Structure And Melt Rheology Of Polystyrene-based Layered Silicate Nano Composites

INSTITUTE OF PHYSICS PUBLISHING

NANOTECHNOLOGY

Nanotechnology 16 (2005) S514–S521

doi:10.1088/0957-4484/16/7/028

Structure and melt rheology of polystyrene-based layered silicate nanocomposites Liang Xu1 , Stacey Reeder2 , Mahesh Thopasridharan2 , Jiaxiang Ren1 , Devon A Shipp2,3 and Ramanan Krishnamoorti1,3 1

Department of Chemical Engineering, University of Houston, 4800 Calhoun, Houston, TX 77204–4004, USA 2 Department of Chemistry, Clarkson University, Potsdam, NY 13699-5810, USA E-mail: [email protected] and [email protected]

Received 18 January 2005, in final form 27 April 2005 Published 2 June 2005 Online at stacks.iop.org/Nano/16/S514 Abstract The melt-state viscoelastic properties of exfoliated in situ polymerized and intercalated solution-blended polystyrene (PS) and organically modified montmorillonite nanocomposites were investigated and compared. The PS nanocomposites prepared by nitroxide-mediated polymerization (NMP) exhibit a stable exfoliated structure whereas the PS nanocomposites prepared by solution mixing exhibit an intercalated structure. The linear viscoelastic properties were strongly correlated with the dispersion state of the nanocomposites. On the other hand, the non-linear oscillatory shear properties exhibited shear thinning character and were consistent with the weak interactions between the polymer and the layered silicate.

1. Introduction Significant interest has recently focused on understanding structure–property relations in polymer nanocomposites and how processing affects both structure and properties [1, 2]. In part, this is due to the dramatic improvements in material properties of nylon-6-layered silicate-based nanocomposites that were demonstrated by the Toyota research group on in situ polymerized hybrids [3, 4]. Subsequently, interest has also focused on trying to reproduce and emulate those results using melt processing of the components [1, 5]. Polystyrene-layered silicate nanocomposites have been prepared by both in situ bulk, solution or emulsion polymerization [6–10] and melt or solution processing [11, 12]. The latter typically results in intercalated materials [11], while the former may give a variety of mixtures of intercalated [7] or exfoliated [8–10] materials, depending on a variety of parameters such as mode of polymerization and type of silicate modifier. Of particular interest when examining structure–property relationships is the speculation that by performing in situ polymerization, with some possibility of 3 Authors to whom any correspondence should be addressed.

0957-4484/05/070514+08$30.00 © 2005 IOP Publishing Ltd

tethering the polymer chains to the layered silicate surface and by controlling the rates of intra-gallery and extra-gallery polymerization, the final equilibrium structure of the hybrids can be controlled [1, 4, 5, 13, 14]. Additionally, the rheological properties of in situ polymerized ionically tethered chains are distinctly different from those of melt-blended nanocomposites and presumed to originate from the differences in polymer conformation in those hybrids [2]. In this work, we exploit a recently developed free radical polymerization route termed nitroxide-mediated polymerization (NMP) for the preparation of partially tethered nanocomposites based on polystyrene and compare and contrast the structure and rheological properties to those from the solution-based intercalation approach. By using NMP, the in situ polymerized polystyrene samples have low polydispersity (<1.3), thus making for more convenient and meaningful comparisons. The melt and solution intercalation of layered silicates by polystyrene has been extensively studied previously and is considered as a model system where only weak thermodynamic interactions are expected [9, 10, 15–18]. Melt state rheological measurements have been shown to be an extremely powerful method to understand the mesoscale structure and the strength of polymer–filler interactions in both

Printed in the UK

S514

Structure and melt rheology of polystyrene-based layered silicate nanocomposites

macrocomposites and nanocomposites [18–20]. In particular, we use linear viscoelastic measurements to examine if the nanoparticles create a network structure that is able to sustain stress and therefore understand the mesoscale dispersion of the nanoparticles. Additionally, we also examine if the presence of a small fraction of tethered chains is able to render the non-linear viscoelastic properties to exhibit strain-hardening or whether the shear-thinning character of the matrix dominates the non-linear response.

2. Experimental methods 2.1. In situ polymerization 2.1.1. Materials. N ,N -dimethyl-n-hexadecylamine (Fluka) was used as received. 4-vinylbenzyl chloride (4VBC; Aldrich) was filtered though alumina to remove the inhibitor then purified via distillation under pressure. Styrene (Fisher) was purified by filtration through alumina followed by distillation under pressure over CaH2 . 2,2
azobis(isobutyronitrile) (AIBN; Kodak) was recrystallized from methanol. 2,2,6,6-tetramethylpiperidinoxy (TEMPO; Aldrich) was used as received. Montmorillonite clay (MMT; Cloisite-Na from Southern Clay Products, ion exchange capacity of 92 mequiv/100 g) was used as received. 2.1.2. Synthesis of N ,N -dimethyl-n-hexadecyl-(4-vinylbenzyl) ammonium chloride (VB16). Synthesis and purification of VB16 were carried out according to the published procedure [9]. To a round bottom flask 10.80 g of N ,N -dimethyl-nhexadecylamine was added with 4.56 g of 4VBC in 40 ml of ethyl acetate. The reaction was stirred at 40 ◦ C overnight. A white precipitate was recovered by filtration and recrystallized using ethyl acetate. 10.34 g of pure product was obtained to give a percentage yield of 71.14%. 2.1.3. Intercalation of VB16 into clay. The intercalation was carried out according to the published procedure [9]. 25 g of unmodified MMT was stirred in 1 l of water overnight. 10.34 g of VB16 was added to 100 ml of water and this solution was then added dropwise to the clay mixture and stirred for 3 h at temperatures between 0–5 ◦ C. The clay was filtered and washed with water several times and vacuum dried over several days. A portion of the clay was then stirred in petroleum ether for 1 h and vacuum dried overnight. 2.1.4. General synthesis of polystyrene. All solids were first added to a Schlenk flask. The flask was subjected to three cycles of alternating vacuum/N2 . Styrene was added via syringe after being purged with N2 for 30 min. The reaction was stirred for 48 h at 120 ◦ C. THF was then added to the flask, and the sample then precipitated into methanol, filtered and dried under vacuum. 2.1.5. Separation of free and tethered polystyrene. Free polystyrene was removed through dissolution of the nanocomposite in THF and then centrifugation. The supernatant solvent was removed by rotary evaporation and the polystyrene yield determined gravimetrically. The tethered

polymer and clay were separated by refluxing in the sample 2.5 wt% LiBr in a methanol/THF mixture (1/3 v/v) and refluxing for 3 h. The solution was then subjected to centrifugation, and the polymer removed from the supernatant liquid by precipitation into methanol. 2.2. Solution intercalation Solution intercalation samples of PS30K (Pressure Chemicals Inc., Mw /Mn < 1.04) with a dimethyl dioctadecyl ammonium modified montmorillonite (2C18M) were prepared with 3, 6.7 and 9 wt% layered silicate by dispersing the silicate as a dilute dispersion in toluene to which the polymer was added and subsequently rapidly dried to remove the solvent. The montmorillonite used here had a charge exchange capacity (CEC) of 0.9 eq kg−1 and a nominal disc diameter of 0.5 µm, as estimated from dilute solution dynamic light scattering and confirmed by quantitative analysis of several tens of transmission electron micrographs of dispersions in epoxy [18]. The description of the organic modification and sample preparation is routine and is provided elsewhere [18, 21]. Samples were subsequently annealed at 100 ◦ C in a vacuum oven for ∼12 h followed by extensive annealing (∼24 h) at 160 ◦ C to remove any remaining solvent and to facilitate complete polymer intercalation between the silicate layers. 2.3. Instrumentation 2.3.1. Gel permeation chromatography (GPC). Samples for GPC analysis were dissolved in a solution of 2.5 wt% LiBr in a methanol/THF mixture (1/3 v/v) and refluxed for 3 h. The solution was then subjected to centrifugation and the supernatant filtered through a 0.2 µm filter. Molecular weights and molecular weight distributions were determined using gel permeation chromatography (GPC) equipped with a Waters 717 autosampler, a Waters 515 HPLC pump, three Waters columns and a Viscotek LR40 laser refractometer with THF as the eluent. The GPC was calibrated using polystyrene standards. 2.3.2. X-ray diffraction analysis (XRD). The dispersion state of the layered silicates was ascertained by x-ray diffraction and using Siemens D5000 diffractometers with Cu Kα radiation (λ = 1.54 Å) generated at 30 mA and 40 kV. Diffraction traces were obtained over a 2θ range of 2◦ –10◦ in steps of 0.02◦ and counting time of 3 s at each angular position. 2.3.3. Thermogravimetric analysis (TGA). Thermogravimetric analysis (TGA) of dried polymer samples was performed on a Perkin-Elmer TGA7. Measurements were performed in flowing air at a heating rate of 10 ◦ C min−1 . 2.3.4. Melt rheology. Melt rheological measurements were performed using a melt state ARES rheometer (TA Instruments) with a torque transducer capable of torque measurements over 0.2–2000 g cm. Dynamic oscillatory shear measurements (120 ◦ C
T
190 ◦ C, with a N2 purge used for all measurements) were performed by applying a sinusoidal strain of the form γ (t) = γ0 sin(ωt) (where γ0 S515

L Xu et al

H3C(H2C)15

N

NMP N

(CH2)15CH3

Styrene + TEMPO + BPO H3C(H2C)15

N O N

N

(CH2)15CH3

O

O

Polymer-silicate nanocomposite

O O

VB16-MMT Scheme 1. (This figure is in colour only in the electronic version)

Table 1. Distribution of free and tethered polymer, and recovered layered silicatea. E-PS-B

87.0 8.8 3.3

87.4 6.6 4.0

a

Percentages based on the mass of recovered material relative to the initial mass of the sample used, and add up to less than 100 because of losses during work-up.

PS30K + 9 wt % 2C18M

Intensity (AU)

Free polymer Tethered polymer Layered silicate

E-PS-A

PS 30K + 6 wt % 2C18M

E-PS-B E-PS-A 2

is the strain amplitude and ω is the frequency with 0.001
ω
100 rad s−1 ), and measuring the resultant shear stress σ (t) (=γ0 {G
sin(ωt) + G

cos(ωt)}), with G
and G

being the storage and loss modulus, respectively. Discs of samples with thickness ∼1–2 mm and diameter 28 mm were prepared by vacuum moulding the samples at 160 ◦ C in a pellet die and heated and pressed in a Carver press with the application of minimal force on the plunger to ensure minimal processrelated orientation in these samples. For the linear viscoelastic properties small strain amplitudes (i.e., γ0 < 0.05) were typically employed and the response was verified to be linear by changing the strain amplitude by a factor of two and observing the invariance of the viscoelastic parameters measured. For shear-alignment tests, large amplitude oscillatory shear over prolonged periods of time was imposed and the resulting shear stress interpreted in the context of the linear response theory. However, no effort was made in these studies to verify that the higher order harmonics were indeed negligible. Finally, for the strain sweep measurements, at constant frequency the sample was subjected to progressively increasing strain amplitudes and followed by a succession of smaller strains. Again the data were interpreted in terms of the linear parameters and should be construed only as a qualitative representation of the true viscoelastic response of these materials.

3. Results and discussion 3.1. Synthesis and structure of nanocomposites The VB16 modified layered silicates exhibited a gallery height of 1.25 nm (d = 2.2 nm) and on the basis of TGA measurements had a total organic content of 27 wt%. The S516

PS 30K + 3 wt % 2C18M

4

6 8 2θ (λ = 1.54 A)

10

12

Figure 1. X-ray diffraction traces for the two in situ polymerized nanocomposites E-PS-A and E-PS-B and the three solution-intercalated nanocomposites based on 2C18M and a model PS30K. All samples were annealed in the melt (160 ◦ C) for at least 24 h prior to XRD measurements. Both E-PS-A and E-PS-B exhibit no peaks and suggest disordered intercalated or exfoliated structures, while all three solution-intercalated samples exhibit peaks corresponding to 001, 002 and 003 reflections, suggesting a well organized intercalated structure.

tethered polystyrene based nanocomposites, samples E-PS-A and E-PS-B, were prepared using a nitroxide mediated scheme as illustrated in scheme 1. From these polymerizations it is expected that some polymer chains will be grafted (or tethered) to the silicates through copolymerization of the styrene with the VB16 that is bound to the silicate layers, while other chains remain unattached to the silicate. The relative amounts of tethered and untethered (‘free’) polymer chains were determined for both EPS-A and E-PS-B, and are summarized in table 1. These data indicate that most of the chains are not tethered, with only 6– 9% bound to the clay. Su and Wilkie [10] have reported similar values (approximately 5%) for bound PS in nanocomposites based on VB16-MMT synthesized using conventional radical polymerization of styrene. The clay content of each sample that was determined experimentally through extraction (3.3 and 4.0 wt% for E-PS-A and E-PS-B, respectively) is similar to the clay content calculated from the initial amount of clay added and the monomer conversion (3.5 and 3.9 wt%, respectively). The development of the polymer molecular weight during nitroxide-mediated polymerization (NMP) is quite different

Structure and melt rheology of polystyrene-based layered silicate nanocomposites

5

10

4

2.0

10

1.5

3

10

1.0 0.5

2

10

10

8

10

7

10

6

10

5

10

4

PS30K + 2C18M wt % 2C18M 0 3 6.7 9 2

10 4

6

8

3

-2 130 140 150 160 170 180

10

o

Silicate content, wt%

1

10 -4 -3 -2 -1 10 10 10 10

2

10

0

1

10

10

2

3

10

0 -1

0.0 0

0 3 6.7 9

1 log(aT)

6

10

PS30K + 2C18M wt % 2C18M 0 3 6.7 9

bTG'' (dynes/cm2)

7

10 bTG' (dynes/cm2)

(b)

8

10

β

(a)

T ( C)

2

10

10 -4 -3 -2 -1 0 1 2 10 10 10 10 10 10 10 o aTω(rad/sec) [To=150 C]

4

o

aTω(rad/sec) [To=150 C] (c) 108

(d) 1.0

10

α

10

0 3 6.7 9

η∗ (poise)

η∗ (poise)

0.2

6

0.0 0

2

4

6

8

10

Silicate content, wt%

5

10

4

10

PS30K + 2C18M wt % 2C18M

0 3

6

10

5

10

4

10

6.7 9

3

3

10 -4 -3 -2 -1 10 10 10 10

4

PS30K + 2C18M wt % 2C18M

0.4

10

10

8

7

0.6

3

10

0.8

7

10

0

10

10

1

10

2

3

10

4

10

o

aTω(rad/sec) [To=150 C]

10 3 10

10

4

5

10

10

6

7

10

8

10

2

G* (dynes/cm )

Figure 2. Time–temperature superposed mastercurves illustrating the frequency dependence of the linear viscoelastic dynamic storage modulus G
(a), loss modulus G

(b) and complex viscosity η∗ (c) for PS30K and the intercalated nanocomposites of PS30K with 2C18M. The data for the 6.7 and 9 wt% nanocomposites demonstrate solid-like character and this behaviour is best observed in a cross-plot of the complex modulus G ∗ versus η∗ (d) and demonstrates that the viscosity for those nanocomposites diverges at a finite and large value of G ∗ . The low-frequency frequency dependence of G
evaluated by the parameter β(G
∝ ωβ ) and of η∗ evaluated by the parameter α(η∗ ∝ ωα ) are captured in the insets of parts (a) and (c) respectively, and clearly demonstrate the similarity to the scaling observed for soft glassy materials with a finite yield stress. The inset in (b) illustrates the temperature dependence of the frequency shift factors (aT ) used to prepare the viscoelastic mastercurves for the polymer and the nanocomposites. The shift factors, within the errors of the measurements, are independent of the silicate loading and are adequately represented by the simple WLF equation.

when compared to conventional radical polymerization. In NMP, all chains begin growth early in the polymerization, producing oligomers that continue to grow throughout the reaction. Ideally, the number average molecular weight (Mn ) increases linearly with monomer conversion, and the polydispersity is less than 1.5, and often less than 1.3. The GPC analysis of the E-PS-A sample demonstrated an increase in the number average molecular weight with reaction time and monomer conversion, with no appreciable change in the polydispersity index (less than 1.35 throughout). This indicates that the polymerizations proceeded as expected. Two samples were synthesized using this method: E-PS-A containing 3.3 wt% silicate, Mn = 12.3 × 103 g mol−1 and Mw /Mn = 1.19 , and E-PS-B containing 4.0 wt% silicate, Mn = 21.8 × 103 g mol−1 and Mw /Mn = 1.32 . The XRD traces for aliquots of samples removed after 5, 24 and 48 h all demonstrated no observable peak corresponding to layer— layer stacking over a scattering angle 2θ of 2◦ –10◦ . The structures of melt-annealed nanocomposites were characterized using XRD and the traces are shown in figure 1. The two nanocomposites prepared by NMP in the presence of the silicate, E-PS-A and E-PS-B, exhibit

no peak while the three solution-intercalated nanocomposites exhibit intercalated structures with well ordered structures as suggested by the presence of 001, 002 and 003 reflections. Both E-PS-A and E-PS-B when examined by electron microscopy demonstrate largely exfoliated structure while the solution-intercalated nanocomposites exhibit ∼10% individual sheets and the rest in tactoids consisting of on average eight to nine layers. 3.2. Linear viscoelasticity The linear viscoelastic time–temperature superposed mastercurves of macroscopically unoriented solution-intercalated PS30K with differing amounts of 2C18M are shown in figure 2. PS30K is a lightly entangled polymer and for reduced frequencies (aT ω) below 100 rad s−1 exhibits liquid-like behaviour with G
scaling as ω2 and η∗ scaling as ω0 as expected from a simple Newtonian liquid [22]. On the other hand, the nanocomposites demonstrate: (i) time–temperature superposed mastercurves, with frequency shift factors comparable to that of polystyrene and moduli shift factors close to unity (0.97
bT
1.03); and (ii) a gradual change in the low S517

L Xu et al

(a)

10

10

6

5

10

6

5

G'

106

G', G" (dynes/cm2)

10

108 E-PS-B PS21.8K + 3.9 wt% MMT 107

5

6

10

10 G"

4

η∗ (Poise)

10

107

E-PS-A PS12.3K + 3.5 wt% MMT

η∗ (Poise)

G', G" (dynes/cm2)

10

7

(b)

7

5

10

10 G'

10

G"

4

10

-2

10

-1

10

0

10

1

10

2

10

3

10

3

4

10 -4 10

4

10 10

-3

aTω (rad/s) [To=130 oC] (c)

10

-1

10

0

10

1

10

2

aTω (rad/s) [To=140 oC] E-PS-B

7

10

η∗ (Poise)

10

-2

o

To=140 C

106

105 E-PS-A To=130 oC

4

10

103

104

105

106

107

2

G* (dynes/cm )

Figure 3. Time–temperature superposed linear viscoelastic mastercurves of G
, G

and η∗ for (a) E-PS-A (PS12.3K + 3.5 wt% MMT; NMP exfoliated) and (b) E-PS-B (PS21.8K + 3.9 wt% MMT; NMP exfoliated). The data for both samples demonstrate solid-like melt rheological properties. The data in part (c) illustrate the solid-like character for both of these nanocomposites with the viscosity diverging at reasonably high values of the complex modulus.

frequency viscoelastic response from that of a liquid to that of a solid-like material. Three signatures are worth considering for the transition from liquid-like to solid-like melt rheological behaviour: (a) the power-law exponent describing the low frequency response of G
(G
∝ ωβ ), β, decreases from two to a value close to zero, with the values for the 6.7 and 9 wt% 2C18M nanocomposites suggestive of a low frequency plateau in G
; (b) the power-law exponent describing the low frequency response of η∗ (η∗ ∝ ωα ), α, increases from near zero to a value close to unity, with the high volume fraction nanocomposites showing large values for α or strong nonNewtonian behaviour even at low frequencies; (c) a cross-plot of G ∗ versus η∗ reveals the divergence of η∗ at a finite value of G ∗ for the higher weight fraction silicate nanocomposites and suggestive of a yield stress material [21]. These signatures of transformation from liquid-like to solidlike material are similar to those observed for other soft glassy materials [23] and in the case of layered silicate nanocomposites have been attributed to the geometrical percolation of dispersed nanoclay sheets or aggregates of the sheets [18, 20, 21, 24]. The location of the geometrical S518

percolation between 3 and 6.7 wt% of the layered silicate for the randomly oriented nanocomposites is consistent with the intercalated nature of the nanocomposites with the presence of <10% (by number) of individualized silicate sheets (obtained by extensive electron microscopy characterization and not shown here [18]) and an effective aspect ratio of the order of ∼30 [21, 25]. The linear oscillatory viscoelastic responses for the exfoliated tethered nanocomposites prepared by nitroxidemediated polymerization are presented in figure 3. The linear viscoelastic response for both samples is capable of excellent time–temperature superpositioning of all viscoelastic functions. These data demonstrate the solid-like character of these materials with a low frequency plateau of G
(β = 0.13 and 0.16 for E-PS-A and E-PS-B respectively) and G
exceeding G

in the terminal region, low frequency divergence of η∗ (α = 0.75 and 0.68 for E-PS-A and E-PS-B respectively) and the divergence of η∗ at a finite (and relatively large) value of G ∗ . Clearly, both E-PS samples are hydrodynamically percolated with the presence of a finite yield stress [20, 21, 26, 27]. The differences in α and β between the two samples suggest that the dispersion state and the development of filler network are much better established for the E-PS-A sample. This is further confirmed from an examination of the value of G ∗

0.6

γ o = 0.75

2.0

γo =

1.5

0.5 1.0 0.5

0.5

γ o = 1.0

0.4 0.3

γ o = 1.5 γ o = 0.75 γ o = 1.0 γ = 1.5 o

η∗ (P), E-PS-A, To=130 oC

2.5

(b)

0.7

γ o = 0.5

0.2 0.1 0.0

6

10

5

10

10 rad/sec

0.1 rad/sec

0.1 rad/sec

5

10

4

10

0.0 0

10 rad/sec

15

10-3

10-2

10-1

100

10

5

10

4

10

3

η∗ (P), E-PS-B, To=160 oC

3.0

G' (dynes/cm2) x 10-3 E-PS-B; 160 oC

(a)

G' (dynes/cm2) x 10-4 E-PS-A; 130 oC

Structure and melt rheology of polystyrene-based layered silicate nanocomposites

γo

-3

t (sec) x 10

10

5

10

4

10

3

4

10 10 rad/sec 0.1 rad/sec

3

10 rad/sec 0.1 rad/sec

10

2

10

10

-3

-2

10

10

γo

-1

10

G' (dynes/cm2), E-PS-B, To =160 oC

G' (dynes/cm2), E-PS-A, To 130 oC

(c)

0

Figure 4. (a) Time variation of the storage modulus G
for E-PS-A and E-PS-B during the application of prolonged large strain amplitudes (γ0 ) at a frequency of 0.1 rad s−1 . The strain amplitude was successively raised from 0.5 to 1.5 as noted in the figure. The values of G
decrease with increasing the amplitudes of strain, indicating the shear-induced alignment of the silicate layers. (b), (c) The strain amplitude dependence of η∗ (b) and G
(c) at specified frequencies for E-PS-A and E-PS-B obtained immediately after prolonged large-strain-amplitude measurements described in (a) and with progressively increasing strain amplitude. Similar data with decreasing strain amplitude with minimal hysteresis was obtained.

at which η∗ diverges—∼3 × 104 dynes cm−2 for E-PS-A as compared to ∼2 × 103 dynes cm−2 for E-PS-B. In contrast, for the PS30K solution-intercalated samples described above the corresponding values of G ∗ are 5 × 103 and 1.3 × 104 dynes cm−2 for the 6.7 and 9 wt% nanocomposites respectively. It is clear from the rheological evidence presented above that in spite of the lower silicate content in E-PS-A as compared to E-PS-B (3.3 versus 4.0 wt%) E-PS-A exhibits a stronger filler network structure, and on the basis of the G ∗ values the network structure of the silicate in E-PS-A is somewhat superior to that of the 9 wt% intercalated 30 K sample and clearly demonstrates the excellent dispersion of the silicate sheets. The most significant difference between E-PS-A and E-PS-B is the lower molecular weight of E-PS-A. On the basis of the mean-field theory developed by Vaia and Giannelis [16, 17], no difference in structure would be anticipated on the basis of the change in molecular weight. On the other hand, the self-consistent theory developed by Balazs and co-workers [28] would anticipate that increasing the molecular weight would lead to poorer dispersions and consistent with the experiments described here. Previous work on PS-based intercalated nanocomposites, prepared by solution processing, indicated that the fraction of individual layers in the nanocomposites increased with

increasing molecular weight of the matrix [18]. Such improvements in the dispersion with increasing molecular weight have been reported by Fornes and Paul [1, 5] on melt-processed nylon nanocomposites and by Manias et al [29] on melt-processed polypropylene nanocomposites. The presence of a higher fraction of individual silicate sheets in the PS systems results in the following observations for the rheological measurements of nanocomposites at comparable levels of added silicate: (i) an increase in the G ∗ value at which η∗ diverges with increasing molecular weight and individualized sheet content; and (ii) an increase in the low-frequency (terminal zone for the matrix) values of η∗ (normalized to that of the zero shear viscosity of the matrix) and G
at comparable values of ω/ωrelaxation [18]. On the basis of these previous experiments, that the results presented here suggest that E-PS-B, having a higher molecular weight and slightly higher silicate loading, exhibits a weaker silicate layer network structure compared to E-PS-A is somewhat surprising. Perhaps the tethering of some of the chains to the silicate and the relatively poor inherent affinity between polystyrene and the silicates [16] results in the poorer dispersion of the silicates with increasing chain length. However, it is important to stress that, unlike the case of thermoset nanocomposites prepared with epoxies where the dispersion state, which can be a function of extent of cure, has been shown to depend S519

L Xu et al

most critically on the relative ratio of inter- and intra-gallery rates of polymerization and is kinetically trapped [14, 30], in these polystyrene nanocomposites the structure formed remains thermodynamically stable throughout its exposure to the melt state. These complexities point to the need for further work in this area. 3.3. Non-linear viscoelasticity The strain amplitude dependences of the dynamic oscillatory shear data for the two exfoliated nanocomposites were studied to examine if the samples strain softened or if the samples strain-hardened as observed in other brush systems endtethered to the silicate sheets. Prior to examining the strain amplitude dependence, the samples were pre-conditioned to achieve a high state of orientation and therefore eliminate the orientation of the silicate layers from the viscoelastic response [31, 32]. The pre-conditioning was achieved by subjecting the samples to prolonged large amplitude oscillatory shear (0.5
γ0
1.5) at low frequencies (ω = 0.1 rad s−1 ) and the during-shear storage modulus as a function of shearing time is shown in figure 4(a). Typically, alignment experiments were stopped after all viscoelastic functions had acquired timeindependent values. Previous studies have demonstrated that such large amplitude oscillatory flows result in the orientation of the silicate sheets with sheet normals parallel to the velocity gradient direction [31, 32]. Immediately following the large-amplitude-strain measurements, strain sweep measurements at fixed frequency and temperature were conducted, first with increasing strain amplitude and then followed by a sweep with decreasing strain amplitude (figures 4(b) and (c) respectively) [27, 32]. At low strain amplitudes i.e., γ0
0.1, both η∗ and G
are independent of strain amplitude. However, progression to higher strain amplitudes (γ0 > 0.1) leads to shear thinning behaviour and results in a decrease in both η∗ and G
. This result is significantly different from those obtained on poly(ε-caprolactone) end-tethered silicate nanocomposites where reversible strainhardening was observed and suggested to arise from a coil to stretch transformation of the polymer chains [32]. In both EPS-A and E-PS-B the fraction of chains tethered is <10% while for the PCL-based nanocomposites the fractions of free polymers were considerably smaller (∼<10% in all cases). Hence in the current measurements the effect of the tethered chains is not observed and the non-linear viscoelasticity dominated by the presence of free untethered shear-thinning PS chains.

4. Conclusions The preparation of in situ polymerized polystyrene nanocomposites using nitroxide-mediated polymerization leads to some tethering of the polymer chains and good dispersion of the silicate sheets to a reasonably exfoliated state. In comparison, solution-blended nanocomposites resulted only in intercalated structures. Further, from melt state linear viscoelastic measurements we infer that in the case of the in situ polymerized materials the dispersion improves with lowering molecular weight and consistent with the weak interactions between the polymer and the silicate. Finally, the non-linear rheological properties for these in situ polymerized nanocomposites S520

exhibit shear-thinning character and appear to suggest that the large fraction of the matrix free polymer chains dominates the large strain response. In this respect, it would be interesting to test the properties of nanocomposites with increasing fraction of chains tethered to the surface to examine the transformation of matrix-dominated to tethered-dominated rheology. We are currently examining this issue, in addition to exploring other questions raised in this study, such as the structure-property dependence on molecular weight.

Acknowledgments We (JR and RK) gratefully acknowledge the ExxonMobil Chemical Company for partial financial support of this work. We (LX and RK) also gratefully acknowledge support in part by the Texas Institute for Intelligent BioNano Materials and Structures for Aerospace Vehicles, funded by NASA Cooperative Agreement No. NCC-1-02038. Further, the Center for Advanced Materials Processing at Clarkson University (a New York State Center for Advanced Technology), and the donors of The Petroleum Research Fund, administered by the American Chemical Society, are gratefully acknowledged for their financial support to DAS.

References [1] Fornes T D and Paul D R 2003 Polymer 44 4993–5013 [2] Krishnamoorti R, Vaia R A and Giannelis E P 1996 Chem. Mater. 8 1728 Giannelis E P, Krishnamoorti R and Manias E 1999 Adv. Polym. Sci. 138 107–47 Alexandre M and Dubois P 2000 Mater. Sci. Eng. 28 1–63 [3] Kojima Y, Usuki A, Kawasumi M, Okada A, Kurauchi T, Kamigaito O and Kaji K 1994 J. Polym. Sci. B 32 625–30 Yano K, Usuki A, Karauchi T and Kamigaito O 1993 J. Polym. Sci. A 31 2493–8 Kojima Y, Usuki A, Kawasumi M, Okada A, Kurauchi T and Kamigaito O 1993 J. Polym. Sci. A 31 983–6 Usuki A, Kato M, Okada A and Kurauchi T 1997 J. Appl. Polym. Sci. 63 137 [4] Usuki A, Koiwai A, Kojima Y, Kawasumi M, Okada A, Kurauchi T and Kamigaito O 1995 J. Appl. Polym. Sci. 55 119–23 Kojima Y, Usuki A, Kawasumi M, Okada A, Kurauchi T, Kamigaito O and Kaji K 1995 J. Polym. Sci. B 33 1039–45 [5] Fornes T D, Yoon P J, Keskkula H and Paul D R 2001 Polymer 42 9929–40 [6] Doh J G and Cho I 1998 Polym. Bull. 41 511–8 Laus M, Cameriani M, Lelli M, Sparnacci K, Sandrolini F and Francescangeli O 1998 J. Mater. Sci. 33 2883–8 Moet A and Akelah A 1993 Mater. Lett. 18 97–102 Noh M W and Lee D C 1999 Polym. Bull. 42 619–26 Zeng C and Lee L J 2001 Macromolecules 34 4098–103 Zhu J, Start P, Mauritz K A and Wilkie C A 2002 J. Polym. Sci. A 40 1498–503 Zhu J, Uhl F M, Morgan A B and Wilkie C A 2001 Chem. Mater. 13 4649–54 Zhu J and Wilkie C A 2000 Polym. Int. 49 1158–63 [7] Akelah A and Moet A 1996 J. Mater. Sci. 31 3589–96 Okamoto M, Morita S, Taguchi H, Kim Y H, Kotaka T and Tateyama H 2000 Polymer 41 3887–90 [8] Fu X and Qutubuddin S 2000 Mater. Lett. 42 12–5 Fu X and Qutubuddin S 2001 Polymer 42 807–13 Weimer M W, Chen H, Giannelis E P and Sogah D Y 1999 J. Am. Chem. Soc. 121 1615–6 [9] Zhu J, Morgan A B, Lamelas F J and Wilkie C A 2001 Chem. Mater. 13 3774–80

Structure and melt rheology of polystyrene-based layered silicate nanocomposites

[10] Su S and Wilkie C A 2003 J. Polym. Sci. A 41 1124–35 [11] Vaia R A, Jandt K D, Kramer E J and Giannelis E P 1995 Macromolecules 28 8180 Vaia R A, Ishii H and Giannelis E P 1993 Chem. Mater. 5 1694 Vaia R A and Giannelis E P 1997 Macromolecules 30 8000–9 [12] Beyer F L, Tan N C B, Dasgupta A and Galvin M E 2002 Chem. Mater. 14 2983–88 Kim T H, Lim S T, Lee C H, Choi H J and Jhon M S 2003 J. Appl. Polym. Sci. 87 2106–12 Park C I, Park O O, Lim J G and Kim H J 2001 Polymer 42 7465–75 [13] Vaia R A, Lincoln D M, Wang Z-G, Hsiao B S and Krishnamoorti R 2001 Polymer 42 9975–84 Lincoln D M, Vaia R A and Krishnamoorti R 2004 Macromolecules 37 4554–61 [14] Wang Z and Pinnavaia T J 1998 Chem. Mater. 10 1820–6 Wang Z, Lan T and Pinnavaia T J 1996 Chem. Mater. 8 2200–4 [15] Hoffmann B, Dietrich C, Thomann R, Friedrich C and Mulhaupt R 2000 Macromol. Rapid Commun. 21 57–61 Zax D B, Yang D K, Santos R A, Hegemann H, Giannelis E P and Manias E 2000 J. Chem. Phys. 112 2945–51 Wang D Y, Zhu J, Yao Q and Wilkie C A 2002 Chem. Mater. 14 3837–43 [16] Vaia R A and Giannelis E P 1997 Macromolecules 30 8000–8 [17] Vaia R A and Giannelis E P 1997 Macromolecules 30 7990–9 [18] Ren J 2002 PhD Thesis Department of Chemical Engineering (Houston: University of Houston) [19] Yurekli K, Krishnamoorti R, Tse M F, McElrath K O, Tsou A H and Wang H-C 2001 J. Polym. Sci. B 39 256–75 Hsieh A J, Moy P, Beyer F L, Madison P, Napadensky E, Ren J and Krishnamoorti R 2004 Polym. Eng. Sci. 44 825–37 [20] Krishnamoorti R and Yurekli K 2001 Curr. Opin. Colloid Interface Sci. 6 464–70 [21] Ren J, Silva A S and Krishnamoorti R 2000 Macromolecules 33 3739–46 [22] Graessley W W 1974 Adv. Polym. Sci. 16 1

[23] Sollich P, Lequeux F, Hebraud P and Cates M E 1997 Phys. Rev. Lett. 78 2020–3 Sollich P 1998 Phys. Rev. E 58 738–59 Bonn D, Kellay H, Tanaka H, Wegdam G and Meunier J 1999 Langmuir 15 7534–6 Bonn D, Tanaka J, Wegdam G, Kellay H and Meunier J 1999 Europhys. Lett. 45 52–7 Bonn D, Tanase S, Abou B, Tanaka H and Meunier J 2002 Phys. Rev. Lett. 89 015701 [24] Krishnamoorti R and Giannelis E P 1997 Macromolecules 30 4097–102 Krishnamoorti R and Silva A S 2000 Polymer–Clay Nanocomposites ed T J Pinnavaia and G W Beall (New York: Wiley) pp 315–43 Mitchell C A and Krishnamoorti R 2001 Polymer Nanocomposites vol 804, ed R Krishnamoorti and R A Vaia (Washington, DC: ACS) pp 159–75 Mitchell C A and Krishnamoorti R 2002 J. Polym. Sci. B 40 1434–43 [25] Garboczi E J, Snyder K A, Douglas J F and Thorpe M F 1995 Phys. Rev. E 52 819–28 [26] Ren J and Krishnamoorti R 2003 Macromolecules 36 4443–51 [27] Krishnamoorti R, Ren J and Silva A S 2001 J. Chem. Phys. 114 4968–73 [28] Ginzburg V V and Balazs A C 1999 Macromolecules 32 5681 Balazs A C, Singh C and Zhulina E 1998 Macromolecules 31 8370–81 Kuznetsov V D and Balazs C A 2000 J. Chem. Phys. 112 4365–75 Balazs A C, Singh C, Zhulina E and Lyatskaya Y 1999 Acc. Chem. Res. 32 651–7 Balazs A C 2000 Curr. Opin. Colloid Interface Sci. 4 443–8 [29] Manias E 2003 private communication [30] Wang M S and Pinnavaia T J 1994 Chem. Mater. 6 2216 [31] Ren J, Casanueva B F, Mitchell C A and Krishnamoorti R 2003 Macromolecules 36 4188–94 [32] Krishnamoorti R and Giannelis E P 2001 Langmuir 17 1448–52

S521