Available online at www.sciencedirect.

com

Bioresource Technology 99 (2008) 4221–4228

Hydrothermal treatment of wheat straw at pilot plant scale using

a three-step reactor system aiming at high hemicellulose recovery,
high cellulose digestibility and low lignin hydrolysis
Mette Hedegaard Thomsen, Anders Thygesen *, Anne Belinda Thomsen
Biosystems Department, Risø National Laboratory, Technical University of Denmark P.O. Box 49, DK-4000 Roskilde, Denmark

Received 27 April 2007; received in revised form 23 August 2007; accepted 26 August 2007
Available online 23 October 2007

Abstract

A pilot plant (IBUS) consisting of three reactors was used for hydrothermal treatment of wheat straw (120–150 kg/h) aiming at co-
production of bioethanol (from sugars) and electricity (from lignin). The first reactor step was pre-soaking at 80 °C, the second extraction
of hemicellulose at 170–180 °C and the third improvement of the enzymatic cellulose convertibility at 195 °C. Water added to the third
reactor passed countercurrent to straw. The highest water addition (600 kg/h) gave the highest hemicellulose recovery (83%). With no
water addition xylose degradation occurred resulting in low hemicellulose recovery (33%) but also in high glucose yield in the enzymatic
hydrolysis (72 g/100 g glucose in straw). Under these conditions most of the lignin was retained in the fibre fraction, which resulted in a
lignin rich residue with high combustion energy (up to 31 MJ/kg) after enzymatic hydrolysis of cellulose and hemicellulose.
Ó 2007 Elsevier Ltd. All rights reserved.

Keywords: Lignocellulose; Pilot plant; Bioethanol; Combustion enthalpy; Enzymatic hydrolysis

1. Introduction lytic cleavage of its partially crystalline structure (Puls

Commercial bioethanol is mainly produced by fermen-
tation of monomeric sugars produced from starch with
for example maize as raw material (1. generation bioetha-
nol). The potential ethanol production can be further
increased by using lignocellulosic material such as wheat
straw as raw material (2. generation bioethanol). The
recent year’s extensive research in bioethanol processes
from lignocellulosic materials has brought the usage of this
structurally less accessible substrate closer to commerciali-
zation (Kabel et al., 2006; Thomsen et al., 2006; Zaldivar
et al., 2001). Straw and leaves from one seasonal agricul-
tural crops consist mainly of cellulose (30–40%), hemicellu-
lose (20–30%) and lignin (10–20%). The cellulose and
hemicellulose fractions can be enzymatic hydrolysed to
monomeric sugars after a pre-treatment aiming at hydro-
*
Corresponding author. Tel.: +45 4677 4279; fax: +45 4677 4122.
E-mail address: anders.thygesen@risoe.dk (A. Thygesen).
and Schuseil, 1993). A number of pre-treatment methods
are now available – dilute acid pre-treatment (Schell
et al., 2003), hydrothermal treatment (Thomsen et al.,
2006), and wet oxidation (Bjerre et al., 1996; McGinnis
et al., 1983). Hydrothermal treatment can be performed
without addition of chemicals and oxygen making it a
potential solution for pre-treatment of large quantities of
straw.
It was decided to scale up the hydrothermal treatment
process in the EU-project called co-production biofuel
(IBUS). Initially the process should be used for extraction
of potassium salts from the straw. Straw with a low potas-
sium content gives less corrosion of power plant boilers
when it is burned and is therefore more useful for combus-
tion. The resulting liquid extract containing extracted
hemicellulose sugars (xylose and arabinose) could then be
fermented to ethanol by C5-sugar fermenting microorgan-
isms. This pilot plant could treat max. 100 kg straw/h and
consisted of two reactors. In the first reactor the straw was

0960-8524/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.biortech.2007.08.054
4222 M.H. Thomsen et al. / Bioresource Technology 99 (2008) 4221–4228

soaked in water at 80 °C to soften it before treatment in the
second reactor with high temperature (190–200 °C). The
solid fraction obtained by the process was highly hydroly-
sable with cellulytic enzymes and therefore useful as raw
material for ethanol production, and a concept was formed
in which all sugars were utilised for ethanol production and
the energy-rich lignin residue from the SSF process was
used for combustion in the power plant. The drawback
of this process was that the hemicellulose recovery was
moderate due to sugar degradation. Fermentation inhibi-
tors like carboxylic acids and furfurals were thereby
formed, which made the biomass difficult and in some cases
impossible to ferment (Thomsen et al., 2006).
This study reports on the scale up of the process to
1000 kg/h. To increase the capacity, a new reactor was
added to the process resulting in three steps (step 1: soak-
ing, step 2: hemicellulose extraction, step 3: preparation
of fibre fraction for enzymatic hydrolysis). The potential
amount of ethanol production from the treated straw was
determined based on chemical composition of the straw
(cellulose and hemicellulose) and enzymatic convertibility
of the cellulose to glucose.
2. Methods
2.1. Raw material
pieces went through a pre-soaking step in a 10 m long
screw conveyer system (reactor 1). The temperature in this
step was 80 °C and the residence time was 20 min. The
straw was fed into the first hot reactor (reactor 2) by means
of a particle pump designed for movement of biomass into
high pressure systems. Reactor 2 was a screw conveyor that
was operated with varied temperature (170 °C and 180 °C)
and residence time (7.5 min and 15 min) in this study. The
conveyor moved the straw up through the reactor and
water from the following and final reactor (reactor 3) was
passed counter currently to obtain leaching of the straw.
From reactor 2 the straw was transported into reactor 3.
Reactor 3 was heated by addition of steam to the upper
part of the reactor to get a temperature of 195 °C and a res-
idence time of 3 min. Biomass was removed from reactor 3
by another particle pump and was used for production of
ethanol by SSF fermentation. Beside steam addition, four
levels of water addition to reactor 3 were tested (0, 200,
420 and 600 kg water/h). Condensated steam resulted in
water flow through the reactor steps when no fresh water
was added resulting in liquid flow between 300 and
1200 kg water/h increasing versus the water addition level
(Table 1). Samples of liquid from the extraction process
were taken in the bottom of reactor 2 and reactor 3. Sam-
ples of fibre fraction were taken at the end of reactor 3. The
raw straw samples were taken just after the debaler.

Wheat straw (Triticum aestivum L.) was grown and har-
vested after a drying period in Denmark during summer
2003. The straw was baled into 500 kg bales and stored
at ambient temperature.
2.2. Description of the three-step pre-treatment pilot plant
The three-step pilot plant was based on Sicco’s proprie-
tary design and was a complete system for pre-treating up
to 1000 kg/h of straw. The straw was fed to the de-baler
with a rate of 120–150 kg straw per hour. After the debaler
unit the straw went through a stone trap to remove sand,
stones, and metal parts. After debaling it was cut into
5 cm pieces in a commercial straw chopper. The straw
2.3. Analysis methods
The dry matter content in the fibre fraction from reactor
3 and the raw straw were determined by drying 100 g sam-
ple at 60 °C for 3 days. Afterward these samples were knife
milled to pass a 1 mm sieve and ready for further analysis.
Ash content in the fibre fractions and dry matter content in
the liquid fraction were determined by drying the samples
(0.5 g for solid milled sample and 5 ml for liquid sample)
at 105 °C overnight in crucibles. The dry weight was mea-
sured followed by incineration of the crucibles at 550 °C
for 3 h (Thygesen et al., 2005).
The composition of the raw and the pre-treated straw
was measured by strong acid hydrolysis of the carbohy-

Table 1
Flow of streams with dry matter into and out of the pilot plant using varied treatment parameters in reactor 2 and at varied water flow
Reactor 1 80 °C, 20 min 80 °C, 20 min 80 °C, 20 min 80 °C, 20 min 80 °C, 20 min
Reactor 2 180 °C, 15 min 180 °C, 15 min 170 °C, 7.5 min 180 °C, 7.5 min 180 °C, 15 min
Reactor 3 195 °C, 3 min 195 °C, 3 min 195 °C, 3 min 195 °C, 3 min 195 °C, 3 min
Water into R3 (kg/h) 0 200 420 420 600
Liquid flow out of R1, R2 and R3 (kg/h) 300 350 800 940 1200
Straw into debaler (kg DM/h) 123 123 96 115 121
Straw into debalera (kg water/h) 27 27 24 25 26
Water-straw ratio (kg/kg) 0 1.62 4.4 3.7 5.0
Fibre fraction out of R3 (kg DM/h) 91 86 63 72 74
Liquid fraction out of R2 (kg DM/h) 14 18 18 25 41
Dry matter out of plant (kg DM/h) 106 104 81 98 116
Dry matter removal 14% 14% 16% 15% 12%
a
The dry matter fraction of the straw was 80–82%.
 M.H. Thomsen et al. / Bioresource Technology 99 (2008) 4221–4228
drates. Dried and milled samples (160 mg) were treated
with 72% (w/w) H2SO4 (1.5 ml) at 30 °C for 1 h. The solu-
tions were diluted with 42 ml of water and autoclaved at
121 °C for 1 h. The hydrolysates were filtered, and the Kla-
son lignin content was determined as the weight of the filter
cake subtracted the ash content. The filtrates (5 ml) were
mixed with 0.50 g Ba(OH)2 Æ 8H2O and after 5 min, the
samples were centrifuged with approx. 3000g for 5 min.
The supernatant was analysed for sugars on HPLC. The
recovery of D-glucose, D-xylose and L-arabinose was deter-
mined by standard addition of sugars to samples before
autoclavation. The sugars were determined after separation
on a HPLC-system (Shimadzu) with a Rezex ROA column
(Phenomenex) at 63 °C using 4 mM H2SO4 as eluent and a
flow rate of 0.6 ml/min. Detection was done by a refractive
index detector (Shimadzu Corp., Kyoto, Japan). Conver-
sion factors for dehydration on polymerization was 162/
180 for glucose and was 132/150 for xylose and arabinose
(Kaar et al., 1991; Thygesen et al., 2005).
Carbohydrates in the liquid extract after pre-treatment
were both polymers and monomers, thus the samples were
hydrolysed using 4% (w/w) H2SO4 at 121 C for 10 min to
determine the total glucose, xylose and arabinose concen-
tration. The sulphate ions in 5 mL acidic filtrate were pre-
cipitated by 0.50 g Ba(OH)2 Æ 8H2O and the supernatant
was diluted 1:1 with 4 mM H2SO4. Glucose, xylose and
arabinose were quantified by HPLC as described above
(Thomsen et al., 2006).
Monomer sugars and acetic acid were measured by
adjusting the sample pH to 2–2.3 with H2SO4. The samples
were centrifuged with approx. 3000g for 10 min. The super-
natant was analysed for glucose, xylose, arabinose and ace-
tic acid by HPLC as described above.
The enzymatic conversion of the solid fraction was
evaluated with 2% DM in 50 mM sodium acetate pH 4.8
using Cellubrix L to obtain 30 FPU/g DM. The solid
fraction was not washed before it was dried, milled and
used in the enzymatic hydrolysis to avoid removal of water
extractable polysaccharides. The solid fibres were enzymat-
ically hydrolysed as milled samples to get a uniform parti-
cle size and thereby comparable hydrolysis time. After 24 h
with a temperature of 50 °C, the reaction was stopped by
centrifugation with approx. 3000g for 10 min. The reaction
was only performed for 24 h since this is used for prehydro-
lysis in SSF experiments (Thomsen et al., 2006). The con-
centrations of glucose, xylose and arabinose were
measured by HPLC as described above (Thygesen et al.,
2003).
2.4. Calculations including combustion enthalpy
Calculation of cellulose recovery, hemicellulose recovery
and hydrolysis yield were performed as outlined by Thom-
sen et al. (2006). The theoretical ethanol production based
on the pre-treatment yield of extracted sugar oligomers and
the enzymatic hydrolysis yield was calculated according to
Eq. (1).
4223
Theoretical ethanol production
¼ TSC  0:51 þ TSH  0:5 ð1Þ
TSC* = Total sugar from cellulose (after pre-treatment and
enzymatic hydrolysis with 95% efficiency which is usually
obtained by SSF fermentation at large scale) (g/100 g raw
material) (data not shown).
TSH* = Total sugar from hemicellulose (after pre-treat-
ment and enzymatic hydrolysis) (g/100 g raw material).
Hemicellulose is defined as the sum of xylan and arabinan.
The pre-treatment severity factor ðR00 Þ was calculated as
outlined by Kabel et al. (2007). This factor includes the
effect of temperature and time and pH obtained in the pro-
cess on the lignocellulose material. The factor is calculated
as the sum of severity in reactor 2 and reactor 3 based on
pH, residence time (t) and temperature (T).
 
pHR2 T R2  100
0
R0 ¼ 10  tR2  exp þ 10pHR3  tR3
14:75
 
T R3  100
 exp ð2Þ
14:75
The energy yield from the solid fraction was calculated as
the combustion enthalpy (Atkins, 1994). The value for glu-
cose was used for cellulose and hemicellulose (14.1 MJ/kg),
the value for phenol was used for lignin (31.1 MJ/kg) and
0 MJ/kg was used for ash. The value for phenol was ap-
plied for lignin since it is a complex network of phenylpro-
pane (Fan et al., 1982). The residual fraction contains
mainly acetate, pectin and protein with energy contents be-
low or similar to that for glucose. Due to uncertainty about
the composition of this residual fraction, the combustion
enthalpy was estimated as a broad interval: 10 ± 4 MJ/
kg. The reaction equations and the values used for the cal-
culation are shown below:
Phenol : C6 H5 OHðsÞ þ 7O2 ðgÞ ! 6CO2 ðgÞ þ 3H2 OðgÞ
Glucose : C6 H12 O6 ðsÞ þ 6O2 ðgÞ ! 6CO2 ðgÞ þ 6H2 OðgÞ
EPhenol ¼ ð6  ð394 kJ=molÞ þ 3  ð242 kJ=molÞ
ð165 kJ=molÞÞ=94 g=mol ¼ 31:1 MJ=kg ð3Þ
EGlucose ¼ ð6  ð394 kJ=molÞ þ 6  ð242 kJ=molÞ
ð1274 kJ=molÞÞ=180 g=mol ¼ 14:1 MJ=kg
ELignocellulose ¼ ðxcellulose þ xhemicellulose Þ  14:1 MJ=kg
þ xlignin  31:1 MJ=kg þ xresidual  ð10  4Þ MJ=kg
3. Results
The pre-treatment conditions were evaluated with
respect to recovery of sugars, extraction of sugars, enzy-
matic hydrolysis of cellulose, and recovery of lignin in
the solid fraction.
3.1. Dry matter balance in the pre-treatment process
The dry matter balance is done by including flow of
liquid out of reactor 2, the addition of straw to the de-baler
4224 M.H. Thomsen et al. / Bioresource Technology 99 (2008) 4221–4228

and the amount of solid fraction leaving reactor 3. The dry Table 2
matter flows are shown in Table 1. The amount of Concentrations of extracted sugars from cellulose and hemicellulose
(xylan + arabinan) after reactor 3 and after reactor 2 (g/l) including sugar
extracted dry matter increased versus the water addition on monomer form
from 14 kg DM/h with no water addition to 41 kg DM/h
Reactor 2 temperature °C 180 180 170 180 180
with 600 kg water/h due to the increased washing effect. Reactor 2 time min 15 15 7.5 7.5 15
The amount of fibre fraction out of reactor 3 thereby Water into R3 kg/h 0 200 420 420 600
decreased versus the water addition from 91 kg DM/h with ðlog R00 ¼ 0:2Þ 0.15 0.20 0.36 0.18 0.21
no water addition to 74 kg DM/h with 600 kg water/h. The Reactor 3
amount of dry matter was 12–16% lower in the streams out pH 3.3 3.3 3.7 3.6 3.6
of the plant than in the stream into the plant (Table 1). a
Glucan (DP P 1) g/l 2.2 2.1 0.9 2.2 0.8
That can be due to the fact that steady state was not yet Hemicellulose (DP P 1) g/l 6.9 6.2 13.0 17.9 9.4
reached or uncertainty on the analysis of dry matter con- Xylan fraction % 90 91 91 92 93
tent of the streams. Formation of gasses like CO2 and Arabinan fraction % 10 9 9 8 7
CH3COOH at the high temperature cannot be neglected Reactor 2
and can partly explain the decrease in dry matter content pH 3.8 3.7 4.1 4.0 4.3
during the process. Loss of dry matter due to leaking and Acetic acid g/l 3.6 3.9 0.9 1.0 1.0
loss of water and dry matter in the particle pump where Glucan (DP P 1) g/l 4.4 4.5 0.9 1.7 1.5
the treated straw leaves the pilot plant has also been Hemicellulose (DP P 1) g/l 20.4 22.9 8.2 10.0 12.6
observed and has subsequently been solved by inserting a Xylan fraction % 87 89 89 89 88
Arabinan fraction % 13 11 11 11 12
cyclone in the plant.
Monomer fraction of:
3.2. Composition of the solid and the liquid stream after the Glucan at DP = 1 % 12 15 12 7 10
Xylan at DP = 1 % 14 16 14 10 10
process Arabinan at DP = 1 % 45 49 37 34 22
The flow rate is included in Table 1 and the severity factor (R00 Þ is calcu-
The concentration of extracted hemicellulose sugars lated using Eq. (2) based on the pH value.
after flow through reactor 3 was lowest in the experiments a
DP = Degree of polymerisation.
with low water addition (0–200 kg water/h; 6–7 g/l) (Table
2). At the higher level of water addition higher concentra-
tions were obtained (9–18 g/l) which can be explained with ars was on monomer form (22–46%). It requires thereby
the higher hemicellulose concentration in the fibre fraction enzymatic hydrolysis to complete the hydrolysis to glucose
leaving reactor 3 (Table 3). The glucose oligomer concen- and xylose.
tration was low at the investigated conditions (1–2 g/l) Table 3 shows the composition of the fibre fraction leav-
due to the thermally stable structure of cellulose. ing reactor 3. The composition of the straw was 34% cellu-
After liquid flow through both reactor 2 and reactor 3, lose, 22% hemicellulose, 16% lignin, 6% ash and 22%
the concentration of extracted hemicellulose sugars was residual components with a dry matter content of 80–
20–23 g/l with low water flow so a very concentrated 82%. The solid fraction after reactor 3 had a dry matter
solution was obtained compared with high water addition content of 24–31%. The cellulose content increased with
(8–13 g/l) due to the dilution effect. The glucose oligomer increasing water flows from 46 g/100 g DM with
concentration was slightly higher after flow through reac- 0 kg water/h to 51 g/100 g DM with 600 kg water/h, due
tor 2 (1–5 g/l) and decreasing versus the water addition to increasing extraction of hemicellulose. The lignin con-
level. The fraction of arabinose in the extracted hemicellu- tent in the solid fraction was highest (33 g/100 g DM) with
lose was 7–10% after flow through reactor 3 and highest 0 kg water/h (steam pre-treatment) while the hemicellulose
with lowest water addition. The arabinose fraction content was low (5 g/100 g DM). This indicates that ther-
increased to 11–13% after flow through reactor 2. The mal degradation of hemicellulose occurs when low water
arabinose fraction is thereby slightly lower than in the flows are used which results in acid insoluble residue iden-
pretreated straw (12%; Table 3). tified as lignin. The arabinose fraction of the hemicellulose
The pH level in reactor 3 was low when low water addi- decreased during the pre-treatment to 3–6% at high water
tion was used (pH 3.3) compared to pH 3.6–3.7 when 420– addition while at low water addition all the arabinose
600 kg water/h was used due to the dilution effect. The pH was extracted and/or degraded.
level was higher in reactor 2 due to dilution caused by con-
densate of the added steam. The pH level varied between 3.3. Sugar balance in the pre-treatment process
pH 3.7 at 0 kg water/h and pH 4.3 at 600 kg water/h and
the acetic acid concentration was highest (3.3–3.9 g/l) at High cellulose recoveries were found in all the experi-
the low pH level (Table 2). The fraction of glucan sugars ments (Close to 100 g/100 g cellulose in straw) independent
on monomer form was 7–15% and highest at low pH val- on the water flow since cellulose is not degraded to a large
ues. The fraction of xylose sugars on monomer form was extent in the applied temperature range (170–195 °C). The
similar (10–16%) while a high fraction of the arabinose sug- reason for cellulose recoveries slightly above 100% at high
 M.H. Thomsen et al. / Bioresource Technology 99 (2008) 4221–4228 4225

Table 3
Chemical composition of the applied wheat straw and the solid fraction rejected after reactor 3
Reactor 2 temperature °C Straw before treatment 180 180 170 180 180
Reactor 2 time min 15 15 7.5 7.5 15
Water into R3 kg/h 0 200 420 420 600
Cellulose g/100 g DM 33.5 46.6 47.7 50.7 53.9 51.2
Hemicellulose g/100 g DM 22.4 4.8 4.8 13.4 9.1 12.0
Xylan fraction % 88 100 100 94 97 96
Arabinan fraction % 12 0 0 6 3 4
Klason lignin g/100 g DM 16.4 32.8 32.3 25.0 24.2 24.0
Ash g/100 g DM 5.8 3.8 3.3 3.1 2.9 3.3
Residual g/100 g DM 21.8 12.0 11.9 7.8 9.9 9.5
Combustion enthalpy MJ/kg DM 15.2 18.6 18.6 17.6 17.4 17.3
Dry matter g DM/100 g 80–82 26 31 24 28 24
The calculated combustion enthalpy and dry matter content are included.

water addition can be incomplete conversion of cellulose to
glucose in the acidic hydrolysis of untreated straw. The
hemicellulose recovery was lower (33–83 g/100 g hemicellu-
lose in straw) since it is less thermally stable than cellulose
(Thomsen and Schmidt, 1999). The hemicellulose recovery
was highest with 600 kg water/h added (83 g/100 g hemicel-
lulose in straw) and lowest with no water addition (33 g/
100 g hemicellulose in straw), which shows that water has
a protecting effect on hemicellulose. The reduced degrada-
tion of hemicellulose results in less formation of inhibitory
compounds like furfural from xylose and arabinose.
Fig. 1 shows the amounts of extracted cellulose and
hemicellulose determined as monomeric sugars after weak
acid hydrolysis. The cellulose extraction was low in all
cases (<4 g/100 g cellulose in straw) due to its crystalline
and thermally stable structure. The extracted fractions of
xylose based and arabinose based sugars were similar at
each pre-treatment condition level since they both originate
from hemicellulose. The hemicellulose extraction was sim-
ilar using 170 °C and 180 °C (27–29 g/100 g) while dou-
bling the retention time increased the extraction to 50 g/
100 g. This high amount of extracted hemicellulose is also
obtained due to the increased washing effect with the high
water addition (600 kg/h). The amount of extracted hemi-
cellulose was lowest with no water addition (steam treat-
ment; 18–24 g/100 g) probably due to insufficient washing
of the fibres and due to degradation of xylose and
arabinose.
3.4. Enzymatic hydrolysis of the solid fraction
Fig. 2 shows the sugar yield after enzymatic hydrolysis
of cellulose to glucose and hemicellulose to xylose and arbi-
nose, from the solid fraction leaving reactor 3. The enzy-
matic hydrolysis of the solid fraction showed the lowest
glucose yield in the experiment with highest water addition
(51 g/100 g glucose in straw). The glucose yield was higher
with no water addition (72 g/100 g glucose in straw). This
shows that the higher pre-treatment process severity
ðlog R00 ¼ 0:2Þ compared with 0.2 at higher water flow
results in more enzymatic convertible cellulose (Table 2).
The enzymatic hydrolysis of hemicellulose was lowest
(14 g xylose/100 g xylose in straw) in fibre fraction with
low hemicellulose content (5 g/100 g DM) due to low hemi-
cellulose recovery. The highest hydrolysis yield (27 g xy-
lose/100 g xylose in straw) was obtained in fibre fraction

100
Extracted polymer sugars in pretreatment

Glucose oligomers from cellulose

Xylose oligomers from xylan
Arabinose oligomers from arabinan
[g/100 g component in straw]

80

60

40

20

0
Reactor 2: 180 ºC, 15 min 180 ºC, 15 min 170 ºC, 7.5 min 180 ºC, 7.5 min 180 ºC, 15 min
Water additíon: 0 kg/h 200 kg/h 420 kg/h 420 kg/h 600 kg/h

Fig. 1. Extracted oligomers of glucose, xylose and arabinose from cellulose, xylan and arabinan, respectively obtained during pre-treatment of the straw.
The x-axis shows increasing water flow into reactor 3 and secondly increasing temperature and time in reactor 2.
4226 M.H. Thomsen et al. / Bioresource Technology 99 (2008) 4221–4228

100
Glucose Xylose Arabinose

Sugar yield after enzymatic hydrolysis

80

[g/100 g component in straw]

60

40

20

0
Reactor 2: 180 ºC, 15 min 180 ºC, 15 min 170 ºC, 7.5 min 180 ºC, 7.5 min 180 ºC, 15 min
Water addition: 0 kg/h 200 kg/h 420 kg/h 420 kg/h 600 kg/h

Fig. 2. Enzymatic hydrolyzed cellulose, xylan and arabinan to monomer sugars from dried fibre fraction from reactor 3. The x-axis shows increasing water
flow into reactor 3 and secondly increasing temperature and time in reactor 2.

with high hemicellulose content (13 g/100 g DM) (Table 3)
due to the high hemicellulose recovery with the higher
water flow (420 kg/h). Arabinose was not released from
hemicellulose by the enzymatic hydrolysis even though
there was 3–6% arabinose in the hemicellulose of the fibre
fraction after the pre-treatment at high water addition.
3.5. Theoretical ethanol yield
Fig. 3 shows the sum of sugars extracted by the pre-
treatment and monomeric sugars produced by enzymatic
hydrolysis of the remaining solid fraction leaving reactor
3. The total glucose yield was lowest in the experiment with
highest water addition (55 g/100 g glucose in straw). The
glucose yield was highest (75 g/100 g glucose in straw) with
no water addition, which is due to the high pre-treatment
severity (Table 2). The hemicellulose yield was lowest
(32 g/100 g xylose in straw) with no water addition and
highest (69 g/100 g xylose in straw) with the highest water
addition.
The theoretical ethanol production was calculated based
on conversion of the entire sugar yield in the process
including glucose from cellulose and xylose and arabinose
from hemicellulose (Fig. 4, Eq. (1)). This yield was highest
when the water addition was low (0–200 kg/h; 184–186 kg
ethanol/ton straw) due to the high enzymatic convertibility
of cellulose in this experiment. Using high water addition
(600 kg/h) the ethanol yield was also high (192 kg etha-
nol/ton straw) but in this case due to the high hemicellulose
recovery and extraction. These values should be compared
to the ethanol yield that is obtained if all cellulose and
hemicellulose in the straw was hydrolysed to monomeric
sugars (318 kg ethanol/ton straw).
Normally hydrolysis of cellulose in pretreated straw
using this pilot plant reach 95–100% yield if simultaneous
saccharification and fermentation with Bakeŕs yeast is per-

100

Glucose Hemicellulose
80

60
Total yield
(%)

40

20

0
Reactor 2: 180 ºC, 15 min 180 ºC, 15 min 170 ºC, 7.5 min 180 ºC, 7.5 min 180 ºC, 15 min
Water addition: 0 l/h 200 l/h 420 l/h 420 l/h 600 l/h

Fig. 3. Sum of pre-treatment extracted cellulose and hemicellulose from the straw and enzymatic hydrolyzed cellulose and hemicellulose to monomer
sugars. The x-axis shows increasing water flow into reactor 3 and secondly increasing temperature and time in reactor 2.
 M.H. Thomsen et al. / Bioresource Technology 99 (2008) 4221–4228 4227

Based on obtained yields after pretreatment and enzymatic hydrolysis

Based on 95% convertion of fibers (in SSF) and obtianed yields of hemicellulose

300

Theoretical ethanol production

250

(kg/ton straw)
200

150

100

50

0
Reactor 2:
180 ºC,15 180 ºC,15 170 ºC, 180 ºC, 180 ºC, 15 Max.
min min 7.5 min 7.5 min min straw
Water addition: 0 l/h 200 l/h 420 l/h 420 l/h 600 l/h

Fig. 4. Theoretical ethanol production from straw: The light-grey bars are based on total conversion of the extracted and enzymatic hydrolyzed cellulose
and hemicellulose to ethanol. The dark-grey bars are based on 95% conversion of cellulose in the straw (the yield usually obtained in SSF of fibres
(Thomsen et al., 2006)) and total conversion of the extracted and enzymatic hydrolyzed hemicellulose to ethanol. The black bar shows the ethanol yield if
all cellulose and hemicellulose in the raw straw is converted to ethanol.

formed since product inhibition is avoided (Thomsen et al.,
2006). Therefore the yield was also calculated based on
95% conversion of cellulose. This ethanol yield was calcu-
lated to 223 kg/ton straw with no water addition increasing
to 270 kg/ton with 600 kg water/h due to increasing hemi-
cellulose recovery.
3.6. Lignin recovery in the solid fraction and energy yield
With low water addition (0–200 kg/h) there was low
extraction of lignin due to the low washing effect and a
large fraction of the produced solid fraction was Klason
lignin (32%). The part of lignin that remained solid
decreased versus the water flow to 80% using 600 kg/h.
The solid fraction after pre-treatment had in this experi-
ment both a high content of cellulose that could be enzy-
matic hydrolysed to glucose (72 g/100 g glucose in straw)
and a high lignin content making it useful for heat produc-
tion by incineration. The combustion energy was calculated
using Eq. (3) and is presented in Table 3. The combustion
energy in the straw was 15 ± 0.5 MJ/kg. The solid fraction
after pre-treatment had generally higher combustion
energy (19 ± 0.5 MJ/kg) with no water addition in pre-
treatment and 17 ± 0.5 MJ/kg with 600 kg water/h. The
increase in energy content is due to increase in the lignin
fraction. The combustion energy for lignin is 31 MJ/kg
according to the calculation.
4. Discussion
A product of hemicellulose degradation is furfural,
which is formed by sugar dehydration during hydrothermal
treatment (Kabel et al., 2007) by an acid-catalyzed reaction
(Antal et al., 1991). The hemicellulose degradation was
thereby lower at the higher pH obtained at high water flow
(Table 2). Since acetic acid is formed in hydrothermal treat-
ment of straw more water addition result in more dilution
of the acid and thereby higher pH and less xylose degrada-
tion which was observed in this study. Kabel et al. (2007)
performed hydrothermal treatment of wheat straw at a
straw/water ratio of 1/10 g/g which is slightly more water
than in the experiments in this study at high water addi-
tion. Similar levels of xylan extraction (40%) with 40%
remaining xylan in the solid fraction were obtained at a
severity factor ðlog R00 Þ of 0.5. The resulting xylan recov-
ery was thereby 80% which is similar to the hemicellulose
recovery in this study at high water addition.
The glucose yield by enzymatic hydrolysis was higher in
the two-step process with 195 °C (60 g/100 g glucose in
straw) than in the three-step process (51 g/100 g glucose
in straw), which is presumably due to the lower water flow.
The process with no water addition resulted in high glucose
yield (72 g/100 g glucose in straw) and appears thereby sim-
ilar to the two-step process with 200 °C due to the higher
process severity (Table 2; Fig. 2). Higher temperatures were
not used since it results in low hemicellulose recovery
(Thomsen et al., 2006). The results further show that a
reduced hemicellulose content gives higher enzymatic con-
vertibility as in agreement with Kabel et al. (2007). It has
been shown by Kabel et al. (2007) and Vincken et al.
(1995) that xylan is adsorbed on cellulose and hinders the
action of cellulase enzymes. It has been found by Kabel
et al. (2006) that the enzymatic convertibility depends on
the enzyme profile used and the low convertibility at high
xylan content indicates that higher xylanase enzyme con-
tent can cause improvement. The arabinan content can also
4228 M.H. Thomsen et al. / Bioresource Technology 99 (2008) 4221–4228
limit the convertibility since the convertibility was highest
in samples with no arabinan content. Furthermore arabi-
nose was not released in the enzymatic hydrolysis (Fig. 2)
so the arabinase content appears to be low or absent.
A high ethanol yield requires both high hemicellulose
recovery and high enzymatic cellulose convertibility. That
was not obtainable in this study due to the low enzymatic
hydrolysis of hemicellulose to xylose and arabinose. The
estimated conversion of cellulose to ethanol on 95%
requires also highly digestible cellulose which was only
obtained using high severity and thereby more hemicellu-
lose degradation. A lower severity would however result
in lower formation of inhibitory compounds but lower cel-
lulose digestibility.
Lignin was retained in the solid fraction when no water
addition was used in the pre-treatment. Klason lignin is
measured as the acid insoluble residue excluding the ash
content. Insoluble degradation products from hemicellu-
lose will be measured as Klason lignin if these are not dis-
solved by the acidic hydrolysis. The lignin that mainly
covers the fibre surface did not seem to prevent enzymatic
hydrolysis of the cellulose which is in agreement with Kabel
et al. (2007).
5. Conclusion
The optimal reaction conditions for high glucose yield
after enzymatic hydrolysis was to avoid water addition to
reactor 3 and to use 195 °C and 3 min residence time in this
step. If high hemicellulose yield is desired the water addition
should be high (600 kg/h) and in reactor 2 the temperature
and time should be 180 °C and 15 min, respectively. The
water addition had a positive effect on the hemicellulose
extraction due to reduced degradation of xylose and arabi-
nose. The highest hemicellulose recovery was 83 g/100 g
hemicellulose in straw obtained with 600 kg water/h. The
lowest recovery was 33 g/100 g hemicellulose in straw
obtained from experiments with no water addition (steam
treatment). The highest enzymatic cellulose convertibility
for the solid fraction obtained by the pre-treatment was
found using no water addition in the process (72 g/100 g
glucose in straw) due to the high process severity. This solid
fraction was also very lignin rich (32%) presumably due to
degradation products from hemicellulose. It possessed a
high combustion enthalpy value (19 MJ/kg) compared to
wheat straw (15 MJ/kg). The high lignin content did not
seem to inhibit the cellulase enzymes due to the high cellu-
lose convertibility in the lignin rich material. However it
was not possible to reach both high hemicellulose recovery
and high cellulose convertibility at the same time.
Acknowledgements
The work was financially supported by EU-contract
ENK6-CT-2002-00650 and Danish project: PSO-F&U-
project 2006-1-6412. Børge Holm Christensen (Sicco K/S,
Denmark), Jan Larsen, Mai Østergaard Petersen, and Erik
Hedahl-Frank (DONG Energy, Denmark) are thanked for
collaboration on pre-treatment experiments on the pilot
plant, and Tomas Fernqvist and Ingelis Larsen (Risø Na-
tional Laboratory) are thanked for technical assistance
on analytical work.
References
Antal Jr., M.J., Leesomboon, T., Mok, W.S., Richards, G.N., 1991.
Mechanism of formation of 2-furaldehyde from D-xylose. Carbohyd.
Res. 217, 71–85.
Atkins, P.W., 1994. Physical Chemistry, fifth ed. Oxford University Press,
Oxford, pp. C8-C12.
Bjerre, A.B., Olesen, A.B., Fernqvist, T., Plöger, A., Schmidt, A.S., 1996.
Pretreatment of wheat straw using combined wet oxidation and
alkaline hydrolysis resulting in convertible cellulose and hemicellulose.
Biotechnol. Bioeng. 49, 568–577.
Fan, L.T., Lee, Y.-H., Gharpuray, M.M., 1982. The nature of lignocell-
ulosics and their pretreatments for enzymatic hydrolysis. Adv.
Biochem. Eng. 23, 158–187.
Kaar, W.E., Cool, L.G., Merriman, M.M., Brink, D.L., 1991. The
complete analysis of wood polysaccharides using HPLC. J. Wood
Chem. Technol. 11, 447–463.
Kabel, M.A., van der Maarel, M.J.E.C., Klip, G., Voragen, A.G.J.,
Schols, H.A., 2006. Standard assays do not predict the efficiency of
commercial cellulase preparations towards plant materials. Biotechnol.
Bioeng. 93, 56–63.
Kabel, M.A., Bos, G., Zeevalking, J., Voragen, A.G.J., Schols, H.A.,
2007. Effect of pretreatment severity on xylan solubility and enzymatic
breakdown of the remaining cellulose from wheat straw. Bioresour.
Technol. 98, 2034–2042.
McGinnis, G.D., Wilson, W.W., Mullen, C.E., 1983. Biomass pretreat-
ment with water and high-pressure oxygen. The wet-oxidation process.
Ind. Eng. Chem. Prod. Res. Dev. 22, 352–357.
Puls, J., Schuseil, J., 1993. Chemistry of hemicelluloses: relationship
between hemicellulose structure and enzymes required for hydrolysis.
In: Coughlan, M.P., Hazlewood, G.P. (Eds.), Hemicellulose and
Hemicellulases. Portland Press Research Monograph, pp. 1–27.
Schell, D.J., Farmer, J., Newman, M., McMillan, J.D., 2003. Dilute-
sulfuric acid pretreatment of corn stover in pilot-scale reactor –
Investigation of yields, kinetics, and enzymatic digestibilities of solids.
Appl. Biochem. Biotechnol. 105, 69–85.
Thomsen, A.B., Schmidt, A.S., 1999. Further development of chemical
and biological processes for production of bioethanol: optimisation of
pre-treatment processes and characterisation of products. Risø
National Laboratory, Risø R1110(EN), pp. 1–73.
Thomsen, M.H., Thygesen, A., Jørgensen, H., Larsen, J., Christensen,
B.H., Thomsen, A.B., 2006. Preliminary results on optimization of
pilot scale pretreatment of wheat straw used in coproduction of
bioethanol and electricity. Appl. Biochem. Biotechnol. 129–132, 448–
460.
Thygesen, A., Thomsen, A.B., Schmidt, A.S., Jørgensen, H., Ahring,
B.K., Olsson, L., 2003. Production of cellulose and hemicellulose-
degrading enzymes by filamentous fungi cultivated on wet-oxidised
wheat straw. Enzyme Microb. Technol. 32, 606–615.
Thygesen, A., Oddershede, J., Lilholt, H., Thomsen, A.B., Ståhl, K., 2005.
On the determination of crystallinity and cellulose content in plant
fibres. Cellulose 12, 563–576.
Vincken, J.-P., De Keizer, A., Beldman, G., Voragen, A.G.J., 1995.
Fractionation of xyloglucan fragments and their interaction with
cellulose. Plant Physiol. 108, 1579–1585.
Zaldivar, J., Nielsen, J., Olsson, L., 2001. Fuel ethanol production from
lignocellulose: a challenge for metabolic engineering and process
integration. Appl. Microbiol. Biotechnol. 56, 17–34.