Bioresource Technology 136 (2013) 102–108

Contents lists available at SciVerse ScienceDirect

Bioresource Technology
journal homepage: www.elsevier.com/locate/biortech

Investigation of gas stripping and pervaporation for improved feasibility

of two-stage butanol production process
Mpho Setlhaku a, Sebastian Heitmann b, Andrzej Górak b, Rolf Wichmann a,⇑
a
TU Dortmund University, Laboratory of Biochemical Engineering, Emil-Figge-Str. 66, 44227 Dortmund, Germany
b
TU Dortmund University, Laboratory of Fluid Separations, Emil-Figge-Str. 70, 44227 Dortmund, Germany

h i g h l i g h t s

" Successful integration of gas stripping to two-stage process for butanol production.
" 160 g/L butanol produced with pervaporation compared to 59 g/L from gas stripping.
" Introduction of a ‘‘window of operation’’ for evaluating ABE processes feasibility.
" Processes with both in situ product removal and immobilization are more feasible.

a r t i c l e i n f o a b s t r a c t

Article history: Gas stripping and pervaporation are investigated for butanol recovery in a two-stage acetone–butanol–
Received 21 November 2012 ethanol (ABE) fermentation process. The first stage is operated in a continuous mode and the second stage
Received in revised form 13 February 2013 as a fed-batch. Gas stripping coupled to the second stage and operated intermittently enabled additional
Accepted 18 February 2013
glucose feeding in the second stage and up to 59 g/L butanol and 73 g/L total ABE solvents in the conden-
Available online 24 February 2013
sate. Concentration of 167 g/L butanol and 269 g/L ABE in the permeate was measured in ex situ pervap-
oration experiments using a PDMS membrane at temperature of 37 °C and pressure of 10 mbars. The
Keywords:
‘‘operating window’’ tool is introduced to evaluate the feasibility of the existing ABE fermentations oper-
Butanol
Clostridium acetobutylicum
ated as continuous with cell recycle, as two-stages, with biomass immobilization or with integrated prod-
Two-stage fermentation uct removal. This tool enables the identification of the most favorable process configuration, which is the
Gas-stripping combination of cell immobilization and integrated product removal.
Pervaporation Ó 2013 Elsevier Ltd. All rights reserved.
Window of operation

1. Introduction ate butanol concentrations up to 23 g/L (Qureshi and Blaschek,

Solvent toxicity, mainly n-butanol, is one of the main chal-
lenges in rendering the acetone–butanol–ethanol (ABE) fermenta-
tion economically feasible. Butanol toxicity leads to microbial
growth inhibition, which results in low final solvent concentra-
tion, productivity and yield. Distillation was traditionally used
as the downstream method for the recovery and separation of
the ABE from the broth and resulted in high-energy costs (Jones
and Woods, 1986; Mariano and Filho, 2012). The options available
to remedy toxicity and low butanol concentrations are bacterial
mutagenesis, metabolic engineering and fermenter integrated re-
moval of ABE. Since clostridia wild type strains can tolerate about
13 g/L butanol and 20 g/L ABE solvents (Jones and Woods, 1986),
engineered strains have been developed that can reach and toler-
⇑ Corresponding author. Tel.: +49 231 755 3205; fax: +49 231 755 5110.
E-mail address: rolf.wichmann@bci.tu-dortmund.de (R. Wichmann).
1999a) and 20 g/L (Xue et al., 2012). In situ or ex situ operated
ABE removal methods coupled to batch, fed-batch or continuous
operated reactors have been extensively investigated. These
methods not only alleviate the toxicity challenge, they also result
in increased volumetric productivities and yields and could re-
duce separation energy significantly. The most common methods
in ABE fermentations are gas stripping (Ezeji et al., 2003; Qureshi
and Maddox, 1991), pervaporation (Gapes et al., 1996; Qureshi
et al., 2001; van Hecke et al., 2012), adsorption (Qureshi et al.,
2005) and liquid–liquid extraction (Shukla et al., 1989). Butanol
recovery energy requirement was predicted to be in the range
of 3–8 and 2–145 MJ/kg-butanol for adsorption and pervapora-
tion, respectively, as compared to gas stripping at 14–31 MJ/kg-
butanol (Qureshi et al., 2005). While adsorption suffers from foul-
ing, gas stripping is relatively easy to perform as compared to
pervaporation and it is also possible to use the CO2 and H2 pro-
duced during fermentation as stripping agents (Ezeji et al.,
2007a; Xue et al., 2012). Compared to gas stripping, pervapora-

0960-8524/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.biortech.2013.02.046
 M. Setlhaku et al. / Bioresource Technology 136 (2013) 102–108
tion has a higher selectivity for butanol and it is not dependent
on vapor–liquid equilibbria. Poly[1-(trimethylsilyl)-1-propyne]
(PTMSP) and zeolith membranes are reported to be fouled by fer-
mentation broth (Fadeev et al., 2000; Weyd et al., 2012) and in a
larger content compared to PDMS membranes (Qureshi and Bla-
schek, 1999b). Qureshi et al. (2001) reported that silicalite–sili-
cone composite membranes connected to an ABE fermentation
do not get fouled.
One of the most promising process configurations for ABE fer-
mentation is the combination of two fermenter stages coupled
with in situ product removal. It allows for significant improvement
of final butanol concentration, volumetric productivity, yield and
operation time (Ennis et al., 1987; Gapes et al., 1996; Richter
et al., 2012; van Hecke et al., 2012). Setlhaku et al. (2012) devel-
oped a two-stage reactor concept where the first stage is operated
continuously and is connected to the second stage operated as a re-
peated fed-batch. A total of 19.5 g/L ABE and 12 g/L butanol was
produced at productivities of 1.47 gABE/L/h and 0.92 gButanol/L/h,
without including any process improvements like in situ product
removal, biomass recycle or cell immobilization. Cell immobiliza-
tion and recycle have been shown to increase butanol productivi-
ties to over 5 g/h/L (Baba et al., 2012; Qureshi et al., 2000;
Tashiro et al., 2005).
In this work, the second stage fed-batch fermenter of our two-
stage process (Setlhaku et al., 2012) is integrated with gas strip-
ping. The fermentation broth from the second stage is also fed to
a pervaporation unit for butanol removal. Testing both gas strip-
ping and pervaporation, provides further information in evaluating
the productivity of ABE fermentations. A very useful bioprocess
comparison tool called the ‘‘operation window’’ (Woodley and
Titchner-Hooker, 1996) is used in this work to compare and assess
the viability of the reported ABE fermentations, which are an
improvement of the conventional batch process. The window of
operation tool has also seen application in fermentations for the
production and isolation of protein alcohol dehydrogenase (Zhou
and Titchener-Hooker, 1999) and to investigate the possible bottle-
necks and key parameters hindering the scale-up of biofilm pro-
cesses (Halan et al., 2012). The four different fermentation modes
compared in this work using the operation window are those oper-
ated as: one stage continuous reactor with cell recycle; two-stage
reactors; processes with cell immobilization and reactor/s with
integrated product removal.
2. Methods
2.1. Strain and medium
Clostridium acetobutylicum ATCC 824 strain was used in all the
fermentations. The method of cell cultivation from precultures to
reactor has been previously described (Setlhaku et al., 2012). The
clostridia growth medium used for all the precultures and main
fermentation preparation comprised the following components
per liter of double distilled water: yeast extract 5 g; KH2PO4
0.75 g; K2HPO4 0.75 g; NaCl 1 g; MnSO4H2O 0.01 g; (NH4)2SO4
2 g; MnSO4H2O 0.2 g; FeSO47H2O 0.01 g; L-asparagine 0.5 g and
glucose 60 g. The yeast extract and salts were autoclaved together
at 120 °C for 20 min; glucose was prepared and autoclaved sepa-
rately and mixed into the final medium according to the concentra-
tions above. The FeSO47H2O and asparagine were sterile filtered.
The final medium had a pH of 6.4 to 6.5.
2.2. Two-stage fermentation process
The two-stage fermentations were conducted in a customized
1L reactor (operated continuously) connected to 4L and 5L reac-
103
tors, alternated for repeated fed-batch. The first stage was first
operated as batch for 16–18 h and then switched to continuous
operation (chemostat). When steady state was reached, achieved
after 4 up to 6 volume changes, it was then connected with the
first second stage reactor. The fermentation broth from the first
stage, containing remaining glucose, cells, formed acids and sol-
vents was pumped to an empty second stage using a peristaltic
pump (Reglo Digital 2 channels, Ismatec SA, Glattbrugg, Switzer-
land). The cells were kept in suspension using a magnetic stirrer.
pH was controlled only in the first stage and maintained at 4.5 by
adding 1 M HCl or 2 M NaOH. The stream from the first stage was
switched to the other second stage fermenter when butanol con-
centration reached a maximum and all the glucose was
consumed.
2.3. Two-stage fermentation coupled with gas stripping
A condenser (62  600 mm, Büchi, Switzerland) fitted to a con-
densate collector was coupled to the second stage reactor similar
to Richter et al. (2012), with the exception that there is no effluent
stream from second stage. Both reactors were maintained at 35 °C
and ABE gases were condensed and collected at 2 °C using 50:50
vol.% mixture of ethyl glycol and water. Gas stripping was tested at
a gas (nitrogen) circulation rate of 4.8–6.6 L/min first with ABE
synthetic mixture comprising 8 g/L butanol, 4 g/L acetone and
1.5 g/L ethanol. The determined butanol removal rate was 0.02 g/
L/h. During the fed-batch, gas stripping was performed intermit-
tently when the butanol concentration in the reactor was above
8 g/L. Condensate samples were collected periodically and ana-
lyzed for ABE concentration. When the butanol concentration in
the reactor decreased below 5.5 g/L gas stripping was stopped
and the fed-batch was continued to allow butanol concentration
to increase. Up to three repeats (fed-batch and gas stripping) were
performed with one of the second-stage reactors before reaching
the maximum liquid level in the reactor and alternating to the
other second stage reactor.
2.4. Second stage fermentation broth tested in a pervaporation unit
Fermentation broth was taken from the second stage (oper-
ated as fed-batch), centrifuged to separate the cells and poured
into the pervaporation lab-scale plant. Pervaporation was oper-
ated as stand-alone process and was not connected to ferment-
ers. A hydrophobic poly(dimethylsiloxane) (PDMS) membrane
was provided by Sulzer Chemtech AG (Pervap 4060) with an
effective diameter of 115 mm. Experiments with binary buta-
nol–water mixtures were performed at butanol concentrations
lower than 30 g/L. Four experiments were performed with origi-
nal fermentation broth and also with broth spiked with acetone,
butanol, and ethanol to extend the concentration range. All per-
vaporation experiments were carried out with one membrane.
Pervaporation using the binary mixture of butanol and water
was performed before and after experiments with fermentation
broth to check if organic acids or other fermentation broth com-
ponents have changed the permeation behavior of the mem-
brane. However, for a later technical application, long term
fouling studies should be carried out to determine the fouling
behavior of the PDMS membranes. All experiments were carried
out at a temperature of 37 °C, corresponding to the fermentation
temperature and a permeate pressure of 10 mbar. Partial fluxes
of the different components were calculated according to Eq.
(1) based on the permeate mass mPerm, permeate mass fractions
wi,Perm, membrane Area AM and the duration of one experiment
texp.
J i ¼ wi;Perm  mPermeate =ðAM  t exp Þ ð1Þ
104 M. Setlhaku et al. / Bioresource Technology 136 (2013) 102–108
2.5. Analytical methods
The concentration of glucose, acetic acid, butyric acid, acetone,
ethanol and butanol in the fermentation broth and gas stripping
condensate were analyzed with high performance liquid chroma-
tography (HPLC) (Knauer GmbH, Berlin, Germany) equipped with
a refractive index detector (Setlhaku et al., 2012). Samples taken
during pervaporation were analyzed by gas chromatography (the
detailed procedure is described elsewhere) (Heitmann et al., 2012).
3. Results and discussion
3.1. Two-stage fermentation with gas stripping
The concentration of acids and butanol in the first fermenter
amounted to 3.5 and 7.5 g/L, respectively (Fig. 1A). Butanol was
kept below toxic concentration. Similar to the technique used by
Richter et al. (2012) in order to overcome clostridia degeneration,
the continuous culture was periodically re-inoculated with heat
shocked spore solutions. This was done at 200, 300 and 380 h, dur-
ing the entire fermentation of 460 h, resulting in an immediate in-
crease of the ABE production, as can be seen in Fig. 1A.
The first stage fermenter was connected to the second stage fer-
menter when steady state was reached at 144 h. The first fed-batch
(FB1, Fig. 1B) was run for 24 h until butanol and ABE concentra-
tions of 12 g/L and 20.8 g/L and productivities of 0.51 g/L/h butanol
and 0.87 g/L/h ABE were reached. Glucose was completely utilized
and 1.8 g/L total acids remained in the reactor (Table 1). An alter-
native and empty 5L reactor, FB2, was connected to the outlet of
first stage and operated as a fed-batch during the period between
177 and 252 h and gas stripping was intermittently operated three
Fig. 1. Two-stage fermentation profiles for ABE using C. acetobutylicum ATCC 824.
(A) Continuous fermentation profile of the first stage. (B) Repeated fed-batch
(second stage) coupled with gas stripping showing product concentrations in both
the reactor and condensate.
times (GS1, GS2, GS3, Fig. 1B), before the fermenter reached its
maximum level and was replaced by a second fermenter in paral-
lel. Gas stripping (GS1, Fig. 1B) was switched on after 192 h of fer-
mentation when the butanol concentration reached 11.2 g/L and
operated for 5 h until butanol in the reactor decreased to 1.9 g/L.
At the end of gas stripping biomass concentration increased to
5.6 g/L and butyric acid decreased to 0.5 g/L, indicating solvento-
genesis and no negative influence gas stripping on cell growth, a
similar observation made by Ezeji et al. (2003). Allowing the buta-
nol concentration to decrease to 1.9 and 2.7 g/L in phase GS1 and
GS2, respectively, instead of maintaining it at 5 g/L, resulted in a
low gas stripping efficiency. The reduced gas stripping efficiency
also explains a decrease of the butanol concentration in the con-
densate to below 2 wt.%.
Gas stripping was switched on at 250 h (GS3 Fig. 1) at a reduced
gas circulation rate to find a balance between ABE formation and
removal rate. The butanol concentration in the reactor was
10.9 g/L and gas stripping was carried on for 4 h until butanol con-
centration reached 5.9 g/L and could not be reduced further. At a
butanol removal rate of 1.6 g/L/h less water was condensed in a
concentrate with 33.9 g/L butanol and 40.6 g/L ABE (Table 1). At
the end of 252 h when the repetitive fed-batch and gas stripping
were stopped, 829 ml condensate with a total butanol mass of
18.4 g was collected. A third fed-batch fermentation was per-
formed in a new fermenter, which was connected to the first stage
fermenter at 256 h and operated until 8 g/L butanol was reached
(FB3, Fig. 1B). At the start of gas stripping at 272 h, glucose and bu-
tyric acid concentration in the reactor were 0.2 and 1.7 g/L, respec-
tively. Additional glucose was manually fed into the second reactor
to increase the concentration to 10 g/L. This resulted in a higher
butanol and ABE condensate of 59 and 73 g/L, respectively. The
butanol concentration in the reactor could be maintained at 5 g/
L. After 300 h the concentrations in the first stage fermenter were
unstable and it was disconnected from the second stage fermenter
(Fig. 1A). Butanol concentration in the first stage fermenter de-
creased to 3 g/L and acids concentration increased to 6.3 g/L indi-
cating clostridia degeneration. Fresh heat-shocked spores were
added in the reactor and it was operated continuously until steady
conditions were reached at 400 h. Two more repeat fed-batches
coupled with gas stripping were performed after 415 and 440 h
(results not shown) and up to 57 g/L butanol and 67 g/L ABE in
the condensate was collected. Acids build up and cessation of prod-
uct formation in the fed-batch stages was not experienced but
rather in the first stage fermenter, which was operated continu-
ously. Cell degeneration is a common, well-known clostridia prob-
lem during continuous fermentation. The method used by Richter
et al. (2012) and also applied in this work of re-inoculating with
heat shocked spores proved successful even though it is a more
reactive approach as compared to co-feeding butyric acid (Lee
et al., 2008; Richter et al., 2012). A different approach in our pro-
cess could be to alternate between two first stage continuous fer-
menters, so that when one reactor suffers from clostridia
degeneration it could be alternated with another. The entire pro-
cess of using and alternating between two fermenters for stage
one and also keeping the concept of two fermenters for stage
two could minimize the downtime experienced at 200, 300 and
380 h (Fig. 1A).
3.2. Separation characteristics of the pervaporation unit
Pervaporation experiments with a mixture of butanol and water
binary solutions and with real fermentation broth have been car-
ried out. Fig. 2A shows the partial fluxes for pervaporation experi-
ments using the real fermentation broth taken from the second
stage fermenter before and after adding extra acetone, butanol
and ethanol to the broth. The partial fluxes of acetone, butanol,
 M. Setlhaku et al. / Bioresource Technology 136 (2013) 102–108 105

Table 1
Fermentation results of the two-stage process (continuous and fed-batch) coupled with gas stripping and pervaporation.

Stages Fed-batch and Gas stripping Pervaporationd

FB1 FB2a + GS1 FB2b + GS2 FB2c + GS3 FB3 + GS1 Ex situ
Process time (h) 24 22 26c 27 23 22
Spent fermentation broth
Butanol (g/L) 12.1 1.9 2.7 5.9 5.1 8.3
Acetone (g/L) 6.8 0.4 1.5 3.5 3.7 3.2
Ethanol (g/L) 1.9 – 0.4 0.8 0.9 0.8
Total ABE (g/L) 20.8 2.3 4.6 10.2 9.7 12.3
Total Acids (g/L) 1.8 3.4 3.1 3.2 3.8 3.3
Volume in fermenter (L) 1.4 0.92 2.3 3.8 0.99 –
Condensate/permeate
Volume collected (ml) – 452 110 267 84 11.7
Butanol (g/L) – 16.4 18.3 33.9 58.9 167.1
Acetone (g/L) – 4.5 4.1 4.5 10.6 97.8
Ethanol (g/L) – 1.9 1.2 2.3 3.4 4.8
Total ABE (g/L) – 22.8 23.6 40.6 72.9 269.4
aButanola – 0.5 2.5 4.2 3.4 26.4
aAcetonea – 1.2 0.41 0.21 0.29 36.7
Overall volumetric productivityb (gABE/L/h) 0.87 0.56 0.22 0.4 0.73 0.68e
Overall volumetric productivityb (gBuOH/L/h) 0.51 0.42 0.14 0.25 0.47 0.45e
a
Selectivity, a, is calculated as = [y/(1  y)]/[x/(1  x)], where x and y are weight fractions of ABE in fermentation broth and condensate, respectively.
b
Calculated from the sum of the mass of butanol in the condensate and remaining reactor, divided by the liquid volume in reactor and process time.
c
Total time it took for 1.9 g/L butanol from FB2b + GS1 to increase to 9.44 g/L (FB2c), including the 4 h of gas stripping (GS3).
d
Mean values for four pervaporation experiments carried out with fermentation broth before addition of ABE to extend the measured range.
e
Calculated from the 20 h required for fed-batch in stage two plus 2 h of pervaporation, which would also be applied intermittently if the pervaporation is connected to the
fed-batch reactor.

Fig. 2. Results for pervaporation of fermentation broth, dependent on the feed concentration of butanol, acetone, ethanol, and acids. (A) Permeate fluxes, (B) mass fractions in
the permeate.

ethanol, butyric acid, and acetic acid were determined. The partial
flux of butanol rises with the feed concentration up to values of
300 g/(m2 h). Compared to butanol the flux of acetone is about
45% higher, while the flux of ethanol is 70% lower. Maximum fluxes
of butyric and acetic acid are 1.6 g/(m2 h) and 0.46 g/(m2 h),
respectively, at mean feed concentrations of 1.7 g/L and 0.5 g/L.
Assuming a linear dependency of the partial fluxes on the feed con-
centrations and an ABE ratio of 3:6:1, the permeate always con-
tains 1.5 times more butanol than acetone and 20 times more
butanol than ethanol. This trend can also be seen in Fig. 2B. Com-
pared to the organic acids contained in the feed, the butanol con-
centration in the permeate is even more than 100 times higher.
This underlines the suitability of pervaporation as an appropriate
method for butanol recovery since butyric and acetic acid, which
are precursors for the production of solvents are hardly removed.
The partial fluxes of butanol for pervaporation experiments
using a binary mixture and real fermentation broth as feed are
shown in Fig. 3A. They increase linearly with the feed concentra-
tion up to 370 g/(m2 h) for each feed while the partial flux of water
remains practically constant (350 and 420 g/(m2 h)). Similar per-
meate fluxes have been reported by Claes et al. (2012) however,
a direct comparison of the results is not possible due to different
pervaporation temperatures, permeate pressures and feed concen-
trations. Based on Fig. 3A it can be concluded that by-products and
precursors like acetone, ethanol and acids as well as further fer-
mentation broth components like salts or sugars do not influence
the butanol separation. Although these components do not influ-
ence the partial flux of butanol, one must consider their influence
on mass fraction of butanol in permeate. The butanol concentra-
tions in permeate change drastically for the real fermentation
106 M. Setlhaku et al. / Bioresource Technology 136 (2013) 102–108
Fig. 3. Results for pervaporation experiments, carried out with binary mixtures and real fermentation broth dependent on the feed concentration of butanol. (A) Partial fluxes
of n-butanol, (B) mass fractions of butanol in the permeate.
broth as compared to a binary butanol–water mixture. For the bin-
ary mixtures the concentration of butanol increases to 415 g/L with
higher feed mass fractions of butanol. In the presence of acetone
and ethanol the butanol concentration in permeate for higher
butanol concentrations in feed was found to be about 20% lower.
The influence of acetone and ethanol ceases at lower feed concen-
trations, because acetone, butanol and ethanol are present in the
feed in a ratio of 3:6:1 and the dilution by acetone and ethanol is
rather small.
With regards to the maximum concentrations of butanol in sec-
ond stage fermenter, which vary between 10 and 12 g/L, the max-
imum concentrations of n-butanol that can be achieved by
pervaporation are in a range between 180 and 200 g/L. The corre-
sponding total ABE concentration ranges between values of 310
and 350 g/L. To ensure that these permeate concentrations can
be reached when connecting the pervaporation unit directly to
the reactor, the concentration in the feed broth should be kept at
the highest level possible in the fermentation. Therefore the mem-
brane area must be adapted to the butanol productivity of the pro-
cess, similar to the gas circulation rate in gas-stripping that was
adapted to butanol productivity as described If more butanol is
separated from the broth than produced, the concentration in the
feed and therefore the concentration of butanol in the permeate
decreases. Because the productivity during a fermentation period
might change, minimum and maximum values for the solvent con-
centrations in the broth have to be determined and pervaporation
should be operated intermittently as described above for the gas
stripping. Otherwise, in case of a continuous operation, the perme-
ate concentrations will be significantly lowered. A direct compari-
son between gas stripping and pervaporation (Table 1) shows that
pervaporation yields higher butanol concentration in the perme-
ate, 167.1 g/L, as compared to gas stripping of up to 60 g/L, and
might simplify downstream processing. In contrast gas stripping
is easier to perform and does not pose operational challenges like
fouling. To be able to eventually choose the best separation meth-
od, total costs of production including further purification of sol-
vents have to be considered.
3.3. Process comparison and feasibility indications using window of
operation
The use of the ‘‘operating window’’ as a tool for bioprocess de-
sign, optimization and feasibility indication was first introduced by
Woodley and Titchner-Hooker (1996). The ‘‘operating window’’ for
butanol fermentations was briefly introduced in the work of Rühl
(2012), using the effective butanol amount per medium volume
in grams per liter as a function of process time as key parameters.
However in this work the butanol volumetric productivity and the
butanol concentration in the final product were chosen as key
parameters as shown in Fig. 4. As a benchmark, the minimum pro-
ductivity and butanol concentration, Pmin = 0.24 g/L/h and
Cmin = 13 g/L, obtained from the best performing batch fermenta-
tions with clostridia (Jones and Woods, 1986; Rühl, 2012). Mariano
and Filho (2012) calculated a minimum butanol target of 36 g/L
(CDSP,min) in the fermentation broth that should result from process
improvements aiming to reduce energy consumption of the distil-
lation unit. Batch and one-stage continuous fermentation pro-
cesses without cell immobilization or in situ product removal are
excluded in the comparison, as they are industrially uneconomical.
This paper deals with continuous fermentation as the base mode of
operation and the minimum operation time is set to 21 days. This
takes into account the fact that continuous fermentations in one or
more reactors require longer operation time to reach steady state
and operate economically, as reported for plants in the former
USSR and China (Ni and Sun, 2009; Nimcevic and Gapes, 2000).
Using the defined parameters marked on the butanol productivity
against concentration plot, Fig. 4, the resulting shaded area repre-
sents the feasible ABE fermentations. A butanol productivity of 5 g/
L/h (Pindustry,min) is chosen as an industrial hurdle rate estimate. As
ABE fermentation is a highly competitive field at present, this pa-
per will focus more on the processes with the highest potential
of fitting in the operating window. These processes are defined
as one-stage continuous with cell recycle (c), two-stage reactors
(cc), fermentations integrated with product removal (int) and pro-
cesses that include cell immobilization (im), Fig. 4. The rest of the
references in Fig. 4 not mentioned in the discussion are supplied as
a Supplementary list.
The majority of processes developed with cell immobilization
managed to increase butanol productivity above Pmin, but a few
meet the industrial hurdle rate of 5 g/L/h (16-im: Qureshi et al.,
2000; 11-im and 12-im: Survase et al., 2012) (Fig. 4). Survase
et al. (2012) reported the highest productivity of 10.94 gButanol/L/
h an butanol concentration of 7.3 g/L (Fig. 4: 12-im) from their con-
tinuous process with C. acetobutylicum DSM 792 immobilized on
wood pulp by using a mixture of sugars (glucose, mannose, galact-
ose, arabinose and xylose) as compared to 7.6 g/L/h from glucose
 M. Setlhaku et al. / Bioresource Technology 136 (2013) 102–108 107

Fig. 4. Window of operation for ABE continuous, two-stage reactors, immobilized and product integrated fermentations. Continuous, h: 1-c (Schlote and Gottschalk, 1986); 2-
c (Baba et al., 2012); 3-c, 4-c (Tashiro et al., 2005). Two-stage process, D: 1-cc, 2-cc (Afschar, 1991); 3-cc, 4-cc (Bahl et al., 1982); 5-cc (Frick and Schügerl, 1986); 6-cc (Godin
and Engasser, 1988); 7-cc (Guo et al., 2012); 8-cc (Lai and Traxler, 1994); 9-cc (Ni and Sun, 2009); 10-cc, 11-c+fb (Setlhaku et al., 2012); 12-cc (van Hecke et al., 2012); 13-cc
(Gapes et al., 1996); 14-cc (Mutschlechner et al., 2000); 15-cc (Survase et al., 2011). Immobilized, : 1-im (Förberg et al., 1983); 2-im (Frick and Schügerl, 1986); 3-im (Huang
et al., 2004); 4-im, 5-im (Largier et al., 1985); 6-im, 7-im (Napoli et al., 2010); 8-im (Park et al., 1990); 9-im (Qureshi and Maddox, 1995); 10-im, 11-im, 12-im (Survase et al.,
2012); 13-im (Welsh et al., 1987); 14-im (Yen and Li, 2011); 15-im (Survase et al., 2011), 16-im (Qureshi et al., 2000); 17-im (Zhang et al., 2009). Integrated, s:1-int (Bankar
et al., 2012); 2-int (Ennis et al., 1987); 3-int (Ezeji et al., 2007a); 4-int (Izák et al., 2008); 5-int (Li et al., 2010); 6-int (Maddox et al., 1995); 7-int (Li et al., 2011); 8-int (Nielson
and Pranther, 2009); 9-int (Qureshi and Maddox, 1991); 10-int, 11-int (Qureshi et al., 1992); 12-int (Qureshi and Blaschek, 1999a,b); 13-int (Qureshi et al., 2001); 14-int
(Qureshi et al., 2006); 15-int, 16-int (This work); 17-int (Shukla et al., 1989); 18-int (Xue et al., 2012); 19-int (Van Hecke et al., 2012); 20-int, 21-int (Ezeji et al., 2003); 22-int
(Ezeji et al., 2007b); 23-int (Gapes et al., 1996); 24-int, 25-int (Mariano et al., 2011); 26-int (Ishizaki et al., 1999); 27-int (Richter et al., 2012).

only (Fig. 4: 11-im). This is also a significant improvement from
0.1 gButanol/L/h in their batch process (Fig. 4: 10-im). The butanol
concentration obtained is much lower than 13 g/L attained from
two-stage processes (3-cc: Bahl et al., 1982; 11-c+fb: Setlhaku
et al., 2012). This shows that two-stage fermenter configurations
result in increased butanol concentration but suffer from low pro-
ductivities (<1 g/L/h) due to low dilution rates required to operate
these processes. Including cell recycle (2-c: Baba et al., 2012; 3-c:
Tashiro et al., 2005) results in higher butanol titers and productiv-
ities similar to those attained with cell immobilization. It is prob-
able that including a product removal stage to the processes
published by Baba et al., (2012), Qureshi et al., (2000), Survase
et al., (2012) and Tashiro et al., (2005) with butanol productivities
above 5 g/L/h, Pindustry,min, and concentration close to Cmin, could
improve their feasibility.
Processes with in situ butanol removal shift the concentration
from Cmin towards CDSP,min, Fig. 4. Pervaporation (3-int: Ezeji
et al., 2007a; 12-int: Qureshi and Blaschek, 1999a; 13-int: Qureshi
et al., 2001; 19-int: van Hecke et al., 2012) and perstraction (10-
int: Qureshi et al., 1992) integrated processes result in butanol ti-
ters of or more than 36 g/L but have lower productivities. The buta-
nol producing processes which lie within the operating window
are 21-int: Ezeji et al. (2003); 22-int: Ezeji et al. (2007b); 9-cc:
Ni and Sun (2009); 9-int: Qureshi and Maddox (1991); 17-int: Shu-
kla et al. (1989), 18-int: Xue et al. (2012), 15-int: gas stripping , this
work and, 16-int: pervaporation, this work. The most dominant
product removal in these feasible processes is gas stripping, fol-
lowed by pervaporation and liquid–liquid extraction. Shukla
et al. (1989) included cell immobilization and in situ extraction
by 2-ethyl-1-hexanol and obtained a butanol concentration of
43.6 g/L at a productivity of 0.55 g/L/h. Qureshi and Maddox
(1991) also included cell immobilization and gas stripping and
were able to produce up to 47.7 g/L butanol at a high productivity
of 4.35 g/L/h rate. This is a much better performing process as com-
pared to Ennis et al. (1987) with 3.7 gbutanol/L and 0.19 gbutanol/L/h
(Fig. 4: 2-int). Both reported the same clostridia strain, substrate
and bone char for immobilization but Ennis et al. (1987) operated
the fermentation continuously in two-stages as compared to the
one stage used by Qureshi and Maddox (1991). Ezeji et al. (2003)
showed that a batch process coupled with gas stripping using the
hyper butanol producing Clostridium beijerinckii BA 101 (Fig. 4:
21-int) could improve the viability of ABE fermentation. They were
later able to show the feasibility of using liquefied cornstarch in a
fed-batch operated process integrated with gas stripping by pro-
ducing 56 g/L butanol (Fig. 4: 22-int). Interestingly, the only feasi-
ble multi-fermenter process is that used in China (Fig. 4: 9-cc),
where up to 76.5 g/L butanol concentrate could be produced from
the fermenter cascades after 170 h operation. In this work, we have
integrated gas stripping and pervaporation to our developed two-
stage process, where a continuous fermenter is coupled to a fed-
108 M. Setlhaku et al. / Bioresource Technology 136 (2013) 102–108
batch operated fermenter. A high butanol concentration of 12.1 g/L
from the two-stage process (Fig. 4:11-c+fb) could be further im-
proved up to 58.9 gbutanol/L and 0.47 gbutanol/L/h (Fig. 4: 15-int) by
including gas stripping and to 167.1 gbutanol/L and 0.45 gbutanol/L/h
when pervaporation was used (Fig. 4: 16-int). The final butanol ti-
ter of 167.1 g/L obtained in our process is the highest reported to
date with pervaporation and could significantly further reduce
downstream energy duty. The feasibility of our process could be
improved further by including cell immobilization as in the case
of Qureshi and Maddox (1991) (Fig. 4: 9-int).
4. Conclusions
ABE producing processes with cell immobilization or recycle are
limited by butanol concentration but result in higher productivi-
ties. Integrating product removal stage to the fermenter/s signifi-
cantly increases process productivity. In this work, we showed an
integrated two-stage process, in which a continuous fermenter is
coupled to a fed-batch operated fermenter and a downstream unit.
Through the integration of gas stripping and pervaporation the
butanol concentration in the broth increased to 59 and 167 g/L,
respectively, above the minimum target, 36 g/L, required to further
reduce distillation duty.
Acknowledgements
‘‘The research leading to these results has received funding
from the Ministry of Innovation, Science and Research of North
Rhine-Westphalia in the frame of CLIB-Graduate Cluster Industrial
Biotechnology, Contract No.: 314 – 108 001 08’’. Sulzer Chemtech
A.G. is kindly acknowledged for providing pervaporation
membranes.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/
j.biortech.2013.02.046.
References
Baba, S., Tashiro, Y., Shinto, H., Sonomoto, K., 2012. Development of high speed and
highly efficient butanol production systems from butyric acid with high density
of living cells of Clostridium saccharoperbutylacetonicum. J. Biotechnol. 157, 605–
612.
Bahl, H., Andersch, W., Gottschalk, G., 1982. Continuous production of acetone and
butanol by Clostridium acetobutylicum in a two-stage phosphate limited
chemostat. Eur. Appl. Microbiol. Biotechnol. 15, 201–205.
Claes, S., Vandezande, P., Mullens, S., De Sitter, K., Peeters, R., Van Bael, M.K., 2012.
Preparation and benchmarking of thin film supported PTMSP–silica
pervaporation membranes. J. Membr. Sci. 389, 265–271.
Ennis, B., Qureshi, N., Maddox, I.S., 1987. In-line toxic product removal during
solvent production by continuous fermentation using immobilized Clostridium
acetobutylicum. Enzyme. Microb. Technol. 9, 672–675.
Ezeji, T.C., Qureshi, N., Blaschek, H.P., 2003. Production of butanol by Clostridium
beijerinckii BA101 and in-situ recovery by gas stripping. World. J. Microbiol.
Biotechnol. 19, 595–603.
Ezeji, T., Qureshi, N., Blaschek, H.P., 2007a. Bioproduction of butanol from biomass:
from genes to bioreactors. Curr. Opinion. Biotechnol. 18, 220–227.
Ezeji, T., Qureshi, N., Blaschek, H.P., 2007b. Production of acetone butanol (AB) from
liquefied cornstarch, a commercial substrate, using Clostridium beijerinckii
coupled with product recovery by gas stripping. J. Ind. Microbiol. Biotechnol. 34,
771–777.
Fadeev, A.G., Meagher, M.M., Kelley, S.S., Volkov, V.V., 2000. Fouling of poly[-1-
(trimethylsilyl)-1-propyne] membranes in pervaporative recovery of butanol
from aqueous solutions and ABE fermentation broth. J. Memb. Sci. 173, 133–
144.
Gapes, J.R., Nimcevic, D., Friedl, A., 1996. Long-term continuous cultivation of
Clostridium beijerinckii in a two-stage chemostat with on-line solvent removal.
Appl. Environ. Microbiol. 62, 3210–3219.
Godin, C., Engasser, J.M., 1988. Improved stability of the continuous production of
acetone butanol by Clostridium acetobutylicum in a two-stage process.
Biotechnol. Lett. 10, 389–392.
Halan, B., Buehler, K., Schmid, A., 2012. Biofilms as living catalysts in continuous
chemical syntheses. Trends in Biotechnol. 30, 453–465.
Heitmann, S., Krings, J., Kreis, P., Lennert, A., Pitner, W.R., Górak, A., Schulte, M.M.,
2012. Recovery of n-butanol using ionic liquid-based pervaporation
membranes. Sep. Purif. Technol. 97, 108–114.
Jones, D.T., Woods, D.R., 1986. Acetone–butanol fermentation revisited. Microbiol.
Rev. 50, 484–524.
Lee, S.M., Cho, M.O., Park, C.H., Chung, Y., Kim, J.H., Sang, B., Um, Y., 2008.
Continuous butanol production using suspended and immobilized Clostridium
beijerinckii NCIMB 8052 with supplementary butyrate. Energy Fuels 22, 3459–
3464.
Mariano, A.P., Filho, R.M., 2012. Improvements in biobutanol fermentation and their
impacts on distillation energy consumption and wastewater generation.
Bioenerg. Res. 5, 504–514.
Ni, Y., Sun, Z., 2009. Recent progress on industrial fermentative production of
acetone–butanol–ethanol by Clostridium acetobutylicum in China. Appl.
Microbiol. Biotechnol. 83, 415–423.
Nimcevic, D., Gapes, J.R., 2000. The acetone–butanol fermentation in pilot plant and
pre-industrial scale. J. Mol. Microbiol. Biotechnol. 2, 15–20.
Qureshi, N., Maddox, I.S., 1991. Integration of continuous production and recovery
of solvents from whey permeate: use of immobilized cells of Clostridium
acetobutylicum in a fluidized bed reactor coupled with gas stripping. Bioproc.
Eng. 6, 63–69.
Qureshi, N., Maddox, I.S., Friedl, A., 1992. Application of continuous substrate
feeding to the ABE fermentation: relief of product inhibition using extraction,
perstraction, stripping and pervaporation. Biotechnol. Prog. 8, 382–390.
Qureshi, N., Blaschek, H.P., 1999a. Production of acetone–butanol–ethanol (ABE) by
hyper-butanol producing mutant strain of Clostridium beijerinckii BA101 and
recovery by pervaporation. Biotechnol. Prog. 15, 594–602.
Qureshi, N., Blaschek, H.P., 1999b. Fouling studies of a pervaporation membrane
with commercial fermentation media and fermentation broth of hyper-
butanol-producing Clostridium beijerinckii BA101. Separ. Sci. Technol. 34,
2803–2815.
Qureshi, N., Schripsema, J., Lienhardt, J., Blaschek, H.P., 2000. Continuous solvent
production by Clostridium beijerinckii BA101 immobilized by adsorption onto
brick. W. J. Microbiol. Biotechnol. 16, 377–382.
Qureshi, N., Meagher, M.M., Huang, J., Hutkins, R.W., 2001. Acetone butanol ethanol
(ABE) recovery by pervaporation using silicalite–silicone composite membrane
from fed-batch reactor of Clostridium acetobutylicum. J. Memb. Sci. 187, 93–102.
Qureshi, N., Hughes, S., Maddox, I.S., Cotta, M.A., 2005. Energy-efficient recovery of
butanol from model solutions and fermentation broth by adsorption.
Bioprocess. Biosyst. Eng. 27, 215–222.
Richter, H., Qureshi, N., Heger, S., Dien, B., Cotta, M.A., Angenent, L.T., 2012.
Prolonged conversion of n-butyrate to n-butanol with Clostridium
saccharoperbutylacetonicum in a two-stage continuous culture with in-situ
product removal. Biotechnol. Bioeng. 109, 913–921.
Rühl, J., 2012. Characterization and Engineering of Pseudomonas putida for Aerobic
n-butanol Production, Ph.D. Thesis. Technische Universität Dortmund,
Germany.
Setlhaku, M., Brunberg, S., del Amor Villa, E.M., Wichmann, R., 2012. Improvement
in the bioreactor specific productivity by coupling continuous reactor with
repeated fed-batch reactor for acetone–butanol–ethanol production. J.
Biotechnol. 161, 147–152.
Shukla, R., Kang, W., Sirkar, K.K., 1989. Acetone–butanol–ethanol (ABE) production
in a novel hollow fiber fermenter-extractor. Biotechnol. Bioeng. 34, 1158–1166.
Survase, S.A., van Heiningen, A., Granström, T., 2012. Continuous bio-catalytic
conversion of sugar mixture to acetone–butanol–ethanol by immobilized
Clostridium acetobutylicum DSM 792 Appl. Microbiol. Biotechnol. 93, 2309–
2316.
Tashiro, Y., Takeda, K., Kobayashi, G., Sonomoto, K., 2005. High production of
acetone–butanol–ethanol with high cell density culture by cell-recycling and
bleeding. J. Biotechnol. 120, 197–206.
van Hecke, W., Vandezande, P., Claes, S., Vangeel, S., Beckers, H., Diels, L., De Wever,
H., 2012. Integrated bioprocess for long-term continuous cultivation of
Clostridium acetobutylicum coupled to pervaporation with PDMS composite
membranes. Biores. Technol. 111, 368–377.
Weyd, M., Richter, H., Troeber, O., 2012. Bio-alcohol concentration by pervaporation
with organophilic zeolite membranes-influence of protective coatings on
fouling tendency. 2012. Proc. Eng. 44, 549–552.
Woodley, J.M., Titchner-Hooker, N.J., 1996. The use of windows of operation as a
bioprocess design tool. Biopr. Eng. 14, 263–268.
Xue, C., Zhao, J., Lu, C., Yang, S.T., Bai, F., Tang, I., 2012. High-titer n-butanol
production by Clostridium acetobutylicum JB200 in fed-batch fermentation with
intermittent gas stripping. Biotechnol. Bioeng. 109, 2746–2756.
Zhou, Y.H., Titchener-Hooker, N.J., 1999. Visualizing integrated bioprocess designs
through ‘‘windows of operation’’. Biotechnol. Bioeng. 65, 550–557.