscos-ity and elasticity play a role [43]. Hence, to understand the

0 1103

109 PαMSAN with NH

2-MWNTs

PαMSAN with PE-MWNTs

η∗ ωev

Conc. of MWN

0 1103

109 PMMA with NH

2-MWNTs

PMMA with PE-MWNTs

Conc. of MWN

Fig. 6. Complex viscosity (g�) and G0 at xeval (0.039 rad/s) of (a) PaMSA

effects of the different MWNTs on the fibril breakup pro-cess, the effects of these MWNTs on the rheology of theblend components should be known. This will be discussedin more detail in Section 3.5.

In summary, the phase separation process facilitates theincrease in effective concentration of MWNTs in one phaseof the blends and the viscoelastic properties are dominatedby the network of MWNTs rather than the concentrationfluctuations.

3.5. Percolation and effects of MWNTs on the phase inversionconcentration

The changes in the viscosity (and elasticity) ratios dueto selective filling above a critical value can affect thephase inversion concentration and the evolution of cocon-tinuous structures by means of fibril breakup processes[43,44]. Fig. 6 compares the complex viscosity (g*) and G0

at xeval (0.039 rad/s) for the components with MWNTs.The zero shear viscosities of PaMSAN and PMMA at220 �C are 10550 Pas and 6065 Pas respectively. It is evi-dent from Fig. 6a that the effective increase in g* and G0

of PaMSAN in case of PE-MWNTs is quite significant incontrast to NH2-MWNTs, particularly above the percola-tion threshold of the filler. For instance at 2 wt%

2 3Ts (wt%)

eval (P

' ωev

2 3Ts (wt%)

N with different MWNTs; and (b) PMMA with different MWNTs.

PE-MWNTs, the effective rise in g� (g�PaMSAN=PE-MWNTs :

2:3E8 Pa s; g�PaMSAN=NH2-MWNTs : 16070 Pa s) and G0

(G0PaMSAN=PE-MWNTs: 7679 Pa; G0PaMSAN=NH2-MWNTs: 60.3 Pa) ofthe PaMSAN phase is several orders of magnitude higheras compared to NH2-MWNTs. This can be attributed tothe larger length of the PE-MWNTs which facilitates theformation of a percolated network. However, if fillerdimensions would be the only factor, similar behaviorwould be expected for the PMMA phase (see Fig. 6b). Onthe contrary, in PMMA a higher percolation threshold forPE-MWNTs and lower values of viscosity and elasticity ascompared to NH2-MWNTs were observed. The -NH2 func-tional moieties on the MWNT surface can presumablyinteract with the AOAC@O of PMMA [45] leading to betterdispersion. Hence, the rheological percolation threshold isgoverned by a complex interplay between the intrinsiccharacteristics of the nanoparticles and the extent of inter-action between the nanoparticles and the polymer matrix.In the case of 60/40 PaMSAN/PMMA blends, the selectivelocalization of the MWNTs in the PaMSAN matrix phaseafter phase separation caused a substantial increase inthe viscosity and elasticity of this phase, especially in thecase of PE-MWNTs. This change in the rheological behaviorhas resulted in a coarser morphology and inhibitedpercolation of the MWNTs. In this context, conductivityspectroscopy measurements will provide additional

10-2 10-1 100 101 102

40/60 40/60+0.5wt% NH

2-MWNTs

40/60+1wt% NH2-MWNTs

Frequency (rad/s)

10-2 10-1 100 101 102

Frequency (rad/s)

60/40 60/40+0.5wt% NH

2-MWNTs

Fig. 7. Evaluation of the elastic criterion maximum of G0 at xeval= 0.039rad/s at 22a) for neat blends; b) for blends with MWNTs (inset showing the corresponding

information on the formation and destruction of theMWNT network in the blends during phase separation(see Section 3.6).

To show the effects of phase separation on the forma-tion of percolative MWNT networks in the different blendsmore clearly, their linear viscoelastic behavior withMWNTs after the complete phase separation i.e. after 5 hat 220 �C is shown in Fig. 7. G0 at low frequencies becomesalmost independent of the test frequency above a criticalconcentration of MWNTs which is a typical solid-like re-sponse of the nanocomposites. It has been reported thatinterconnected structures of anisometric fillers result in ayield stress [27] and a plateau in G0 and G00 manifestingthe transition from ‘liquid-like’ to ‘solid-like’ elastic behav-ior in the filled blends. The G0 as a function of frequency for40/60 blends with different MWNTs are shown in Fig. 7aand b respectively. A plateau in the low frequency regimein G0 is observed at 0.5 wt% MWNTs in 40/60 cocontinuousblends, irrespective of their type (Fig. 7a and b). Similarobservations were made above 1 wt% in 60/40 blends withNH2-MWNTs (Fig. 7c). However, for 60/40 blends withPE-MWNTs the linear viscoelastic behavior remains almostunaffected with increasing concentration of MWNTs(Fig. 7d).

The phase separation process significantly affects theoverall state of dispersion of MWNTs, which can further

10-2 10-1 100 101 102

40/60 40/60+0.5wt% PE-MWNTs 40/60+1wt% PE-MWNTs 40/60+2wt% PE-MWNTs

Frequency (rad/s)

10-2 10-1 100 101 102

Frequency (rad/s)

60/40 60/40+0.5wt% PE-MWNTs 60/40+2wt% PE-MWNTs

0 �C dependent on the volume fraction of the PaMSAN phase of the blendsmorphologies for blends with NH2-MWNTs).

0.0 0.2 0.4 0.6 0.8 1.0

φPα MSAN

Phase inversion

0.0 0.2 0.4 0.6 0.8 1.00

30000(b) blends with NH2-MWNTs

blends with PE-MWNTs

φPαMSAN

G' extra

,ωe v

Fig. 8. Evaluation of the elastic criterion maximum of G0 at xeval = 0.039 -rad/s at 220 �C dependent on the volume fraction of the PaMSAN phase ofthe blends (a) for neat blends; (b) for blends with MWNTs (inset showingthe corresponding morphologies for blends with NH2-MWNTs).

affect the viscosity and elasticity ratios in the blends. This,in its turn, has an effect on the blend morphology andhence its viscoelasticity. A quantitative estimation of theadditional viscoelasticity of the blend after phase separa-tion can help in predicting the phase inversion concentra-tion (ØPI). We followed the procedure adopted bySteinmann et al. [44] to estimate ØPI.

The extra contribution to G0 due to selective filling at agiven evaluation frequency (xeval, 0.039 rad/s) can beestimated from the following relations:

G0extra ¼ G0blend � G0mix ð10Þ

G0mix ¼ ØPaMSANG0PaMSAN þ ð1� ØPaMSANÞG0PMMA ð11Þ

In Fig. 8, the extra G0 of the blends are plotted as a functionof volume fraction of the PaMSAN phase. G0extra increasesfrom both sides and meet at a maximum, manifesting theØPI (Fig. 8a). For neat blends, the maximum occurs at60 wt% PaMSAN. According to this criterion, the ØPI shiftedtowards lower content of PaMSAN due to selective filling(see Fig. 8b). In earlier investigations [30,33], we observedthat the interconnected domains start developing in theblends even at 15 wt% dispersed phase (of PaMSAN) inpresence of PE-MWNTs and NH2-MWNTs (see insets ofFig. 8b). It is envisaged that for neat blends, the co-contin-uous structures that are formed at lower concentrations ofthe minor phase are the thinnest and are destroyed. How-ever, selective filling retards the fibril break-up process. Itis worth pointing out that interfacial tension driven coales-cence does not destroy the co-continuous structures butrather leads to coarsening of the domains. The kinetics ofthe fibril break-up process is strongly influenced by inter-facial tension and matrix viscosity (or elasticity). Above acritical value of the viscosity (or elasticity) ratio the inter-connected domains can break up completely into droplet–matrix type of morphology; often characterized by a bimo-dal distribution of the droplets [46]. These factors lead to ashift in the phase inversion concentration and the onset ofco-continuity. Similar observations were reported for SMAcompatibilized PA6/(PPE/PS) blends [46]. In case of filledblends, especially with large aspect ratio fillers, the effec-tive changes in viscosity (and elasticity) ratio dominatethe overall kinetics of the fibril break-up process as theircontribution to interfacial tension (effective) is not signifi-cant unlike for compatibilizers at the interface [47].

3.6. Electrical conductivity: selective localization of MWNTs

It is evident from the TEM images that irrespective oftheir type, MWNTs migrate to the PaMSAN phase duringphase separation, driven by thermodynamic forces. Suchselective distribution results in MWNT-rich domains inthe blends which preset an alternative strategy for moreeffective percolation. To investigate this, electrical conduc-tivity spectroscopy measurements were performed on theblends with MWNTs, both before phase separation (i.e. atroom temperature, Fig. 9a) and during phase separationat an elevated temperature (220 �C) (Fig. 9b). Interestingly,irrespective of their type, both 40/60 and 60/40 PaMSAN/PMMA blends with 2 wt% MWNTs show insulating

properties at room temperature (i.e. when the blends arehomogeneous, see Fig. 9a), which, later on reveal anincrease in the melt conductivity (at 220 �C) by severalorders of magnitude (see Fig. 9b). At an elevated tempera-ture, the blends phase separate and allow MWNTs tomigrate to the PaMSAN phase. Such migration of MWNTsduring the phase separation leads to an increase in theirlocal concentrations which assists in the formation of apercolative network-like structure as manifested fromthe plateau (dc conductivity).

As a general observation, the 40/60 PaMSAN/PMMAblends which show co-continuous morphologies have rea-sonably good electrical conductivities in the melt. Amongthe 40/60 blends, the one with 2 wt% PE-MWNTs has aconductivity which is two orders of magnitude higher thanthat of the NH2-MWNTs filled one. The higher conductivityof the 40/60 PaMSAN/PMMA blend with PE-MWNTs is inagreement with its larger storage modulus (Fig. 5) andcan be explained on the basis of the larger length of thePE-MWNTs as compared to NH2-MWNTs. In the case of60/40 PaMSAN/PMMA blends, the one with 2 wt%NH2-MWNTs shows an impressive increase in conductivity

during phase separation, reaching a final conductivityhigher than 10�3 S/cm. The 60/40 PaMSAN/PMMA blendwith PE-MWNTs exhibits much lower conductivity which,instead of increasing, even slightly decreases as the phaseseparation proceeds (see inset of Fig. 9b). A close look atthe time scales revealed that the time scale of the decreasein conductivity is in reasonably good agreement with thetime scale of the decrease in G0 during time-sweep exper-iments (see inset of Fig. 9b). As already discussed in Section3.5, a plausible explanation for both the low value of con-ductivity and its slight decrease during phase separation ofthe 60/40 sample with 2 wt% PE-MWNTs is the less pro-nounced network build up of PE-MWNTs and the morepronounced coarsening.

A cartoon (Fig. 10) further illustrates how phaseseparation in polymeric blends can be used as a template

10-2 100 102 104 10610-14

40/60 + 2 wt% MWNT-NH 60/40 + 2 wt% MWNT-NH 40/60 + 2 wt% MWNT-PE

σ AC (S

Frequency (Hz)

0 100 20010-10

100 40/60 + 2wt% MWNT-NH

2 60/40 + 2wt% MWNT-NH

Time (min)

40/60 + 2wt% MWNT-PE 60/40 + 2wt% MWNT-PE PαMSAN + 2wt% MWNT(b)

Fig. 9. Electrical conductivities (r, S/cm) of the 60/40 and 40/60 blends withseparation for 5 h at 220 �C (inset showing the magnified view (as indicated) ofMWNTs and also comparing the evolution of G0 with time for this particular com

to increase the effective concentration of MWNTs whichotherwise is difficult from processing viewpoint. Process-ing polymer nanocomposites involving higher fractionsof MWNTs becomes practically very difficult as, mostoften, the end to end distance (R) of a flexible polymerchain is greater than the surface to surface distancebetween two MWNTs (van der Waals’ distance). TheMWNTs which are randomly dispersed in a mono-phasicsystem, migrate or re-distribute upon phase separationdriven by thermodynamic forces. This leads to transitionfrom insulating materials at room temperature(one-phasic) to a highly conducting material in the melt(two-phasic) as a result of phase separation. Byquenching such morphologies, a highly conductivematerial with controlled phase microstructures can bedeveloped.

1,2E-7

1,5E-7

60/40 + 2wt% PE-MWNTs

0 50 100 150 200 250 300100

Time (min)

0 50 100 150 200 250 300Time (min)

60/40 + 2wt% PE-MWNT

2 wt% MWNTs: (a) before phase separation at 25 �C; (b) during phasethe evolution of conductivity with time for 60/40 blends with 2 wt% PE-position).

Fig. 10. Cartoon illustrating the effect of MWNTs on the resulting morphology of the blends during phase separation (the gray regions indicate the PaMSANphase; the white regions represent the PMMA phase; hairy like structures represent MWNTs).

4. Conclusions

Thermally induced phase separation in 40/60 and 60/40LCST-type blends of poly[(a-methyl styrene)-co-acryloni-trile]/poly(methyl methacrylate) (PaMSAN/PMMA) weremonitored in the presence of surface functionalised multi-wall carbon nanotubes (MWNTs) by differential scanningcalorimetry, melt rheology, conductivity spectroscopyand microscopic techniques. The temperature interval inwhich phase separation occurred is independent of thetype and amount of MWNTs added. The evolution ofdynamic moduli (G0) as a function of temperature was usedas a probe to investigate the structural development as thesystem transits from the mono-phasic to the bi-phasicregime. The rheological phase separation temperature(Trheo) was observed to be shifted to lower temperaturesfor 60/40 and 40/60 blends filled with MWNTs.

The phase separation kinetics was investigated in pres-ence of two different types of MWNTs by isochronal time-sweep measurements at a deep quench depth in the unsta-ble region. The evolution of the viscoelastic properties wasdominated by the network formation of MWNTs after theirselective localization in the PaMSAN phase and did not re-veal the time dependence of the phase separation process.

Electron microscopic images revealed that the phaseseparation resulted in selective localization of MWNTs inthe PaMSAN phase of the blends which were initially

randomly distributed in the mono-phasic materials. Thisled to significant changes in the viscosity (and the elastic-ity) ratios in the blends which further resulted in a shift inthe phase inversion concentration towards lower PaMSANcontent. The resulting morphology of the blends was foundto be strongly governed by the type of MWNTs which inturn, affected the overall state of dispersion of MWNTs,more specifically the ‘network-like’ structure. This wassupported by both time sweep experiments and conductiv-ity spectroscopy measurements.

The phase separation induced selective localization ofMWNTs in the blends also resulted in a dramatic transitionfrom an insulating one-phasic material at room tempera-ture to a highly conducting material in the melt. Byquenching such morphologies, effective percolation ofMWNTs can be achieved at room temperature in contrastto filled homopolymers or mono-phasic materials.

Acknowledgements

Onderzoeksfonds KU Leuven (Postdoctoral Fellowshipfor S. Bose), Agency for Innovation by Science and Technol-ogy – IWT (Postdoctoral Fellowship for C. Özdilek) and Re-search Foundation Flanders – FWO (PostdoctoralFellowships for J. Leys and R. Cardinaels) are gratefullyacknowledged for financial support. We would also liketo thank Mr. Danny Winant, Dep. of Metallurgy and Mate-

rials Engineering, KU Leuven for his kind assistance in dy-namic mechanical thermal analysis and Jan Vandenbosschefor his contributions to the rheological part of the work.

References

[1] Nesterov AE, Lipatov YS, Horichko VV, Gritsenko OT. Polymer1992;33:619–22.

[2] Li W, I’Abee RMA, Goossens JGP. Macromol Chem Phys2013;214:2705–15.

[3] Lipatov YS, Nesterov AE, Ignatova TD, Nesterov DA. Polymer2002;43:875–80.

[4] Ginzburg VV. Macromolecules 2005;38:2362–7.[5] Laradji M, MacNevin G. J Chem Phys 2003;119:2275–84.[6] Ginzburg VV, Qiu F, Paniconi M, Peng G, Jasnow D, Balazs AC. Phys

Rev Lett 1999;82:4026–9.[7] Du M, Wu Q, Zuo M, Zheng Q. Eur Polym J 2013;49:2721–9.[8] Xavier P, Bose S. J Phys Chem B 2013;117:8633–46.[9] Balazs AC, Ginzburg VV, Qiu F, Peng G, Jasnow D. J Phys Chem B

2000;104:3411–22.[10] Sharma K, Sharma M, Chandra A, Bose S. Macromol Chem Phys

2013;214:2651–69.[11] Gharachorlou A, Goharpey F. Macromolecules 2008;41:3276–83.[12] Huang C, Gao J, Yu W, Zhou C. Macromolecules 2012;45:8420–9.[13] Gao J, Huang C, Wang N, Yu W, Zhou C. Polymer 2012;53:1772–82.[14] Vleminckx G, Bose S, Leys J, Vermant J, Wübbenhorst M, Abdala AA,

et al. ACS Appl Mat Interfaces 2011;3:3172–80.[15] Yurekli K, Karim A, Amis EJ, Krishnamoorti R. Macromolecules

2003;36:7256–67.[16] El Mabrouk K, Bousmina M. Rheol Acta 2006;45:877–89.[17] Dreezen G, Groeninckx G, Swier S, Van Mele B. Polymer

2001;42:1449–59.[18] Zuo M, Peng M, Zheng Q. Polymer 2005;46:11085–92.[19] Jin J, Song M. Thermochim Acta 2005;426:151–5.[20] Kapnistos M, Hinrichs A, Vlassopoulus D, Anastasiadis SH, Stammer

A, Wolf BA. Macromolecules 1996;29:7155–63.[21] Carreau PJ, Bousmina M, Ajji A. Rheological properties of blends:

Facts and Challenges. In: Progress in pacific polymer science 3, 1994,Springer-Verlag, Berlin pp. 25–39.

[22] Mani S, Malone MF, Winter HH. J Rheol 1992;36:1625–49.

[23] Ajji A, Choplin L, Prud’homme RE. J Polym Sci Part B: Polym Phys1991;29:1573–8.

[24] Vinckier I, Laun H. M. Rheol Acta 1999;38:274–86.[25] Polios IS, Soliman M, Lee C, Gido SP, Schmidt-Rohr K, Winter HH.

Macromolecules 1997;30:4470–80.[26] Kim JK, Son HW, Lee Y, Kim J. J Polym Sci Part B: Polym Phys

1999;37:889–906.[27] Bose S, Khare RA, Moldenaers P. Polymer 2010;51:975–93.[28] McNally T, Pötschke P. Polymer carbon nanotube composites:

preparation, properties and applications. Woodhead PublishingLimited; 2011. p. 848.

[29] Baudouin AC, Devaux J, Bailly C. Polymer 2010;51:1341–54.[30] Bose S, Özdilek C, Leys J, Seo JW, Wübbenhorst M, Vermant J, et al.

ACS Appl Mat Interfaces 2010;2:800–7.[31] Laun HM. Pure Appl Chem 1998;70:1547–66.[32] Madbouly SA, Ougizawa T. J Macromol Sci Part B Phys

2002;B41:271–87.[33] Özdilek C, Bose S, Leys J, Seo JW, Wübbenhorst M, Moldenaers P.

Polymer 2011;52:4480–9.[34] http://nanocyl.com.[35] Nuriel S, Liu L, Barber AH, Wagner HD. Chem Phys Lett

2005;404:263–6.[36] Barber AH, Cohen SR, Wagner HD. Phys Rev Lett 2004;92. 186103/1-4.[37] Wu S. Interfacial and surface tensions of polymer melts and liquids.

In: Polymer interface and adhesion, 1982, Marcel Dekker, New York,pp. 67-129.

[38] Pötschke P, Pegel S, Claes M, Bonduel D. Macromol Rapid Commun2008;29:244–51.

[39] Leys J, Wübbenhorst M, Menon CP, Rajesh R, Thoen J, Glorieux C,et al. J Chem Phys 2008;128. 064509/1-7.

[40] Molnar A, Eisenverg A. Macromolecules 1992;25:5774–82.[41] Altstädt V, De Lucca Freitas L, Schubert DW. Pure Appl Chem

2004;76:389–413.[42] Fredrickson GH, Larson RGJ. Chem Phys 1987;86:1553–60.[43] Tol RT, Groeninckx G, Vinckier I, Moldenaers P, Mewis J. Polymer

2004;45:2587–601.[44] Steinmann S, Gronski W, Freidrich C. Polymer 2003;43:4467–77.[45] Zhang K, Lim JY, Choi H. J Diamond Related Mater 2009;18:316–8.[46] Utracki LA. Polym Comp 1986;7:274–82.[47] Levine S, Bowen BD. Colloids Surfaces A Physicochem Eng Aspects

1993;70:33–45.

Effect of multiwall carbon nanotubes on the phase separation of concentrated blends of...

Documents

Transcript of Effect of multiwall carbon nanotubes on the phase separation of concentrated blends of...

POLY-USAGE EPS GEOFOAM - polymos.com - Poly-Usage.pdf · POLY-USAGE EPS GEOFOAM Minimum compression RESISTANCE (σ) of 93 kPa to 275 kPa (Depending on type of block, when tested at

V34 methyl acetate

Anti-5-Methyl Cytidine (5MC) made in

Poly(lactide-co-ε-caprolactone) · Electronic Supplementary Information Poly(lactide-co-ε-caprolactone) Copolymers byBis-Thioetherphenolate Group 4 Metal Complexes: Synthesis, Characterization

Nano thick poly (ε-caprolactone)-poly (ethylene glycol ... · †Electronic Supplementary Information (ESI) Nano thick poly (ε-caprolactone)-poly (ethylene glycol) coatings developed

Melt electrospinning writing of poly-Hydroxymethylglycolide- co … · pHMGCL-fibre scaffolds in comparison to electrospun poly(ε-caprolactone)-based scaffolds. Overall, this study

Poly(butylene succinate-co–caprolactone) Copolyesters ...

Production of Poly(-Caprolactone)/Hydroxyapatite Composite ...

Endomorphin-1 analogs containing α-methyl-β-amino acid ... ·

Investigation of poly(γ-glutamic acid) production via ...

Poly-β-hydroxybutyrate and glycogen metabolism in ...€¦ · Poly-β-hydroxybutyrate and glycogen metabolism in Rhizobium leguminosarum E.M.Lodwig1, K. Findlay2, A.J.Downie2, ...

Methyl 4-O-(2-chlorobenzoyl)-α-L-rhamnopyranosides ...

ε-Poly-l-Lysine Peptide Chain Length Regulated by the ... · ε-Poly-L-Lysine Peptide Chain Length Regulated by the LinkersConnecting the Transmembrane Domains of ε-Poly-L-LysineSynthetase

Poly SA2005 2006 - unis.ens-lyon.fr

Highly Branched Poly(α-Methylene-γ-Butyrolactone) from …file.scirp.org/pdf/OJPChem_2017112914172525.pdf · 2017-12-01 · ... (3.00 g, 0.013 mol), and L-valinol [(S)-(+)-2-Amino-3-methyl-1-butanol]

Inducing Phase Crystallinity in Block Copolymers of ...€¦ · Inducing Phase Crystallinity in Block Copolymers of Vinylidene Fluoride with Methyl Methacrylate or Styrene Nahal Golzari

Polydisperse methyl β-cyclodextrin–epichlorohydrin ... · PDF file2785 Polydisperse methyl β-cyclodextrin–epichlorohydrin polymers: variable contact time 13C CP-MAS solid-state

Metabolic Planar Imaging Using 123I-β-Methyl-Iodophenyl ......113 Metabolic Planar Imaging Using 123I-β-Methyl-Iodophenyl Pentadecanoic Acid Identifi es Myocardial Ischemic Memory

Fine-tuning Thermoresponsive Functional Poly(ε ...

ε-caprolactone)/Poly(3-hydroxybutyrate- hydroxyvalerate ...