The Chemical Engineering Journal, 32 (1986) B7 - B17 B7

Bubble Break-up in Gas-Liquid Bioreactors: Break-up in Turbulent Flows

JAMES F. WALTER and HARVEY W. BLANCH

Department of Chemical Engineering, University of California, Berkeley, CA 94720 (U.S.A.)

(Received March 5, 1985)

ABSTRACT High liquid viscosity, commonly occurring in

fermentation systems, will reduce liquid mix-
The breakage of gas bubbles in a turbulent ing, oxygen diffusivities and liquid velocities,
flow field is explored. Break-up was found to and leads to large mass transfer resistances.
occur by a dumbbell elongation of the bubble. The mass transfer rate depends on the
The elongational viscosity was found to overall mass transfer coefficient (K,a) and
correlate the observed dependence of maxi- the concentration driving force. The difficulty
mum bubble size with the nature of the in the design and scale-up of gas-liquid con-
continuous phase. The effect of surfactants tactors is in the evaluation of the parameters
on the break-up mechanism was examined KL and a. The overall mass transfer rate is in-
and the rate of replenishment of surfactants fluenced by a number of physical parameters
at the interface and their subsequent orienta- and operating conditions. The key physical
tion were found to determine the surface parameters are; liquid properties including
elasticity and thus the rate of break-up. Small viscosity, density, interfacial tension, vessel
inorganic ions were found not to influence geometry, sparger design and agitator type.
the maximum bubble size. Important operating conditions are the
superficial gas and liquid velocities and the
impeller speed. These factors are highly
1. INTRODUCTION interactive and control the reactor perfor-
mance, as depicted in Fig. 1. This paper
Gas-liquid contactors, with or without examines the role of liquid viscosity, inter-
mechanical agitation, are used in a variety of facial tension and the presence of surfactants
industrial applications ranging from stripping on bubble break-up.
and absorption columns to three phase slurry
beds and activated sludge ponds. Gas-liquid
MASSTRANSFER IN GAS-LIWID CONTACTORS

contactors of significant economic impor-

tance include aerobic biochemical reactors,
where ensuring an adequate supply of oxygen
to a growing organism is of prime importance.
For many aerobic fermentations the produc-
tivity of the system is controlled by the rate
of oxygen transfer from the gas to the liquid
phase and hence to organisms suspended in
the liquid. Due to its low solubility, oxygen,
usually added in the form of air, must be Fig. 1. Relationship between the various factors
continually supplied to the medium so that affecting the mass transfer rates in a gas-liquid
the oxygen absorption rate at least equals the reactor.
oxygen consumption rate by the cells. Even
temporary depletion of dissolved oxygen
could mean irreversible cell damage. There are 1.1. Bubble break-up
two major resistances to oxygen transfer in Bubble break-up controls the maximum
the reactor, namely, the gas-liquid interface stable bubble size and can be greatly in-
and the cell-fluid interface. Both of these fluenced by the liquid hydrodynamics. Hinze
resistances are affected by the hydrodynamics. [l] suggested that bubble break-up is caused

0300-9467/86/$3.50 0 Elsevier Sequoia/Printed in The Netherlands

B8

BUBBLE BREAK-UP

COWROLLED BY

HYDRODYNAMIC CONDITIONS

TURBULENT FLOW LAMINAR FLOW STAGNANT LxQuroS

I 1
WAKE
INTROPIC r---l
IMPELLER v1sc0us SURFACE
TURBULENCE ISOTROPIC ELONGATION DRAG INSTABILITY VORTICITY

Fig. 2. Classification of break-up regimes.

by hydrodynamic stresses that attempt to stirred systems and that the turbulence may
disrupt the bubble, and that these are opposed be considered isotropic if D % 2001. Since
by surface tension forces. At the point of industrial impellers can be several meters in
breakage these forces must balance. Thus diameter, this criterion does not guarantee
u isotropy on the bubble scale. A more logical
-i- cc --
(1) approach would assume that the microscale
d
is much smaller than the maximum stable
This balance leads to the prediction of a bubble size. This suggests that isotropy may
critical Weber number We = rd/o where the be characterized by
shear stress term will reflect the hydrodynam-
2001< d,
ic conditions responsible for bubble break-up.
The mechanisms of bubble break-up can thus and
be related to the hydrodynamics (Fig. 2). In 314

the case of turbulent flow, bubble break-up is I= hl

(2)
(P/vy4p2
caused by fluctuating eddies that bombard
the bubble surface. In laminar flow, viscous If the maximum bubble size is larger than
shear at the bubble surface will elongate the the microscale, eddies of this wavelength are
bubble and cause break-up. In a stagnant in the inertial subrange. Then forces on the
liquid, bubble break-up is caused by surface bubble surface are Reynolds stresses and can
instabilities and wake vortices. The mecha- be expressed as
nism of bubble break-up varies, as a result of
I- = p($i2)/2 (3)
the hydrodynamic stresses in the region of the
bubble. The fluctuating velocity has been expressed
by Batchelor [ 31 as
1.2. Break-up in turbulent flows
,,‘$ = (P/V)“3(d/p)“3 (4)
Many gas-liquid contactors operate under
what may be considered turbulent conditions. It is assumed that bubble break-up is caused
In this regime bubble break-up can be mod- by eddies of the same scale as the maximum
eled using isotropic turbulence theory. Most stable bubble size. Eddies larger than the
practical examples of bubble break-up are in bubble cause translational motion while
non-isotropic bulk flows; however, if the smaller eddies do not possess enough power
volume under consideration is small enough, to disrupt the bubble surface. At the point of
the turbulent eddies may be considered both breakage, the hydrodynamic stresses over-
isotropic and homogeneous. come the surface tension forces so that
Flow may be considered isotropic if the
(P/V)2’3(d/p)2’3 = Ku/d (5)
microscale of turbulence (I) is much smaller
than the primary eddy size (L). Moo-Young and on rearranging
and Blanch [2] have suggested that the
o0.6
primary eddy size may be approximated by d, = K
the impeller diameter (D) for mechanically (P/ v)0.4p0.2
(6)
 BQ

where K is a function of any other forces that
may stabilize the bubble surface. It is ex-
pected that liquid rheology and the presence
of surfactants may have a significant effect
on the maximum stable bubble size.
2. EXPERIMENTAL METHODS
In order to examine the range of validity of
eqn. (6), it is important to have flexible,
well-controlled flow conditions, and turbulent
pipe flow was chosen. The experimental
apparatus is shown in Fig. 3. Bubbles were
injected into turbulent flowing liquid and
photographed by a high speed tine camera
(HY-CAM K2001) as they passed down the
r I
tube. The upper leg of the flow loop was
constructed of clear Plexiglass pipe and
encased in a Plexiglass photographic box to
eliminate optical distortions. A variety of
fluids were investigated, including distilled
water and solutions of glycerol, carbopol,
polyox and various surfactants. These solu-
tions were used to simulate various physical
properties found in industrial fermentations.
These properties are outlined in Table 1.
Water-glycerol solutions were used to ex-
amine the influence of liquid viscosity and
density on bubble break-up. Carbopol-water
solutions are pseudoplastic and polyox-water
solutions show a visco-elastic behavior. Flow
conditions were varied to simulate the power
requirements (0.25 - 2.0 HP 100 gals-i) found
in industrial fermenters.

3. MECHANISM OF BUBBLE BREAK-UP IN ISO-

TROPIC TURBULENCE (NEWTONIAN LIQUIDS)

Since smaller bubble sizes are generally

Fig. 3. Schematic diagram of apparatus used to
photograph bubble break-up in turbulent flows: A,
holding tank; B, high speed tine camera; C, light
source; D, gas injector post; E, recirculating pump;
F, photographic box; G, flow controller; H, manom-
eter.
desired in gas-liquid contacting, methods for
increasing bubble break-up are of interest. A
single mechanism of break-up was observed
in the present experiments, and most of the
break-up events occurred in the isotropic core
of the tube. This mechanism can be modeled
as consisting of three stages: oscillation,
dumb-bell stretching and pinching off [ 4 1. As
a bubble larger than the maximum stable size
travels down the center of the tube, its
surface oscillates and ripples due to fluctuat-
ing pressure forces. The bubble then stretches
into the shape of a dumb-bell with two large
centers of mass connected by a thin strand of

TABLE 1
Properties of solutions

Solution Symbola PC K N
I s-2) rg cmp3 ) (g cm-’ 6-l) (g cm-l S-(2--N9

Distilled Hz0 73 1.00 0.01

H20-0.4% butanol : 61 1.00 0.0099
H20-62% glycerol v 70 1.16 0.12
H@-76% glycerol 0 67 1.18 0.20
H20-76% glycerol 0 65 1.2 0.40
H@-0.05% carbopol 0 73 1.00 0.034 0.982
H20-0.675% carbopol A 13 1.00 0.078 0.94
H20-0.1% carbopol 0 73.2 1.00 0.48 0.83
H20-0.1 5% carbopol 0 73.3 1.00 6.60 0.59

aSymbols refer to those used in subsequent figures.

B10

(b)

(e)
Fig. 4. Break-up of a bubble in the turbulent core of pipe flow. Total time for break-up 25 ms.

fluid. This strand is pinched off and two

independent bubbles are formed. The dumb-
bell stretching step seems to be rate control-
ling and break-up requires a total tune of
about 25 ms. The photographs in Fig. 4
depict this mechanism. Several investigators,
e.g. Karabelas [ 51 and Sleicher [6], have
considered droplet break-up at the tube walls
to dominate in liquid-liquid systems. Thus,
the observed mechanisms of break-up appear
to vary between gas-liquid and liquid-liquid
systems.
The time required for break-up was found
to vary slightly between bubbles in the same
flow field, but a general trend was recognized
that indicates an increase in break-up time
with an increase in continuous phase viscosity.
This is intuitively expected from the mecha-
nism observed. The viscosity of the con-
tinuous phase will increase resistance to
I
elongation, thus increasing the break-up time
I I I
02 0.6
and the maximum stable bubble size. The
results of varying viscosity and surface tension
Fig. 5. Effect of liquid viscosity and surface tension
are presented in Fig. 5 and suggest that
on the maximum bubble size in turbulent flow. Solid
d, -pc ‘.I. Equation (6) was found to repre- line is a plot of eqn. (6), corrected for the effects of
sent the experimental data if this viscosity viscosity. From the data K = 1.12. Symbols refer to
dependence is added, as is shown in Fig. 5. Table 1.

(b)
Cd)
Fig. 6. Bubble break-up in a 3 m bubble column.
Photographs taken in the center of the column at
various times (a) t = 0 s; (b) t = 8 ms; (c) t = 20 ms;
(d) t = 22 ms; (e) and (f) t = 25 ms.
The bubble size was increased by surface
tension and liquid viscosity and decreased by
power input. Linear regression of the experi-
Bll
mental data suggests exponents within 10% of
those obtained from eqn. (6).
It is important to determine whether the
mechanism of break-up in actual gas-liquid
contactors is similar to that observed under
idealized conditions in the horizontal pipe.
IIigh speed photography was used in a bubble
column in order to observe bubble mechanics
and the mechanism of bubble break-up.
Typical results are depicted in Fig. 6. It was
concluded that bubble break-up did occur by
a dumb-bell stretching mechanism and that
the break-up time was of the order of 25 ms.
This suggests that the results obtained from
the horizontal tube are applicable to other
forms of gas-liquid contactors.
3.1. The effect of pseudoplasticity and
elasticity on bubble break-up
Fermentation broths frequently exhibit a
pseudoplastic or elastic nature which can
have a dramatic effect on bubble stability.
The influence of pseudoplasticity on bubble
break-up is similar to that observed for
Newtonian fluids and can be accounted for by
incorporation of an apparent liquid viscosity.
The viscosity of pseudoplastic fluids is shear
dependent but it is difficult to predict the
true shear at the bubble surface in a turbulent
field. In isotropic turbulence, motion close to
the bubble surface is controlled by turbulent
eddies with a wavelength smaller than the
bubble diameter. Outside this turbulent shell,
flow is controlled by the bulk liquid motion.
The thickness of this shell is of the order of
the largest important wave length. Thus the
shear rate can be written as
au
y = ar = 1.35{(P/V)(l/pd,)}1’3 (7)
and the apparent viscosity can be calculated
from
PaP= K[1.35{(P/V)(l/pd,))1’3JN-’ (3)
This approach credibly models the experi-
mental data in predicting the influence of
pseudoplasticity on bubble break-up (using
carbopol as a model fluid) and maintains the
dependence on apparent viscosity
d, a (pap)**’ (9)
The results using carbopol as a model pseudo-
plastic fluid are included in Fig. 5, with the
apparent viscosity used in place of pC.
B12

The influence of viscoelasticity on bubble polyox solutions and the data are presented
stability is much more difficult to predict. in Table 2. The effect of elasticity on bubble
Polyox solutions were used to examine these break-up is presented in Fig. 7. From these
effects. The maximum stable bubble size limited results it appears that d, - pea.’
observed experimentally was larger than which is similar to the previously determined
would be expected from fluids with a similar dependence for Newtonian and power law
apparent shear viscosity (Table 2). This fluids (d, - pap O.‘). Increasing the elonga-
suggests that the elastic nature of the fluid tional viscosity of the liquid in terms of
stabilizes the maximum bubble size. In order either viscosity or elasticity will increase the
to explain the phenomena, the mechanism of maximum stable bubble size, while pseudo-
break-up must be examined. During the plasticity decreases the apparent viscosity,
dumb-bell stretching phase the bubble under- which results in a smaller bubble size.
goes an elongation. The resulting viscous Together with the liquid viscosity, it may
forces that resist this motion are of an elonga- be expected that the gas viscosity will have an
tional nature and not the shear forces com- influence on bubble break-up. In order to
monly encountered. This suggests that the evaluate this effect, nitrogen, hydrogen and
elongational viscosity is responsible for the argon as well as air bubbles were examined in
increase in bubble size. In the case of a the experimental system. The gases did not
Newtonian fluid, elongational viscosity is change the break-up mechanism. While the
simply three times the shear viscosity. In interfacial tension of these gases does not
viscoelastic solutions, the elongation viscosity measurably change, the viscosity of the gas
is a function of the fluid relaxation time and phase changes by an order of magnitude
can be much greater than the apparent shear (Table 3). The maximum stable size was
viscosity. found to decrease slightly with increased gas
The ductless siphon method [ 71 was used viscosity. This supports the use of the gas
to determine the elongational viscosity of the viscosity to normalize the elongational liquid
viscosity.
This observation is supported by the work
TABLE 2
of Vermuelen et al. [8] who examined bubble
The effect of elongational viscosity on bubble break-
up TABLE 3
The effect of gas viscosity on maximum bubble size
Solution Sym- Pap D
bol (gem s-l) ytcrn s-r) (c=y
Gas rue d,
0.043% Polyox 0 0.012 0.014 0.365 (g cm-l s-l) ;Idyne cm-‘) (cm)
0.075% Polyox n 0.014 0.054 0.385
0.1% Polyox 0 0.015 0.08 0.395 Air 1.8 x 10-J 73.1 0.452
0.15% Polyox 0 0.022 0.35 0.42 Argon 2.2 x 10-d 73.0 0.446
0.2% Polyox 0 0.037 0.60 0.44 Hydrogen 8.6 x 10-s 73.4 0.462
0.3% Polyox rJ 0.08 0.71 0.46 Nitrogen 2.0 x 10-e 73.0 0.45

0.1 - I I I Illlll I I I Illll

0.01 01
pE/3 (g/cm set)

Fig. 7. The effect of elongational viscosity on the maximum stable bubble size. Symbols refer to those in Table 2.
 B13

and droplet size distributions in agitated

vessels and found smaller average sizes for
droplets than for bubbles under similar
conditions, suggesting a similar dependence
on the dispersed phase viscosity. The experi-
mental results suggest that the maximum
stable bubble size can be calculated from
-0.6 r _. lo.1

3.2. The effect of surfactants on bubble

break-up
Fermentation broths contain a variety of
surface active materials, including extra-
cellular proteins, antifoam agents and biologi-
cal products. Reducing the surface tension of
the liquid medium may be the most efficient
method of reducing the maximum stable
bubble size and increasing the interfacial area.
Equation (10) suggests that reducing the
liquid surface tension by 50% will reduce the
maximum stable bubble size by 34% and
0.26c,
8, / l

IO
, I

12
I J
similarly will increase the interfacial area per a0~6(g/sec2P6

unit volume. This prediction ignores the
possible influence of the surface excess and
surface elasticity on the mechanism of bubble
break-up.
A series of surfactants was examined to
determine these effects. Surfactants, including
alcohols ranging from propanol to lauryl
alcohol, sodium lauryl sulfate and hexanoic
acids were investigated, as well as Tween 80
and silicone antifoam, which are commonly
added to fermentation systems. The surface
tension of the aqueous system was varied
from 25 to 73 dynes cm-’ while other factors
were held constant. Two different types of
Fig. 8. The effect of surface active substances on bub-
ble break-up: 0, distilled H,O; 0, salts; o, n-butanol;
0, n-propanol; @, n-octanol; 0, hexanoic acid; 0,
dodecanol; m, Tween 80; 6, SDS; 0, antifoam; Y,
yeast broth; L, Loctobocillus broth. The properties of
the solutions are given in Table 4.
experimental results were observed and the
results are presented in Table 4 and Fig. 8.
For short carbon chain length surfactants,
up to octanol, the maximum stable bubble
size decreased as predicted by isotropic
turbulence theory (d, = uO-~), as shown in
Fig. 8. The surface tension was varied by a
factor of 2 and the results show good agree-

TABLE 4
The effect of surfactants on bubble break-up

Solution concentrations D max D max

;rdyn cm-‘) rmol cmd2) (observed) (predicted
from eqn. 10)

5 x lop2 M n-butanol 63 3.8 x lo-lo 0.338 0.4

1.8 x 10-l M n-propanol 61 2.6 x lo-lo 0.35 0.35
2 x 10P3 M n-octanol 43 5.8 x lo-lo 0.30 0.29
2 x 10P3 M bexanoic acid 61 0.36 0.35
5 x lop4 M dodecanol 25 ; ; ;;I:; 0.30 0.22
1.6 X lo@’ M dodecanol 30 4.1 x 10-10 0.32 0.24
3 x 1O-4 M SDS 45 0.42 0.30
1 x 1O-3 M SDS 40 0.385 0.27
B14
ment with the results for pure liquids. The
higher molecular weight and longer chain
surfactants such as dodecanol, antifoam,
sodium lauryl sulfate (SDS) and Tween 80,
show a maximum stable bubble size which is
much larger than that predicted by eqn. (10)
(Table 4). In comparing results obtained with
0.002 M n-octanol and 0.003 M SDS, it is
seen that although these solutions have sim-
ilar surface tensions, the maximum stable
bubble size differs by 40%. This suggests that
other factors, in addition to surface tension,
are acting to stabilize the bubble surface.
It was observed that the presence of the
larger molecular weight surfactants decreased
rippling of the bubble surface. Similar ob-
servations were made for elastic liquids, sug-
gesting that surface elasticity may be respon-
sible for stabilizing the maximum bubble
size.
There are several possible explanations for
the role of surface elasticity on bubble
break-up. Surface elasticity can be described
by the Gibbs equation
au al?
E=- =-_RT-
(11)
dlnA a In A
where
au
r=-- - (12)
RT alnC,
where r is the surface excess and is a measure
of the surfactant concentration at the inter-
face. Organic surface active agents are prefer-
entially adsorbed onto the gas-liquid interface
and the surface excess is positive. Assuming
that the surface of the bubble is at equilib-
rium prior to break-up, as the surface of the
bubble elongates, the amount of surfactant
per unit area decreases. This increases the
apparent interfacial tension unless the amount
of surfactant on the surface is replenished.
Thus, two important factors may influence
the generation of elastic forces during bubble
break-up; the magnitude of the surface excess
and the rate of surface replenishment. Surface
excesses were measured by the Wilhelmy plate
technique and the results are presented in
Table 4. All the surface excesses measured
were of the same order of magnitude, differ-
ing at most by a factor of three and did not
correlate with the observed results. Thus it
appears that surface replenishment may
control the surface elasticity. The results
suggest that the surface of the bubble is
rapidly replenished by low molecular weight
surfactants (<C,) and more slowly by longer
chain length surfactants (Xs). Surface
tension gradients may be created on the
surface. However, in a turbulent system the
surface region will be well mixed, resulting in
very little mass transfer resistance. Thus the
rate of surface adsorption probably controls
the influence of surface elasticity on bubble
break-up.
There are two factors that influence the
rate of surface replenishment; the rate of mass
transfer to the surface and the rate of orienta-
tion and adsorption of the surfactants at the
surface. Since the diffusivities of the sur-
factants are of the same magnitude, mass
transfer is not expected to control elastic
effects, but the rate of surface orientation
appears to be important. A similar observa-
tion has been reported in stagnant systems
where the interfacial tension of the solution
is found to be time dependent. Some solu-
tions containing large surfactants exhibit a
surface relaxation time where the surface
tension of an initially fresh surface changes
with time as surfactants are adsorbed onto the
surface. Kimizuka et al. [9] has shown that
this phenomenon is not solely diffusion
dependent and the rate of orientation of a
particular surfactant may be significant. It is
expected that short chain molecules will more
easily adsorb onto the surface and thus no
elastic effects are observed, while long chain
molecules will take longer to properly orient
themselves and adsorb at a much slower rate.
With these assumptions the influence of
surface elasticity can be simply modeled.
In order to evaluate the influence of
surface elasticity on bubble break-up, it will
be assumed that mass transfer resistances at
the surface of the bubble can be ignored and a
Langmuir isotherm describes the adsorption
equilibrium. The rate of change of the surface
excess during dumb-bell stretching can then
be expressed as
dr
- =r,-r (13)
dt
where
r, = k,c,(r’,- r) - h_,r
r, is the rate of replenishment of species at
the surface; kzlCs(I’, - r) is the rate of ad-
 B15

sorption and k_,I’ the rate of desorption. The TABLE 5

rate of surface depletion by elongation is P, Adsorption constants
assumed to be equal to a’l’ where the surface
is assumed to elongate at a linear rate A = Surfactont k, (1 g mol-’ s-l)
Ao(l + a’t). If the surface concentration of
adsorbing species is initially at the equilibrium Butanol >104
Propanol >104
value, eqn. (13) yields
Octanol > 105
Hexanoic acid > 105
k,C, + (a’ + k-1)
r = r, X Dodecanol >104
k,C, + k_, + a’ Sodium dodecylsulfate >104
Tween 80 > 104

X exp{-(k,C, + K-i + a’)t} (14)

The elasticity can be expressed as
E+u
7=-
au d
E=-
(15)
a In A
For isotropic turbulence, eqn. (10) can be
and thus rearranged to predict the maximum bubble
size
E=- i + ~‘t _a0 (16)
i-1
d at d = l.l2(E + u)Oa6 E.C, O-l
m (21)
Expressing the surface tension as (P/v)0*4p0** 3/.lg

a Evaluating the magnitude of the surface

-=l-!Zr (17) elasticity is very difficult, but for the larger
(Jo (JO surfactants a value of 20 dynes cm-’ fits the
The elasticity varies as data well.

3.3. Effects of ions on bubble break-up

E = -(l + a’t) -
X exp{-(k,C, + k_, + a’)t}
For slow desorption k ,C, S k_l and if a’ % k-,
at short times eqn. (18) reduces to
E = RT(1 + a’t)r, exp{-(klC, + a’)t}
Thus the magnitude of the exponential terms
determines whether surface elasticity affects
bubble break-up and the magnitude of the
surface excess determines the extent of this
effect. Thus trace amounts of surfactant do
not influence bubble break-up. The surface
excess of a dilute solution is so small that
although the surface renewal rate may be
Together with organic surfactants, ions are
known to have a great effect on surface
(18) phenomena. Several investigators [lo] have
reported that the addition of salts to a gas-
liquid contactor can greatly decrease the
mean bubble size. This can be due to either a
decrease in the maximum stable size or a
(19)
decrease in coalescence frequency. In order
to investigate this phenomenon, KC1 solu-
tions up to 0.4 M were examined and the
results are presented in Fig. 9. It was observed
that ions have no perceptible effect on bubble
break-up. This is expected because the salts
I I
I I I

0.4
slow, no elastic forces are generated. Further-
more, for small surfactants the rate of adsorp- 0mox
-A-_A-*---___-___-A-

tion is very fast and the elastic forces are (cm)

t 1

negligible. Larger surfactants are adsorbed at

much slower rates and elastic forces are
considerable. Table 5 presents the estimated 0.3 C
1 I I I I
values of kl for the system investigated.
I
0 0.2 0.4
Equation (1) may be modified to include r (mole-equivalents/liter)
the influence of surface elasticity on bubble Fig. 9. The maximum stable bubble size as a function
break-up by suggesting that of ionic strength of solutions of KC1 in water.
B16

have a very small, negative surface excess ACKNOWLEDGMENTS

(more salt in the bulk than at the surface)
and are rather small so no elastic influence
is anticipated. These results suggest that the
decrease in mean bubble size is due to a
decrease in coalescence frequency.
This work was supported by a grant from
the National Science Foundation (CPE
8210724), Division of Engineering, Chemical
and Biochemical Processes program.

3.4. The influence of fermentation media on

bubble break-up
Although eqn. (21) predicts the perfor-
mance of model solutions, it was unclear how REFERENCES
actual fermentation broths would behave. In
order to examine this, typical broths from a J. 0. Hinze, AIChE J., 1 (1956) 289.
yeast and a Lactobaccillus fermentation were M. Moo-Young and H. W. Blanch, Adv. Biochem.
analyzed. Cells had to be removed from the Eng., 19 (1981) 1.
G. K. Batchelor, Proc. Camb. Phil. Sot., 47
broths in order to use the photographic (1951) 359.
technique employed. The broths contained J. F. Walter and H. W. Blanch, Paper 71d, Pre-
considerable surfactants including alcohols, sented at AIChE Annual Meeting, New Orleans,
acids, proteins and salts. The influence of LA, November, 1981.
A. J. Karabelas, AIChE J., 2 (1978) 176.
surface elasticity is evident from the results
C. A. Sleicher, AZChE J., 4 (1962).
shown in Fig. 8. Equation (21) predicts the C. A. Child, M.Sc. Thesis, Univ. of California,
maximum stable bubble size (if the surface Berkeley, 1982.
elasticity of 20 dynes cm-’ is inserted) more T. Vermeulen, G. M. Williams and G. E. Langlois,
closely than eqn. (lo), indicating the need to Chem. Eng. Prog., 2(85-F) (1955).
H. Kimizuka, L. G. Abood, T. Tahara and K.
include surface elastic measurements as part
Kaibara, J. Colloid and Interface Sci., 40 (1)
of the characterization of the liquid. (1972) 27.
10 C. W. Robinson and C. R. Wilke, AIChE J., 20
(1974) 285.

4. CONCLUSIONS

In a turbulent flow field, bubble break-up

is caused by fluctuating eddies which bom-
bard the bubble surface. Break-up occurs by a
dumb-bell stretching mechanism involving
elongation of the bubble surface. Both
viscous and surface tension forces will resist
this elongation and stabilize the bubble
surface. Elongational viscous forces are
responsible for bubble stabilization and result
in a larger maximum stable bubble size.
Surfactants have one of two possible effects
on bubble break-up depending upon the rate
of surfactant replenishment at the surface. As
the bubble surface elongates, the amount of
surfactant per unit area decreases unless more
surfactant is transported to the surface. For
small molecular weight surfactants, the
replenishment rate is very fast and these
substances act to reduce the surface tension
and the maximum stable bubble size. In
general the maximum stable bubble size in an
isotropic turbulent field can be calculated by
eqn. (21).
APPENDIX A: NOMENCLATURE
a interfacial area per unit volume
(cm2 cmM3)
I
rate of surface elongation (s-l)
i surface area (cm2)
Ao surface area at t = 0 (cm2)
G surfactant concentration in bulk
(g mol 1-l)
d bubble diameter (cm)
d, maximum stable bubble diameter (cm)
D impeller diameter (cm)
D max maximum stable bubble diameter
(cm)
E surface elasticity
k, adsorption rate constant (1 g mol-i)
s-l)
k-1 desorption rate constant (s-i)
K constant (defined by eqn. (5))
K consistency index, non-Newtonian
fluids (g cm-’ s-(2--N))
K, mass transfer coefficient (cm s-l)
 B17

L length scale of primary eddies (cm) Greek symbols

1 Kolmogaroff length scale of turbulent P density (g cmm3)
eddies (cm) P shear viscosity (g cm-’ s-l)
N power law index for non-Newtonian I& extensional viscosity
fluids PaP apparent viscosity (g cm-’ s-l)
p/v power per unit volume (W me3) 0 surface tension (dyn cm-‘)
r rate of surface depletion by elonga- 00 surface tension at t = 0 (dyn cm-‘)
tion Y shear rate (s- ‘)
re rate of replenishment of species at 7 shear stress (dyn cmP2)
surface r ionic strength (mol-equiv 1-l)
R Universal gas constant r surface excess (g mol 1-l)
t time (s) r, steady state surface excess (g mol I--1 1
T absolute temperature (K)
u velocity (cm s-l) Subscripts
ii2 root mean square fluctuating velocity C continuous phase
(cm s-l) D dispersed phase
We critical Weber number g gas phase