Oxygen Transfer

Oxygen Transfer Through Gas-Liquid Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

Oxygen and Carbon Dioxide Concentration in Medium. . . . . . . . . . . . . . . . . . . . . 2
Oxygen Concentration in Culture and Oxygen Transfer. . . . . . . . . . . . . . . . . . . . . . 3
Oxygen Diffusion and Order of Magnitude Analysis. . . . . . . . . . . . . . . . . . . . . . . . . 4
Oxygen Transfer through Interface of Two Phases. . Not covered in class . . . . . . . 7
Oxygen Transfer in bioreactors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Oxygen Consumption and CO2 Production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Experimental Measurement of KLa and OUR. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Mass Transfer Resistance in Cell Immobilization Reactor . . . . . . . . . . . . . . . . . . . . . . . . 11
Intraparticle Diffusion Limitation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

Oxygen Transfer Through GasLiquid Interface

Oxygen transfer rate (OTR) = KLa (c*-c) =KLa C
[mmole/lhr] [=] [cm/hr] [1/cm] [mmole/l]
c*: solubility of oxygen,
C: oxygen concentration gradient (i.e., the driving force)
Three factors affecting OTR
Mass transfer coefficient (KL)
Specific transfer area (i.e. the interfacial area)
Driving force (i.e. the gradient across interface)

To improve oxygen transfer

Increase KL (make the interface more turbulent)
Increase a (use smaller bubbles for sparging, or use
silicone tubing)
Increase C (use oxygen enriched air, or dont maintain
dissolved oxygen (C) at an unnecessarily high level.

In cell cultivation, oxygen is transferred from the gas phase

(it could be gas bubbles or the gas space in the flask) through
gas liquid interface into the medium. In order for oxygen
molecules to diffuse to the liquid phase there must be a
concentration difference; the gas phase and liquid phase must
not be at equilibrium. If the two phases are in equilibrium
(i.e. the medium is already saturated with oxygen), there will
not be more net transfer of oxygen into the medium. The
concentrations of oxygen in the gas phase and liquid phase
in a culture are depicted in the figure. In the bulk liquid (i.e.
some distance away from the interface) the concentration of
oxygen is CL. Customarily, oxygen levels in the gas phase
are described as percentages, mole percent or mole fraction
(YO2). If the liquid phase is at equilibrium with the gas
phase, its oxygen concentration is denoted as C*. Closer to
the interface there is a boundary layer or an imaginary film
around each side of the interface wherein the concentration
is different than the bulk. At the edge of the gas film (or gas
side boundary layers) the concentration of oxygen is thus YO2
or C*. On the other side of the gas bubble the concentration
of oxygen is CL. If the two phases are at equilibrium (i.e. C* =
CL), then there is no oxygen transfer. Conversely, the speed at

Oxygen Transfer 1

which oxygen gets transferred is dependent on three factors.

First, the magnitude of the difference between C* and CL has
a major effect on the oxygen transfer rate. The difference
in concentration is a measure of how far the system is from
equilibrium. It is the driving force for the mass transfer.
The second factor is the area of gas-liquid interface that is
available for oxygen transfer. The larger the interfacial area
is, the larger the transfer rate. Normally, the surface area is
expressed as interfacial area per volume of liquid. It thus
has a unit of inverse of length. The third factor is the mass
transfer coefficient. This is a measurement of the degree of
turbulence at the interface. The magnitude of resistance at
the boundary layer is also affected by the molecules that are
being transferred. The mass transfer coefficient (KL) has the
same units as velocity.

Oxygen and Carbon Dioxide

Concentration in Medium
Calculation from Henrys Law
Henrys Law

xA =

PA
H

xA: mole solute A/mole solution

PA: partial pressure of A
H: Henrys law constant
Henrys Law Constants for O2 (atm/mole 02/mole H20)

To be able to estimate how much oxygen can be transferred,

it is important to calculate the saturation level or equilibrium
concentration of the gas species being transferred. For
a sparingly soluble solute, the equilibrium relationship
between gas phase and liquid phase is described by Henrys
Law. It says that concentrations in the liquid phase and
gas phase are linearly related. One can use a variety of
ways to describe the concentration in each phase. Many
handbooks use partial pressure (which is mole fraction of
the gas solute multiplied by total pressure) for gas phase and
mole fraction for liquid phase. For general use it is more
convenient to express liquid phase concentration in mmole/l.
In the medical profession, the gas phase composition is often
expressed in mmHg, (760 mmHg= 1atm).

Henrys Law applies well to an ideal solution. Cell culture

medium contains salts and other nutrients. In spite of all the
4.01
4.38 4.75
5.07 5.18
10-4 x H 2.91
components in it, cell culture medium is still close to an ideal
solution. However, the cell culture medium is frequently
Example: at 37 C, in 1 atm air (P02 = 0.21 atm),
used under 1-20% CO2 environment. Furthermore, in a
O2 concentration in H20 is:
bioreactor, the air pressure is usually greater than 1 atm.
55.5 mole H 2 0 Thus, corrections on oxygen solubility under culture
0.21atm
c* =

conditions need to be made.

H 0
5.18 x 104 atm/moleO /moleH O
T, C

20

25

30

35

37

= 0.22 mmole O2/ H20

In ambient air, (oxygen partial pressure = 0.21 atm), c* ~
0.22 mmole/l in H20 at 37 C = 160 mmHg = 5.8 mg/l (5.8
ppm)

2 Oxygen Transfer

Blood Gas Analyzer Measurement

Temp. (C)

Water
pH
PO2, mmHg
a
6.46
21.5
194.4
6.51
37
158.0a
a. In open air, without CO2
b. In 5% CO2 humidified incubator

PBS
PO2, mmHg
194.8a
157.1a

pH
7.37
7.37

Culture Medium (15 mm NaHCO3

and 10 mm HEPES)
PO2, mmHg PCO2 mmHg pH
131.9b

46.4

7.27

Oxygen solubility is not affected by other dissolved

species in medium. Its solubility is virtually identical
to PBS and water.
Solubility in medium should be adjusted by CO2 used
in gas phase (i.e., the fraction of O2 in gas phase is
usually not 0.21).
Henrys Law Constants for CO2
(atm/mole CO2/mole H2O)
T, C

10

15

20

25

30

35

40

10-3 x H 0.728 0.876 1.04 1.22 1.42 1.64 1.80 2.09 2.33 The solubility of CO2 can be calculated in the same way
using Henrys law. However, CO2 associates with water-

CO2 ( g ) CO2 (aq )

CO2 (aq ) + H 2O H 2CO3 (aq )

H 2CO3 HCO3 + H +
Example: Using10% CO2 in air, CO2
concentration at 37C is

C=

molecules and then dissociates into HCO3 and HT. The

value calculated using Henrys law is the sum of CO2 (aq)
and HCO3 in pure water. One can see that the solubility
of CO2 will be affected by the pH of the solution and the
counter ion. It is not like oxygen for which the solubility is
very similar in pure water and in medium.

PCO 2
H

55.5 mole H 2 O
0.10atm
=

3
2.2 10 atm/mole CO 2 /mole H 2 0
H2 0

=2.2mmole/l

Oxygen Concentration in Culture and

Oxygen Transfer
Oxygen has a low solubility in aqueous medium, as seen
in the previous section. Its solubility under ambient air is
only about 7 mg/l as compared to over 900 g/l for glucose
at room temperature. If cells are grown in a medium saturated with glucose they would have had to undergo osmotic
shock. Therefore, typical glucose concentration for cell cultivation is a few grams per liter. The specific oxygen consumption rates, in terms of mole, of oxygen and glucose are
typically a few fold higher for oxygen. In a solution with
saturated oxygen levels and with a typical glucose concentration, oxygen will get depleted by cell consumption much

Oxygen Transfer 3

faster than glucose. Thus, oxygen will need to be continuously supplied for an aerobe while glucose and other nutrients can only be supplied at the beginning of the culture.

Resistances for O2 transfer

Diffusion to the gas-liquid interface
Solution of the gas into the liquid at the
interface
Diffusion from the liquid side of the interface
into the liquid bulk
Diffusion from the bulk liquid near the surface
of gas bubble to bulk liquid near the
surface of cells
Diffusion from the liquid bulk through liquid
film surrounding the cell
Diffusion through cell clump, pellet or mycelia
Consumption reaction, internal diffusion
Is intracellular oxygen diffusion important?

Oxygen Diffusion and Order of

Magnitude Analysis

The supply of oxygen to cultured organisms serves not

only to prevent it from depleting but also to maintain it in
an optimal range. Most aerobic microorganisms grow well
in a very wide range of dissolved oxygen concentration, up
to saturation with ambient air. However, some cells from
higher organisms grow better under unsaturated concentration. For example, stem cells from mammalian species
have been reported to grow better when the gas used for
cultivation contains 5% O2 instead of 21%. Some tissues,
such as bone marrow, have a lower concentration and cells
from such tissues may prefer somewhat lower oxygen concentration. However, except obligative or facultative anaerobes, all other organisms require oxygen levels be sustained
about certain level, otherwise adverse effects on growth and
survival is seen. This critical level of oxygen is typically
about 20% of saturation with air for aerobic microorganisms (or about 0.01 mM). For mammalian cells, such lung
fibroblasts 3T3 cells, Chinese Hamster Ovary Cells, the
critical level is around 10% of saturation with air at 1atm.
The supply of oxygen to cells in culture is typically achieved
through a gas phase to the liquid phase (i.e. the medium
in which cells are growing). There are a number of steps
that oxygen molecules need to traverse through in order to
reach the cells and eventually the intracellular sites where
the reaction cosuming oxygen occurs. This is described in
the panel. A question to ask is tha among all the steps involved which one is rate limiting or is the controlling step?
A very effective way of identifying the rate controlling step
without resorting to solving a set of differential equations is
by order of magnitude analysis of time constant. In this type
of analysis instead of considering the detailed concentration
changes over time in a system of fixed geometry, one tries
to examine the amount of time (called characteristic time
or time constant) it takes to change a set of concentrations
of a chemical species (characteristic concentration) over a
distance (characteristic length). In such order of magnitude
analysis, one compares the characteristic time needed for
mass (or heat) transfer in different systems that have different
characteristic lengths or a different transfer mechanism. The
step that has the longest time constant is the slowest one.
Consider a solute that is soluble in two different phases
(liquid-gas, gas-solid, liquid-solid, etc.) Given the two phases
co-exist for a long enough time, the concentrations in the two
phases will be in equilibrium. If suddenly the concentration
in one phase is increased to a new level, the solute will be
transferred through the interface into the other phase and the

4 Oxygen Transfer

concentration in the other phase will continue to increase.

For simplicity, let us assume that the concentration at the
interface will remain constant. In other words, the transfer of
solute into the other phase does not cause the concentration
in the first phase to decrease. This would be the case, that
the first phase has a very large capacity (e.g. large volume)
relative to the second phase, so that even some solute is being
transferred into the second phase, the amount transferred is
relatively small to the total in the first phase.
If we look at a slice of the second phase perpendicular to the
direction of transfer and do a material balance on this slice,
the balance will be:
rate of solute
rate of change of solute concentration rate of solute transported rate of solute transported

=
+
+ concentration change
in the control volume
by fluid flow
by diffusion

due to reaction

*Slice is referred to as control volume

Its equation form is shown in the panel. In general, whenever

a solute is carried by a fluid (this type of transport is
convection), its rate of transport in the direction of fluid flow
is far greater (by orders of magnitude) than that from diffusion
unless the flow rate is extremely slow. We will consider
only the case that convection is not involved in transfer and
especially in the case of transfer across the interface. The
convection caused by solute transfer is generally negligible.
So the case we consider is the transport into a stagnant
liquid of a low solubility chemical species (so that the
transport of solute does not cause convection) or in a solid.
By neglecting convection, the equation is simplified.
We first examine a condition that in the medium (i.e. the
stagnant liquid or the solid) no reaction occurs, i.e. as solute
is traveling down the medium it does not get reacted (i.e.
consumed) nor does the medium produce any solute by itself.
The diffusion term is generally described by Ficks Law. The
diffusion is driven by concentration difference between two
points in space. If the concentration in position 2 is greater
than position 2, the solute will diffuse in the direction toward
2 or in the negative direction. In equation form, it says

J (flux of solute) -

c2 c1
C
=
x2 x1
x

C / x is the concentration difference or the concentraton

gradient. J is the flux of the solute. The negative sign accounts

for the direction of transfer being opposite the sign of the
slope. The proportionality between flux and the concentration
gradient is the diffusion coefficient as shown below:
J is a measure of transfer rate (per surface area)
(m / L3 ) / ( L2 T ) = M / ( L T )

Oxygen Transfer 5

normal to the direction of transfer. It has units of

(m / L3 ) / ( L2 T ) = M / ( L T )

The diffusion coefficient has units of L / t . Its magnitude

is dependent on the molecular species and the medium it
diffuses in. For example, oxygen and glucose and small
proteins have very different diffusion coefficients in water
(approximately 10-4 : 10-6: 10-8 cm2/sec for oxygen, glucose
and protein respectively), which the diffusion coefficient
of oxygen in air, lipid and water decreases progressively.
Considering only diffusion in the medium, the change of the
oxygen concentration in the control volume is the difference
between the diffusion at positions x and x + x (assume
diffusion is one dimensional, thus ignoring bottom and top
surface). At surface x, the diffusion rate is -D D

While at x+ox it is D

x
x

C
x + ox .
x

Taking differential between the two planes, and assuming D

is constant, we obtain D

.
x x

The equation for the control volume thus becomes:

2c
22 C
2c
22 C
= D 2 = 0 or
=D
.
2t
2x
2t
2x
Instead of solving the partial differential equation, we can
do an order of magnitude estimation on each term. The left
hand side

2c
can be estimated as
2t

( x )

The order of magnitude for x is C1-C0. Again that can

be estimated as S, the penetration depth or the characteristic
length. So, overall, combining right and left hand sides:

C
C
D 2 .
td
g
g
So, td
D

6 Oxygen Transfer

The characteristic length is usually chosen to represent the

penetration depth. For a long slab with diffusion from the
two surfaces, a long cylinder (both neglecting the diffusion
from the two ends) and a sphere, the characteristic length is
shown in the table. One can calculate the concentration of the
solute as % of the final equivalent to a characteristic time. As
can be seen, after one characteristic time, the concentration
at a characteristic length is very close to equilibrium.
One can estimate the time constant for convection and different
types of reaction kinetics similarly as shown in the Table.

Oxygen Transfer through Interface of

Two Phases

Going back to oxygen transfer problems in bioreactors,

the step that incurs most resistance is the transfer between
gas phase and liquid phase. Microscopically the transfer
can be depicted as oxygen molecules passing through
two stagnant films, one on the gas side, the other on the
liquid side as shown in Figure 1. We have introduced the
equation for the rate of transfer. This overall transfer is
the net result of transfer through the two films. For each
film we can write an equation to describe the rate, using
kL and kg for liquid-side and gas-side transfer coefficient
respectively. A can be considered the same for both gas
film and liquid film. At the interface between the two films,
the two phases can be considered to be in equilibrium,
i.e. Cli=HPgi (recall that at equilibrium the concentration
at the two phases is related by Henrys Law constant).
One can write the equation for oxygen transfer rate
(OTR) in three different forms: overall, liquid-side and
gas-side. Since there is no accumulation in the films, the
rate expressed by the three equations should be equal.
As shown in the panel, by combining liquid side and gas
concentration gradient to give the overall concentration
gradient, one can show that the inverse of Kl is the sum
of the inverse of kl and hg. If one considers K, kl, kg as the
conductivity the 1/KL, 1/kl, 1/kg gives the resistance. In other
words, it shows that the overall mass transfer resistance
1/kl is the sum of the two film resistances, 1/kl and 1kg.
In using kl or kg for calculating OTR, one has to use
the interface concentration (Cli, Pgi) for calculating
concentration gradient. The concentration at interface
cannot be determined readily. By using overall transfer,
the driving force used becomes the P*or C* or the
concentration of the two phases were at equilibrium. For
this case it is a measure of how far away the system is
from the equilibrium. P* and C* can be readily calculated.
Note that one can also use Kg for overall mass
transfer coefficient, that gives OTR=Kga(Pg-P*).
But traditionally Kl is used. We just discussed that the overall
resistance 1/kL is the sum of 1/kL and 1/kg. But what is the

Oxygen Transfer 7

relative magnitude of 1/kL and 1/kg? The magnitude of mass

transfer coefficient is dictated by the diffusion coefficient
and the film thickness (D/). The film thickness for gas
and liquid side is similar. While the diffusion coefficient
on the gas side and liquid side are orders of magnitude
dififerent. It can be shown that the liquid side resistance
dominates the overall resistance. It is thus unusual that kLis
used in place of kL even when overall driving force is used.

Oxygen Transfer in bioreactors

KL has units of L/t (like velocity), and the interfacial area

has units of 1/L. After multiplying by a concentration, the
mass transfer equation gives a concentration perimeter that
is the rate of transfer. The interfacial area term is interfacial
area for oxygen transfer per liquid volume, a/V. A is the total
amount of transfer area in the reactor, V is the total liquid
volume. In a large bioreactor aerated by sparging air bubbles,
the bubbles are not necessarily evenly distributed and local
kL is not necessarily the same. The oxygen transfer rate
discussed in literature is mostly an average over the reactor.
OUR is the product of specific oxygen concentration and
cell concentration (q0). Obviously if OTR is lower than
OUR then dC/dt will be negative and dissolved oxygen
level will decrease. In a culture the dissolved oxygen
level decreases as cell concentration increases. With a
dissolved oxygen, the controller manipulates the agitation
rate, air flow rate or the oxygen content (i.e. by influence
kL or triangle C0 to maintain dissolved oxygen levels at
the set point. But even under uncontrolled conditions that
change of dissolved oxygen level is not fast all the time.
Consider the three terms in liquid phase oxygen balance
equations. If the accumulation term (dc/dt) is small
compared to OTR and OUR, one can consider the system
is at a quasi steady state, i.e. dC/dt ~), thus OTR=OUR.
In a typical bioreactor the oxygen level in the gas phase
(Pg or C*) decreases from the level at the inlet to the
level at the outlet while it traverses through the reactor
as oxygen is being transferred into liquid phase. The
question thus arises as to which C* should be used, the one
corresponding to gas at the inlet or something else. If the
reactor is a small stirred tank that one can consider to be
well mixed, then CL will be the same in the entire liquid
phase, and so is the gas phase concentration. In a well
mixed reactor, one assumes that the sample taken at any
point is the same as that in the well mixed phase. Therefore
C* should be that of the exit gas stream, not the inlet.
In a large reactor, the gas phase is likely to behave closer to
plug flow. One then uses logrithmic mean to calculate the
driving force. This is analogous to using logrithmic mean
driving force in heat transfer in a double-pipe heat exchanger.
Basically, it is an average (but a simple geometrical

8 Oxygen Transfer

average) of the driving force at the inlet and the outlet.

In addition to balance on the liquid, the gas supplying the
oxygen also conforms to material balance. The balance on the
amount of oxygen supplied to the reactor at the intlet and the
amout at the outlet is the amount that has been transferred to
the liquid phase. In principle, in most reactors above the liquid
there is also a headspace with gas and should be considered for
balance, but in general the gas composition in the head space
is considered to be the same as that in the exit. Thus oxygen
transferred into the liquid phase by gas phase is one line
OTRV=(oxygen concentration at inlet)(inlet airflow rate)-(oxygen concentration in outlet)(outlet airflow rate)
OTR (dot V-(oxygen concentration at inlet) dot (inlet air flow
rate) (oxygen concentration at outlet)dot (outlet airflow
rate)=NO2 in dot Qin-NO2out-Qout (asme as in slide)
With a quasi steady state
transferred into liquid phase
as that received by liquid
Furthermore, OTR can be

assumption, the oxygen

by gas phase is the same
phase, both being OTR.
assumed equal to OUR.

If the exit gas (or falled off gas) composition is measured

on line using mass spectrometer or other oxygen gas sensor,
one can easily obtain the volumetric oxygen uptake rate. In
some cases kLa is relative constant, such in the case of some
cells, can be calculated from C if C* is kept constant. If offgas is not measured, one can resort to a dynamic method.
The method involves switching gas to a composition that is
an equivalent to the current dissolved oxygen leve. At that
moment no net transfer of oxygen may occur (because C=0).
As a result the change of dissolved oxygen is only caused
by cell consumption. From the slope one can obtain OUR.

Oxygen Consumption and CO2

Production
Optimal Oxygen Concentration
Most cells grow well at a dissolved oxygen level of 30%
of saturation with air.
When oxygen-enriched gas is used, sometimes the
dissolved oxygen level goes above saturation. A short
duration of above saturation level of oxygen is usually
not detrimental to cells.
The OTR efficiency is higher with a larger oxygen
concentration gradient between gas and medium
(driving force). Maintaining DO at an unnecessarily
high level decreases the driving force.

The objective in oxygen transfer is to supply oxygen at a rate

that is enough to maintain the dissolved oxygen at a level
suitable (or optimal) for cell growth and production. For
most mammalian cells a 30% saturation with ambient air is
sufficient for optimal growth. The amount of oxygen cells
consumed per culture volume is referred to as Oxygen Uptake
Rate (OUR). As you may recall from the kinetics chapter,
OUR is dependent on the specific oxygen consumption
rate (qO2) and the cell concentration. In culture, the change
of oxygen concentration with time is thus the balance
between oxygen transfer rate (OTR) and oxygen uptake rate.
Most oxygen taken up by cells is used to oxidize carbon
sources. The most important carbon sources are glucose
and glutamine. For both the oxidation reaction yield a

Oxygen Transfer 9

Oxygen Demand by Cells

On the average q02 1.0 x 10-10 mmole/cell-hr
dC
= OTR V OUR V
dt
= K L a V (C * C2 ) qO 2 x V

CO2 Production by Cells

Respiratory quotient of cells is about 1.0, i.e. 1 mole CO2
is produced for every mole of O2 consumed, or qCO2
1.0 x 10-10 mmole/cell-hr
A high CO2 concentration is inhibitory
Inhibitory level usually starts at 15% to 20% CO2

Experimental Measurement of KLa and

OUR

Integrate

ln(C * C ) = K L a t

10 Oxygen Transfer

stoichiometric ratio of one for CO2 to O2 (called respiratory

quotient [RQ]). The CO2 produced by cells will need to be
removed from the culture to avoid excessive accumulation
Excess amount of either dissolved O2 or CO2 can
be growth inhibiting or toxic to the cells. However,
the growth inhibitory effect of O2 and CO2 take an
exposure time to be seen. A short exposure of high
concentration (>100% saturation of O2 or nearly 150
mmHg of CO2) up to a few hours, may not be detrimental.
The mass transfer coefficient can be estimated by using some
empirical correlations developed in the form of dimensionless numbers. There has been a report that KL is relatively
insensitive to bubble size except in small scale reactors and
under the conditions that the liquid flow pattern is relatively
simple, like most bubbles rise in a simple pattern; the estimated KL is probably not reliable. Furthermore, except under those conditions, a is also difficult to estimate, as relied
on direct experimental measurement of KLa is often lumped
together and called volumetric mass transfer coefficient, instead of having KL and a estimated or measured separately.
Many bioprocess plants measure gas composition from
the biotreactor inlet and outlet. From the balance equation one can see that the OUR and OTR calculated from
the gas balance can also be used to calculate the average
KLa in the reactor, since C*and C are both measured. One
can also use a dynamic method to measure KLa. One way
to do this is by stripping the dissolved oxygen off using
N2 to replace air (or the original gas phase). In this case

Mass Transfer Resistance in Cell

Immobilization Reactor
Intraparticle Diffusion Limitation

Criteria for assessing the magnitude of mass-transfer effects

on overall kinetics:
Criterion

value

Limiting rate
process

Extent of
mass-transfer
limitation

< 0.3*

~1

Chemical reaction

Negligible

>3

-1

Diffusion

Large

*Or critical, the critical value at which = 1.

Oxygen Transfer 11

The shape of the curve varies with reaction kinetics

and the shape of the particles
Zero order kinetics is often a good assumption
Oxygen is usually the first compound to become
limiting because of its low solubility

12 Oxygen Transfer