Langmuir 2007, 23, 4405-4413
4405
Aggregation, Adsorption, and Surface Properties of Multiply End-Functionalized Polystyrenes Imtiyaz A. Ansari,† Nigel Clarke,† Lian R. Hutchings,† Amilcar Pillay-Narrainen,†,‡ Ann E. Terry,§ Richard L. Thompson,*,† and John R. P. Webster§ Department of Chemistry, Durham UniVersity, Science Site, South Road, Durham, DH1 3LE, U.K. and ISIS Facility, CCLRC Rutherford Appleton Laboratory, Chilton, Didcot OX11 0QX, U.K. ReceiVed December 11, 2006. In Final Form: January 24, 2007 The properties of polystyrene blends containing deuteriopolystyrene, multiply end-functionalized with C8F17 fluorocarbon groups, are strikingly analogous to those of surfactants in solution. These materials, denoted FxdPSy, where x is the number of fluorocarbon groups and y is the molecular weight of the dPS chain in kg/mol, were blended with unfunctionalized polystyrene, hPS. Nuclear reaction analysis experiments show that FxdPSy polymers adsorb spontaneously to solution and blend surfaces, resulting in a reduction in surface energy inferred from contact angle analysis. Aggregation of functionalized polymers in the bulk was found to be sensitive to FxdPSy structure and closely related to surface properties. At low concentrations, the functionalized polymers are freely dispersed in the hPS matrix, and in this range, the surface excess concentration grows sharply with increasing bulk concentration. At higher concentrations, surface excess concentrations and contact angles reach a plateau, small-angle neutron scattering data indicate small micellar aggregates of six to seven F2dPS10 polymer chains and much larger aggregates of F4dPS10. Whereas F2dPS10 aggregates are miscible with the hPS matrix, F4dPS10 forms a separate phase of multilamellar vesicles. Using neutron reflectometry (NR), we found that the extent of the adsorbed layer was approximately half the lamellar spacing of the multilamellar vesicles. NR data were fitted using an error function profile to describe the concentration profile of the adsorbed layer, and reasonable agreement was found with concentration profiles predicted by the SCFT model. The thermodynamic sticking energy of the fluorocarbon-functionalized polymer chains to the blend surface increases from 5.3kBT for x ) 2 to 6.6kBT for x ) 4 but appears to be somewhat dependent upon the blend concentration.
Introduction Demand for polymers with bespoke surface and bulk properties has resulted in a large volume of research into additives, fillers, blends, coatings, and surface treatments. (See, for example, ref 1.) Common to all of these approaches is the need for good understanding and control of surface and interfacial processes such as the adhesion between filler particles or fibers and a host polymer or the chemistry of the exposed polymer surface. In a situation where it is desirable to enhance a surface property, such as water repellency, one would ideally wish to achieve this without detriment to the bulk properties such as mechanical strength of the host polymer. Small-molecule additives are seldom ideal for this purpose because they commonly reduce the glass-transition temperature of the polymer in which they are dispersed.2 Homopolymer additives having a different molecular structure than the matrix polymer are seldom compatible with the matrix, and whereas polymer-polymer incompatibility can enhance surface segregation,3 the incompatibility also leads to the presence of sharp interfaces between components with poor mechanical strength. Block copolymers in blends with homopolymers are the obvious polymeric analogues of small-molecule surfactants in solution and at equilibrium should adsorb extremely efficiently * Corresponding author. E-mail:
[email protected]. † Durham University. ‡ Present address: Materials Science Centre, The University of Manchester, Grosvenor Street, Manchester M1 7HS, U.K. § CCLRC Rutherford Appleton Laboratory. (1) Mark, H. F., Bikales, N. M., Overberger, C. G., Menges, G., Eds. Encyclopedia of Polymer Science and Engineering, 2nd ed.; John Wiley and Sons: New York, 1985; Vol. 3. (2) Thomas, N. L.; Windle, A. H. Polymer 1982, 23, 529-542. (3) Jones, R. A. L.; Kramer, E. J.; Rafailovich, M. H.; Sokolov, J.; Schwarz, S. A. Phys. ReV. Lett. 1989, 62, 280-283.
to surfaces and interfaces.4 However, experimental studies of block copolymer adsorption in blends have commonly shown that these excellent equilibrium properties cannot be attained on accessible experimental timescales because of the formation of micelles that are either slow to diffuse to interfaces or unable to dissociate.5-7 End-functionalized polymers (EFPs) have been studied at surfaces and interfaces by numerous authors over the last two decades.7-21 Because of their inherent compatibility with (4) Lyatskaya, Y.; Gersappe, D.; Gross, N. A.; Balazs, A. C. J. Phys. Chem. 1996, 100, 1449-1458. (5) Bucknall, D. G.; Higgins, J. S.; Penfold, J. Physica B 1992, 180, 468-470. (6) Hu, W. C.; Koberstein, J. T.; Lingelser, J. P.; Gallot, Y. Macromolecules 1995, 28, 5209-5214. (7) Kiff, F. T.; Richards, R. W.; Thompson, R. L. Langmuir 2004, 20, 44654470. (8) Narrainen, A. P.; Hutchings, L. R.; Ansari, I. A.; Clarke, N.; Thompson, R. L. Soft Matter 2006, 2, 126-128. (9) van de Grampel, R. D.; Ming, W.; Gildenpfennig, A.; van Gennip, W. J. H.; Laven, J.; Niemantsverdriet, J. W.; Brongersma, H. H.; de With, G.; van der Linde, R. Langmuir 2004, 20, 6344-6351. (10) van de Grampel, R. D.; Ming, W.; Gildenpfennig, A.; Laven, J.; Brongersma, H. H.; De With, G.; Van der Linde, R. Langmuir 2004, 20, 145149. (11) Koberstein, J. T. J. Polym. Sci., Part B: Polym. Phys. 2004, 42, 29422956. (12) O’Rourke-Muisener, P. A. V.; Jalbert, C. A.; Yuan, C. G.; Baetzold, J.; Mason, R.; Wong, D.; Kim, Y. J.; Koberstein, J. T. Macromolecules 2003, 36, 2956-2966. (13) O’Rourke-Muisener, P. A. V.; Koberstein, J. T.; Kumar, S. Macromolecules 2003, 36, 771-781. (14) Hutchings, L. R.; Richards, R. W.; Thompson, R. L.; Bucknall, D. G.; Clough, A. S. Eur. Phys. J. E 2001, 5, 451-464. (15) Hopkinson, I.; Kiff, F. T.; Richards, R. W.; Bucknall, D. G.; Clough, A. S. Polymer 1997, 38, 87-98. (16) Schaub, T. F.; Kellogg, G. J.; Mayes, A. M.; Kulasekere, R.; Ankner, J. F.; Kaiser, H. Macromolecules 1996, 29, 3982-3990. (17) Clarke, C. J.; Jones, R. A. L.; Clough, A. S. Polymer 1996, 37, 38133817. (18) Clarke, C. J. Polymer 1996, 37, 4747-4752.
10.1021/la0635810 CCC: $37.00 © 2007 American Chemical Society Published on Web 03/20/2007
4406 Langmuir, Vol. 23, No. 8, 2007
Ansari et al.
Table 1. Polymer Molecular Weight Data sample
Mn/kg mol-1
Mw/Mn
F2dPS10 F2dPS15 F3dPS5 F3dPS11 F4dPS10 hPS
10.5 15.1 5.5 11.2 10.2 232.8
1.23 1.26 1.28 1.33 1.21 1.05
unfunctionalized polymers of the same monomer repeat unit, EFPs are appealing as possible additives to modify the properties of polymer surfaces and interfaces. EFPs are also of academic interest for studies on polymer brushes because they can be synthesized with excellent control over molecular weight and functionality. However, the adsorption of singly end-functionalized polymers is restricted by the “sticking energy” of the functional group to the surface or interface of interest, and for polymers with just a single functional group, the grafting densities that are attained can be too small to have a significant influence on surface properties such as the contact angle.22 To some extent, this problem may be overcome by increasing the size of the functional group, but in many cases, this approach is limited by the solubility of the functionality required for the synthesis. Recently, the possibility of attaching multiple fluorocarbon groups to a single polymer chain has shown great promise as a means to achieve highly functionalized surfaces.8,23,24 Poly(arylether) dendron end-functionalized materials exhibited a modest level of surface activity with eight CF3 units on a generation 2 (G2) dendron.8 However, dendrimer chemistry is notoriously timeconsuming; therefore, for practical applications, it is desirable to minimize the generation number. In this report, we focus on smaller G0 and G1 dendrons onto which two, three, or four C8F17 fluorocarbon units were attached. These dendrons were used to prepare a series of EFPs. The selforganization behavior of these materials, when added to highmolecular-weight unfunctionalized polystyrene, is described in three sections. First, we explore the ability of these additives to interdiffuse with unfunctionalized polystyrene, and the influence of EFP structure on this behavior. In the second section, the aggregation behavior is explored using small-angle neutron scattering and TEM techniques. These measurements provide a basis for understanding the interdiffusion experiments as well as indicating the likely limits beyond which adding more additives should have a negligible effect on the surface properties. In the final section, the surface modification is discussed in relation to the adsorbed surface excess of EFP. Experimental Section Materials. Fluorocarbon-functionalized poly(arylether) dendroninitiated deuteriopolystyrenes (FxdPSy) were synthesized, where x denotes the number of C8F17 fluorocarbon units per polymer chain and y is the molecular weight of the dPS chain in kg/mol. Unfunctionalized linear hydrogenous polystyrene (hPS) was synthesized using living anionic polymerization, and molecular weight data for all of these materials are summarized in Table 1. The synthesis of multiply end-functionalized polymers is possible for a variety of (19) Elman, J. F.; Johs, B. D.; Long, T. E.; Koberstein, J. T. Macromolecules 1994, 27, 5341-5349. (20) Jones, R. A. L.; Norton, L. J.; Shull, K. R.; Kramer, E. J.; Felcher, G. P.; Karim, A.; Fetters, L. J. Macromolecules 1992, 25, 2359-2368. (21) Shull, K. R. J. Chem. Phys. 1991, 94, 5723-5737. (22) Hutchings, L. R.; Richards, R. W.; Thompson, R. L.; Bucknall, D. G. Eur. Phys. J. E 2002, 8, 121-128. (23) Li, H.; Zhang, Y. M.; Zhang, H.; Xue, M. Z.; Liu, Y. G. J. Polym. Sci., Part A: Polym. Chem. 2006, 44, 3853-3858. (24) Narrianen, A. P.; Hutchings, L. R.; Ansari, I.; Thompson, R. L.; Clarke, N. Macromolecules 2007, 40, 1969-1980.
Figure 1. Structure of fluorocarbon-functionalized deuteriopolystyrenes. polymers and is described in greater detail elsewhere.8,24,25 Structures of the three different fluorocarbon-functionalized polymers are shown in Figure 1. Thin Film Preparation. Thin films were prepared comprising hPS and one of the FxdPSy polymers. For the interdiffusion experiments, bilayer samples were prepared in which the two polymer components were initially present as discrete layers. A thin film (70 nm) of FxdPSy was spin-coated onto a silicon substrate from toluene solution. A thicker film of hPS (1200 nm) was spin-coated onto a large microscope slide and then transferred onto the FxdPSy-coated silicon wafer by flotation and pickup from deionized water. To explore the surface adsorption, blended films were prepared in which FxdPSy and hPS were co-dissolved in toluene in the desired ratio prior to spin-coating onto a silicon substrate. Silicon wafer substrates (0.28 mm thick) were used for nuclear reaction analysis experiments, and these were broken into 2 cm2 sections prior to annealing. Samples for neutron reflectometry were prepared on silicon blocks of 50 mm diameter and 5 mm thickness that had successively been cleaned with toluene and then permanganic acid. The absence of surface contamination was verified by ellipsometry to confirm that the native oxide layer on the silicon substrates did not exceed 3.0 nm. Annealing was carried out at temperatures of up to 160 °C under vacuum to minimize the risk of oxidation. Small-Angle Neutron Scattering (SANS). Polymer blend samples (12 mm diameter, 1 mm thickness) were prepared for SANS experiments as follows: the polymers were co-dissolved in toluene in the required ratio prior to precipitation into cold stirred methanol. Samples were placed in a 12-mm-diameter mold and dried overnight under ambient conditions before being transferred to a vacuum oven. Any remaining solvent was then carefully removed over several days with the oven temperature being gradually increased to 160 °C. Particular care was required with this procedure because the blended polymers were prone to bubble formation as the glass-transition temperature of the blend was approached. SANS experiments were (25) Pillay-Narrainen, A.; Hutchings, L. R.; Feast, W. J.; Thompson, R. L.; Ansari, I. A.; Clarke, N. Macromol. Symp. 2006, 231, 103-109.
Multifunctionalized Polystyrenes
Langmuir, Vol. 23, No. 8, 2007 4407
carried out at the ISIS pulsed neutron facility, Rutherford Appleton Laboratories, U.K. The experimental equipment and technique are described in greater detail elsewhere.26 Transmission Electron Microscopy (TEM). Thin sections (thickness 70-90 nm) for TEM were cut from the blend samples examined in SANS experiments using a Leica EM UC6 ultramicrotome fitted with a cryoattachment. Sections were picked up onto carbon film support copper grids. TEM experiments were performed on samples with a Hitachi H-7600 transmission electron microscope. The relatively high electron density of fluorocarbons in comparison to that of styrene units of the polymer chains is sufficient to obtain TEM images without staining. Nuclear Reaction Analysis (NRA). NRA experiments were carried out using the Durham ion beam accelerator facility. A 2-mmdiameter 0.7 MeV 3He+ beam was delivered to the sample surface. The nuclear reaction He + 2H f 4He + 1H + energy (18.352 MeV)
3
(1)
was used to determine the depth distribution of the deuterium-labeled polymer27,28 from the energy of the protons ejected at 170° (backscattered) with respect to the incident beam. The near-surface region of the film was examined using the beam incident at 83° with respect to the sample normal. With this setup, the surface resolution is approximately 6 nm and is therefore sufficient to quantify any surface excess of dPS due to FxdPSy adsorption. Although the proton spectrum is sensitive only to the deuterated polymer in the blend, 3He+ recoils from the C atoms due to both polymers present in the film are also detected. A comparison of these two signals confirmed that the composition of the polymer films was not altered measurably by the spin-coating process. Interdiffusion in hPS/FxdPSy bilayer samples was determined via the broadening of the polymer-polymer interface. To probe the hPS/FxdPSy interface 1200 nm below the film surface, it was necessary to orient the samples perpendicular to the incident ion beam. The depth resolution of NRA experiments carried out at this interface was approximately 50 nm. Neutron Reflectometry (NR). For a complete description of surface adsorption, it is desirable to obtain a description of the depth distribution within the adsorbed layer. The extent of mixing between adsorbed polymer chains and the matrix polymer may have important ramifications for the durability of modified surfaces. The depth resolution required for this analysis is available only using neutron reflectometry.29 Specular neutron reflectometry was carried out using the SURF reflectometer at the ISIS spallation neutron source, Rutherford Appleton Laboratories, U.K.30 Two angles of incidence were used (θ ) 0.35 and 0.8°) to provide specular reflectivity data from the critical edge, R(Q) ) 1, to the point where the signal could not be readily distinguished from the background, R(Q) ≈ 10-6. The neutron beam was collimated to maintain a resolution in δQ/Q of 5%, where Q is the scattering vector defined by Q ) (4π/λ)sin θ and λ is the neutron wavelength. Contact Angle Analysis. Advancing contact angles were measured manually using water and dodecane as polar and nonpolar probe liquids, respectively. In each experiment, six values were measured, and the average value was found with a typical uncertainty of (2°. Hysteresis in contact angles was found to be small, as would be expected for smooth spin-cast films, with advancing contact angles exceeding receding contact angles by approximately 4°. (26) King, S. M. In Modern Techniques for Polymer Characterisation; Pethrick, R. A., Dawkins, P., Eds.; Wiley: Chichester, England, 1999. (27) Payne, R. S.; Clough, A. S.; Murphy, P.; Mills, P. J. Nucl. Instrum. Methods Phys. Res., Sect. B 1989, 42, 130-134. (28) Chaturvedi, U. K.; Steiner, U.; Zak, O.; Krausch, G.; Schatz, G.; Klein, J. Appl. Phys. Lett. 1990, 56, 1228-1230. (29) Russell, T. P. Mater. Sci. Rep. 1990, 5, 171-271. (30) Penfold, J.; Richardson, R. M.; Zarbakhsh, A.; Webster, J. R. P.; Bucknall, D. G.; Rennie, A. R.; Jones, R. A. L.; Cosgrove, T.; Thomas, R. K.; Higgins, J. S.; Fletcher, P. D. I.; Dickinson, E.; Roser, S. J.; McLure, I. A.; Hillman, A. R.; Richards, R. W.; Staples, E. J.; Burgess, A. N.; Simister, E. A.; White, J. W. J. Chem. Soc., Faraday Trans. 1997, 93, 3899-3917.
Figure 2. Concentration versus depth profiles obtained for bilayers of hPS on F2dPS10 (circles) and F4dPS10 (squares). Results after annealing for 2 h at 140 °C (gray symbols) and 150 °C (open symbols) are compared with data for unannealed samples (black symbols). The solid curves indicate fits to the experimental data obtained using a bilayer model with a diffuse interface and surface excess.
Results Interdiffusion. Interdiffusion between FxdPSy and hPS was determined from the evolution of the deuterium concentration profiles by nuclear reaction analysis of annealed bilayer films. Volume fraction profiles for FxdPS10 are shown in Figure 2 as a function of annealing temperature. Because the rates of interdiffusion of these materials were completely unknown, a very broad range of effective annealing times was accessed by annealing for 2 h at different temperatures, followed by timetemperature superposition31 to convert results to an equivalent annealing time at 120 °C. Using the shift factor calculated by Tassin et al.,32 annealing for 2 h at 130, 140, and 150 °C corresponds to annealing at 120 °C for 27 h, 8.7 days, and 44 days respectively. For both of the unannealed specimens, the dPS due to FxdPS10 was initially buried beneath an hPS layer, which is approximately 1200 nm thick, such that the concentration of deuterium apparent at depths of less than 1000 nm is negligible. Although the FxdPS10 layer initially contains no hPS, the apparent volume fraction of dPS is somewhat less than unity because the layer thickness (70 nm) is comparable to the instrumental resolution at this depth. Data were analyzed using a model in which the volume fraction of dPS, φ(z), as a function of depth, z, in the region of the buried interface was given by
φ(z) )
1 h+z-l h-z+l erf + erf 2 w w
[ (
)
(
)]
(2)
where l is the total thickness of the film and h is the thickness of the FxdPSy layer. The width of the interface w is related to the interdiffusion coefficient, D, and annealing time, t, by w ) x4Dt. A thin excess layer of dPS at z ) 0 (the film surface) was incorporated into the model to allow for FxdPSy adsorption in annealed samples. The development of this concentration profile as a function of annealing time is shown schematically in Figure 2. The model concentration profile was convolved with the instrumental resolution and fitted to the experimental data to quantify the polymer interdiffusion, yielding a diffusion (31) Williams, M. L.; Landel, R. F.; Ferry, J. D. J. Am. Chem. Soc. 1955, 77, 3701-3707. (32) Tassin, J. F.; Monnerie, L.; Fetters, L. J. Macromolecules 1988, 21, 24042412.
4408 Langmuir, Vol. 23, No. 8, 2007
Figure 3. Absolute differential scattering cross-section, dΣ/dΩ, versus scattering vector, Q, for blends of 8% (downward-facing triangles), 12% (upward-facing triangles), 15% (right-facing triangles), and 20% (circles) F2dPS10 in hPS. The dotted curves indicate RPA simulations of scattering due to unperturbed chains in the absence of any interactions. The solid curves show RPA simulations for “star” polymers in which each arm was defined by the average dimensions a single F2dPS10 macromolecule and the number of arms was varied to obtain the best possible fit to the data.
coefficient of 0.15 nm2 s-1 for F2dPS10 at 120 °C. There was no systematic broadening of the F4dPS10/hPS interface with annealing; therefore, it was not possible to obtain a diffusion coefficient from this measurement. Aggregation. Small-angle neutron scattering data are shown as the absolute differential scattering cross-section, dΣ/dΩ, versus Q in Figure 3 for various blends of F2dPS10 with hPS. For compatible polymer blends with a small interaction parameter, the random phase approximation (RPA) provides a nearquantitative description of SANS data.33 The scattered intensity, dΣ/dΩ, is a function of the scattering-length density of each component, the structure factor, which includes the molecular weight and connectivity of each component, and the FloryHuggins interaction parameter, χ, which describes interactions between monomers of different species. This technique has been generalized to include block copolymers of arbitrary architecture34-36 and multiple-component systems.37 The main limitation of the RPA occurs when strong interactions between polymer chains cause demixing, as is the case for microphase separation or phase separation. Good agreement between experimental data and RPA simulations with χ ) 0 was found for F2dPS10 concentrations of up to 12 wt %. However, at the two higher concentrations, large deviations between experiment and simulation were seen, particularly at low scattering vectors, where the large scattering cross-section in the experimental data indicates comparatively large structures in the blend. This result indicates a significant deviation from the random organization of these materials in blends with a high concentration of F2dPS10, which would not be seen for unfunctionalized dPS blends with hPS (χ ≈ 0). Although the RPA approach is not valid for strongly segregated systems in which polymer chains are not randomly mixed, we can use this method to represent the soluble micelles as “star diblock copolymers”35 with a fluorinated core and the (33) deGennes, P. G. Scaling Concepts in Polymer Physics; Cornell University Press: Ithaca, New York, 1979. (34) Read, D. J. Macromolecules 1998, 31, 899-911. (35) Kiff, F. T.; Richards, R. W.; Thompson, H. L.; Bucknall, D. G.; Webster, J. R. P. J. Phys. II 1997, 7, 1871-1891. (36) Benoit, H.; Hadziioannou, G. Macromolecules 1988, 21, 1449-1464. (37) Balsara, N. P.; Jonnalagadda, S. V.; Lin, C. C.; Han, C. C.; Krishnamoorti, R. J. Chem. Phys. 1993, 99, 10011-10020.
Ansari et al.
Figure 4. Absolute differential scattering cross-section data for blends of 4, 8, 12, 15, and 20% (w/w) F4dPS10 in hPS. The dashed line indicates dΣ/dΩ ≈ Q-4.
Figure 5. Transmission electron micrograph for 8% (w/w) F4dPS10 in hPS.
number of arms defined by the aggregation number of the micelle. With this approach, good fits to the experimental data at 15 and 20 wt % F2dPS10 in hPS were obtained with aggregation numbers of approximately 6 and 7, respectively. These unaggregated and micellar states are sketched in Figure 3. Whereas the scattering due to F2dPS10 is consistent with the presence of either free polymer chains or micellar aggregates, dΣ/dΩ for the corresponding F4dPS10/hPS blends is very different. The very large dΣ/dΩ values as Q tends to 0 in Figure 4 together with the Q-4 dependence of dΣ/dΩ both indicate that this is not a single-phase blend. The small Bragg peak at Q ≈ 0.050-0.055 Å-1 and the weak secondary peak at double this value suggest localized ordering on a length scale of approximately 12 nm that is lamellar in character. Transmission electron microscopy was used to examine the self-organised structure of the 8% F4dPS10/hPS blend. The micrograph shown in Figure 5 confirms the findings of the SANS experiments. There appear to be a number of discrete objects present in the F4dPS10 blends that are on the order of 100-200 nm in size and resemble multilamellar vesicles. The spacing between layers of these
Multifunctionalized Polystyrenes
Langmuir, Vol. 23, No. 8, 2007 4409
Figure 7. Surface excess versus concentration for F2dPS10 (open symbols) and F4dPS10 (solid symbols) as a function of annealing time at 160 °C. Data for bulk concentrations of 7.5, 10, and 15% are shown as triangles, circles, and squares, respectively. The surface excess of 10% F2dPS10 obtained by neutron reflectometry after annealing for 720 min is shown for comparison.
Figure 6. Nuclear reaction analysis data and fits for (a) 15% F2dPS10 in hPS and (b) 15% F4dPS10 in hPS. Data obtained for specimens prior to annealing are indicated by open symbols, and data obtained following annealing for 1 h at 160 °C, by solid symbols.
structures is consistent with the 12 nm spacing inferred from the SANS data. These comparatively large aggregated structures appeared to be present only in blends for which the SANS data contained the Bragg peaks. Adsorption. Preliminary evidence for the surface activity of FxdPSy materials was apparent from both the surface excess found in the 150 °C annealed bilayer films in Figure 2 and the tendency of toluene solutions containing these polymers to form foams when agitated. Near-surface concentration profiles of FxdPSy were determined by NRA and are shown in Figure 6. The surface activity of these materials in toluene solutions is further underlined by the large surface excess of FxdPS10 found at the surface of blends even before annealing the films above their glass-transition temperature. Immediately adjacent to the surface excess in the unannealed films, there is a small depletion zone in which the concentration of FxdPS10 is slightly lower than the bulk concentration. A three-layer model comprising a surface excess layer, a depletion layer, and a bulk layer was used to quantify the depth distribution of dPS in the blended films. This model is shown in schematic form in the inset of Figure 6. The surface excess for this three-layer model is given by
z* ) hs(φs - φ2)
(3)
where hs is the thickness of the surface layer and φs and φ2 are the volume fractions of FxdPSy in the surface and adjacent layers of the model, respectively. Derived surface excess values are shown in Figure 7 as a function of annealing time and FxdPS10 concentration. Surface excess concentrations appear to grow appreciably over the first 15 min of annealing at 160 °C. Annealing for much longer periods appears to have little effect on the surface
Figure 8. Neutron reflectometry data and fits (solid lines) obtained using eq 4. The dotted curves are the calculated reflectivties from SCFT model concentration profiles.
excess concentration. For bulk concentrations between 7.5 and 15% (w/w), the final surface excess concentrations appear to be quite sensitive to F2dPS10 concentration but remarkably independent of F4dPS10 concentration. The near-surface distribution of some miscible FxdPSy blends was examined in greater detail using neutron reflectometry. Data were analyzed by fitting the calculated reflectivity for a surface excess layer of dPS conforming to an error function profile of
φ(z) ) φ∞ +
(
)[
)]
φ s - φ∞ h-z 1 + erf 2 w
(
(4)
where z is depth, w is the width of the interface, h is the distance between the film surface and the center of the interface, and φ∞ is the bulk volume fraction of dPS. The adjustable parameters in the fit were φs, h, and w. NR data are shown in Figure 8 in the form of RQ4 versus scattering vector, Q, along with the best fits that could be obtained using eq 3 to describe the concentration profile of the adsorbed layer. The small peak at Q ≈ 0.01 Å-1 is due to the critical edge beyond which R(Q) falls rapidly from 1. The broad peak at higher Q arises from the adsorbed layer of FxdPS10 at the film surface. The film surfaces appeared to have a rms roughness of
4410 Langmuir, Vol. 23, No. 8, 2007
Ansari et al.
there was a critical concentration region beyond which further addition of FxdPSy gives rise to relatively small increases in contact angle. In the case of the F4dPS10/hPS blend, the maximum contact angle appears to be found at bulk concentrations of 4% and above.
Discussion
Figure 9. Concentration profiles of FxdPS10 obtained by fitting eq 4 to neutron reflectometry data. The curves correspond to SCFTcalculated concentration profiles having the same surface excess as the experimentally derived values. The thermodynamic sticking energy, β, is shown adjacent to each curve.
Figure 10. Contact angles for water (solid squares) and dodecane (open squares) as a function of F4dPS10 concentration in blends with hPS. Derived values for the polar, dispersive, and total solid surface energies are shown as open triangles, open circles, and solid triangles, respectively.
approximately 1.5 nm, which is slightly greater than the typical roughness of spin-cast polystyrene films (0.5-1.0 nm). It is likely that this number is an upper limit to the true surface roughness and that it contains a contribution due to fluorocarbon groups, which are present in high concentration on the film surface. Because the scattering-length density of C8F17 (3.86 × 10-6 Å-2) is intermediate between values for air (0.0) and dPS (6.47 × 10-6 Å-2), the presence of a dense fluorocarbon layer on the film surface effectively broadens the interface between air and the dPS-rich adsorbate. The concentration profiles corresponding to the calculated reflectivities are shown in Figure 9, where the depth scale, z, has been normalized with respect to the radius of gyration of the dPS chain, 2.7 nm. Contact Angle Analysis. Results for the contact angles of water and dodecane on several blends of FxdPSy in hPS have been described previously,24 and typical data for these experiments are shown in Figure 10. All of the FxdPSy/hPS blends show qualitatively similar behavior. In the absence of FxdPSy, dodecane spreads on hPS surfaces, giving an effective contact angle of zero, and the contact angle of water is approximately 90°. A rapid increase in the contact angles of both liquids was found with increasing FxdPSy concentration, demonstrating appreciable hydrophobicity and lipophobicity due to the adsorption of fluorocarbon groups onto the blend surface. For each blend,
Interdiffusion. Despite the similarity in molecular weight of the two functionalized materials, it is clear that their interdiffusion behavior with hPS differs dramatically. F2dPS10 appears to interdiffuse readily with hPS, and the hPS/dPS interface broadens continuously with annealing, indicating that these polymers are fully miscible. In contrast, F4dPS10 appears to be only partially miscible with hPS because there is very little change evident in the width of the hPS/dPS interface with annealing. In both bilayer films, it is clear that after annealing for 2 h at 150 °C excess FxdPS10 begins to appear at the film surface (depth ) 0). This observation confirms that F4dPS10 must be at least partially miscible with hPS to have reached the upper surface of the film. Aggregation of FxdPSy in Blends with hPS. The dendritic functionality, although it is a comparatively small component of any FxdPSy molecule that we have examined, has a significant bearing on the phase behavior in binary blends. It appears that the bulk organization has three distinct manifestations. At relatively low concentrations of F2dPS10, the small-angle neutron scattering data are consistent with unaggregated unperturbed chains, and are quantitatively similar to that which would be expected in the absence of any interactions between the fluorocarbon groups and the matrix. When the fraction of fluorocarbon in the blend is increased by increasing the concentration of F2dPS10, the mixture remains in a single phase, but the significant increase in dΣ/dΩ at low scattering vectors confirms that the small-angle scattering is due to objects that are somewhat larger than the unperturbed chain dimensions. This is a clear indication that as the concentration of FxdPSy increases there is a tendency toward aggregation that is driven by the fluorocarbon groups. Whereas this is to our knowledge the first report of end-functionalized polymer aggregation in blends, we note that the analogous behavior in small-molecule surfactants based on hydrocarbon and fluorocarbon moieties in solution is quite well established.38,39 Although it is not possible to describe the aggregated structure explicitly from the SANS data alone, it appears most plausible that the scattering is due to micelles comprising a fluorocarbon core and a dPS corona because this would minimize unfavorable contact between fluorocarbon groups and polystyrene while permitting the polystyrene components to remain fully mixed. The anticipated scattering law for such an aggregate may be calculated using an RPA simulation for a dPS star polymer having a fluorocarbon core and arms of the same degree of polymerization as in the dPS chain in the FxdPSy material. For this model to be valid, it is essential that the proposed structure does not preclude the possibility of mixing between aggregated F2dPS10 chains and the polymer matrix. The radius of gyration of the dPS chains was 26.6 Å, which corresponds to a rms end-to-end distance, Re2e, of 65 Å. A spherical aggregate of this radius comprising seven F2dPS10 chains has a volume of 1.15 × 106 Å3, or 1.64 × 105 Å3 per dPS chain. The calculated space-filling volume of one dPS chain is approximately 1.66 × 104 Å.3 It therefore follows that nearly 90% of the volume available to an aggregate of seven F2dPS10 chains is occupied by polymer (38) Turberg, M. P.; Brady, J. E. J. Am. Chem. Soc. 1988, 110, 7797-7801. (39) Binks, B. P.; Fletcher, P. D. I.; Kotsev, S. N.; Thompson, R. L. Langmuir 1997, 13, 6669-6682.
Multifunctionalized Polystyrenes
Langmuir, Vol. 23, No. 8, 2007 4411
chains that do not belong to the aggregate, satisfying the RPA requirement that polymer chains are not strongly segregated. When the fluorocarbon content of the blend is increased by increasing the proportion of fluorocarbon per FxdPSy macromolecule, quite different behavior is observed. In F4dPS10 blends, dΣ/dΩ diverges at low scattering vectors and is not consistent with the scattering behavior of a single-phase blend. The Q-4 dependence of dΣ/dΩ is consistent with a two-phase blend, with an interfacial area per unit volume, ST, given by the Porod law,
Q4
dΣ ) 2πbv2ST dΩ
(4)
in which bv is the contrast in the neutron-scattering-length density. The dashed line in Figure 4, (Q4 dΣ/dΩ ) 1.9 × 1026 cm-5) suggests that the interfacial area of dispersed material (assuming complete segregation of each component) is approximately 1.2 m2 cm-3 in the 8% (w/w) blend. This area corresponds to uniform spheres of approximately 195 nm radius, which is larger than but on the same order as the size of multilamellar vesicles found by TEM. The discrepancy between the radius inferred from the Porod law analysis and the particles detected by TEM indicates that the contrast in scattering-length density between the two phases is somewhat lower than if they were completely segregated and is therefore consistent with our evidence from the diffusion experiments, which shows that these polymers are partially miscible. Very similar structures to the multilamellar vesicles (“onions”) of F4dPS10 have previously been reported for blends of block copolymers with homopolymers. In common with our structures, the aggregates reported by Koizumi et al. are somewhat elliptical in character; however, the origin of this anisotropy is not clear at present.40,41 When the molecular weight of the homopolymer far exceeds that of the copolymer, macrophase separation generally precedes microphase separation, and it is thought that droplets of phase-separated copolymer form, which later undergo microphase separation. This situation appears to describe the F4dPS10 blends with hPS quite well because the molecular weight of hPS is much greater than that of the end-functionalized polymer. The microphase separation is again driven by the antipathy between the fluorocarbon groups of the functionalized dendron and the polystyrene components of the blend, resulting in the formation of fluorocarbon-rich layers that are separated by lamellae of polystyrene. It is interesting that such a small functional group (∼9% of the F4dPS10 polymer volume is fluorocarbon) appears to promote lamellar self-assembly. Where similar structures have been observed with block copolymers, the incompatible block has been approximately 40% of the volume of the copolymer.40 It is likely that the formation of lamellae by polymers with such asymmetric volume fractions is a feature conferred by the highly branched nature of the dendron, which has a marked effect on the intrinsic curvature of aggregated structures.42 Although the SANS and TEM data provide the clearest evidence of phase separation, they do not readily provide a measure of the extent of miscibility between F4dPSy and hPS. The clearest indication of the true extent of miscibility is available from the concentration profiles obtained for the annealed blends using nuclear reaction analysis. Prior to annealing, the concen(40) Koizumi, S.; Hasegawa, H.; Hashimoto, T. Macromolecules 1994, 27, 6532-6540. (41) Hamley, I. W. The Physics of Block Copolymers; Oxford University Press: Oxford, England, 1998. (42) Cho, B. K.; Jain, A.; Nieberle, J.; Mahajan, S.; Wiesner, U.; Gruner, S. M.; Turk, S.; Rader, H. J. Macromolecules 2004, 37, 4227-4234.
tration profiles of all of the blends were qualitatively identical, having a surface excess of FxdPSy and a depletion zone adjacent to this, which was sparse in FxdPSy. Miscible polymers are characterized by a positive diffusion coefficient, and two-phase blends are characterized by a negative diffusion coefficient. Upon annealing, the depletion zone of miscible blends disappears as polymer from the bulk of the film diffuses into this region. However, in the immiscible blends, the reverse behavior is observed, and the depletion zone grows as phase separation occurs. The volume fraction of dPS in this region of the annealed films corresponds to the binodal concentration of FxdPSy. The binodal concentrations of F4dPS10 and F3dPS11 were 4.5 and 5.4%, respectively, and although rather similar in value, they are consistent with the anticipated result of decreasing miscibility with increasing fluorocarbon content per dPS chain. Surface Organization and Modification. The glass-transition temperature of polystyrene is approximately 100 °C; therefore, the surface excess observed in unannealed films must have formed during the few seconds over which toluene evaporates during the spin-coating process. A small increase in surface concentration was generally found with further annealing, which may be driven by two factors. First, the depletion layer of low FxdPSy concentration indicates that equilibrium adsorption had not been attained. This depletion is most likely to arise from the rapid adsorption of FxdPS10 polymer chains from the near-surface region of the film to the surface during the spin-coating process. Second, the surface energy of polystyrene, 40 mJ m-2, is significantly higher than the surface tension of toluene, ∼28 mJ m-2. It therefore follows that there is a greater free energy of adsorption when fluorocarbons adsorb to the surface of a polystyrene blend rather than to the surface of a toluene solution. The concentration dependence of the surface excess values of annealed blends clearly has its roots in the bulk behavior. The surface excess of F2dPS10, which remains unaggregated up to 12% (w/w) and forms only small aggregates at higher concentrations, appears to increase continuously over this concentration range. In contrast, there appears to be very little concentration dependence for the surface excess of F4dPS10 blends over the same range, for which large aggregates are seen. This behavior is exactly analogous to the classical behavior of surfactant adsorption in the region of the critical micelle concentration. The surface excess continues to grow with increasing concentration up to the point at which aggregation in the bulk occurs. Further increasing the bulk concentration leads to an increase in the number of aggregated particles but has relatively little effect on the surfactant activity. Little increase in surface excess would be expected beyond this point. Polymer adsorption in blends is complicated by the fact that equilibrium adsorption is frequently hindered by kinetic factors. As the surface excess increases, adsorbed polymer chains overlap and must be stretched relative to their unperturbed dimensions. The entropic penalty of stretching adsorbed polymer chains into “brushes” ultimately limits the maximum surface excess that can be obtained by adsorption from the bulk. Neutron reflectometry can provide insight into the extent of stretching of adsorbed polymer chains. The fits to the neutron reflectometry data in Figure 8 show that eq 3 can provide a reasonable description of the actual concentration profiles of end-functionalized polymers in the adsorbed layers. Surface excess values derived from these model concentration profiles were in excellent agreement with those obtained by NRA. Self-consistent field theory calculations were used to predict the concentration profile of an adsorbed layer as a function of the attraction of the functional group to the film surface. The thermodynamic sticking energy
4412 Langmuir, Vol. 23, No. 8, 2007
Ansari et al.
Table 2. Composition and Limiting Surface Properties of a Fluorocarbon End-Functionalized Polymer material
% (w/w) fluorocarbon
θmax (dodecane)
θmax (water)
γmin
F2dPS10 F2dPS15 F3dPS5 F3dPS11 F4dPS10
8.0 5.5 22.8 11.2 16.4
35 25 60 56 63
103 102 112 104 110
21.9 ( 1.1 23.9 ( 1.0 14.8 ( 1.1 16.9 ( 1.1 14.3 ( 1.2
of adsorption, β, is defined as
()
β δ ) χb - χs + 1.1 ln kBT Rg
(5)
where χb - χs is the free energy of transfer of the functional group from the bulk of the film to the surface and the logarithmic term describes the entropic penalty of confining one end of the polymer chain to the first layer of the lattice, in which the lattice spacing is given by δ. Concentration profiles derived from both experiment and selfconsistent field theory show that with increasing surface excess concentration both the z ) 0 surface concentration and the extent of the adsorbed layer from the surface into the film increase. The midpoint of the adsorbed polymer layer in the 1% F4dPS10 and 10% F2dPS10 blends is on the order of 2.5Rg, or 6.5 nm, which is notably consistent with half of the lamellar spacing of the aggregates found in the F4dPS10 blends by SANS and TEM. The relationship between values for the thermodynamic sticking energy, β, is not trivial. One might expect β to be independent of concentration, yet it appears that in the F4dPS10 blends, β increases with increasing concentration from 0.3 to 1.0 wt %. The concentration dependence of β may be a sign of a hindered approach to equilibrium adsorption, which is more noticeable at low concentrations where the flux of F4dPS10 toward the surface is limited by its low bulk concentration. It could also be expected that χb - χs is directly proportional to the number of fluorocarbon groups per polymer chain of a given size. However, doubling the number of fluorocarbon groups causes a relatively small increase in β, possibly as a result of intramolecular aggregation of these groups in the bulk, which would reduce the driving force for F4dPS10 adsorption. It is also possible that steric restrictions to the organization of bulky groups at the blend surface contribute to the relatively low value of β. The problem of fitting fluorocarbon groups at the film surface would be expected to be most severe for tetrafunctional F4dPS10, and this may explain why the SCFT simulation (which neglects this factor) tends to overestimate the surface concentration of dPS for this material. At any weight fraction of additive, the fluorocarbon content may be increased by increasing x, the number of fluorocarbon groups per additive molecule, or by decreasing y, the molecular weight of the additive molecule. Both of these adjustments were found to lead to the anticipated increases in hydrophobicity and lipophobicity as determined from the contact angles of water and dodecane on the blend surface, respectively. The equilibrium plateau values for contact angles with water and dodecane are given in Table 2. The critical concentration of F4dPS10 in hPS, beyond which no further increase in contact angle is observed (∼4% (w/w)), is consistent with the binodal concentration of F4dPS10 derived from the NRA data. Using contact angle analysis, it is possible to assess the change in blend surface energy due to adsorption of the functionalized material. By using at least two contact fluids of widely differing polarity, the polar and dispersive contributions to the surface energy or any surface may be estimated. The contact angle, θ,
is a function of the liquid surface energy, γl, the dispersive components of the solid and liquid surface energies, γds and γdl , and the polar components of the solid and liquid surface energies, γ ps and γ pl :
cos θ )
2xγds γdl 2xγps γpl + -1 γl γl
(6)
In our analysis, we used deionized water as the polar contact fluid (γl ) 72.8 mJ m-2, γdl ) 21.8 mJ m-2, γpl ) 51.0 mJ m-2) and dodecane, which is almost completely nonpolar (γl ) 25.4 mJ m-2, γdl ) 25.4 mJ m-2, γpl ) 0). It appears from the tabulated results that several blends have a surface energy lower than that of PTFE, which is approximately 20 mJ m-2.43 Because the NR results showed that our blend surfaces are quite smooth and we observed only small differences between advancing and receding contact angles, the very low surface-energy values obtained arise from the chemical composition and not the surface topography. Even in glassy polymers, some local rearrangement of functional groups is possible, generally leading to a reduction in contact angles after immersion in water for many hours. Although there was some evidence of this behavior in our earlier work with smaller, weakly adsorbing CF3 functionalities, this did not appear to be significant over the course of measurements with the larger C8F17-functionalized dendrons, possibly because their size inhibits local reorganization. Other polymer materials have been reported with surface-energy values that are well below that of PTFE, and it appears that the surface energy of these FxdPSy blends approaches values recently reported by DeSimone et al. for perfluoropolyethers.44
Conclusions Polystyrenes, functionalized with multiple fluorocarbon groups, are highly surface-active in blends with homopolymer polystyrene (hPS) and are sufficiently mobile to diffuse to surfaces, resulting in surface properties similar to those of PTFE while retaining bulk properties similar to those of polystyrene. We have rationalized the surface properties of these materials in terms of their bulk behavior and have found clear parallels between the aggregation and adsorption of these functionalized polymers in blends and the corresponding behavior of surfactants in solution. The surface activity, measured in terms of the surface excess concentration and surface-energy reduction, generally increases with increasing fluorocarbon content per functionalized polymer chain. This behavior is clearly analogous to the effect of increasing the size of the hydrophobic group in aqueous surfactant solutions, as is the tendency of these end-functionalized polymers to form aggregates at higher concentrations. Over the molecular weight range considered, polystyrene functionalized with two C8F17 fluorocarbon groups was fully miscible with hPS but appeared to form soluble micelles at concentrations above 12% (w/w). Polystyrenes functionalized with more than two C8F17 groups were only partially miscible with hPS and formed multilamellar vesicles. The binodal composition of the immiscible blends was determined from the near-surface concentration profiles using nuclear reaction analysis and was found to decrease with increasing fluorocarbon content per functionalized polymer chain. The plateau in the surface properties was found to coincide with the binodal concentration. Neutron reflectometry was used to examine the concentration profile of the adsorbed functionalized (43) Chen, J. R.; Wakida, T. J. Appl. Polym. Sci. 1997, 63, 1733-1739. (44) Yarbrough, J. C.; Rolland, J. P.; DeSimone, J. M.; Callow, M. E.; Finlay, J. A.; Callow, J. A. Macromolecules 2006, 39, 2521-2528.
Multifunctionalized Polystyrenes
polymer. Surface excess values derived from neutron reflectometry were in excellent agreement with our NRA results. The extent of the adsorbed layer was consistent with the lamellar separation found in phase-separated blends, and the composition profile was in reasonable agreement with those predicted using self-consistent mean-field theory. An analysis of the thermodynamic sticking energy, β, per adsorbed functional group revealed some dependence on the blend composition, which suggests that equilibration may be slow, particularly at low concentrations. The value of β increased with increasing fluorocarbon content but not in a linear manner, suggesting that steric factors may limit the efficiency of adsorption for bulky fluorinated groups.
Langmuir, Vol. 23, No. 8, 2007 4413
Acknowledgment. We are very grateful to Professor Jim Feast, Department of Chemistry, Durham University, for many useful discussions, Mrs. Christine Richardson, Department of Biology, Durham University, for her assistance with the TEM experiments, and Professor Ken Shull, Northwestern University, for providing the SCFT analysis software. We thank OneNorthEast, the DTI, and ERDF for funding this research through the UIC in Nanotechnology and EPSRC for providing the neutronscattering and reflectometry facilities at the Rutherford Appleton Laboratories, U.K. LA0635810
