Analysis of Surfactant Laden LiquidLiquid

Dispersion Using an Axisymmetric Laminar Jet

Justin R. Walker and Richard V. Calabrese*
Department of Chemical and Biomolecular Engineering, University of Maryland, College Park, MD 20742-2111, U.S.A.

The contacting of multiple liquid phases is a complex process; and one that is often difcult to study experimentally. However, many multiphase
mixers function by forcing an immiscible uid through an orice or nozzle, which can be modelled using an axisymmetric jet. Despite the extensive
literature on jets, there is little information on the role played by surfactants in determining jet breakup length and drop diameter.
Experiments involving the discharge of oil jets into otherwise quiescent aqueous surfactant solutions were performed over a broad range of
conditions. Jet length was found to increase with surfactant concentration, while droplet diameter was found to decrease (dependent on jet
regime). These results compare favourably with, and can aid the interpretation of similar experiments in high shear mixers.

Keywords: multiphase mixing, surfactants, liquidliquid dispersion, jet breakup, hydrodynamics

INTRODUCTION droplets can be obtained using an axisymmetric oil jet into an

aqueous system, and that these results are comparable to those

T
he contacting of multiple liquid phases to form emulsions
obtained using a Silverson L4R batch rotorstator mixer. There-
or dispersions is a complex process, and one that is difcult
fore, we are able to perform parametric studies of the effect of
to study experimentally, requiring expensive specialised
surfactants on liquid dispersions using the jet system and use our
equipment, and difcult and lengthy cleaning and decontami-
results to better understand the high shear mixer system. The
nation procedures, due to the complex geometry of the mixing
results of parametric jet studies are presented and discussed, and
devices. It is therefore desirable to compliment experiments on
then used to interpret previous mixer results.
mixers with a geometrically and hydrodynamically simpler ana-
logue on which many experiments can be more easily performed.
Many multiphase contactors (such as high shear rotorstator JET BREAKUP THEORY
mixers, jet mixers and valve homogenisers) function by forcing
A cylindrical jet of uid is hydrodynamically unstable, having
an immiscible uid through an orice or nozzle, thereby destabil-
a larger surface area than a sphere of equal volume. As such,
ising deformed droplets. Many of these processes can be mimicked
surface tension forces cause the jet to break up, and the dynam-
on a fundamental level by the discharge of an immiscible axisym-
ics of the system drive the jet to yield a series of spheres, rather
metric laminar jet. Despite extensive literature on jets, there is
than a single one. According to Rayleigh (1879), if the column
little information on the effect of surfactants on jet breakup
becomes varicose with wavelength
exceeding the circumference
(Eggers and Villermaux, 2008). Figure 1 shows RANS (FLUENT)
2r of the cylinder, the cylinder will decay into droplets such that
simulation results for a Silverson L4R batch rotorstator mixer
each uid node comprises one drop. Figure 4 shows the relevant
equipped with a slotted stator head and 4-bladed rotor operating
geometry.
at 4000 rpm (reproduced from Yang and Calabrese, 2009). The
geometry of the mixing head is shown in Figure 2. A strong jet is
observed discharging from one of the stator slots. Figure 3 shows
a Volume of Fluid FLUENT simulation from our study showing Author to whom correspondence may be addressed.
a laminar liquid jet discharging from a circular capillary tube E-mail address: rvc@umd.edu
into a second immiscible phase. These ow patterns show strong Can. J. Chem. Eng. 89:10961103, 2011
qualitative similarities. 2011 Canadian Society for Chemical Engineering
DOI 10.1002/cjce.20624
This study shows that reliable results for the effect of aqueous- Published online 22 July 2011 in Wiley Online Library
phase non-ionic surfactants on the breakup of liquid threads and (wileyonlinelibrary.com).

| 1096 | THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING | | VOLUME 89, OCTOBER 2011 |
 If the breakup of an inviscid uid cylinder is assumed to be
driven only by capillarity (e.g. the inuence of the outer uid is
ignored), and we work in the jets reference frame (e.g. the jet
velocity in the axial direction is zero), the NavierStokes equa-
tions may be linearised and rearranged to form Equation (1) (see
Eggers and Villermaux (2008) for a detailed derivation):

 I1 (kr0 )
2 = (kr0 )[1(kr0 )2 ] (1)
r03 I0 (kr0 )

The most unstable wavenumber is found by locating the value

of k which maximises .
No longer assuming the jet to be inviscid, a similar derivation
results in Equation (2) (Eggers and Villermaux, 2008):

 I1 () 3 k 2
2 = 3
[12 ] (2)
r0 I0 () 

Equation (2) can be recast in dimensionless form:

Figure 1. FLUENT RANS simulation results for a rotorstator
mixer velocity vectors. Colour scale indicates mean velocity magnitude I1 ()
w2 = (12 ) 32 Oh w (3)
in m/s. White edged black voids are the solid stator teeth. Rotor I0 ()
speed = 4000 rpm, rotor diameter = 28.12 mm, ReI = 5.23 104 .
Drop size is predicted by nding the value of  which maximises
w (known as max ), and then using Equation (4), which is derived
from a simple geometric reduction. The initial jet diameter djet,i is
equal to the capillary inner diameter:

1/3
ddrop 3
D = = (4)
djet,i 2max

The other relevant quantity when discussing jet breakup is the

breakup length. Using the instability growth rate found above,
coupled with knowledge of the jets geometry and velocity, an
estimate of the time until the instability grows to exceed the jet
Figure 2. Details of mixing head from a Silverson L4R batch rotorstator diameter can be made. This quantity is known as the breakup
mixer with slotted stator head and 4-bladed rotor. time. From the breakup time, the breakup length can be calcu-
lated from knowledge of the jet velocity. Das (1997), working
from the earlier results of Rayleigh (1879), derived the following
expression for jet breakup length Lb :


Un r0
Lb = ln (5)
0

Since the value of 0 cannot be determined experimentally,

Das (1997); and Scheele and Meister (1968) both found that
using ln(r0 /0 ) = 6 successfully correlated breakup lengths for
liquidliquid systems. For this prediction, the value of from
Equation (1) or (2) is used, as appropriate. When is made
dimensionless using w and Un using We1/2 , the dimensionless
breakup length Lb is found by dividing by the initial jet diameter:


We 1/2 r0
Lb = ln (6)
Figure 3. FLUENT 2D axisymmetric Volume of Fluid simulation results for 2w 0
a laminar jet velocity vectors. Colour scale indicates velocity magnitude
in m/s. Capillary inner diameter is 0.84 mm. Re = 25, We = 2.
This expression is similar to other expressions for jet breakup
length which also show a Weber number to the one-half power
dependence. Middleman (1998) proposes Equations (7a) and (7b)
for an inviscid and viscous jet, respectively. The constant C is
entirely empirical:
Figure 4. Schematic diagram of laminar axisymmetric jet with relevant
geometry.
Lb = C We 1/2 (7a)

| VOLUME 89, OCTOBER 2011 | | THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING | 1097 |
Figure 5. Predicted effect of jet velocity on breakup length of a
liquidliquid jet. Results are from a direct numerical simulation utilising a
front-tracking volume of uid method, utilising approximately 1 million
nodes. Reproduced with permission from Homma et al. (2006).


We 1/2
Lb = C We 1/2
1+ (7b)
Re

Homma et al. (2006) performed a direct numerical simulation

of a jet of one liquid discharging into a second immiscible liquid,
similar to the current experimental study. Their results showed
three distinct regimes, as seen in Figure 5. At low ow rate (or
Re), dripping occurred, where droplets are produced one at a time
at the capillary tip. Above a critical ow rate (indicated by the
rst dashed line), a stable jet begins to form, which increases in
length with increasing jet velocity (as predicted by Equation 5).
Jets in this regime exhibit axisymmetric wave patterns. However,
another transition occurs where asymmetric waves begin to occur
(labelled 3-Dimensional Flow in Figure 5), which destabilise the
jet and reduce its breakup length. The jet ultimately transitions
into atomising ow at very high jet velocity (well beyond the
range used in this study). Sample images taken from the high
speed camera footage in our study illustrate these regimes, and
are presented in Figure 6.
In our study, theoretical predictions are only made for clean-
water systems (ones without surfactants). This is due to the fact
that the Ohnesorge and Weber numbers cannot be dened for a
system where the interfacial tension is a function of interfacial
age and interfacial surfactant concentration. Currently available
jet breakup theory is unable to reconcile the dynamic effect of
surfactants.
Figure 6. Sample images from the three jet breakup regimes. Each row
Experimental Setup and Procedure represents a series of snapshots of the same experimental conditions at
different times. Note the variability of breakup length in the
A schematic diagram of the experimental apparatus is shown in axisymmetric jet case.
Figure 7. Silicone oil (Dow Corning 200 Fluid) with a nominal
viscosity of 10 cSt (9.8 cP) and density of 935 kg/m3 was injected windows and lled completely with deionised water (or aqueous
into otherwise still deionised water or aqueous surfactant solu- surfactant solutions). The uid reservoir was lled with deionised
tions. Three capillaries were used to form the jets, having inner water in order to maintain the calibration of the rotameter, and the
diameters of 0.42, 0.60 and 0.84 mm, respectively. The capillaries secondary reservoir at the bottom of the test cell was lled with
were all 51 mm long; sufcient to assure fully developed laminar silicone oil. In this manner, water owed from the uid reservoir,
ow proles at the capillary tip. A 1 L uid reservoir, maintained through the metering valve and ow meter into the secondary
at constant pressure through the use of a regulated nitrogen gas chamber, where it displaced oil and slowly lled the secondary
cylinder, was used to initiate the ow. The ow rate of the jet chamber. The displaced oil was forced out of the capillary, creating
phase was controlled using a needle valve and monitored using an oil jet in the upper water-lled primary chamber. An overow
a rotameter. Flow rates were varied from 5 to 30 mL/min, corre- keeps the cell chamber at constant pressure. The entire apparatus
sponding to Reynolds numbers from 20 to 2000, depending on the was placed on top of a pneumatic isolation table to prevent any
capillary and uid in use. The test cell was tted with watertight ambient vibrations from being transmitted to the jet surface.

| 1098 | THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING | | VOLUME 89, OCTOBER 2011 |

Figure 7. Schematic diagram of experimental apparatus.
Two distinct CCD cameras were used to image droplets. A low
frame rate (30 images/s) Pulnix TM-1405GE, high resolution cam-
era was used to measure breakup length and to quantify droplet
population statistics. A high frame rate (up to 20,000 images per
second) Vision Research Phantom V9, moderate resolution cam-
era was employed to observe detailed breakup phenomena. High
quality Nikkor SLR (F-mount) lenses of various focal lengths
were utilised with both cameras. For lighting, a diffuse uores-
cent lamp with an 85 kHz driver (StockerYale ILT06S3KDDTC)
was utilised to ensure even backlighting and eliminate the usual
60 Hz oscillations in lighting intensity common with AC lamps.
An automated image processing algorithm was used to acquire
droplet geometry and population statistics. An automated algo-
rithm developed in ImageJ is employed to analyse the images
and report the equivalent spherical diameter of the droplets. A
similar automated algorithm is utilised to calculate the breakup
length. Each presented measurement is based on a minimum
of 1000 images, with a total of up to 10,000 individual droplet
measurements.
The non-ionic surfactant used in this study was octyl phenol
ethoxylate (hereinafter referred to by the trade name Triton X-
100). This surfactant is insoluble in silicone oils and was added to
the aqueous phase, meaning surfactant must diffuse from within
the surrounding matrix phase to the oil jet interface. Physical and
interfacial properties of this surfactant were determined using the
Pendant Drop method, and are listed in Table 1. The equilibrium
interfacial tension between silicone oil and the aqueous surfactant
decreases linearly with the natural logarithm of the surfactant
concentration until the critical micelle concentration (CMC) is
reached, after which it is constant. Further details of the pendant
drop technique, including the measurement procedure, are given
by Padron (2004).
RESULTS AND DISCUSSION
Clean (Surfactant-Free) Systems
Figure 8 shows the effect of Reynolds number on the dimension-
less breakup length for a 10 cSt silicone oil jet discharged into
deionised water. The line shown in this and subsequent plots is a
| VOLUME 89, OCTOBER 2011 | |
Table 1. Physical and interfacial properties of aqueous surfactant
system
Surfactant Triton X-100
Critical micelle concentration (CMC) 2.19 104 mol/L
Diffusion coefcient 5.02 1010 m2 /s
Interfacial tension of clean system,
0 36 mN/m
Equilibrium interfacial tension at CMC,
4 mN/m
trend line. At low Re, these jets are in the dripping ow regime,
meaning individual drops are produced one by one on the tip of
the capillary. As such, the measured jet length will periodically
grow from nearly zero to the major axis of the hanging large drop
until it detaches. These droplets are generally many times the
diameter of the capillary on which they form, which results in
an average dimensionless breakup length on the order of 510.
Once the Reynolds number exceeds about 50 (although some-
what dependent on the capillary diameter), a stable jet forms.
The length of this jet then decreases with increasing Reynolds
number. This agrees with the results seen in the direct numeri-
cal simulation of a liquidliquid jet published by Homma et al.
(2006) (Figure 5). Homma explained that at Reynolds numbers
below the maximum jet length, the waves present on the jet sur-
face are primarily axisymmetric, and the jet length will increase
rapidly with increasing ow rate. At the maximum jet length,
asymmetric waves start to appear on the surface of the jet, pri-
marily in the form of periodic wake type structures (see Figure
6). These structures destabilise the jet and subsequently reduce
its length with increasing Reynolds number.
Figure 9 shows the effect of Reynolds number on the droplet
size. Droplet sizes are represented by the dimensionless Sauter
mean diameter, D32
, which strongly weights the mean toward
the larger droplets, ignoring the inuence of satellite drops. This
is desirable since experiments in the dripping ow regime will
Figure 8. Effect of Reynolds number on dimensionless breakup length
for a 10 cSt silicone oil jet discharged into deionised water.
Corresponding Weber numbers are indicated on the top axis. Closed
symbols indicate stable jet formation. Open symbols indicate dripping
ow or unstable jet formation. Error bars indicate one standard deviation
in the measured values.
THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING | 1099 |

Figure 9. Effect of Reynolds number on droplet size for a 10 cSt silicone
oil jet discharging into deionised water. Closed symbols indicate a jetting
ow regime (stable or unstable). Open symbols indicate a dripping ow
regime. Error bars indicate one standard deviation in the measured
values.
not produce satellite drops. Using the Sauter mean diameter
allows dripping ow and jetting ow experiments to be compared
directly.
Figure 9 shows that while long jets were not found to occur
until around Re = 50, the mean droplet size resolves to smaller
Figure 10. Effect of surfactant concentration on jet breakup length. Surfactant concentration is normalised by the critical micelle concentration (CMC).
(A) Results using the 0.41 mm capillary, (B) the 0.60 mm capillary and (C) the 0.84 mm capillary.
| 1100 | THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING | |
size droplets at lower Rearound 25. Further analysis of the high
speed video imagery revealed that in this intermediate regime
(25 < Re < 50), a very short, unstable jet formed that was of com-
parable length to the large hanging droplets found in the dripping
ow regime. These short jets produced small droplets similar in
size to those produced by the longer stable jets. In the stable jet-
ting ow regime, droplet size was correlated to jet length, with
longer jets producing larger droplets, and droplet size decreas-
ing as jet length decreased at higher Reynolds number. This is
contrary to the linear breakup theory presented above, which
predicts that droplet size should be independent of jet velocity.
However, our experimental observations reveal that jet breakup is
not driven exclusively by capillarity and that shear induced waves,
due to the viscosity of the surrounding aqueous phase, also play a
role in determining the ultimate breakup frequency and breakup
time. From the high speed video, it appears as though shear layer
instabilities form very near the capillary tip. The effect of these
shear layer instabilities is strongly dependant on the jet length.
For longer length jets, these instabilities are damped out by vis-
cous dissipation and the jet breakup is dominated by capillary
(Rayleigh) instabilities. For very short jets, the breakup is domi-
nated by shear layer instabilities. For jets of moderate length, the
dominant instabilities are a superposition of the two waves.
Surfactant Laden Systems
Figure 10 shows the effect of surfactant concentration on the
breakup length of a 10 cSt silicone oil jet discharged into aqueous
solutions of Triton X-100. The surfactant concentration of 1E4
VOLUME 89, OCTOBER 2011 |
represents the deionised water case. Since the Reynolds number
poorly correlates the experiments performed with different cap-
illary diameters, the gures are separated by capillary diameter.
Results using the smallest capillary are presented in Figure 10A,
the intermediate capillary in Figure 10B and the largest capillary
in Figure 10C.
Figure 10A shows the effect of surfactant concentration on
breakup length for the experiments with the largest Re. As pre-
dicted by Figure 5, the intermediate ow rate (circular symbols)
produces the longest jet due to the presence of asymmetric waves
on the higher velocity jet (triangles).
In all cases, a pronounced increase in jet breakup length is
noted at increased surfactant concentration. Jet length is almost
constant at concentrations below the CMC, and then increases
strongly with concentration above the CMC. The primary reason
for this is that surfactant concentration gradients on the jet sur-
face become more pronounced. This increases the magnitude of
the Marangoni stresses, which act to oppose any change in the sur-
face area. Marangoni stresses arise when new interface is created
(in our case due to the formation of a surface wave). These new
interfaces do not contain surfactant, causing gradients in inter-
facial tension between regions of old surface area (which has
a certain amount of surfactant adsorbed) and this new surface
area. The Marangoni stress can either be relieved by this new sur-
face area being removed (i.e. the surface returning to its previous
shape) or by the new surface becoming populated with an equal
concentration of surfactant diffusing from the bulk. Thus, when
a capillary wave attempts to cause the jet to break, the Marangoni
stress resists the capillary forces and maintains the jets diameter.
As the capillary wave grows in magnitude further downstream,
it is able to overcome the Marangoni stress; but this results in a
jet which is signicantly longer than one which does not have
sufcient surfactant coverage. Additionally, increasing surfactant
concentration downstream on the jets surface causes a decrease
in interfacial tension, which increases the Weber number. As seen
in Equation (6), increasing Weber number should result in longer
jet breakup lengths.
Since surfactant must diffuse from the surrounding aqueous
phase to the jet surface, the effect of the bulk surfactant con-
centration strongly inuences the jet breakup phenomena even
at concentrations exceeding the CMC, where the static interfa-
cial tension will not change. The effect is synergistic in the sense
that increasing surfactant concentration allows more surfactant
to reach the jets surface in the short breakup time (the increased
concentration gradient results in a larger mass transfer ux from
the bulk to the surface). This, in turn, stabilises the jet, increasing
the breakup time, which allows for yet more surfactant to diffuse
to the surface, which further increases the breakup time, and so
on. This is the reason for the very rapid increase in breakup length
with increasing surfactant concentration. Clearly, at some point a
plateau should be reached when the jet surface becomes saturated
with surfactant so rapidly that further increases in surfactant con-
centration no longer have any benecial effect on the jet breakup
length. It appears as though that plateau concentration has nearly
been reached in certain conditions (e.g. Figure 10C), as the dimen-
sionless breakup length did not increase as much between the 20
CMC to 50 CMC concentration range as for the 1 CMC to 10 CMC
and 10 CMC to 20 CMC ranges. Testing at higher concentrations
is not possible with the current experimental setup as at higher
concentrations the continuous phase starts to become opaque as
the concentration approaches the cloud point of the surfactant.
One of the most interesting results is observed most clearly in
Figure 10. At low surfactant concentration, at the lowest ow rates
| VOLUME 89, OCTOBER 2011 | |
tested for each capillary (indicated by the black square symbols
on plots A, B and C), the jets fall within the dripping ow regime,
as evidenced by the extremely short jet lengths (producing results
similar to the top row of Figure 6). However, once the surfactant
concentration exceeds the CMC, the jet is sufciently stabilised to
transition into the jetting regime, producing jets with dimension-
less breakup lengths similar to conditions at much higher ow
rates despite the fact that the Reynolds number is unchanged,
comparable to the middle row of images in Figure 6.
Another interesting result is that for the largest jet at its highest
ow rate (most easily observed in Figure 10C, and the bottom
row of images in Figure 6), the effect of surfactants is much
less pronounced. Upon examining the high speed video of these
experiments, there were strong shear layer instabilities and asym-
metric waves present. Unlike the axisymmetric capillary waves,
the asymmetric waves do not signicantly change the interfa-
cial area, and so are not as strongly opposed by the Marangoni
stress. Once the asymmetric waves gain strength, they are able
to induce strong recirculation zones in the outer uid which tear
large chunks off of the jet. These recirculation zones do not appear
to be signicantly opposed by the Marangoni stress and so, under
these conditions, only a moderate increase in jet length (with a
signicant amount of data scatter) is observed. A similar, but less
pronounced, effect is observable in Figure 10B when looking at
the highest ow rate for this 0.6 mm capillary. The image analysis
in this case showed some asymmetry to the jet, but not as much
as with the 0.84 mm capillary.
Figure 11 shows the effect of surfactant concentration on the
dimensionless Sauter mean diameter of droplets produced by
the breakup of a 10 cSt silicone oil jet discharged into aqueous
solutions of Triton X-100. Once again the 1E4 concentration
represents the deionised water case. For the larger capillary diam-
eters, dripping ow was observed at the lowest ow rate (see
Figure 11B,C), resulting in the production of very large droplets.
As surfactant concentrations increase, these jets transition into
the jetting regime and produce much smaller droplets. For jetting
phase breakup, droplet size was found to increase with surfactant
concentration. The same stabilisation mechanism that allows the
jets to grow to such large lengths is responsible for these large
droplets. Downstream of the breakup event, the troughs of the jet
waves are suppressed due to the Marangoni stress. Instead of the
drop forming on the tip of the jet pinching off, this causes the
second uid node to be pushed into it, doubling the size of
the resulting drop. Under certain conditions, several uid nodes
can be pumped into the forming droplet in this manner until the
droplet, assisted by its own buoyancy, detaches from the jet. This
is much more like the droplet formation in a dripping ow mode
than standard jetting ow. As with the jet length, concentrations
above the CMC are required to effect signicant change in droplet
diameter.
For jets breaking up in the asymmetric regime, no signicant
change in droplet diameter is observed. Figure 11C clearly shows
that for the largest diameter jet, a slight increase in droplet diam-
eter is observed for the 20 mL/min ow rate, and no increase
is observed for the 30 mL/min ow rate. It appears as though
the uid thread detachment mechanism observed in the asym-
metric jetting regime is not sensitive to surfactant concentration.
It is also notable that this regime produces signicantly larger
drops than the axisymmetric jetting regime. This can be clearly
seen by examining the scale on the vertical axis in Figure 11,
where the 0.84 mm capillary produces droplets with dimension-
less diameters two to three times of those produced by the 0.41 mm
capillary.
THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING | 1101 |
Figure 11. Effect of surfactant concentration on droplet size. Surfactant concentration is normalised by the critical micelle concentration (CMC). (A)
Results using the 0.41 mm capillary, (B) the 0.60 mm capillary and (C) the 0.84 mm capillary.

Application to LiquidLiquid Dispersions in High high shear mixer. In our jet experiments, no signicant effect was
Shear Mixers observed with jets operating in the asymmetric jetting regime.
This implies that in the high shear mixer, a different breakup
Figure 12 (adapted from Padron, 2004) shows the effect of sur-
mechanism is occurring at high rotor speed than at lower rotor
factant concentration on the Sauter mean diameter of droplets
produced in a Silverson L4R batch rotorstator mixer equipped
with a slotted stator head (see Figure 2). When the mixer is
operating at low rotation speed (3000 rpm), an initial decrease in
droplet diameter is seen with surfactant concentration, up to the
CMC. Increasing surfactant concentration above the CMC resulted
in a slight increase in droplet diameter. This parallels the results
seen in our experiments, especially for moderate ow rates when
no ow regime transitions are observed (e.g. Figure 11B where
the minimum droplet size is observed just below the CMC). The
changes are more pronounced in our current experiment, which
is likely due to the fact that in the high shear mixer there is strong
turbulent convective transport of surfactant to the droplet, so the
rate limiting step for resolving Marangoni stresses is the transport
of surfactant through the drops boundary layer, while in our case
there is little convective surfactant transport due to the primarily
quiescent aqueous phase, so we must rely on diffusion through
the bulk aqueous phase.
When the mixer is rotating at a high speed (7000 rpm open
symbols in Figure 12), the results achieved in the rotor-stator
mixer for the 10 cSt oil again closely match the results shown Figure 12. Effect of surfactant concentration on droplet diameter for a
in the laminar jet; in this case the results for the high Re jets Silverson L4R batch rotorstator mixer. Closed symbols indicate slow
observed in Figure 11C. At all surfactant concentrations, there rotor speed (3000 rpm), open symbols indicate fast rotor speed
is essentially no effect on the droplet diameter observed in the (7000 rpm). Modied from Padron (2004).

| 1102 | THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING | | VOLUME 89, OCTOBER 2011 |
speed, and this secondary breakup mechanism is relatively insen-
sitive to surfactant concentration. In the case of the jet we observe
that for the asymmetric jetting case the bulk uid motion is what
drives breakup, while at lower jet velocities (the dripping and
axisymmetric jetting regimes), it is the dispersed phase motion
that is ultimately responsible for jet breakup.
CONCLUSIONS
This study has shown several signicant results. Consistent with
existing literature, three distinct ow regimes were observed
over the range of conditions studied. However, the traditional
prediction criteria (based on Weber number) are insufcient to
predict the transitions between these breakup modes because the
interfacial tension is a function of interfacial age and surfactant
surface excess concentration. Transitions between breakup modes
were observed when surfactant concentration was changed, even
though hydrodynamic conditions were held constant.
Increasing surfactant concentration was found to increase jet
breakup length for all cases studied, however the largest increase
was found at concentrations well above the CMC. This was
attributed to the short jet breakup time with respect to the time
required for surfactant to diffuse to the oilwater interface from
the bulk aqueous phase.
Current theory regarding jet breakup is insufcient to predict
droplet diameters or jet lengths in a system with surfactants, since
the Weber number varies along the length of the jet. Development
of a model to address this, utilising either an empirical effective
viscosity or interfacial tension term, or some form of diffusive ux
model, is essential for further quantication of the data.
The axisymmetric laminar jet has been shown to be a useful tool
for analysing surfactant laden liquidliquid ows. Similar results
have been observed in surfactant laden ows in rotorstator mix-
ers as with liquidliquid jets. At moderate ow rate or rotor speed,
droplet size was found to decrease with increasing surfactant con-
centration, with droplet size increasing (or levelling off) at higher
concentrations. At higher ow rates or rotor speeds, droplet size
was not found to change with increasing surfactant concentration.
In jets this was attributed to the fact that these high ow rate jets
broke up due to asymmetric waves which are dominated by bulk
phase motion.
NOMENCLATURE
3
CMC critical micelle concentration (mol/m )
D* dimensionless droplet diameter (ddrop /djet,i )
D32   2 Sauter mean diameter (D32 = (1/djet )
dimensionless
( di3 / di ))
ddrop droplet equivalent spherical diameter (m)
di diameter of droplet i (m)
djet,i initial jet diameter or capillary inner diameter (m)
I0 modied Bessel function of the rst kind of order zero
I1 modied Bessel function of the rst kind of rst order
k wavenumber (1/m)
Lb breakup length (m)
Lb dimensionless breakup length (Lb /djet,i )
Oh Ohnesorge number (Oh = / r0 )
3
Q volumetric ow rate (m /s)
r0 initial radius of jet (m)
ReI Reynolds number (Re = Un djet,i / = 4Q/djet,i ),
Reynolds number based on impeller diameter and tip
speed
Un initial jet velocity normal to the capillary opening (m/s)
| VOLUME 89, OCTOBER 2011 | |
w dimensionless growth rate of capillary instabilities
3/2
(w = (1/2 r0 / 1/2 ))
We Weber number (We = Un2 djet,i / = 16Q2 /2 djet,i
3
)
Greek Symbols
0 amplitude of initial perturbation responsible for jet breakup
(m)
 dimensionless wavenumber ( = kr0 )
max most amplied dimensionless wavenumber
viscosity (cP)
 density (kg/m3 )
 interfacial tension (mN/m)
growth rate of capillary instabilities (1/s)
REFERENCES
Das, T. K., Prediction of Jet Breakup Length in LiquidLiquid
Systems Using the RayleighTomotika Analysis, Atomisation
Sprays 7(5), 549559 (1997).
Eggers, J. and E. Villermaux, Physics of Liquid Jets, Rep. Prog.
Phys. 71(3), 179 (2008).
Homma, S., J. Koga, S. Matsumoto, M. Song and G. Tryggvason,
Breakup Mode of an Axisymmetric Liquid Jet Injected Into
Another Immiscible Liquid, Chem. Eng. Sci. 61(12),
39863996 (2006).
Middleman, S., An Introduction to Fluid Dynamics: Principles
of Analysis and Design, Wiley, New York (1998), pp.
325336.
Padron, G. A., Effect of Surfactants on Drop Size Distributions
in a Batch, RotorStator Mixer, Ph.D. Dissertation,
University of Maryland, College Park, Maryland (2004).
Rayleigh, L., On the Capillary Phenomena of Jets, Proc. R. Soc.
Lond. 29(196199), 7197 (1879).
Scheele, G. F. and B. J. Meister, Drop Formation at Low
Velocities in LiquidLiquid Systems: Part II. Prediction of
Jetting Velocity, AIChE J. 14(1), 1519 (1968).
Yang, M. and R. V. Calabrese, Simulation of Flow Field and
Circulation Time Phenomena in a Batch RotorStator Mixer,
AIChE Annual Meeting, Nashville, TN, November (2009),
Paper no. 468b.
Manuscript received September 3, 2010; revised manuscript
received May 29, 2011; accepted for publication June 8, 2011.
THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING | 1103 |