Article pubs.acs.org/Langmuir
Colloidal Aggregation in Mixtures of Partially Miscible Liquids by Shear-Induced Capillary Bridges Niek Hijnen* and Paul S. Clegg* School of Physics and Astronomy, The University of Edinburgh, Mayfield Road, Edinburgh EH9 3JZ, United Kingdom S Supporting Information *
ABSTRACT: We have studied shear-induced aggregation of colloidal silica particles suspended in a variety of partially miscible liquid mixtures. The shared characteristic of the investigated systems is that after liquid−liquid phase separation of the binary liquid mixtures one phase completely wets the particles. We have explored compositions where there are insufficient quantities of the particle wetting component to induce phase separation. As the proportion of the wetting component is increased, we find a significant concentration range where shear-induced aggregation takes place. The macroscopic characteristics of this phenomenon are demonstrated, for which observations were greatly facilitated by mostly using liquid pairs partially miscible at room temperature. Measurements revealing the adsorption of the minority component to colloidal particles show that capillary condensation between particles causes the observed aggregation. The likely microscopic features underlying this aggregation behavior are then discussed. Finally, the overall picture of these systems is sketched as a nonequilibrium liquid−liquid phase diagram, in which outside the binodal there is a region of shear-induced aggregation.
■
INTRODUCTION Stable colloidal suspensions rely on repulsive interactions between constituent particles to prevent them from flocculating. In contrast, when some attractive interparticle interaction dominates, colloids can assemble (aggregate) into larger structures to result in a material with potentially useful properties.1−4 Being able to tune colloidal interactions in a straightforward way is therefore very attractive. Interest in using partially miscible liquid mixtures to control interparticle interactions was initiated by observations of temperature-reversible wetting-induced colloidal aggregation in such a solvent.5 An important characteristic of the studied system was that after the mixture phase separated there was complete wetting of the particles by one of the phases. Adjacent to the miscibility gap of the pure liquid mixture, a narrow region was observed where aggregation of suspended particles occurs. Subsequent studies have shown that in the vicinity of the critical point the aggregation is caused by the confinement of concentration fluctuations in the solvent between particles.6,7 Further away from the critical composition, where the component that completely wets the particles is in the minority, a different mechanism is responsible for the aggregation. Here liquid bridges form between particles due to the coalescence of wetting layers present on particles.6−9 Alternatively, solid particles in partially miscible liquid mixtures have also been interpreted as true ternary mixtures.10 The aggregation in these systems can be temperaturereversible provided that particle surfaces remain repulsive, hence preventing aggregation into the primary minimum of their interaction potential. This means particles can be redispersed upon changing the temperature, removing the critical fluctuations or the liquid bridges by remixing. In © 2014 American Chemical Society
aggregates, large distances separating particles, and even particles arranging themselves in crystalline structures, have indeed been observed. 9,11 It has also been proposed theoretically that crystallization is possible for one of the investigated systems in the case where a liquid bridge forms from adsorbed layers.12,13 Moreover, the interparticle interactions will not only be affected by adsorption but also in general depend on liquid composition as appears already from DLVO theory.14,15 So far this aggregation behavior is observed and studied in several systems.16 The two types of binary liquid mixtures that have mostly been used in experiments are water and pyridine derivatives (2,6-lutidine and 3-methylpiridine)5,7,17−19 and water and n-alkyl polyglycol ethers (nonionic surfactants)10,11,20 which both have a LCST above 25 °C. In this case, to approach the miscibility gap, samples of predefined compositions can be prepared at room temperature and then heated in a controlled manner. In addition to these liquids, one report also uses water and isobutyric acid as a qualitative comparison.19 In this article we present a study of a remarkable colloidal aggregation process in several similar systems using a distinctively different approach. The colloidal particles in our samples are charge-stabilized in both individual liquid components. As in the reports mentioned above, after phase separation of the liquid mixture, they are located exclusively in one of the liquid phases. The main component of this phase, which completely wets the particles, is hereafter referred to as Received: March 21, 2014 Revised: April 30, 2014 Published: May 6, 2014 5763
dx.doi.org/10.1021/la501060w | Langmuir 2014, 30, 5763−5770
Langmuir
Article
Preparation and Handling of Samples. Silica particles were washed several times with water (basic pH) after synthesis and then stored in ethanol. Particles were dried at 100 °C (Binder VD23 oven) for 30 min to let bulk solvent evaporate. Then, the resulting cake was ground into a fine powder using a spatula, and subsequently this was dried at 100 °C under vacuum for 60 min. The preparation of samples depended slightly on the required liquid composition. Samples with a liquid composition for which the components were completely mixed at room temperature were prepared by dispersing the colloids directly into the liquid mixture using an ultrasonic bath (VWR). When preparing a sample with a liquid composition for which the liquids were phase-separated at room temperature, the particles were first dispersed into the majority component before adding the minority component that preferentially wets the particles. Samples were photographed in a thermostated (21 ± 1 °C) room using an SLR camera (Canon EOS 40D with a Tamron 70−300 mm Tele-Macro lens), controlled by a computer using the canon EOS utility software. Time-lapse photography was used to monitor sedimentation. Samples were studied before and after applying shear by 30 s of vigorous shaking using a vortex mixer. For determining the fraction of shear-aggregated colloids, samples were left for 10 min after vortex mixing, allowing aggregates to settle to the bottom of the vial. A known amount (by weight) of supernatant was then transferred into a weighed vial, and the particles were spun down using a centrifuge, after which a clear supernatant was removed. Remaining solvent was subsequently removed by drying in an oven (100 °C, 15 min), and finally, the mass fraction of colloids in the supernatant was determined after weighing the vial with particles. This was then related to the initial sample composition to obtain the fraction of shear-aggregated colloids. ζ-Potential Measurements. Samples of, typically, a few milliliters were prepared by dispersing the colloids (0.2 vol %) directly into mixtures of various liquid compositions. Measuring the ζ-potential was done using a Malvern Zeta Sizer Nano Z. For the measurements, 2butanol/water samples were transferred into disposable folded capillary cells, while for propanal/water samples a dip cell was used in combination with a glass square cuvette. Solvent properties of the majority solvent were used to extract the ζ-potential from the electrophoretic mobility. Fluorencence Spectroscopy. In order to be able to measure fluorescence spectra of Nile Red in the bulk propanal/water solvent of dispersions, first stock mixtures at 4, 8, and 12 wt % water (∼25 mL)
the colloid-wetting component Cw. The approach adopted in this work differs from those of previous studies in the use of liquid composition instead of temperature as the adjustable parameter to approach the miscibility gap. To this end, liquid pairs were used that have not yet been studied in this context and which are (with one exception) only partially miscible at room temperature. This conveniently allows the binodal to be approached by increasing the concentration of Cw in the liquid mixture, as indicated in Figure 1.
Figure 1. Schematic liquid−liquid phase diagram with an arrow indicating samples which are probed in this experimental approach.
Below, first the experimental methods are detailed, followed by an overview and discussion of the results.
■
EXPERIMENTAL SECTION
Materials. Propanal (propionaldehyde, ≥97%, FG, Kosher, SigmaAldrich), 2-butanol (99%, Fisher Scientific), nitromethane (99+%, stored under nitrogen, Acros Organics), ethylene glycol (anhydrous, 99.8%, Sigma-Aldrich), propylene glycol propyl ether (PGPE, 1propoxy-2-propanol, 99%, Sigma-Aldrich), and Nile Red (Technical grade, Sigma-Aldrich) were used as received. Particles used were silica spheres prepared by the Stöber method.21,22 Nonfluorescent particles were used with the water/propanal and water/2-butanol (from TEM images, the average particle radius RTEM = 0.38 μm with polydispersity σ = 7%), and FITC-labeled particles were used with nitromethane/ ethylene glycol. Particles used with water/PGPE were the same FITC labeled particles but coated with unlabeled Stöber silica (RTEM = 0.35 μm with σ = 12%).
Figure 2. (a) Phase diagram of water and propanal23 with an inset demonstrating partitioning of particles (1 vol %) into the water-rich phase (liquid mixture: 30 wt % water). (b) Vials containing suspensions of silica particles (1 vol %) in propanal/water mixtures at various liquid compositions (given in wt % of Cw in the liquid mixture): initially stable suspensions (middle frame) and the resulting state of the same samples after vortex mixing (top frame) and tilting them backward after particle sediments have formed over ∼24 h under gravity (bottom frame). (c) Estimates of the fraction of aggregated particles by measuring the concentration in the supernatants of the samples shown in the top frame of (b). (d) Estimates of the fraction of aggregated particles plotted as a function of particle concentration for intermediate water concentrations. 5764
dx.doi.org/10.1021/la501060w | Langmuir 2014, 30, 5763−5770
Langmuir
Article
Figure 3. (a) Phase diagram of 2-butanol and water24 with an inset demonstrating partitioning of particles into the water-rich phase. (b) Vials containing silica particles (1 vol %) dispersed into 2-butanol/water mixtures at various liquid compositions (given in wt % of Cw in the liquid mixture) using sonication, and (c) a plot of ζ-potentials at various water concentrations. (d) The same samples as shown in (b), but now the suspensions were shaken vigorously using a vortex mixer, and (e) the fraction of aggregated particles for the relevant samples is plotted as a function of water concentration. (f) Shows the fraction of aggregated particles plotted against the total particle volume fraction for several water concentrations. were prepared containing 1 μM Nile Red. Subsequently, particles were dispersed at varying concentration in the stock mixtures to obtain samples of ∼5 mL. After shearing the samples for 30 s using a vortex mixer, particles were spun down and the clear supernatant transferred into a square cuvette (10 mm path length, optical glass, from Starna Scientific) for the measurements. The remaining stock mixtures were used as blanks and measured in between supernatant samples of the dispersions. For measuring the spectra a Varian Cary Eclipse fluorescence spectrophotometer was used, and for every measurement multiple (typically five) spectra were recorded. Processing and peakfitting were done in Origin 8.5 using batch processing, outputting the relevant fit parameters and measures of statistical accuracy.
suspensions were obtained up to 16 wt % water, with the subsequent sample at 18 wt % water being phase-separated. The particles are charge-stabilized and carry a negative surface charge (ζ-potential ≈ −50 mV, see Figure S1b). As expected from their size and density mismatch with the solvent, they were observed to sediment slowly to the bottom of the vial, where eventually all particles are collected within a concentrated layer. While this behavior is consistent over the entire range of samples, the response of the formed sediments to gravity is not. In the absence of water and at very low water concentrations, the dense sedimented layer of charge-stabilized particles flows in response to gravity when tilting the vial backward (Figure 2b, bottom frame). Simply shaking the vials redisperses the particles to again obtain a stable suspension. As the water concentration is increased, a more solidlike “cake” is formed, which does not flow under gravity. In this case, particles cannot be redispersed properly by shaking, and aggregates remain as evidenced by rapid sedimentation into voluminous sediments. Curiously, this indicates that within the concentrated layer the particles have aggregated, as opposed to remaining wellstabilized against aggregation due to their charged surfaces. Another surprising result is obtained when the stable suspensions shown in Figure 2b are vigorously shaken, instead of being left to sediment. There is no effect on colloidal stability for suspensions at low water concentration (Figure 2b, top frame), as can be expected with the particles being chargestabilized. However, at higher concentration of the Cw, the particles in the initially stable dispersions (Figure 2b, middle frame) aggregate and quickly settle into voluminous sediments (Figure 2b, top frame). The supernatants above sediments clearly stay turbid for immediate concentrations (8 and 10 wt %). Plotting the fraction of aggregated particles as a function of the Cw concentration shows a relatively sharp increase between 6 and 10 wt % water (Figure 2c).
■
RESULTS First, the general behavior of suspensions with liquid compositions relatively poor in the Cw is presented for the various systems studied. Those investigated more extensively are aqueous mixtures and will be presented first. For completeness, they are supplemented by a nonaqueous liquid mixture and an aqueous mixture that is fully miscible at room temperature, similar to that used in previous studies.5,10,18 Macroscopic Observations. Water and Propanal. Water and propanal are partially miscible at room temperature (Figure 2a). In the left-hand vial of the inset in Figure 2a a dispersion of silica particles in propanal is added on top (lower density) of some water; the latter makes up 30 wt % of the total liquid in the vial. At this composition the liquids phase-separate at room temperature, and after vigorously shaking the sample all the particles transfer into the bottom (since higher density), waterrich phase, indicating that water in this case is the Cw. Alternatively, this is confirmed by observing the colloids locating themselves into the water-rich phase after a temperature quench leading to phase separation of the liquids (Figure S1a). A series of suspensions with increasing water concentration were prepared, using sonication to disperse the particles into the liquid mixtures (Figure 2b, middle frame). Stable 5765
dx.doi.org/10.1021/la501060w | Langmuir 2014, 30, 5763−5770
Langmuir
Article
Thus, it appears that for intermediate Cw concentration only a fraction of particles has aggregated. To further explore this regime, the particle concentration was varied for the range of water concentrations 6−10 wt %. The fraction of aggregated particles appears to be independent of the particle concentration, as demonstrated in Figure 2d. Only at 8 wt % water the fraction of aggregated particles is observed to suddenly drop again for the highest particle concentration. Water and 2-Butanol. The phase diagram of 2-butanol and water shows that these liquids are also partially miscible at room temperature (Figure 3a). Silica particles present in such mixtures are located in the water-rich phase after liquid−liquid phase separation upon heating or mixing, in an analogous way to the propanal/water system (inset of Figure 3a and Figure S2). Hence, in this system water is again the Cw. In the 2-butanol/water system, approaching the miscibility gap by increasing the Cw concentration, an onset of colloidal aggregation is already observed at relatively low water concentrations (Figure 3b). Aggregation vanishes again at higher water concentration. These observations are consistent with ζ-potentials, which show that when increasing the water concentration the colloids go through a point of zero charge (PZC) (Figure 3c). In pure 2-butanol the silica particles are positively charged, and since they are of opposite charge in water, adding water eventually changes the sign of the total particle charge. Closer to the binary liquid’s binodal, aggregation is seen again. The dispersion with a liquid composition where the two components only just mix (35 wt % water) is unstable and the particles aggregate to form a voluminous sediment (Figure 3b). At somewhat lower Cw concentration, stable dispersions can be obtained (Figure 3b), but shear-induced aggregation is observed (Figure 3d,e) after vigorous shaking. In contrast to the propanal/water system, the fraction of aggregated colloids increases with particle concentration (Figure 3f). Taking into account the amount of Cw available per particle by rescaling the data in Figures 2d and 3f did not provide any other insight. Ethylene Glycol and Nitromethane. The nonaqueous liquid mixture included in this study consists of ethylene glycol (Cw) and nitromethane. FITC-labeled silica particles dispersed in nitromethane completely transfer into the ethylene glycol-rich phase after mixing with an appropriate amount of ethylene glycol (far right vial in Figure 4a25). These liquids are also partially miscible at room temperature, but with a fairly low solubility of the Cw (∼8 wt %) and a low UCST (∼40 °C).23 Using sonication, stable suspensions were obtained for ethylene glycol concentrations running up to maximum solubility of the Cw at room temperature, while crossing into the miscibility gap results in phase-separated samples (Figure 4a, top). As in the propanal/water system, the stable suspensions slowly sediment into compact concentrated layers, which upon adding some Cw appear more solidlike (i.e., do not flow under gravity when tilting the vials). Vigorous shaking of some of these samples results in shear-induced aggregation, similar to the previously discussed systems (Figure 4a, bottom). At 7 wt % ethylene glycol the situation appears to be different to its neighbors on both sides. No clear supernatant and sharp sedimentation front is observed, 3 h after dispersal by sonication. After some time the supernatant clears up, and the sedimentation front is visible as in the samples with lower Cw concentration. Once subjected to shear, a compact and very viscous phase is collected at the bottom of the vial, much like what is observed in the phase-separated 8 wt % sample. It was
Figure 4. (a) Suspensions of silica particles (1 vol %) in various mixtures of ethylene glycol (concentrations given in wt %) and nitromethane, which were initially mixed and dispersed by sonication and subsequently either (top frame) left undisturbed to sediment for 3 h, or (bottom frame) first vigorously shaken and then left undisturbed to sediment for 3 h. (b, left frame) Two suspensions of silica particles (1 vol %) in different mixtures of water (concentrations given in wt %) and propylene glycol propyl ether heated from room temperature to 50 °C in a water bath and left to sediment for 1 h. (b, right frame) The same heated samples, 1 h after being vortex mixed.
found that some Cw-rich phase separates out of the bulk solvent, but that not many of the present particles are collected within it. The ethylene glycol-rich droplets then simply float to the top of the sample. Experiments demonstrating this behavior for a range of liquid compositions are presented in Figure S3. Water and Propylene Glycol Propyl Ether. A final system, more similar to those of previous studies mentioned in the Introduction, was studied briefly to check for shear-induced aggregation observed in the results presented before. The liquid mixture consists of propylene glycol propyl ether and water (Cw), which have a LCST just above room temperature.26 This nicely complements the other systems, which were all partially miscible at room temperature. Two samples were prepared with a liquid composition on the water-poor side of the miscibility gap and heated to 50 °C where the liquids are partially miscible. The dispersions remain stable after being heated, and the colloids slowly sediment as evidenced by sharp sedimentation fronts (Figure 4b, left panel). Again, subjecting them to shear by vigorous shaking, shear-induced aggregation is observed (Figure 4b, right panel). It should be noted that temperatureinduced aggregation that was found in similar systems16 also occurs here, close to the miscibility gap.27 Tracking the Bulk Solvent Composition. Clearly shearinduced aggregation is a pronounced feature of the investigated systems. Aggregation caused by shear has been observed previously in various systems with quite different characteristics,28,29 the majority being destabilized and weakly stabilized suspensions.30,31 Compared to the wetting-induced aggregation observed in studies of similar systems,16 the aggregation observed here occurs relatively far away from the coexistence curve where the liquid composition is poor in Cw. In relation to 5766
dx.doi.org/10.1021/la501060w | Langmuir 2014, 30, 5763−5770
Langmuir
Article
Figure 5. (a) Fluorescence spectra of Nile Red in propanal/water mixtures at various water concentrations (wt %). (b) The spectra in (a) normalized to demonstrate the red-shift for increasingly polar mixtures. (c) Plots of all x̅ for various water concentrations, with linear fits shown as dashed lines that serve as a guide to the eye. (d) Plots of the amount of adsorbed water against particle concentration for the three different water concentrations (deduced from (c) using the calibration curve shown in Figure S6), where dashed lines are linear fits to the data with the intercept set to 0 wt % adsorbed water, and mainly serve as guide to the eye. (e) Estimates of the thickness of the adsorbed water layer plotted against the liquid composition, with an inset showing a sketch of the increase of the layer of adsorbed water (dark blue ring) around a particle (light blue), corresponding to these estimates and the measured size of the particles. The fit serves as a guide to the eye.
previous findings, this concentration regime corresponds most closely to cases where aggregation is caused by liquid bridges (rich in Cw) that form between particles through capillary condensation.6,7,32 If this were also to be the underlying cause for the shear-induced aggregation, analyzing the composition of the bulk solvent in aggregated samples would reveal a related decrease in the concentration of Cw with respect to the initial liquid mixture. For measuring changes in composition of the bulk solvent, the fluorescent dye Nile Red was employed as a composition probe. The fluorescence of this dye is known to strongly depend on the solvent and generally shifts to longer wavelengths in more polar chemical environments.33 Therefore, the aqueous systems are chemically suitable, while those partially miscible at room temperature are most convenient for measurements. Nile Red fluorescence spectra for varying water concentration show a red-shift and a decreasing fluorescence intensity for increasing water content, as demonstrated for propanal/water (Figure 5a). The red-shift in fluorescence spectra of Nile Red in propanal/water mixtures (Figure 5b)
ultimately appeared the best option, being sufficiently sensitive and not dependent on the dye concentration. To determine changes in the composition of the bulk solvent, fluorescence spectra of Nile Red in supernatants of sheared propanal/water samples at varying particle and water concentration were measured. Liquid compositions at 4, 8, and 12 wt % water (around the onset of aggregation) were chosen, where respectively, no, partial, and complete aggregation of the silica suspension was observed (Figure 2c). The peak positions x of these spectra, determined by fitting a part of the curves to an asymmetric Gaussian (Supporting Information, Figure S4), were used to track the composition. For each sample an averaged peak position x̅ was obtained from several measurements (Figure 5c). A detailed description of how this was done is given in the Supporting Information. Using a calibration curve (Figure S6), these were then converted into the amount of water adsorbed by the particles (Figure 5d) as well as the mass ratio of adsorbed water to particles (Figure S7). Some clear trends are observed. The amount of adsorbed water clearly increases with the particle concentration for all three liquid compositions (Figure 5d). Linear fits to the data 5767
dx.doi.org/10.1021/la501060w | Langmuir 2014, 30, 5763−5770
Langmuir
Article
particles were allowed to slowly settle to the bottom of the vial (Figure 2b, bottom panel), indirectly demonstrate that the influence of adsorbed water increases for increasing Cw concentration. It shows that, with increasing water concentration, liquid bridges start to form between particles within the dense sediment sticking them together. This is consistent with the assumption that the thickness of the adsorbed layers increases. The increasing fraction of aggregation, occuring at higher water concentration (8 and 12 wt % water, Figure 2b,c), was observed to be associated with an increased adsorption (Figure 5). This is most likely due to thicker adsorption layers on individual particles, but possibly also additional water separates out of the bulk solvent when liquid bridges form. The latter seems a realistic scenario; however, whether this occurs cannot clearly be deduced from the presented measurements. In either case, the measurements support the idea of aggregation via liquid bridges, as found previously in stationary samples of similar systems,7,16 but this is now brought about by shearing the samples. Thus, the act of vigorously shaking a stable dispersion of colloids carrying an adsorbed Cw-rich layer (Figure 6a) induces coalescence of the adsorbed layers and instantly flocculates the dispersion (Figure 6b). Apparently, by shearing (at a shear rate γ̇), the energy of particle collisions, Ucol(γ̇), can be increased sufficiently to overcome a barrier that otherwise prevents aggregation. This barrier originates from electrostatic repulsions between the charged particles. As the Cw concentration increases, the adsorbed layer grows and extends its reach further into the bulk solvent. It is also possible that the adsorbed layer reduces the range of the electrostatic repulsions. Overall, there is a balance between the reach of the adsorbed layer (rCw) and the effective range of the electrostatic repulsion (rU) (Figure 6c). There is an energy barrier U(rCw) for the adsorbed layers to meet, and thus a liquid bridge to form, set by the energy of the interparticle repulsion at rCw. The aggregation rate, analogous to the rate at which liquid bridges form, can then roughly be described as
obtained at the various liquid compositions with an intercept at 0 wt % reveal that there is at least nearly direct proportionality of the amount of adsorbed water to the particle concentration. This agrees with water being adsorbed on the particles. The total amount of adsorbed water increases with increasing water content (Figure 5d) and thus also with the amount of aggregation (as becomes clear from Figure 2c). The same trend can therefore be observed for the amount of water adsorbed by a fixed amount of particles (Figure S7). Additionally, corresponding to the amount of adsorbed water being close to directly proportional to particle concentration, this mass ratio of adsorbed water to particles remains roughly constant for increasing particle concentration. By taking the averages of the mass ratio of adsorbed water to particles for each water concentration, estimates34 of the thickness of the adsorbed water layers can be plotted (Figure 5e).
■
DISCUSSION The fluorescence measurements, which demonstrate the adsorption of Cw (water) on silica particles in propanal/water mixtures, support the idea that liquid bridges are the underlying cause of aggregation. These results together with the observation of the aggregation characteristics and previous reports allow a general picture of this type of system to be deduced. At low concentrations, in the absence of (shear-induced) aggregation, particles are already observed to adsorb Cw (4 wt % water, Figure 5d,e). This confirms that even in the stable suspensions obtained by sonication (Figure 2b, middle panel) water is adsorbed onto the particles, although it does not have any effect on the colloidal stability of the dispersion. For these samples, the state of the system can be sketched as in Figure 6a.
⎛ −U (rC ) ⎞ w ⎟⎟ k = N (φ , γ )̇ exp⎜⎜ ⎝ Ucol(γ )̇ ⎠
where N(φ,γ̇) is the frequency of attempt to aggregate, which will depend on the shear rate γ̇ and the particle concentration φ. Thus, once the adsorbed layer reaches far enough into the bulk solvent, the barrier U(rCw) decreases and becomes insufficient to prevent liquid bridges from forming. The situation is sketched in Figure 6c. Quantitatively, this will depend strongly on the surface chemistry and the properties of the liquid mixture. When increasing the Cw concentration, there is an onset of shear-induced aggregation with a clear regime of partial aggregation (Figures 2c and 3e). This would then be related to a higher barrier, resulting in fewer successful aggregation attempts. For propanal/water the fraction of aggregated colloids appears independent of particle concentration (Figure 2d), while in 2-butanol/water samples it increases with increasing particle concentration (Figure 3f). The difference in the dependence on φ could well be related to the proximity to the binodal where aggregation takes place. For propanal/ water, aggregation happens much closer to the miscibility gap, possibly related to better charge-stabilized particles in this system, as is suggested by the ζ-potential measurements
Figure 6. (a) Adsorbed layers of Cw (dark blue rings) are present at the surface of the particles for low Cw concentration approaching the miscibility gap. (b) Upon applying shear liquid bridges are formed between particles, causing instantaneous flocculation of the suspension. (c) Sketch of the situation at the particle surface, showing the adsorbed layer (Cw(ads)) and a sketch of the pair potential (U).
Another observation suggesting the adsorption of Cw on particles, far away from the binodal and well before the occurrence of shear-induced aggregation, is the PZC in 2butanol/water at low water concentrations (Figure 3c). Water apparently associates with charged groups on the surface and therefore already manages to reverse the charge at low concentrations. The adsorbed Cw-rich layers would be the precursors for the formation of liquid bridges between particles, causing them to aggregate. Since measurements were conducted after applying shear, the adsorbed water in the stable suspensions for higher Cw concentration was not directly measured. However, the increasingly solidlike sediments of stable suspensions in which 5768
dx.doi.org/10.1021/la501060w | Langmuir 2014, 30, 5763−5770
Langmuir
Article
bridges between particles. This happens upon approaching the miscibility gap of the liquid components, and particles can be redispersed by sonication, which apparently removes the liquid bridges by local remixing. The study has for the first time taken the control parameter to be composition rather than temperature. Additionally, several different liquid mixtures with varied characteristics were used, all different from the few liquid pairs that were used previously, emphasizing how generic the aggregation behavior is. The general picture sketched in previous reports shows a narrow region where temperature-induced aggregation happens, right next to the miscibility gap. This picture can now be extended to include a region where shear-induced aggregation is observed (Figure 7).
(Figure S1b and Figure 3c). Closer to the binodal the particle concentration will influence less the amount of aggregation, since plenty of Cw is present. After aggregation particle surfaces will, through the liquid bridges, still repel each other via electrostatic interactions, since they form charge-stabilized suspensions in both individual liquids. However, the liquid bridges form quite strong bonds between particles that are difficult to break. Sonication appears to achieve this, restoring the suspension to its stable form. This possibly also involves mixing some Cw back into the bulk solvent. The high frequency pressure waves involved in this method of agitating samples allows breaking of the bridges and intimate mixing of the components to go back to the initial situation of well-dispersed particles, carrying an adsorbed Cwrich layer (Figure 6a,b). Finally, we note it is likely that shear-induced aggregation was presumably not observed in previous studies16 on similar systems due to the experimental approach, where stationary samples were simply heated. In these cases, resulting sketches of the situation simply consisted of a narrow region next to the binodal to indicate where aggregation was observed. The shearinduced aggregation demonstrated above, however, occurs already much further away from the binodal. For a given system, the onset of this aggregation behavior also depends on the applied shear rate. Furthermore, adsorption of Cw onto particles was observed already before the onset of shearinduced aggregation. Thus, taking into account the observations presented here, the sketch of situation should be extended to include a region of shear-induced aggregation (Figure 7).
■
ASSOCIATED CONTENT
S Supporting Information *
Supporting experimental results and details on how data from fluorescence measurements were processed to extract quantitative data on water adsorption. This material is available free of charge via the Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Authors
*E-mail [email protected] (N.H.). *E-mail [email protected] (P.S.C.). Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS We thank Dr. Andy Schofield for providing silica particles and Dr. Jeroen van Duijneveldt for helpful suggestions. This work was supported by the Marie Curie Initial Training Network COMPLOIDS No. 234810 and EPSRC grant EP/J007404/1.
■
REFERENCES
(1) Zaccarelli, E. Colloidal gels: equilibrium and non-equilibrium routes. J. Phys.: Condens. Matter 2007, 19, 323101. (2) Campbell, A. I.; Anderson, V. J.; van Duijneveldt, J. S.; Bartlett, P. Dynamical arrest in attractive colloids: The effect of long-range repulsion. Phys. Rev. Lett. 2005, 94, 208301. (3) Wood, T. A.; Lintuvuori, J. S.; Schofield, A. B.; Marenduzzo, D.; Poon, W. C. K. A self-quenched defect glass in a colloid-nematic liquid crystal composite. Science 2011, 334, 79−83. (4) Chen, Q.; Bae, S. C.; Granick, S. Directed self-assembly of a colloidal kagome lattice. Nature 2011, 469, 381−384. (5) Beysens, D.; Esteve, D. Adsorption phenomena at the surface of silica spheres in a binary liquid mixture. Phys. Rev. Lett. 1985, 54, 2123−2126. (6) Hertlein, C.; Helden, L.; Gambassi, A.; Dietrich, S.; Bechinger, C. Direct measurement of critical Casimir forces. Nature 2008, 451, 172− 5. (7) Gambassi, A.; Maciolek, A.; Hertlein, C.; Nellen, U.; Helden, L.; Bechinger, C.; Dietrich, S. Critical Casimir effect in classical binary liquid mixtures. Phys. Rev. E 2009, 80, 061143. (8) Gögelein, C.; Brinkmann, M. Controlling the formation of capillary bridges in binary liquid mixtures. Langmuir 2010, 26, 17184− 17189. (9) Gurfein, V.; Beysens, D.; Perrot, F. Stability of colloids and wetting phenomena. Phys. Rev. A 1989, 40, 2543−2546. (10) Jayalakshmi, Y.; Kaler, E. W. Phase behavior of colloids in binary liquid mixtures. Phys. Rev. Lett. 1997, 78, 1379−1382. (11) Koehler, R. D.; Kaler, E. W. Colloidal phase transitions in aqueous nonionic surfactant solutions. Langmuir 1997, 13, 2463− 2470.
Figure 7. Schematic liquid−liquid phase diagram wherein areas bordered by dashed lines indicate where enough Cw is present to dress all particles with an adsorbed layer as well as where particles in stationary samples aggregate as reported by previous studies. A dotted line in between borders a light gray area where shear-induced aggregation can occur.
■
CONCLUSIONS We have presented a study of the colloidal stability of particles suspended in partially miscible binary liquid mixtures. More specifically, when the liquid mixture is phase-separated, in all cases there is complete wetting of the colloidal particles by one of the liquid phases. The strong “attraction” of Cw, the main component of this phase, toward the surface of the colloidal particles can result in aggregation. When Cw only makes up a small fraction of the liquid mixture, it adsorbs onto the suspended colloidal particles, as shown by fluorescence of a dye that acts as composition probe. If there is sufficient adsorption on the colloidal particles, shear can make the adsorbed layers coalesce, resulting in shear-induced aggregation via liquid 5769
dx.doi.org/10.1021/la501060w | Langmuir 2014, 30, 5763−5770
Langmuir
Article
(12) Petit, J.-M.; Law, B.; Beysens, D. Adsorption-induced aggregation of colloidal particles in binary mixtures: Modeling the pair free energy. J. Colloid Interface Sci. 1998, 449, 441−449. (13) Law, B.; Petit, J.-M.; Beysens, D. Adsorption-induced reversible colloidal aggregation. Phys. Rev. E 1998, 57, 5782−5794. (14) Vincent, B.; Király, Z.; Emmett, S.; Beaver, A. The stability of silica dispersions in ethanol/cyclohexane mixtures. Colloids Surf. 1990, 49, 121−132. (15) Kline, S. R.; Kaler, E. W. Colloidal interactions in water/2butoxyethanol solvents. Langmuir 1994, 10, 412−417. (16) Beysens, D.; Narayanan, T. Wetting-induced aggregation of colloids. J. Stat. Phys. 1999, 95, 997−1008. (17) Guo, H.; Narayanan, T.; Sztuchi, M.; Schall, P.; Wegdam, G. H. Reversible phase transition of colloids in a binary liquid solvent. Phys. Rev. Lett. 2008, 100, 188303. (18) Gallagher, P. D.; Kurnaz, M. L.; Maher, J. V. Aggregation in polystyrene-sphere suspensions in near-critical binary liquid mixtures. Phys. Rev. A 1992, 46, 7750−7755. (19) Gallagher, P. D.; Maher, J. V. Partitioning of polystyrene latex spheres in immiscible critical liquid mixtures. Phys. Rev. A 1992, 46, 2012−2021. (20) Rathke, B.; Grüll, H.; Woermann, D. Stability of dilute colloidal silica suspensions in the vicinity of the binodal curve of the system 2butoxyethanol/water. J. Colloid Interface Sci. 1997, 192, 334−337. (21) Stö ber, W.; Fink, A.; Bohn, E. Controlled growth of monodisperse silica spheres in the micron size range. J. Colloid Interface Sci. 1968, 26, 62−69. (22) van Blaaderen, A.; Vrij, A. Synthesis and characterization of colloidal dispersions of fluorescent, monodisperse silica spheres. Langmuir 1992, 8, 2921−2931. (23) Sørenson, J.; Arlt, W. In Liquid-Liquid Equilibrium Data Collection; Behrens, D., Eckermann, R., Eds.; Dechema: Frankfurt am Main, 1979. (24) Stephenson, R.; Stuart, J. Mutual binary solubilities: wateralcohols and water-esters. J. Chem. Eng. Data 1986, 31, 56−70. (25) Ethylene glycol is slightly less dense than nitromethane; hence, in phase-separated samples, some of the colloid-containing phase is rising to the top of the sample, making the top phase appear turbid. (26) Bauduin, P.; Wattebled, L.; Schrödle, S.; Touraud, D.; Kunz, W. Temperature dependence of industrial propylene glycol alkyl ether/ water mixtures. J. Mol. Liq. 2004, 115, 23−28. (27) Hijnen, N.; Clegg, P. S. Percolating networks of colloidal rods assembled by a liquid-liquid phase separation. In preparation, 2014. (28) Guery, J.; Bertrand, E.; Rouzeau, C.; Levitz, P.; Weitz, D. A.; Bibette, J. Irreversible shear-activated aggregation in non-Brownian suspensions. Phys. Rev. Lett. 2006, 96, 198301. (29) Lips, A.; Westbury, T.; Hart, P. M.; Evans, I. D.; Campbell, I. J. In Food Colloids and Polymers: Stability and Mechanical Properties; Dickinson, E., Walstra, P., Eds.; RSC Special Publications: London, 1993; Vol. 113; Chapter on the physics of shear-induced aggregation in concentrated food emulsions, pp 31−44. (30) Le Berre, F.; Chauveteau, G.; Pefferkorn, E. Perikinetic and orthokinetic aggregation of hydrated colloids. J. Colloid Interface Sci. 1998, 199, 1−12. (31) Husband, J. C.; Adams, J. M. Shear-induced aggregation of carboxylated polymer latices. Colloid Polym. Sci. 1992, 270, 1194− 1200. (32) Dobbs, H. T.; Darbellay, G. A.; Yeomans, J. M. Capillary condensation between spheres. Europhys. Lett. 1992, 18, 439−444. (33) Greenspan, P.; Mayer, E. P.; Fowler, S. D. Nile red: a selective fluorescent stain for intracellular lipid droplets. J. Cell Biol. 1985, 100, 965−73. (34) Densities were taken to be 1.8 g/cm3 for the particles and 1 g/ cm3 for the adsorbed water layer, and the particle radius (see Experimental Section) was assumed to be 0.38 μm.
5770
dx.doi.org/10.1021/la501060w | Langmuir 2014, 30, 5763−5770
Colloidal Aggregation in Mixtures of Partially Miscible Liquids by Shear-Induced Capillary Bridges Niek Hijnen* and Paul S. Clegg* School of Physics and Astronomy, The University of Edinburgh, Mayfield Road, Edinburgh EH9 3JZ, United Kingdom S Supporting Information *
ABSTRACT: We have studied shear-induced aggregation of colloidal silica particles suspended in a variety of partially miscible liquid mixtures. The shared characteristic of the investigated systems is that after liquid−liquid phase separation of the binary liquid mixtures one phase completely wets the particles. We have explored compositions where there are insufficient quantities of the particle wetting component to induce phase separation. As the proportion of the wetting component is increased, we find a significant concentration range where shear-induced aggregation takes place. The macroscopic characteristics of this phenomenon are demonstrated, for which observations were greatly facilitated by mostly using liquid pairs partially miscible at room temperature. Measurements revealing the adsorption of the minority component to colloidal particles show that capillary condensation between particles causes the observed aggregation. The likely microscopic features underlying this aggregation behavior are then discussed. Finally, the overall picture of these systems is sketched as a nonequilibrium liquid−liquid phase diagram, in which outside the binodal there is a region of shear-induced aggregation.
■
INTRODUCTION Stable colloidal suspensions rely on repulsive interactions between constituent particles to prevent them from flocculating. In contrast, when some attractive interparticle interaction dominates, colloids can assemble (aggregate) into larger structures to result in a material with potentially useful properties.1−4 Being able to tune colloidal interactions in a straightforward way is therefore very attractive. Interest in using partially miscible liquid mixtures to control interparticle interactions was initiated by observations of temperature-reversible wetting-induced colloidal aggregation in such a solvent.5 An important characteristic of the studied system was that after the mixture phase separated there was complete wetting of the particles by one of the phases. Adjacent to the miscibility gap of the pure liquid mixture, a narrow region was observed where aggregation of suspended particles occurs. Subsequent studies have shown that in the vicinity of the critical point the aggregation is caused by the confinement of concentration fluctuations in the solvent between particles.6,7 Further away from the critical composition, where the component that completely wets the particles is in the minority, a different mechanism is responsible for the aggregation. Here liquid bridges form between particles due to the coalescence of wetting layers present on particles.6−9 Alternatively, solid particles in partially miscible liquid mixtures have also been interpreted as true ternary mixtures.10 The aggregation in these systems can be temperaturereversible provided that particle surfaces remain repulsive, hence preventing aggregation into the primary minimum of their interaction potential. This means particles can be redispersed upon changing the temperature, removing the critical fluctuations or the liquid bridges by remixing. In © 2014 American Chemical Society
aggregates, large distances separating particles, and even particles arranging themselves in crystalline structures, have indeed been observed. 9,11 It has also been proposed theoretically that crystallization is possible for one of the investigated systems in the case where a liquid bridge forms from adsorbed layers.12,13 Moreover, the interparticle interactions will not only be affected by adsorption but also in general depend on liquid composition as appears already from DLVO theory.14,15 So far this aggregation behavior is observed and studied in several systems.16 The two types of binary liquid mixtures that have mostly been used in experiments are water and pyridine derivatives (2,6-lutidine and 3-methylpiridine)5,7,17−19 and water and n-alkyl polyglycol ethers (nonionic surfactants)10,11,20 which both have a LCST above 25 °C. In this case, to approach the miscibility gap, samples of predefined compositions can be prepared at room temperature and then heated in a controlled manner. In addition to these liquids, one report also uses water and isobutyric acid as a qualitative comparison.19 In this article we present a study of a remarkable colloidal aggregation process in several similar systems using a distinctively different approach. The colloidal particles in our samples are charge-stabilized in both individual liquid components. As in the reports mentioned above, after phase separation of the liquid mixture, they are located exclusively in one of the liquid phases. The main component of this phase, which completely wets the particles, is hereafter referred to as Received: March 21, 2014 Revised: April 30, 2014 Published: May 6, 2014 5763
dx.doi.org/10.1021/la501060w | Langmuir 2014, 30, 5763−5770
Langmuir
Article
Preparation and Handling of Samples. Silica particles were washed several times with water (basic pH) after synthesis and then stored in ethanol. Particles were dried at 100 °C (Binder VD23 oven) for 30 min to let bulk solvent evaporate. Then, the resulting cake was ground into a fine powder using a spatula, and subsequently this was dried at 100 °C under vacuum for 60 min. The preparation of samples depended slightly on the required liquid composition. Samples with a liquid composition for which the components were completely mixed at room temperature were prepared by dispersing the colloids directly into the liquid mixture using an ultrasonic bath (VWR). When preparing a sample with a liquid composition for which the liquids were phase-separated at room temperature, the particles were first dispersed into the majority component before adding the minority component that preferentially wets the particles. Samples were photographed in a thermostated (21 ± 1 °C) room using an SLR camera (Canon EOS 40D with a Tamron 70−300 mm Tele-Macro lens), controlled by a computer using the canon EOS utility software. Time-lapse photography was used to monitor sedimentation. Samples were studied before and after applying shear by 30 s of vigorous shaking using a vortex mixer. For determining the fraction of shear-aggregated colloids, samples were left for 10 min after vortex mixing, allowing aggregates to settle to the bottom of the vial. A known amount (by weight) of supernatant was then transferred into a weighed vial, and the particles were spun down using a centrifuge, after which a clear supernatant was removed. Remaining solvent was subsequently removed by drying in an oven (100 °C, 15 min), and finally, the mass fraction of colloids in the supernatant was determined after weighing the vial with particles. This was then related to the initial sample composition to obtain the fraction of shear-aggregated colloids. ζ-Potential Measurements. Samples of, typically, a few milliliters were prepared by dispersing the colloids (0.2 vol %) directly into mixtures of various liquid compositions. Measuring the ζ-potential was done using a Malvern Zeta Sizer Nano Z. For the measurements, 2butanol/water samples were transferred into disposable folded capillary cells, while for propanal/water samples a dip cell was used in combination with a glass square cuvette. Solvent properties of the majority solvent were used to extract the ζ-potential from the electrophoretic mobility. Fluorencence Spectroscopy. In order to be able to measure fluorescence spectra of Nile Red in the bulk propanal/water solvent of dispersions, first stock mixtures at 4, 8, and 12 wt % water (∼25 mL)
the colloid-wetting component Cw. The approach adopted in this work differs from those of previous studies in the use of liquid composition instead of temperature as the adjustable parameter to approach the miscibility gap. To this end, liquid pairs were used that have not yet been studied in this context and which are (with one exception) only partially miscible at room temperature. This conveniently allows the binodal to be approached by increasing the concentration of Cw in the liquid mixture, as indicated in Figure 1.
Figure 1. Schematic liquid−liquid phase diagram with an arrow indicating samples which are probed in this experimental approach.
Below, first the experimental methods are detailed, followed by an overview and discussion of the results.
■
EXPERIMENTAL SECTION
Materials. Propanal (propionaldehyde, ≥97%, FG, Kosher, SigmaAldrich), 2-butanol (99%, Fisher Scientific), nitromethane (99+%, stored under nitrogen, Acros Organics), ethylene glycol (anhydrous, 99.8%, Sigma-Aldrich), propylene glycol propyl ether (PGPE, 1propoxy-2-propanol, 99%, Sigma-Aldrich), and Nile Red (Technical grade, Sigma-Aldrich) were used as received. Particles used were silica spheres prepared by the Stöber method.21,22 Nonfluorescent particles were used with the water/propanal and water/2-butanol (from TEM images, the average particle radius RTEM = 0.38 μm with polydispersity σ = 7%), and FITC-labeled particles were used with nitromethane/ ethylene glycol. Particles used with water/PGPE were the same FITC labeled particles but coated with unlabeled Stöber silica (RTEM = 0.35 μm with σ = 12%).
Figure 2. (a) Phase diagram of water and propanal23 with an inset demonstrating partitioning of particles (1 vol %) into the water-rich phase (liquid mixture: 30 wt % water). (b) Vials containing suspensions of silica particles (1 vol %) in propanal/water mixtures at various liquid compositions (given in wt % of Cw in the liquid mixture): initially stable suspensions (middle frame) and the resulting state of the same samples after vortex mixing (top frame) and tilting them backward after particle sediments have formed over ∼24 h under gravity (bottom frame). (c) Estimates of the fraction of aggregated particles by measuring the concentration in the supernatants of the samples shown in the top frame of (b). (d) Estimates of the fraction of aggregated particles plotted as a function of particle concentration for intermediate water concentrations. 5764
dx.doi.org/10.1021/la501060w | Langmuir 2014, 30, 5763−5770
Langmuir
Article
Figure 3. (a) Phase diagram of 2-butanol and water24 with an inset demonstrating partitioning of particles into the water-rich phase. (b) Vials containing silica particles (1 vol %) dispersed into 2-butanol/water mixtures at various liquid compositions (given in wt % of Cw in the liquid mixture) using sonication, and (c) a plot of ζ-potentials at various water concentrations. (d) The same samples as shown in (b), but now the suspensions were shaken vigorously using a vortex mixer, and (e) the fraction of aggregated particles for the relevant samples is plotted as a function of water concentration. (f) Shows the fraction of aggregated particles plotted against the total particle volume fraction for several water concentrations. were prepared containing 1 μM Nile Red. Subsequently, particles were dispersed at varying concentration in the stock mixtures to obtain samples of ∼5 mL. After shearing the samples for 30 s using a vortex mixer, particles were spun down and the clear supernatant transferred into a square cuvette (10 mm path length, optical glass, from Starna Scientific) for the measurements. The remaining stock mixtures were used as blanks and measured in between supernatant samples of the dispersions. For measuring the spectra a Varian Cary Eclipse fluorescence spectrophotometer was used, and for every measurement multiple (typically five) spectra were recorded. Processing and peakfitting were done in Origin 8.5 using batch processing, outputting the relevant fit parameters and measures of statistical accuracy.
suspensions were obtained up to 16 wt % water, with the subsequent sample at 18 wt % water being phase-separated. The particles are charge-stabilized and carry a negative surface charge (ζ-potential ≈ −50 mV, see Figure S1b). As expected from their size and density mismatch with the solvent, they were observed to sediment slowly to the bottom of the vial, where eventually all particles are collected within a concentrated layer. While this behavior is consistent over the entire range of samples, the response of the formed sediments to gravity is not. In the absence of water and at very low water concentrations, the dense sedimented layer of charge-stabilized particles flows in response to gravity when tilting the vial backward (Figure 2b, bottom frame). Simply shaking the vials redisperses the particles to again obtain a stable suspension. As the water concentration is increased, a more solidlike “cake” is formed, which does not flow under gravity. In this case, particles cannot be redispersed properly by shaking, and aggregates remain as evidenced by rapid sedimentation into voluminous sediments. Curiously, this indicates that within the concentrated layer the particles have aggregated, as opposed to remaining wellstabilized against aggregation due to their charged surfaces. Another surprising result is obtained when the stable suspensions shown in Figure 2b are vigorously shaken, instead of being left to sediment. There is no effect on colloidal stability for suspensions at low water concentration (Figure 2b, top frame), as can be expected with the particles being chargestabilized. However, at higher concentration of the Cw, the particles in the initially stable dispersions (Figure 2b, middle frame) aggregate and quickly settle into voluminous sediments (Figure 2b, top frame). The supernatants above sediments clearly stay turbid for immediate concentrations (8 and 10 wt %). Plotting the fraction of aggregated particles as a function of the Cw concentration shows a relatively sharp increase between 6 and 10 wt % water (Figure 2c).
■
RESULTS First, the general behavior of suspensions with liquid compositions relatively poor in the Cw is presented for the various systems studied. Those investigated more extensively are aqueous mixtures and will be presented first. For completeness, they are supplemented by a nonaqueous liquid mixture and an aqueous mixture that is fully miscible at room temperature, similar to that used in previous studies.5,10,18 Macroscopic Observations. Water and Propanal. Water and propanal are partially miscible at room temperature (Figure 2a). In the left-hand vial of the inset in Figure 2a a dispersion of silica particles in propanal is added on top (lower density) of some water; the latter makes up 30 wt % of the total liquid in the vial. At this composition the liquids phase-separate at room temperature, and after vigorously shaking the sample all the particles transfer into the bottom (since higher density), waterrich phase, indicating that water in this case is the Cw. Alternatively, this is confirmed by observing the colloids locating themselves into the water-rich phase after a temperature quench leading to phase separation of the liquids (Figure S1a). A series of suspensions with increasing water concentration were prepared, using sonication to disperse the particles into the liquid mixtures (Figure 2b, middle frame). Stable 5765
dx.doi.org/10.1021/la501060w | Langmuir 2014, 30, 5763−5770
Langmuir
Article
Thus, it appears that for intermediate Cw concentration only a fraction of particles has aggregated. To further explore this regime, the particle concentration was varied for the range of water concentrations 6−10 wt %. The fraction of aggregated particles appears to be independent of the particle concentration, as demonstrated in Figure 2d. Only at 8 wt % water the fraction of aggregated particles is observed to suddenly drop again for the highest particle concentration. Water and 2-Butanol. The phase diagram of 2-butanol and water shows that these liquids are also partially miscible at room temperature (Figure 3a). Silica particles present in such mixtures are located in the water-rich phase after liquid−liquid phase separation upon heating or mixing, in an analogous way to the propanal/water system (inset of Figure 3a and Figure S2). Hence, in this system water is again the Cw. In the 2-butanol/water system, approaching the miscibility gap by increasing the Cw concentration, an onset of colloidal aggregation is already observed at relatively low water concentrations (Figure 3b). Aggregation vanishes again at higher water concentration. These observations are consistent with ζ-potentials, which show that when increasing the water concentration the colloids go through a point of zero charge (PZC) (Figure 3c). In pure 2-butanol the silica particles are positively charged, and since they are of opposite charge in water, adding water eventually changes the sign of the total particle charge. Closer to the binary liquid’s binodal, aggregation is seen again. The dispersion with a liquid composition where the two components only just mix (35 wt % water) is unstable and the particles aggregate to form a voluminous sediment (Figure 3b). At somewhat lower Cw concentration, stable dispersions can be obtained (Figure 3b), but shear-induced aggregation is observed (Figure 3d,e) after vigorous shaking. In contrast to the propanal/water system, the fraction of aggregated colloids increases with particle concentration (Figure 3f). Taking into account the amount of Cw available per particle by rescaling the data in Figures 2d and 3f did not provide any other insight. Ethylene Glycol and Nitromethane. The nonaqueous liquid mixture included in this study consists of ethylene glycol (Cw) and nitromethane. FITC-labeled silica particles dispersed in nitromethane completely transfer into the ethylene glycol-rich phase after mixing with an appropriate amount of ethylene glycol (far right vial in Figure 4a25). These liquids are also partially miscible at room temperature, but with a fairly low solubility of the Cw (∼8 wt %) and a low UCST (∼40 °C).23 Using sonication, stable suspensions were obtained for ethylene glycol concentrations running up to maximum solubility of the Cw at room temperature, while crossing into the miscibility gap results in phase-separated samples (Figure 4a, top). As in the propanal/water system, the stable suspensions slowly sediment into compact concentrated layers, which upon adding some Cw appear more solidlike (i.e., do not flow under gravity when tilting the vials). Vigorous shaking of some of these samples results in shear-induced aggregation, similar to the previously discussed systems (Figure 4a, bottom). At 7 wt % ethylene glycol the situation appears to be different to its neighbors on both sides. No clear supernatant and sharp sedimentation front is observed, 3 h after dispersal by sonication. After some time the supernatant clears up, and the sedimentation front is visible as in the samples with lower Cw concentration. Once subjected to shear, a compact and very viscous phase is collected at the bottom of the vial, much like what is observed in the phase-separated 8 wt % sample. It was
Figure 4. (a) Suspensions of silica particles (1 vol %) in various mixtures of ethylene glycol (concentrations given in wt %) and nitromethane, which were initially mixed and dispersed by sonication and subsequently either (top frame) left undisturbed to sediment for 3 h, or (bottom frame) first vigorously shaken and then left undisturbed to sediment for 3 h. (b, left frame) Two suspensions of silica particles (1 vol %) in different mixtures of water (concentrations given in wt %) and propylene glycol propyl ether heated from room temperature to 50 °C in a water bath and left to sediment for 1 h. (b, right frame) The same heated samples, 1 h after being vortex mixed.
found that some Cw-rich phase separates out of the bulk solvent, but that not many of the present particles are collected within it. The ethylene glycol-rich droplets then simply float to the top of the sample. Experiments demonstrating this behavior for a range of liquid compositions are presented in Figure S3. Water and Propylene Glycol Propyl Ether. A final system, more similar to those of previous studies mentioned in the Introduction, was studied briefly to check for shear-induced aggregation observed in the results presented before. The liquid mixture consists of propylene glycol propyl ether and water (Cw), which have a LCST just above room temperature.26 This nicely complements the other systems, which were all partially miscible at room temperature. Two samples were prepared with a liquid composition on the water-poor side of the miscibility gap and heated to 50 °C where the liquids are partially miscible. The dispersions remain stable after being heated, and the colloids slowly sediment as evidenced by sharp sedimentation fronts (Figure 4b, left panel). Again, subjecting them to shear by vigorous shaking, shear-induced aggregation is observed (Figure 4b, right panel). It should be noted that temperatureinduced aggregation that was found in similar systems16 also occurs here, close to the miscibility gap.27 Tracking the Bulk Solvent Composition. Clearly shearinduced aggregation is a pronounced feature of the investigated systems. Aggregation caused by shear has been observed previously in various systems with quite different characteristics,28,29 the majority being destabilized and weakly stabilized suspensions.30,31 Compared to the wetting-induced aggregation observed in studies of similar systems,16 the aggregation observed here occurs relatively far away from the coexistence curve where the liquid composition is poor in Cw. In relation to 5766
dx.doi.org/10.1021/la501060w | Langmuir 2014, 30, 5763−5770
Langmuir
Article
Figure 5. (a) Fluorescence spectra of Nile Red in propanal/water mixtures at various water concentrations (wt %). (b) The spectra in (a) normalized to demonstrate the red-shift for increasingly polar mixtures. (c) Plots of all x̅ for various water concentrations, with linear fits shown as dashed lines that serve as a guide to the eye. (d) Plots of the amount of adsorbed water against particle concentration for the three different water concentrations (deduced from (c) using the calibration curve shown in Figure S6), where dashed lines are linear fits to the data with the intercept set to 0 wt % adsorbed water, and mainly serve as guide to the eye. (e) Estimates of the thickness of the adsorbed water layer plotted against the liquid composition, with an inset showing a sketch of the increase of the layer of adsorbed water (dark blue ring) around a particle (light blue), corresponding to these estimates and the measured size of the particles. The fit serves as a guide to the eye.
previous findings, this concentration regime corresponds most closely to cases where aggregation is caused by liquid bridges (rich in Cw) that form between particles through capillary condensation.6,7,32 If this were also to be the underlying cause for the shear-induced aggregation, analyzing the composition of the bulk solvent in aggregated samples would reveal a related decrease in the concentration of Cw with respect to the initial liquid mixture. For measuring changes in composition of the bulk solvent, the fluorescent dye Nile Red was employed as a composition probe. The fluorescence of this dye is known to strongly depend on the solvent and generally shifts to longer wavelengths in more polar chemical environments.33 Therefore, the aqueous systems are chemically suitable, while those partially miscible at room temperature are most convenient for measurements. Nile Red fluorescence spectra for varying water concentration show a red-shift and a decreasing fluorescence intensity for increasing water content, as demonstrated for propanal/water (Figure 5a). The red-shift in fluorescence spectra of Nile Red in propanal/water mixtures (Figure 5b)
ultimately appeared the best option, being sufficiently sensitive and not dependent on the dye concentration. To determine changes in the composition of the bulk solvent, fluorescence spectra of Nile Red in supernatants of sheared propanal/water samples at varying particle and water concentration were measured. Liquid compositions at 4, 8, and 12 wt % water (around the onset of aggregation) were chosen, where respectively, no, partial, and complete aggregation of the silica suspension was observed (Figure 2c). The peak positions x of these spectra, determined by fitting a part of the curves to an asymmetric Gaussian (Supporting Information, Figure S4), were used to track the composition. For each sample an averaged peak position x̅ was obtained from several measurements (Figure 5c). A detailed description of how this was done is given in the Supporting Information. Using a calibration curve (Figure S6), these were then converted into the amount of water adsorbed by the particles (Figure 5d) as well as the mass ratio of adsorbed water to particles (Figure S7). Some clear trends are observed. The amount of adsorbed water clearly increases with the particle concentration for all three liquid compositions (Figure 5d). Linear fits to the data 5767
dx.doi.org/10.1021/la501060w | Langmuir 2014, 30, 5763−5770
Langmuir
Article
particles were allowed to slowly settle to the bottom of the vial (Figure 2b, bottom panel), indirectly demonstrate that the influence of adsorbed water increases for increasing Cw concentration. It shows that, with increasing water concentration, liquid bridges start to form between particles within the dense sediment sticking them together. This is consistent with the assumption that the thickness of the adsorbed layers increases. The increasing fraction of aggregation, occuring at higher water concentration (8 and 12 wt % water, Figure 2b,c), was observed to be associated with an increased adsorption (Figure 5). This is most likely due to thicker adsorption layers on individual particles, but possibly also additional water separates out of the bulk solvent when liquid bridges form. The latter seems a realistic scenario; however, whether this occurs cannot clearly be deduced from the presented measurements. In either case, the measurements support the idea of aggregation via liquid bridges, as found previously in stationary samples of similar systems,7,16 but this is now brought about by shearing the samples. Thus, the act of vigorously shaking a stable dispersion of colloids carrying an adsorbed Cw-rich layer (Figure 6a) induces coalescence of the adsorbed layers and instantly flocculates the dispersion (Figure 6b). Apparently, by shearing (at a shear rate γ̇), the energy of particle collisions, Ucol(γ̇), can be increased sufficiently to overcome a barrier that otherwise prevents aggregation. This barrier originates from electrostatic repulsions between the charged particles. As the Cw concentration increases, the adsorbed layer grows and extends its reach further into the bulk solvent. It is also possible that the adsorbed layer reduces the range of the electrostatic repulsions. Overall, there is a balance between the reach of the adsorbed layer (rCw) and the effective range of the electrostatic repulsion (rU) (Figure 6c). There is an energy barrier U(rCw) for the adsorbed layers to meet, and thus a liquid bridge to form, set by the energy of the interparticle repulsion at rCw. The aggregation rate, analogous to the rate at which liquid bridges form, can then roughly be described as
obtained at the various liquid compositions with an intercept at 0 wt % reveal that there is at least nearly direct proportionality of the amount of adsorbed water to the particle concentration. This agrees with water being adsorbed on the particles. The total amount of adsorbed water increases with increasing water content (Figure 5d) and thus also with the amount of aggregation (as becomes clear from Figure 2c). The same trend can therefore be observed for the amount of water adsorbed by a fixed amount of particles (Figure S7). Additionally, corresponding to the amount of adsorbed water being close to directly proportional to particle concentration, this mass ratio of adsorbed water to particles remains roughly constant for increasing particle concentration. By taking the averages of the mass ratio of adsorbed water to particles for each water concentration, estimates34 of the thickness of the adsorbed water layers can be plotted (Figure 5e).
■
DISCUSSION The fluorescence measurements, which demonstrate the adsorption of Cw (water) on silica particles in propanal/water mixtures, support the idea that liquid bridges are the underlying cause of aggregation. These results together with the observation of the aggregation characteristics and previous reports allow a general picture of this type of system to be deduced. At low concentrations, in the absence of (shear-induced) aggregation, particles are already observed to adsorb Cw (4 wt % water, Figure 5d,e). This confirms that even in the stable suspensions obtained by sonication (Figure 2b, middle panel) water is adsorbed onto the particles, although it does not have any effect on the colloidal stability of the dispersion. For these samples, the state of the system can be sketched as in Figure 6a.
⎛ −U (rC ) ⎞ w ⎟⎟ k = N (φ , γ )̇ exp⎜⎜ ⎝ Ucol(γ )̇ ⎠
where N(φ,γ̇) is the frequency of attempt to aggregate, which will depend on the shear rate γ̇ and the particle concentration φ. Thus, once the adsorbed layer reaches far enough into the bulk solvent, the barrier U(rCw) decreases and becomes insufficient to prevent liquid bridges from forming. The situation is sketched in Figure 6c. Quantitatively, this will depend strongly on the surface chemistry and the properties of the liquid mixture. When increasing the Cw concentration, there is an onset of shear-induced aggregation with a clear regime of partial aggregation (Figures 2c and 3e). This would then be related to a higher barrier, resulting in fewer successful aggregation attempts. For propanal/water the fraction of aggregated colloids appears independent of particle concentration (Figure 2d), while in 2-butanol/water samples it increases with increasing particle concentration (Figure 3f). The difference in the dependence on φ could well be related to the proximity to the binodal where aggregation takes place. For propanal/ water, aggregation happens much closer to the miscibility gap, possibly related to better charge-stabilized particles in this system, as is suggested by the ζ-potential measurements
Figure 6. (a) Adsorbed layers of Cw (dark blue rings) are present at the surface of the particles for low Cw concentration approaching the miscibility gap. (b) Upon applying shear liquid bridges are formed between particles, causing instantaneous flocculation of the suspension. (c) Sketch of the situation at the particle surface, showing the adsorbed layer (Cw(ads)) and a sketch of the pair potential (U).
Another observation suggesting the adsorption of Cw on particles, far away from the binodal and well before the occurrence of shear-induced aggregation, is the PZC in 2butanol/water at low water concentrations (Figure 3c). Water apparently associates with charged groups on the surface and therefore already manages to reverse the charge at low concentrations. The adsorbed Cw-rich layers would be the precursors for the formation of liquid bridges between particles, causing them to aggregate. Since measurements were conducted after applying shear, the adsorbed water in the stable suspensions for higher Cw concentration was not directly measured. However, the increasingly solidlike sediments of stable suspensions in which 5768
dx.doi.org/10.1021/la501060w | Langmuir 2014, 30, 5763−5770
Langmuir
Article
bridges between particles. This happens upon approaching the miscibility gap of the liquid components, and particles can be redispersed by sonication, which apparently removes the liquid bridges by local remixing. The study has for the first time taken the control parameter to be composition rather than temperature. Additionally, several different liquid mixtures with varied characteristics were used, all different from the few liquid pairs that were used previously, emphasizing how generic the aggregation behavior is. The general picture sketched in previous reports shows a narrow region where temperature-induced aggregation happens, right next to the miscibility gap. This picture can now be extended to include a region where shear-induced aggregation is observed (Figure 7).
(Figure S1b and Figure 3c). Closer to the binodal the particle concentration will influence less the amount of aggregation, since plenty of Cw is present. After aggregation particle surfaces will, through the liquid bridges, still repel each other via electrostatic interactions, since they form charge-stabilized suspensions in both individual liquids. However, the liquid bridges form quite strong bonds between particles that are difficult to break. Sonication appears to achieve this, restoring the suspension to its stable form. This possibly also involves mixing some Cw back into the bulk solvent. The high frequency pressure waves involved in this method of agitating samples allows breaking of the bridges and intimate mixing of the components to go back to the initial situation of well-dispersed particles, carrying an adsorbed Cwrich layer (Figure 6a,b). Finally, we note it is likely that shear-induced aggregation was presumably not observed in previous studies16 on similar systems due to the experimental approach, where stationary samples were simply heated. In these cases, resulting sketches of the situation simply consisted of a narrow region next to the binodal to indicate where aggregation was observed. The shearinduced aggregation demonstrated above, however, occurs already much further away from the binodal. For a given system, the onset of this aggregation behavior also depends on the applied shear rate. Furthermore, adsorption of Cw onto particles was observed already before the onset of shearinduced aggregation. Thus, taking into account the observations presented here, the sketch of situation should be extended to include a region of shear-induced aggregation (Figure 7).
■
ASSOCIATED CONTENT
S Supporting Information *
Supporting experimental results and details on how data from fluorescence measurements were processed to extract quantitative data on water adsorption. This material is available free of charge via the Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Authors
*E-mail [email protected] (N.H.). *E-mail [email protected] (P.S.C.). Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS We thank Dr. Andy Schofield for providing silica particles and Dr. Jeroen van Duijneveldt for helpful suggestions. This work was supported by the Marie Curie Initial Training Network COMPLOIDS No. 234810 and EPSRC grant EP/J007404/1.
■
REFERENCES
(1) Zaccarelli, E. Colloidal gels: equilibrium and non-equilibrium routes. J. Phys.: Condens. Matter 2007, 19, 323101. (2) Campbell, A. I.; Anderson, V. J.; van Duijneveldt, J. S.; Bartlett, P. Dynamical arrest in attractive colloids: The effect of long-range repulsion. Phys. Rev. Lett. 2005, 94, 208301. (3) Wood, T. A.; Lintuvuori, J. S.; Schofield, A. B.; Marenduzzo, D.; Poon, W. C. K. A self-quenched defect glass in a colloid-nematic liquid crystal composite. Science 2011, 334, 79−83. (4) Chen, Q.; Bae, S. C.; Granick, S. Directed self-assembly of a colloidal kagome lattice. Nature 2011, 469, 381−384. (5) Beysens, D.; Esteve, D. Adsorption phenomena at the surface of silica spheres in a binary liquid mixture. Phys. Rev. Lett. 1985, 54, 2123−2126. (6) Hertlein, C.; Helden, L.; Gambassi, A.; Dietrich, S.; Bechinger, C. Direct measurement of critical Casimir forces. Nature 2008, 451, 172− 5. (7) Gambassi, A.; Maciolek, A.; Hertlein, C.; Nellen, U.; Helden, L.; Bechinger, C.; Dietrich, S. Critical Casimir effect in classical binary liquid mixtures. Phys. Rev. E 2009, 80, 061143. (8) Gögelein, C.; Brinkmann, M. Controlling the formation of capillary bridges in binary liquid mixtures. Langmuir 2010, 26, 17184− 17189. (9) Gurfein, V.; Beysens, D.; Perrot, F. Stability of colloids and wetting phenomena. Phys. Rev. A 1989, 40, 2543−2546. (10) Jayalakshmi, Y.; Kaler, E. W. Phase behavior of colloids in binary liquid mixtures. Phys. Rev. Lett. 1997, 78, 1379−1382. (11) Koehler, R. D.; Kaler, E. W. Colloidal phase transitions in aqueous nonionic surfactant solutions. Langmuir 1997, 13, 2463− 2470.
Figure 7. Schematic liquid−liquid phase diagram wherein areas bordered by dashed lines indicate where enough Cw is present to dress all particles with an adsorbed layer as well as where particles in stationary samples aggregate as reported by previous studies. A dotted line in between borders a light gray area where shear-induced aggregation can occur.
■
CONCLUSIONS We have presented a study of the colloidal stability of particles suspended in partially miscible binary liquid mixtures. More specifically, when the liquid mixture is phase-separated, in all cases there is complete wetting of the colloidal particles by one of the liquid phases. The strong “attraction” of Cw, the main component of this phase, toward the surface of the colloidal particles can result in aggregation. When Cw only makes up a small fraction of the liquid mixture, it adsorbs onto the suspended colloidal particles, as shown by fluorescence of a dye that acts as composition probe. If there is sufficient adsorption on the colloidal particles, shear can make the adsorbed layers coalesce, resulting in shear-induced aggregation via liquid 5769
dx.doi.org/10.1021/la501060w | Langmuir 2014, 30, 5763−5770
Langmuir
Article
(12) Petit, J.-M.; Law, B.; Beysens, D. Adsorption-induced aggregation of colloidal particles in binary mixtures: Modeling the pair free energy. J. Colloid Interface Sci. 1998, 449, 441−449. (13) Law, B.; Petit, J.-M.; Beysens, D. Adsorption-induced reversible colloidal aggregation. Phys. Rev. E 1998, 57, 5782−5794. (14) Vincent, B.; Király, Z.; Emmett, S.; Beaver, A. The stability of silica dispersions in ethanol/cyclohexane mixtures. Colloids Surf. 1990, 49, 121−132. (15) Kline, S. R.; Kaler, E. W. Colloidal interactions in water/2butoxyethanol solvents. Langmuir 1994, 10, 412−417. (16) Beysens, D.; Narayanan, T. Wetting-induced aggregation of colloids. J. Stat. Phys. 1999, 95, 997−1008. (17) Guo, H.; Narayanan, T.; Sztuchi, M.; Schall, P.; Wegdam, G. H. Reversible phase transition of colloids in a binary liquid solvent. Phys. Rev. Lett. 2008, 100, 188303. (18) Gallagher, P. D.; Kurnaz, M. L.; Maher, J. V. Aggregation in polystyrene-sphere suspensions in near-critical binary liquid mixtures. Phys. Rev. A 1992, 46, 7750−7755. (19) Gallagher, P. D.; Maher, J. V. Partitioning of polystyrene latex spheres in immiscible critical liquid mixtures. Phys. Rev. A 1992, 46, 2012−2021. (20) Rathke, B.; Grüll, H.; Woermann, D. Stability of dilute colloidal silica suspensions in the vicinity of the binodal curve of the system 2butoxyethanol/water. J. Colloid Interface Sci. 1997, 192, 334−337. (21) Stö ber, W.; Fink, A.; Bohn, E. Controlled growth of monodisperse silica spheres in the micron size range. J. Colloid Interface Sci. 1968, 26, 62−69. (22) van Blaaderen, A.; Vrij, A. Synthesis and characterization of colloidal dispersions of fluorescent, monodisperse silica spheres. Langmuir 1992, 8, 2921−2931. (23) Sørenson, J.; Arlt, W. In Liquid-Liquid Equilibrium Data Collection; Behrens, D., Eckermann, R., Eds.; Dechema: Frankfurt am Main, 1979. (24) Stephenson, R.; Stuart, J. Mutual binary solubilities: wateralcohols and water-esters. J. Chem. Eng. Data 1986, 31, 56−70. (25) Ethylene glycol is slightly less dense than nitromethane; hence, in phase-separated samples, some of the colloid-containing phase is rising to the top of the sample, making the top phase appear turbid. (26) Bauduin, P.; Wattebled, L.; Schrödle, S.; Touraud, D.; Kunz, W. Temperature dependence of industrial propylene glycol alkyl ether/ water mixtures. J. Mol. Liq. 2004, 115, 23−28. (27) Hijnen, N.; Clegg, P. S. Percolating networks of colloidal rods assembled by a liquid-liquid phase separation. In preparation, 2014. (28) Guery, J.; Bertrand, E.; Rouzeau, C.; Levitz, P.; Weitz, D. A.; Bibette, J. Irreversible shear-activated aggregation in non-Brownian suspensions. Phys. Rev. Lett. 2006, 96, 198301. (29) Lips, A.; Westbury, T.; Hart, P. M.; Evans, I. D.; Campbell, I. J. In Food Colloids and Polymers: Stability and Mechanical Properties; Dickinson, E., Walstra, P., Eds.; RSC Special Publications: London, 1993; Vol. 113; Chapter on the physics of shear-induced aggregation in concentrated food emulsions, pp 31−44. (30) Le Berre, F.; Chauveteau, G.; Pefferkorn, E. Perikinetic and orthokinetic aggregation of hydrated colloids. J. Colloid Interface Sci. 1998, 199, 1−12. (31) Husband, J. C.; Adams, J. M. Shear-induced aggregation of carboxylated polymer latices. Colloid Polym. Sci. 1992, 270, 1194− 1200. (32) Dobbs, H. T.; Darbellay, G. A.; Yeomans, J. M. Capillary condensation between spheres. Europhys. Lett. 1992, 18, 439−444. (33) Greenspan, P.; Mayer, E. P.; Fowler, S. D. Nile red: a selective fluorescent stain for intracellular lipid droplets. J. Cell Biol. 1985, 100, 965−73. (34) Densities were taken to be 1.8 g/cm3 for the particles and 1 g/ cm3 for the adsorbed water layer, and the particle radius (see Experimental Section) was assumed to be 0.38 μm.
5770
dx.doi.org/10.1021/la501060w | Langmuir 2014, 30, 5763−5770
