You are on page 1of 9

Langmuir 2001, 17, 2647-2655 2647

Coalescence in Surfactant-Stabilized Emulsions Subjected


to Shear Flow
Arup Nandi, D. V. Khakhar, and Anurag Mehra*
Department of Chemical Engineering, Indian Institute of TechnologysBombay,
Powai, Bombay-400076, India

Received October 20, 2000. In Final Form: January 29, 2001

Experiments were carried out to study the coalescence dynamics of a neutrally buoyant liquid-liquid
emulsion subjected to a simple shear flow in a Couette device. The effect of shear rate and dispersed phase
holdup were studied. The surfactant-stabilized emulsions were prepared in a stirred tank at high shear
rates which were applied for long times so as to obtain reproducible equilibrium drop size distributions.
Low shear rates were used in the Couette device to prevent drop breakup. Sufficiently large drops and
density matching ensured that coalescence due to Brownian motion and creaming was negligible and thus
the change in drop size distribution with time was entirely due to shear-induced coalescence. The evolving
volume density distributions were measured using optical microscopy and image analysis. A population
balance model based on Smoluchowski’s result for the rate of shear-induced coalescence and an empirical
form for the coalescence efficiency was used to describe the system. Shearing results in stabilization of
the emulsion, and the coalescence rate reduces with increasing shear rates. The stabilizing effect due to
shearing is greater for higher holdups. The drop size distribution for high values of the holdup becomes
bimodal at long times. The coalescence efficiency, back-calculated from experimental data, is independent
of drop size for low holdups but is size dependent for higher holdups. This work demonstrates a simple
experimental method for evaluating the coalescence efficiency for emulsions with high holdups where
theoretical models are not available.

Introduction the drops to “collide” (interfaces approach within some


Liquid-liquid emulsions constitute a wide variety of specified small length) as a result of the velocity gradient
products and are formed in many processes of industrial (Figure 1). The drops form a doublet and rotate in a plane
relevance. Most lubricants, paints, and cosmetics and that is coplanar to the drop relative velocity vector (u)
many food products are emulsions.1 Processes such as and the line joining their centers (e). The interfacial film
liquid-liquid extraction, petroleum refining, and many between the drops thins initially, and it may deform due
areas of polymer processing involve handling and convey- to the pressure within the film. The process of film thinning
ing of emulsions.1-3 During these operations the emulsion continues until the following condition is satisfied
drops are subjected to flow fields which may cause them
to break or coalesce. The simultaneous occurrence of the u‚e e 0 (1)
two phenomena controls the drop size distribution, which
in turn has a significant effect on the processing conditions Beyond this time the drops move apart and the film begins
and characteristics of the product. Most emulsions contain to thicken. If the film has achieved the critical thickness
surface active impurities or added surfactants, which act within this time, it may rupture, leading to coalescence.
as emulsion stabilizers. Thus, a fundamental understand- Otherwise, the drops continue their respective motions
ing of the phenomena of drop coalescence and breakage with the bulk flow (Figure 1). A review of previous studies
in surfactant-stabilized emulsions is of wide interest. related to this process is given below.
This work explores the effect of shear rate and dispersed The simplest model for coalescence of drops in an
phase holdup on the coalescence kinetics of a surfactant- emulsion subjected to a simple shear flow is based on
stabilized emulsion subjected to a simple shear flow. The Smoluchowski’s theory,6 with the coalescence rate given
shear rates used are small enough such that no drop by
breakup occurs.4,5 The drops in the emulsions used are
large enough so that coalescence due to Brownian motion γ
is negligible. Also, the densities of the drops and the Cs(ν,ν′) ) [ν1/3 + ν′1/3]3n(ν) n(ν′) (2)
π
continuous phase are matched so that creaming-induced
coalescence is insignificant. Thus, the evolution of the drop where ν and ν′ are the volumes of the coalescing drops,
size distribution in the system occurs due to shear-induced γ is the shear rate, and n(ν) is the number density of the
coalescence alone. drops. Smoluchowski’s equation is based on the assump-
Consider the physical processes during the interaction tion that the drops move along the flow streamlines and
of a pair of drops in a shear flow. The shear flow causes hence does not account for hydrodynamic interactions.
* Corresponding author. E-mail: mehra@che.iitb.ernet.in. The effects of physicochemical parameters affecting in-
(1) Venugopal, B. V.; Wasan, D. T. In Encyclopedia of Emulsion terdrop film stability are also unaccounted for in the
Technology; Becher, P., Ed.; Marcel Dekker: New York, 1983; Vol. 2. theory. Inclusion of such effects generally results in a
(2) Mana-Zloczower, I. In Mixing and Compounding of Polymers-
Theory and Practice; Mana-Zloczower, I., Tadmor, Z., Eds.; Hanser coalescence rate that is lower than the Smoluchowski
Publishers: Munich, 1994; p 5. collision rate and is typically expressed as
(3) Ottino, J. M.; et al. Adv. Chem. Eng. 1999, 25, 105.
(4) Grace, H. P. Chem. Eng. Commun. 1982, 14, 225.
(5) Khakhar, D. V.; Ottino, J. M. J. Fluid Mech. 1986, 166, 265. C(ν,ν′) ) Cs(ν,ν′) (3)

10.1021/la001473m CCC: $20.00 © 2001 American Chemical Society


Published on Web 04/05/2001
2648 Langmuir, Vol. 17, No. 9, 2001 Nandi et al.

diameter ratio becomes smaller with decreasing viscosity


ratio (µd/µc), where µd is the viscosity of the dispersed phase
and µc that of the continuous phase.
Only relatively few experimental studies of coalescence
in emulsions subjected to a laminar shear flow have been
previously reported. Hazlett and Schechter16 subjected
an emulsion of methyl methacrylate in a mixture of
methanol and water to tangential Couette flow. The
Figure 1. Schematic view of the stages of the collision between coalescence efficiency, calculated using a population
two drops in a shear flow where u is the relative velocity vector balance model, was studied at temperatures close to the
and e is the unit vector along the line joining the centers. critical point of the emulsion. Their results indicate that
there is no variation of the coalescence efficiency at
where  is the coalescence efficiency, defined as the fraction temperatures near the critical point. Vinckier et al.17
of collisions which result in coalescence. studied shear-induced coalescence in a polymer blend
The drainage of the film between closely spaced drops, produced at a shear rate higher than that at which the
which may determine coalescence efficiency, has been emulsion was broken. Their results, obtained from mea-
studied in several works, and a review is given by sured average drop diameters, matched the coalescence
Chesters.7 Expressions for the coalescence efficiency based efficiency model for partially mobile interfaces given by
on models for the rate of thinning of films for fully mobile, Chesters.7 Experiments at high dispersed phase holdups
partially mobile, and immobile drop interfaces are de- gave a coalescence efficiency which increased with dis-
rived.7 An assumption of the models is that films below persed phase holdup. This was explained on the basis of
a critical thickness rupture, thereby leading to coalescence. an increased critical film thickness for coalescence at
Several more recent works focus on the details of film higher holdups. Mishra et al.18 subjected a surfactant-
thinning between drops. free emulsion of pentadecane in an aqueous solution of
Davis et al.8 carried out a lubrication analysis of film NaCl to a simple shear flow using a concentric cylinder
thinning for nondeformable drops pressed together by a apparatus. The evolving drop size was measured using
constant force. Hartland et al.9 considered both the viscous laser Doppler anemometry. Their results follow the hard
and inertial components for a flat circular interface with sphere predictions of Zeichner and Schowalter13 and Feke
partially mobile interfaces. Later work by Jeelani and and Schowalter.14 They also found the drop size distribu-
Hartland10 included the effect of van der Waals forces on tion to be self-similar, in agreement with the analysis of
the interfacial film thinning rate during the axisymmetric Swift and Friedlander,19 for systems with a constant
motion of a pair of equal-sized liquid drops pressed together coalescence efficiency. The coalescence rate was found to
by a constant force. Abid and Chesters11 carried out increase with shear rate as well as with electrolyte
detailed calculations of the film thinning rate for a pair concentration.
of deformable drops with partially mobile interfaces. Their Shear-induced coalescence in the presence of surfactant
results are close to the rates predicted by the model of was studied by Mousa and van de Ven.20 An emulsion of
Chesters7 for the same case. Hartland and Jeelani12 also silicone oil in water, stabilized by SDS (sodium dodecyl
studied the effect of interfacial tension gradients on the sulfate) was sheared between rotating circular plates. The
rate of thinning of the intervening film. coalescence efficiency was back-calculated using the
Computations of the coalescence efficiency, taking into population balance equation incorporating an assumed
account the detailed hydrodynamics around a pair of drops, functionality for the coalescence efficiency given by

{
have been carried out in a few works. Zeichner and

[ 4q
]
Schowalter13 studied the coagulation of equal-sized rigid 5.0
R0 , if ν and ν′ < νc
particles with immobile interfaces in shear and extensional ) (1 + q)2 (4)
flow taking into account van der Waals forces. Particles if ν or ν′ > νc
which approach closer than a critical distance form a
0,
permanent doublet (coagulate) as a result of van der Waals
forces. This work was extended by Feke and Schowalter14 In the above equation, q ) (ν/ν′)1/3 is the ratio of the
to include the case of particles with different diameters. diameter of the colliding drops, and νc is the maximum
Wang et al.15 calculated the coalescence efficiency for rigid volume a drop can attain due to coalescence with other
drops with partially mobile interfaces under the action of drops. The index value (5.0) was selected on the basis of
a simple shear and uniaxial extensional flow. Their results phenomenological arguments. Experiments were carried
indicate that the highest probability of coalescence exists out at a dispersed-phase holdup of 0.2%, and the volume-
for drops which are of the same size. For a given emulsion averaged diameters of the coalescing emulsion were found
system, their results predict a critical drop diameter ratio from the transmitted light intensity measurements. Their
(smaller drop diameter to larger drop diameter) below results indicate that for a surfactant-free emulsion the
which there is no coalescence. The value of this critical coalescence efficiency term (R0) first decreased with shear
rate and then showed an abrupt increase beyond a certain
(6) Smoluchowski, M. V. Z. Phys. Chem. 1917, 92, 129. value of the shear rate. A similar qualitative behavior
(7) Chesters, A. K. Trans. Inst. Chem. Eng. 1991, 69, 259. was reported earlier in the theoretical study by van de
(8) Davis, R. H.; Schonberg, J. A.; Rallison, J. M. Phys. Fluids A Ven and Mason.21 The addition of salts such as KCl and
1989, 1, 77.
(9) Hartland, S.; Jeelani, S. A. K.; Suter, A. Chem. Eng. Sci. 1989, AlCl3 resulted in a rise in the coefficient R0 with shear
44, 387.
(10) Jeelani, S. A. K.; Hartland, S. J. Colloid Interface Sci. 1993, 156, (16) Hazlett, R. D.; Schechter, R. S. Colloid Surf. 1988, 29, 71.
467. (17) Vinckier, I.; et al. AIChE J. 1998, 44, 951.
(11) Abid, S.; Chesters, A. K. Int. J. Multiphase Flow 1994, 20, 613. (18) Mishra, V.; Kresta, S. M.; Masliyah, J. H. J. Colloid Interface
(12) Hartland, S.; Jeelani, S. A. K. Colloids Surf., A 1994, 88, 289. Sci. 1998, 197, 57.
(13) Zeichner, G. R.; Schowalter, W. R. AIChE J. 1977, 23, 243. (19) Swift, D. L.; Friedlander, S. K. J. Colloid Sci. 1964, 19, 621.
(14) Feke, D. L.; Schowalter, W. R. J. Fluid Mech. 1983, 133, 17. (20) Mousa, H.; van de Ven, T. G. M. Colloids Surf., A 1991, 60, 39.
(15) Wang, H.; Zinchenko, A. Z.; Davis, R. H. J. Fluid Mech. 1994, (21) van de Ven, T. G. M.; Mason, S. G. Colloid Polym. Sci. 1977, 255,
265, 161. 468.
Coalescence in Surfactant-Stabilized Emulsions Langmuir, Vol. 17, No. 9, 2001 2649

rate and then a drop beyond a critical value. The addition


of sodium dodecyl sulfate (SDS) to the emulsion gave a
similar result. Schokker and Dalgleish22 reported floc-
culation of drops in an oil-in-water emulsion stabilized by
sodium casseinate. Their experiments were conducted in
a tangential Couette device at very high shear rates (670-
740 s-1). They showed that the coalescence rate was
strongly influenced by the storage time as well as by the
apparatus used for making the emulsion.
We have recently studied coalescence in a neutrally
buoyant, surfactant-stabilized emulsion subjected to a
simple shear flow in a Couette device.23 The experiments
were done for a low dispersed phase holdup (1%) at which
multidrop collisions are negligibly small. The coalescence
efficiency was determined by fitting the theoretically
obtained drop size distribution profile to the experimental
data and found to be independent of colliding drop sizes.
The model of Chesters7 was found to overpredict the rate
of coalescence by 2 orders of magnitude. A model for the
computation of coalescence efficiency, based on the
stochastic nature of film rupture, gave good agreement
with experimental data. Figure 2. Initial volume density distributions for all three
The stability of the film between flat interfaces was holdups.
studied by Nikolov and Wasan;24 they found that the
coalescence time in the presence of a surfactant was much Table 1. Time of Stirring (ts) and Rotation Speed (ωs) for
higher than that for the corresponding surfactant-free Making the Emulsion and Resulting Volume-Average
h ) for the Starting Emulsions for the Three
Diameters (D
system as well as the value predicted by models for film Holdups (O), Together with the Interfacial Tension
thinning. Coalescence times for identical drops were Values (σ)
distributed over a range of values, and the stability of the
φ ts (h) ωs (rpm) h (µm)
D σ (mN/m)
interface was found to be more important than the fluid
mechanics governing the thinning of the film. Dreher et 1% 2.5 2090 ( 40 8.41 ( 0.09 16.99 ( 0.01
al.25 conducted a similar study in a surfactant-free system 5% 5.5 2200 ( 50 8.07 ( 0.24 16.95 ( 0.05
and found that the coalescence time was log-normally 15% 8.5 2450 ( 40 8.83 ( 0.28 16.92 ( 0.12
distributed. Ghosh26 has demonstrated similar results for
surfactant loaded systems using different surfactants. the ratio of glycerol to water, to match that of the dispersed
phase to an accuracy of (0.005 g/cm3. All experiments were
The objective of the current work is to study the effect
conducted at room temperature, which was maintained at 25 (
of shear rate on surfactant-stabilized emulsions at high 1 °C. The continuous-phase constituted 0.525 L of glycerol and
dispersed-phase holdups. Despite the practical importance 10 g of sodium chloride (0.1131 M) dissolved in 1 L of double-
of emulsions with high holdup, no theoretical or experi- distilled water. The addition of sodium chloride masks the trace
mental work for the case of well-defined laminar flows ionic impurities and suppresses the electric double layer, resulting
has been previously reported. We present here the results in a characteristic Debye-Huckel length of 3.35 × 10-2 µm.27
of an experimental study of the evolution of the drop size The surfactant was initially dissolved in the dispersed phase,
distribution in a neutrally buoyant, surfactant-stabilized and a surfactant concentration of 0.85% (w/w) of the dispersed
emulsion subjected to a shear flow. The evolving drop size phase was used in all the emulsions.
distributions obtained from a population balance model Preparation of Emulsion. Emulsions were stirred for long
times to achieve a dynamic equilibrium between drop breakup
simulation are fitted to the experimentally obtained drop
and coalescence and thus to obtain a steady state for the drop
size distributions to back-calculate the coalescence ef- size distribution. This ensured reproducibility of the starting
ficiency. Experimental details are given in the next section drop size distributions in the shear flow experiments. Further,
followed by a description of the model and computational the impeller rotation speed was adjusted to obtain nearly similar
procedures used. drop size distributions for the different cases. The emulsion-
making conditions for the various holdups are summarized in
Experimental Section Table 1, where φ is the volume fraction of the dispersed phase,
ts is the time of stirring, and ωs is the impeller rotational speed.
Materials and Method. Monochlorobenzene (analytical The interfacial tension (σ) between the two phases was measured
grade, Glaxo Chemicals, India) was dispersed in a mixture of using a drop volume tensiometer after equilibrating the two
glycerol (LR grade, Merck, India), double-distilled water, and phases (oil and aqueous) for 36 h. The values of the interfacial
sodium chloride (Loba Chemie, India), using TWEEN-80 (spec- tensions are quite close to each other (Table 1) for the three
troscopic grade, SD Fine Chemicals, India) as the surfactant. holdups.
Emulsions were prepared in a 1 L beaker fitted with stainless The initial volume average diameters (D h ) 〈D3〉1/3) are highly
steel baffles. A shrouded turbine impeller was used for emul- reproducible and are quite close for all three holdups (Table 1).
sification. The beaker was covered in order to avoid dust Figure 2 shows the initial drop size distributions for the three
contamination as well as to minimize the evaporation of holdups, which are very similar. The distributions are shown on
chlorobenzene. To avoid gravity-driven motion of the drops, the a volume basis with (πD3/6)f(D) dD being the volume fraction of
specific gravity of the continuous phase was adjusted, by varying drops in the size range (D, D + dD). The distributions are
normalized by the total volume fraction (φ) for comparison. The
(22) Schokker, E. P.; Dalgleish, D. G. Colloids Surf., A 1998, 145, 61. error bars, which indicate the standard deviation over a minimum
(23) Nandi, A.; Mehra, A.; Khakhar, D. V. Phys. Rev. Lett. 1999, 83, of three runs, are small and show the reproducibility of the initial
2461.
(24) Nikolov, A. D.; Wasan, D. T. Ind. Eng. Chem. Res. 1996, 34,
distributions.
3653.
(25) Dreher, T. M.; et al. AIChE J. 1999, 45, 1182. (27) Stokes, R. J.; Evans, D. F. Advances in Interfacial Engineering:
(26) Ghosh. Ph.D. Thesis, I. I. T.sBombay. Wiley-VCH: New York, 1997.
2650 Langmuir, Vol. 17, No. 9, 2001 Nandi et al.

Figure 3. Schematic diagram of the tangential Couette flow


assembly.

Figure 5. Schematic diagram showing a rectangular cavity


made using cover slips for viewing the emulsion sample in the
microscope. The typical height of the rectangular cavity varied
Figure 4. Plot of the volume-average diameter versus the from 270 to 300 µm, which was significantly higher than the
number of drops counted. diameter of the largest drop in the experiments.
Shear Flow Apparatus. The emulsion was sheared in a
tangential Couette flow apparatus with the inner cylinder
rotating. A schematic view of the apparatus along with dimen-
sions is shown in Figure 3. The rotation of the inner cylinder
generates a shear flow in the gap between the cylinders which
is small compared to the diameter of the cylinders to ensure a
uniform shear rate. The ratio of the width of the gap to the length
of the cylinder was >20, which resulted in minimization of
longitudinal instabilities28 and negligible end effects. The
cylinders were made of brass to obtain a smooth surface and to
prevent corrosion. The bottom surface of the lower cylinder was
cut in the form of a cone with an angle of 6° to ensure a uniform
shear rate at the bottom. The inner cylinder was powered by an
ac motor, through a set of gears with a high reduction ratio, so
that small fluctuations in the shaft speed resulted in minimal
changes of the rotation speed of the inner cylinder ((0.5 rpm).
The critical angular speed of the inner cylinder for transition
from simple shear flow to Taylor vortices is given by29

[3390(d22 - d12)
]
0.5
Ωc ) ν 2 4
(5)
4d1 (d2 - d1)

where d1 and d2 are the diameters of the outer and the inner
cylinders, respectively (Figure 3). The critical angular velocity Figure 6. Volume density distribution of the same emulsion
calculated using the kinematic viscosity (ν) of the continuous sample taken separately on two different slides: (A) measure-
phase is 5.0 rad s-1 (48 rpm). ment made 8 min after sampling and again after 2.5 h; (B)
Measurement of Drop Size Distribution. The drop size measurement made 8 min after sampling from the same
distribution was obtained by using an Image analysis system emulsion stock from which A was withdrawn.
consisting of a CCD camera (SONY 94C) mounted on an optical
microscope (Olympus BX60). Various combinations of objective gravity of the diluting phase was maintained around 0.4%-0.6%
lenses were used to accurately measure drops of diameter between higher than that of the dispersed phase. This resulted in all the
2 and 200 µm. The emulsion samples were diluted to prevent drops rising and resting lightly against the top cover slip, and
overlap of drops during viewing in the microscope and to control consequently, all drops were in the plane of focus of the objective
agglomeration as well as coalescence of the drops during lens. Each image typically contained 80-200 drops, and several
microscopy. The diluent used was the pure continuous phase, nonoverlapping images (at least 156) were taken to capture a
equilibrated with chlorobenzene, and containing 1% (w/w of the sufficiently large number of drops.
continuous phase) surfactant. The large quantity of surfactant Consistency Tests. Several tests were performed to check
present prevented the drops from coalescing during storage on the consistency and accuracy of the experimental procedure. The
the slide. A drop of the diluted emulsion was placed on a glass minimum number of drops to be counted in order to obtain a
slide with spacers, and a cover slip was carefully placed over it. good description of the drop size distribution was examined by
Thus, the emulsion sample was enclosed in a rectangular cavity, counting an increasing number of drops. Figure 5 shows the
as shown in Figure 4. This procedure ensures that the top cover variation of volume-average diameter with the number of drops
slip does not press the drops and deform them. Nor does the counted for an emulsion sample. The value of the volume-average
emulsion flow out due to the weight of the cover slip. To ensure diameter becomes nearly constant beyond 15 000 drops, and
that all drops are within the same plane of focus, the specific beyond this limit reasonably smooth distributions were obtained
as well. Hence, in all the reported data, the number of drops
(28) Snyder, H. D. Phys. Fluids 1968, 11, 1599. counted was in excess of 15 000. The stability of the emulsion,
(29) Chandrashekhar, S. In Hydrodynamic and hydromagnetic withdrawn for making drop size measurements, after dilution
stability; Dover publications: New York, 1981; pp 272-339. was tested as follows. Two samples (samples A and B) were taken
Coalescence in Surfactant-Stabilized Emulsions Langmuir, Vol. 17, No. 9, 2001 2651

on the size ratio. The last two terms account for the fact
that there may be a critical drop diameter (Dc) at which
the coalescence efficiency is maximum or minimum,
depending on the value of m. The prefactor R0 is an overall
coalescence efficiency term which is assumed to depend
only on the shear rate and holdup. The parameters of the
model (R0, C, m, and Dc) are obtained by a regression
procedure applied to the experimental drop size distribu-
tions as described below.
The population balance equation (eq 6) along with the
coalescence efficiency (eq 7) was solved using a 4th-order
Runge-Kutta method with a variable time step. The
experimentally obtained initial drop size distribution was
used as the initial condition in the computations. There-
after, at each integration time step, the error (E) between
the theoretical and the experimental distributions was
calculated using
N
E) ∑
i)0
[(Di3n(Di)|theor - Di3n(Di)|exp)∆Di]2 (8)
Figure 7. Volume density distribution of emulsion samples
taken from the top sampling port and the bottom of the Couette
apparatus after 2.5 h of shearing at 10 rpm, for holdup φ ) 1%. Computations were carried out for fixed values of m, C,
and Dc, and the value of R0 was obtained at which the
from the same emulsion stock and then diluted separately. The error was minimum for an experimental drop size
volume density distribution was measured first for sample A distribution corresponding to a particular time. The
within 8 min of sampling and then again after it had been left parameters m, C, and Dc were varied in the ranges C ∈
standing for 2.5 h, while sample B was assessed for the drop size
[-4.0,4.0] in steps of 0.5, m ∈ [-1.0,2.5] in steps of 0.5, and
distribution only once, after 8 min of sampling. The distributions
are shown in Figure 6 and are nearly identical. This indicates Dc ∈ [3.0,81] µm in steps of 3.0 µm, and the set of values
the consistency of the sampling method as well as the stability at which the error is a minimum was obtained for the
of the emulsion after it has been diluted and transferred to the distribution. This was repeated for each experimental
slide. The third test consisted of shearing an emulsion of 1% distribution in a run and for all runs corresponding to
holdup in the tangential Couette flow apparatus for a period of various shear rates (for a fixed holdup). The arithmetic
2.5 h. Samples were then withdrawn from the sampling port and averages of the values obtained for all samples for a
also from the top and bottom of the apparatus. The volume density particular holdup were used as the final values of C, m,
distributions for each of the samples were found to be nearly the and Dc. Thereafter, the computation was repeated with
same (Figure 7). This shows that the shear rate along the length these average values of m, C, and Dc and the value of R0
of the cylinder is uniform and that creaming and settling of the
drops during the course of the experiment are negligible.
was recalculated. For a given shear rate, the average R0
was the mean of the values obtained (for constant m, C,
Theoretical Analysis and Dc corresponding to that particular holdup) calculated
over the four samples for that particular shear rate and
The time evolution of the drop size distribution for a holdup.
purely coalescing system is given by the population balance
equation Results and Discussions
We first present the experimental results and discuss
∂n 1
)
∂t 2 ∫0νCs(ν,ν-ν′) dν′ - ∫0∞Cs(ν,ν′) dν′ (6) the significant qualitative features of the data. The model
predictions and their implications are discussed next.
where n(ν) is the number density of drops of volume ν and The time evolution of the drop size distributions for the
Cs is the Smoluchowski coalescence rate given by eq 1 for three holdups studied are shown in Figures 8-10, and
drops of size ν and ν′. An assumption in the above equation the volume-averaged drop diameters for all three holdups
is that only pairwise collisions are considered. If all are shown in Figure 11. Each figure shows data for three
collisions result in coalescence,  ) 1. However, as will be different rotational speeds used. Some trends are common
seen below, the experimental results indicate that  , 1 to all the cases, and we list these first. The peak in the
and the efficiency values also depend on the sizes of the drop size distributions falls and the tail lengthens with
colliding drops. Here we assume the following expression increasing time of shearing. The rate of coalescence
for the coalescence efficiency, which is a generalization of decreases with increasing shear rate, and the coalescence
that proposed by Mousa and van de Ven,20 rate is highest at the lowest shear rate. Both features are
evident from the extent of change of the distributions with

[ ][ ( )] [ ( )]
C Di 2 -m Dj 2 -m time for the different shear rates as well as from the slopes
4q
 ) R0 1+ 1- 1+ 1- of the average diameter versus time curves.
(1 + q)2 Dc Dc There are also significant differences in the results for
(7) the different volume holdups studied. From Figure 10 it
can be seen that the most significant difference is the
The first two terms on the right-hand side of the expression formation of the second peak at larger drop sizes in the
are the same as those assumed by Mousa and van de Ven.20 high-holdup case (φ ) 15%) whereas at φ ) 5% (Figure 9)
The second term captures the dependence of the coales- just a shoulder appears. At, φ ) 1% there is simply a
cence efficiency on the ratio of the diameters (q ) Di/Dj) smooth tail. The rate of coalescence also increases with
of the colliding drops. The value of the exponent C indicates holdup, and this is most clearly evident from the time
the strength of the dependence of the coalescence efficiency variation of the average diameters. However, the rate of
2652 Langmuir, Vol. 17, No. 9, 2001 Nandi et al.

Figure 8. Time variation of the volume density distributions Figure 9. Time variation of the volume density distributions
for various shear rates at 1% holdup: (symbols) experimental for various shear rates at 5% holdup: (symbols) experimental
data, error bars show standard deviation over 3 runs; (solid data, error bars show standard deviation over 3 runs; (solid
lines) predictions of the population balance model using lines) predictions of the population balance model using
parameters given in Table 2 and Figure 14. parameters given in Table 2 and Figure 14.

increase is not proportional to φ2 (as predicted by Smolu- were scaled using the analysis of Swift and Friedlander.19
chowski6 theory) but slower. Thus, the coalescence ef- A plot of the cumulative oversize number fraction (1 -
ficiency decreases with increasing holdup. Finally, from Nνi/NT) versus the normalized drop volume (νi/〈ν〉) for φ )
Figure 11 it can be observed that the nature of the average 1% and for three rotation speeds is shown in Figure 12
diameter versus time curves are different. For φ ) 1%, along with that for φ ) 15% for the lowest shear rate.

the variation is almost linear with time, and this is Here, Nν ) ∑k)1i
nk, NT ) ∑i)1 ni, and 〈ν〉 ) NT/φ. For the
indicative of a coalescence efficiency that is independent lowest holdup (φ ) 1%) the scaled distributions at different
of the sizes of the colliding drops, as shown in Nandi et times of shearing superimpose (Figure 12a-c), showing
al.23 However, at φ ) 5%, the rate initially rises with time that the distributions are self-similar and that the
and then slows down toward the end. For the dispersed- coalescence efficiency is independent of drop size. However,
phase holdup φ ) 15%, the rate of coalescence goes up the curves for φ ) 15% and γ ) 10.8 s-1 (Figure 12d) do
with time for the entire period of the experiment. This is not scale, indicating that the coalescence efficiency in this
evident from the increasing slope of the plot showing the case is dependent on drop size.
nondimensionalized volume-average diameter with time. The solid lines in Figures 8-10 are the model predictions
These two cases indicate a dependence of the coalescence obtained using the fitted parameters. The predictions
efficiency upon the diameters of the colliding drops. match the experimental distributions within the limits of
A reduction in the coalescence rate at later times (as experimental error. Note that the development of the
seen from the 5% holdup) was also observed by Mishra et shoulder for the 5% emulsions and the formation of the
al.18 in their experimental results with a surfactant-free second peak for the 15% holdup are well-described by the
emulsion system. The slowing was attributed to the model. The prediction of the variation of the volume-
flattening of the surfaces of the colliding drops, resulting average drop diameter with time also matches well with
in a slow rate of film thinning.7 This does not occur during experimental data for all cases except for the 5% holdup
the initial stages of the process when drops are small, at long times.
since interfacial tension prevents significant flattening. The values of the fitted parameters (C, DC, and m) along
A similar effect could be active in the present system as with the standard deviations for each value of the
well. Possible reasons why this effect is not seen in the dispersed-phase holdup are given in Table 2. The values
lower and higher holdup systems are as follows. In the 1% of these parameters obtained from fitting the model to
holdup system the drops never become large enough. In the experimental distributions at different time steps
the 15% holdup case, chaining of drops due to shear flow (during any single run) were nearly the same. The values
may provide the extra force to overcome the additional were also found to be nearly independent of shear rate at
resistance to thinning of the film due to flattening of the a particular holdup. The parameter values provide an
larger drops. More detailed drop-level studies, however, insight into the coalescence at the scale of individual drops.
are required to confirm these explanations. At the lowest volume holdup, C ) 0 and m ) 0, implying
The drop size distributions for the case with φ ) 1% that the coalescence efficiency is independent of the sizes
Coalescence in Surfactant-Stabilized Emulsions Langmuir, Vol. 17, No. 9, 2001 2653

Figure 10. Time variation of the volume density distributions Figure 11. Variation of the nondimensionalized volume-
for various shear rates at 15% holdup: (symbols) experimental averaged diameters with time for various shear rates (corre-
data, error bars show standard deviation over 3 runs; (solid sponding to each holdup as shown in Figures 8-10).
lines) predictions of the population balance model using
parameters given in Table 2 and Figure 14. rate. Further, R0γ decreases with increasing holdup,
indicating that the probability of coalescence is lower in
of the colliding drops and depends only on the shear rate, the case of higher holdups. Thus, the coalescence rate
as reported earlier.23 With increasing holdup, the pa- increases with holdup but at a rate that is lower than the
rameter C decreases while m increases. The critical drop φ2 proportionality, as suggested by Smoluchowski’s theory.6
diameter is found to be about 18 µm for both 5% and 15% The magnitude of coalescence efficiency obtained is 2
holdup. A lower value of C indicates a higher probability orders of magnitude lower than that predicted by Ches-
of coalescence for drops of unequal sizes. The coalescence ters’7 model.
efficiency attains a maximum (since m is positive) at the Experiments were also performed in which the emulsion
critical diameter (Dc). However, coalescence efficiency is was kept standing in a beaker and samples were with-
a product of these two terms, and the variation of the drawn at specified times and the size distributions
coalescence efficiency with the diameters of the colliding analyzed. For the case of 1% holdup, the extent of increase
drops is shown in Figure 13 for the highest dispersed- of the volume-averaged diameter was slightly higher (D h /D
h0
phase holdup. This figure shows that coalescence is most ) 1.337 after standing for 4.5 h) than that at the lowest
likely when one of the pair of colliding drops is close to the shear rate studied. For the holdup of 5%, the dimensionless
critical diameter and the other is small (∼2.0 µm). While volume-averaged diameter increased to 3.7 after standing
the existence of an optimal drop size for coalescence has for 4.5 h, which is much larger than the result for the
been proposed earlier,30 it is not clear whether the same lowest shear rate. In the 15% holdup case, the emulsion
effects are applicable here. We speculate that the observed separated into two distinct layers on standing for 0.5 h,
maximum could be the result of long-lived clusters of one the bottom layer being the dispersed phase, as its specific
large and several small drops which eventually coalesce. gravity was higher than that of the continuous phase by
The clusters comprising 18 µm and several 2 µm drops 0.003 g/cm3. The high coalescence rate in standing
are perhaps the most stable under shear flow. We note emulsions is due to the formation of long-lived aggregates
that a bimodal distribution is not produced if m ) 0 (which of drops. The thin films in these aggregates have sufficient
corresponds to the model of Mousa and van de Ven20), for time to rupture. Weak Brownian motion and small flow
any value of the exponent C. Thus, the existence of a critical currents generated due to handling cause the drops to
diameter seems essential to explain this feature of the collide and form aggregates which can be seen under the
results. microscope if a sample is carefully taken. Formation of
The shear rate dependence of the coalescence rate is such aggregates was reported by Bibette et al.31
given by the product of R0 with the shear rate (γ) (eq 3). The reduction in the coalescence efficiency is primarily
Figure 14 shows that the coalescence rate decreases due to the fact that, with higher shear rates, the interdrop
monotonically with shear rate, indicating that coalescence contact time (tc) goes down approximately as the inverse
efficiency decreases faster than the inverse of the shear of the shear rate.7 As a result of this, the time available
for the film between two drops to rupture is reduced as
(30) Kumar, S.; Kumar, R.; Gandhi, K. S. Chem. Eng. Sci. 1993, 48,
2025. (31) Bibette, J.; et al. Phys. Rev. Lett. 1992, 69, 981.
2654 Langmuir, Vol. 17, No. 9, 2001 Nandi et al.

Figure 14. Variation of R0γ with shear rate (γ) for different
holdup emulsions: (symbols) values back-calculated from
experimental distributions; (solid line) prediction of eq 10.

Table 3. Back-calculated Dimensionless Contact Times


(γtc) for Colliding Drops Based on the Results Given in
Figure 14
φ γ ) 10.8 s-1 γ ) 21.5 s-1 γ ) 43.0 s-1
1% 1.00 1.00 1.00
5% 0.61 0.75 1.04
15% 0.54 0.64 0.83
Figure 12. Scaled cumulative oversize distribution for different
shear rates and holdups. The distributions are self-similar for be stochastically distributed, with a log-normal distribu-
the emulsion having φ ) 1% tion25,26
.

( [
1 ln tb - µb
])
2
1
P(ln tb) ) exp - (9)
σx2π 2 σb

Assuming that the interdrop contact time is tc ) 1/γ7 for


dilute emulsions (φ ) 1%), the coalescence efficiency is
given by

R0 ) ∫-∞ln t
c
P(ln tb) d ln tb )
1
2
erfc
{
µb + ln γ
x2σb } (10)

The above equation was fitted to the values of R0 obtained


for φ ) 1% emulsions and for different shear rates to yield
µb ) 4.35 and σb ) 1.99. The solid line in Figure 14 shows
Figure 13. Variation of the normalized coalescence efficiency that the model adequately describes the experimental
(/R0) with diameters of the colliding drops (Di,Dj) obtained from data. In the case of high holdups, although film stability
eq 7 for the 15% emulsion (Table 2). may be well-described by the log-normal probability
density (eq 9), the average contact time is difficult to
estimate independently. Fitting eq 10 to the data for R0
Table 2. Fitted Values of the Parameters C, m and Dc for
the Three Holdups Studied (Refer to Eq 7) and Standard given in Figure 14 for the higher holdups yields estimates
Deviations of Each of the Parameters over the Three of the average contact times (tc). The results obtained for
Shear Rates for Each Holdup different shear rates are given in Table 3. The increased
φ C M Dc (µm)
stabilization (compared to that for the lower holdups) may
be due to multidrop collisions which reduce the contact
1% 0.0 0.0 times as shown in Table 3. A quantitative understanding
5% -1.5 ( 0.43 0.5 ( 0.5 18.75 ( 2.54 of such phenomena is complicated, and as yet models to
15% -2.0 ( 0.90 1.75 ( 0.66 18.00 ( 3.5
calculate parameters such as average contact time are
unavailable in the literature.
The lack of dependence of the coalescence efficiency on
the shear rate is increased and thus the probability of the sizes of the colliding droplets, for the case of a
coalescence is also reduced. A model based on this dispersed-phase holdup of 1%, can be explained from the
mechanism was presented in ref 23. The time (tb) for model of Chesters7 for droplets with immobile interfaces.
which the film between the drops survives is assumed to Calculations show that, on reducing the critical thickness
Coalescence in Surfactant-Stabilized Emulsions Langmuir, Vol. 17, No. 9, 2001 2655

at which the film can rupture, the coalescence efficiency ing the dependence of the coalescence efficiency on the
becomes weakly dependent on the sizes of the colliding size of the coalescing droplets.
species. In the case of our emulsion system, the critical A population balance model was used to describe
thickness for the rupture of the film is low, since the film transient evolution of the experimentally obtained drop
is stabilized by the presence of surfactant. Further, only size distribution and to back-calculate the coalescence
“near head-on” collisions give contact times long enough efficiency. The coalescence efficiency was found to decrease
to allow for film rupture and coalescence. The contact time with increasing shear rate in all the cases. At low holdup
may be divided into two regimes. Initially, the film thins (1%) the coalescence efficiency was found to be independent
rapidly, and thereafter, the film thinning becomes very of drop size, in agreement with the scaling result. At high
slow. The second part may be termed as the “drop rest holdups the coalescence efficiency exhibited dependence
time”. For near head-on collisions, the period for the drop on colliding drop sizes and the highest efficiency was
rest time is significantly higher than the time for which obtained for a pair of drops with one drop size being about
the film thins. The rate of film thinning is dependent on 18 µm and the other approximately 2 µm. The coalescence
the sizes of the colliding species, however, the drop rest efficiency was found to reduce significantly with increased
time is not. Thus, for collisions in which the time of holdup. Thus, the coalescence rate increases at a rate
thinning is small compared to the drop rest time, the slower than φ2.
dependence of the coalescence efficiency on the drop sizes An estimate of the coalescence efficiency was obtained
becomes insignificant. At higher dispersed-phase holdups, considering the process of film rupture to be controlling
hydrodynamic, multibody interactions are likely to affect and assuming a log-normal distribution of the rupture
the interdroplet contact times and the situation of a size time. The analysis gave a good description of the variation
independent coalescence efficiency will therefore not hold, of coalescence efficiency with shear rate at low holdups.
as has been observed in this work. With increasing holdups, the drop-drop contact times
were found to decrease, indicating the importance of
Conclusions interdrop interactions.
This work demonstrates an experimental method for
A neutrally buoyant emulsion, stabilized by surfactant back-calculating the shear-induced coalescence efficiency
and containing an electrolyte, was subjected to a tangential for emulsions, for which quantitative models are as yet
Couette flow in the simple shear regime. The drop size unavailable. The experimental results and analysis
distributions were measured using optical microscopy and provide a qualitative insight into drop level processes
image analysis. Studies were conducted for varying shear during shear-induced coalescence. More detailed models
rates and holdups. An increasing shear rate was found to and computations including multidrop interactions are
stabilize the emulsion in all the cases. Increasing the perhaps required to obtain a quantitative description of
dispersed-phase holdup gave rise to a bimodal distribution, the process.
unlike in the case of the lowest holdup (1%), where the
Acknowledgment. The financial support of the BRNS
drop size distribution curves showed an elongating tail
through Project No. 36/7/94/R&D-II is gratefully acknowl-
with time and a diminishing peak. On scaling the
edged. D.V.K. acknowledges the financial support from
experimental distributions, it was noted that the distri-
DST, India, through the Swarnajayanti fellowship. The
butions of samples withdrawn at various times for the
authors thank Dr. B. K. Das and Dr. P. D. Sawant of the
case of 1% holdup showed self-similarity, indicating that
Alchemie Research Center, Bombay, for the help in
the coalescence efficiency was independent of drop size.
measuring the interfacial tension values.
A similar graph for the case with 15% holdup showed a
wide difference between the scaled distributions, indicat- LA001473M

You might also like