Chenucal Engineering Science. Printed in Great Britain.

Vol. 43. No. 3, pp. b49-654.

1988.

0009-2509!88 80 1988 Pergamon

$3.00+0.00 Journals Ltd.

EFFECT

OF SURFACTANTS ON DROP BREAKAGE TURBULENT LIQUID DISPERSIONS

IN

Department

ABRAHAM KOSHY, T. R. DAS and R. KUMAR* of Chemical Engineering, Indian Institute of Science, Bangalore 560012, India (Received 1 September 1986; accepted for publication 1 June 1987)

Abstract-The existing models of drop breakage in stirred dispersions grossly overpredict the maximum drop size when surface active agents are present inspite of using the lowered value of interfacial tension. It is shown that the difference in the values of dynamic and static interfacial tension, aids the turbulent stresses in drop breakage. When the difference is zero, e.g. for pure liquids and for high concentration of surfactants, the influence of the addition of surfactant is merely to reduce the interfacial tension and can be accounted for by existing models. A modified model has been developed, where the drop breakage is assumed to be represented by a Voigt element. The deforming stresses are due to turbulence and the difference between dynamic and static interfacial tensions. The resisting stresses arise due to interfacial tension and the viscous flow inside the drop. The model yields the existing expressions for d,,, as special cases. The model has been found to be satisfactory when tested against experimental results using the styrene-water-teepol system.

lNTRODUCTlON Mechanical agitation is a commonly employed method for generating and maintaining a dispersion of one liquid in another. The rate of transport processes in such dispersions depends on the interfacial area, transfer coefficient and the state of mixing of the two phases. The interfacial area can be calculated from the sauter mean diameter, which has generally been found to be directly proportional to the diameter of the largest stable drop (d,,) existing in the stirred dispersion. A number of reports exist in literature which predict d,,, from the operating parameters of the equipment and system properties. Hinze (1955) suggested that a drop would break if the ratio of the deforming inertial stress across its diameter and the restoring stress due to the interfacial tension, crossed a critical value. Shinnar (1961) used the above criterion and derived the following expression for d,,,
d max = constant D

increase in viscosity tends to increase the d,,, in a stirred vessel. Lagisetty et al. (1986) have proposed a phenomenological model for drop breakage and have tested it not only for Newtonian liquids, but also for power law fluids and Bingham plastics as dispersed
phases. pression, Their model reduces to Shinnar s (1961) ex-

( We)-.6

(1)

where We is the Weber number. Sprow (1967) has suggested the same expression as above and has found the value of the constant to be between 0.126 and 0.15. Lagisetty et al. (1986) have found the value of the constant to be 0.125. Equation (1) has been employed by a number of investigators and Coulaloglou and Tavlarides (1976) have discussed the various correlations available in literature, based on eq. (1). Equation (1) is applicable when the viscosity of the dispersed phase is small, since it does not take into account the Row of the dispersed phase during breakage. The viscosity of the dispersed phase has been reported to have a significant effect on drop size (Arai et al., 1977; Konno et al., 1982; Lagisetty et al., 1986). Apart from the interfacial tension which tends to restore the original drop shape, viscosity also tends to lower the rate of deformation of a drop. Thus an
649

when the viscous properties of the dispersed phase are neglected. Addition of a surfactant to the liquid-liquid dispersion reduces the droplet size as in the case of emulsification. The main reason for this would be the reduction of interfacial tension caused by the surfactant. To verify if existing models can adequately explain the effect of surfactant on drop breakage, preliminary experiments were conducted in a stirred vessel. Low viscosity dispersed phases were used, as all the models yield Sprow s (1967) expression for these. Such a comparison is shown in Fig. 1, which contains experimental results for two systems-one without a surfactant and the other with it. The water-octanol system is representative of one without a surfactant. For this system d maxvalues were measured as a function of rps and are shown as solid circles. The system with surfactant was water-styrene. It has an interfacial tension of 0.034 N/m when no surfactant is used. Teepol was added to water until the interfacial tension of the system came down to the same value as for water-octanol (0.0083 N/m). The values of d,,, for this system have been shown as solid triangles in the figure. It is seen that the values of d,,, with the surfactant are much lower than without surfactant, though the relevant system property, i.e. the interfacial tension is the same. The broken line gives the values of d,., calculated from Sprow s expression using the constant as 0.125. It is seen that this equation predicts for the water-octanol system exthe values of d,,, tremely well, but grossly overpredicts those for the water-surfactant solution-styrene system, though the properties influencing the drop volume are the same.

650

ABRAHAM

KOSHY

et

aI.

They used the following

150

expression for8<

for t, and 5, 1 (3)

~~=:f3(1-Q) =o

for B 3 1

E CD P x ii D

100

and

TO= fix.

de

(4)

50

I 7

10

11

Impeller speed (rps) Fig. 1. Comparison of experimental results of a surfactant and a pure system having the same interfacial tensions. Interfacial tension: 0.0083 N m I, Systems: water-octanol, experimental: 0; surfactant solution-styrene, experimental: A. Sprows equation - - - - - .

Experiments that drop active the

performed existing models

on other

systems

also the when

indicate surface that the

overpredict dispersion, would This

maximum

size obtained agents

in a stirred

are present.

indicate

not only acts through the reduction of interfacial tension but also influences in some other fashion, which has not been considered by any of the existing models. An attempt has been made in the present investigation to develop a model incorporating the effect of the surfactant.
surfactant
DEVELOPMENT OF THE MODEL

The recent model of Lagisetty et al. (1986) assumes that the deformation of the drop can be represented by a Voigt element. The drop deformation is caused by
the external pressure inertial stress arising out of the turbulent

The deformation is resisted by the stress due to interfacial tension and the viscous stress (due to internal flow). Thus
fluctuations.

t =

7s+7,.

(2)

The stress due to interfacial tension is assumed to pass through a maximum and then reach zero at 9 = 1. The model described above captures the gross essential features of the breakage process, but bypasses the details of the flow both inside and outside the breaking drop. Thus, the presentation of the applied turbulent stress and the resisting stresses due to interfacial tension and viscosity through a spring and dashpot assembly significantly simplifies the complex flow problem. As the addition of the surfactant is not likely to change the mechanism of breakage in a drastic way, the basic features of the model of Lagisetty et al. (1986) can still be retained while the role of surfactant is delineated in more detail. The surfactant molecules have a tendency to get adsorbed at the liquid-liquid interface, thereby reducing the interfacial tension. When a pressure fluctuation due to an eddy is experienced by the drop across its diameter, the drop starts deforming. The deformation most probably starts by the formation of a depression on the drop interface and this depression propagates resulting in breakage. When the surfactants are present at the interface, the pressure fluctuation apart from causing depression at the interface, also removes the adsorbed surfactant molecules thereby exposing a fresh interface. This fresh interface has dynamic interfacial tension which is higher than the static interfacial tension. Thus, at the base of the depression, the interfacial tension is higher. This difference in interfacial tension, causes a flow towards the base and this adds to the flow already taking place due to the pressure fluctuation. Thus, an extra stress is generated due to the difference in dynamic and static interfacial tension, which adds to the stress due to the turbulent pressure fluctuation. The assumed mechanism through which the extra stress gets generated has been shown in Fig. 2. Here
Applied turbulent stress

La) Drop before with eddy.

Fig. 2. Generation

interaction

(b) Formation of the depression. in dynamic and equilibrium interfacial tensions.

of the extra

stress due to difference

Effects of surfactants on drop breakage again we bypass the complex flow field and simplify the analysis by adding the extra stress to the inertial stress. The extra stress may be expressed as An/d, where Aa is the difference between the dynamic and static interfacial tensions. Equating the deforming and resisting stresses, we obtain

651

Substituting the above expression in eq. (14) we obtain ~(0 = 1) = (Re/We)(D/d,,)3

(16)

where Re = D2N p,/,u. Equation (13) has been integrated with C = 8 (Lagisetty et al., 1986), obtaining the following expression for d,,,
4

r+-

Od -

CT

=
32 We(d,,,lD) tan-l

The od value corresponds to the interfacial tension at the freshly created interface. The difference between od and u is function of the surfactant concentration. Hinze (1955) has given the following expression for r r cc P~U (d) (6)

5/3+4(?!$)_l]

32 We(d,,,/D)5 3
= (Re/We)(d,,,/D)- /3. (17)

u (d) can be expressed in terms of parameters associated with the stirred vessel (Coulaloglou and Tavlarides, 1977) as

u2 (d) = C,C2 3d2 3.

Further Thus, E = 0.4Q7N3D2. r = Cp,N= D4 3d2/3.

(7)
(8) (9)

Substituting this expression for r in eq. (5) we obtain

CpcN2D Vd2 3 +%j? =

For pure liquids, trd = IT and under this condition, eq. (17) reduces to the proper limit, i.e. the equation given by Lagisetty et al. (1986). Further, the equation is applicable for Newtonian dispersed phases, but can be easily extended to non-Newtonian dispersed phases by substituting appropriate rheological expression in eq. (5). Thus, for power law liquid having the rheological expression
7

%e(l -t,)+p%p (10)

and defining in non-dimensional

=k

( >
g

(18)

Reynolds He, =

number for any value of n as D(ND) -PC 2-3(3 + l/n)k when n = 213 works out to be

Equation form as
-

(10) can be expressed

d0 d?

= CWe(d/D)5 3+

the equation time given by

for d,,,, 20

where PI is the non-dimensional

(12)
Equation (11) may be written as d0
(6 - 4) + a where a = CWe(d/D)5 3+

32 We(d,,,/D)513 = drj (13)

= (Re /

We)3 2

(d,,,ax/D)-5 6.

EXPERIMENTAL

Equation

(13) has been solved with initial condition 8=0 at q=O.

For breakage to occur, the deformation (0) should be at least unity so that the interfacial tension cannot bring it back to the original state. Further, this minimum deformation for breakage should be achieved within the life time of an eddy, r. We can therefore obtain maximum stable drop diameter, if
q(tl = 1) = F;

(
Coulaloglou and Tavlarides of an eddy as

-> max

(14)

(1977) have given the life

(15)

T = d2 3/(ND2 3).

The stirrer assembly employed in the present investigation is shown in Fig. 3. A glass stirred vessel of 0.135 m i.d. and 0.2 m height was equipped with four equi-spaced baffles at its wall. Each baffle had a width equal to l/lOth the vessel diameter. A six bladed Rushton type impeller, shown in the figure, was employed for stirring. The impeller was normally located at the centre of the liquid column and was run through a l/5 hp motor. The speed was controlled through a dimmerstat and measured through a digital tachometer connected to the stirrer shaft through a belt and pulley arrangement. To start a run, the continuous phase was first charged in the required quantity. The stirrer was started and its speed adjusted to the required rpm. The dispersed phase was then added. In all the experiments

652

ABRAHAM

KOSHY

et al.

V. S. F. T. I.

Vessel Shaft Frame Tachometer Impeller

B. Baffle C. Coupling M. Motor Ii. Bearing house are

in m Cont.

1.C

of

surfactant

solution

(Kg/m?

Fig. 4. Effect of surfactant on dynamic and equilibrium interfacial tension for water-teepokkstyrene system.

Dimensions

Fig. 3. Schematic diagram of stirring assembly.

here, the dispersed phase fraction was less than 0.02 to minimize coalescence. It took about an hour to reach steady state. After this period, samples were withdrawn and the diameters measured using a microscope. There was no coalescence of drops during withdrawal and measurement because of the very low dispersed phase fraction employed. A minimum of 200 droplets were measured to decide on d,,,. The dynamic interfacial tension was determined through the photographic method of Andreas et al. (1938). A drop of the surfactant solution was formed from a capillary tip and was photographed at known intervals of time. The drop shape varied as a function of time because of the change in interfacial tension brought about by adsorption of the surfactant molecules at the liquid-liquid interface. The interfacial tension at each time was found by the method given by Andreas et al. (1938). Interfacial tension was plotted as a function of time and extrapolated to zero time to
reported

obtain dynamic values. Static interfacial tension value was obtained when the slope no longer changed as a function of time. Static interfacial tension was also alternatively obtained by the drop weight method of Harkins and Brown (1919). A typical set of both equilibrium and dynamic values of interfacial tension vs the surfactant concentration for the system water-teepol-styrene has been presented in Fig. 4. It should be noted that after a certain concentration of surfactant, the equilibrium and dynamic interfacial tensions become identical. The rheological parameters of the liquids were measured using a rotational viscometer.
RESULTS AND DISCUSSION

Figure 5 presents two sets of experimental data in the range of surfactant concentrations where the dynamic and equilibrium interfacial tensions differ. The system is styrene-water and the surfactant used is teepol. For points corresponding to the equilibrium

Fig. 5. Comparison ofexperimental results with present model when the dynamic and equilibrium interfacial tensions differ. Equilibrium interfacial tension: 0.0253 N/m; AC: 0.0011 N/m; experimental: A; Lagisetty et al. model:-.----.-; present modek -_ Equilibrium interfacial tension: 0.011 N/m; Ao: 0.0014 N/m; experimental: l; Lagisetty et al. model: - - ~ --; present model: -p.

Effects of surfactants on drop breakage

653

8 speed

9 ( rps)

10

11

Impeller

Fig. 6. Comparison of experimental results with those predicted by the present model at zero and high concentrations of the surfactant. System: water-styrene; interfacial tensionl 0.034 N/m; experimental: A; present model -. System: water-teepol-styrene; interfacial tension: 0.0032 N/m; experimental: l ; present model p_

Impeller

speed

(rps)

Fig. 7. Comparison of experimental results with the prediction from the present model for a power-law fluid. Dispersed Equilibrium interfacial tension: polystyrene. phase: 0.0144 N/m; Acr: 0.0015 N/m; K: 4.1 kg/ms4j3; n = 2/3. Experimental: A; present model: __.

interfacial tension value of 0.011 N/m the A5 is 0.0014 N/m. The dotted line corresponds to the model of Lagisetty et al. (1986). This line also represents the earlier models as they are special cases of the model of Lagisetty et al. It is clear that the existing models significantly overpredict the values of d,, over the whole range of rps studied. The present model, shown by the solid line, however, predicts the data quite satisfactorily. Similar conclusions can be drawn from the other set of data presented in the figure. Here, the values of 5 and A5 are 0.0253 N/m and 0.0011 N/m respectively. In this instance, the overprediction by earlier models is relatively small because of the low value of A5 when compared to u. It has been shown earlier in Fig. 4 that at zero and high concentrations of the surfactant, there is no difference between dynamic and equilibrium values. To verify, if the model could be applied to these extreme conditions also, experiments were conducted with zero and high concentrations of the surfactant for the water-styrene system. These results are presented in Fig. 6. Here again, it is seen that the present model explains the data well both for zero and high concentrations of the surfactant. The success of the model to predict the high concentration results further strengthens the assumption that the difference between the two interfacial tensions indeed plays an important role in facilitating drop breakage.
Eflect of Au on breakage of viscous drops To study the breakage of viscous drops, a solution of polystyrene was employed as the dispersed phase and water as the continuous phase. The polystyrene solution was highly viscous and followed power law characteristics with k = 4.1 and n = 2/3. The surfactant teepol was added to the aqueous phase to reduce value to equilibrium interfacial tension the 0.0144 N/m. The A5 was found to be 0.0015 N/m. The results for this system are presented in Fig. 7. Along with the experimental points, predictions by the model

have been shown as solid line. The model is found to predict the experimental results for this system also. Calculations were made by putting A5 = 0 instead of 0.0015. The calculated values were the same as shown by the solid line. As the viscosity of the dispersed phase is increased, the breakage is increasingly controlled by viscous forces and small changes in interfacial tension would not be expected to change the value of d,,,.
CONCLUSIONS

The breakage of drops in stirred dispersions is influenced by the presence of surfactant not only through the reduction of interfacial tension, but also by the generation of an interfacial tension gradient across the drop. This results in an extra stress which adds to the turbulence stress and hence results in the formation of smaller droplets. The effects are small when the difference in dynamic and equilibrium interfacial tension is small when compared to the equilibrium value for low viscosity liquids. As the viscosity of the dispersed phase is raised the effect becomes increasingly smaller and finally vanishes. It is possible to predict the d,, values by adding this new stress to the turbulence stress causing the drop breakage.
NOTATION c, d
Cl

constants drop diameter, m Sauter mean diameter,

d 32 d InBX D k
n

23nid:

Cn,df

N
Re We

maximum drop diameter, impeller diameter, m non-Newtonian viscosity constant (power law) imoeller soeed. rDs
1 _ ,*

Reynolds Weber

number,

D (ND) 2-(3+

- p< l/nyk

number, (pcD3 N2)/o

654

ABRAHAM

KOSHY

er al.

t T
Greek & ? Pr F CL 0 CT Gd 5
T.

time, s
average letter energy density viscosity, constant dimensionless strain static interfacial tension, dynamic average N/m2 elastic interfacial value stress. N/m N/m inertial stress, time period, R

dissipation time

rate per unit mass, m2/s3 phase, kg/m3

dimensionless Ns/m

of the continuous

tension,

of the external N /m2

/

viscous stress, N/m difference between the inter-facial tensions, N/m

dynamic

and

static

REFERENCES

Andreas, J. M., Hauser, E. A. and Tucker, W. B., 1938, Boundary tension by pendant method. J. Phys. Chem. 42, 1001.

Arsi, K., Konno, M., Matunga, Y. and Saito, S., 1977, Effect of dispersed phase viscosity on the maximum stable drop size for break-up in turbulent flow. J. Chem. Engng Jap. 10, 325-330. Coulaloglou, C. A. and Tavlarides, L. L., 1976, Drop size distribution and coalescence frequencies of liquid-liquid dispersion in flow vessels. A.1.Ch.E. J. 22, 289-297. Coulaloglou, C. A. and Tavlarides, L. L., 1977, Description of interaction Processes in agitated liquid-liquid dispersions. Chem. Engng Sci. 32, 1289-97. Harkins, W. D. and Brown, F. E., 1919, The determination of surface tension (free surface energy) and the weight of falling drops: The surface tension of water and benzene by the capillary height method. J. Am. Chem. Sot. 41,499-524. Hinze, J. O., 1955, Fundamentals of the hydrodynamic mechanism of splitting up in dispersion process. A.1.Ch.E. J. 1, 289-295. Konno. M.. Arai, K. and Saito. S.. 1982. The effect of stabilizer on coalescence of dispersed drops in suspension polymerlzation of slyrene. J. Chem. Engng Jap. 15, 131-135. Lagisetty, 3. S., Das. P. K., Kumar, R. and Gandhi, K. S., 1986, Breakage of viscous and non-Newtonian droos in stirred dispersions. Chem. Engng Sri. 41, 65-72. Shinnar, R., 1961, On the behaviour of liquid dispersions in mixing vessels. J. Fluid Mech. IO, 259-275. Sprow, F. 3., 1967, Distribution of drop sizes produced in turbulent liquid-liquid dispersion. Chem. Engng Sci. 22, 435442.