Biocatalysis and Biotransformation, JanuaryFebruary 2009; 27(1): 2735

ORIGINAL ARTICLE

Modelling ethanol production from cellulose: separate hydrolysis and

fermentation versus simultaneous saccharification and fermentation

R. E. T. DRISSEN1, R. H. W. MAAS2, J. TRAMPER1, & H. H. BEEFTINK1

Wageningen University and Research Center, 1Food and Bioprocess Engineering Group, PO Box 8129, 6700 EV Wageningen
and 2Agrotechnology and Food Sciences Group, PO Box 17, 6700 AA Wageningen, The Netherlands
(Received 19 August 2007; accepted 25 February 2008)

Abstract
In ethanol production from cellulose, enzymatic hydrolysis, and fermentative conversion may be performed sequentially
(separate hydrolysis and fermentation, SHF) or in a single reaction vessel (simultaneous saccharification and fermentation,
SSF). Opting for either is essentially a trade-off between optimal temperatures and inhibitory glucose concentrations on the
one hand (SHF) vs. sub-optimal temperatures and ethanol-inhibited cellulolysis on the other (SSF). Although the impact of
ethanol on cellobiose hydrolysis was found to be negligible, formation of glucose and cellobiose from cellulose were found to
be significantly inhibited by ethanol. A previous model for the kinetics of enzymatic cellulose hydrolysis was, therefore,
extended with enzyme inhibition by ethanol, thus allowing a rational evaluation of SSF and SHF. The model predicted SSF
processing to be superior. The superiority of SSF over SHF (separate hydrolysis and fermentation) was confirmed
experimentally, both with respect to ethanol yield on glucose (0.41 g g
1 for SSF vs. 0.35 g g
1 for SHF) and ethanol
production rate, being 30% higher for an SSF type process. High conversion rates were found to be difficult to achieve since
at a conversion rate of 52% in a SSF process the reaction rate dropped to 5% of its initial value. The model, extended with
the impact of ethanol on the cellulase complex proved to predict reaction progress accurately.

Keywords: SSF, SHF, ethanol inhibition, glucose inhibition, model

Introduction waste materials, such as grass or wheat straw, which

With decreasing oil availability and the negative
effects of traditional fossil fuels on the environment,
initiatives have started to look for alternative sources
of liquid fuel. One of these promising sources is
ethanol, which can be produced through fermenta-
tion of a sugar source, such as sugar cane or glucose
syrup from starch. However, for introduction of
ethanol for fuel purposes to be successful, the price
of ethanol should be competitive with the price of
petrol. Until the recent oil price rises, ethanol from
traditional sugar sources, such as starch has been
more expensive than petrol because of the relatively
high sugar cost. Furthermore, as demand for ethanol
rises, a dramatic increase in sugar price is expected
to occur (Lynd et al. 1999) with consequences for
the price of food.
Therefore, research over the past years has focused
on making available glucose from (ligno)cellulosic
can act as an abundant and cheap source of fermen-
table sugars (Wyman 1994). In addition to a neces-
sary pretreatment of the cellulosic material, this will
involve two consecutive catalytic steps, an enzymatic
conversion of the cellulose to fermentable sugars and
a fermentative conversion of these sugars to ethanol
(Sheehan &d Himmel 1999). These hydrolysis and
fermentation steps may be operated sequentially
(separate hydrolysis and fermentation, SHF) or
concurrently (simultaneous saccharification and fer-
mentation, SSF; Wingren et al. 2003). Diluted
ethanol is finally concentrated by, for instance,
distillation or pervaporation (O’Brien et al. 2000).
When opting for either SSF or SHF, one faces a
trade-off in terms of process variables (Takagi et al.
1977). With a sequential SHF process, both the
hydrolysis and the fermentation may be performed
at their optimum temperature: 378C for yeast

Correspondence: R.E.T. Drissen, Food and Bioprocess Engineering Group, PO Box 8129, 6700 EV Wageningen, The Netherlands.
E-mail: rik.beeftink@wur.nl

ISSN 1024-2422 print/ISSN 1029-2446 online # 2009 Informa UK Ltd

DOI: 10.1080/10242420802564358
28 R. E. T. Drissen et al.
fermentation and 508C for a cellulase complex
(Stenberg et al. 2000). The downside, however, is
accumulation of the hydrolysis product glucose in
the enzyme reactor. As a result, the cellulase com-
plex would, at increasing glucose concentration,
greatly suffer from product inhibition. Such product
inhibition may be avoided by continuous sugar
removal by a yeast, i.e. in an SSF type of process.
In the case of SSF, however, both the hydrolysis and
the fermentation are performed at some sub-optimal
temperature (the pH optimum is approximately 5 for
both reactions). In addition, possible effects of
ethanol on the performance of the cellulase have to
be considered.
Previous work has often limited itself to proof of
principle without actually weighing the two pro-
cesses on a quantitative basis (Mohagheghi et al.
1992; Alkasrawi et al. 2002). Some comparisons
were based on rather limited mathematical models of
both processes and for comparison. Previous math-
ematical models of an SSF-type process (Philippidis
et al. 1992; Philippidis & Wyman 1992; Wright et al.
1987; Wyman et al. 1986) have often taken only
parts of the complete trade-off into consideration,
which should include the impact of various sugar
inhibitions, as well as ethanol inhibition, an impact
of reaction temperature and a changing substrate
reactivity (South et al. 1993, 1995; Desai & Con-
verse 1997; Yang et al. 2006).
In this article both the enzymatic hydrolysis and
fermentation to ethanol will be topic of investigation.
A previously developed model, adequately describ-
ing hydrolysis and glucose production of both a
pretreated lignocellulosic substrate and Avicel, a
model substrate, will be expanded to describe
ethanol production from cellulosic material. There-
fore, effects of ethanol on cellulase are presented.
Secondly, SSF and SHF are weighed against one
another, and a model presented to simulate both
processes. This might be used for an economical
assessment and process optimization of the produc-
tion process of ethanol from cellulosic sources.
Materials and methods
Enzyme and enzyme assay
Cellubrix (Novozymes Corp. Denmark) was used as
cellulase preparation for experiments presented in
this article. Its cellulolytic activity was 56 filter paper
units (FPU) per mL of solution as determined
according to Ghose (1987). The concentration of
protein per millilitres of Cellubrix was determined
according to Bradford (1976) and found to be 32.3
mg/mL. All experiments were conducted at pH 4.8
(citrate buffer).
b-Glucosidase activity in 1.0 mL of enzyme
preparation was determined by addition of 1.0
mL of a 15 mM solution of cellobiose in 0.05 M
citrate buffer (pH 4.8), and incubation for 30 min
at 508C. The enzyme was then inactivated (5 min
at 1008C) after which glucose was measured. The
b-glucosidase activity was found to be 204 units per
mL of Cellubrix (1 unit forming 1 mmol glucose per
min, i.e. a maximum turnover rate of 0.0108 g
glucose unit
1 h
1).
Substrates
As model cellulose, Avicel was used. This was
supplied by Merck (order nr. 1.02330.0500).
The additional nutrients used for fermentation,
were: 5 g L
1 (NH4)2SO4, 3 g L
1 KH2PO4, 0.5 g
L
1 MgSO4 ×/7H2O, 1 mL L
1 trace elements
solution, 1 mL L
1 vitamin solution, 1.25 mL
L
1 ergosterol solution, 0.5 mL L
1 anti-foam.
HPLC analysis
Samples were centrifuged (1 min, 13,000 rpm); the
supernatant was inactivated by diluting 1:1 v/v
with 1 M sulphuric acid. and filtered (0.2 mm) for
removal of any residual proteins or yeast. Samples
were then analysed on a Waters (USA) HPLC with a
refractive index detector and a IOA-1000 column
(Alltech, The Netherlands) at 908C and 0.4 mL
min
1 of 3.0 mM as mobile phase.
Ethanol inhibition
Six shake flasks with 40 g L
1 of Avicel were
incubated in 0.1 M citrate buffer (pH 4.8) with
Cellubrix (20 FPU g
1 cellulose) at a temperature
of 378C. Ethanol was added to each of the shake
flasks to create a series of increasing ethanol con-
centrations of 0, 20, 40, 60, 80 and 100 g L
1
(density ethanol determined to be 0.790 kg L
1).
Samples were taken at t 1, 3, 7, 15 and 30 min, and
at t2, 3, 4, 20 and 28 h. Sample analysis was done
by HPLC as described above. Initial glucose pro-
duction rates were calculated as the slope of a linear
fit to the data points of the first 30 min. During this
period, glucose was seen to accumulate linearly and,
therefore, enzymatic activity was not yet influenced
by inhibitory effects and inactivation.
SSF in shake flasks
Two 200 mL bottles were filled with 100 mL of
suspension of 40 g L
1 Avicel (being 95% pure
cellulose) with pH 4.8 (citrate buffer) and incubated
in a 378C water bath with 7.4 and 74 g L
1 yeast
(equalling 2 g L
1 and 20 g L
1 based on dry mass,

respectively). The micro-organism was a commer-
cially available Saccharomyces cerevisiae (DSM Delft,
The Netherlands). Samples were taken every hour
for HPLC analysis on glucose and ethanol concen-
trations. The cellulose concentration was chosen to
mimic the maximum workable dry-mass substrate
concentration found in previous research (Drissen et
al. 2007), i.e. 100 g L
1 of wheat straw with an
approximately 40% cellulose concentration. Limita-
tions in handling were due to high viscosity.
SSF and SHF in fermentors
For these experiments two 3-L fermentors with a
working volume of 2 L (Applikon BV, The Nether-
lands, type ADI 1020) were used. For both experi-
ments, the initial suspension contained 40 g L
1
Avicel and additional nutrients in 0.1 M citrate
buffer (pH 4.8) and Cellubrix (12 FPU g
1
cellulose).
The SHF-process was operated at 508C for the
first 42 h to allow for the build-up of fermentable
glucose. After 42 h, 2 g L
1 (based on dry mass) of
yeast was added and the process was continued at
378C. For the SHF-process all concentrations were
monitored by measuring of samples by HPLC.
The SSF-process was continuously operated at
378C and 2 g L
1 yeast (based on dry mass) was
added at the start of the experiment. For the SSF-
type fermentation, cumulative carbon dioxide pro-
duction was measured with an online gas analyser
(1440, Servomex), using 1 L min
1 N2 as carrier
gas. A condenser (set at 28C) was used to avoid
ethanol stripping from the broth. Glucose and
ethanol concentrations were also monitored by
HPLC analysis of samples.
Numerical procedure
For modelling purposes and solution of differential
equations, a first-order Euler forward algorithm was
used (Press 2007).
Results and discussion
Ethanol inhibition of cellulase activity
When contemplating a process for bio-ethanol pro-
duction where both the enzymatic conversion of
cellulose to glucose and the parallel fermentation of
glucose to ethanol are combined, one must take into
consideration possible effects of ethanol on the
performance of the cellulase complex, as has been
suggested previously (Wu & Lee 1997). In the present
research, initial glucose production rates were mea-
sured for incubations of Avicel and Cellubrix at
SHF vs. SSF 29
various ethanol concentrations. Values for R2 were
0.95 and higher.
Initial glucose production rates from Avicel were
found to be inhibited by ethanol (Figure 1). Regres-
sion with a general type of product inhibition
equation (eqn 1) yielded an inhibition constant
Ki 95 g L
1.
vv(0) × [Ki;EtOH =(Ki;EtOH CEtOH )] (1)
The model fit using eqn 1 and an inhibition constant
of 95 g L
1 is also plotted in Figure 1. This ethanol
inhibition is somewhat less severe than reported
elsewhere: Phillipidis and Hatzis (1997) found a 1st-
order inhibition constant of approximately 50 g L
1,
whereas Wu and Lee (1997) found almost complete
inhibition of the reaction at approximately 65 g L
1
ethanol. Apparently, the current Cellubrix prepara-
tion is less susceptible to ethanol inhibition.
Previously (Drissen et al. 2007), it was shown that
cellobiose activity is in excess in Cellubrix. Potential
effects of ethanol were further investigated by study-
ing the accumulation of the intermediate cellobiose
over time for various reaction mixtures with Cellu-
brix, Avicel and a specific ethanol concentration
(Figure 2). In all cases, the cellobiose concentration
was found to rise quickly due to initial formation of
the dimeric cellobiose from polymeric cellulose by
the cellulase activity of Cellubrix. Due to the
increasing cellobiose concentration, however, cello-
biose degradation then started to increase and the
net formation of cellobiose decreased,, until, finally,
production and degradation of cellobiose were
balanced and a plateau value was reached.
40
glucose production rate (mg.l-1.h-1)
30
20
10
0
0 20 40 60 80 100
ethanol concentration (g/l)
Figure 1. Effect of ethanol concentration on the production rate
of glucose by Cellubrix (20 FPU g
1 cellulose) from Avicel (40 g
L
1), experimental data (j) and model fit (**) at 378C.
30 R. E. T. Drissen et al.
v1
1.0 cellulose 0 cellobiose
v2
cellobiose 0 glucose
v3
0.8 cellulose 0 glucose (2)
cellobiose concentration (g.l-1)

It will be used as a starting point for this paper.

Combined with the effects of ethanol on cellulase
0.6 (eqn 1), the hydrolysis model is:
kmax;1 × CE e
Ea =(R×T)
KD (T)×t
v1 (t; T) × ×e ×CC (t)
0.4 KL  CE e
Ea =(R×313)
 
CC (t)

Krec × 1

CC (0)
Ki;G Ki;EtOH

×e × (3)
0.2 Ki;G  CG (t) Ki;EtOH  CEtOH

0.0
0 200
Figure 2. Accumulation of cellobiose from Avicel (40 g L
1) after
incubation with Cellubrix (20 FPU g
1 cellulose) and various
concentrations of ethanol (", 0 g L
1, 2, 20 g L
1, ', 40 g
L
1, ^, 60 g L
1, %, 80 g L
1, \, 100 g L
1).
Lower plateau values for cellobiose were observed
at higher ethanol concentrations. Given the fact that
the plateau values are below 1.92 g L
1, the Km of
the cellobiose (Dekker 1986), the conversion of
cellobiose to glucose can be considered to be first
order in cellobiose. A plateau value means the
production rate of glucose is equal to consumption
rate.
The ratio between glucose production rate and
maximum glucose consumption rate by cellobiose
equals Cg(t)/[Cg(t)Km]. At increasing ethanol con-
centrations, this ratio becomes smaller; therefore,
the ratio between glucose production rate and
maximum glucose consumption rate becomes smal-
ler, leading to the conclusion that the cellulase
activity is more strongly inhibited by ethanol than
the cellobiose activity. The impact of ethanol on the
conversion reaction of cellobiose to glucose can
therefore be neglected.
Model development
A generic model for cellulose hydrolysis and glucose
production from various cellulose sources by a
commercial cellulase complex has been studied
previously (Drissen et al. 2007). During this model
development, the impact of various enzyme concen-
trations, various temperatures, and the effects and
kinetics of using various cellulose sources (both
Avicel as model substrate and pretreated wheat
straw) were examined. The model assumes three
enzyme-catalysed reactions (reaction rates v1, v2 and
v3) for cellulose hydrolysis:
e
Ea =(RT)
v2 (t; T)kmax;2 eg etotal × ×e
KD (T)t
e
Ea =(R×313)
400 1000 1200 1400 1600
time (min)
CCb (t)

(4)
Km (1  CG (t)=Ki;2 )  CCb
kmax;3 × CE e
Ea =(RT)
KD (T)×t
v3 (t; T) × ×e
KL  CE e
Ea =(R×313)
 
CC (t)

Krec × 1

CC (0)
Ki;G

CC (t)× ×e
Ki;G  CG (t)
Ki;EtOH

(5)
Ki;EtOH  CEtOH
For reactions with cellulose as a substrate (reaction 1
and 3), the active amount of enzyme was assumed to
be determined by enzyme adsorption onto the
cellulose substrate (adsorption constant KL). For
reactions with glucose as a reaction product (reac-
tions 2 and 3), glucose product inhibition was
assumed (inhibition constants Ki,G and Ki,2). For
all three reactions, the zero-order rate constant was
given as a function of temperature (activation energy
Ea); in addition, all enzyme activity was assumed to
be subject to thermal inactivation (kD). The nature
of the cellulose substrate was assumed to be conver-
sion-dependent; to this end, a recalcitrance constant
Krec was used. Finally, ethanol inhibition was
assumed to affect the rates of reactions 1 and 3
(inhibition constant Ki,EtOH).
For the thermal inactivation constant kD(T), the
following temperature dependency was assumed:
kD AD ×e
DH=RT (6)
Values for the relevant kinetic parameters for
Cellubrix on Avicel were obtained from the literature
(Dekker 1986; Drissen et al. 2007): kmax,1 0.081
h
1; KL 18.2 FPU L
1; Ea 29.8 kJ mol
1;
AD 3.64 ×/1018 h
1; DH 148 kJ mol
1; Ki 6.3
g L
1; Krec 2.8 (
); kmax,2 0.0108 g U
1 h
1;
eg 6320 U g
1; Km 1.92 g L
; Ki,2 0.54 g L
;

kmax,3 0.058 h
1. Ce and etotal depend on the
specific enzyme dosage for a specific experiment.
Yeast (Saccharomyces cerevisiae) metabolism has al-
ready been topic of extensive research. For modelling
glucose consumption and biomass formation, stan-
dard Monod kinetics (Groot et al. 1993; Oliveira et al.
1999) were assumed, expanded to include ethanol
inhibition (Philippidis & Hatzis 1997) on yeast.
This yields the following overall model for the yeast
growth rate (v4) and substrate consumption rate (v5):
mmax × Cg (t)
v4 (t) ×Cx (t)
Kg  Cg (t)
Kiy;EtOH

(biomass) (7)
Kiy;EtOH  CEtOH (t)
v4 (t)
v5 (t)

Cx (t)×ms (substrate) (8)
Yxs
The following values for kinetic constants were
adopted from the literature (Groot et al. 1993; Hatzis
et al. 1996; Oliveira et al. 1999): Kiy,EtOH 50 g L
;
Kg 0.0252 g L
; ms 0.02 g g
1 h
1; Yxg 0.11 g
g
1; mmax 0.250 h
1.
Overall, changes in concentrations of cellulose,
cellobiose, and glucose in a batch reactor are now
described by:
dCC =dt
[v1 (t)v3 (t)] (9)
dCCb =dt 1; 053×v1 (t)
v2 (t) (10)
dCG =dt1; 053×v2 (t)1; 111×v3 (t)v5 (t) (11)
dCx =dt v4 (t) (12)
The formation of ethanol and CO2 are calculated
by closing the overall carbon mass balance (Hatzis et
al. 1996). It is assumed that all carbon, not used for
biomass formation, is used for energy production
according to:
C6 H12 O6 0 2 CO2 2 C2 H5 OH:
Taking into consideration the slightly higher molar
mass of ethanol (46 g mol
1) vs. that of carbon
dioxide (44 g mol
1), on a weight base 0.511 g of
ethanol and 0.489 g of CO2 will be produced from
every gram of glucose.
The overall mass balance describing ethanol con-
centration with time as a function of yeast and
cellulase activity becomes:
Cethanol 
0:511×½1:111×(Cc (t)
Cc (0))
1:053×(Ccb (t)
Ccb (0))
(Cg (t)
Cg (0))(Cx (t)
Cx (0)) (13)
Model testing
Experiments have been conducted to validate the
model for a simultaneous saccharification and fer-
mentation (SSF) as presented in this paper. A
SHF vs. SSF 31
requirement for an effective SSF-type of process is
that the production rate of glucose is the rate limiting
step in the process, i.e. the catalytic activity of the
yeast is in excess in comparison with the catalytic
activity of the cellulase. In this way, the concentra-
tion of the inhibiting sugars will remain negligibly
low, thus avoiding product inhibition of the en-
zymes. The experiments have been executed with
various yeast concentrations, both an excessive (i.e.
20 g L
1) concentration and a concentration that
would, according to initial model predictions, be just
sufficient to avoid glucose accumulation (i.e. 2 g
L
1 dry yeast). Figure 3 shows the production of
ethanol over time and model predictions.
Final analysis of the reaction mixture showed that
possible other components (i.e. sugars, acetic acid)
were negligible (i.e. concentrations below 0.25 g
L
1).
As can be seen in Figure 3, the model for
production of ethanol from cellulose, with a com-
mercial cellulase complex and S. cerevisiae accurately
predicts ethanol formation. Final glucose accumula-
tion has been determined, being 0.5 g L
1 and 0.8 g
L
1 for the experiments with 20 g L
1 yeast and 2 g
L
1 yeast, respectively. The production of ethanol
with time using 2 g L
1 of yeast was almost identical
to that using 20 g L
1 of yeast. Apparently, only a
small amount of yeast is sufficient to conduct an
SSF-type process, therefore adding to the ease of
implementing this type of process. The production
of glucose itself can accurately be described by the
proposed model for cellulose degradation to glucose,
12
10
ethanol concentration (g.l-1)
8
6
4
2
0
0 5 10 15 20
time (h)
Figure 3. Ethanol concentration in time for an SSF process using
Avicel (40 g L
1) at 378C, experimental data (20 g L
1 yeast 
and 2 g L
1 yeast k) and model prediction (20 g L
1
L
1 ). *
and 2 g *
32 R. E. T. Drissen et al.

expanded with the effect of ethanol on cellulase

16
activity. Therefore, the yeast has no negative impact
on cellulase activity through, e.g. cellulase protein 14
consumption for biomass formation.
12

Comparing SSF and SHF
The model as described by eqns 214 allows a
comparison between an SSF and an SHF process.
Input parameters for model calculations are listed in
Table I.
In order to compare both configurations, SSF and
SHF experiments were conducted in a 2-L fermen-
tor, and were compared with model predictions.
Figure 4 shows the measured glucose and ethanol
concentrations (dots) as a function in time for an
SHF process and the corresponding model predic-
tions. After 42 h, i.e. shortly after the yeast was
added, the glucose concentration dropped very
rapidly and the ethanol concentration rose rapidly.
At t 44 h, all accumulated glucose had been
converted; from then on, the process continued as
an SSF-type process. Cellulase activity continued to
produce new glucose, which was directly converted
to ethanol by the yeast. This can be seen by the
glucose concentration remaining low (BB1 g L
1)
and a slow further increase in ethanol concentration,
compared with the rapid initial increase. Calculated
ethanol yield was 0.35 g ethanol g
1 glucose.
Figure 5 shows the result for an SSF process in a
2-L fermentor. Both the ethanol concentration,
measured through sampling and HPLC analysis,
and the ethanol concentration, measured indirectly
through cumulative CO2 production, are shown and
compared with a model prediction. The glucose
concentration (not shown in Figure 5) was mon-
itored by HPLC analysis of samples taken every 3 h
and remained negligibly low (i.e. below 0.3 g L
1).
After completion of the experiment a complete
carbon analysis and carbon balance were made.
The final ethanol yield on glucose was 0.41 g g
1.
Comparing Figure 4 and 5 reveals that after 48 h,
the SSF process produced 30% more ethanol than
concentration (g.l-1)
10
8
6
4
2
0
0 10 20 30 40 50
time (h)
Figure 4. Experimental glucose (m) and ethanol (k) concentra-
tions in time, and model predictions ( ). *
the SHF process, i.e. with respect to ethanol yield
from a cellulose source, an SSF-type process seems
the preferable set-up. The negative effects of product
inhibition of sugars on cellulase in an SHF process
therefore outweigh the negative effects of ethanol
inhibition and a sub-optimal saccharification tem-
perature in an SSF-type process, especially given the
fact that all other factors, such as variance in
substrate crystallinity were identical for both pro-
cesses.
Secondly, the ethanol yield in an SHF process is
lower than in an SSF process. This is in line with
conclusions from previous research, although the
benefits of an SSF process are more pronounced
than found by, e.g. Wingren et al. (2003), but less
than found by Stenberg et al. (2000).
In an SHF process, shortly after yeast addition,
the excess glucose allows for biomass formation. In
an SSF process, glucose concentration is continu-
ously low and will mainly be used for cell main-
tenance, i.e. energy and, therefore, ethanol
production. For this, it will, however, be imperative

Table I. Input parameters for model comparison of SSF and SHF.

Parameter SSF SHF

Total process time modelled 48 h 48 h

Moment of yeast addition t 0 h t 40 h
Temperature 378C 508C (saccharification first 40 h)
378C (fermentation, last 8 h)
Yeast concentration 2 g L
1 2 g L
1
Avicel concentration 38 g L
1 38 g L
1
Enzyme loading
k1 and k3 12 FPU g
cellulose 12 FPU g
1 cellulose
0.032 h
1 and 0.023 h
1 0.032 h
1 and 0.023 h
1
 SHF vs. SSF 33
100
14

12 80
ethanol concentration (g.l-1)

relative reaction rate (%)

10
60
8

6
40

4
20
2

0
0
0 10 20 30 40 50 0 10 20 30 40 50
time (h) time (h)
Figure 5. Ethanol concentration as a function of time for an SSF Figure 6. Relative CO2 production rate (%) for an SSF process,
process, experimental results (k) and model prediction (black); indirectly indicating overall ethanol production rate.
direct EtOH measurement (dots) and calculated through CO2-
measurement (grey line).

during an SSF process that the catalytic activity of
the yeast at all times is larger than the catalytic
activity of the cellulase complex used, in order to
avoid sugar accumulation and therefore product
inhibition. Figure 3 demonstrates yeast’s potential
for an SSF process in this respect. This has further
been affirmed by the results presented in Figure 4
and 5.
The online measurement of CO2-production in
the SSF configuration enabled a constant determi-
nation of the reaction rate during the course of the
process (Figure 6).
Clearly, the reaction proceeded at only approxi-
mately 5% of the original reaction rate after 48 h. Due
to this rapid decline in conversion rate, it was
considered impractical to progress further with the
reaction. As Figure 5 shows, this correlates with
approximately 11 g L
1 ethanol, i.e. a conversion of
53%. This drop in reaction rate is attributed to a
change in the nature of the substrate. Since the yeast
can consume all of the sugar and ethanol concentra-
tions are still non-inhibitory (Figure 1), product
inhibition does not explain the drop in the reaction
rate. Enzyme inactivation was excluded as an ex-
planation since addition of fresh enzyme showed no
increase in reaction rate. However, the addition of
fresh cellulose substrate showed an increase in reac-
tion rate, leading to the conclusion that the enzyme
still had catalytic activity and the presence of less
recalcitrant cellulose could restart glucose produc-
tion.
Conclusions
A model to describe the hydrolysis of cellulose by a
commercial cellulase was successfully expanded to
describe the conversion of cellulose to ethanol
through the process of simultaneous saccharification
and fermentation. Ethanol has been demonstrated to
cause first-order inhibition of cellulase with a Ki of
95 g L
1. No impact of ethanol on conversion of
cellobiose to glucose could be found. It was shown
that with a relatively small inoculum of yeast,
glucose accumulation in a SSF-type process could
be avoided. Furthermore, it was shown that yeast
had no negative impact on the activity of cellulase.
The SSF type process proved to be superior to the
SHF type process, both with respect to ethanol yield
per gram of glucose produced (0.41 for SSF vs. 0.35
for SHF) and ethanol production rate, being 30%
higher for an SSF type process. However, for the
cellulose source used it was difficult to obtain a high
conversion rate and yield. Due to the increasing
recalcitrance of the substrate the conversion rate
dropped to 5% of its initial value at a conversion of
approximately 53%.
Notation
A Arrhenius constant for (h
1)
increase of initial rate at
increasing temperatures
AD Arrhenius constant for (h
1)
thermal enzyme deactivation
CC Cellulose concentration (g L
1)
CCb Cellobiose concentration (g L
1)
34 R. E. T. Drissen et al.
Acknowledgements
CE Enzyme concentration (FPU L
1)
CEtOH Ethanol concentration (g L
1) The project was financially supported through a
CG Glucose concentration (g L
1) grant from the Programme Economy, Ecology and
CX Yeast concentration (dry (g L
1) Technology (EET) by the Netherlands’ Department
weight) of Economic Affairs, the Department of Public
eg b-glucosidase activity per g of (units g
1) Housing, Spatial Planning and Environmental Pro-
protein in the enzyme tection, and the Department of Education, Cultural
preparation Affairs and Sciences. The authors thank Arianne
etotal Protein (enzyme) (g L
1
) Huijnen, Iris Camp and Steven Rothuizen for help-
concentration per L reaction ing to determine the relevant kinetic parameters.
volume
Ea Activation energy (J mol
1) Declaration of interest: The authors report no
h Planck’s constant (J s) conflicts of interest. The authors alone are respon-
(6.626 ×/10
34) sible for the content and writing of the paper.
ki Reaction rate constant (h
1)
(i 13) References
k2 Reaction rate constant of g U
1 ×/h
1
Alkasrawi M, Galbe M, Zacchi G. 2002. Recirculation of process
b-glucosidase

1 streams in fuel ethanol production from softwood based on
kmax,i Maximum specific rate (h ) simultaneous saccharification and fermentation. Appl Biochem
constant (i 1 or 3) Biotechnol 98:849861.
kD Thermal deactivation constant (h
1) Bradford, MM. 1976. A rapid and sensitive method for quantita-
Kiy,EtOH Inhibition constant of ethanol (g
1) tion of microgram quantities of protein utilizing the principle of
protein-dye-binding. Anal Biochem 72:24854.
on yeast
Dekker RFH. 1986. Kinetic, inhibition, and stability properties of
Kg Glucose saturation constant (g L
1) a commercial beta-D-glucosidase (cellobiose) preparation from
for yeast Aspergillus niger and its suitability in the hydrolysis of lignocel-
Ki,g Inhibition constant of glucose (g L
1) lulose. Biotechnol Bioengin 28:14381442.
on cellulase Desai SG, Converse AO. 1997. Substrate reactivity as a function
of the extent of reaction in the enzymatic hydrolysis of
Ki,2 Inhibition constant of glucose (g L
1)
lignocellulose. Biotechnol Bioengin 56:650655.
on b-glucosidase Drissen RET, Maas RHW, van der Maarel MJEC, Kabel MA,
KL Langmuir adsorption constant (FPU L
1) Schols HA, Tramper J, Beeftink HH. 2007. A generic model
Km Michaelis constant for (g L
1) for glucose production from various cellulose sources by a
b-glucosidase commercial cellulase complex. Biocatal Biotransform
25(6):419429.
Krec Recalcitrance constant (
)
Ghose TK. 1987. Measurement of cellulase activities. Pure Appl
ms Maintenance requirement (g g
1 h
1) Chem 59:257268.
for yeast Groot WJ, Kraayenbrink MR, Van der Lans RGJM, Luyben
NA Avogadro’s number (mol
1) KCAM. 1993. Ethanol-production in an integrated fermenta-
(6.23 ×/1023) tion membrane system-process simulations and economics.
Bioproc Engineer 8(56):189201.
R Gas constant (8.3142) (J mol
1 ×/K
1)
Hatzis C, Riley C, Philippidis GP. 1996. Detailed material balance
t Time (h) and ethanol yield calculations for the biomass-to-ethanol
T Temperature (K) conversion process. Appl Biochem Biotechnol 57(8):443459.
v1 Production rate of cellobiose (mg ×/ L
1 h
1) Lynd LR, Wyman CE, Gerngross TU. 1999. Biocommodity
from cellulose by cellulase engineering. Biotechnol Progr 15(5):777793.
Mohagheghi A, Tucker M, Grohmann K, Wyman C. 1992. High
v2 Production rate of glucose (mg ×/ L
1 h
1)
solids simultaneous saccharification and fermentation of pre-
from cellobiose by treated wheat straw to ethanol. Appl Biochem Biotechnol
b-glucosidase 33(2):6781.
v3 Production rate of glucose (mg ×/ L
1 h
1) O’Brien DJ, Roth LH, McAloon AJ. 2000. Ethanol production by
from cellulose by cellulase continuous fermentation-pervaporation: a preliminary eco-
nomic analysis. J Membrane Sci 166:105111
v4 Production rate of biomass (g ×/ L
1 h
1)
Oliveira SC, De Castro HF, Visconti AES, Giudici R. 1999.
v5 Consumption rate of glucose (g ×/ L
1 h
1) Continuous ethanol fermentation in a tower reactor with
by yeast flocculating yeast recycle: scale-up effects on process perfor-
Yxg Anaerobic yield of cell (g g
1) mance, kinetic parameters and model predictions. Bioproc
mass on glucose Engineer 20(6):525530.
Philippidis GP, Hatzis C. 1997. Biochemical engineering analysis
DH Deactivation enthalpy (J mol
1)
of critical process factors in the biomass-to-ethanol technology.
mmax Maximum growth rate (h
1) Biotechnol Progr 13(3):222231.

Philippidis GP, Spindler DD, Wyman CE. 1992. Mathematical
modeling of cellulose conversion to ethanol by the simultaneous
saccharification and fermentation process. Appl Biochem Bio-
technol 34(5):543556.
Philippidis GP, Wyman C. 1992. Production of alternative fuels:
modeling of cellulosic biomass conversion to ethanol. In:
Vardar-Sukan F, Suha Sukan S, eds, Recent advances in
biotechnology (pp. 405411). Dordrecht: Kluwer Academic
Publishers.
Press WH. 2007. Numerical recipes: the art of scientific comput-
ing. New York: Cambridge University Press.
Sheehan J, Himmel M. 1999. Enzymes, energy, and the environ-
ment: A strategic perspective on the US Department of
Energy’s Research and Development Activities for Bioethanol
Biotechnol Progr 15:817827.
South CR, Hogsett DA, Lynd LR. 1993. Continuous fermenta-
tion of cellulosic biomass to ethanol. Appl Biochem Biotechnol
3940:587600.
South CR, Hogsett DAL, Lynd LR. 1995. Modeling simulta-
neous saccharification and fermentation of lignocellulose to
ethanol in batch and continuous reactors. Enzyme Microb
Technol 17(9):797803.
Stenberg K, Bollok M, Reczey K, Galbe M, Zacchi G. 2000.
Effect of substrate and cellulase concentration on simultaneous
This paper was first published online on iFirst on 9 December
2008.
SHF vs. SSF 35
saccharification and fermentation of steam-pretreated softwood
for ethanol production. Biotechnol Bioengineer 68:204210.
Takagi M, Abe S, Suzuki S, Emert G, Yata N. 1977. A method for
production of alcohol direct from cellulose using cellulase and
yeast. In: Proceedings of the Bioconservation Symposium,
Ghose T, ed. (pp. 551571). New Delhi. IIT.
Wingren A, Galbe M, Zacchi G. 2003. Techno-economic evalua-
tion of producing ethanol from softwood: Comparison of SSF
and SHF and identification of bottlenecks. Biotechnol Progr
19:11091117.
Wright JD, Wyman CE, Grohmann K. 1987. Simultaneous
saccharification and fermentation of lignocellulose: Process
evaluation. Appl Biochem Biotechnol 18:7590.
Wu Z, Lee YY. 1997. Inhibition of the enzymatic hydrolysis of
cellulose by ethanol. Biotechnol Lett 19:977979.
Wyman CE. 1994. Ethanol from lignocellulosic biomass-technol-
ogy, economics, and opportunities. Bioresource Technol 50:3
16.
Wyman CES, Grohmann K, Lastick SM. 1986. Simultaneous
saccharification and fermentation of cellulose with the yeast
Brettanomyces clausenii. Biotechnol Bioengineer 17:221238.
Yang B, Willies DM, Wyman CE. 2006. Changes in the enzymatic
hydrolysis rate of avicel cellulose with conversion. Biotechnol
Bioengineer 94:11221128.