Oxygen Transfer Properties of Bubbles in

Animal Cell Culture Media

S. Zhang, A. Handa-Corrigan,* and R.E. Spier
School of Biological Sciences, University of Surrey, Guildford, Surrey GU2 5XH,
United Kingdom
Received September 16, 1991Accepted January 15, 1992

Oxygen transfer rates were determined in a bubble aerated
animal cell bioreactor. It was found that the oxygen trans-
fer rates increased in the following order: large bubbles
(=5 mm diameter) < intermediate bubbles (4 mm diame-
ter) < micron-sized bubbles ( 4 0 0p m diameter).Under cer-
tain conditions, the micron-sized bubbles were capable of
achieving oxygen transfer rates up to 100 h-', a 10-20-fold
higher transfer rate than the large bubbles. The effects of
medium composition on oxygen transfer rates were differ-
ent for the three ranges of bubbles studied. For the large
bubbles, oxygen transfer rates decreased with increasing
medium complexity. The lowest oxygen transfer rate was
found in new-born calf serum (NBCS)and/or Pluronic F-68
supplemented media. For the intermediate and micron-sized
bubbles, supplementation with NBCS into the culture media
resulted in decreased oxygen transfer rate. However, fur-
ther supplementation with Pluronic F-68 enhanced oxygen
transfer rate greatly for both types of bubbles. The highest
oxygen transfer rate was found for micron-sized bubbles in
Pluronic F-68 supplemented media containing antifoam
agent and NBCS.
Key words: animal cell culture bubble K,a surfactant
INTRODUCTION
The supply of oxygen to animal cells in culture can still
cause problems. Although most animal cells have a
lower specific oxygen uptake rate compared with that of
other microbial cells, the oxygen demand rate increases
significantly when the cell concentration and culture
scale increases. A simple and efficient way to meet the
oxygen demand is bubble oxygenation. However, bubble
oxygenation has often been seen as unsuitable in animal
cell culture because of shear damage and foaming prob-
lems.9.1 1,1724 Considerable efforts have been devoted to
the development of alternative complex oxygenation
techniques such as membrane oxygenati~n,'~"~ intensi-
fied surface aeration,'0v23and use of oxygen enriching
chemical^.^,^ Among these techniques, only bubble oxy-
genation is amenable to scaleup despite the problems
associated with it.
In a bubble column bioreactor, it has been demon-
strated that bubble damage of cells occurred predomi-
nantly in the bubble-bursting region at the medium
~ u r f a c e .The
~ same phenomenon was observed in our
laboratory in different stirred and sparged bioreactors.
Part of this article was presented at the 4th Chemical Congress
of North America, August 25-30, 1991, in New York City.
* To whom all correspondence should be addressed.
By the optimization of culture medium formulation and
bubble generation, both cell damage/losses and foaming
problems due to direct air sparging were minimized. In
a recent article by Handa-Corrigan,*it was proposed that
the combination of an appropriate medium formulation
and the use of micron-sized bubbles was the solution to
the problems associated with the oxygenation of high-
density cell cultures. Her reasons for this proposal are
summarized below:
(1) An appropriate medium formulation is required
to combat the foaming problems in sparged animal cell
cultures. Specifically, the foam properties have to be
controlled to prevent losses of cells via a foam flotation
process.'
(2) The micron-sized bubble is most suitable for ani-
mal cell cultures primarily because the surface of the
bubble is saturated with surface-active agents and, there-
fore, behaves predictably as a rigid sphere. In addition,
coalescence problems would be eliminated in surfactant-
rich media by using micron-sized bubbles. Finally, the
oxygen transfer capability of such bubbles would be ex-
pected to be higher than for larger bubbles.
Cultures with cell concentration in the order of
10' cells/mL require 200 pg O2/h," which, from oxygen
transfer point of view, requires a volumetric oxygen
transfer coefficient of 80 h-' for air bubble aeration.
This can only be provided by either a combination of
high-shear turbine and perforated sparger or sinter
sparger that can generate micron-sized bubbles. Because
most animal cells are very sensitive to hydrodynamic
shear stresses," the application of high-shear turbine is
infeasible; hence, micron-sized bubble-generating sinter
sparger has to be used. Although considerable bubble
coalescence was observed at sinter spargers,'6v18it is ex-
pected that, in surfactant-rich animal cell culture media,
bubble coalescence could be effectively reduced due to
the adsorption of surfactants on both the sinter sparger
and the bubble surfaces. By the combination of an ap-
propriate medium formulation (which contains effective
bubble coalescence inhibiting surfactants) and a fine,
sinter sparger, bubble size in the range of 50 p m could
be produced under low-shear conditions in animal cell
bioreactors. In this article we are specifically concerned
with the effects of medium composition on the oxygen
transfer properties of various sizes of bubbles that are
currently used for animal cell cultures.

Biotechnology and Bioengineering, Vol. 40, Pp. 252-259 (1992)

8 1992 John Wiley & Sons, Inc. CCC 0006-3592/92/020252-08$04.00
MATERIALS AND METHODS
Bioreactor
A 2-L (working volume 1.5-1.6 L) round-bottomed bio-
reactor (LH series 210, LH Fermentation, UK) with a
pitched, four-bladed impeller of 58 mm diameter was
used in the studies. The reactor had a diameter of
120 mm and a height of 210 mm. Temperature was main-
tained at 36.5”C by a circulating water jacket around the
vessel. pH was controlled at 7.1-7.4 by sparging COZ
between each K/a measurement. Agitation speed was
controlled automatically from the control panel.
Spargers
Three different spargers were employed to generate
bubbles of different sizes. These included a perforated
stainless steel ring sparger, a G-1 and a G-4 sinter glass
spargers. The ring sparger was supplied by LH Fermen-
tation. It had a diameter of 40 mm with six orifices of
0.5 mm diameter evenly distributed along the ring. The
G-1 and G-4 sinter spargers were made in house from
grade-1 and grade-4 sintered glass disks (Pyrex) to pro-
vide orifice diameters in the ranges of 100-120 pm and
5- 10 pm, respectively. The G-1 sinter sparger had a
diameter of 46 mm and a thickness of 4 mm. The G-4
sinter sparger had a diameter of 35 mm and a thickness
of 2 mm. All these spargers were situated 10 mm under
the impeller. After each experiment the spargers were
thoroughly cleaned using phosphate-free detergent and
tissue culture quality water.
Media
The oxygen transfer rates of bubbles in various media
were examined. The media included:
1. tissue culture quality water (SuperQ, Millipore)
2. 0.8M NaCl solution
3. 50% Dulbecco’s modified Eagle medium (DMEM) +
50% Roswell Park Memorial Institute (RPMI) basal
medium (Flow Laboratory, Irvine, Scotland) (B)
4. basal medium + 50 pprn antifoam agent (B + AF)
5 . basal medium + 50 ppm antifoam agent + 5% New-
born calf serum (NBCS) (B + AF + NBCS)
6. basal medium + 50 ppm antifoam agent + 5%
NBCS + 0.2% Pluronic F-68 (B + AF + NBCS +
Plu .)
All cell culture media were maintained sterile during
oxygen transfer studies.
Preparation of Pluronic F-68 Stock Solution
Pluronic F-68 (BASF, Ludwigshafen, Germany) was dis-
solved in water to give a 10% (wthol) solution. The re-
sultant solution was autoclaved at 121°C for 20 min. The
stock solution was kept at 4°C in a cold room for later
uses. Pluronic was added aseptically to culture medium
to give a final concentration of 0.2% (wthol).
ZHANG, HANDA-CORRIGAN, AND SPIER: OXYGEN TRANSFER OF BUBBLES
Preparation of Silicone Antifoam Stock Solution
Three milliliters of antifoam C emulsion (30%, SIGMA
Cat.# A-8011) was made up to 90 mL in water. The
resultant emulsion was autoclaved at 121°C for 20 min.
The stock solution was kept at 4°C in a cold room until
required for use. A final concentration of 50 ppm was
used in cell culture media.
Measurement of K,a
The oxygen transfer rate of bubble oxygenation can be
evaluated by the following equation:
dC
-= Kla(C* - C)
dt
where Kr is the liquid side oxygen transfer coefficient, a
is the volume-specific gas-liquid interfacial area, C is
the oxygen concentration in the liquid phase, and C* is
the saturation oxygen concentration. The product KIU,
which is termed volumetric oxygen transfer coefficient,
is used to characterize the oxygen transfer capabilities
of bubbles.
There are a number of methods for measuring Kla, of
which the dynamic gassing-out method is the most popu-
lar one due to its simplicity and relative accuracy. The
method consists of lowering the oxygen concentration of
the medium by gassing it with nitrogen, and the increase
in dissolved oxygen concentration is monitored following
the start of aeration. The accuracy of this method de-
pends greatly upon the dissolved oxygen (DO) probe re-
sponse time (the time needed to record 63% of a 100%
stepwise change in DO from O%), the liquid film resis-
tance, and the liquid and gas phase dynamics6 A variety
of models have been proposed to account for the effects
of these factors on K p measurement. Comparison stud-
ies showed that different models gave similar results for
KIUvalues below l/tp (where tp is the DO probe response
time measured under the same conditions as for Klu
measurement), and the effects of DO probe response
time and liquid film resistance on KIUmeasurement were
not significant, especially for medium of low viscosity
such as water.*’ Furthermore, it is expected that in a
small and low shear bioreactor, such as used in the pre-
sent study, the effect of gas phase dynamics should be
negligible because of the very short gas residence time in
the bioreactor. Another possible source of error in KIU
measurement is the experimental starting condition.l5
However, this effect could be minimized by maintaining
the same starting condition during each KIU measure-
ment, which makes the comparison of KI(Ivalues in dif-
ferent media possible. Complying with the conditions
described previously, KIUvalues can be calculated by the
following equation, as suggested by van’t Riet;”
1
Kla =- (2)
t0.63
where 10.63 is the time needed to reach 63% of air satura-
tion starting from 0%.
253
 In this study, nitrogen (oxygen-free) was sparged first
to bring DO level down to 0%. After adjustment of DO
monitor (LH Fermentation) to 0% dissolved oxygen, air 10 -
sparging was initiated under controlled air flow rate
n
and agitation speed. Air flow rate was controlled by an i
-
a-
air flow meter (Platon, UK, 0-600 mL/min). DO \
F
change was followed on a chart recorder (BBC, Goerz- 8-
Metrawatt, Austria). Two or three measurements were Q
carried out for each Kla value. E 4-

2-
DO Probe Characteristics
The DO probe (Ingold) was inserted into the bioreactor 0'
0 100 200 300 400 500 800 700
in between the impeller and the bioreactor wall with the
probe membrane being approximately 70 mm away from Air flow rate (ml/min.)
the bottom of the bioreactor. The response time of the
7
probe was tested before each experiment in order to
ensure the reproducibility of the experimental data. 6-
1 (b)
The DO probe was transferred quickly from a nitrogen-
saturated medium into an air-saturated medium under n -

the same condition as for Kla measurement. Probe re- \

sponse times ranging from 10-13 s were found in the Y
l- 4-

experiments . Further more, the probe was calibrated

ca 3-
every day to prevent any drift effects. si
2-

Measurement of Bubble Size I

A high-speed (& s ) , high-magnification (Xl-120)

black-and-white video camera system (Olympus-KMI,
KeyMed, UK) and a color video printer (YV-lSOE,
Hitachi, Japan) were used to measure bubble sizes in
various media. The video camera system consisted of a
CCD video camera, a X l TV probe adaptor, a X 2 TV
probe adaptor, a high-magnification TV probe, a moni-
tor, and a light source. The video signal can be freeze-
framed by the video printer on the monitor. Satisfactory
frames were printed out from the video printer. Bubble
sizes were measured manually from the prints obtained.
A minimum of 250 bubbles was counted for each sparger
in each medium formulation.
RESULTS
K,8 with the Ring Sparger
Figures l(a) and (b) show the Kla data with the ring
sparger in various media under constant agitation speed
of 60 rpm with variable air flow rate, and under constant
air flow rate of 200 mL/min with variable agitation
speed, respectively. Kla values in the range of 2-6 h-'
were obtained. These are in agreement with the data
reported by Lavery et aI.l3Kla decreased with increasing
medium complexity. The highest Kla values were ob-
tained in 0.8M NaCl solution and water, whereas the
lowest was found in complex culture media with surfac-
tants. The negligible difference in Kla values between
the less-coalescing 0.8M NaCl solution and the coalesc-
0 20 40 80 80 100 120 140
Agitation speed (rpm)
Figure 1. Values of Kla for the ring sparger in various media:
(a) constant agitation speed of 60 rpm; (b) constant air flow rate of
200 mL/min. Symbols show basal medium (closed squares),
B + AF (crosses), B + A F + NBCS (stars), B + A F + NBCS +
Plu. (open diamonds), water (open squares), and 0.8M NaCl solu-
tion (open triangles).
ing water suggests that no significant bubble coalescence
has occurred because the ring sparger had widely spaced
orifices, and a relatively low air through put. Kla in-
creased with increasing agitation speed and air flow rate.
K,a with the G-1 Sinter Sparger
Figures 2(a) and (b) show the Kla data with the G-1
sinter sparger in various media under constant agitation
speed of 60 rpm with variable air flow rate, and under
constant air flow rate of 200 mL/min with variable agi-
tation speed, respectively. Kla values in the range of 5-
50 h-' were found in the media studied. These were
about fivefold higher than those obtained with the ring
sparger. Unlike the ring sparger, Kla did not decrease
consistently with the medium complexity. Under con-
stant agitation condition [Fig. 2(a)], the highest Kla val-
ues were found in basal medium, B + AF + NBCS +
Plu., and 0.8M NaCl solution. The lowest Kla value was
found in water. Addition of 5% NBCS to the 50 ppm

254 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 40, NO. 2, JUNE 20, 1992
 K,a with the G-4 Sinter Sparger
(a)
50 - Figures 3(a) and (b) show the Kla data with the G-4
sinter sparger in various media under constant agitation
n
40- speed of 60 rpm with variable air flow rate, and under
f
7
constant air flow rate of 200 mL/min with variable agi-
w 30- tation speed, respectively. Kla values in the range of 20-
cp
100 h-' were obtained, which were 10-20-fold higher
2 20- than those obtained with the ring sparger. The highest
Kla value was found in 0.8M NaCl solution, while the
10 - lowest was found in water. This suggests that bubble co-
alescence occurring at the sparger and/or in the bulk has
" an important effect on K,u for the G-4 sinter sparger,
0 100 200 300 400 500 600 700 which is in contrast to the results obtained with the ring
Air flow rate (ml/min.) sparger. Unlike the ring and G-1 sinter spargers, ad-
dition of 50 ppm antifoam to basal medium increased
120 Kla. Further supplementation with 5% NBCS induced a
(b) significant drop in Kla. However, inclusion of 0.2%
100 - Pluronic F-68 into the culture medium increased Kla
n
L greatly. Theoretically, Pluronic F-68 containing culture
f 60- medium and sparging with the micron-sized bubbles
r
W could supply enough oxygen to a cell concentration in
60 -
(0
z 40 -

20 -

0 20 40 60 80 100 120 140 160

Agitation speed (rpm)

Figure 2. Values of Kla for the G-1 sinter sparger in various media:
(a) constant agitation speed of 60 rpm; (b) constant air flow rate of
200 mL/min. Symbols as in Figure 1.

antifoam containing basal medium induced a 33% drop

in &a. However, further supplementation with 0.2%
0
Pluronic F-68 increased Kla significantly. Different Kla 0 100 200 300 400 500 600 700
profiles were observed under constant air flow rate con-
dition [Fig. 2(b)]. Very high &a values were observed in
Air flow rate (ml/min.)
basal medium, B + AF, and 0.8M NaCl solution when
agitation speed was higher than 80 rpm. This is at- (b)
tributed to the increased mixing in the bioreactor as the 100 -
agitation speed increased. Under agitation speed lower
than 60 rpm, the bubbles rose directly upward through
the middle of the bioreactor to the medium surface
without any contact with the DO probe membrane.
When the agitation speed was increased over 80 rpm,
- 601

the bubbles were distributed more uniformly in the 401

whole bioreactor because of the increased mixing, which
created the possibility of direct bubble contact with the
DO probe membrane. In the three media mentioned
previously, some bubbles were found attached to the DO
probe membrane, which could explain the very high Kla
2ot
"
o 20 40 60 eo 100 120 140

values observed at the higher agitation speed. Similar Agitation speed (rpm)
observations were reported by Linek and Vacek.15 Gen- Figure 3. Values of Kla for the G-4 sinter sparger in various media:
erally, however, Kla increased with increasing agitation (a) constant agitation speed of 60 rpm; (b) constant air flow rate of
speed and air flow rate. 200 mL/min. Symbols as in Figure 1.

ZHANG, HANDA-CORRIGAN, AND SPIER: OXYGEN TRANSFER OF BUBBLES 255

the order of lo8 cells/mL assuming a constant cell spe-
cific oxygen uptake rate of 2 pg 02/106 cells h.*’

Bubble Size and Shape

Bubbles generated from the different spargers in various
media were photographed. For the ring sparger, large
ellipsoidal bubbles with surface movements (observed on
the video monitor) were found in water, 0.8M NaCl so-
lution, and basal medium only, while ellipsoidal bubbles
with no observable surface movement were observed in
all the other media. The ellipsoidal bubbles were char-
acterized by the diameter of the corresponding volume-
equivalent sphere, which is defined as

where V, is the volume of the ellipsoidal bubble.

Spherical bubbles of different sizes were found for
both the G-1 and G-4 sinter spargers in all the media
studied. These bubbles rose through the media without
any observable surface movement. For the G-4 sinter
sparger, culture medium containing 0.2% Pluronic F-68
was “milky white” because of the formation of micron-
sized bubbles. Figure 4 shows the comparison of micron-
sized bubbles generated in Pluronic F-68 containing
culture medium with the bubbles generated in medium
without Pluronic F-68 from the G-4 sinter sparger. Ap-
proximate normal distributions of bubble sizes were
found for the different spargers, as shown in Figure 5.
Sauter mean bubble size, which is defined as,

Figure 4. Comparison of bubbles generated from the G-4 sinter

(4) sparger: (a) micron-sized bubbles in B + AF + NBCS + Plu.;
(b) bubbles in B + AF + NBCS.

was used to represent the bubble size data, where ni is
the number of bubbles with diameter of di. Table I shows
the Sauter mean sizes of bubbles generated from the dif-
ferent spargers in various media. For the ring sparger,
there is no significant difference in bubble sizes between
water and 0.8M NaCl solution. This supports the sugges-
tion that bubble coalescence is negligible in small, low-
shear animal cell bioreactors fitted with perforated
spargers having widely spaced orifices, and relatively low
air throughput. More than 100-fold reduction in bubble
size from the ring sparger was effected when the G-4
sinter sparger, producing micron-sized bubbles, was used
in culture medium containing Pluronic F-68.
DISCUSSION
Ring Sparger
For the large, ellipsoidal bubbles generated from the
ring sparger, the changes in Kla in various media are
attributed to the variations of bubble surface movement
caused by the adsorption of surfactants on the bubble
surfaces.2As the bubbles rise through the medium, the
adsorbed surfactants are swept toward the rear part of
the bubble due to the tangential shear force.’ The non-
uniform distribution of surfactants results in the buildup
of a surface pressure resisting further compression of
surfactants along the bubble surface. The relative sur-
face area covered by surfactants increases with in-
creasing surfactant concentration in the medium and
decreasing bubble size (because of the reduced tangen-
tial shear force). In other words, the surface mobility
decreases with increasing surfactant concentration and
decreasing bubble size. According to Danckwerts’ sur-
face renewal theory: reduced surface mobility leads to
decreased liquid side oxygen transfer coefficient K,. In
the relatively clean water and 0.8M NaCl solution, where
bubbles have the highest surface mobility, the highest
Kla was obtained. In the basal medium, where bubble
surface movement is partially retarded, lower Kla was
obtained. In basal medium supplemented with 50 ppm
antifoam agent, the complete elimination of bubble sur-

256 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 40, NO. 2, JUNE 20, 1992
 0.6 I 1 ing surfactant molecules will not have any available gas-
liquid interface for adsorption. Once fully saturated with
medium components, a minimum Kl is reached for the
large, ellipsoidal bubbles. Therefore, this is reflected in
the similar Kla values and bubble sizes observed for all
culture media other than basal.

G-1 Sinter Sparger

”
2.6-3.0 3.0-3.5 3.6-4.0 4.0-4.5 4.5-6.0 Adsorbed surfactants have two rather conflicting effects
Bubble size (mm) on K,a. One is the reduced K, caused by the retarded
surface movement and possibly increased resistance to
oxygen transfer. The other is the reduced bubble coales-
cence that leads to an increased a. The eventual effect
on Kla will depend on the relative magnitude of reduc-
tion in KI and increase in a. The reduction in K,a after
addition of 50 ppm antifoam agent into the basal me-
dium is believed to be due to the increased bubble co-
alescence as indicated by the large bubbles generated
(Table I). Incorporation of NBCS caused further reduc-
tion in K,a, even though no significant change in bubble
260-300 300-350 360-400 400-460 450-600 size was found. This result indicates that in the presence
Bubble size (urn) of NBCS, a large reduction in K , occurs. NBCS is com-
posed of various surface active proteins that are readily
0.6, 1 adsorbed at the gas-liquid interfaces. Surfactants ad-
sorbed at the surfaces of small, spherical bubbles gener-
ated from the G-1 sinter sparger face less tangential
shear force compared with the large, ellipsoidal bubbles.’
As a result, a condensed surfactant film is easily built
up as the surfactant concentration increases in the me-
dium. Proteinaceous films have unusually high surface
viscosity that could reduce not only the bubble surface
movement but also the oxygen diffusivity. The combined
”
46-60 50-00 80-70 70-80 80-90 90-100 effect results in the observed drop in &a. Meanwhile,
Bubble size (urn) supplementationwith 0.2% Pluronic F-68 into the NBCS-
Figure 5. Bubble size distributions in B + AF + NBCS + Plu.:
containing medium increased KIa greatly. This is at-
(a) ring sparger; (b) G-1 sinter sparger; (c) G-4 sinter sparger. tributed to the overriding increase in a as indicated by
the smaller bubbles generated (Table I). Pluronic F-68 is
Table I. Sauter mean bubble sizes in various media (mm). a block copolymer of polyoxyethylene (POE) and poly-
oxypropylene (POP) with a molecular weight of about
Ring G-1 sinter G-4 sinter
8400. It has the following general structure:’’

Water 3.82 20.63 1.60 k0.55 1.25 20.53 HO(CH,CH,O)a(CH(CH,)CH,O)p(CH~CHZO)TH

0.8M NaCl solution 3.84 20.72 0.53 k0.09 0.14 -1-0.04
Basal medium 3.54 20.50 0.78 20.22 0.52 20.18 where LY and T are statistically equal. In the molecule,
B + AF 4.01 20.42 1.17 20.28 0.63 20.22 POE and POP are the lyophilic and hydrophobic parts,
B + AF + NBCS 3.98 50.45 1.23 20.30 0.61 20.16 respectively. Pluronic F-68 molecule consists of 80% of
B + AF + NBCS + Plu. 3.71 20.40 0.37 20.04 0.067 20.01 POE and 20% POP. Because of its spatial configuration
in aqueous media and its relatively low molecular weight,
face movement resulted in the further reduction in Kla. Pluronic F-68 is more surface active than NBCS proteins,
Addition of NBCS and Pluronic F-68, however, did not which allows it to be adsorbed more easily at gas-liquid
cause any significant change in Kla, because the added interfaces. The hydrophobe (POP) in the molecule is re-
NBCS or Pluronic did not affect either the bubble surface jected from the medium onto the gas-liquid interface,
mobility or the bubble sizes (Table I). It is expected that leaving the lyophilic POE moieties extending into the
in culture media other than basal, a saturated, rigid sur- medium in a state of hydrated random coil,” which in-
factant film is formed around the bubbles, and further hibits effectively the bubble coalescence at the sparger
increases in surfactant concentration will not change and/or in the bulk. Consequently, very small bubbles are
the bubble surface composition because the newcom- generated. However, the increase in K,a is less than that

ZHANG, HANDA-CORRIGAN, AND SPIER: OXYGEN TRANSFER OF BUBBLES 257

expected from the reduction of bubble sizes. It is pro-
posed that as the bubble size decreases, the Kl is also
decreased by the accumulation of surfactants on bubble
surfaces.
G-4 Sinter Sparger
Except in water, very small spherical bubbles were gen-
erated from the G-4 sinter sparger in all the media stud-
ied. Because of the very small size, and consequently
the relatively high residence time, the bubble surface is
expected to be rapidly saturated with surfactants and
surface movement effects can therefore be neglected.*In
this case, changes in Kla can only be effected by varia-
tions in a and/or oxygen diffusivity due to the adsorbed
surfactants. Unlike with the ring and G-1 sinter sparg-
ers, addition of 50 ppm antifoam into the basal medium
resulted in an increased &a. This is suggested to be
the result of an increased oxygen diffusivity across the
bubble surface. In the presence of antifoam agent, the
bubble surfaces are preferentially covered by a layer of
the adsorbed antifoam molecules that imparts its own
property to the bubble surfaces. Silicone oil antifoam
has not only very low surface viscosity but also high oxy-
gen diffusivity and solubility. All these properties are in
favor of increasing the liquid side oxygen transfer coef-
ficient, Kl. Further supplementation with NBCS, how-
ever, reduced Kla by a factor of two even though no
obvious change in bubble size was observed (Table I). In
the presence of NBCS, a rather dense and viscous
proteinaceous film is likely to be formed around the al-
ready rigid bubbles. Compared with the more fluid and
oxygen soluble antifoam film, significant reduction in
oxygen diffusivity and solubility will be expected in the
presence of NBCS, which leads to the reduced K p . Ad-
dition of 0.2% Pluronic F-68 into the NBCS-containing
medium greatly increased the k a . This is attributed to
the nearly 10-fold reduction in bubble sizes, i.e., the for-
mation of micron-sized bubbles. The adsorbed Pluronic
molecules are likely to not only decrease the bubble co-
alescence rate but also to reduce Kl, because a much
smaller increase in k a was obtained than that expected
from the reduction in bubble size. Further experiments,
which measure the oxygen diffusivity in various media,
are needed to be carried out in this respect.
Kla Correlation
Power input per unit volume and superficial gas veloc-
ity have been widely used as the correlating parameters
for Kla in stirred and sparged reactors.25Comparison of
the results obtained in this work with Kla values esti-
mated from published correlations, however, is inap-
propriate because of the extremely low power inputs
(estimated to be in the range of 0.1-1 W/m3) and rela-
tively low gas flow rates employed. In fact, in our work
very different K@values were obtained for different me-
258 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 40, NO. 2, JUNE 20, 1992
dia formulations under the same power input and super-
ficial gas velocity. We have attempted to correlate the
K,u data from this work with power inputs, superficial
gas velocities, and bubble diameters. Although consis-
tent correlations have been obtained for each type of
medium (correlations not shown), a satisfactory cor-
relation, which could be applied to all the media for-
mulations, cannot be realized. We suggest that other
important parameters, such as oxygen permeability and
surface viscosity (not bulk viscosity), need be evaluated
in order to take media properties into consideration in a
K/a correlation for animal cell cultures.
CONCLUSION
Different K/a profiles were obtained for the three size
ranges of bubbles studied. For the large ellipsoidal
bubbles, &a decreased consistently as the medium com-
plexity increased until minimum &a values were reached
in the fully formulated culture media. For large bubbles,
in cell culture media &a is controlled by bubble surface
movement or the liquid side oxygen transfer coefficient
K , rather than bubble coalescence. For the intermediate-
sized and micron-sized bubbles, addition of NBCS into
the antifoam containing basal medium resulted in a sig-
nificant drop in K,a under the same agitation and air
flow conditions. Further supplementation with Pluronic
F-68, however, gave Kla values higher than those ob-
served for all other cell culture media tested (basal me-
dium, B + AF and B + AF + NBCS). Theoretically,
Pluronic F-68 containing medium and sparging with the
micron-sized bubbles could supply enough oxygen to
support a cell concentration in the order of 10' cells/mL.
From an oxygen transfer point of view, large ellipsoidal
bubbles are not appropriate for oxygenating animal cell
cultures because the inevitable supplementation of basal
medium with serum, antifoam agent, or Pluronic F-68
results in decreased oxygen transfer rates. Micron-sized
bubbles are recommended for use in the oxygenation of
animal cell cultures because of their high oxygen supply
capabilities. To exploit the oxygen transfer potential of
micron-sized bubbles fully, it is suggested that Pluronic
F-68 be incorporated into the culture medium, while
keeping serum concentration to a minimum.
The financial support of the Chinese Government and the
British Council are acknowledged. The technical assistance
of Marion Chadd is most appreciated.
NOMENCLATURE
specific gas-liquid interfacial area (m-')
oxygen concentration (pg/mL)
saturation oxygen concentration (pg/mL)
bubble diameter (mm)
Sauter mean bubble diameter (mm)
volume equivalent sphere diameter (mm)
liquid side oxygen transfer coefficient (s-' m)
volumetric oxygen transfer coefficient (h-')
 N agitation speed (rpm)
Q air flow rate (mL/min)
t time (s)
tp dissolved oxygen probe response time (s)
v, volume of an ellipsoidal bubble (mm’)
1. Bavarian, F., Fan, L. S., Chalmers, J. J. 1991. Microscopic visu-
alization of insect cell-bubble interactions. I: rising bubbles,
air-medium interface, and the foam layer. Biotechnol. Prog. 7:
140-150.
2. Clift, R., Grace, J. R., Weber, M. E. 1978. Bubbles, Drops, and
Particles. Academic Press, New York.
3. Damiano, D., Wang, S. S. 1985. Novel use of perfluorocarbon
for supplying oxygen to aerobic submerged cultures. Biotechnol.
Lett. 7: 81-86.
4. Danckwerts, P.V. 1951. Significance of liquid-film coefficients
in gas absorption. Ind. Eng. Chem. 43: 1460-1465.
5. Davies, J.T., Rideal, E. K. 1963. Interfacial Phenomena. 2nd
edition. Academic Press, New York.
6. Dunn, I. J., Einsele, A. J. 1975. Oxygen transfer coefficients by
the dynamic model. J. Appl. Chem. Biotechnol. 25: 707-720.
7 . Hamamoto, K., Tokahiki, M., Ichikawa, Y., Murakami, H .
1987. High cell density culture of a hybridoma using perfluoro-
carbon to supply oxygen. Agric. Biol. Chem. 51: 3415-3416.
8. Handa-Corrigan, A. 1990. Oxygenating animal cell cultures:
the remaining problems. p. 123-132 In: R. E. Spier and J. B.
Griffiths (ed.), Animal cell biotechnology, vol. 4. Academic
Press, London.
9. Handa-Corrigan, A., Emery, A. N., Spier, R. E. 1989. Effect of
gas-liquid interfaces on the growth of suspended mammalian
cells: mechanisms of cell damage by bubbles. Enzyme Microb.
Techno]. 11: 230-235.
10. Hu, W. S., Meier, J., Wang, D. I. C. 1986. Use of surface aerator
to improve oxygen transfer in cell culture. Biotechnol. Bioeng.
28: 122-125.
11. Kilburn, D. G., Webb, F. C. 1968. The cultivation of animal
cells at controlled dissolved oxygen partial pressure. Biotech-
nol. Bioeng. 10: 801-814.
12. Kuhlmann, N. 1987. Optimization of a membrane oxygenation
system for cell culture in stirred tank reactors. Develop. Biol.
Standard. 66: 263-268.
ZHANG, HANDA-CORRIGAN, AND SPIER: OXYGEN TRANSFER OF BUBBLES
13. Lavery, M., Nienow, A.W. 1987. Oxygen transfer in animal cell
culture medium. Biotechnol. Bioeng. 30: 368-373.
14. Lehmann, J., Piehl, G.W., Schulz, R. 1987. Bubble free cell
culture aeration with porous moving membranes. Develop.
Biol. Standard. 66: 227-240.
15. Linek, V., Vacek, V. 1977. Dynamic measurement of the volu-
metric mass transfer coefficient in agitated vessels: effect of
the start-up period on the response of an oxygen electrode.
Biotechnol. Bioeng. 19: 983-1008.
16. Marrucci, G., Nicodemo, L. 1967. Coalescence of gas bubbles
in aqueous solutions of inorganic electrolytes. Chem. Eng. Sci.
22: 1257-1265.
17. Papoutsakis, E.T. 1991. Fluid-mechanical damage of animal
cells in bioreactors. Trends Biotechnol. 9: 427-437.
18. Prakash, A., Briens, C.L. 1990. Porous gas distributors in
bubble columns. Effect of liquid presence on distributor pres-
sure drop. Effect of start-up procedure on distributor perfor-
mance. Canad. J. Chem. Eng. 68: 204-210.
19. Rosen, M . J. 1989. Surfactants and Interfacial Phenomena. 2nd
edition. John Wiley & Sons, New York.
20. Ruchti, R., Dunn, I. J., Bourne, J. R. 1981. Comparison of dy-
namic electrode methods for the measurement of Kra. Biotech-
nol. Bioeng. 23: 277-290.
21. Schmolka, I. R. 1967. Polyalkylene oxide polymers. p. 308-332
In: M. J. Schick (ed.), Nonionic surfactants. Marrel Dekkar
Inc., New York.
22. Spier, R. E., Griffiths, B. 1984. An examination of the data and
concepts germane to the oxygenation of cultured animal cells.
Develop. Biol. Standard. 55: 81-92.
23. Sybert, E. M., Winter, E. L., Cadman, T.W., Conner, M. A.
1989. Effect of agitation rate and impeller design on oxygen
transfer coefficients in small bioreactors using surface aeration.
BioTechniques. 7: 368-373.
24. Tramper, J., Smit, D., Straatman, J., Vlak, J. M. 1988. Bubble-
column design for growth of fragile insect cells. Bioprocess
Eng. 3: 37-41.
25. van’t Riet, K. 1979. Review of measuring methods and results
in nonviscous gas-liquid mass transfer in stirred vessels. Ind.
Eng. Chem. Process Des. Dev. 18: 357-363.
259