Materials Chemistry and Physics 164 (2015) 206e213
Contents lists available at ScienceDirect
Materials Chemistry and Physics journal homepage: www.elsevier.com/locate/matchemphys
Electrical percolation and crystallization kinetics of semi-crystalline polystyrene composites filled with graphene nanosheets Chi Wang a, *, Yen-Chang Chiu a, Chien-Lin Huang b a b
Department of Chemical Engineering, National Cheng Kung University, Tainan 701, Taiwan Department of Fiber and Composite Materials, Feng Chia University, Taichung 407, Taiwan
h i g h l i g h t s
g r a p h i c a l a b s t r a c t
Composites of sPS/GNS and aPS/GNS have been compared. sPS/GNS composites have a higher percolation threshold for electrical conductivity. Composites containing GNS with a larger aspect ratio have a lower percolation threshold. To enhance sPS crystallization, 1D CNT is more effective than 2D GNS.
a r t i c l e i n f o
a b s t r a c t
Article history: Received 8 February 2015 Received in revised form 6 August 2015 Accepted 19 August 2015 Available online 28 August 2015
Syndiotactic polystyrene (sPS) is a semi-crystalline polymer with high melting temperature and good mechanical strength. Composites of sPS filled with different contents of graphene nanosheets (GNS) are prepared by coagulation method. Two types of GNS with different thicknesses (denoted as G1 and G10) are studied to unveil the effect of aspect ratio on electrical conductivity and crystallization kinetics of the composite. Atomic force microscopy and transmission electron microscopy (TEM) show that G1 is a wrinkled sheet with an average thickness of ~2 nm and that G10 is a smooth flake with a thickness of ~50 nm; both possess a similar basal dimension of ~5 mm. The percolation thresholds for electrical conductivity (4c) of the G1-filled and G10-filled composites are 0.46 and 3.84 vol%, respectively. At a given GNS content, the electrical conductivity of the G1-filled composites is higher than that of the G10filled composites. Both findings are attributed to the larger GNS aspect ratio of G1 compared with G10. The deduced 4c of the G1-filled composites is significantly larger than that of GNS-filled amorphous atactic PS composites, indicating that the crystallizability of the matrix has an important influence on formation of GNS networks. Both G1 and G10 nanofillers are found to be good nucleating agents for the heterogeneous nucleation of sPS. Because of its wrinkled surface, G1 is less effective than G10 in inducing sPS crystallization. Compared with 2D sheet-like GNS, 1D CNTs are more effective in enhancing sPS crystallization through surface-induced nucleation as well as the chain-tube wrapping behavior in the sPS/CNT composites. © 2015 Elsevier B.V. All rights reserved.
Keywords: Composite materials Polymers Crystallization Electrical conductivity X-ray scattering
1. Introduction * Corresponding author. E-mail address:
[email protected] (C. Wang). http://dx.doi.org/10.1016/j.matchemphys.2015.08.046 0254-0584/© 2015 Elsevier B.V. All rights reserved.
Syndiotactic polystyrene (sPS) is an engineering thermoplastic with an apparent melting temperature of ~270 C and a
C. Wang et al. / Materials Chemistry and Physics 164 (2015) 206e213
rapid crystallization rate. By contrast, atactic polystyrene (aPS) is an amorphous thermoplastic with a service temperature lower than its glass transition of ~100 C. In composite applications, the thermal stability of sPS composites is much better than that of aPS composites. Depending on the crystallization conditions, sPS chains can crystallize into at least four crystalline modifications (a, b, g, and d forms) because of their polymorphic nature [1]. The mechanical strength of sPS samples is dependent on the crystalline modification and degree of crystallinity developed within the sample. Polymer composites containing nanoscaled fillers have elicited considerable attention in the field of nanotechnology. Among nanoscaled fillers, carbon nanotubes (CNTs) and their associated materials, such as fullerene and graphene nanosheets (GNS), are the most promising for advanced applications because of their exceptional thermal, mechanical, and electrical properties [2e7]. In general, CNT and graphene have large aspect ratios but different apparent geometries. CNTs are considered as 1D carbon nanoparticles, whereas GNSs are considered as 2D carbon nanoparticles. In composites, agglomeration of GNS is likely to occur because of the large area-tovolume ratio of the nanofiller. To characterize filler dispersion, a percolation scaling law is adopted to determine the minimum filler content (threshold), above which a significant enhancement of physical properties is expected [8,9]. For practical applications, electrical conductivity is the primary focus of property characterization in sp2-carbon-filled polymer composites because only a small amount of CNT (or GNS) is sufficient to develop a conductive percolation path for enhanced electron transport within the insulating polymer matrix. These GNS-filled polymer composites show promising applications in electrostatic discharge and electromagnetic interference-shielding materials [4]. According to Stankovich et al. the lowest electrical conductivity threshold (4c) of aPS/GNS composites is 0.1 vol% [4]. Despite extensive studies on the amorphous composite of aPS/ GNS, the 4c of semi-crystalline sPS/GNS composites has yet to be studied. We are interested to determine whether or not the crystallizability of the matrix (sPS and aPS) will affect the 4c of GNS-reinforced composites. Our previous studies on CNT-filled composites of sPS and aPS [10,11] have disclosed that the overgrowing crystals at the CNT surface leads to higher 4c. In this study, we aim to distinguish whether 2D GNS or 1D CNT is more effective in structuring the nanofiller network for electrical conductivity, and which has a lower 4c for the sPS matrix. The addition of GNS improves the electrical properties of polymers; however, it may also alter the crystalline modification, morphology, and crystallization kinetics of the polymers used. Crystallization and morphologies of GNS-filled polyethylene [12] and isotactic polypropylene [13] have been investigated. Results of these studies show that GNSs, like CNTs, act as effective nucleating agents and increase the overall crystallization rate significantly. Interestingly, 2D GNSs can accelerate the crystallite growth kinetics of poly(L-lactide) (PLLA), although their effect is less pronounced than that of 1D CNTs [14]. We have reported on the influence of CNT [11] and carbon nanocapsules (CNC) [15] on crystal morphology and crystallization kinetics of sPS; both nanofillers serve as good nucleating agents. To date, the crystallization kinetics of GNS-filled sPS composites has not been explored, and the mechanism underlying the nucleating ability of GNS remains unclear. In addition, whether or not 2D GNSs are a better nucleating agent for sPS chains than 1D CNTs for later crystal growth remains unknown. As part of our series of investigations on nanocarbonreinforced sPS composites [10,11,15], the present study investigates in detail the effect of GNS with two different aspect ratios on crystal modification, crystallization kinetics, and melting behavior of the sPS matrix.
207
2. Experimental 2.1. Materials and composite preparation The sPS pellets were obtained from Dow Chemical Co. The weight-average molecular weight and polydispersity were 2.49 105 g/mol and 2.37, respectively. Two types of GNS with different thicknesses, coded by N002-PDR and N006-P, were purchased from Angstron Materials, LLC. According to the manufacturer, both fillers have a density of 2.20 g/cm3; the former has an average thickness lower than 1 nm and was denoted as G1, whereas the latter has an average thickness of 10 nme20 nm and was denoted by G10. Both G1 and G10 were used without any surface treatment. In producing well-dispersed GNS-filled sPS composites by solution blending, ortho-dichlorobenzene (o-DCB) is more appropriate than other solvents because of its higher solubility for sPS chains and the pep interactions between the graphene surface and the phenol group [16]. After sonication, GNSs in the sPS/o-DCB solution exhibited a uniform dispersion for days without visible precipitation or aggregation on the vial bottom. Thus, sPS/GNS composites with various compositions were prepared by dissolving the sPS pellets in o-DCB solvent at 140 C and adding GNS to the homogeneous solution. This step was followed by sonication for 3 h. Subsequently, the uniform suspension with 1% (w/v) was precipitated dropwise into a 20-fold excess volume of methanol. Recovered powders were dried continuously in a vacuum oven until the residual solvent was removed. In this work, the code 99/1 represents the weight ratio of sPS to GNS, equivalent to a GNS weight fraction (4w) of 0.01. The volume fraction of nanofillers (4v) was calculated using the respective density of GNS and sPS (~1.06 g/ cm3). Melt-quenched samples 20 mm thick were obtained by holding the dried powders at 300 C for 10 min in a hot stage (THMS600, Linkam), followed by quenching in liquid N2. 2.2. Composite characterization Raman spectroscopy was used to probe the chemical difference between G1 and G10. The Raman spectra of the as-received G1 and G10 nanofillers were obtained using DXR Raman Microscope (ThermoScientific). The surface and dimensions of the G1 and G10 particles were investigated through atomic force microscopy (AFM, MMAFM-2, Digital Instruments). Tapping mode imaging was conducted under ambient conditions. Suspensions of GNS were prepared in o-DCB at 0.5 mg/mL. After sonication at room temperature for 10 min or 3 h, the suspensions were deposited onto a mica substrate and a carbon-coated copper grid, respectively. The former was used for AFM study to measure the GNS dimensions, whereas the latter was used for the TEM observation to reveal the GNS morphology. The TEM micrographs shown in this work were obtained by using a Jeol JEM-2000FX electron microscope operated at 80 kV. To reveal the microstructure of GNS-filled sPS composites under TEM, ultrathin films (ca. 50 nm) were prepared by sectioning the samples at room temperature with an Ultracut UCT (Leica) microtome. The sPS lamellar morphology was observed by staining the ultrathin films to enhance the contrast between the amorphous and crystalline layers. Staining was carried out with ruthenium tetraoxide (RuO4) vapor at room temperature. WAXD intensity profiles were obtained using a Bruker diffractometer (NanoSTAR Universal System, Cu Ka radiation). The crystallization and melting behavior of sPS/GNS composites were investigated using a PerkinElmer DSC7 under N2 atmosphere. The melt-quenched samples were heated to 300 C at 10 C/min for nonisothermal crystallization to reveal the cold-crystallization kinetics (first heating scan). After holding at 300 C for 10 min, the samples were cooled to room temperature at a rate of 10 C/min to disclose the melt-
208
C. Wang et al. / Materials Chemistry and Physics 164 (2015) 206e213
crystallization kinetics (first cooling scan) [15]. For isothermal crystallization, the precipitated powders were first held at 300 C for 10 min and then rapidly cooled to the desired crystallization temperature (Tc) until complete phase transformation was reached. A subsequent heating trace was obtained at a heating rate of 10 C/ min to determine their melting behavior. Samples for electrical conductivity measurements were prepared by hot pressing the precipitated powders in a rectangular steel mold (thickness: 0.30 mm) at 290 C for 3 min and then air cooling to room temperature. Measurements of samples with high electrical conductivity (>104 S/m) were carried out using a Keithley 2400 sourcemeter. The standard four-probe technique was applied to reduce the effects of contact resistance. A Keithley 6487 electrometer equipped with a Keithley 8009 resistivity fixture was used for samples with low conductivity (<104 S/m) following the ASTM D257 standard.
3. Results and discussion 3.1. GNS characterization SEM shows that the as-received GNSs of G1 and G10 are thick, with numerous stacking graphene layers. Moreover, the Raman spectra reveal that the G and D bands at 1573 and 1348 cm1 of the G10 nanofiller are sharp/strong and weak, respectively (Fig. 1). The G band at 1582 cm1 of the G1 nanofiller is not only broad but also smaller than the D band at 1341 cm1. The intensity ratios of the G/ D band are approximately 0.73 and 11.9 for the G1 and G10 nanofillers, respectively. Thus, the G1 nanofillers possess more defects than the G10 nanofillers; these defects are associated with the inplane vibration graphene layers as indicated by the lower G/D band ratio. Sonication of GNS suspension in o-DCB with concentration of 0.5 mg/mL was carried out at various times to effectively separate the stacking graphene layers in the as-received GNS. However, significant overlays of graphene are still observed after a short period (10 min, upper row) of sonication, as shown by the TEM images in Fig. 2. The G1 nanofillers contain numerous wrinkled features on their surfaces, whereas the G10 nanofillers possess smooth and uniform surfaces. After 3 h of sonication, layer separation of G1 and G10 is significantly improved. Most GNS are individually observed, whereas several GNS inevitably collapse to overlay one another after dispersion in o-DCB, followed by solvent removal during TEM sample preparation. Lateral and thickness dimensions of GNS were analyzed using AFM. The thickness of G10 averaged over 20 unaggregated particles is ~50 nm, which is larger than the datum supplied by the manufacture (ca. 10 nme20 nm). The G1 nanofillers show different thicknesses even within a single sheet because of their wrinkled and buckled nature, with a minimum thickness of 1.8 nm. The lateral dimension of G1 ranges from
Fig. 1. Raman spectra of G1 and G10 nanofillers.
1 mm to 3 mm, whereas that of G10 ranges from 3 mm to 8 mm. Our AFM results, except for the thickness of G10 nanofillers, agree with those provided by the manufacturer. Thus, two types of GNS with different aspect ratios, ~103 for G1 and ~102 for G10, are available for further investigations. 3.2. Electrical conductivity of GNS-filled sPS composites The performance of the polymer composites depends strongly on the state of filler dispersion in the matrix. TEM images of microtomed thin films show that G1 nanofillers disperse more uniformly than G10 nanofillers in the sPS composites (Fig. 3), probably because of their pronounced difference in aspect ratio. Although electron microscopy may provide local information on the composites, electrical conductivity data, coupled with percolation laws, are regarded as more appropriate for describing the global dispersion of GNS in solid state. The percolation threshold, used to characterize the dispersion state of fillers, is the filler content above which the formation of a filler-associated network is developed. For a given polymer/filler pair, a better filler dispersion leads to a lower percolation threshold. Fig. 4 shows the electrical conductivity (s) of the composites with various GNS contents. The insulating sPS matrix has a s of ~1016 S/m. After slight G1 addition of 0.48 vol%, s increases to 107 S/m. At a G1 loading of 0.61 vol%, s becomes 9 105 S/m, which is approximately two orders higher than the criterion (~106 S/m) used for antistatic thin films [4]. As the G1 content is further increased to 1 and 2.5 vol%, s increases to ~1 and 82 S/m, respectively. The final saturated value approaches ~102 S/m, which is suitable for conductive material applications. The pronounced increase in s is attributed to the formation of a GNS network at an extremely low content of G1, which facilitates electron transportation. For the G10-filled composites, a higher GNS content is required to form the conducting network; at a loading of 4 vol%, s is increased to 2 107 S/m. Further addition of G10 mildly increases the electrical conductivity. Finally, s becomes saturated at 2 S/m after a G10 loading of 9.6 vol%. At a fixed GNS content, the s of G1-filled composites is higher than that of the G10 containing samples although more defects are found in the G1 nanofillers. In agreement with theoretical calculation, these results validate that the aspect ratio of GNS, rather than its intrinsic nature, is important in determining the electrical conductivity of composites. The s of the composites increases with increasing GNS aspect ratio [17,18]. The details for a logical explanation are provided in the following. Since sPS matrix is an electrical insulator, the s of GNS-filled composite depends mainly on the intrinsic properties of GNS as well as the GNSeGNS network structure developed in the samples. To evaluate the quality of the GNSeGNS network for conduction, the specific GNS contact area (interfacial GNS area per unit volume of composite) has to be considered. In other words, a network structure with more GNSeGNS junction area provides more conduction paths for electrons, giving rise to a higher s of the composite. G1 possesses a wrinkled surface and more defects with a sp3 structure (Figs. 1e3). This may imply that individual G1 flakes have a lower intrinsic s than individual G10 sheets. Nevertheless, G1 flakes possess an aspect ratio 10 times larger than G10 sheets (Table 1), indicating that the specific GNS contact area in the G1filled composites is higher. Our results show that above the percolation threshold the saturated electrical conductivity of sPS/ G1 composites is higher than that of sPS/G10 composites (Fig. 4). It indicates that the weighting factor of the GNSeGNS network structure outperforms the intrinsic properties of GNS. Nevertheless, the intrinsic properties of GNS may be prevailing provided that sPS becomes the dispersed phase. This issue is beyond the scope of this study and deserves future work. The percolation scaling law is applied to quantitatively estimate
C. Wang et al. / Materials Chemistry and Physics 164 (2015) 206e213
209
Fig. 2. TEM micrographs of GNS of G1 (left column) and G10 (right column) after sonication treatment in o-DCB solvent for 10 min and 3 h.
Fig. 3. TEM micrographs of melt-quenched composites; (a), (b) sPS/G10 ¼ 99/1, and (c), (d) sPS/G1 ¼ 99/1. Samples are in absence of RuO4 staining to clearly reveal the dispersion of fillers, and filler morphologies.
the minimum GNS concentration required to enhance the s of sPS composites (inset of Fig. 4). Good linear fittings are achieved from the logelog plots; the 4c of the G1-filled and G10-filled composites are 0.46 and 3.84 vol%, respectively. However, the G1-filled and G10-filled composites possess similar exponents, namely, 4.71 and 4.86, respectively. These exponents are higher than those of sPS/ CNT composites (~2.87) [10], suggesting the formation of a GNS network that is more effective in enhancing electron conduction. As expected, sPS composites containing G1 nanofillers with higher aspect ratio possess lower 4c, which agrees with theoretical prediction [17,19]. Zhang et al. [20] investigated the electrical conductivity of poly(ethylene terephthalate) (PET) composites filled with different amounts of graphene and graphite. They obtained a scaling exponent of 4.22. In addition, the PET/graphite composites possess a higher 4c than the graphene-filled PET, which can be attributed to the larger aspect ratio of the latter. Huang et al. [21]
have studied PLLA/CNT composites and concluded that matrix crystallization plays a role in the electrical properties. For the PLLA composites with a CNT concentration lower than the percolation threshold, the electrical conductivity is increased with increasing PLLA crystallinity. For the present composites studied, the degree of crystallinity of sPS is similar since the melting enthalpy derived from the DSC heating trace is ca. 30.7 J/g for all composites. Thus, the enhancement of electrical conductivity with increasing GNS content is due to the formation of a GNS network, and not the crystallinity of sPS matrix. 3.3. Effects of matrix and filler on conductive threshold of PS composites For the present sPS/G1 system, the low 4c of 0.46 vol% indicates a good G1 dispersion in the sPS matrix by the simple coagulation
210
C. Wang et al. / Materials Chemistry and Physics 164 (2015) 206e213
Fig. 4. Electrical conductivity of the sPS/G1 (triangle symbols) and sPS/G10 (circle symbols) composites as a function of filler volume fraction. The inset shows the percolation scaling law between s and (4v 4c) for the composite solids at 25 C. Solid lines in the inset are based on the fitting of the measured data to the effective conductivity equation: s~(4v 4c)b. Fitting parameters are also displayed in the inset.
method. An even lower 4c of ~0.1 vol% has been observed in GNS composites with amorphous aPS matrix using a similar approach [4]. Apparently the crystallizability of a polymer matrix influences the formation of a GNS network for electrical conduction. A similar matrix effect has been found on sPS/CNT and aPS/CNT composites [10], which possess 4c values of 0.49 and 0.12 vol%, respectively, although they have a similar “rheological threshold” of ~0.10 vol% as determined from dynamic rheological measurements on the composite melts. These findings suggest that the GNS (and CNT) network is altered during sPS crystallization. GNS serves as an effective nucleating agent for sPS chains. During solidification from the melt, the average separation distance between GNS nanofillers is gradually enlarged because of pronounced crystal growth of sPS from the GNS surface. For a given matrix, a similar 4c can be obtained regardless of whether 1D CNT or 2D GNS is applied provided that the aspect ratio is sufficiently large. 3.4. Non-isothermal crystallization of sPS/GNS composites The precipitated composite powders were first melted at 300 C for 10 min and then quenched into liquid N2 to prepare meltquenched samples for a nonisothermal crystallization study. Fig. 5 shows the WAXD intensity profiles of the melt-quenched composites. The sPS crystalline phase is readily developed in G1-filled composites with 4w higher than 1.0 wt%. For the G10-filled composites, crystalline reflections are clearly discernible at an even lower loading of 0.5 wt%. The signature reflections at 2q ¼ 6.34 , 10.7, 12.59 , and 13.77 show that the crystalline modification developed in the melt-quenched composites is in the orthorhombic b form [1]. In addition to diffraction peaks relevant to b-form sPS crystals, a small peak at 2q ¼ ~27.16 is also detected for the sPS/G10 99/1 and 95/5 composites. This diffraction peak is associated with
the (002) d-spacing of GNS. The absence of GNS layer diffraction in the G1-filled composites (even at a G1 loading of 5 wt%) confirms that only a few graphene layers exist in the G1 nanofiller. After deconvolution of the diffraction curves [1], the crystalline fraction (4WAXD) is estimated from the ratio of the integrated intensities from all the crystalline peaks to the total intensity curve. The 4WAXD values of the G1-filled composites with filler contents of 0.5, 1.0, and 5.0 wt% are 3.5%, 11.5%, and 27.5%, respectively. The 4WAXD values of the G10-filled composites with filler contents of 0.5, 1.0, and 5.0 wt% are 11.2%, 14.2%, and 28.1%, respectively. The amount of b-form crystals increases with increasing GNS content. Moreover, G10 is slightly more effective than G1 in producing b-form crystals under rapid cooling. Enhanced sPS crystallization has also been observed in sPS/CNT composites obtained by using the same experimental procedure [11]. These results indicate that GNS and CNT are effective nucleating agents in inducing the crystallization of sPS even under extreme cooling conditions. By comparison, melt-quenched CNT-filled sPS develops a 4WAXD of 11% at a CNT concentration of 0.1 wt% [11], suggesting that 1D CNT is more effective than 2D GNS in enhancing the crystallization of the sPS matrix. The crystal morphology developed in the melt-quenched composites was observed by TEM of ultrathin films sectioned from the bulk sample, followed by RuO4 staining to enhance the imaging contrast. TEM micrographs of the sPS/G10 99/1 composite are shown in Fig. 6. In (a), the light and dark regions are associated with the G10 nanofillers and sPS matrix, respectively. In (b), on top of the selected G10 nanofiller, sPS lamellae are observed to cover the G10 surface. In (c), edge-on lamellae are identified in the absence of lamellar stacks. Moreover, the lateral dimensions of sPS lamellae are short, and no preferential growth direction is detected. These lamellar growth characters are extremely different from those of neat sPS crystallized in the b-form [15], exhibiting long and parallel stacks of lamellae. The DSC heating curves of the melt-quenched sPS/G10 composites are shown in Fig. 7a. The glass transition temperature (Tg) is determined from the mid-point of the heat capacity jump; the peak temperature of the cold crystallization is referred to as Tp,cc, and the crystallization enthalpy determined from the integral area of the exotherm is denoted as DHcc. Cold crystallization produces only aform crystals regardless of the presence of b-form crystals prior to heating [11]. We observed a single melting endotherm with a peak temperature at ca. 269.0 C and a normalized melting enthalpy of ~30 J/g, although mixed crystals were developed prior to melting (i.e., pure a-form for the neat sPS and nearly pure b-form for the 5 wt% composite). However, the melting curve becomes sharper as the content of b-form crystals is increased in composites with higher GNS content. A shallow exotherm can observed in the 95/5 composites before melting, suggesting recrystallization. After being held at 300 C for 10 min, the composite melts were cooled at a rate of 10 C/min to 30 C to study the melt crystallization. The cooling curves are shown in Fig. 7b, with the crystallization peak temperature denoted by Tp,mc. A subsequent heating scan of the meltcrystallized samples shows the melting of the sPS b-form
Table 1 Carbon nanofiller effects on the electrical conductivity percolation and crystallization kinetics of sPS composites. Carbon filler type
2D 2D 1D 1D
GNS GNS CNT CNC
Thickness (nm)
1.8 50 12 40
Aspect ratio
~103 ~102 ~102 ~6
Crystallizationa
Conductivity 4c (vol%)
b (e)
k (min1)
0.46 3.84 0.49 6.13
4.71 4.86 2.87 2.42
2.83 3.41 3.55 2.38
103 102 101 101
Source n (e) 2.80 1.98 2.06 2.11
This work, G1 filler This work, G10 filler Refs. [10,11] Refs. [10,15]
a The parameters of k and n are the crystallization rate and Avrami exponent of 1 wt% composites crystallized isothermally at 256 C. For the neat sPS sample, k and n are 7.68 104 min1 and 2.80, respectively.
C. Wang et al. / Materials Chemistry and Physics 164 (2015) 206e213
211
Fig. 5. WAXD profiles of melt-quenched (a) sPS/G1, (b) sPS/G10 composites with different filler concentrations (wt.%): a, 0; b, 0.05; c, 0.1; d, 0.5; e, 1.0; and f, 5.0. The reflection peaks at 2q ¼ 6.34, 10.7, 12.59 and 13.77 are due to the (020), (110), (040) and (130) planes of orthorhombic b form of sPS crystalline modification. The composites are prepared by holding at 300 C for 10 min, followed by rapidly quenching into liquid nitrogen.
(Fig. 7c), where all sPS composites exhibit dual-melting peaks. The high melting peak centers at ca. 270.0 C regardless of the G10 content, but the low melting peak gradually moves to a high temperature when the G10 content is increased. A small exotherm can be found between the two melting peaks for composites with 4w lower than 0.1 wt%, verifying that the dual melting behavior is attributed to the meltingerecrystallizationeremelting phenomenon [1]. Thus, the low melting temperature is due to the lamellae developed prior to crystal melting, and the high melting peak is associated with the recrystallized and/or thickened lamellae. Evaluation of Tp,mc indicates that composites with higher G10 content experience a lower degree of supercooling as the equilibrium melting temperature of the b-form crystals remains intact (to be discussed later). Consequently, the thicker lamellae that melt at higher temperatures are produced during heating. A similar filler effect on the melting behavior of CNT-filled sPS composites has been reported [11]. The measured thermal parameters (Tg, Tp,cc, DHcc, and Tp,mc) are plotted in Fig. 8 as a function of G10 content, together with those obtained from sPS/G1 composites, for a detailed comparison. DHcc was normalized with the sPS weight fraction to represent the crystallizability of the sPS matrix in the presence of GNS. No significant variation in Tg was observed regardless of the GNS content. G1-filled composites possess a slightly higher Tg than G10-filled composites, but the difference vanishes at a GNS loading of 5 wt%. A low G10 content of 0.05 wt% reduces Tp,cc by ~3 C, thereby enhancing the cold crystallization kinetics. A maximum Tp,cc reduction of ~10 C can be observed in the composites with 4w of 5 wt%. Considering that Tg is unchanged, we can attribute the enhanced cold crystallization to the increase in nucleation sites and
Fig. 6. TEM micrographs of melt quenched sPS/G10 ¼ 99/1 composites with different magnifications of (a) 30 K, (b) 100 K, (c) 250 K. Image (c) is the enlargement of the selected area in image (b), and image (b) is the enlargement of the selected area in image (a). Samples are stained with RuO4 for 15 min at room temperature to reveal the lamellar morphology of sPS. In (a), the light and dark areas are associated with the GNS and sPS matrix, respectively. In (b), on top of a selected GNS, sPS lamellae are observed to cover the GNS surface. In (c), edge-on lamellae are evidently identified. The lateral dimensions of lamellae are short, and no preferential growth direction is observed.
to the better thermal conduction of the composites. By contrast, the G1-filled composites show a milder decrease in Tp,cc possibly because of the slightly higher Tg. As expected, sPS crystallizability decreases with increasing GNS content because some crystallites have already developed in the sample prior to heating. As shown in Fig. 8d, Tp,mc continuously shifts to higher temperature with increasing GNS content, and a maximum shift of 10 C is obtained in the 5 wt% G10-filled composite. By contrast, the 5 wt% G1-filled composite only shows 4 C elevation in Tp,mc. Fig. 8d also shows the results for the sPS composites filled with various amounts of CNT [11]. The sPS/CNT 95/5 composite possesses a Tp,mc of 251.6 C, which corresponds to a 14 C elevation in Tp,mc. These nonisothermal crystallization results demonstrate that tube-like CNTs are more effective than sheet-like graphenes in producing sPS nuclei for further crystallization. The thicker G10 nanofiller exhibits better nucleating ability than the thinner G1 nanofiller despite the lower surface area of G10 that can interact with the crystallizing sPS chains at a fixed GNS loading. Overall, the addition of GNS enhances
212
C. Wang et al. / Materials Chemistry and Physics 164 (2015) 206e213
Fig. 8. Variations of the (a) glass transition temperature, Tg, (b) peak temperature of cold crystallization, Tp,cc, (c) normalized cold-crystallization enthalpy, DHcc/(1 4w), (d) peak temperature of melt crystallization, Tp,mc of sPS composites with different filler concentrations. In (d), filled circles are results of sPS/CNT composites obtained from Ref. [11].
Fig. 7. (a). DSC heating traces of neat sPS and G10-filled composites for cold crystallization studies. The melt-quenched samples are heated at 10 C/min to 300 C. (b) DSC cooling curves for composites for melt crystallization studies. After holding 300 C for 10 min, the samples are cooled at a rate of 10 C/min. (c) 2nd heating traces of the melt-crystallized composites.
the nonisothermal crystallization kinetics in both cold- and melto are hardly changed crystallization processes, although Tg and Tm regardless of the GNS content. The shift in crystallization temperature (Tp,cc, Tp,mc) demonstrates the presence of a favorable interaction between the GNS and sPS chains possibly because of the presence of phenyls in sPS in which the carbon atoms have the same sp2 hybridization as those in the GNS [22]. 3.5. Isothermal crystallization and crystal melting of sPS/GNS composites To describe the kinetics of isothermal crystallization, the Avrami equation was applied to derive the overall crystallization rate (k) and Avrami exponent (n). For an in-depth comparison, the crystallization parameters derived for various composites crystallized at 256 C are tabulated in Table 1. The k and n of the neat sPS at 256 C are 7.68 104 min1 and 2.8, respectively. Compared with the neat sPS, all nanocarbons enhance k but the level of
enhancement is different. That is, 4-fold for G1, 45-fold for G10, 450-fold for CNT, and 300-fold for CNC. Evidently, 1D tube-like CNT is more effective in enhancing the crystallization of sPS than 2D sheet-like GNS. Judging from their respective 4c, both G1-and CNTfilled composites, but not the G10-and CNC-filled composites, already possess a fillerefiller network within the sample after 1 wt % filler addition. The melting behavior of isothermally crystallized composites was also studied. The classical HoffmaneWeeks approach was applied to determine the equilibrium melting temo ) of sPS composites. Linear extrapolation with the perature (Tm o by plotting the observed T as a Tm ¼ Tc line gives the value of Tm m function of Tc. Regardless of the GNS content, all composites show a superposition of the measured Tm. This result suggests that GNS addition has no influence on the melting of the crystalline lamellae, o and that the lamellar thickness remains intact. The determined Tm is about 290.3 C in the GNS-filled sPS composites. Since the glass transition and equilibrium melting temperatures of sPS remain relatively unchanged after the addition of GNS, the enhanced crystallization kinetics can be attributed to the heterogeneous nucleation induced by the GNS. Surface-induced nucleation is often observed in CNT-filled composites [9,11]. It is likely to be associated with the epitaxial mechanism. Epitaxial nucleation can take place when periodicity exists on the substrate surface and matches the particular spacing of the crystallizing plane of the matrix. In other words, it is the substrate surface, rather than substrate interior, that plays an important role in determining the surface-induced nucleation. Both the isothermal crystallization (Table 1) and non-isothermal crystallization (Figs. 5 and 8d) clearly show that GNSs of G10 with a lower aspect ratio are more effective than G1 to enhance the heterogeneous nucleation of sPS matrix. This is unexpected because at a fixed GNS content the interfacial area of G10-filled composites is lower than that of G1-filled ones. Thus, the nucleating difference could be attributed to the surface roughness of GNSs. The wrinkled surface of G1 may prohibit the incubation of sPS nuclei. Conversely, with its mechanical integrity, G10 has a relatively smooth surface
C. Wang et al. / Materials Chemistry and Physics 164 (2015) 206e213
for effectively hosting the deposition of sPS chains for epitaxial nucleation. Based on the epitaxy mechanism, the reason that the nucleating ability of G10 is better than that of the thinner G1 nanofiller could be that G1 nanofillers possess more surface defects than the G10 nanofillers. For the CNT-filled composites, surface-induced crystallization of sPS also occurs [11]. In addition, the 1D tube-like CNT with a diameter of several nanometers is readily wrapped by the surrounding sPS chains prior to crystallization. Thus, when a sufficient number of sPS chains anchor between neighboring CNTs, the segments of these wrapping chains around the CNTs may serve as heterogeneous nucleation sites because of their low mobility. This additional mechanism leads to the present findings (as shown in Table 1 and Fig. 8d) that 1D tube-like CNT is more effective in enhancing the crystallization of sPS than 2D sheet-like GNS. 4. Conclusions Semi-crystalline sPS composites filled with well-dispersed GNS with different aspect ratios (G1 and G10) were prepared through a coagulation method. A higher electrical conductivity threshold is required for thicker GNS of G10 with a smaller aspect ratio. The sPS/ GNS composites possess a higher electrical conductivity threshold than the amorphous aPS/GNS composites. Moreover, a similar percolation threshold is found for the sPS/GNS and sPS/CNT composites. The melt-quenched composites possess the dominant bform crystal, and the content increases with increasing GNS concentration. Both G1 and G10 are good nucleating agents in enhancing the crystallization of sPS. However, the nucleating ability of G1 is weaker than that of G10 because of its wrinkled surface. Thus, sPS nuclei are more difficult to incubate in G1 for triggering heterogeneous nucleation. Our results show that the nucleating ability of 1D CNT toward sPS crystallization is better than that of 2D GNS.
213
Acknowledgments The authors are grateful to the National Science Council of Taiwan (ROC) for the research grant (NSC-101-2221-E-006-069MY2) that supported this work. This research was, in part, supported by the Ministry of Education, Taiwan, R.O.C. The Aim for the Top University Project to the National Cheng Kung University (NCKU).
References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22]
E.B. Gowd, K. Tashiro, C. Ramesh, Prog. Polym. Sci. 34 (2009) 280. M. Moniruzzaman, K.I. Winey, Macromolecules 39 (2006) 5194. W. Bauhofer, J.Z. Kovacs, Compos. Sci. Technol. 69 (2009) 1486. S. Stankovich, D.A. Dikin, G.H.B. Dommett, K.M. Kohlhaas, E.J. Zimney, E.A. Stach, R.D. Piner, S.T. Nguyen, R.S. Ruoff, Nature 442 (2006) 282. H. Kim, A.A. Abdala, C.W. Macosko, Macromolecules 43 (2010) 6515. D. Cai, M. Song, J. Mater. Chem. 20 (2010) 7906. M.J. Allen, V.C. Tung, R.B. Kaner, Chem. Rev. 110 (2010) 132. A. Bunde, S. Havlin, Fractals and Disordered Systems, second ed., Springer, New York, 1996, pp. 27e28. F. Du, R.C. Scogna, W. Zhou, S. Brand, J.E. Fischer, K.I. Winey, Macromolecules 37 (2004) 9048. C.L. Huang, C. Wang, Carbon 49 (2011) 2334. C.L. Huang, C. Wang, Eur. Polym. J. 47 (2011) 2087. S. Cheng, X. Chen, G. Hsuan, Y. Li, Macromolecules 45 (2012) 993. J.Z. Xu, C. Chen, Y. Wang, H. Tang, Z.M. Li, B.S. Hsiao, Macromolecules 44 (2011) 2808. J.Z. Xu, T. Chen, C.L. Yang, Z.M. Li, Y.M. Mao, B.Q. Zeng, B.S. Hsiao, Macromolecules 43 (2010) 5000. C. Wang, C.L. Huang, Y.C. Chen, G.L. Hwang, S.J. Tsai, Polymer 49 (2008) 5564. N. Grossiord, H.E. Miltner, J. Loos, J. Meuldijk, B.V. Mele, C.E. Koning, Chem. Mater 19 (2007) 3787. J. Hicks, A. Behnam, A. Ural, Appl. Phys. Lett. 95 (2009) 213103. G. Eda, M. Chhowalla, Nano. Lett. 9 (2009) 814. J. Li, J.K. Kim, Compos. Sci. Technol. 67 (2007) 2114. H.B. Zhang, W.G. Zheng, Q. Yan, Y. Yang, J.W. Wang, Z.H. Lu, G.Y. Ji, Z.Z. Yu, Polymer 51 (2010) 1191. C. Huang, H. Bai, H. Xiu, Q. Zhang, Q. Fu, Compos. Sci. Technol. 102 (2014) 20. M. Yang, V. Koutsos, M. Zaiser, J. Phys. Chem. B 109 (2005) 10009.