Direct Imaging of Surface and Bulk Structures in ... - ACS Publications

Department of Chemistry, University of Toronto, Toronto, Ontario, Canada M5S 3H6, ... of Engineering Science, University of Western Ontario, London, O...
0 downloads 0 Views 353KB Size
Download PDF
Langmuir 1997, 13, 2483-2489

2483

Direct Imaging of Surface and Bulk Structures in Solvent Cast Polymer Blend Films Eugenia Kumacheva,*,† Lin Li,† Mitchell A. Winnik,† Doug M. Shinozaki,‡ and P. C. Cheng§ Department of Chemistry, University of Toronto, Toronto, Ontario, Canada M5S 3H6, Faculty of Engineering Science, University of Western Ontario, London, Ontario, Canada N6A 5B9, and Department of Electrical and Computer Engineering, State University of New York at Buffalo, Buffalo, New York 14260 Received November 8, 1996. In Final Form: February 11, 1997X We have used laser confocal fluorescent microscopy and atomic force microscopy to study surface and bulk morphologies in solvent cast poly(methyl methacrylate) (PMMA)/polystyrene (PS) films. The solvent evaporation rate strongly affects the surface morphology. Sufficient suppression of solvent evaporation results in a periodic distribution of highly monodisperse PMMA-rich domains. Formation of the surface pattern is interpreted in terms of diffusion-driven coarsening of the minor phase and flow induced by the diffusive transport of PMMA molecules from the bulk of the liquid film to the air-film interface.

Introduction Over the last 10 years, strong attention has been given to thermodynamic, interfacial, and rheological properties of polymer mixtures, both in the bulk and at surfaces.1-4 Interfacial effects play an important role in determining the morphology of multicomponent thin polymer films. For example, the phase-separation mechanism is strongly affected by the thermal instability of macromolecules at the surface, surface-induced or suppressed nucleation of the new phase, and surface enrichment with a lower freeenergy component.5-9 As a result, the dynamics of surface domain coarsening and, consequently, surface morphology may differ from those in the bulk. Besides specific surface effects, films obtained from ternary polymer solutions are sensitive to the gradual change in the solvent concentration across the film due to solvent evaporation. As a result, the composition of the film and, hence, the phase behavior of the system can depend on the distance from the air-film interface. Generally this effect is neglected in films formed relatively fast, e.g., by spin casting or evaporation of the volatile solvent. The entire system rapidly enters an unstable * To whom correspondence should be addressed. E-mail: [email protected]. † University of Toronto. ‡ University of Western Ontario. § State University of New York at Buffalo. X Abstract published in Advance ACS Abstracts, April 1, 1997. (1) (a) Polymer blends; Paul, D. R., Newman, S., Eds.; Academic: San Diego, 1990; Vol. 1, 2. (b) Utracki, L. A. Polymer Alloys and Blends: Thermodynamics and Rheology; Wiley: New York, 1989. (2) Bates, F. S. Science 1991, 251, 898. (3) (a) Pincus, P. J. J. Chem Phys. 1981, 75, 1986. (b) Binder, K. J. Chem Phys. 1983, 79, 6387. (4) (a) Hashimoto, T. In Materials Science and Technology; Thomas, E. L., Ed.; VCH: New York, 1993; Vol. 12, pp 251-300. (b) Hashimoto, T.; Sasaki, K.; Kawai, H. Macromolecules 1984, 17, 2812. (5) (a) Green, P. F.; Christensen, T. M.; Russel, T. P.; Jerome, R. Macromolecules 1989, 22, 2189. (b) Anastasiadis, S. H.; Russel, T. P.; Satija, S. K.; Majkrzak, C. F. J. Chem. Phys. 1990, 92, 5677. (6) (a) Dierker, S. B.; Wiltzius, P. Phys. Rev. Lett. 1991, 66, 1185. (b) Liu, A. J.; Durian, D. J.; Herbolzheimer, H.; Safran, S. A. Phys. Rev. Lett. 1990, 65, 1897. (c) Liu, A.; Grest, G. S. Phys. Rev. A 1991, 44, 44. (7) (a) Sagui, C.; Somosa, A. M.; Roland, C.; Desai, R. C. J. Phys. A: Math Gen. 1993, 26, L1163. (b) Rogers, T. M.; Desai, R. C. Phys. Rev. B 1989, 39, 11956. (8) (a) Cumming, A.; Witzius, P.; Bates, F. S.; Rosedale, J. H. Phys. Rev. A 1992, 45, 885. (b) Cumming, A.; Wiltzius, P.; Bates, F. S. Phys. Rev. Lett. 1990, 65, 863. (9) (a) Wiltzius, P.; Cumming, A. Phys. Rev. Lett. 1991, 66, 3000. (b) Shi, B. Q.; Harrison, C.; Cumming, A. Phys. Rev. Lett. 1993, 70, 206.

S0743-7463(96)01089-X CCC: $14.00

region. Phase separation occurs by spinodal decomposition until the system reaches the glass composition with the subsequent “freezing” of the nonequilibrium structure. When film formation is prolonged in time (i.e., with slow solvent evaporation), then other factors can affect film morphology. Under these circumstances, the concentration of the solvent at the air-liquid interface and the layer adjacent to it remains lower than that in the bulk for sufficient lengths of time. Phase separation can begin at the surface, with little or no phase separation in the bulk. The solvent-rich interior can act as a reservoir of material for the domains growing at the surface. This effect has been studied for off-critical systems via Monte Carlo simulations by Sagui et al.7a They found that under certain quench conditions, ordered domains of the minor phase were formed in the surface layer which then grew faster than the bulk domains. In the experiments described here, we use fluorescence microscopy to study the surface and bulk effects in film morphology obtained on slow quenching of a ternary system below the coexistence curve. We consider the solvent evaporation rate to be a quench parameter, which allows us to modify the phase separation process. To study in situ both the surface and the bulk structure of the films, we apply laser confocal fluorescent microscopy (LCFM).13 The advantage of confocal fluorescent imaging is that only light from a particular focal plane (ca. 1 µm thick) reaches the detector. This provides the information from a specific level of the specimen, eliminating contribution of light from adjacent layers. Imaging slices at different depths below the surface of the material enables one to detect the transition from surface to bulk morphology. Recently, several publications appeared10-12 reporting that LCFM can be successfully used to study the morphology of multicomponent polymer systems. Li et al.10 used LCFM to study the surface and bulk morphologies of PS/PMMA films and found a remarkable difference in the structure. Jinnai and co-authors11 and White and Wiltzius12 employed LCFM with subsequent three-dimensional recon(10) Li, L.; Sosnowski, S.; Chaffey, C. E.; Balke, S. T.; Winnik, M. A. Langmuir, 1994, 10, 2495. (11) Jinnai, H.; Nishikawa, Yu.; Koga, T.; Hashimoto, T. Macromolecules 1995, 28, 4782. (12) White, W. R.; Wiltzius, P. Phys. Rev. Lett. 1995, 75, 3012. (13) (a) Multidimensional Microscopy; Cheng, P. C., et al, Eds.; Wiley: New York, 1994. (b) Confocal Microscopy; Wilson, T., Ed.; Academic: London, 1990.

© 1997 American Chemical Society

2484 Langmuir, Vol. 13, No. 9, 1997

Kumacheva et al. tion was calculated as

∆P ) msRT/MV

(1)

struction analysis to determine the structure in the bulk of phase-separating two-component polymer blends. In the present paper, we employ the LCFM approach for nondestructive depth profiling in polymer blend films cast from solution at different solvent evaporation rates. We use LCFM and atomic force microscopy (AFM) to show that suppression of solvent evaporation from polystyrenepoly(methyl methacrylate) [PS-PMMA]-toluene liquid films significantly modifies the morphology of the top surface (air-film interface). The surface morphology exhibits a uniform distribution of nearly monodisperse particles of the PMMA-rich phase. A depletion layer (ca. 18 µm) almost free of the PMMA particles is located beneath the top surface, while the rest of the film exhibits a random distribution of the polydisperse domains. These phenomena are interpreted in terms of an evaporationcondensation coarsening mechanism and flux induced by the diffusion of PMMA molecules to the air-film interface. The bulk structure in films formed by the controlled slow evaporation as well as the surface and bulk structures in films formed by “fast” solvent evaporation shows a nonordered distribution of polydisperse PMMA particles.

where ms is the weight of evaporated solvent placed into the container prior to film casting, M ) 92 is the molecular mass of toluene, V ) 2.85 × 10-6 m3 is the volume of the sealed container, R is the gas constant, and T is the temperature, T ) 292 ( 0.5 K. Figure 1 shows the effect of the excess vapor pressure on the mean toluene evaporation rate from the film. The evaporation rate k was determined as msf/τ, where msf is the amount of solvent in the liquid film and τ is the time of complete film drying. By increasing the amount of extra solvent placed into the container prior to film casting, we could decrease the toluene evaporation rate and expand the film formation time from 10 min at free evaporation in air to periods ranging from 40 min to 8 h. In this paper, we discuss the results of morphology studies in PS-PMMA films obtained primarily at two toluene evaporation rates, 0.022 and 0.16 g/h, designated as “slow” and “fast” modes, respectively. In a very few experiments, films were formed at a slower evaporation rate of 0.013 g/h, and by free evaporation in air at a faster evaporation rate of 0.52 g/h. Prior to the morphology studies, blend films were dried in a vacuum oven overnight at 90 °C. Methods. Both confocal microscopes employed the 488-nm line from an Ar ion laser for the dye excitation. The surface and bulk morphologies of the composite films were studied with a Bio-Rad MRC 600 confocal microscope. Lateral and vertical resolutions were on the order of 0.3 and 0.7 µm, respectively. The cross-section microstructure of the film was examined by carefully fracturing the specimen in a plane perpendicular to the specimen surface and looking at the edge of the fracture surface. For these experiments, we used an Olympus GB 200 confocal microscope with an oil immersion objective lens. The lateral resolution in the cross-section plane was 0.15 µm, while resolution in the z axis (in this case lying parallel to the original specimen plane) was 0.6 µm. The image analysis program Global Lab Image was used to analyze the average radii of the PMMA-rich domains in the LCFM images. To study the topology of the top film surface, the samples were imaged in air by atomic force microscopy using either a Nanoscope II or Nanoscope III (Digital Instruments, Inc.) in the contact, constant-force mode. The spring constant of the Si3N4 100-µm cantilever was 0.58 N/m.

Experimental Section

Results and Discussion

Materials. Both polymers were synthesized by standard surfactant-free emulsion polymerization. Experimental details of the polymer synthesis are reported by Sosnowski et al.14 To achieve contrast between PS and PMMA, the latter polymer was covalently labeled with the fluorescent dye 4-amino-7-nitrobenzo2-oxa-1,3-diazol (NBD). The labeling content of NBD uniformly distributed among the PMMA chain was in the range 0.015-1 mol %. The molecular weights determined by gel permeation chromatography were for PS Mw ) 234 000 and Mw/Mn ) 6.3 and for PMMA Mw ) 306 000 and Mw/Mn ) 2.3. Samples Preparation. Polymer films were prepared by casting a filtered (0.2-µm Millipore PTFE filter) toluene solution with a total polymer concentration 4 wt % onto clean quartz plates. The weight ratio of PMMA/PS in the solution was 1:9. The rate of solvent evaporation from the liquid film was controlled through the known vapor-phase pressure of toluene in the following way: a calculated amount of toluene was placed into the sealed container shown in Figure 1 (inset) and allowed to evaporate. Then a known amount of the polymer solution was cast onto a quartz plate positioned into the same hermetically sealed container. The total amount of solvent (both added as a pure liquid and that in the solution), after evaporation, was less than the amount that would saturate the container with toluene vapor. The excess pressure ∆P caused by pure solvent evapora-

Surface Morphology. Figure 2 displays images of the air-film surface in the composite PS-PMMA films formed at different solvent evaporation rates. PMMArich particles labeled with the fluorescent dye NBD are prominent against the dark PS background. The surface morphology of the film obtained at the evaporation rate 0.52 g/h is shown in Figure 2a. Two different length scales can be clearly seen in the image: a great number of small (3-6 µm) PMMA particles randomly distributed in the PS matrix accompanied by large PMMA-rich domains with dimensions in the range 15-30 µm. Many of the domains are nonspherical in shape, and some comprise distinct dimers and trimers of individual particles. These films were formed by free toluene evaporation on the stage of the confocal microscope, where we could directly visualize the coalescence of small PMMA-rich drops into larger domains which then “froze” into the structure observed. Decreasing the solvent evaporation rate to 0.16 g/h (Figure 2b) results in a few effects. The domains are more circular in Figure 2b compared to those in Figure 2a. Both the number and the surface area fraction of the PMMA-rich domains at the air-film interface decrease. The changes occur in the large domains, which exhibit reduced dimensions (8-12 µm), while small particles do not significantly change their size. Dramatic changes in film structure occur at sufficiently suppressed toluene evaporation rates (0.022 g/h). Figure

Figure 1. Effect of excess vapor pressure on the mean evaporation rate of toluene from the PMMA-PS-toluene film. Inset: scheme of the method used to control toluene evaporation from the cast film.

(14) Sosnowski, S.; Feng, J.; Winnik, M. A. J. Polym. Sci., Part A: Polym. Chem. 1994, 32, 1497. (15) (a) Venugopal, G.; Krause, S. Macromolecules 25, 4626. (b) Law, W. W. Y.; Burns, C. M.; Huang, R. Y. M. J. Appl. Polym. Sci. 1985, 30, 1187.

Direct Imaging of Surface and Bulk Structures

Langmuir, Vol. 13, No. 9, 1997 2485

Figure 2. Surface morphology of the cast from toluene PMMA-PS films imaged by LCFM. Bright domains represent the dyelabeled PMMA-rich phase. Films are obtained at the following mean toluene evaporation rates, g/h: (a) 0.52; (b) 0.16; (c) 0.022; (d) 0.013.

2c displays the surface morphology of one of these films. Two remarkable features are clearly seen in this image. First, an ordered pattern of PMMA-rich domains regularly distributed in the PS matrix appears at the air-film interface. The mean closest neighbor-to-neighbor distance is ca. 12 µm. Second, the PMMA-rich particles become highly monodisperse, with a mean particle radius of 2.2 µm. Further suppression of the toluene evaporation rate to 0.013 g/h (Figure 2d) does not result in visible changes in the film structure. To quantify the effect of solvent evaporation rate on the film surface morphology, we calculated the distribution of areas of the PMMA-rich domains. The area contribution of each size class of particles was divided by the area occupied by the total number of domains at the surface. The area distribution function (Σiniri2/Σniri2) is displayed in Figure 3. Rapid evaporation (0.52 g/h) produces a structure characterized by a bimodal area distribution with the peaks corresponding to 3.0 and 9 µm (curve 1). The small particles exhibit a relatively narrow distribution in the range 1.5-4 µm, while the large domains are polydisperse. Reducing the solvent evaporation rate to 0.16 g/h (curve 2) shifts the area distribution curve toward smaller particle sizes with a single maximum at 2.1 µm. Further suppression of the toluene evaporation rate (0.022 g/h) results in an extremely narrow area distribution, displayed by curve 3. However, if the evaporation rate is reduced to 0.013 g/h, the change of morphology is not significant: the area distribution curve slightly broadens

Figure 3. Area distribution function calculated for PMMArich domains at the air-film interface in films obtained at different mean evaporation rates, g/h: 0.52 (s); 0.16 (- -); 0.022 (----); 0.013 (‚‚‚).

and shifts toward domains of larger radius (curve 4, inset to Figure 3). Bulk Morphology. To distinguish between the surface and bulk effects, we studied the bulk morphology of the films formed at slow and fast solvent evaporation rates. The images shown in Figure 4 present the structure of these films at a depth equal to ca. one-third of the film thickness. The striking difference observed between these films in their surface structure disappears in the bulk. Both films formed at fast (Figure 4a) and slow (Figure 4b)

2486 Langmuir, Vol. 13, No. 9, 1997

Figure 4. Confocal fluorescent images of the bulk structure in the cast films obtained in the “fast” (a) and “slow” (b) evaporation modes (see in text). Films were examined at a depth of ca. 0.3 of the total film thickness.

evaporation exhibit a random distribution of polydisperse particles. There are many small PMMA-rich domains with dimensions smaller than those on the surface. However, in contrast to the films formed in the fast evaporation mode, most of the particles in the films obtained by suppressed solvent evaporation are spherical in shape and larger in size. It is obvious that slow solvent evaporation has a large effect on surface morphology, while at depths of more than 20 µm from the surface the evaporation rate does not play an important role. To see how far in depth this influence penetrates the film structure, we studied the morphology in the cross section of a film obtained from slow solvent evaporation. Figure 5 shows the cross-section structure of this fractured film, in which three zones can be identified: (1) The film-air interface with half-spherical PMMA-rich particles, a zone that corresponds to the image shown in Figure 2c; (2) a depletion layer in this image of about 18 µm, almost free of PMMA-rich particles; (3) a bulk area with a random distribution of the small domains, whose morphology compares well with that shown in Figure 4b. The mean particle radii and interparticle distances in the first and third zones are very similar to those observed in the images obtained through optical slices from the film surface. However, the presence of the depletion layer below the air-film interface is a surprising phenomenon. We never observe this layer in films obtained in the “fast” evaporation mode, where polydisperse spherical particles were randomly distributed across the film.

Kumacheva et al.

To interpret the experimental results, we refer to the phase behavior of the PMMA-PS-toluene ternary mixture. Prior to film casting, the PMMA-PS-toluene solution, with a ratio of concentrations 0.4:3.6:96, respectively, does not show any evidence of phase separation. Phase diagrams of similar systems15 also show one-phase behavior in this concentration region. On removal of toluene, the system shifts from a three-component to a two-component region. In principal, the ultimate film morphology is determined by the superposition of effects related to three- and two-component systems. However, since the samples were not annealed prior to LCFM measurements, we assign all features of the structure to three-component effects. In systems characterized by a large ratio of the majorto-minor component, as in the composition considered in the present work, phase separation is known to proceed in a metastable regime via nucleation and growth.7-9 Nonetheless, it is the solvent evaporation rate that determines how long the system stays between the coexistence and spinodal decomposition curves and when it moves to instability. When solvent evaporation is fast, the system rapidly passes through the metastable region. Phase separation initiated by nucleation of the PMMArich phase is, at later times, overtaken by spinodal decomposition. The well-known bicontinuous structure formed by spinodal decomposition breaks into isolated domains due to surface-tension-driven effects.1-4 Latestage particle growth via this mechanism is determined mainly by droplet coalescence. Two different length scales in the surface structure seen in Figure 2a can be explained by the dual origin PMMA-rich domain coalescence. The small particles which form initially collide via Brownian motion and adhere to form dimers, trimers, and larger aggregates which subsequently merge into PMMA-rich drops. The very large PMMA domains seen in Figure 2a are created as a result of multiple-particle aggregation induced by local surface flow. Such flow has its origin in thermo- or mechanical instabilities caused by rapid solvent evaporation. In a different experimental work, we examined flow-induced formation of large PMMA-rich domains in the same system. First, arrays of small droplets appeared at the surface, then they merged to form noncircular domains, and later this domain acquired a circular shape. Solvent evaporation can “freeze in” the developing structure at any stage; as a consequence, large domains in the blend film may preserve a nonspherical shape as shown in Figure 3, curve 1. Decreasing the evaporation rate reduces surface flow in the liquid film. One result is that no very large aggregates are formed (Figure 2b). On the other hand, the longer time for film formation provides more efficient droplet coalescence as a result of Brownian motion and leads to the formation of particles which are spherical in shape. The situation changes on significant suppression of the solvent evaporation rate. The time interval in which the system remains in the metastable region increases. Phase separation begins at the air-film interface, while the bulk of the specimen is still in the one-phase regime. There are at least two reasons for the preferred phase separation at the surface. First, due to toluene evaporation from the film surface, the two-dimensional concentration of the constituent components moves below the coexistence curve of the phase diagram. Surface-phase separation proceeds by the diffusion of free PMMA molecules from the supersaturated medium to nuclei of condensation. In

Direct Imaging of Surface and Bulk Structures

Langmuir, Vol. 13, No. 9, 1997 2487

Figure 5. Cross-section morphology of the PMMA-PS film prepared at a solvent evaporation of 0.022g/h.

addition, heterogeneous nucleation at the top liquid film surface is generated by wetting effects.16 The monodispersity of the surface domains (Figure 2c) was to us an unexpected feature. Our most compelling explanation is that a limited number of nucleation centers at the interface eliminate the appearance of the new droplets while incipient nuclei grow. We also have to assume that the coarsening droplets do not interact with each other. A similar effect has been observed for offcritical quenches in the bulk of the polyisoprenepoly(ethylene-propylene) polymer blend.8a In that system, extremely monodisperse droplets of the minor phase proceeded growth with the radius R(t) ∼ t1/2. The authors explained this effect by a heterogeneous nucleation of the minor-phase droplets on a fixed number of sites in the bulk of the blend. However, in contrast to our work, Cumming et al.8a do not report specific surface effects for off-critical concentrations, and their entire system was characterized by a single length scale. In our system, as time progresses, PMMA-rich droplets emerge beneath the top film surface, at first in the layers adjacent to the interface and then later in the bulk. These droplets are smaller than the surface domains. Because of this, they are unstable with respect to the surface droplets, and this drives the interaction between the surface and bulk PMMA-rich domains via the Ostwald (16) Supersaturation required for the nucleation of the new phase at the interface is essentially affected by the wetting of the interface by the incipient phase. For the nuclei with the same radius Wchet ) f(θ)Wchom, where Wchet and Wchom are the energies of nucleus formation in heterogeneous and homogeneous condensation, respectively, and f(θ) is a function of the wetting angle θ of the new phase on the surface.17. f(θ) ) 1/4(1 - cos θ)2(2 + cos θ). When θ varies from 0° (complete wetting) to 180° (nonwetting conditions), f(θ) increases from 0 to 1. This implies that in good wetting conditions, heterogeneous nucleation of the new phase starts at a lower solution supersaturation than needed for the homogeneous nucleation. In our experiments, the wetting angle of PMMA-rich phase measured at the air-film interface changes in a range 88-92° which provides f(θ) ) 0.47-0.52 (in liquid films, this value should be even lower), so the nucleation of the PMMA-rich phase at the air-film interface occurs earlier than in the bulk. (17) Hiemenz, P. C. Principles of Colloid and Surface Chemistry; Marcel Dekker: New York, 1986; p 815.

ripening mechanism. Small droplets dissolve while large particles grow at the expense of the small ones due to the higher solubility of the small particles.18 More information is provided by the image shown in Figure 5, where we observe that the PMMA domains at the surface (zone 1) have a hemispherical shape. The depletion layer (zone 2) occurs as the result of transfer of PMMA molecules from the bulk to the surface of the film. This process includes both free-molecule diffusion from the supersaturated solution and the diffusive interaction between the bulk and surface droplets. In this sense, the layer of the liquid film below the surface plays the role of a reservoir of material for the surface PMMA-rich domains. At the moment, we cannot resolve the contribution of each process in the surface film morphology. An increase in the area fraction of the larger domains and a broadening of the size distribution curve in films obtained at the slowest evaporation rate k ) 0.013 g/h (Figure 3, inset, curve 4) may indicate the increasing influence of Ostwald ripening at the later times of film formation. Cumming et al.8a also observed an increase in polydispersity at later stages of domain growth and explained it by the transition from free diffusion to the evaporation-condensation mechanism. However, in our work, the difference in area distribution of the PMMArich domains in films formed at k ) 0.22 and 0.13 g/h is too small to make a conclusion about the switch between different growth mechanisms. The evaporation-condensation mechanism as an alternative to coalescence in polymer blends has been recently discussed in a number of publications.12,19 Limited by low polymer diffusivity (∼10-11-10-15 cm2/s), this mechanism rarely plays a major role in particle coarsen(18) Livshitz, I. M.; Slyosov, V. V. J. Phys. Chem. Solid 1961, 19, 35. (19) (a) Crist, B., Nesarikar, A. R. Macromolecules 1995, 28, 890. (b) Mirabella, F. M., Jr. J. Polym. Sci. Part B: Polym. Phys. 1994, 32, 1205. (c) Mirabella, F. M., Jr.; Barley, J. S. J. Polym. Sci., Part B: Polym. Phys. 1994, 32, 2187. (20) Tanaka, K., Takahara, A., Kajiyama, T. Macromolecules 1996, 29, 3232.

2488 Langmuir, Vol. 13, No. 9, 1997

ing.21 However, in ternary polymer systems, two features promote Ostwald ripening. First, polymer solubility is enhanced. In addition, the diffusion coefficient increases to 10-6-10-7 cm2/s, which, for the system studied here, compensates for the ca. 102 reduction in the interfacial tension.15 As toluene evaporates, the diffusion coefficient of PMMA molecules decreases to the point where mass transfer between the bulk and surface of the film finally ceases. The thickness of the depletion layer is determined by the critical diffusion length beyond which mass transfer from the bulk to the surface becomes ineffective; thus, this parameter is also affected by the rate of solvent evaporation. The structure seen in the bulk of the film below the depletion layer (Figures 4a and 5, zone 3) is a result of a bulk PMMA-rich phase nucleation, followed by domain interaction via both Ostwald ripening and coalescence. The dimensions of the bulk domains are 2-2.5 times smaller than those at the surface. A similar difference in length scales corresponding to the surface and bulk morphologies has been reported for slow quenching in the critical phase separating mixtures of low- and highmolecular weight components.8a,9b In the works of Cumming et al.8a and Wiltzius and Cumming,9b surface domain length scales increased proportionally to t3/2, whereas bulk domain growths scaled t1/3; as a result, different structures were formed in the bulk and at the surface of the specimen. The authors attributed the surface-preferred growth of the minority phase to the wetting phenomena. However, in contrast to our work, these effects were detected for critical polymer compositions which proceeded distinct spinodal decomposition. In addition, the domain distribution was far from monodisperse. Surprisingly, in none of our experiments do we observe complete surface coverage with the lower surface energy PS, as reported by Tanaka et al.20 On the contrary, the surface of our films obtained by free toluene evaporation in air was enriched with the PMMA-rich phase: the surface area fraction was 49%, which is much larger than the bulk ratio of PMMA/PS (Figure 2a). We explain this effect in terms of a flux of droplets from the bulk of the film to the surface driven by rapid solvent evaporation. Lowering the toluene evaporation rate decreases the surface concentration of the minor phase. For example, in parts b, c, and d of Figure 2, we find surface amounts of 8.0, 7.1, and 6.6%, respectively, for the PMMA-rich phase. Such a decrease in the surface area fraction with the solvent evaporation rate may suggest that at a significantly long time of film formation, the surface of the specimen may be covered with PS. Periodicity of the Surface Structure. The suggested mechanism of the PMMA-rich domain growth in films obtained by slow solvent evaporation does not explain the formation of the ordered pattern at the surface of the film. Periodic domain distribution of the minor phase generated by spinodal decomposition was observed in critical or close-to-critical two-component polymer systems.11 However, in the present work, the concentration ratio of the components suggests a nucleation-and-growth mechanism of phase separation rather than spinodal decomposition. Two-dimensional particle ordering observed at the fluid-liquid interface can be attributed either to capillary (21) According to the calculations,18 the increase of the particle mean radius in time drm3/dt in systems with the low miscible components is given by drm3/dt ) 8γDcVm/9RT, where γ is the interfacial tension, Vm is the molar volume, D is the diffusion coefficient, c is the solubility of the minor phase in the major phase, R is a gas constant, and T is the absolute temperature.

Kumacheva et al.

Figure 6. AFM surface plot (a) and sectional view (b) for PMMA-PS films obtained at a toluene evaporation rate of 0.022 g/h.

interactions22 or convection induced by surface-tension gradients.23 To estimate the effect of capillary forces, we examined the surface topography of films obtained by fast and slow evaporation with AFM. A surface image and a cross-sectional plot of the film formed at slow toluene evaporation rates is shown in Figure 6. Here the surface roughness is regular in both the lateral and normal directions. The mean peak-to-peak distance corresponding to the particle center-to-center distance is about 11 µm. The PMMA domain height is 135 ( 15 nm; i.e., the minor-phase particles protrude slightly above the PS matrix. Films obtained at fast solvent evaporation always had an irregular surface topography in both the normal and horizontal directions: domain heights ranged from 90 to 180 nm and the interparticle distances from 2 to 10 µm. The upthrust of PMMA-rich particles in air has been reported by Tanaka et al.20 in thin PMMA/PS films obtained from toluene solution by spin coating. The authors attributed this effect to the formation of a nonequilibrium structure frozen by rapid solvent evaporation. In our work, regardless of the solvent evaporation rate, we always observed that the PMMA-rich domains protrude in air. The images shown in Figures 5 and 6 allow us to draw a sketch of PMMA-rich drops floating at the air-liquid (22) (a) Chan, D. Y.; Henry, J. D.; White, L. R. J. Colloid Interface Sci. 1981, 79, 410. (b) Denkov, N.; Velev, O. D.; Kralchevsky, P. A.; Ivanov, I. B.; Yoshimura, H.; Nagayama, K. Langmuir 1992, 8, 3183. (c) Kralchevsky, P. A.; Paunov, V. N.; Denkov, N.; Velev, O. D.; Ivanov, I. B.; Nagayama, K. J. Colloid Interface Sci. 1993, 155, 420. (23) (a) Koschmieder, E. L. Adv. Chem. Phys. 1974, 26, 177. (b) Tenan, M. A.; Teschke, O.; Kleinke, M. U. Langmuir 1990, 6, 1640. (c) Bragard, J.; Slavtchev, S. G.; Lebon, G. J. Colloid Interface Sci. 1994, 168, 403. (d) Teschke, O.; Kleinke, M. U.; Tenan, M. A. J. Colloid Interface Sci. 1992, 151, 477.

Direct Imaging of Surface and Bulk Structures

Langmuir, Vol. 13, No. 9, 1997 2489

Figure 7. Schematic drawing of the PMMA-rich drops at the fluid interface in liquid PMMA-PS films formed in the “slow” evaporation mode.

film surface (Figure 7). When a liquid drop is placed at the fluid interface, the interfacial tensions are described by the equation of Neuman-Young

b γ PMMA-air + b γ PS-air + b γ PS-PMMA ) 0

(2)

where b γPMMA-air is the interfacial tension between the PMMA-rich phase and air, b γPS-air is the interfacial tension between the PS-rich phase and air, and b γPS-PMMA is the interfacial tension between the PMMA-rich and PS-rich phases. At equilibrium, the equations for the vertical and horizontal components of the interfacial tension are

γPMMA-air sin θ1 ) γPS-PMMA sin θ2

(3)

γPMMA-air cos θ1 + γPS-PMMA cos θ2 ) γPS-air

(4)

The protrusion of the PMMA-rich drops satisfies thermodynamic equilibrium, as it compensates for the vertical component of γPMMA-air. As solvent evaporates, the droplet shape can either be frozen or it can evolve, depending on the solvent evaporation rate. To estimate the energy of the capillary interaction between PMMA-rich drops, we applied eq 47 of Chan et al.22a We used reasonable values of the system parameters, i.e., the mean radius of the PMMA-rich droplet 2.2 µm, the interfacial tension and density of the PS-rich phase of ca. 35 erg/cm2 and 0.9 g/cm3, respectively, and an interparticle distance of ca. 12 µm; θ2, measured from Figure 5, is in the range 88-92°, and θ1, calculated using eqs 3 and 4, is ca. 4.5-5°. We found that the interaction energy of the minor-phase droplets located at the airliquid film interface is 2 orders of magnitude lower than kT. It is very unlikely that the capillary interaction contributes to particle ordering. We prefer to ascribe the formation of the periodic surface pattern to the solutal instability in the liquid film due to mass transfer of PMMA from the bulk to the surface. Nonlinear convection of the dissolved molecules or of small colloidal particles in thin liquid films is known to result in surface pattern formation ranging from hybrid cells to hexagons and polygons.23 The driving force is a mass gradient across the film, which is analogous to Marangoni thermal convection. A measure of the competition between the dissipative forces and the driving forces acting in the film is the solutal Marangoni number

Ma ) (∂γ/∂C)d2/D2η

(6)

Figure 8. LCFM image of the surface morphology for the film formed at a solvent evaporation rate of 0.022 g/h. The film thickness is 22 µm.

where (∂γ/∂C) is the derivative of the film-air interfacial tension with respect to the concentration of the component at the interface, d is the thickness of the film, and D is the diffusion coefficient of the component in the liquid medium with viscosity η. We have not as yet carried out a detailed study of the role of Marangoni solutal effect in PMMA/PS/toluene films. As a preliminary experiment, to evaluate the importance of this effect in films formed at slow evaporation rates, we repeated our experiment with a reduced film thickness d, here 22 µm, keeping all other parameters the same. The evaporation rate was 0.022 g/h, i.e., the same as for the 70-µm films. Since the Marangoni number is proportional to d2, formation of a thinner film should lead to a strong change in the surface pattern. Figure 8 presents an LCFM image of the air-film interface in the 22-µm film. One can see that PMMA-rich domains remain monodisperse, but the periodicity in the particle distribution in PS matrix has disappeared. To conclude, we found that solvent evaporation rate strongly influences the surface morphology in solventcast PMMA-PS films but causes insignificant changes in the bulk structure. Rapid solvent evaporation results in two length scales in the PMMA-PS surface morphology: large domains of ca. 15-30 µm and small domains (3-6 µm) of the PMMA-rich phase. Sufficient suppression of the evaporation rate provides a periodic distribution of highly monodisperse PMMA-rich domains at the air-film interface. Diffusive transfer of PMMA molecules from the bulk of the liquid film to the surface results in a depletion layer free from PMMA-rich particles just below the surface and also leads to the formation of an ordered surface pattern. Acknowledgment. We acknowledge the Ontario Centre for Materials Research and NSERC Canada for their support of this research. Thanks are due Peter Markiewicz for his assistance in the AFM experiments. LA961089I