Bioprocess Engineering 3 (1988) 123 127
9 Springer-Verlag 1988
Studies of insufficient mixing in bioreactors: Effects of limiting oxygen concen-
trations and short term oxygen starvation on Penicillium chrysogenum
G. Larsson and S.-O. Enfors, Stockholm
Abstract. The effect of oxygen limitation on the respiration rate of
Penicillium chrysogenum was studied. The results show that mea-
surements of critical oxygen tensions within a process that on mor-
phological or on physical grounds exhibits an inhomogenous struc-
ture are not likely to resemble the Monod model.
In order to study the effects of short term oxygen starvation on
the respiratory capacity of Penicillium chrysogenum, a two com-
partment fermenter was constructed. This fermenter consists of one
well mixed aerobic part (CSTR) and one minor anaerobic part
(CPFR). In the latter the circulation time as well as the volume can
be varied. After passage of the whole cell culture volume through the
anaerobic part, irreversible inhibition of the respiration was ob-
served. This was caused by a circulation time of 5 and t0 min in the
plug flow reactor and with a volume of 6% of the stirred tank
reactor volume. However, circulation times of 1 and 2 min with an
anaerobic zone of t % of the stirred tank reactor volume did not give
any irreversible effects on the respiratory capacity.
This was compared with the results of the previously established
model ln(t -/ouR/t00)-1 = k t [1]. The Iou~ is the percentage irre-
versible inhibition of the respiration, t is the anaerobic circulation
time and k is a constant. The two compartment fermenter results
agree with the earlier model at circulation times of 5 and 10 rain, but
not with the shorter times, and this suggests that a lag phase exists
in the inactivation kinetics.
1 Introduction
As an effect of insufficient mixing, the oxygen transfer rate in
the fermenter will vary. In c o m b i n a t i o n with the low solubil-
ity of oxygen in fermentation broths and high oxygen con-
sumption rates this will give local variations in the oxygen
concentration in a fermenter. The culture may therefore
suffer from oxygen limitation (i.e. a decrease in oxygen con-
centration gives a reduction of the maximal respiration rate),
or even oxygen starvation (no oxygen present). It has been
shown by Oosterhuis and Kossen [2] that this is true even for
a low viscous broth.
The cells are exposed to oxygen limitation to the extent
determined by the residence time distribution in these zones
in the fermenter. In order to understand the performance of
a fermenter, it will therefore be necessary to know this distri-
b u t i o n and the actual concentration in the zones. However,
it is also necessary to take into consideration the biological
Bi0processEngineering
response of the cell to these changes of the microenviron-
ment. Vardar and Lilly [3] show that cycling oxygen concen-
trations reduce the product formation rate, and when the
respiration rate is reduced to 90% of the m a x i m u m value no
product is formed at all. It has also been shown by Sweere
et al. [4] that, using a two fermenter system, fast fluctuations
in substrate concentration can to a large extent influence the
production of baker's yeast.
The usual way of visualizing limiting oxygen conditions
is by using the M o n o d model. In practice, the oxygen con-
centration is changed by reducing the stirrer speed and by a
mass balance of oxygen the respiration rate is calculated.
However, this results not only in a reduction of the oxygen
concentration in the bulk phase, but also in an interference
with the total mass transfer situation. One effect is an in-
crease in the thickness of all the b o u n d a r y layers, and there-
by also an interference with the oxygen transfer to the cells.
There will also be a disturbance of the transport of other
substances that could effect the results, for example the
transport of carbon dioxide.
In production scale fermentation, one also has to con-
sider the effect of oxygen starvation. It was shown in earlier
experiments with Penicillium chrysogenum [1] that the sensi-
tivity of the respiration system to oxygen starvation varies in
the process, and in the most sensitive phase, the tropophase,
it follows first order kinetics according to eq. (1):
l n ( l - I O U R / 1 0 0 ) - 1 = k t. (1)
It was not possible to measure the effect of shorter anaerobic
residence times than 10 min with this method. These shorter
periodes are of course of most interest, because they are of
the same order of magnitude as the expected residence times
in dead zones in a fermenter.
This study shows a way of determining the critical oxygen
concentration without interference with the mixing, and by
the use of the two-compartment fermenter a simulation of
the situation in a large insufficiently mixed fermenter is pro-
vided to show the effect of repeated short term oxygen star-
vation on Penicillium chrysogenum in a small part of a fer-
menter.
124 Bioprocess Engineering 3 (1988)

2 Materials and methods % 02 %

2.1 Cultivation .~_4 9ZOo~

Three strains of Penicillium chrysogenum have been used in ~3
o pH 19~
this study, all kindly provided by Fermenta AB, Sweden.
They are all indicated in the results. Spores were produced ~ 2 18
on rice cultures and harvested with distilled water and stored c.

in a 10% skim milk solution at - 2 0 ~ until used. 105 spores 1 17

per cm 3 of broth were inoculated directly in the fermenters. 5-
The early progress of a cultivation is shown in Fig. 1. -- 0 3'0 4'0 h 5'0 16
The media used was: 8.5% cornsteep liquor, 0.6% Timet
(NH~) 2 SO4, 0.7% CaCO3, 0.064% H3POr 0.15% maize oil, Fig. 1. Batch cultivation of Penicillium chrysogenum. Progress of
0.25% phenoxyacetic (pH 10), 2% glucose, and 0.00625% pH, CO z and O z in the outlet gas. Time taken from inoculation with
Adecanol, which was provided by Fermenta AB for foam spores in the fermenter
control. The precursor and the glucose were autoclaved N2
separately and fed to the broth before inoculation. Before
sterilization, pH was adjusted to 6.0 and the inoculation was 02-el.
9~ ! 02-et.
made at pH 5.9. The fermentation temperature was 25 ~
]~ ) Outlet

!2
During the spore germination phase, the stirrer speed was
550 rpm and the aeration rate was 0.25-0.5 vvm. At about
30 h, when the vegetative growth started, the stirrer speed , ~ Mogneticstirrer
and the aeration rate were increased to 1.500 rpm and
1 vvm, respectively. When p H reached 6.8, it was kept con-
stant by controlled addition of a 2.5 M H z S O 4 solution until
about 70 h, when p H tended to decline. Then a 2.5 M N a O H
solution was used for further control. When pH reached
]"
6.8, a glucose solution (45% w/v) containing (N}-I4)2SO 4
(1.6% w/v) was fed intermittently. A timer controlled addi-
tion of 65 cm 3 of solution every second hour.
Wb
.! Dr
.? =@==

2.2 Fermenters Fig. 2. Experimental setup for the two compartment reactor. The
anaerobic plug flow reactor is shaded in the figure. For dimensions
The fermenters used were, for the critical oxygen deter- see section Materials and methods
mination, a standard fermenter with a volume of 10 dm 3 and
a working volume of 8 dm 3 equipped with two impellers
(Chemoferm AB, Sweden) and for the oxygen starvation
on the line to be able to take samples for controlling the
study, two identical fermenters with a volume of 20 dm 3 and
oxygen level within the inlet gas stream. The same was done
a working volume of 15 dm a (Electrolux Fermentation AB,
at the outlet stream. The total respiration rate was calculated
Sweden). These latter fermenters (Fig. 2) have the following
by a mass balance of oxygen. The study was carried out
dimensions (in mm): W = 10, Wb = 25, H~ = 520, D i = 105,
immediately after a sugar dose, when the response of the
H x = 3 1 0 , H i = 1 9 0 , H s = 4 1 0 and D 1 = 3 1 5 . The sparger
dissolved oxygen signal had become constant. This effect
outlet was 2 m m in diameter. All the fermenters were
was followed on the p H monitor. All through the experi-
equipped with pH- and oxygen electrodes. Galvanic elec-
ment, the stirrer speed and the total flow rate were kept
trodes of laboratory design were used in the critical oxygen
constant. By mixing the air in the inlet gas stream with
determinations and polarographic electrodes (Ingold) were
nitrogen, it was possible to reduce the oxygen content. At
used in the other experiments. The exit gases were analyzed
each level of oxygen tension, the oxygen in the inlet and in
for oxygen by a paramagnetic oxygen analyzer (Servomex
the outlet were analyzed as well as the dissolved oxygen
540A, Taylor Instruments Comp., N. Y., USA), and for car-
content in the medium. The limitation study was carried out
bon dioxide by a Beckmann 864 infrared analyzer (Beck-
during the production phase. It was done at 95 h after inocu-
mann Instr., Inc., CA, USA).
lation and at 0.75 vvm and 850 rpm.
2.3 Determination of the critical oxygen tension
2.4 Determination of the effect of oxygen starvation
A setup of flowmeters (rotameter) was situated before the air
inlet in the fermenter, one for air, one for nitrogen, and as the The equipment for this study consists of two fermenters, one
two lines met, one for the total flow. A by-pass was installed for the experiment and an identical one as a reference. The
G. Larsson et al.: Effects of limiting oxygen concentrations and short term oxygen starvation on Penicillium chrysogenum 125

experimental fermenter is shown in Fig. 2. It consists of two 50

mM/h
parts, one aerobic with a volume (V) of 15 dm 3, and one
anaerobic with a volume (VAA) of 900 or 150 cm 3, respec- 40
tively. This anaerobic volume is indicated in the figure. The
anaerobic chamber consists of a silicon tubing, that was kept 4 30
in a water bath of 25 ~ together with sodium sulphite and
copper sulphate to avoid oxygen diffusion into the tubing.
Before the experimental start the oxygen transfer was 9& 20 f,v
adjusted to avoid critical concentrations during the experi- c~ r
ment according to the earlier oxygen limitation studies.
The mycelium was pumped through a stainless steel tube
from the bottom part of the fermenter, through the tubings,
and was led via a nitrogen chamber, which had an outlet at 0 10 20 30 40 50 60 70 % 80
the side near the bottom. As the mycelium dripped through Dissolved oxygen tension(in % of oir soturotion)
the chamber, the dissolved oxygen was rapidly consumed Fig. 3. Total respiration rate in a 10 dm 3 fermenter as a function of
according to the oxygen electrode in the chamber. Traces of the dissolved oxygen tension changed by mixing the inlet air with
oxygen could be found in the gas phase and it was therefore nitrogen
carefully checked that no gas bubbles would go down the
anaerobic coil by keeping the media level above the outlet. 100 o
This amount of the broth was carefully stirred with a magne-
tic stirrer to avoid settling of the mycelium. The circulation
time through the anaerobic part was varied by changes of
the pump speed. The mycelium was then pumped back to ~~ 6o
~,~
the fermenter and released at the top of the head space to be
"~ ~ 40
immediately oxygenized.
The respiration rate, 0c. q), was calculated for both fer- ~ 20
menters by a mass balance. The quotient of the respiration
rate between the test and the reference culture was used to
calculate the inhibition of the respiration according to
o ~o io 3b Zo 5~ rain 60
Time t
Eq. (2):
Fig. 4. Irreversible inhibition of the oxygen utilization rate as a
Redresp = 100 -- (Ft/F~) (qt/qr)" 100 (2) function of time without oxygen.
9 Experimental observations with strain E t5
[] Experimental observations with strain PC 174
where q is the respiration rate and F is the air flow rate of the The lines are fitted to eq. (t) with k = 0.045 (upper curve) and
fermenters. Indexes r and t refer to reference and test culture, k = 0.065 (lower curve).
respectively.
After the experiments, a recalibration was made of the
pump flow rates and the anaerobic fermenter volume with this study is included in Fig. 4, together with the earlier
the mycelium to give the actual residence time.
results. Both strains were inactivated according to first order
kinetics. In the figure the data are fitted to the function in
eq. (1), with k = 0.045 for strain E 15 and k = 0.065 for strain
3 Results PC 174. With the strain used in the mixing study (PC 180),
60 min starvation gave rise to an inhibition of 73%.
In Fig. 3 the result of a critical oxygen tension study is The first experiment with the two compartment fer-
shown. Above 35% of air saturation changes in the oxygen menter, was performed with an anaerobic volume, VAA, of
concentration had no or little effect on the respiration rate. 900 cm 3, i.e. 6% of the aerobic fermenter volume, V. After
Decrease of the oxygen tension from 35 to 25% reduced the starting the experiment, the respiration rate was immediately
respiration by 50% and this level of respiration was stable at reduced and continued to fall (Fig. 5). With a circulation
oxygen concentration values down to 5%. This pattern was time, tc, of 11 min and the number of circulations, n = t/tc,
repeated in several cultivations with variations only in the equal to 16.7, the inactivation of the respiration was 39%.
s-shape of the curve. The flow rate was then increased to give a circulation time
We have earlier demonstrated [2] the sensitivities towards of 5 min with n = 16.7. The inhibition then increased further
oxygen starvation of one strain of Penicillium chrysogenum, to a total of 43%.
strain E 15. The validity of the kinetics was tested by re- In the second part of the study, the anaerobic volume was
peated experiments with another strain (PC 174), that has 150 cm 3, i.e. 1% of the aerobic volume V, and the circulation
been used for commercial penicillin production. The result of time 1 and 2 min, respectively (n = 100). The outlet oxygen
126 Bioprocess Engineering 3 (1988)

]9 i centrations one would still have a respiration rate above

% zero (e.g. from cell growing on walls in the fermenter). It
could also be due to the fact that the true critical value is very
low, as could be seen by extrapolation of the curve to zero.
o This way one would get two possible critical oxygen concen-
trations. It is p r o p o s e d that this is due to effects of biological
~18
~J
' 3 ,
inhomogeneities of the system caused by the coexistence of
diffused mycelium and pellets or to inhomogenities caused
by insufficient mixing. N o effort is made in this study to
\~ '
distinguish between these two effects. W h e n a microorgan-
ism is subject to mass t r a n s p o r t restrictions, as in the case of
immobilization or pellet formation, the saturation constant
17 1
5'0 5'z 5'4 % (ks) is changed and an a p p a r e n t k s can be defined. This k~. apv
Air flow rote ;n the test fermenter Ft I is a function of the true k~, the mass t r a n s p o r t coefficient and
~8 5o s'2 54 the reaction rate [5, 6].
Air flow rote in the reference fermenter F~ Fractions exposed to differences in oxygen content due to
F i g . 5. Progress of the oxygen signal in the outlet gas ofF,. (reference insufficient mixing behave in the same way. also as a conse-
fermenter) and Ft (test fermenter). Area 1 : before experimental start- quence of transfer resistances. Therefore, results from a criti-
up; Area 2: Ft used as a part of the two compartment fermenter with cal oxygen tension study in heterogenous broths are difficult
a circulation time (to) of 10 rain; Area 3: As area 2, but t c = 5 min
to interpret. Thus, if one has a large signal from the oxygen
electrode it is still possible that large fractions in the fer-
20 menter suffer from substrate limitation, because of the loca-
% tion of the electrode. This has been d e m o n s t r a t e d in pilot
plant experiments with E. coli, where hydrogen evolution
I I I
from anaerobic metabolism could be observed long before
the oxygen electrode signals a p p r o a c h e d zero [7].
In p r o d u c t i o n scale fermenters, one would also have to
I ~ , deal with oxygen starvation in these zones. Extended kinetic
studies (Fig. 4) of the irreversible inhibition of the respiration
rate shows different constants (k in eq. (1)) for different
strains, but will still give first order kinetics. However, it is
I I
I [ difficult in these measurements to determine the exact time
of zero, when no oxygen is present. As described earlier [2],
16 , l the head space and the sparger inlet are at a certain time
go 57 54 h 5f supplied with nitrogen, but the response is not immediately
Time t
seen and the oxygen electrodes might also have a different
Fig. 6. Progress of the oxygen signal in the outlet gas ofF, (reference response between the different experiments.
fermenter) and Ft (test fermenter). Area 1 : before experimental start-
up; Area 2: Ft used as part of the two compartment fermenter with One can see a clear effect of the starvation represented in
a circulation time of i min; Area 3 : As area 2 Fig. 5. The two curves are moved in time to a resemblance
according to the well k n o w n shift in p H and oxygen concen-
tration. The effect is calculated according to eq. (2) with a
concentration from the reference and the test fermenters are flow ratio of 0.6, shown in Fig. 7. The non-linear response is
shown in Fig. 6. N o irreversible effect could be observed due to the residence time distribution of the cells in the
within 200 min, during which two culture volumes (2* V) was system.
transferred through the anaerobic part. In order to investigate the effect of short circulation times
in a small oxygen free area in a fermenter, starvation for 1
and 2 min were imposed on the system. The volume of the
4 Discussion anaerobic zone, VaA, was in this case 1% of the aerated
volume (150 cm3). To attain 1 min circulation time in VAA,
As can be seen from Fig. 3, the M o n o d model cannot be the a m o u n t of cycles through the plug flow zone (n) will
applied without restriction. increase from 16.7 to 100 and thereby give an additional
Even when the dissolved oxygen tension was reduced to effect on the residence time distribution. The results show no
4 % according to the electrode, a considerable respiration irreversible effect of the respiration (Fig. 6).
could be observed. This could be due to the fact that the The results of the expected inhibition, according to eq. (1),
electrode is situated in a zone of the fermenter with low and the t w o - c o m p a r t m e n t reactor results are shown in
oxygen concentration and that in other zones of higher c o n - Table 1. It shows that results derived from eq. (1) correlate
G. Larsson et al.: Effects of limiting oxygen concentrations and short term oxygen starvation on Penicillium chrysogenurn 127

50
%
tc:llmin
~40

.~ 30
~w

g
l
I
I Fig. 7. Reduction of the respiration as a func-
i
tion of circulation time in the anaerobic part of
I
I I the two compartment fermenter as calculated
by eq. 2. Areas 1-3 correspond to Fig. 5
3 h
Time t

Table 1. Comparison of the results of eq. (1) and eq. (2) as the limiting substrate and its effect on the capacity or
respiration.
VAA tc /OUR [eq. (1)] Redrosv [eq. (2)]
The study also shows that a circulation time of 2 min in
6 11 41 39 a small oxygen free area will not do any irreversible damage
6 6* 14 4
1 1 5 0 to the respiratory capacity of Penicillium chrysogenum. This
1 2** 9 0 will add a lag phase to the kinetics described in eq. (1), i.e.
the two first minutes give no effect on lov a. The two-com-
VAA volume of the anaeorbic plug flow zone of the two compart- p a r t m e n t model will be used for further studies, This is due
ment reactor (in percent of the aerated volume)
tc circulation time through VAA (in min) to the suitability of this model to give the exact circulation
* indicates that this time is imposed above the foregoing time in time in the smaller plug flow zone, where the limitation not
the table necessarily has to be oxygen, but can also be other substrate
** two times one minute in VAA followed immediately after each components.
other

with the t w o - c h a m b e r reactor for the longer circulation
times. The shorter times, as earlier only a calculated effect,
were too large in comparison.
Parallels with former results according to eq. (1) must be
done somewhat carefully, since the inhibition effect was
100% for 60 min of oxygen starvation with the earlier used
strain E 15 [2] and for this strain (PC 180) it is 73%. There
seems to be a metabolic difference and the new organism
seems m o r e resistant to the starvation. H o w this works is not
yet fully understood.
5 Conclusions
D e t e r m i n a t i o n of the critical oxygen tension in an inhomo-
genous system is very complex. It is clearly seen in this study
that the a p p a r e n t critical tension is not a constant for an
organism, but depends on the physical and biological situa-
tion in the fermenter. It will not be possible to get a single
value from a cultivation, since it suffers from different condi-
tions in different fractions. W i t h insufficient mixing condi-
tions, the possible way to discover oxygen limitation with
the oxygen electrode will be very limited.
A two c o m p a r t m e n t fermenter was used in order to simu-
late the inhomogeneous situation within a fermenter with
respect to insufficient mass transfer.
This model gives a possibility to study the short <'ir~t~l:l-
tion times in restricted zones, for which we haxe used ox3gcn
Acknowledgements
This work was sponsored by the National Swedish Board for
Technical Development. We also want to thank Dr. ~ke Und~n,
Fermenta AB, for his interest and support.
References
1. Larsson, G. ; Enfors, S.-O. : Influence of oxygen starvation on the
respiratory capacity of Penicillium chrysogenum Appl. Micro-
biol. Biotechnol. 21 (1985) 228-233
2. Oosterhuis, N. M. G.; Kossen, N. W. F.: Dissolved oxygen con-
centration profiles in a production scale bioreactor. Biotechnol.
Bioeng. 36 (1984) 546-550
3. Vardar, F. ; Lilly, M. D.: Effect of cycling oxygen concentrations
on product formation in penicillin fermentations. Eur. J. Appl.
Microbiol. Biotechnol. 14 (1982)203-211
4. Sweere et al.: Proc. 4th European Congress on biotechnology 1
(1987) 180-183
5. Wang, D., et al.: Fermentation and enzyme technology, pp.
324-326, New York: Wiley&Sons Inc. 1979
6. Kobayashi, T.; Moo-Young, M. : Backmixing and mass transfer
in the design of immobilized-enzyme reactors. Biotechnol.
Bioeng. 13 (1971) 893-910
7. Cleland, N.; Enfors, S.-O.: A biological system for studic~ on
mixing in bioreactors. Bioprocess Engineering 2 (1987) 115-120
Received September 29, 1987
G. Larsson and S.-O. Enfors
Dept. of Biochemistry and Bioteehnology
Royal Institute of Technology
S-t00 44 Stockholm, Sweden