Size Selection in Viscoelastic Emulsions under Shear
X. Zhao and J. L. Goveas* Department of Chemical Engineering, MS 362, Rice University, 6100 Main Street, Houston, Texas 77005 Received January 8, 2001. In Final Form: May 2, 2001
We investigate the deformation of a dilute emulsion with a viscoelastic continuous phase under shear flow. Initially, polydisperse drops are deformed into thin cylinders. The breakup of these liquid cylinders appears to be suppressed until they reach a certain width. The resulting ruptured drops are uniform along the length of the cylinder. As a control, we also study drop breakup for a Newtonian emulsion, at a similar viscosity ratio. In this case, there is no corresponding size selection for the cylinder radius or daughter drops.
Introduction surface tension, and a is the diameter of the initial
Emulsions are usually formed by dispersing drops of spherical drop. [We consider only low (µ < 1) viscosity one immiscible fluid in another under shear. If the shear ratio drops in this letter.] Much work has been done on rate is large enough to overcome the interfacial tension quiescent drop breakup; that is, a spherical drop is between the two fluids, the drops deform and rupture deformed by an elongational flow, but drop breakup occurs into smaller drops, leading to a polydisperse emulsion. only after cessation of the flow.8 Studies in simple shear These smaller drops are prevented from coalescing by flows have been restricted to the low-deformation regime9 localizing surfactants at the drop interfaces, which provide and to the analysis of breakup times and final size strong short-ranged repulsion and give the emulsion long- distributions.7,10 term stability. Emulsions are ubiquitous in the food, In this paper, we describe the drop deformation and cosmetics, and paint industries,1 where product properties breakup of dilute emulsions of surfactant-covered drops and emulsion rheology are largely controlled by the mean in steady shearing flows. An initially spherical drop is droplet size and size distribution.2 deformed into an ellipsoid which rotates with the flow. Some years ago, Mason and Bibette3 (MB) found that The ellipsoid stretches into a liquid cylinder, becoming certain oil-in-water emulsions become monodisperse under progressively more aligned with the flow direction. Capil- shear. The mechanism that causes this monodispersity lary waves grow on the cylinder surface and are convected was not found. These emulsions are special in three ways: away, as the cylinder continues to thin. When the (i) the continuous phase consists of a 40 wt % aqueous amplitude of these waves becomes equal to the cylinder surfactant solution (nonionic Tergitol NP-7) (i.e., it is radius, the drop breaks up into daughter drops separated highly viscoelastic); (ii) the volume fraction of the dispersed by smaller satellite drops. The daughter drops may be phase (silicone oil) corresponds to close packing of the nonuniform along the length of the cylinder. Similar drops; (iii) the emulsion must be sheared in a thin (100- processes at the cylinder ends cause them to bulb up and 200 micron) gap in order to prevent nonuniform flow. The break off, a process known as “end-pinching”.8 End-pinched surfactant plays a dual role in the MB system, by drops generally form in advance of the daughter drops stabilizing the emulsion and making it viscoelastic. In and may have different sizes. turn, the high oil concentration both causes hydrodynamic The size distribution resulting from the breakup of even interactions between the drops and contributes to the a single drop is thus polydisperse. However, the small emulsion elasticity. The coupling between viscoelasticity satellite drops and the few end-pinched drops (for long and concentration makes the MB emulsion quite a cylinders) do not contribute greatly to the volume poly- complicated system. In this letter, we isolate the effects dispersity. A dilute emulsion contains multiple drops of of viscoelasticity in producing monodispersity. different sizes, which provides an additional source of Most fundamental studies on drop breakup have focused polydispersity for the ruptured emulsion. The initial drop on single drops in Newtonian systems.4-7 This is because size may affect the cylinder width at breakup, inducing the rupture of a single drop is itself a complex, transient a dependence of the daughter drop radii on the mother process. In such systems, only two dimensionless param- drop radii. eters control the physics of drop breakup: the viscosity Here, drop breakup in both viscoelastic and Newtonian ratio µ ) ηi/ηo, and the capillary number, Ca. matrixes is investigated. We find that a viscoelastic continuous phase greatly narrows the distribution of ηoγ̆ cylinder widths at breakup as well as the daughter drop Ca ) (1.1) size distribution. Because the daughter drop size distri- σ/a (4) Stone, H. A. Annu. Rev. Fluid Mech. 1994, 26, 65. Here, ηi and ηo refer to the dispersed and continuous phase (5) Acrivos, A. Ann. N.Y. Acad. Sci. 1983, 404, 1. viscosity, respectively, γ̆ is the shear rate, σ is the drop (6) Rallison, J. M. Annu. Rev. Fluid Mech. 1984, 16, 45. (7) Grace, H. P. Chem. Eng. Commun. 1982, 14, 225. (8) Stone, H. A.; Bentley, B. J.; Leal, L. G. J. Fluid Mech. 1986, 173, * To whom correspondence should be addressed. 131. Stone, H. A.; Leal, L. G. J. Fluid Mech. 1989, 206, 223. (1) Becher, P. Emulsions: Theory and Practice, 2nd ed.; Reinhold (9) (a) Guido, S.; Greco, F.; Villone, M. J. Colloid Interface Sci. 1999, Publishing Co.: New York, 1965. 219, 298. (b) Guido, S.; Greco, F.; Villone, M. J. Colloid Interface Sci. (2) Mason, T. M. Curr. Opin. Colloid Interface Sci. 1999, 4, 231. 2000, 224, 209. (3) Mason, T. M.; Bibette, J. Phys. Rev. Lett. 1996, 77, 3481; Langmuir (10) Bigio, D. I.; Marks, S. R.; Calabrese, R. V. Int. Polym. Process. 1997, 13, 4600. 1998, 13, 2.
Published on Web 06/01/2001 Letters Langmuir, Vol. 17, No. 13, 2001 3789
and incorporating it into the emulsion by gentle stirring with a
spatula. This procedure minimizes the formation of very small drops which can be ruptured only by very strong shear flows. Apparatus. A simple shear flow field is generated by a shear cell built on the stage of an inverted microscope (Nikon, Eclipse TE-300). The experimental apparatus is shown in Figure 2. The cell consists of two glass plates, between which the sample is placed. The lower plate is stationary, and the upper plate is driven by an air cylinder, which moves smoothly along linear rollers. The upper plate speed can be regulated between 1 and 50 cm/s. The gap between the plates is maintained between 100 and 200 microns. To ensure uniformity of the flow, it is essential that the two plates be parallel to each other. The gap is first zeroed by bringing the plates in contact with each other. The parallelism of the plates is acertained by counting the interference fringes visible on the top plate. The plates are then locked in place and separated to the desired gap width, using three micrometers equipped with electronic sensors. Recording of the Deformation Process. The deformation and breakup of emulsion droplets is filmed using a high-speed camera (Kodak, Ektapro 1000HRC), which can record images Figure 1. The continuous phase of the viscoelastic emulsion up to 1000 frames per second. The experiment usually lasts about is a 12 wt % solution of PVP and 2 wt % SDS in water, which is shear-thinning. The emulsion is sheared at 750 s-1, where a second. The images are then downloaded to a computer and the continuous phase has a viscosity of 701 cP. processed (using Adobe Photoshop), to measure the drop size and thread width. Results bution is the most significant source of polydispersity, we believe that this cylinder size selection leads to the The deformation processes for the NE and VE are monodispersity observed in the MB experiments. compared by shearing both emulsions at fixed viscosity ratios and external stresses. Both experiments are per- Experimental Section formed at a shear rate of 750 s-1 in a 100 micron gap. At this shear rate, the continuous phase of the VE emulsion Model Emulsions. We study two emulsions in this study: a purely Newtonian emulsion (NE) and a viscoelastic emulsion has a viscosity of 701 cP. The viscosity ratios of the NE (VE). The dispersed phase in both emulsions is silicone oil, which and VE emulsions are thus 0.453 and 0.482, respectively. is a Newtonian fluid with a viscosity of 338 cP. A small amount In the NE case, we observe qualitatively the same of surfactant is used to emulsify both systems. The continuous phenomena under shear flow as has been previously phase of the NE is 2 wt % of the nonionic surfactant TWEEN-80 observed for quiescent drop breakup.4 The drop deforms (Polyoxyethylene Sorbitan Monooleate, Sigma) in glycerine (a into a long, thin thread which continuously stretches in Newtonian fluid with a viscosity of 747 cP). [Pure glycerine has the flow. Drops pinch off from the ends, and the thread a viscosity of 945 cP; the lower viscosity measured here reflects breaks up into a set of daughter drops and satellite drops. some absorption of water.] The VE continuous phase consists of We observe that the drops deform into cylinders which 2 wt % SDS (sodium dodecyl sulfate) in an aqueous solution of 12 wt % polyvinyl pyrrolidine (PVP) (MW ) 1.3M, Aldrich). This have very different widths at breakup and that the is a viscoelastic, shear-thinning solution (see Figure 1). The cylinders break up at different times. Figure 3 shows the polymer solution rheology is measured in a controlled stress NE before and during shear. rheometer using a cone-and-plate geometry. The emulsions are We observe similar deformation and breakup mecha- prepared by adding the oil drop by drop to the continuous phase nisms in the VE case. However, the cylinder width at
Figure 2. Schematic of experimental apparatus used to shear emulsions. The three-dimensional view shows the linear shear cell mounted onto a microscope stage: (A) microscope stage; (B) aluminum fixed frame with linear rollers (shown as dotted lines); (C) upper frame which moves along linear rollers; (D) upper glass plate, which is affixed to C; (E) micrometer. The inset is a side view of the shear cell showing, in addition, (F) the sample placed between upper and lower plates and (G) the lower stationary glass plate. The gap between the upper and lower plates is set using the three micrometers. 3790 Langmuir, Vol. 17, No. 13, 2001 Letters
Figure 3. (a) Presheared Newtonian emulsion containing 2 Figure 4. (a) Presheared viscoelastic emulsion containing 2 wt % silicone oil in glycerol with 2 wt % TWEEN-80, which wt % silicone oil in an aqueous solution of 12 wt % PVP and corresponds to a viscosity ratio of 0.453. (b) Newtonian emulsion 2 wt % SDS, which corresponds to a viscosity ratio of 0.482. (b) under shear. Drops of different sizes are deformed into cylinders Viscoelastic emulsion under shear, showing that cylinders have which have different widths at breakup. The resulting daughter similar widths at breakup. The resulting daughter drops are drops are nonuniform along the length of the cylinder. very uniform.
breakup is almost the same for all drops (see Figure 4). decreasing the shear rate from 750 to 500 s-1 in the NE In addition, the daughter drops resulting from the breakup has little effect on the cylinder width distribution. Also, of these cylinders are uniform along the cylinder length. there is a broad distribution of initial drop sizes (and Because the lower plate in the shear cell is stationary, consequently of Ca’s) in both emulsions, and we have drops constantly move in and out of the field of view. estimated a substantial overlap in Ca. Therefore, the size Although we cannot follow the motion of a single drop, it selection in the VE is not likely to be due to Ca differences is clear that the cylinders originate from different-sized between the emulsions. mother drops, because they have similar radii but different Effect of Surfactant. Notice that different surfactants lengths. [Because only the plane containing the flow and have been used for the NE and VE. In addition, anionic vorticity directions can be observed, the cylinder cross surfactants are known to complex with neutral polymers.12 section is assumed to remain spherical.] In the VE, all the Consequently, we have repeated the VE experiments using drops also break up at different times. TWEEN-80 as the surfactant, but we find qualitatively To quantify the narrowing of the cylinder width the same phenomena as for the SDS/PVP system. distribution for the VE, we have measured the diameters Effect of Gap Thickness. The drops in the presheared of 40 cylinders for each emulsion. The cylinder width is emulsions range in size from 5 to 100 microns, so that the recorded when capillary waves on the surface of the larger drop diameters can become comparable to the cylinder first become discernible. These waves will system size. Doubling the gap width has no effect on the eventually grow and result in the breakup of the cylinder. thread width distribution, so the size selection probably For both emulsions, drops do not rebreak in the flow. cannot be attributed to confinement effects. Figure 5 shows that the polydispersity (the ratio of the standard deviation to the mean width) for the NE is 3 Discussion times larger than that for the viscoelastic case. We have investigated drop breakup in a viscoelastic Effect of Ca. There is a substantial variation in emulsion and a “control” Newtonian emulsion. Under interfacial tension between the NE and VE. We measure shear flow, dilute drops deform into cylinders which break σVE ) 27.3 dyn/cm and σNE ) 4.7 dyn/cm, using the spinning up into smaller drops. At breakup, the distribution of drop technique.11 This means that equal-sized drops have cylinder widths is much narrower in the viscoelastic a smaller Ca in the VE than in the NE. However, emulsion compared to the Newtonian case. The resulting Letters Langmuir, Vol. 17, No. 13, 2001 3791
quiescent breakup of Newtonian drops in a viscoelastic
matrix has found a narrowing of the dispersion relation for the fastest growing unstable mode. Because linear stability analyses give surprisingly good predictions for final drop sizes,4 this process would lead to uniform daughter drops. We suspect that such wavelength selection is present for dynamical breakup as well and is responsible for the cylinder and daughter drop uniformity. In recent years, drop deformation studies in blends of polymer melts have produced many exotic phenomena such as cylinder-folding14 and vorticity alignment.15 We do not observe these effects in our system, probably because the dispersed phase is Newtonian. However, second normal stress differences have been shown to widen the diameter of deformed drops in polymeric emulsions.14 Similarly, first or second normal stress differences in the continuous phase may cause the cylinder width suppres- sion observed here. We find that decreasing the polymer concentration of the outer phase (while maintaining a semidilute solution) or decreasing the shear rate broadens the cylinder width distribution. This could be due to decreasing normal stresses but may also be attributed to lowering the capillary number. In future work, we will focus on single drops in viscoelastic phases under shear, to correlate the normal stresses, viscosity ratio, and capillary number to the cylinder width at breakup. Returning to the MB experiments, how does concentra- tion affect breakup? A suggestive experiment was done by Elemans et al.16 on concentrated, monodisperse, molten polymer fibers. They showed that the quasi-quiescent breakup of these fibers was synchronous and resulted in uniformly sized daughter drops. We speculate that co- operativity between cylinder size selection due to vis- coelasticity and simultaneous breakup due to concentra- tion results in a monodisperse distribution in the MB emulsions. Figure 5. Probability distribution for thread widths in (a) viscoelastic emulsion and (b) Newtonian emulsion. Notice that the thread widths have been normalized by the mean thread Acknowledgment. We thank R. Chronister for help width. Forty threads were measured in each case. with the design and construction of the linear shear cell, H. Zhang for making interfacial tension measurements, daughter drops are also more uniform along the length of and C. A. Miller, G. J. Hirasaki, M. Pasquali, T. G. Mason, the cylinder for the viscoelastic system. We believe that and H. A. Stone for useful discussions. J.L.G. thanks A. this mechanism is largely responsible for the production Imhof, V. Manoharan, D. Pine, D. Ternet, R. Stuber, and of monodisperse emulsions under shear in other vis- L. G. Leal at UCSB for invaluable assistance in initiating coelastic emulsions.3 A linear theory by Lee et al.13 for the this work. We acknowledge the Petroleum Research Foundation under Grant No. 34808-G9 for financial (11) Miller, C. A.; Neogi, P. Interfacial Phenomena, Equilibrium and support. Dynamic Effects; Marcel Dekker: New York, 1985; p 31. (12) Goddard, E. D.; Ananthapadmanabhan, K. P. In Polymer- LA0100439 Surfactant Systems; Kwak, J. C. T., Ed.; Marcel Dekker: New York, 1998; p 22. (13) Lee, W. K.; Yu, K.; Flumerfelt, R. W. Int. J. Multiphase Flow (15) Migler, K. B. J. Rheol. 2000, 44, 277. 1981, 7, 385. (16) Elemans, P. H. M.; van Wunnik, J. M.; van Dam, R. A. AIChE (14) Levitt, L.; Mackosko, C. W. Polym. Eng. Sci. 1996, 36, 1647. J. 1997, 43, 1649.
