Appl Microbiol Biotechnol (1991) 34:463-468
0175759891000078
Production of 2,3-butanediol in a membrane bioreactor
with cell recycle
A.-P. Zeng, H. Biebl, and W.-D. Deckwer
GBF - National Research Institute for Biotechnology, Biochemical Engineering Division, Mascheroder Weg 1, D-3300 Braunschweig,
Federal Republic of Germany
Received 23 July 1990/Accepted 14 September 1990
Summary. The production of 2,3-butanediol by Entero-
bacter aerogenes DSM 30053 was studied in a cell recy-
cle system with a microfiltration module. Emphasis was
put on the influence of oxygen supply, cell residence
time, dilution rate, and pH. Under optimal conditions a
productivity as high as 14.6 g butanediol + acetoin/1 per
hour was achieved with a product concentration o f
54 g/1 and a product yield of 88%. This productivity is
three times higher than that of an ordinary continuous
culture. The achievable final product concentration of a
cell recycle system was limited by the accumulation o f
the inhibiting by-product acetic acid, which increased
very rapidly at low dilution rate. To maximize product
concentration a fed-batch fermentation was carried out
with stepwise p H adaption at high cell density. A final
product concentration of 110 g/l was obtained with a
productivity of 5.4 g/1 per hour and a yield of 97%.
Introduction
2,3-Butanediol (BD) is an appealing product because of
its diverse potential use (Ledingham and Neish 1954;
Emerson et al. 1982; Magee and Kosaric 1987). The
economic feasibility of the microbial production of BD
depends on the achievable final product concentration
and the productivity. In conventional batch and contin-
uous cultures BD concentrations of about 40 g/1 could
be obtained with a productivity between 2 and 5 g/l per
hour (Sablayrolles and G o m a 1984; Zeng et al. 1990a).
In fed-batch fermentation a concentration of BD + ace-
toin as high as 113g/1 has been reported (Yu and
Saddler 1983). However, the productivity was only
about 0.7 g/l per hour. In a two-stage continuous cul-
ture 77 g/l BD ( + 4 . 4 g/1 acetoin) was reached with a
productivity o f 2.3 g/1 per hour (Tran-Dinh et al. 1987).
In general, the low productivity of conventional anae-
robic and microaerobic fermentations is mainly due to
Offprint requests to: W.-D. Deckwer
App//ed
Microbiology
Biotechnology
© Springer-Verlag 1991
the relatively low cell concentration in the systems. In
order to increase the cell concentration various tech-
niques have been proposed. Among those, cell recycle
with hollow fibre membranes has been successfully
used in the production of ethanol (Lee and Chang
1987; Tegtmeier 1989) and organic acids (Ohleyer et al.
1985; Park et al. 1989).
In a previous study on the continuous culture of En-
terobacter aerogenes in the usual way it has been shown
that BD production is not associated with growth under
microaerobic conditions (Zeng et al. 1990a). To achieve
high volumetric productivities it may therefore be bet-
ter to make use of cell recycle and cell immobilization
systems, respectively. Recently, BD production was
studied with cell recycle in a membrane reactor (Rama-
chandran and G o m a 1988; Qureshi and Cheryan 1989).
Ramachandran and Goma (1988) worked with a total
recycle of biomass and increased the productivity to
9.8 g/1 per hour. However, no steady-state operation
was reached in their study. In the work of Qureshi and
Cheryan (1989) unexpectedly poor performance was
observed with a cell recycle operation. The reason for
this is not clear.
In this work BD production by E. aerogenes was
studied in a membrane reactor with partial recycle of
biomass. Emphasis was put upon the factors affecting
the final product concentration and productivity as well
as stability of the system. Experiments were also con-
ducted in fed-batch mode with high cell density.
Materials and methods
Or~Tanism and culture methods. The strain E. aerogenes DSM
30053 was used in this study. Culture medium, culture conditions
and analytical methods were reported by Zeng et al. (1990a). The
measurements were carried out in glucose-sufficient culture under
oxygen-limited conditions.
The experimental set-up of the membrane cell recycle system
is shown in Fig. 1. The fermentations were carried out in a 4.5-1
Setric bioreactor (type SET00 4V, Setric Genie Industriel, Tou-
louse, France) with a working volume of 1.5 to 3.5 1. The bio-
reactor was equipped with pH, Po2, temperature, antifoam and
464

~2
Antifoam ~
si~2~ "~-Exhaust gas

Agitation
I~1 a~s dryin9
p~ Sterite I~] bottle
• ~i.~r #'5
Te~oerature
• • - ~ ~ ~ C ~ t e n s a t e
,~ ~ B i ~ s s breed

3: - "2i

i,o Thero,t.t
-i . . . . . . . . . . .

i ,
I~
Reactor

,
S
~ ~. .; .; .~.[. i. ". .~. i. ~. . . .f.t. ~. .t.e. .r /
fitter
~

°
c~ntrolter
:-
:.

Mag~t ic ~tirrer
. Air 02 N2
~ ..........................................................................

Fig. 1. Experimental set-up of the membrane cell recycle system; AF, antifoam; MF, microfiltration; P, pressure gauge

agitation speed controls which were connected to a process com-

puter (RTOS/PEARL, Institut ftir Regelungstechnik, University
of Hannover, FRG). A steam-sterilizable microfiltration module 8~ ._.o
(type MD 020 CP 2N, ENKA, Wuppertal, FRG) with 50 micro-
porous polypropylene capillaries and 0.1 m 2 filtration area, XO2
1.8 mm in diameter and 0.2 lxm in pore diameter, was attached to
the bioreactor for cell recycling. Cell broth was recirculated by a
gear-pump (type V288, Verder, Dtisseldorf, FRG) with a velocity
of about 0.7 m/s. The filtration module was back-flushed with
permeate daily. Peristaltic pumps were used for feed, permeate o~ o o u,~..

flows and for the cell bleed. The flow rates of feed, product and
cell bleed were controlled with a Philips weight control unit (type
PR 1592, Philips, Hamburg, FRG). Dilution rates were calculated ~o ~o 1~o 1~o
on the basis of the total volume of the system including the work- Time ( h )
ing volume of the reactor, the circulation loop and the cell side of
the filtration module. Fig. 2. Control of oxygen uptake rate (OUR) with a real-time op-
erating computer system at high cell density (set values of OUR:
Control of oxygen uptake rate (OUR) with a real-time computer sys- 150, 75 and 100 mmol O2/1 per hour)
tem. In a previous paper (Zeng et al. 1990a) it has been shown
that the O U R of an oxygen-limited culture depends not only on
one can preset the O U R over a wide range. It is also possible to
the aeration and agitation of the bioreactor but also strongly on
control the respiration rate or the specific O U R of the culture.
the product composition of the fermentation broth. The latter var-
ies with the fermentation conditions and is especially sensitive to
the oxygen supply of the culture. In order to obtain a definite O2
supply of the culture, an O U R control system was developed with R e s u l t s and discussion
a real-time operating computer (RTOS/PEARL). For this pur-
pose, the actual O U R was calculated from the 02 and CO2 con-
centrations measured in the effluent gas and compared with the Performance o f the cell recycle system
set value. Aeration rate a n d / o r impeller speed were then regul-
ated by the computer to achieve the desired OUR. With the OUR control system the effect of OUR on
Figure 2 shows the result of an O U R control at different p e r f o r m a n c e o f t h e cell r e c y c l e s y s t e m w a s s t u d i e d . F i g -
O U R set values. A reliable control of O U R over a long fermenta- u r e 3 s h o w s t y p i c a l r e s u l t s at a d i l u t i o n r a t e ( D ) o f
tion time was achieved with this real-time operating computer sys- 0.23 h - I a n d a m e a n c e l l r e s i d e n c e t i m e o f 57 h. As in
tem. When changing the set value of O U R or in the case of distur-
bance the O U R became quickly stable. With this control system t h e c a s e o f a c o n t i n u o u s c u l t u r e w i t h o u t cell r e c y c l e a n
 465
o
BD+Acoto in ~ ~ ~ Gluaone utilizotion
0 Biomoss
[3 BD+Acetotn

NO ~

~/~ Acetoin ~ g
o ~=o.~
8= o~-
~
Q , ~.~-,--~.~, ~=~- ~
55 85 115 145 •
OUR(~ol/L, h) ~- ~
2b ~'&'~'
Fig. 3. Effect of OUR on cell and product (2,3-butanediol, BD) Cell residence time (h)
concentration in a cell recycle system (dilution rate, D = 0.23 h-~,
cell residence time = 57 h) Fig. 4. Effect of cell residence time on glucose utilization, bio-
mass and product concentration (OUR= 100 mmol/1 per hour,
D=0.23 h - ' )
optimal O U R for the production of BD was found at
each cell residence time (Zeng et al. 1990a), amounting
to about 100 mmol/1 per hour under the conditions ~- D Spec. productivity
given in Fig. 3. At the optimal O U R a product concen- - ASpec. glucose u t i l i z a t i o n
O Spec. OUR
tration of 54.4 g/1 was obtained with a productivity of ~- i
12.5 g / l per hour, a glucose consumption of 131 g / l and ~ -
a product yield of 0.42g p r o d u c t / g glucose. The ~ * - ~ -~q ~'~-
N
achieved productivity is higher than any literature ~ -
•~ ,..~ :7,
-~ ~ ,,.._
value reported so far. The biomass concentration in- ~ ~-
creased continuously with O U R and reached 56.6 g/1 at g - ~ ~~ .
•

an O U R of 100 mmol/1 per hour. This is about sixfold ~ - ~~ .

~ ~ " ° ~ o
higher than that of a conventional continuous culture
without cell recycle. However, the increase in the volu- ~ ~ ~,

metric productivity was only about two- to threefold, g'g'~'~o

Cell r~i~n~ ~i~ (~)
indicating that the specific productivity o f cells is re-
duced in the cell recycle system. Fig. 5. Effect of cell residence time on specific rates of product
In spite of the relative low specific O U R (between formation, glucose and oxygen utilization (OUR= 100 mmol/l per
1.7 to 2.4 m m o l / g per hour) the percentage of acetoin hour, D = 0.23 h- 1)
in the product was rather high. In contrast, the ethanol
concentration remained low under all conditions. This
indicates that the producing cells had a higher actual Table 1. Fermentation results at different dilution rates with cell
specific O U R than that calculated for the total dry bio- recycle
mass. The reason m a y be that the viability a n d / o r activ-
D Glucose Biomass BD + Product- Yield
ity of cells were reduced due to the high product con- (h-') (g/l) (g/l) Acetoin ivity (g/g)
centration and the long cell residence time. Fond et al. (g/l) (g/~ h)
(1985) reported that at a BD concentration of 65 g / l
and a concentration of undissociated acetic acid of 0.31 119.2 63.2 48.48 15.03 0.41
0.3 g/1 only 35% of the total cells are viable. 0.27 124.6 62.2 54.23 14.64 0.44
In order to examine the effect of cell residence time 0.25 126.1 56.7 52.01 13.03 0.43
0.23 130.9 56.6 54.39 12.51 0.42
fermentations were carried out at different biomass 0.13a -- -- 74.50 9.69 --
bleed rates. The results are shown in Fig. 4. As expected
biomass, product concentration, and glucose utilization Oxygen uptake rate (OUR)= 100 mmol/1 per hour; cell residence
increased with the cell residence time. However, as time = 57 h; D, dilution rate; BD, 2,3-butanediol
shown in Fig. 5, the specific rates of product formation, a NO steady-state data
oxygen and glucose utilization decreased steadily, indi-
cating a loss of cell viability a n d / o r activity. An experi-
ment with total cell recycle did not lead to stable oper- summarized in Table 1. Between D = 0 . 3 1 to 0.23 h -1
ation: cell lysis was serious when the cell residence time the biomass concentration decreased slightly. The
exceeded 70 h. For stable operation a controlled bleed B D + a c e t o i n concentration remained at about 50 g/l.
of biomass is necessary. However, as the total biomass The highest productivity was obtained at D = 0 . 3 1 h - l ,
concentration and the productivity increased with the amounting to 15 g/1 per hour. At the lowest D exam-
cell residence time, the bleed of biomass should be kept ined (0.13h -1) the B D + a c e t o i n concentration in-
at a m i n i m u m rate. creased at first, but started to fall when it reached
To increase the final product concentration D was 74.5 g/l. The changes in OUR, RQ and Po2 of the cul-
varied in a further fermentation run. The results are ture (Fig. 6) indicate that respiration was strongly inhi-
466
0
C~
Fed-batch fermentation with pH adaption at high cell
density
Seo )
Continuous cultivations with addition of acetic acid
(Zeng et al. 1990b) showed that the strength of the ace-
c-- x Xx X 902....~ .... tic acid inhibition on the growth of E. aerogenes de-
pends exclusively on the concentration of its undisso-
~0 ~ ((}-
ciated form, [HAc]. This finding allows the application

"t"
~

+
~

~ ~~ 0~-
,~-_
of a strategy for maximizing the product concentration.
In normal batch or fed-batch culture, growth and meta-
bolism cease completely once the pH-specific critical
N ~ acetic acid concentration is reached (Zeng et al. 1990b).
However, if the pH is increased instantly, the inhibition
' ~'0 ' 10 ' ~ ' ~'0 ' ~'0 ' ~0 ' g0 ' 9b '
Time ( h) is expected to be suspended temporarily, as [HAc]
strongly depends on the pH. Hence, growth and meta-
Fig. 6. BD fermentation in a cell recycle system: effect of the di- bolism can continue until the new pH-dependent equil-
lution rate ( O U R = I 0 0 mmol/1 per hour, cell residence ibrium is established. Indeed, this could be shown in a
time = 57 h - 1) fed-batch fermentation at high cell density in which the
pH was periodically raised.
Figure 7 shows the results of such a fed-batch fer-
bited. The effect results from the accumulation of acetic mentation. Before the culture was switched to fed-batch
acid in the culture (Zeng et al. 1990b). As shown in Fig. operation, it had been running as a continuous culture
6, the acetic acid concentration increased rapidly from for about 100 h (Fig. 6). The pH of the culture was in-
0.7 g/1 to a critical concentration of 2.7 g/l (corre- creased stepwise from 5.5 to 6.3 during the fed-batch
sponding to 0.42 g undissociated acetic acid/l; [HAc]) operation. Correspondingly the OUR was reduced
when D was lowered from 0.31 h -1 to 0.13 h -1. from 100 to 30 mmol/1 per hour. Glucose was added to
Ramachandran and Goma (1988) found in their to- the reactor in a concentration of 354 g/l. Within 17 h
tal recycle of biomass that the production of BD started 568 g glucose was consumed, and 274 g BD+acetoin
to decrease when the cell concentration in the bioreac- was produced. The product yield was 97%. A BD + ace-
tor built up to 40 g/1. The authors postulated that at toin concentration as high as 110 g/l was obtained with
high cell concentration the organism may switch from an average productivity of 5.4 g/l per hour. Acetic acid
BD to polysaccharide production due to severe 02 lim-
itation. Qureshi and Cheyan (1989) also claimed that
the unexpectedly poor performance of their cell recycle
operation at high biomass concentration is due to nu- ~,~<~ .~/
trient or mass transfer limitations. Our results suggest
_~ z ~--~j:o
that the unstable operation is a consequence of the ac-
cumulation of acetic acid at high biomass concentra-
tion and/or low dilution rate.
~ _ ~ 0
In a recent study in ordinary continuous culture it ~ ~ ~ ~J B~Ace,oln ,
has been shown that the strength of acetic acid inhibi- ~ ~alo=os. /~--
tion depends only on [HAc], irrespective of pH (Zeng et ,--,.
al. 1990b). In order to assess the effect of pH on the = _~ ~ o~ / O~m--m
~ ~ g~ m~O-e
performance of a cell recycle system experiments were '°~ T X ~='=~
carried out at elevated pH values. Table 2 shows the d~_ ~~
~ ~
~ ~
~ .e~
results. The biomass concentration increased slightly ~ ~ ~ ~, ~, ~'~'~,{;(<~'
with increase in pH. As in ordinary continuous culture
Time ( h )
the acetic acid concentration increased rapidly (Zeng et
al. 1990b). Glucose utilization, product concentration Fig.% Eed-batch fermentation at high cell density with pH adap-
and product yield decreased clearly with the pH. tion

Table 2. Results of fermentation with cell recycle at different pH values

pH Glucose Biomass BD + Acetoin Productivity Yield Acetate

(g/l) (g/l) (g/l) (g/1 h) (g/g) (g/l)

5.5 133.9 51.3 52.45 12.06 0.40 0.94

5.7 127.3 57.5 50.54 11.60 0.40 6.37
6.0 111.1 61.0 40.89 9.40 0.37 11.86
6.4 116.1 63.9 29.17 6.71 0.25 15.88

O U R = 100 mmol/l per hour; D=0.23 h - l ; cell residence t i m e = 4 0 h


Table 3. Comparison of different cultivation systems for BD production
Dilution rate (h-~) Continuous culture
With recycle No recycle
0.27
Glucose utilization (g/l) 124.6 111.5
BD + Acetoin (g/l) 54.0
Biomass (g/l) 62.2
Yield (g/g) 0.44
Productivity (g/l h) 14.6
Specific productivity (g/g h) 0.24
a Mean value
(dissociated and undissociated) accumulated up to 8 g /
1. The achieved final BD + acetoin concentration equals
the highest B D + a c e t o i n concentration ( l 1 3 g / 1 ) ob-
tained by Yu and Saddler (1983) in a double fed-batch
fermentation.
The experiment illustrates that the final product
concentration obtained in a continuous cell recycle sys-
tem is lower than that achievable in a fed-batch cultiva-
tion. It also confirmed that limitation of the final prod-
uct concentration in a cell recycle system cannot be at-
tributed to the accumulation of BD and acetoin but to
that of acetic acid.
Comparison of different cultivation systems
Table 3 shows a comparison of different cultivation
systems. The highest glucose utilization and highest
product concentration can be obtained in fed-batch fer-
mentation. However, the productivity of fed-batch fer-
mentation is rather low c o m p a r e d to continuous cul-
ture. In ordinary continuous culture the productivity is
three times as high as that of a simple fed-batch system.
Continuous fermentation with cell recycle allows excel-
lent biomass concentration, high productivity and good
glucose utilization. Product concentration and product
yield are fairly high for the cell recycle system. Nev-
ertheless, the achievable final product concentrations is
limited by the accumulation of acetic acid, as revealed
by the low specific productivity.
As BD is not easily recovered from the broth, a high
final product concentration is desirable for industrial
production. In this respect fermentation with p H adap-
tion appears to be most favourable. It can be done in
batch, fed-batch a n d / o r continuous cascade culture. As
shown above, the incorporation of a m e m b r a n e filtra-
tion unit is advantageous for increasing the productivi-
ty. An alternative solution m a y be repeated batch or re-
peated fed-batch fermentation with a m e m b r a n e - t y p e
bioreactor. In some fermentation processes repeated
batch or fed-batch operations have been found to be
superior to continuous operation. Recently, Watanabe
et al. (1990) used such a system to produce ethanol and
obtained an ethanol concentration of 140 g/1 for each
batch with a relatively high productivity of 3.5 g/1 per
hour.
467
Fed-batch
Simple pH Adaption
0.1 0.28 -- --
62.0 204.3 223.0
43.0 20.0 85.3 110.0
6.9 9.1 13.5 30.0
0.39 0.32 0.43 0.49
4.3 5.6 1.8 5.4
0.62 0.60 0.14a 0.18a
Another technique to improve BD production could
be on-line removal of products, especially acetic acid.
As shown in Table 3, the capability of this strain to pro-
duce BD is not fully utilized in cultures with cell recy-
cle a n d / o r in fed-batch fermentation. It might be possi-
ble to increase the specific productivity at high cell con-
centration by means of on-line removal of the inhibit-
ing products, so that the productivity could be further
increased. Acetic acid can be removed, for example, by
means of electrodialysis, which has been successfully
applied in the fermentation of lactic acid (Hongo et al.
1986) and acetic acid ( N o m u r a et al. 1988).
References
Emerson RR, Flickinger MC, Tsao GT (1982) Kinetics of dehy-
dration of aqueous 2,3-butanediol to methylethyl ketone. Ind
Eng Chem Prod Res Dev 21:473-477
Fond O, Jansen NB, Tsao GT (1985) A model of acetic acid and
2,3-butanediol inhibition of the growth and metabolism of
Klebsiella oxytoca. Biotechnol Lett 7:727-732
Hongo M, Nomura Y, Iwahara M (1986) Novel method of lactic
acid production by electrodialysis fermentation. Appl Environ
Microbiol 52:314-319
Ledingham GA, Neish AC (1954) Fermentative production of 2,3-
butanediol. In: Underkofler LA (ed) Industrial fermentations,
vol 2. Chemical Publications, New York, pp 27-93
Lee CW, Chang HN (1987) Kinetics of ethanol fermentation in
membrane cell recycle fermentors. Biotechnol Bioeng
29:1105-1112
Magee RJ, Kosaric N (1987) The microbial production of 2,3-bu-
tanediol. Adv Appl Microbiol 32:89-161
Nomura Y, Iwahara M, Hongo M (1988) Acetic acid production
by electrodialysis fermentation method with a computerized
control system. Appl Environ Microbiol 54:137-142
Ohleyer E, Blanch HW, Wilke CR (1985) Continuous production
of lactic acid in a cell recycle reactor. Appl Biochem Biotech-
nol 11:317-332
Park YS, Ohtake H, Fukaya M, Okumura H, Kawamura Y, Toda
K (1989) Acetic acid production using a fermentor equipped
with hollow fiber filter module. Biotechnol Bioeng 33:918-
923
Qureshi N, Cheryan M (1989) Production of 2,3-butanediol in a
membrane recycle bioreactor. Process Biochem. 1989:172-
175
Ramachandran KB, Goma G (1988) 2,3-Butanediol production
from glucose by Klebsiellapneumoniae in a cell recycle system.
J Biotechnol 9:39-46
468
Sablayrolles JM, Goma G (1984) Butanediol production by Aero-
bacter aerogenes NRRL B199: effect of initial substrate con-
centration and aeration agitation. Biotechnol Bioeng 26:148-
155
Tegtmeier U (1989) Einsatz von Mikrofiltrationsmembranen zur
Zellrtickhaltung in einem kontinuierlichen Fermentationspro-
zess. Chem Ing Techn 61:570-571
Tran-Dinh K, Chin CW, Hill F (1987) Zweistufiges kontinuier-
liches Verfahren zur Production von 2,3-Butandiol mit Kleb-
siella pneumoniae. Dechema-Jahrestagung der Biotechnolo-
gen, Frankfurt/Main
Watanabe T, Aoki T, Honda H, Taya M, Kobayashi T (1990) Pro-
duction of ethanol in repeated-batch fermentation with mem-
brane-type bioreactor. J Ferment Bioeng 69:33-38
Yu EKC, Saddler JN (1983) Fed-batch approach to production of
2,3-butanediol by Klebsiella pneumoniae grown on high sub-
strate concentrations. Appl Environ Microbiol 44:777-784
Zeng A-P, Biebl H, Deckwer W-D (1990a) 2,3-Butanediol produc-
tion in continuous culture of Enterobacter aerogenes: role of
oxygen supply. Appl Microbiol Biotechnol 33:264-268
Zeng A-P, Biebl H, Deckwer W-D (1990b) Effect of pH and acetic
acid on growth and 2,3-butanediol production of Enterobacter
aerogenes in microaerobic culture. Appl Microbiol Biotechnol
33:485-489