Mobile nanoparticles and their effect on phase separation dynamics in thin-film polymer blends

Transcription

1 EUROPHYSICS LETTERS 15 October 2004 Europhys. Lett., 68 (2), pp (2004) DOI: /epl/i Mobile nanoparticles and their effect on phase separation dynamics in thin-film polymer blends H.-J. Chung, A. Taubert( ),R.D.Deshmukhand R. J. Composto( ) Department of Materials Science and Engineering and Laboratory for Research on the Structure of Matter, University of Pennsylvania Philadelphia, PA , USA received 7 June 2004; accepted in final form30 August 2004 PACS Ja Liquid-liquid transitions. PACS Np Nanoparticles in polymers. PACS a Thin film structure and morphology. Abstract. We present a systematic study of mobile nanoparticles (NP) and their effect on phase separation dynamics in polymer blend films. Starting with a homogeneous dispersion of silica NP in PMMA : SAN films, the NP are observed to partition into the PMMA-rich phase and stratify during phase separation. The correlation length ξ between phases grows as ξ t 1/3 for blends containing 0, 2, and 5 wt% NP. However, phase size decreases as the wt% of NP increases. This behavior agrees with a coalescence model, which incorporates an increase in viscosity due to the NP. Materials science relies on combining dissimilar, yet complementary components to create materials with unique properties [1]. For example, the fracture toughness of nickel-based superalloys is enhanced by precipitation hardening. Similarly, many properties of polymers can be improved by adding low concentrations of nanoparticles (NP) [2]. Despite their potential advantages, including processibility and cost, a fundamental understanding of phase separation and wetting in polymer blend nanocomposites is lacking relative to the pure polymer blends [3, 4]. This understanding is of practical importance because the final morphologies of multi-component materials are typically kinetically trapped (i.e., non-equilibrium) and, therefore, the material properties are ultimately dictated by phase evolution and wetting. In pioneering experiments by Tanaka et al. [5], micron-sized glass beads similar in size to the filmthickness were found to slow down phase separation of binary liquids. Although recent experiments [6, 7] support this result, the mobility and partitioning of the particles, as well as the mechanism of slowing-down, are not yet understood. Several models predict how polymer blend phase separation is effected by silica networks [8], as well as immobile [9, 10] and mobile particles [10 15]. The effect of shear [13], hydrodynamic flow [9], particle wetting [8 15], ( ) Current address: Department of Chemistry, University of Basel - CH-4056 Basel, Switzerland. ( ) composto@lrsm.upenn.edu c EDP Sciences

2 220 EUROPHYSICS LETTERS particle-particle interactions [14] and particle-substrate interactions [15] have also been considered. Relevant to this letter, modified Ginzburg-Landau [11] and molecular-dynamics studies [10,12] show that the wetting of NP and their partitioning into domains play an important role in slowing down phase evolution. Motivated by these fundamental issues and potential applications, we have performed the first systematic experimental study of phase separation dynamics in polymer blends containing mobile NP. A fundamental understanding of phase evolution in polymer blend films with NP requires prior knowledge of blend behavior. Previous studies show that 50 : 50 (bulk) films of poly(methyl methacrylate) (PMMA) and poly(styrene-ran-acrylonitrile) (SAN) undergo three distinct stages of morphology evolution [16]. An early stage (ES) is dominated by hydrodynamic-flow driven wetting of PMMA through a 3D bicontinuous morphology (e.g., tubes) to the free surface and substrate [17]. During the intermediate stage (IS), discrete, PMMA-rich (denoted PMMA) domains span the SAN-rich (denoted SAN) mid-layer and grow as 2D disks. Domain growth scales with time and thickness as t 1/3 and d 2/3, respectively, in agreement with a coalescence model [18]. In the late stage, the PMMA/SAN/PMMA trilayer structure ruptures due to interfacial fluctuations resulting in a rough film[19]. These studies provide the necessary background for a quantitative study of phase separation dynamics of NP containing films during the IS. A model system has been chosen to systematically investigate the effect of mobile NP on phase separation in polymer blends [10 12]. This system displays the three stages of evolution observed for pure blends. Using lateral and depth profiling techniques, this letter demonstrates that NP are initially homogeneously dispersed (i.e., well-defined starting state), mobile (i.e., not pinned to substrate), and partition (i.e., wet) into the PMMA phase during phase separation. By measuring the correlation length ξ during the IS, the addition of NP is found to slow down phase growth without changing the growth exponent observed for blends, ξ t 1/3. Using a coalescence model [18], this slowing-down is attributed to an increase in viscosity of the wetting component, PMMA, due to the sequestered NP. The weight average molecular weights and polydispersities of PMMA and SAN (33 wt% AN) are 82.5 K and 1.05, and 118 K and 2.24, respectively. The glass transition temperatures are 125 C and 115 C, respectively. The radii of gyration are 7.3 nmand 9.5 nm, respectively [20]. This blend displays lower critical solution behavior with a critical temperature of 160 C. At 185 C and 195 C, deep quenches, the coexisting compositions are nearly pure PMMA and SAN [21]. The NP are monodisperse, 22 nm diameter colloidal silica (MIBK-ST, Nissan Chemical), surface modified by trimethyl silane to produce miscibility in methyl-isobutylketone (MIBK) [22]. The PMMA : SAN (50 : 50) and NP were mixed in MIBK, spun-cast on silicon and dried at 120 C in vacuumfor 24 h. The filmthickness values ranged from120 nmto 700 nm. Samples were annealed on a hot stage (Mettler FP-82, Mettler Toledo) at 185 C or 195 C in argon. The surface and interface morphologies were obtained using scanning force microscopy (SFM) (Dimension 3000, Digital Instruments). To reveal the interface, PMMA was removed by exposing samples to 2.0 MeV He +, followed by immersion in acetic acid. Images were obtained in tapping mode. Grazing-incidence Rutherford backscattering spectrometry (RBS) with enhanced surface resolution (18 nm) was used to determine the NP depth profile [23]. The lateral distribution of NP was determined by transmission electron microscopy (TEM) using a JEOL 2010F with a field emission gun and Gatan image filter. The acceleration voltage, condenser aperture, and energy slit width were 197 kv, 50 µm, and 20 ev, respectively. Figure 1 shows plane-view TEM images of 120 nm PMMA : SAN films containing NP. For the as-cast filmwith 0.5 wt% NP, fig. 1(a) shows that the NP (black dots) are nearly uniformin size ( 22 nm) and homogeneously dispersed in the matrix. Upon increasing

3 H.-J. Chung et al.: Effects of nanoparticles on phase separation 221 Fig. 1 TEM images of PMMA : SAN films with (a) 0.5 wt% and (b, c) 5 wt% NP. NP (black dots) are uniformly dispersed after spin coating (a, b). After phase separation (185 C, 900 min), PMMA (white) and SAN (gray) phases are observed (c). Scale bars are 200 nm. The cartoon in (d) represents the morphology during the IS. Note, the film thickness and domain size are not to scale. loading to 5.0 wt%, the NP maintain their uniform distribution, although some small clusters are observed as shown in fig. 1(b). However, no evidence of phase separation is observed. Upon annealing at 185 C for 900 min, isolated PMMA domains (white) form with a diameter of 200 nmas shown in fig. 1(c). The lower electron density is attributed to the degradation (i.e., chain scission) of PMMA under the electron beam. The NP appear to partition into the PMMA domains relative to the SAN matrix (gray), indicating that the NP are free to diffuse. Because plane-view TEM only shows a projection of the NP through the film, a complementary technique is required to determine the depth distribution of NP and thus whether the NP in the gray region are located in the PMMA wetting layers or SAN mid-layer (cf. fig. 1(c)). The silica NP depth distribution is determined from the silicon profile measured by RBS. For 700 nm films containing 5 wt% NP, fig. 2 shows that the NP are uniformly dispersed in the bulk (solid squares), although some surface segregation occurs during spin coating. To observe the entire 700 nm film, normal incidence RBS (not shown) was performed and revealed a uniform concentration of NP throughout the film. After 90 min (i.e., end of ES) [16], NP segregation to the surface is enhanced. Studies are underway to investigate NP surface segregation during the ES [24]. Under similar annealing conditions, PMMA : NP and SAN : NP blends show no evidence of NP surface segregation. After 480 min (filled circles), the NP surface excess decreases, consistent with backflow of PMMA which signals the beginning of the IS [16]. A comparison of the NP profiles at 480 min and 7200 min (open circles) shows that the surface excess remains relatively constant throughout the IS, suggesting that the PMMA wetting layer is pinned due to the high density of NP (i.e., high viscosity). During the IS,

4 222 EUROPHYSICS LETTERS silica conc. [wt%] Depth [A] As Cast 90min 480min 7200min Fig. 2 The near-surface silica depth profiles of 700 nm films with 5 wt% NP annealed at 195 C for 0, 90 min, 480 min, and 7200 min. The IS begins at 90 min and ends at 7200 min. the bulk NP concentration decreases from5.0 to 2.5 wt%. Using this bulk value, a NP surface excess of 7.3 nm, and mass conservation, the interfacial excess of NP at the oxide surface is 10.2 nm. In summary, depth profile analysis establishes that NP are mobile and partition into the PMMA wetting layers. Combining the TEM and RBS studies from figs. 1 and 2, the NP distribution can be represented by the cross-sectional and plane-view images shown in fig. 1(d). The PMMA and SAN are gray and black, respectively, whereas the open circles are the NP. In plane view, the NP in the PMMA wetting layers are superimposed on the SAN mid-layer (cf. fig. 1(c)). Although fig. 1(d) assumes complete partitioning of NP into PMMA, a minority of NP may be trapped in the mid-san layer. Note, fig. 1(d) represents the morphology only during the IS. The morphology of PMMA : SAN films (650 nm) without and with 2 wt% NP were analyzed by SFM. Representative images are shown in figs. 3(a-c) and 3(d-f), respectively. Figures 3(a-b) and (d-e) reflect the surface features whereas figs. 3(c) and (f) show the interface morphology after removing PMMA. The height images show a homogeneous distribution of 10 nmhigh hills (light) in both figs. 3(a) and (d). Over the same scan areas as figs. 3(a) and (d), the phase images in figs. 3(b) and (e) suggest that the hills and surroundings have similar moduli for PS : PMMA but differ upon adding NP [25]. The rings in fig. 3(b) are artifacts caused by abrupt height changes [26]. In fig. 3(e), the hill itself is stiffer (light) than the surroundings (dark). The height images in figs. 3(c) and (f) show a continuous SAN matrix (light) punctured by mainly circular holes (dark). Surface hills formin response to capillary-force driven flow of PMMA towards the surface during the ES [16]. Because flow occurs along PMMA tubes oriented perpendicular to the wetting layer, the hills map onto the PMMA domains as observed by comparing feature sizes in figs. 3(a), (c), and 3(d), (f), respectively. Furthermore, from fig. 3(e), the hills exhibit a higher modulus than the surroundings, consistent with NP partitioning into the PMMA domains. This interpretation is consistent with the lateral partitioning and surface enrichment of NP observed by TEM and RBS; namely, the SFM studies support the morphology depicted in fig. 1(d). The phase separation dynamics is quantified by the correlation length, ξ(t) =2π/k max (t), where k max (t) is the dominant wave vector. In SFM experiments, k max (t) is determined from the fast-fourier transform(fft) (inset) of the interface morphology in figs. 3(c) and (f) [18].

5 H.-J. Chung et al.: Effects of nanoparticles on phase separation 223 Fig. 3 SFM images (40 µm scan) of 650 nm films with 0 wt% (a-c) and 2 wt% (d-f) NP at 195 C for 1440 min. Height ((a) and (d)) and phase ((b) and (e)) images were from the same area, whereas (c) and (f) are interface morphologies (PMMA removed). The insets in (c) and (f) are FFTs of the respective morphologies. The ξ values are plotted vs. t 1/3 in fig. 4. A striking feature is that phase growth substantially slows down with a small amount of NP. For example, after 2880 min, the addition of 5 wt% ( 2 vol%) reduces ξ by 50% compared to the PMMA : SAN blend. For all systems, ξ scales as Kt 1/3, where K decreases as wt% NP increases. This scaling behavior is in agreement with correlation length (ξ) [nm] no silica 2wt% 5wt% (annealing time) 1/3 [min 1/3 ] Fig. 4 The in-plane domain correlation length, ξ, vs. t 1/3 at 195 C. The dashed lines are linear fits for 0 wt%, 2 wt% (650 nm) and 5 wt% (700 nm) NP. Data are determined from FFTs (cf. fig. 3).

6 224 EUROPHYSICS LETTERS a coalescence model [18, 27], ξ (σ/η) 1/3 d 2/3 t 1/3, (1) where σ is the PMMA : SAN interfacial energy and η is the viscosity of PMMA. The mechanism of slowing-down can be understood by applying eq. (1). For example, if NP segregate to the domain/matrix interface similar to surfactants [28], σ will be reduced resulting in a decrease in the driving force for domain growth. However, TEM (fig. 1c) and RBS (fig. 2) studies strongly suggest that particles partition fairly uniformly within the PMMA domains and do not accumulate at the PMMA/SAN boundary. Furthermore, in SAN : NP/PMMA bilayers, the NP freely diffuse into PMMA without accumulating at the interface. Several models account for the effect of viscosity on phase separating blends containing NP. Ginzburg et al. [11] argued that particles impede interface motion and thus slow down domain growth. Similarly, Tang and Ma [12] argued that particles inhibit the shape relaxation of domains due to excluded volume. Tanaka et al. [5] proposed that glass particles perturb the fluid flow inside the domain, causing friction and slower domain motion. Utilizing these concepts, we propose that an increase in the effective viscosity of PMMA (η eff ) due to NP accounts for the slowing-down of domain growth. In most experiments, the addition of NP increases the polymer viscosity [29, 30]. Theoretical studies attribute this behavior to an attraction between polymer and NP [31, 32], and confinement [33]. In the case of PMMA : NP, the viscosity of PMMA increases with NP concentration [34]. Using eq. (1) and σ = const, we can test whether η eff accounts for the experimental scaling behavior of ξ. Fromfig. 4, the ratios of slopes with and without NP, K/K 0, are 0.64 and 0.43 for 2 wt% and 5 wt% NP, respectively. Using eq. (1), the scaling behavior of ξ predicts that η eff /η 0 are 3.7 and 12.6, respectively. Because of partitioning during phase separation, the effective NP concentrations in the PMMA phase are 4 wt% and 10 wt%. Thus, the extracted values of η eff /η 0 are in reasonable agreement with bulk measurements on PMMA : NP blends [34]. In conclusion, the behavior of ξ shown in fig. 4 is consistent with a viscosity controlled phase separation mechanism. In summary, the first quantitative experimental study of phase separation dynamics in a polymer blend film containing mobile NP is presented. In a homogeneous blend of PMMA and SAN, silica NP are shown to be initially uniformly dispersed. During phase separation, the NP are mobile and partition into PMMA wetting layers and PMMA domains. Domain growth slows down with the addition of NP, and the kinetics of phase separation is in agreement with a coalescence model, which predicts ξ η 1/3 t 1/3. Thus, for PMMA : SAN : NP nanocomposites undergoing IS phase separation, the slowing-down of domain growth is mainly attributed to the enhancement of the PMMA viscosity due to the addition of NP. The authors thank Prof. K. Winey and Mr. F. Du for insightful discussions, Ms. J. Deng for rheology measurements, and Nissan Chemical for NP. Work was supported by the NSF, Polymer and MRSEC Programs. REFERENCES [1] Callister W. D., Materials Science and Engineering: An Introduction, 5th edition (Wiley & Sons) [2] Schmidt G. and Malwitz M. M., Curr. Opin. Colloid Interface Sci., 8 (2003) 103. [3] Balazs A. C., Curr. Opin. Colloid Interface Sci., 4 (2000) 443.

7 H.-J. Chung et al.: Effects of nanoparticles on phase separation 225 [4] Gunton J. D., San Miguel M. and Sahni P. S., inphase Transition and Critical Phenomena, edited by Domb C. and Lebowitz J. L., Vol. 8 (Academic Press, London) 1983, pp [5] Tanaka H., Lovinger A. J. and Davis D. D., Phys. Rev. Lett., 72 (1994) [6] Gubbels F. et al., Macromolecules, 28 (1995) [7] Yurekli K. et al., Macromolecules, 36 (2003) [8] Chakrabarti A., J. Chem. Phys., 111 (1999) [9] Qiu F. et al., J. Chem. Phys., 115 (2001) [10] Laradji M. and MacNevin G., J. Chem. Phys., 119 (2003) [11] Ginzburg V. V. et al., Phys. Rev. Lett., 82 (1999) [12] Tang Y.-L. and Ma Y.-Q., J. Chem. Phys., 116 (2002) [13] Qiu F. et al., Langmuir, 15 (1999) [14] Balazs A. C. et al., J. Phys. Chem. B, 104 (2000) [15] Ma Y.-Q., Phys. Rev. E, 62 (2000) [16] Wang H. and Composto R. J., J. Chem. Phys., 113 (2000) [17] Wang H. and Composto R. J., Phys. Rev. E, 61 (2000) [18] Chung H.-J. and Composto R. J., Phys. Rev. Lett., 92 (2004) [19] Wang H. and Composto R. J., Macromolecules, 35 (2002) [20] Wang H., PhD Thesis, University of Pennsylvania (1999). [21] Newby B. Z. and Composto R. J., Macromolecules, 33 (2000) [22] Yoshitake K. and Yokoyama T., U.S. Patent, A (2000). [23] Composto R. J., Walters R. M. and Genzer J., Mat. Sci. Eng. R, 38 (2003) 107. [24] Chung H.-J., Deng J. D., Ohno K., Fukuda T. and Composto R. J., in preparation. [25] Knoll A., Magerle R. and Krausch G., Macromolecules, 34 (2001) [26] Dimension 3100 User s Manual v4.43b (Digital Instruments) [27] Tanaka H., J. Phys. Condens. Matter, 13 (2001) [28] Kawakatsu T. et al., J. Phys. Condens. Matter, 6 (1994) [29] Krishnamoorti R., Ren J. and Silva A. S., J. Chem. Phys., 114 (2001) [30] Steinmann S., Gronski W. and Friedrich C., Polymer, 43 (2002) [31] Smith G. D., Bedrov D. and Borodin O., Phys. Rev. Lett., 90 (2003) [32] Salaniwal S., Kumar S. K. and Douglas J. F., Phys. Rev. Lett., 89 (2002) [33] Semenov A. N., Phys. Rev. Lett., 80 (1998) [34] Preliminary rheology experiments show that η eff /η 0 is 8 for 10 wt% NP.