3296
Langmuir 2005, 21, 3296-3302
Inversion of Silica-Stabilized Emulsions Induced by Particle Concentration Bernard P. Binks,* John Philip, and Jhonny A. Rodrigues Surfactant & Colloid Group, Department of Chemistry, University of Hull, Hull HU6 7RX, U.K. Received December 15, 2004. In Final Form: January 27, 2005 Emulsions of equal volumes of a cyclic silicone oil and water stabilized by fumed silica nanoparticles alone can be inverted from oil-in-water (o/w) to water-in-oil (w/o) by simply increasing the concentration of particles. The phenomenon is found to be crucially dependent both on the inherent hydrophobicity of the particles and on their initial location. Inversion only occurs in systems with particles of intermediate hydrophobicity when dispersed in oil; emulsions prepared from the same particles but initially dispersed in water remain o/w at all particle concentrations. The stability and drop size distributions in the different emulsions are compared. Various hypotheses are put forward and argued to explain this novel inversion route including adsorption of oil onto particle surfaces, hysteresis of contact angle affecting particle wettability in situ, and the structure of particle dispersions in oil or water prior to emulsification inferred from rheology and light scattering measurements. We propose that the tendency for particles to behave more hydrophobically at higher concentrations in oil is due to the reduction in the effective silanol content at their surfaces as a result of gel formation via silanol-silanol hydrogen bonds. In water, solvation of particle surfaces prevents this from occurring and particles behave as hydrophilic ones at all concentrations. A concentration-induced change in particle wettability is thus advanced.
Introduction In emulsions stabilized solely by solid particles, their concentration in the system is an important parameter influencing the size of droplets and the stability of the emulsion. This was recognized in the early work of Bechhold et al.1 who found that with increasing amounts of solid powder, smaller droplets covered with particles could be formed. Tambe and Sharma,2 investigating preferred emulsions of decane-in-water stabilized by calcium carbonate particles, observed that the stability to coalescence increased as the particle concentration was raised. Particle coverage on drop interfaces increased progressively providing a steric barrier preventing fusion of drops. For water-in-toluene emulsions (1:1) stabilized by partially hydrophobic silica particles, Binks and Lumsdon3 showed that the stability to both sedimentation and coalescence increased progressively with particle concentration. Meanwhile, for silica-stabilized emulsions of silicone oil-in-water, Binks and Whitby4 observed that the average drop diameter decreased with increasing particle concentration reaching a limit, while the stability to creaming increased. The enhanced stability at low concentrations was argued to be due to the formation of smaller drops, while at high concentrations excess particles in the continuous phase formed a network raising its viscosity impeding creaming. Although inversion of solid-stabilized emulsions from oil-in-water (o/w) to water-in-oil (w/o) can be effected by varying either the oil:water ratio,3 the ratio of hydrophilic to hydrophobic particles in mixtures,5 or the aqueous phase * To whom correspondence may be addressed. E-mail:
[email protected]. (1) Bechhold, H.; Dede, L.; Reiner, L. Kolloid-Z. 1921, 28, 6. (2) Tambe, D. E.; Sharma, M. M. J. Colloid Interface Sci. 1993, 157, 244. (3) Binks, B. P.; Lumsdon, S. O. Langmuir 2000, 16, 2539. (4) Binks, B. P.; Whitby, C. P. Langmuir 2004, 20, 1130. (5) Binks, B. P.; Lumsdon, S. O. Langmuir 2000, 16, 3748.
pH,6 inversion by changing particle concentration alone has not been reported previously. We describe here a study of the inversion of preferred emulsions of silicone oil and water by simply increasing the concentration of fumed silica particles. The initial location of particles in either oil or water and the inherent hydrophobicity of the particles are crucial in observing the phenomenon. In previous work, we already discovered that the phase in which particles were first dispersed became the continuous phase of the emulsion prepared using them. Thus, for example, for silica particles possessing 67% SiOH groups on their surface, preferred emulsions containing toluene were o/w from 1 wt % dispersions in water but w/o from 1 wt % dispersions in oil.7 Similarly, tricaprylin-water emulsions stabilized by particles with 80% SiOH groups were o/w from 2 wt % dispersions in water and w/o from 2 wt % dispersions in oil.8 These findings are consistent with the new data presented here. Using contact angle measurements, we argued that particles were more hydrophilic in situ stabilizing o/w emulsions when in contact first with water but were more hydrophobic stabilizing w/o emulsions when contacted first by oil.7,8 A hint to this contact angle hysteresis idea was given by Ramsden in 19289 and alluded to by Bartell and Hatch investigating the wetting characteristics of galena (PbS).10 They found that if the powdered material was first wetted by water, it behaved similarly to hydrophilic silica, whereas if it was first wetted with an oil it behaved as hydrophobic carbon. In addition to this effect, we discuss the inversion phenomenon in relation to the structure of silica aggregates in dispersions before emulsification as inferred from rheology and light-scattering measurements. (6) Binks, B. P.; Rodrigues, J. A. Angew. Chem., Int. Ed. 2005, 44, 441. (7) Binks, B. P.; Lumsdon, S. O. Phys. Chem. Chem. Phys. 2000, 2, 2959. (8) Binks, B. P.; Rodrigues, J. A. Langmuir 2003, 19, 4905. (9) Clayton, W. The Theory of Emulsions and Their Technical Treatment, 2nd ed.; Churchill: London, 1928; p 240. Appendix written by W. Ramsden. (10) Bartell, F. E.; Hatch, G. B. J. Phys. Chem. 1935, 39, 11.
10.1021/la046915z CCC: $30.25 © 2005 American Chemical Society Published on Web 03/01/2005
Silica-Stabilized Emulsions
Langmuir, Vol. 21, No. 8, 2005 3297
Experimental Section Materials. Water was first passed through an Elga reverse osmosis unit and then a Milli-Q reagent water system (pH ) 5.8). Sodium chloride (BDH, 99.5%) was used as received. The oil was a cyclic silicone, decamethylcyclopentasiloxane or D5, obtained from Dow Corning as 245 fluid (99%). It has a density at 25 °C of 0.95 g cm-3 and was passed though chromatographic alumina to remove polar impurities. The fumed silica powders were a gift from Wacker-Chemie, Burghausen. Their primary particle diameters range from 20 to 50 nm with specific surface areas of 200 m2 g-1, and grades possessing 87 (most hydrophilic), 71, 57, and 36% (most hydrophobic) SiOH groups on their surfaces are used here. The hydrophobicity is increased using dichlorodimethylsilane as coating reagent.3 Methods. Silica powder was dispersed in either silicone oil or 0.01 M aqueous NaCl using an ultrasonic vibracell processor (Sonics & Materials, 0.3 cm tip diameter) operating at 20 kHz and up to 10 W for 2 min. The vessel was kept cool with ice during sonification. Emulsions were prepared from equal volumes of oil dispersion (aqueous dispersion) and water (oil) using an Ultra Turrax (Janke & Kunkel) rotor-stator mixer operating at 13000 rpm for 2 min. The total emulsion volume was 10 cm3, and a 1.8 cm dispersing head was used. In one system, emulsification was also achieved by ultrasonification at 55 W for 1 min. The emulsion conductivity was measured immediately after formation using Pt/Pt black electrodes and a Jenway 4310 digital conductivity meter. The emulsion type was inferred by observing whether a drop of emulsion dispersed when added to a small volume of neat water or oil. The stability at 25 °C of water continuous emulsions to creaming or coalescence was assessed by monitoring the change with time of the position of the clear water-emulsion or oil-emulsion interfaces, respectively. For oil continuous emulsions, the downward movement of the oilemulsion boundary was used as a measure of the stability to sedimentation, and the position of the water-emulsion interface was used as an indicator of coalescence. Drop size distributions of emulsions were determined using a Malvern MasterSizer 2000 light diffraction instrument. Approximately 0.1 cm3 of emulsion was diluted to 100 cm3 with either water or oil containing a low concentration of dispersed particles. A Nikon Labophot microscope equipped with a DIC-U camera was used to observe emulsions also. A drop of emulsion was diluted with its own continuous phase on a microscope slide and covered with a coverslip. Images were processed using Adobe Photoshop 5.0 LE software. The structure of silica particle dispersions in oil was visualized using transmission electron microscopy (TEM) in which a drop of dispersion was placed on a carbon-coated copper grid (0.3 cm diameter, mesh size 200 holes/cm) and allowed to dry overnight at room temperature. The instrument used was a JEOL JEM2011 with an acceleration voltage of 120 kV. The size of silica particle aggregates in water was determined using a Malvern ZetaSizer 3000 HS instrument. The flow rheology of oil or aqueous dispersions of silica particles possessing 71% SiOH was determined at 20 °C using a Bohlin CVO 120 high-resolution rheometer in the controlled rate mode. A truncated cone (4° cone angle, 4 cm cone diameter) and plate geometry (150 µm gap width) was employed. They were cleaned by rinsing in 2-propanol (Aldrich, 99%). A solvent trap was used to prevent evaporation during measurement. Typically, 2 cm3 of dispersion was placed in the gap and left 15 min to equilibrate. A preshear was applied at a shear rate of 1 s-1 for 1 min. The shear rate was increased in a series of steps from 0.017 to 100 s-1. For each step, shear was applied for 10 s while no information was recorded (“delay” time) and then for 10 s during which the shear stress was measured and the viscosity was calculated (“integration” time).
Results and Discussion (i) Emulsions from Particles Initially Dispersed in Oil. The four powdered samples of silica of different hydrophobicity disperse in neat D5 silicone oil, and emulsions containing water can be prepared from them. In Figure 1, the emulsion conductivities are plotted as a function of silica particle concentration initially in the oil
Figure 1. Conductivity of emulsions of D5 silicone oil-water (1:1 by vol.) stabilized by silica particles of different hydrophobicity (given) as a function of particle concentration in oil.
phase for the silica particle types. For the most hydrophobic silica (36% SiOH), the conductivities are below 9 µS cm-1 and the drop test indicated w/o emulsions at all particle concentrations. For the most hydrophilic silica (87% SiOH), the conductivities are high (between 300 and 650 µS cm-1) and the drop test confirmed the formation of o/w emulsions at all particle concentrations. These results are as expected and in line with the previously observed effect of the particle hydrophobicity on emulsion type.11 However, for particles of intermediate hydrophobicity (57 and 71% SiOH), emulsion inversion occurs with increasing particle concentration, from o/w at low concentrations (high conductivities) to w/o at high concentrations (low conductivities). This is unexpected and has not been reported before. The concentration of particles required for inversion is less in the case of the more hydrophobic silica (≈1 vs 2 wt %, respectively). This simple route to phase inversion could have significant potential industrially. It is presumed that particles of intermediate hydrophobicity exhibit contact angles at the oil-water interface around 90°, and emulsions prepared from them are therefore very sensitive to the prevailing conditions. Optical micrographs of the noninverting emulsion systems stabilized by particles of extreme hydrophobicity are given in Figure 2 at a particle concentration of 5 wt %. A simple w/o emulsion in which drops are characteristically flocculated is seen for particles of 36% SiOH (upper), whereas a simple o/w emulsion containing discrete drops is observed for particles of 87% SiOH (lower). By contrast, images of both emulsion types depending on particle concentration are given in Figures 3 and 4 for particles of 57 and 71% SiOH, respectively. In Figure 3, an o/w emulsion comprising discrete spherical drops forms at 0.5 wt % particles (upper), while w/o emulsions occur at 1 and 5 wt % in which drops are flocculated and some appear nonspherical and distorted (middle and lower). In Figure 4, spherical oil drops in simple o/w emulsions are easily identified at 2 wt % (upper) whereas distorted water drops are observed at higher particle concentrations (middle and lower). In fact we see multiple water globules in oil-in-water-in-oil (o/w/o) emulsions at concentrations (11) Binks, B. P.; Lumsdon, S. O. Langmuir 2000, 16, 8622.
3298
Langmuir, Vol. 21, No. 8, 2005
Binks et al.
Figure 2. Optical microscope images of D5 silicone oil-water emulsions (1:1) stabilized by silica particles at 5 wt % in oil: (upper) water drops in a w/o emulsion with particles of 36% SiOH; (lower) oil drops in an o/w emulsion with particles of 87% SiOH.
near inversion (middle) which become a mixture of multiple o/w/o and simple w/o emulsions away from inversion (lower). The existence of multiple emulsions stabilized by only one type of silica particle is relatively new,8 it having been shown earlier that binary mixtures of particles of different hydrophobicity are normally required.12 The stability after 3 months of w/o emulsions stabilized by silica particles with 36% SiOH is represented in Figure 5. All emulsions were completely stable to coalescence with no release of water being visible. The emulsion drops sediment however (open points), but the extent of sedimentation decreases progressively with increasing particle concentration, becoming zero by 4 wt %. One reason for the increased stability to sedimentation observed is the decrease in the median drop diameter with particle concentration, also shown in Figure 5 (filled points). The major change in size occurs up to 3 wt % (from 175 to 13 µm), after which the diameter remains constant. A second reason for the enhanced stability is due to the increased viscosity with particle concentration of the continuous oil phase (see later), in which excess particles form a network structure trapping the drops within. Interestingly, the uniformity of the drop size distribution, defined in ref 8, increases progressively with the decreasing median diameter from 0.22 to 0.95. Unlike surfactant-stabilized emulsions, the most monodisperse ones (low uniformity) are those of largest drop size, in this case over 170 µm. The stability of emulsions prepared with silica particles possessing 71% SiOH was also investigated in detail, and Figure 6a shows the influence of the particle concentration in oil on the release of both water and oil 3 months after (12) Barthel, H.; Binks, B. P.; Dyab, A. K. F.; Fletcher, P. D. I. Patent assigned to Wacker-Chemie (Munich), US 2003/0175317 A1, Sept. 2003.
Figure 3. Optical microscope images of D5 silicone oil-water emulsions (1:1) stabilized by silica particles with 57% SiOH from oil dispersions at different particle concentrations: (upper) oil drops in an o/w emulsion formed at 0.5 wt %; (middle) water drops in a w/o emulsion formed at 1 wt %; (lower) water drops in a w/o emulsion formed at 5 wt %.
emulsion formation. It will be recalled that emulsion inversion occurs between 2.0 and 2.5 wt %. For o/w emulsions prior to inversion, the creaming extent was gradually reduced with increasing particle concentration (open points, left ordinate). In addition, the coalescence of oil drops was completely prevented for concentrations above or equal to 0.5 wt % (filled points, right ordinate). For oil continuous emulsions after inversion, the sedimentation of water drops and globules decreased slightly by 5 wt % of particles (filled points), without any coalescence (open points). The majority of these stability changes are linked to the changes in the median diameter of emulsion drops in the separate systems, shown in Figure 6b. As before, the drop diameter increases noticeably approaching low particle concentrations for both emulsion types. With sufficient particles to cover the newly created interfaces (by 2 wt % in oil initially for o/w and 4 wt % in oil initially for w/o), the average drop diameter is similar in the two cases. (ii) Emulsions from Particles Initially Dispersed in Water. Silica particles of intermediate hydrophobicity
Silica-Stabilized Emulsions
Langmuir, Vol. 21, No. 8, 2005 3299
Figure 5. Stability to sedimentation after 3 months (open points, left ordinate) and median drop diameter (filled points, right ordinate) of emulsions of water-in-D5 silicone oil stabilized by silica particles with 36% SiOH as a function of particle concentration in oil.
Figure 4. As in Figure 3 but for silica particles with 71% SiOH: (upper) oil drops in an o/w emulsion formed at 2 wt %; (middle) water globules in a o/w/o emulsion formed at 2.5 wt %; (lower) water globules and water drops in a mixture of o/w/o and w/o emulsions formed at 5 wt %.
can be dispersed initially in water as well as in oil. In systems containing either toluene7 or tricaprylin oil,8 we found earlier that preferred emulsions stabilized by such particles at relatively high concentrations were o/w from water-borne dispersions but w/o from oil-borne ones. It is of interest therefore to investigate the emulsion type as a function of particle concentration from aqueous dispersions of particles possessing 71% SiOH. Figure 7 shows the effect of particle concentration on the conductivity of emulsions (1:1) where particles originate in the aqueous phase (open points). Their conductivity remains high at all particle concentrations and, combined with results of the drop test, we conclude that emulsions are always o/w and no inversion occurs. This is in sharp contrast to that described above from oil-based dispersions (Figure 1). Increasing the particle concentration in water leads to a gradual decrease in the median diameter of emulsion drops (filled points), which reaches a limit for these emulsification conditions of 20 µm by 4 wt %. By comparing the drop sizes for o/w emulsions prepared from oil dispersions (Figure 6b) and aqueous dispersions (Figure 7), we see that the median diameter is more or less independent of the initial location of particles. The microscope images of o/w emulsions prepared from aqueous particle dispersions, Figure 8, show drop sizes in good agreement with those measured by light diffraction,
being 295, 60, and 20 µm at 0.1, 1, and 5 wt % particles, respectively. A characteristic of solid-stabilized emulsions, due to the phenomenon of limited coalescence,13,14 is that their monodispersity surprisingly increases with the average drop diameter.4,8 This is well demonstrated in Figure 8, where the uniformity of the drop distribution is as low as 0.17 for the largest drops (upper). Such low monodispersity for relatively concentrated emulsions leads to near perfect hexagonal close packing of droplets, where 6 drops envelop the central one. (iii) Rheology and Structure of Particle Dispersions. The main question arising from the foregoing description for partially hydrophobic silica particle systems is why do emulsions invert from o/w to w/o with increasing particle concentration when initially dispersed in oil but remain o/w when initially dispersed in water? There are several hypotheses which can be put forward to explain this marked difference. One possibility is that the hydrophobicity of the silica particles in situ is different in the two cases as a result of the adsorption of oil molecules onto their surfaces for the oil-borne dispersions. If this does occur for the cyclic silicone oil used here, it is not clear whether particles would be rendered more hydrophilic or more hydrophobic as a result. Further, it may explain the opposite emulsion type seen at high particle concentrations (>2.5 wt %), but the argument breaks down in attempting to explain the same emulsion type (o/w) at low particle concentrations. In addition, studies of the adsorption of polymeric silicone molecules onto silica particles dispersed in the polymeric oil15 showed that adsorption at room temperature was very slow (months), with the extent decreasing significantly for a decrease in polymer molecular weight and an increase in particle hydrophobicity.16 It is therefore unlikely that adsorption of this low molecular weight oligomeric silicone oil onto such particles will occur to any significant degree in the 15 min the two phases are in contact before emulsification. (13) Wiley: R. M. J. Colloid Sci. 1954, 9, 427. (14) Arditty, S.; Whitby, C. P.; Binks, B. P.; Schmitt, V.; Leal-Calderon, F. Eur. Phys. J. E 2003, 11, 273. (15) DeGroot, J. V., Jr.; Macosko, C. W. J. Colloid Interface Sci. 1999, 217, 86. (16) Barthel, H.; Ro¨sch, L.; Weis, J. Surf. Rev. Lett. 1997, 4, 873.
3300
Langmuir, Vol. 21, No. 8, 2005
Binks et al.
Figure 7. Conductivity (open points, left ordinate) and median drop diameter (filled points, right ordinate) of D5 silicone oilin-water emulsions (1:1) stabilized by silica particles with 71% SiOH as a function of particle concentration in water.
Figure 6. (a) Percentage of water resolved (open points, left ordinate defined as volume of water/total volume) and oil resolved (filled points, right ordinate defined as volume of oil/ total volume) after 3 months of emulsions of D5 silicone oilwater stabilized by silica particles with 71% SiOH as a function of particle concentration in oil. (b) Median drop diameter of emulsions described in (a).
A second possibility for the marked influence of particle location on emulsion type, argued before,17 is that the same particles exhibit a different contact angle with the oil-water interface in the two situations and is linked to contact angle hysteresis. For particles originally in oil, the equivalent of the advancing angle θa (measured into water) is preferred upon contact with water, in which water advances over the particle surface. By contrast, if particles originate in water, the equivalent of the receding angle θr is adopted upon contact with oil, in which water recedes over the particle surface. As demonstrated by us earlier,8 θa is always greater than θr for a range of partially hydrophobic glass surfaces (mimicking silica particles) in triglyceride oil-water systems, in line with the prediction that particles behave more hydrophobically (larger angle) and prefer to stabilize w/o emulsions when initially in oil but are more hydrophilic (lower angle) and prefer o/w emulsions when initially in water. The same conclusion that surfaces of galena are hydrophilic when wetted first by water but more hydrophobic when wetted first by oil was arrived at in a different context.10 Again however, this argument does not explain the experimental data (17) Aveyard, R.; Binks, B. P.; Clint, J. H. Adv. Colloid Interface Sci. 2003, 100-102, 503.
presented here at low particle concentrations, in which emulsions are always o/w. A third and most plausible origin of the difference relates to the structure and ensuing rheology of the particle dispersion in oil or water before emulsification, influencing the nature of the adsorbed film. Fumed silica is prepared by the flame hydrolysis of SiCl4. Initially, spherical primary particles of silica (20-50 nm diameter) are formed, but these can fuse irreversibly into aggregates (a few hundred nanometers). Silanol (SiOH) groups are generated on the silica surface during preparation. When dispersed in liquids, adjacent aggregates can interact via silanol-silanol hydrogen bonding giving rise to larger structures called flocs or agglomerates. At high enough silica concentrations, a three-dimensional network of flocs extends throughout the volume of the sample and the suspension is considered to be a gel. The flocculated microstructure therefore has multiple levels. Fumed silica is thus widely employed as a filler in silicone rubber-based products and lubricating greases modifying their viscosity.18,19 The rheology of fumed silica dispersions in a variety of liquids has been investigated by, inter alia, Otsubo et al.,20 Khan and co-workers,21-23 and Barthel.24 The dependence of the viscosity on particle concentration depends on both the hydrophobicity of the particles and the polarity of the liquid. For partially hydrophobic particles in a nonpolar liquid like mineral oil, particleparticle interactions dominate through hydrogen bonds between silanol groups on neighboring particle aggregates giving rise to a noticeable gel-like behavior even at low solids concentration. In a polar medium like water, particle-solvent interactions dominate and the liquid forms a solvation layer around particle aggregates rendering them inactive for network formation.23 Areas on particle surfaces not wetted by the solvent, e.g., dimethyl (18) Young, G. J.; Chessick, J. J. J. Colloid Sci. 1958, 13, 358. (19) Iler, R. K. The Chemistry of Silica; Wiley: New York, 1979; p 588. (20) Otsubo, Y.; Horigome, M.; Umeya, K. J. Colloid Interface Sci. 1981, 83, 240. (21) Khan, S. A.; Zoeller, N. J. J. Rheol. 1993, 37, 1225. (22) Raghavan, S. R.; Khan, S. A. J. Rheol. 1995, 39, 1311. (23) Raghavan, S. R.; Walls, H. J.; Khan, S. A. Langmuir 2000, 16, 7920. (24) Barthel, H. Colloids Surf., A 1995, 101, 217.
Silica-Stabilized Emulsions
Figure 8. Optical microscope images of emulsions described in Figure 7. Oil drops of o/w emulsions formed at 0.1 wt % (upper, note monodispersity), 1 wt % (middle), and 5 wt % (lower) of particles in water.
groups, may interact with similar areas on adjacent particles by hydrophobic bonding to give some bridging between particles, which is however much weaker. Such linkages between aggregates, of whatever type, are easily disrupted by shearing forces but readily reestablished when the mass is at rest. A high degree of thixotropy is thus present in these dispersions.19 To see if this different rheological behavior is observed in our systems, we have measured the flow rheology of dispersions of silica particles possessing 71% SiOH groups in both silicone oil and water as a function of particle concentration. Parts a and b of Figure 9 show how the apparent viscosity of these dispersions varies as a function of applied shear rate in the case of oil and water, respectively, 15 min after preparation. For this Newtonian oil (a), we recover the viscosity quoted in the literature in the absence of particles of 4.5 mPa s. Increasing the particle concentration raises the viscosity and the dispersions become increasingly shear thinning. In water however (b), dispersions remain of low viscosity and Newtonian up to 3 wt % of particles, shear thinning at higher concentrations. The apparent viscosity at three selected shear rates is plotted versus particle concentration in Figure 10 for both liquids. The viscosity of the oil dispersions is always greater than that of aqueous dispersions and the former
Langmuir, Vol. 21, No. 8, 2005 3301
Figure 9. (a) Apparent viscosity versus shear rate of dispersions of fumed silica (71% SiOH) in D5 silicone oil at 20 °C at different concentrations (given in wt %). (b) Apparent viscosity versus shear rate of dispersions of fumed silica (71% SiOH) in water at 20 °C at different concentrations (given in wt %).
Figure 10. Apparent viscosity versus particle concentration at 20 °C for D5 silicone oil dispersions (open points) and aqueous dispersions (filled points) of fumed silica (71% SiOH) at shear rates of 10 (O), 30 (0), and 100 s-1 (4). Schematic representation of isolated particle aggregates in water (lower) and connected particle aggregates in oil (upper). Photograph of 5 wt % dispersion in oil (upper, gel-like does not flow) and water (lower, fluid which flows) 5 min after inverting the vessels.
increase more markedly with particle concentration. At 5 wt % in oil, a gel forms which resists flow upon inverting
3302
Langmuir, Vol. 21, No. 8, 2005
Figure 11. TEM image after drying a dispersion of silica particles (71% SiOH) at 2 wt % in D5 silicone oil.
the vessel, in contrast to the fluid dispersion in water (see photographs of vessels on right). In agreement with earlier studies18-24 and represented schematically on the figure, the volume filling gel which forms in oil arises as particle aggregates hydrogen bond to each other, whereas such aggregates remain discrete in water. A TEM image of a dried dispersion of particles at 2 wt % in oil is given in Figure 11 where the network of aggregates is clearly seen. Attempts to measure the size of such connected aggregates in oil using light scattering (ZetaSizer) were unsuccessful since, for dispersions diluted from between 1 and 5 wt % to 0.016 wt %, they sedimented within 1 min in the cuvette. For dispersions diluted similarly in water, no sedimentation occurred and the average aggregate diameter increased from 170 to 330 nm between 1 and 5 wt %, respectively. On the basis of these results, we propose that the effective hydrophobicity of the silica particles in oil dispersions increases with particle concentration as more and more silanol groups become involved in hydrogen bonds with each other. The free SiOH content, important for determining the wettability of such particles at the oil-water interface,11 thus decreases such that these particle aggregates then prefer to stabilize w/o emulsions. The above idea also explains why emulsions stabilized by particles of inherently lower silanol content (57%) invert to w/o at lower particle concentration (see Figure 1) than more hydrophilic particles (71%). This is because being more hydrophobic initially requires a smaller number of aggregate linkages to reduce the effective silanol content to the level imparting sufficient hydrophobicity to stabilize w/o emulsions. An additional set of experiments was designed to further aid our understanding of these results, in which emulsions were prepared by another means. Ultrasonic irradiation has recently been employed to prepare surfactantstabilized emulsions. Abismaı¨l et al.25 investigated the properties of kerosene-in-water emulsions containing
Binks et al.
ethoxylated sorbitan monostearate prepared by mechanical agitation or using ultrasound. The average drop diameter formed using ultrasound was smaller than that obtained with an Ultra Turrax mixer and emulsions were more stable to creaming as a result. Behrend et al.26 likewise compared the efficiency of using ultrasound to that of high-pressure homogenization for o/w emulsions of sodium dodecyl sulfate and w/o emulsions of polyglycerin polyricinoleate. Since the energy density input into the system is higher using ultrasound compared with the Ultra Turrax,26 ultrasound should disrupt connected aggregates of particles in oil to a higher degree causing the effective SiOH content to increase toward its initial level. In addition, such disrupted aggregates are then available for immediate adsorption at freshly created oil-water interfaces. We tested this hypothesis by preparing 1:1 emulsions of D5 silicone oil and water using an ultrasonic probe (55 W for 1 min) as a function of the particle concentration in oil using particles of 71% SiOH. Reassuringly, emulsions were of high conductivity (400-560 µS cm-1), dispersed in water and were confirmed by microscopy as simple o/w for all particle concentrations between 0.1 and 5 wt %, in contrast to those which phase inverted if prepared using an Ultra Turrax homogenizer (Figure 1). The average drop diameters of emulsions made using ultrasound were smaller than those made with mechanical mixing, and the uniformity of the distribution was relatively low (
