9260
Langmuir 2006, 22, 9260-9263
Self-Assembly of Tin Oxide Nanoparticles: Localized Percolating Network Formation in Polymer Matrix Atsumi Wakabayashi,*,† Yuki Sasakawa,† Toshiaki Dobashi,‡ and Takao Yamamoto§ AdVanced Materials DiVision, Sumitomo Osaka Cement Co., Ltd., Funabashi, Chiba 274-8601, Japan, and Department of Biological and Chemical Engineering, and Department of Physics, Faculty of Engineering, Gunma UniVersity, Kiryu, Gunma 376-8515, Japan ReceiVed April 25, 2006. In Final Form: July 13, 2006 A novel method to prepare transparent antistatic films by trapping nanoparticles in thin surface layers is proposed, and high performance of the product is demonstrated. Coating solutions consisting of surface-active antimony-doped tin oxide (ATO) nanoparticles, an organic solvent mixture of high affinity with poly(methyl methacrylate) substrate, and an ultraviolet curable resin are used for the film formation. Antistatic property of the layer is obtained at ATO concentrations above the critical concentration for percolation. A scaling analysis of the data shows that the critical concentration (0.004 volume fraction) is extremely low as compared to the value predicted by the percolation theory for randomly packed systems as well as the values in ever developed composite films filled with ATO particles. Microscopic observation of the deposited layers indicates that percolating clusters of ATO particles are localized on the layer surfaces. The excellent electrical and optical properties of the layer are attributed to the characteristic microstructure.
Introduction Optically transparent antistatic composite films consisting of conductive transparent metal oxide particles and organic compound matrixes are industrially very important.1-3 The antistatic properties of these films are obtained at conductive particle concentrations above the percolation threshold, the critical concentration for forming an infinite percolating network of the particles. Because the optical transparency of the films is higher as the particle concentration is lower, a reduction of the threshold value has been a target of recent investigations. As long as we restrict ourselves to randomly packed isotropic composites, however, we cannot exceed a certain theoretical limit for threedimensional structures.4,5 Because antistatic property of the films can be achieved even in very thin surface layers of nanometerscale thickness, it is challenging to design films in which conductive particles are confined to two-dimensional spaces. Trapping of colloidal particles at air-liquid and liquid-liquid interfaces can be a method to realize the anisotropic microstructures. The trapping of colloidal particles at the interfaces is one of the recent studies in colloid and surface science.6 Pioneering work in this area was done by Pieranski.7 He noted that polystyrene * Corresponding author. E-mail:
[email protected]. † Sumitomo Osaka Cement Co., Ltd. ‡ Department of Biological and Chemical Engineering, Gunma University. § Department of Physics, Gunma University. (1) Texter, J.; Lelental, M. Langmuir 1999, 15, 654. (2) Wang, Y.; Anderson, C. Macromolecules 1999, 32, 6172. (3) Sun, J.; Velamakanni, B. V.; Gerberich, W. W.; Francis, L. F. J. Colloid Interface Sci. 2004, 280, 387. (4) McLachlan, D. S.; Blaszkiewicz, M.; Newnham, R. E. J. Am. Ceram. Soc. 1990, 73, 2187. (5) Stauffer, D.; Aharony, A. Introduction to Percolation Theory; Taylor & Francis: London, 1992. (6) (a) Tarimala, S.; Ranabothu, S. R.; Vernetti, J. P.; Dai, L. L. Langmuir 2004, 20, 5171. (b) Tarimala, S.; Dai, L. L. Langmuir 2004, 20, 3492. (c) Nicolaides, M. G.; Bausch, A. R.; Hsu, M. F.; Dinsmore, A. D.; Brenner, M. P.; Gay, C.; Weitz, D. A. Nature 2002, 420, 299. (d) Aveyard, R.; Clint, J. H.; Nees, D.; Paunov, V. N. Langmuir 2000, 16, 1969. (e) Aveyard, R.; Clint, J. H.; Nees, D.; Quirke, N. Langmuir 2000, 16, 8820. (f) Binks, B. P.; Lumsdon, S. O. Langmuir 2001, 17, 4540. (g) Onoda, G. Y. Phys. ReV. Lett. 1985, 55, 226. (h) Stankiewicz, J.; Vı´lchez, M. A. C.; Alvarez, R. H. Phys. ReV. E 1993, 47, 2663.
particles can be trapped at an air-water interface by surface tension. A contribution of electrostatic force to the trapping phenomena was also pointed out.8 Hurd and Schaefer9 have observed the fractal aggregate formation of silica particles trapped at an air-water interface. They described the aggregate formation on the basis of diffusion-limited cluster aggregation (DLCA) models.10 According to the DLCA models, Brownian particles irreversibly join together upon contact to form aggregates. The aggregates and single particles continue to diffuse and form large clusters (agglomerates) by binding together. Paunov et al.11 have developed a thermodynamic approach to adsorption of charged colloidal particles at air-water and oil-water interfaces. An analytical expression for the distribution coefficient of particles between the interfaces has been derived and discussed in terms of the effects of electrolyte, contact angle, surface charge, and oil-water interface tension. Here, we report on a novel antistatic composite film with excellent conductive and transparent properties by self-assembly (and self-organization) of colloidal particles at an air/liquid interface. Our strategy to form the composite film is to utilize a convective flow12 of a surfaceactive nanoparticle suspension and trapping of colloidal particles at an air/liquid interface in the process of the film formation. Experimental Section Nanoparticulate antimony-doped tin oxide (ATO, SUMICE FINE ATO OF103, Sumitomo Osaka Cement) was used in this study. The particles were prepared by a wet-chemical synthesis (doped with antimony at 10.3 mol %), treated with cationic surfactant, and dispersed in organic solvents (methyl ethyl ketone (MEK), diacetone alcohol (DAA), and water (75:15:10 weight ratio)) at 20 wt %. The nanoparticle suspension was mixed with UV curable acrylic resin (PCD-20, Dainichi Seika) at a total solid weight percentage of 30 (7) Pieranski, P. Phys. ReV. Lett. 1980, 45, 569. (8) Earnshaw, J. C. J. Phys. D: Appl. Phys. 1986, 19, 1863. (9) Hurd, A. J.; Schaefer, D. W. Phys. ReV. Lett. 1985, 54, 1043. (10) (a) Witten, T. A.; Sander, L. M. Phys. ReV. Lett. 1981, 47, 1400. (b) Meakin, P. Phys. ReV. Lett. 1983, 51, 1119. (11) Paunov, V. N.; Binks, B. P.; Ashby, N. P. Langmuir 2002, 18, 6946. (12) Chandrasekhar, S. Hydrodynamic and Hydromagnetic Stability; Oxford University Press: London, 1961.
10.1021/la061116x CCC: $33.50 © 2006 American Chemical Society Published on Web 09/21/2006
Self-Assembly of Tin Oxide Nanoparticles
Figure 1. Phase diagram of the ternary system DAA-MEK-water. The symbols O and × denote homogeneous and inhomogeneous states, respectively, and the boundary is the cloud point curve. to prepare coating solutions. Antistatic layers were deposited by applying the coating solutions onto a poly(methyl methacrylate) (PMMA) sheet using a number 14 Meyer wire-wound rod (R. D. Specialties) to yield a layer thickness of 32 µm (manufacturer’s value), followed by drying (60 °C for 10 min) and curing with ultraviolet (UV) irradiation (using a high-pressure mercury lamp, 700 mJ/cm2 of UV dose). The layer depositions were carried out at controlled temperature and humidity conditions. Here, DAA was used for raising the solubility of MEK with water and for increasing the affinity between the resin and PMMA sheet. The resin is dissolved in MEK but not in water. The total systems were designed to keep a homogeneous state all through the drying process of the coating solutions by trial and error. The phase diagram of the ternary system MEK-DAA-water was determined by naked eye observation of the cloud point for mixtures with different compositions. The distribution of ATO particles in cross-sectional layer samples was observed using a HITACHI H-800 transmission electron microscope (TEM) operated at an acceleration voltage of 200 kV. The cross-sectional samples were prepared using a SEIKO SMI2050 focused ion beam (FIB) system operated at an acceleration voltage of 30 kV. In the system, a high-energy ion source, liquid gallium, was accelerated down a vacuum column and scanned across the sample surfaces. Distribution of ATO particles on the layer surfaces was observed using a HITACHI S-4000 field emission scanning electron microscope (FE-SEM) operated at an acceleration voltage of 15 kV. Surface resistivity of the layers prepared was determined according to ASTM D257 protocol applying a 500 V potential using a resistivity meter (Mitsubishi Chemical, HIRESTA-IP with concentric ring URS probe). The surface resisitivity value Fs is related to bulk resisitivity F by the equation F ) Fs × film thickness. Haze and total light transmittance measurements were carried out according to ASTM D1003 protocol using a haze meter (Tokyo Densyoku, TC-HIII DP). Haze value H is related to total transmittance Tt and diffuse transmittance Td as H ) (Td/Tt) × 100.
Results and Discussion Among the components of the coating solutions, water, DAA, and MEK are volatile, and MEK is assumed to have the fastest evaporation rate. The dominant components of the coating solutions near the final state in the drying process should be DAA and water. The time course of composition of the coating solutions in the drying process could be drawn in the ternary phase diagram (Figure 1), if nanoparticles and resin do not affect the phase diagram of the ternary system water-DAA-MEK significantly. The diagram indicates that the coating solutions can be kept homogeneous all through the drying process, which is consistent with our observation by the naked-eye; the deposited layers were homogeneous during the drying process. Figure 2shows representative TEM images of the crosssectional layer samples. In all of the images, localized clusters of ATO particles formed at the layer surfaces are observed. At
Langmuir, Vol. 22, No. 22, 2006 9261
Figure 2. TEM images of thin cross sections of deposited layers: ATO volume fraction φ ) 0.0042 (a) and φ ) 0.0083 (b, c). Samples for the TEM observation were prepared using a FIB system.
Figure 3. FE-SEM images of the surface of deposited layers prepared with ATO volume fraction φ ) 0.0083 observed at low (a), medium (b), and high (c) magnifications.
the lower ATO volume fraction of φ ) 0.0042 (near the percolation threshold as shown below), the thickness of this localized cluster is about 10 nm (Figure 2a). At the higher ATO volume fraction of φ ) 0.0083, the thickness of the cluster exceeds 20 nm (Figure 2b). On the other hand, small aggregates, which do not participate in this localized cluster formation, are also observed in inner layers (Figure 2c). The fraction of ATO localized on the surface layer, which is 25 nm thick in total ATO, in the film is around 12 vol %. Typical FE-SEM images for the layer with ATO volume fraction of φ ) 0.0083 are shown in Figure 3 at different magnifications. Clusters of similar irregularly shaped morphology formed by densely packed ATO particles are observed at every magnification (Figure 3a-c).
9262 Langmuir, Vol. 22, No. 22, 2006
Wakabayashi et al.
Figure 4. Surface resistivity as a function of ATO volume fraction for deposited layers. Controlled relative humidity during layer deposition: 54% RH (b), 40% RH (O). Inset: log-log plot of surface resistivity Fs versus ATO volume fraction φ. Relative humidity during layer deposition: 54% RH.
These microscopic observations clearly suggest that the localized clusters of ATO particles form parts of percolating networks. Because the networks consist of branched long chains of nanoparticles, the localized conductive pathway confined in a thin surface layer could allow an effective dissipation of electrostatic-charge without causing a visible light scattering. The most intriguing result we would like to emphasize is that localized clusters of ATO particles formed on (or near) the air/ liquid (air/solid after drying and UV curing) interfaces. A large amount of suspended ATO particles (the primary and aggregated particles) seems to migrate in deposited layers (initially wet) and be trapped in the thin surface layers during the drying process, although the mechanism of migrating and trapping will be discussed later. At this stage, we can only describe that the localized clusters were formed by collisions among the trapped ATO particles, and the fluidity of the wet layers, before UV irradiation during the layer-drying process, allowed this particle migration. Figure 4 shows the surface resistivity Fs of the layers as a function of ATO volume fraction φ. A sharp decrease in surface resistivity Fs, that is, a transition from noninsulating to insulating, is observed at a very low ATO concentration (less than 0.005 volume fraction). The inset in Figure 4 shows a double logarithmic plot of the surface resistivity data. A nonlinear least-squares fit of the data suggests that, over a wide range of ATO concentrations, the surface resistivity value Fs follows the power law of the form,
Fs ) A(φ - φc)-t
(1)
where t is the critical exponent and φc is the critical concentration for percolation. Because the range of the power law can be generally applied to a limited range near the critical threshold, we tried a least-squares fit with varying ATO volume fraction range and determined the coefficient and exponent in eq 1 at the minimum χ2. The analysis has yielded a relatively high value (t ) 1.98 ( 0.17) for the critical exponent and low value (φc ) 0.00395 ( 0.00009) for the critical concentration with A ) 5.90 ( 0.34 in the range φ < 0.034. The linear relationship in the double logarithmic plot of resistivity and nanoparticle concentration indicated no specific size in the range of analysis, which characterizes the structure. The resultant low critical threshold obtained in the analysis is very important in application study.
Figure 5. Optical clarity as a function of ATO volume fraction for deposited layers. Inset: total light transmittance versus ATO volume fraction.
The value of φc is remarkably lower than the values2,3 reported in ever-developed antistatic films filled with tin oxide particles and smaller than the values predicted by the percolation theories for the systems with random sites or random bonds.4 It is, however, meaningless to compare the critical value with that predicted by percolation theories directly because the concentration of the surface layer is much different from the overall one. The nanoparticles localized at the surface layer of 25 nm thickness were measured as around 12% of total nanoparticles, and this layer undertakes the conductive property of the film. The correction for the localization of the nanoparticle distribution is required to understand the meaning of the value of φc correctly. Here, we note that the layer deposition was performed under the circumstances of ambient temperature (25 °C) and relatively high humidity (54% relative humidity (RH)). However, the layer that was deposited from the same coating solutions in a lower atmospheric humidity condition (40% RH) exhibits a lower conductivity (Figure 4). Additionally, the layers deposited at extremely low atmospheric humidity condition (lower than 30% RH) are almost insulating (data not shown). Therefore, the conductivity of the obtained layer was found to be highly dependent on the atmospheric humidity condition in the course of the layer deposition process, suggesting that water molecules in the air could take part in forming the conductive network formation at the air/liquid interface. Optical clarity of the layers was evaluated by haze and total light transmittance measurements. Haze results for the layers prepared by a deposition at 25 °C and 54% RH are shown as a function of ATO volume fraction φ in Figure 5. The haze of the layers showed nonlinear additive behavior, indicating that the intensity of the scattered light increased drastically at ATO concentrations above a threshold value. The threshold value (about 0.03 volume fraction) is, however, significantly higher than the above-mentioned critical concentration where the conductive network is formed. Total transmittance values also show nonlinear additive behavior, as is given in the inset of Figure 5. The transmittance decreases with increasing ATO concentration. However, the layer maintains the high transparency (lower than 1% total transmittance reduction) at ATO concentrations, even considerably higher than the percolation threshold φc. Thus, we conclude that the high conductivity (less than 1 × 1010 Ω square) is obtained, while still maintaining the extremely high optical transparency.
Self-Assembly of Tin Oxide Nanoparticles
Langmuir, Vol. 22, No. 22, 2006 9263
to a small effect of gravitational force on nanoparticles. As shown in the experiment, humidity in the environment plays an important role for the present specific structure. Thus, we speculate that the lowering temperature at the layer surface (air/liquid interface) by the evaporation of MEK and DAA results in the moisture condensation on the surface; the condensed water molecules induce an anisotropic distribution of polarity in the deposited wet layer and maintain fluidity of the layer. The anisotropy and the fluidity of matrix could be the driving forces of ATO particle localization or trapping. The proposed model is schematically shown in Figure 6. This model encompasses an interaction between water molecules in the air and in the deposited layer to explain the experiment results consistently. Another candidate of the mechanism could be derived from an analogy to directional crystal growth13 at the melt/solid surface. When resin is separated out from the solution by lowering MEK concentration by evaporation, the nanoparticles in the supernatant fluid could be accumulated to the growth front to form a specific structure, although the exact shape of the interface should be considerably smoothed in comparison to the crystal growth on the melt/solid surface.
Conclusions
Figure 6. Schematic illustration of the layer deposition process. (a) The layer is deposited from a coating solution consisting of surfaceactive ATO particles, UV curable matrix materials, and an organic dispersing medium. At the early stage of the drying process, the coated wet layer maintains its fluidity to a certain extent. ATO particles coalesce to form small aggregates. (b) The small aggregates and single particles are transported to the air/liquid interface by convective flow due to the temperature gradient induced by evaporation from the surface. The lowering temperature results in moisture condensation on the surface. The primary and aggregated particles are trapped at the surface to develop large clusters (agglomerates). (c) The percolating clusters are fixed and embedded in polymer matrix through drying and UV curing processes.
Finally, the mechanism of particle migration and trapping at the interface of the air and deposited layer is discussed, although clear evidence cannot be provided only from the present experiment. It is well known that a convective flow12 results from a temperature gradient in the solution during the evaporation process. The nanoparticles are transported by the convection due
We have developed a novel method to prepare transparent antistatic films by trapping nanoparticles in thin surface layers. The obtained antistatic layers have exhibited both high electrical conductivity and high optical transparency. A nonlinear least-squares fit of the data has yielded an extremely low value (φc ) 0.00395 ( 0.00009) for the critical concentration for percolation. The microscopic observations clearly indicate that the localized clusters of ATO particles are formed on (or near) layer surfaces. Conductivity or antistaticity of materials is given by locating conductive components continuously. On the other hand, transparency of materials is achieved by making them homogeneous on submicrometer scale. Therefore, to obtain both conductive and transparent properties, the size of the conductive component must be less than submicrometer, and its concentration must be low to keep high transparency. Excellent transparency and conductivity of films are derived by locating nanoparticles continuously and homogeneously at submicrometer scale in a self-organized way, the characteristic localized cluster formation. This percolating cluster formation can be closely related to the fractal dimension.5 We will report on the fractal analysis of the system soon. LA061116X (13) (a) Langer, J. S. ReV. Mod. Phys. 1980, 52, 1. (b) Ungar, L. H.; Brown, R. A. Phys. ReV. B 1984, 29, 1367. (c) Ungar, L. H.; Brown, R. A. Phys. ReV. B 1984, 30, 3993. (d) Ungar, L. H.; Brown, R. A. Phys. ReV. B 1985, 31, 5931. (e) Kessler, D. A.; Levine, H. Phys. ReV. A 1989, 39, 3041. (f) Saito, Y.; Misbah, C.; Mu¨ller-Krumbhaar, H. Phys. ReV. Lett. 1989, 63, 2377. (g) Hadji, L. Phys. ReV. E 2002, 66, 041404.
