Fuel 136 (2014) 349357

Contents lists available at ScienceDirect

Fuel
journal homepage: www.elsevier.com/locate/fuel

Alkaline hydrogen peroxide pretreatment, enzymatic hydrolysis

and fermentation of sugarcane bagasse to ethanol
Sarita C. Rabelo , Rafael R. Andrade 1, Rubens Maciel Filho, Aline C. Costa
School of Chemical Engineering, University of Campinas, UNICAMP, P.O. Box 6066, 13083-970 Campinas, SP, Brazil

h i g h l i g h t s

 Evaluation of sugarcane bagasse pretreated with hydrogen peroxide.

 Evaluation of necessity of particle size reduction.
 Inuence of increasing solids loadings in the pretreatment and hydrolysis stages.
 Performance evaluation of batch fermentation of pure hydrolysate.
 Evaluation of continuous fermentation of hydrolysate with sugarcane molasses.

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

Article history: The pretreatment of sugarcane bagasse with alkaline hydrogen peroxide was evaluated for second gen-
Received 10 March 2014 eration ethanol production via enzymatic hydrolysis and fermentation using Saccharomyces cerevisiae.
Received in revised form 11 July 2014 Factorial designs were used to determine the need for particle size reduction as well as to optimize pre-
Accepted 15 July 2014
treatment conditions and enzymes loadings in the hydrolysis. The inuence of increasing solids loadings
Available online 2 August 2014
in the pretreatment and hydrolysis stages was determined; batch fermentation of pure hydrolysate, as
well as continuous fermentation of hydrolysate concentrated with sugarcane molasses were performed.
Keywords:
Furthermore, mass balances were used to determine the mass of ethanol obtained by mass of raw bagasse
Sugarcane bagasse
Pretreatment
in different operational conditions. The pretreatment increased bagasse enzymatic digestibility without
Alkaline hydrogen peroxide the need for prior size reduction. In the optimal pretreatment (1 h, 25 C, 1.84 mL hydrogen peroxide/g
Enzymatic hydrolysis, Ethanol bagasse) and hydrolysis conditions (3.5 FPU/g bagasse of cellulase and 25 CBU/g bagasse of b-glucosi-
dase), 416.7 kg glucose/ton of raw bagasse were obtained. Fermentation of pure hydrolysate led to an
ethanol yield of 187.85 kg/ton of raw bagasse.
2014 Elsevier Ltd. All rights reserved.

1. Introduction trying to develop a cost-effective technology that leads to maxi-

Pretreatment is an important stage in the production of ethanol
from biomass. Different pretreatments lead to materials with dif-
ferent characteristics, which inuence the enzymatic hydrolysis
and fermentation steps. The appropriate choice of a pretreatment
can be determinant for the process economical and technical via-
bility and should be made considering aspects of the whole process
[1], including the energy input and output and the environmental
impact [2]. Thus, in the last years many researchers have been
Corresponding author. Present address: Laboratrio Nacional de Cincia e
Tecnologia do Bioetanol CTBE/CNPEM, Caixa Postal 6170, 13083-970 Campinas,
So Paulo, Brazil.
E-mail address: sarita.rabelo@bioetanol.org.br (S.C. Rabelo).
1
Present address: Department of Exact and Earth Sciences, Federal University of
So Paulo, UNIFESP, 09972-270 Diadema, SP, Brazil.
mum fermentable sugars recovery, with minimum inhibitors pro-
duction and energy input, low demand of post-pretreatment
processes and low costs for reactors, water and chemicals [3].
Alkaline hydrogen peroxide has been shown to be a good choice
for the pretreatment of lignocellulosic biomass [48], as it leads to
high glucose yields and can be carried out in conditions of moder-
ate temperature and pressure without acids, which leads to minor
inhibitors formation [9]. It is a typical environment-friendly agent
used for delignication in wood pulping processes [10] and leaves
no residues in the biomass, as it degrades in oxygen and water,
thus minimizing the need for waste treatment.
In this work the process of second generation ethanol produc-
tion from sugarcane bagasse was studied considering alkaline
hydrogen peroxide as the pretreatment agent. All the steps in the
study were performed using bagasse from a same harvesting as
the used in the work of Rabelo et al. [11], which evaluated lime

http://dx.doi.org/10.1016/j.fuel.2014.07.033
0016-2361/ 2014 Elsevier Ltd. All rights reserved.
350 S.C. Rabelo et al. / Fuel 136 (2014) 349357
as the pretreatment agent. The objective was to compare the
pretreatments considering their inuence in all the stages of the
production process. First, the need for particle size reduction
before pretreatment was assessed using a complete factorial
design. Then, a central composite design was performed to
determine the optimal values for time, temperature and peroxide
concentration in the pretreatment. The inuence of increasing
solids loading in the pretreatment was determined and the compo-
sition of bagasse pretreated at different solids loadings were
compared. The lignin in the pretreatment liquor was recovered
and characterized by differential scanning calorimetry and gel per-
meation chromatography. The optimization of enzymes loading in
hydrolysis was performed using a central composite design and the
inuence of increasing solids loading in the hydrolysis was
assessed. Finally, the hydrolysate was fermented using Saccharo-
myces cerevisiae in batch mode and the hydrolysate concentrated
with sugar cane molasses was fermented in continuous mode for
21 days using the same microorganism. Mass balances for the
process considering different process alternatives were performed
and compared to the obtained in the work of [11].
2. Materials and methods
2.1. Raw material
Fresh sugarcane bagasse, from a single harvest (2005/06), was
provided by the sugar plant Usina So Luiz Dedini S/A (Pira-
ssununga/SP, Brazil). The material was obtained by manual har-
vesting of burnt sugarcane and from the last milling resulting
after juice extraction without washing. It was dried at 45 C for
48 h, left for 48 h at room temperature, and stored into plastic
bags. The average dry matter content (DM) was 95.0%.
2.2. Particle size
In order to study the inuence of particle size on the release of
fermentable sugars, the bagasse was divided into two parts. One
part was used as it comes from the mill, without prior screening,
and presented highly heterogeneous particle sizes, with
85.59 2.89% of the mass presenting average particle size larger
than 1.397 mm (non-screened bagasse). The other part of the
material was screened (without grinding), and the portion which
passed a 12 mesh and was retained by a 60 mesh screen (average
particle size between 0.248 and 1.397 mm) was selected. Material
of ner mesh corresponded to sand mainly and was not used.
2.3. Pretreatment
Pretreatment was carried out in 250 ml asks in an orbital
shaker MA-832 (Marconi, Piracicaba, SP, Brazil), agitated at
150 rpm. 4.00 0.01 g of bagasse was used in each pretreatment
run, at a concentration of 4% (w/w) DM, except for the runs per-
formed to evaluate the inuence of solids concentration in the
pretreatment. Hydrogen peroxide concentration, temperature
and pretreatment time were varied according to factorial designs
as shown in the results and discussion section. After the pretreat-
ment step, the liquors were separated from the solid fraction by
ltration. They were then reserved for the precipitation and
recovery of lignin and determination of sugars (monomeric and
oligomeric fraction), furanic aldehydes, and organic acids concen-
trations. The solids fraction was washed several times for
removal of the water-soluble solids (WS) and used to determine
the water insoluble solids (WIS) recovery (pretreatment yield)
and chemical composition [12,13].
2.4. Batch enzymatic hydrolysis
The enzymatic hydrolysis of the washed material was per-
formed using 3.00 0.01 g substrate at solid concentration of
3.0% (w/w) DM (except for the tests to determine the inuence
of solids loading in hydrolysis conversion) in 250 ml asks incu-
bated in an orbital shaker MA-832 (Marconi, Piracicaba, SP, Brazil)
agitated at 100 rpm at 50 C. The pH was adjusted to 4.8 with
0.05 mol/L sodium citrate buffer. Initially, cellulase from Tricho-
derma reesei ATCC 26921 (SigmaAldrich Corporation, St. Louis,
MO, USA) was added at a loading corresponding to 3.5 FPU/g
pretreated bagasse and b-glucosidase from Aspergillus niger
(SigmaAldrich Corporation, St. Louis, MO, USA) was added at a
loading corresponding to 1.0 CBU/g pretreated bagasse. After the
evaluation of enzymes loading, experiments were performed at
the enzymes concentrations determined.
Cellulase activity was determined as lter paper units per mil-
liliter, as recommended by the International Union of Pure and
Applied Chemistry [14]. b-glucosidase activity was determined
through a solution of cellobiose 15 mmol/L and expressed in units
per milliliter (CBU/mL) [15]. Enzyme activity was 47.44 FPU/mL for
cellulases and 343.63 CBU/mL for b-glucosidase.
2.5. Fed-batch enzymatic hydrolysis
The hydrolysis started with 5% (w/w) DM (5.00 0.01 g) in
250 ml asks incubated in an orbital shaker MA-832 (Marconi,
Piracicaba, SP, Brazil) agitated at 100 rpm at 50 C. The pH was
adjusted to 4.8 with 0.05 mol/L sodium citrate buffer. Cellulase
and b-glucosidase were added at a concentration corresponding
to 3.5 FPU/g pretreated bagasse and 25 CBU/g pretreated bagasse,
respectively. After 6 and 12 h of reaction, 2.5 g of pretreated
bagasse and the amount of enzymes to maintain enzymatic loads
of 3.5 FPU/g and 25 CBU/g were added. The total solids concentra-
tion reached 10% (w/w) DM.
2.6. Fermentation
Hydrolysate obtained by fed-batch enzymatic hydrolysis of pre-
treated sugarcane bagasse (approximately 70 g/L glucose) was
used for ethanol batch fermentation. In addition, fermentation
was conducted using standard glucose (70 g/L) for comparison.
The microorganism was an industrial strain of S. cerevisiae pro-
vided by Santa Adlia ethanol plant, SP, Brazil. Fermentation was
performed in shaker at 34 C and 150 rpm.
Continuous fermentation was performed by using sugarcane
molasses to increase the hydrolysate sugars concentration
(sucrose + glucose) to around 70 g/L. The hydrolysate was from
hydrolysis performed at 3% (w/w) solids. Fermentation was con-
ducted in bioreactors Bioo III (New Brunswick Scientic Co., Inc.,
Edison, NJ) with 700 mL of working volume, stirred by two at blade
turbines, with six blades each, at 300 rpm. The ow rate was 0.6 mL/
min and the temperature 34 C. In order to evaluate the inuence of
adding hydrolysates in different proportions on the kinetics of
molasses fermentation, the fermentation was initiated with pure
molasses and in predetermined time intervals different percentages
(mass) of hydrolysate was added to the feeding medium.
2.7. Inoculum and culture media
The media composition for inoculum growth was 50 kg/m3 of
glucose, 5 kg/m3 of KH2PO4, 1.5 kg/m3 of NH4Cl, 0.7 kg/m3 of
MgSO4
7H2O, 1.2 kg/m3 of KCl, and 5 kg/m3 of yeast extract. The
inoculum growth was performed in erlenmeyers asks maintained
in a shaker at 30 C and 150 rpm for 24 h. Prior to fermentation, the
molasses sterilization was performed in an autoclave Phoenix
 S.C. Rabelo et al. / Fuel 136 (2014) 349357
(Araraquara, SP, Brazil) at 121 C for 20 min. The hydrolysate broth
was cold sterilized using a sterile system of membrane ltration
after centrifuging (Sorvall centrifuge) at 8000 rpm (15,810 g) for
20 min. The cellulose ester material membrane was 0.45 lm pore
diameter.
2.8. Characterization of raw and pretreated biomass
For compositional analysis, extractives, structural carbohy-
drates and lignin were analyzed in accordance with Sluiter et al.
[12,13]. Prior to the analysis the samples were nely ground in a
knife mill MA-630/1/E (Marconi, Piracicaba, SP, Brazil). Sugars
and organic acids, hydroxymethylfurfural (HMF) and furfural were
analyzed with an HPLC system (Waters Corporation, Milford, MA,
USA) equipped with a refractive index detector. Cellobiose, glucose,
xylose and arabinose were separated on a Sugar-Pak I column
(Waters Corporation, Milford, MA, USA) run at a ow rate of
0.5 mL/min at 70 C, with water as the eluent. Acetic acid, HMF,
furfural, lactic acid, formic acid and levulinic acid were separated
using an Aminex HPX-87H column (Bio-Rad Laboratories Inc.,
Hercules, CA, USA) at 65 C with 0.5 mL/min of H2SO4 at 5 mM as
the eluent.
2.9. Pretreatment and hydrolysis liquors composition
The pretreatment and hydrolysis liquors were analyzed by an
HPLC equipped with a refractive index detector, in accordance with
the National Renewable Energy Laboratory (NREL) standard
procedure [12,13]. In the case of pretreatment liquor, one part of
the ltrate was analyzed for monomeric sugars content and the
other part was used to determine the total sugars in the pretreated
liquor as monomers and oligomers by post-hydrolysis as described
elsewhere [16].
2.10. Lignin characterization
Lignin samples obtained by the pretreatment were analyzed for
moisture content in accordance with the NREL standard procedure
[12,13]. Samples of lignin were also analyzed by Differential
Scanning Calorimetry (DSC) (Mettler-Toledo, Columbus, OH, USA)
and had their molar weight distribution determined by gel perme-
ation chromatography (GPC) using a CLASS-LC10 data analyzer and
a series of 500, 103 and 104 PLGel columns (Shimadzu, Kyoto,
Japan).
Fig. 1. Pareto chart of effects.
351
2.11. Fermentation analysis
Yeasts concentrations were gravimetrically determined, after
centrifuging the media for 15 min at 3300 rpm (2690 g), washing
and drying the precipitate in an oven at 70 C.
The supernatant was used for determination of sugar (sucrose,
glucose, fructose and xylose), acetic acid, ethanol, furfural and
HMF concentrations in a high-performance liquid chromatograph
Varian model 9010 (Varian Inc. Scientic Instruments, Palo Alto,
CA). For sugar, acids and ethanol separation, an Aminex HPX-87-
H column (Bio-Rad Laboratories Inc., Hercules, CA) at 30 C and
eluent ow of 0.6 mL/min (degassed and ultrapure water with
pH adjusted to 1.4 with H2SO4) and refraction index detector was
used. Furfural and HMF were separated by a Nova-Pak C18 column
(Waters Co., Milford, MA), with an eluent ow of 0.8 mL/min. (ace-
tonitrile/water 1:8) and detected by UV at wavelength of 276 nm.
The column was maintained at 30 C.
3. Results and discussion
3.1. Inuence of different particle sizes
Feedstock particle sizing is a crucial point regarding the econ-
omy of second generation ethanol production because although
it usually affects hydrolysis glucose yields positively, it consumes
signicant amounts of energy. The need for particle size reduction
depends on the biomass and pretreatment applied, and pretreat-
ments that do not require particle reduction should be benecial
for process economy.
In this work the need for a particle size reduction before pre-
treatment was assessed by comparing the hydrolysis yields of
screened bagasse (average particle size between 0.248 and
1.397 mm) and not screened bagasse (approximately 85.59% of
the mass with average particle size larger than 1.397 mm).
A 24 + central point full factorial design was performed consid-
ering time, temperature, hydrogen peroxide concentration and
screening (or not screening) as factors. The range of factor values
was based on previous works [10]: pretreatment time from 6 to
24 h, temperature from 20 to 60 C and hydrogen peroxide concen-
tration from 1% to 5% (v/v) at pH 11.5. Glucose concentration (g/g
raw bagasse) obtained after enzymatic hydrolysis was the
response considered. As screening is a qualitative factor, the soft-
ware Statistica 7.0 (Statsoft, Inc., Tulsa, OK, USA), used to analyze
the data, performed a full factorial design for this factor at the
352 S.C. Rabelo et al. / Fuel 136 (2014) 349357
central point, resulting in two central points. These central points
were replicated three times.
Fig. 1 shows the Paretto chart of effects. It can be seen that
hydrogen peroxide concentration, temperature and screening are
signicant at the 95% condence level (p < 0.05), but neither the
pretreatment time nor its interactions with the other factors inu-
ence glucose concentration. Hydrogen peroxide concentration is
the factor with the higher inuence on glucose concentration and
its main effect is positive, which means that, in general, increasing
hydrogen peroxide concentration favors hydrolysis. However, the
interactions between temperature and peroxide concentration (2
by 3) and between peroxide concentration and screening (3 by 4)
are negative, which means that for high temperatures or screened
bagasse, high hydrogen peroxide concentrations decreases glucose
concentration obtained after hydrolysis. This is due to cellulose
solubilization in the pretreatment stage, which is undesirable, as
only the solid material goes to the hydrolysis step in the process
proposed in this work.
The response surfaces of glucose concentration after hydrolysis
as a function of the signicant factors are shown in Fig. 2 (the statis-
tical model was signicant and showed no lack of t in the analysis
of variance, ANOVA). It can be seen from the gures that the higher
glucose concentrations were obtained for high hydrogen peroxide
concentrations and that the inuence of temperature was weak,
although slightly higher glucose concentrations were obtained for
lower temperatures. This is because although the main effect of tem-
perature is positive, its interaction with hydrogen peroxide is nega-
tive and more signicant, as can be seen in Fig. 1.
It can be also noticed from Fig. 2 that higher glucose concentra-
tions were obtained for non-screened bagasse, as was expected,
since screening has a negative effect on glucose concentration
(see Fig. 1).
Many works in the literature argues that decreasing particle
size improves hydrolysis yields because it increases the specic
surface area [17]. In a previous work we have shown that screening
had a positive effect on glucose concentration when working with
lime pretreatment [11]. Liu et al. [18] have concluded that sugar
conversions and yields were higher during enzymatic hydrolysis
of steam exploded corn stover for larger biomass particles.
When compared to the work of Rabelo et al. [11], the present
results show that this analysis depends on the pretreatment used.
For alkaline hydrogen peroxide, the decrease in particle size caused
an increase in cellulose solubilization during the pretreatment,
which resulted in a lower mass of glucose recovered by mass of
raw biomass (mass before pretreatment) after hydrolysis. Lime
pretreatment [11] is milder and there was no increase in cellulose
solubilization during pretreatment, so the improvement in enzyme
accessibility for smaller particles led to higher conversions.
Fig. 2. Glucose concentration as a function of pretreatment time and temperature. (a) Non-screened bagasse. (b) Screened bagasse.
As the milling process is an expensive unit operation and should
provide a signicant improvement in enzymatic hydrolysis yield to
be justied, non-screened bagasse was chosen as the raw material
for the following steps.
3.2. Optimization of pretreatment conditions
In order to determine the pretreatment conditions that maxi-
mize glucose yields, a central composite design (CDD) with three
replicates in the central point was performed varying temperature
(C) and hydrogen peroxide concentration (%, v/v) in the pretreat-
ment. As pretreatment time was not signicant it was xed at 6 h.
Fig. 3a shows the percentage of delignication and glucose glo-
bal yield (g glucan hydrolyzed/100 g glucan in raw bagasse) after
hydrolysis for all the runs of the central composite design. The
results for the central point (45 C, 7% (v/v)) are the average of
three replicates (standard deviations of 3% for percentage of delig-
nication and 1% for global glucose hydrolysis yield). It can be seen
that, in general, the hydrolysis global yield increases with the del-
ignication achieved in the pretreatment stage, as can be seen in
Fig. 3b, which shows the percentage of hydrolysis global yield ver-
sus the percentage of delignication in the pretreatment. Other
authors have reported the same behavior for other biomasses
and pretreatments [19].
The data for glucose concentration (g/g raw bagasse), directly
calculated from the global yield, were analyzed using the software
Statistica 7.0 (Statsoft, Inc., Tulsa, OK, USA). Only the hydrogen per-
oxide concentration was signicant at the 90% condence level
(p < 0.1) (linear and quadratic terms), thus a model was proposed
to describe glucose concentration as a function of this factor. The
model was signicant at the 90% condence level according to
the ANOVA (data not shown) and was used to determine the opti-
mal concentration of hydrogen peroxide, which was 7.36% (v/v).
In order to evaluate if pretreatment time was really not
signicant, experiments varying this factor from 1 to 24 h and
maintaining the other factors constant (50 C and 7.36% (v/v) of
hydrogen peroxide) were performed and there was no signicant
inuence on hydrolysis global yields. In fact, during the pretreat-
ment there is a high rate of O2 production due to peroxide decom-
position that lasts approximately 45 min and after there is no
noticeable O2 formation. Thus, pretreatment time was xed at 1 h.
The inuence of temperature was further evaluated by varying
temperature from 25 C to 70 C while maintaining time and
peroxide concentration xed (1 h and 7.36%). No statistically
signicant inuence on hydrolysis global yield was detected and
temperature was xed at 25 C.
The concentration obtained at the determined conditions (1 h,
25 C and 7.36% (v/v) or 1.84 mL peroxide/g bagasse) is 0.38 g
 S.C. Rabelo et al. / Fuel 136 (2014) 349357
100
(a) Delignification Global yield
90
80
70
60
Global yield (%)
%
50
40
30
20
10
0 50
Fig. 3. (a) % Delignication and glucose global yield after hydrolysis (g glucan hydrolyzed/100 g glucan raw bagasse) as functions of the conditions of pretreatment
(temperature and hydrogen peroxide concentration). (b) % of global yield versus % delignication.
glucose/g raw dry bagasse, corresponding to a global enzymatic
yield of 84% (g glucan hydrolyzed/100 g glucan in raw bagasse). This
yield was obtained after hydrolysis performed with 3% (w/w) solids
and enzymes loadings of 3.5 FPU/g pretreated material and 1.0 CBU/
g pretreated material of cellulase and b-glucosidase, respectively.
Other authors have pretreated different biomasses with alkaline
hydrogen peroxide achieving high enzymatic yields. Yamashita
et al. [6] pretreated bamboo with 1.0 mL hydrogen peroxide/g
bamboo at pH 11.4 and reached global enzymatic yield of 79.2%
after hydrolysis at 3% solids concentration and cellulase loading
of 20 FPU/g substrate. Correia et al. [4] pretreated cashew apple
bagasse with 0.86 mL peroxide/g biomass at pH 11.5 and achieved
70% of global enzymatic yield when the hydrolysis was performed
at approximately 2% solids and 30 FPU/g cellulose.
A mass balance for the pretreatment at the optimal conditions
is shown in Fig. 4. It can be seen that 92.6% of the cellulose is recov-
ered at the solid fraction and that 71.4% of the hemicelluloses and
73.5% of the lignin are recovered in the pretreatment liquor. The
pretreatment yield is 45.6 g pretreated solid/100 g raw bagasse.
The same gure shows the total mass balance, including enzymatic
hydrolysis in conditions different from the used in this rst evalu-
ation, and fermentation. These results will be discussed in the next
sections of the paper.
3.3. Characterization of the lignin recovered from the pretreatment
liquor at the optimal conditions
H2SO4 1 mol/L was added to the pretreatment liquor obtained
at the optimal conditions (1 h, 25 C and 7.36% (v/v) hydrogen
peroxide) until pH 2 to precipitate lignin. The lignin was washed
with water, dried at room temperature and analyzed. The mass
of lignin precipitated was of 14.73 0.016 g/100 g bagasse with
19.43 0.12% of humidity. The degradation of lignin isolated from
alkaline hydrogen peroxide pretreatment of sugarcane bagasse
occurs at 400.6 C and its enthalpy is 52.5 J/g. Gel permeation
chromatography (GPC) analysis of the isolated lignin was also
performed. The measured average molecular weight (Mw) was
952, the number average molecular weight (Mn) was 502 and the
polydispersity index (PDI), dened as Mw/Mn was 3.88.
3.4. Inuence of the solids loading in the pretreatment
Solids concentration has a strong inuence on the process, in the
economical and sustainable aspects, as it determines the size of the
pretreatment reactor and water usage. The results of the previous
sections considered a solid concentration of 4% (w/w) in the pre-
treatment step, and in this section a study on increasing solids load-
ing was performed. Fig. 5a shows the percentage of delignication
353
(b)
90
85
80 y = 0,911x + 2,7009
R = 0,911
75
70
65
60
55
45
40
45 50 55 60 65 70 75 80 85
Delignification (%)
and cellulose and hemicelluloses solubilization in the pretreatment,
as well as the glucose global hydrolysis yield (g glucan hydrolyzed/
100 glucan in raw biomass) as a function of solids loading in the pre-
treatment. It can be seen that in increasing the solids loading (from
4% to 15%) and maintaining pretreatment time, temperature and
hydrogen peroxide in the optimal values (1 h, 25 C and 7.36% (v/
v)), the solubilization of lignin and hemicelluloses decreases (from
86.8% to 65.9% and from 77.4% to 41%, respectively), resulting in a
decrease in the glucose hydrolysis global yield, from 83.9% to
64.8%. The maximum cellulose solubilization in the pretreatment
was of 5.5% when the solids concentration was 4% or 5%. The same
behavior for the solubilization of hemicelluloses and lignin was
observed when the same biomass (sugarcane bagasse from the
same harvest) was submitted to lime pretreatment [11], but for
lime pretreatment increasing solids loading in the pretreatment
had no signicant effect on hydrolysis yield.
When comparing the two pretreatments, alkaline hydrogen
peroxide is more effective than lime, as the maximum glucose
global hydrolysis yield obtained after lime pretreatment (53.7%,
4% solids) was lower than the minimum global yield obtained after
alkaline hydrogen peroxide pretreatment (65.9%, 15% solids). In
fact, the composition of bagasse after hydrogen peroxide pretreat-
ment at 15% solids, was the same as the composition of bagasse
after lime pretreatment in the optimal conditions at 5% solids, as
can be seen in Fig. 5b. Even with the same composition, the hydrogen
peroxide pretreated bagasse was more susceptible to enzymatic
hydrolysis, achieving 65.9% global glucose hydrolysis yield. Lime
pretreated bagasse (5% solids) reached global glucose yield of 50.3%
when hydrolysis was performed under the same conditions.
3.5. Inuence of enzymes loading
The results of the previous section were obtained with low
enzymes loading (3.5 FPU/g pretreated material of cellulase and
1.0 CBU/g pretreated material of b-glucosidase). In order to evalu-
ate the inuence of varying enzymes loading, a central composite
design (CDD) with three replicates in the central point was per-
formed varying the loads of cellulase (FPU/g pretreated bagasse)
and b-glucosidase (CBU/g pretreated bagasse) and the responses
considered were glucose (g glucan hydrolyzed/100 g glucan in
the pretreated bagasse) and xylose (g xylan hydrolyzed/100 g
xylan in the pretreated bagasse) conversions (%) in the hydrolysis.
Pretreatment was performed with 4% (w/w) solids and hydrolysis
with 3% (w/w) solids. The responses for all the runs are shown in
Fig. 6a. The run corresponding to the central point (35 FPU/g
pretreated material of cellulase and 25 CBU/g pretreated material
of b-glucosidase) was performed in triplicate and the results
shown are the average. Standard deviations were of 0.31% for
354 S.C. Rabelo et al. / Fuel 136 (2014) 349357

Fig. 4. Mass balances for pretreatment at the optimal conditions, hydrolysis at different solids concentrations and enzymes loadings and ethanol fermentation.

100
90 Without 5% 15% Alkaline
Bagasse
80 pretreatment Ca(OH)2 H2O2
70 Mass 100.0 63.6 62.3
60 Ashes 3.8 0.1 - -
50
Extractives 2.6 0.1 - -
%

Total Lignin 25.8 1.6 8.9 0.2 8.8 0.1

40
Cellulose solubilization Cellulose 39.6 0.9 39.6 0.4 39.7 0.6
30
Hemicellulose solubilization Hemicellulose 23.9 0.3 14.4 0.1 14.1 0.5
20 Delignification
Global hydrolysis yield Total mass 95.4 1.8 62.9 0.7 62.6 1.2
10
0
4 5 6 7 8 9 10 11 12 13 14 15 16
% Solids in the pretreatment

Fig. 5. (a) Variation of the percentage of cellulose solubilization, delignication, hemicelluloses solubilization and global yield as a function of solids loading in the
pretreatment. (b) Comparison of bagasse composition after alkaline hydrogen peroxide pretreatment at 15% solids and lime pretreatment at 5% solids, both at the determined
optimal conditions. Lime results are from Rabelo et al. [11].

glucose conversion and 0.53% for xylose conversion. It can be seen
that glucose conversion in the hydrolysis is between 90% and 100%
for all the enzyme loadings considered. Xylose conversion varied
from 48% to 99%.
A model was proposed to describe glucose conversion as a
function of cellulase and b-glucosidase loadings, but it presented
evidence of lack of t (F lack of t > F listed value) (data not shown).
When the model was used to determine the optimal enzymes
loadings, the values obtained were 3.5 FPU/g pretreated material
of cellulase and 25 CBU/g pretreated material of b-glucosidase. As
the model was not considered accurate, new experiments were
performed varying cellulase concentration from 1 to 12.7 FPU/g
pretreated material, without the addition of b-glucosidase and with
25 CBU/g pretreated material of b-glucosidase. The results are
shown in Fig. 6b. It can be seen that 100% glucose conversion was
obtained using 25 CBU/g pretreated material of b-glucosidase and
cellulase loadings from 3.5 FPU/g pretreated material.
Fig. 4 shows the mass balances for the hydrolysis considering
hydrolysis at 3% (w/w) solids, 3.5 FPU/g pretreated material of cel-
lulase and 25 CBU/g pretreated material of b-glucosidase. Although
all the glucan in the pretreated bagasse was converted to glucose,
only 36.6% of the xylan was converted to xylose. If high xylose con-
version is desirable, the cellulase concentration of 12.7 FPU/g pre-
treated material should be used, in which case 98.7% of the xylan in
the pretreated bagasse is converted to xylose. This condition is also
shown in Fig. 4.
 S.C. Rabelo et al. / Fuel 136 (2014) 349357
(a)
Fig. 6. (a) Glucose and xylose conversions with varying cellulase and b-glucosidase loadings. (b) Glucose conversion with varying cellulase loadings with and without
addition of b-glucosidase from A. niger.
To calculate the global glucose hydrolysis yield (g glucan hydro-
lyzed/100 g glucan in raw biomass), the mass of glucose obtained
after hydrolysis (41.67 g) has to be compared to the mass of glu-
cose in the raw biomass (40.5 (180.2/162.2) = 45 g glucose). Table 1
shows the global hydrolysis yields and the overall yields, which
considers also the sugars recovered from the pretreatment liquor,
for glucose and xylose. The calculations for xylose yields use a
stoichiometric factor of 132.1 g xylan/150.1 g xylose.
The glucan yields in Table 1 are higher than the obtained for the
hydrolysis of the same biomass pretreated with lime [11], in which
case the maximum hydrolysis global yield was 87.21% and the
maximum overall yield was 91.16% using a much higher cellulase
loading of 50 FPU/g bagasse and the same b-glucosidase loading.
The pretreatment with lime, however, resulted in higher xylan
yields. The minimum xylan hydrolysis global yield for lime pre-
treated bagasse was 43.08%, with corresponding overall yield of
94.51% for enzymatic loads of 12.7 FPU/g pretreated material of
cellulase and 7.3 CBU/g pretreated material of b-glucosidase.
3.6. Inuence of the solids loading in hydrolysis
Low solids loading in hydrolysis leads to low sugar concentra-
tion in the hydrolysis liquor and, consequently, to low ethanol
concentration in the fermentation, which increases the costs of dis-
tillation. However, increasing solids loading impacts on hydrolysis
glucose yields.
In the previous sections the hydrolysis solids concentration was
3% (w/w). In this section the inuence of increasing solids loading
was evaluated by varying solids loading from 1% to 10% (w/w).
Enzymes loadings were of 3.5 FPU/g pretreated material of cellu-
lase and 25 CBU/g pretreated material of b-glucosidase. From 1%
to 5% of solids the hydrolysis was performed in batch mode and
the 10% solids loading was achieved by fed-batch hydrolysis, start-
ing with 5% of solids and feeding bagasse and enzymes at 6 and
12 h.
Fig. 7a shows the concentration of glucose as a function of time
in the hydrolysis of alkaline hydrogen peroxide pretreated bagasse.
The proles of glucose for batch hydrolysis of lime pretreated
bagasse at 10% (w/w) solids and fed-batch hydrolysis of lime pre-
treated bagasse (20% total solids) were added for comparison [11].
The hydrolysis of lime pretreated bagasse was performed at the
Table 1
Enzymatic yields and overall yields for different enzymatic loads. Hydrolysis with 3% (w/w) solids, pretreatment at 25 C, 1 h, 7.36% (v/v) hydrogen peroxide, pH = 11.5, 4% solids.
Enzymatic loading Hydrolysis global yields (%)
g hydrolyzed/100 g raw bagasse
Glucan
3.5 FPU/g, 25 CBU/g 92.59
12.7 FPU/g, 25 CBU/g 92.59
355
(b) 100
Glucose conversion (%)
80
60
40 Without -glucosidase
25 CBU/g bagasse
20
0
0 5 10 15
Cellulase (FPU/g bagasse)
optimal enzymatic loadings determined for this pretreated mate-
rial (50 FPU/g pretreated material of cellulase and 25 CBU/g pre-
treated material of b-glucosidase) [11].
Comparing the results of fed-batch hydrolysis in Fig. 7a it can be
seen that even though the cellulase load used for lime pretreated
bagasse was more than 14 times higher (50 FPU/g pretreated
material versus 3.5 FPU/g pretreated material), two times the mass
of lime pretreated bagasse (20%) was needed in comparison to per-
oxide pretreated bagasse (10%) to reach virtually the same nal
glucose concentration at the end of the hydrolysis (around 70 g/
L). This is because peroxide pretreated bagasse has more cellulose
in its composition (82.2%) than lime pretreated bagasse (65.9%)
and is more susceptible to hydrolysis (the glucose yield obtained
in the hydrolysis of pretreated biomass with same composition
was higher for peroxide pretreated bagasse, as shown in the Inu-
ence of Solids Loading in the Pretreatment section). Also, the
hydrolysis conversion of peroxide pretreated bagasse is less
affected by the increase in solids concentrations, as can be seen
in Fig. 7b. Increasing solids loading from 1% to 5% in batch hydro-
lysis decreased glucose conversion by 15.7% for peroxide pre-
treated bagasse, and 34.0% for lime pretreated bagasse.
These results show that analysis of the whole process should be
performed before choosing among pretreatments. In this case,
although alkaline hydrogen peroxide pretreatment is more expen-
sive than lime pretreatment, the need for much less enzyme and
the possibility of using half the solids load to achieve the same glu-
cose concentration are enormous advantages. It is well known that
enzymes prices are a barrier to the economic viability of second
generation ethanol production process and that the operation with
high solids loads increase energy costs with stirring and causes
heat transfer difculties and this should be taken into account in
an economic analysis.
The decrease in conversion with increased solids loadings has
been reported for the hydrolysis of many biomasses submitted to
various pretreatments [20,21]. This decrease has been attributed
to end product inhibition [22], a decline in adsorption capacity at
high solids concentrations [21,23,24], among others. In fact, the
decreased adsorption capacity at high solids loadings is probably
caused by mass transfer limitations. End product inhibition appar-
ently play a role in decreasing conversion, as cellobiose concentra-
tion increased with higher solids loading despite a proportional
Overall yields (%)
g hydrolyzed/100 g raw bagasse
Xylan Glucan Xylan
8.07 95.31 79.50
21.75 95.31 93.18
356
Fig. 7. (a) Glucose concentrations versus time and (b) glucose conversion (g glucan hydrolyzed/100 g pretreated glucan in bagasse) with varying solids loading for bagasse
pretreated with lime or alkaline hydrogen peroxide. Data for lime pretreated bagasse is from Rabelo et al. [11].
increase in b-glucosidase concentration, for peroxide pretreated
bagasse (data not shown) as well as lime pretreated bagasse (data
shown in Rabelo et al. [11]). This is due to the high glucose concen-
trations attained, which inhibit b-glucosidase. However, end prod-
uct inhibition is not the only cause, as the decrease in conversion
was more pronounced for lime pretreated bagasse, even though
lower glucose concentrations were obtained in the hydrolysis of
this material for a given solid load.
Fig. 4 shows the mass balance considering hydrolysis at 10%
solids concentration and starting from 100 g of raw dry bagasse.
30.88 g of glucose are obtained from 100 g of raw bagasse, thus
to reach the concentration of 70 g/L in Fig. 7a, 219.30 g/L of raw
bagasse were needed. This is much lower than the concentration
of raw bagasse needed to reach 70 g/L of glucose in the hydrolysis
of lime pretreated bagasse, which was 338.98 g/L (Rabelo et al.
[11]).
3.7. Batch fermentation of pure hydrolysate
The hydrolysate of alkaline hydrogen peroxide pretreated
bagasse was fermented by S. cerevisiae. A fed-batch hydrolysis
until 10% (w/w) solids was performed to reach high sugar con-
centration. For comparison, a synthetic medium of glucose was
also fermented in the same conditions. The composition of the
hydrolysate was 67.74 g/L glucose, 8.94 g/L xylose and 2.96 g/L
acetic acid. There was no HMF or furfural in the hydrolysate,
(a) 80
70
60
Glucose, Ethanol (g/L)
Pure Glucose
Peroxide hydrolysis liquor
50
40
30
20
10
0 0
0 5 10 15 20
Time (h)
Fig. 8. (a) Batch fermentation of pure hydrolysate. (b) Continuous fermentation of hydrolysate concentrated with molasses for different hydrolysate percentages in the
fermentation medium.
S.C. Rabelo et al. / Fuel 136 (2014) 349357
which is expected for pretreatments at mild conditions (low tem-
peratures, atmospheric pressure and no acid). The main inhibitor
present was acetic acid, which can totally inhibit yeasts growth in
concentrations from 1.4 to 8 g/L, depending on the microorgan-
ism lineage [25].
Fig. 8a shows the time proles of glucose and ethanol concen-
trations for both fermentations. The ethanol yield of the hydroly-
sate fermentation was of 88.4%, while for glucose fermentation
the yield was 89.2%. The high yield can be explained by the low
concentration of inhibitors in the hydrolysate.
A mass balance starting from 100 g of raw bagasse, performing
the pretreatment at 4% solids at the optimal conditions and hydro-
lysis considering different solids concentrations and enzymes load-
ings can be observed in Fig. 4. It can be seen that the highest
ethanol concentration obtained (for low solids hydrolysis) is
18.78 g ethanol/100 g raw bagasse, which corresponds to 41.20 g
ethanol/100 g raw pretreated bagasse. For high solids loading
13.92 g ethanol/100 g raw bagasse or 30.53 g ethanol/100 g
pretreated bagasse were obtained. Both results were obtained for
low cellulase loading in the hydrolysis (3.5 FPU/g pretreated
bagasse). Rabelo et al. [11] obtained 17.5 g ethanol/100 g raw
bagasse (29.7 g ethanol/100 g pretreated bagasse) for hydrolysis
at low solids loading and 8.6 g ethanol/100 g raw bagasse (14.6 g
ethanol/100 g pretreated bagasse) for high solids loading consider-
ing pretreatment with lime and cellulase loading of 50 FPU/g
pretreated bagasse in hydrolysis.
50
(b) no hydrolysate
5% 25% 35% 45% 60%
15%
40
Ethanol (g/L)
30
20
10
25 30 35 40 0 100 200 300 400 500
Time (h)
 S.C. Rabelo et al. / Fuel 136 (2014) 349357
Conde-Meja et al. [26] calculated the targets for bioethanol pro-
duction considering liquid hot water, steam explosion, dilute acid,
lime, AFEX and organossolv pretreatments and they varied from
229.75 L (or 181.27 kg) ethanol/ton dry biomass to 372.53 L (or
293.93 kg) ethanol/ton dry biomass considering ethanol production
from both glucose and xylose, with ethanol yield of 95% from glucose
and 60% from xylose, which may not be attained in the presence of
inhibitors present in the hydrolysate. In the present work we
obtained 187.85 kg ethanol/ton raw bagasse only from glucose
and considering a more realistic ethanol yield of 88.4%, obtained
experimentally. This value, however, was obtained considering
low solids load in hydrolysis. Increasing the solids loading in hydro-
lysis to 10%, 139.22 kg ethanol/ton raw biomass was obtained.
3.8. Continuous fermentation of hydrolysate concentrated with
molasses
Considering the extreme decrease in conversion when the
hydrolysis is performed with high solids loading and the need for
high ethanol concentration to minimize separation costs, one
alternative is to integrate the second generation ethanol produc-
tion process to the rst generation process by using molasses to
concentrate the hydrolysate. The results of adding hydrolysate to
a fermentation initiated using pure molasses is shown in Fig. 8b.
It can be seen that the ethanol concentration was not affected by
increasing hydrolysate percentage up to 60% in the feeding
medium even for high operation times (around 21 days).
4. Conclusion
The pretreatment of sugarcane bagasse signicantly increased
its enzymatic digestibility without the need for prior size reduc-
tion. The optimal conditions determined for the pretreatment were
1 h, 25 C, 1.84 mL hydrogen peroxide/g bagasse (7.36% (v/v)
peroxide, 4% solids) at pH 11.5. Although the concentration of per-
oxide to obtain the highest global glucose yield (92.59%) was high,
the cellulase loading in hydrolysis was low (3.5 FPU/g pretreated
material). The lowest hydrogen peroxide concentration evaluated,
of 0.49 mL peroxide/g bagasse (7.36% (v/v) peroxide, 15% (w/w)
solids), led to a global glucose yield of 65.9% after hydrolysis.
The optimal cellulase load for the hydrolysis of alkaline hydro-
gen peroxide pretreated bagasse (3.5 FPU/g pretreated material)
was much lower than for lime pretreated bagasse (50 FPU/g
pretreated material) [11] and the global glucose yield higher
(92.59% for peroxide and 87.21% for lime), which shows that
alkaline hydrogen peroxide is a more effective pretreatment.
Increasing solids loading in hydrolysis increases sugars concen-
trations in the hydrolysis liquor, but decreases hydrolysis conver-
sions. The decrease in conversion is less severe than for lime
pretreatment. When lime pretreated bagasse loading was
increased from 1% to 10% there was a decrease of 41.6% in conver-
sion and when peroxide pretreated bagasse was considered the
decrease was of only 26.3%.
The lower decrease of conversion for high solids loading and the
higher digestibility of peroxide pretreated bagasse allowed to
obtain high glucose concentrations in hydrolysis with half the
solids load (when compared with lime pretreatment), which
improves process economics by minimizing stirring energy costs
and raw material utilization.
The fermentability of the hydrolysate from peroxide pretreated
sugarcane bagasse was as good as that of lime pretreated bagasse,
achieving ethanol yield similar to the obtained by fermenting
glucose in the same conditions. The main inhibitor present in the
hydrolysate was acetic acid.
357
Acknowledgment
Authors acknowledge CNPq and Fapesp (project 07/01525-9)
for nancial support.
References
[1] Tao L, Aden A, Elander RT, Pallapolu VR, Lee YY, Garlock RJ, et al. Process and
technoeconomic analysis of leading pretreatment technologies for
lignocellulosic ethanol production using switchgrass. Bioresour Technol
2011;102(24):1110514.
[2] Barakat A, Chuetor S, Monlau F, Solhy A, Rouau X. Eco-friendly dry chemo-
mechanical pretreatments of lignocellulosic biomass: impact on energy and
yield of the enzymatic hydrolysis. Appl Energy 2014;113:97105.
[3] Hu F, Ragauskas A. Pretreatment and lignocellulosic chemistry. Bioenerg Res
2012;5(4):104366.
[4] Correia JADC, Marques Jnior JE, Gonalves LRB, Rocha MVP. Alkaline hydrogen
peroxide pretreatment of cashew apple bagasse for ethanol production: Study
of parameters. Bioresour Technol 2013;139:24956.
[5] Ayeni AO, Hymore FK, Mudliar SN, Deshmukh SC, Satpute DB, Omoleye JA, et al.
Hydrogen peroxide and lime based oxidative pretreatment of wood waste to
enhance enzymatic hydrolysis for a biorenery: process parameters
optimization using response surface methodology. Fuel 2013;106:18794.
[6] Yamashita Y, Shono M, Sasaki C, Nakamura Y. Alkaline peroxide pretreatment
for efcient enzymatic saccharication of bamboo. Carbohyd Polym 2010;79(4):
91420.
[7] Karagz P, Rocha IV, zkan M, Angelidaki I. Alkaline peroxide pretreatment of
rapeseed straw for enhancing bioethanol production by same vessel
saccharication and co-fermentation. Bioresour Technol 2012;104:34957.
[8] Banerjee G, Car S, Liu T, Williams DL, Meza SL, Walton JD, et al. Scale-up and
integration of alkaline hydrogen peroxide pretreatment, enzymatic hydrolysis,
and ethanolic fermentation. Biotechnol Bioeng 2012;109(4):92231.
[9] Rabelo SC, Maciel Filho R, Costa AC. A comparison between lime and alkaline
hydrogen peroxide pretreatments of sugarcane bagasse for ethanol
production. Appl Biochem Biotech 2008;144(1):87100.
[10] Mou HY, Heikkil E, Fardim P. Topochemistry of alkaline, alkaline-peroxide
and hydrotropic pretreatments of common reed to enhance enzymatic
hydrolysis efciency. Bioresour Technol 2013;150:3641.
[11] Rabelo SC, Maciel Filho R, Costa AC. Lime pretreatment and fermentation of
enzymatically hydrolyzed sugarcane bagasse. Appl Biochem Biotech 2013;169(5):
1696712.
[12] Sluiter A, Hames B, Ruiz R, Scarlata C, Sluiter J, Templeton D, et al.
Determination of structural carbohydrates and lignin in biomass
determination of structural carbohydrates and lignin in biomass; 2011.
<http://www.nrel.gov/biomass/pdfs/42618.pdf>. [accessed 19.01. 14].
[13] Sluiter A, Hames B, Ruiz R, Scarlata C, Sluiter J, Templeton D. Determination of
sugars, byproducts, and degradation products in liquid fraction process samples;
2008. <http://www.nrel.gov/biomass/pdfs/42623.pdf>. [accessed 19.01. 14].
[14] Ghose TK. Measurement of cellulase activities. Pure Appl Chem 1987;59:
25768.
[15] Brethauer S, Studer MH, Yang B. The effect of bovine serum albumin on batch
and continuous enzymatic cellulose hydrolysis mixed by stirring or shaking.
Bioresour Technol 2011;102(10):62958.
[16] Jacobsen SE, Wyman CE. Xylose monomer and oligomer yields for uncatalyzed
hydrolysis of sugarcane bagasse hemicellulose at varying solids concentration.
Ind Eng Chem Res 2002;41(6):145461.
[17] Khullar E, Dien BS, Rausch KD, Tumblesona ME, Singha V. Effect of particle size on
enzymatic hydrolysis of pretreated Miscanthus. Ind Crop Prod 2013;44:117.
[18] Liu ZH, Qin L, Pang F, Jin MJ, Li BZ, Kang Y, et al. Effects of biomass particle size
on steam explosion pretreatment performance for improving the enzyme
digestibility of corn stover. Ind Crop Prod 2013;44:17684.
[19] Yu Z, Jameel H, Chang HM, Park S. The effect of delignication of forest
biomass on enzymatic hydrolysis. Bioresour Technol 2011;102(19):90839.
[20] Shen F, Zhong Y, Saddler JN, Liu R. Relatively high-substrate consistency
hydrolysis of steam-pretreated sweet sorghum bagasse at relatively low
cellulase loading. Appl Biochem Biotech 2011;165(34):102436.
[21] Wang W, Kang L, Wei H, Arora R, Lee YY. Study on the decreased sugar yield in
enzymatic hydrolysis of cellulosic substrate at high solid loading. Appl
Biochem Biotech 2011;164(7):113949.
[22] Yang J, Zhang X, Yong Q, Yu S. Three-stage enzymatic hydrolysis of steam-
exploded corn stover at high substrate concentration. Bioresour Technol
2011;102(7):49058.
[23] Kristensen JB, Felby C, Jrgensen H. Yield-determining factors in high-solids
enzymatic hydrolysis of lignocellulose. Biotechnol Biofuels 2009;2(1):11.
[24] Xue Y, Jameel H, Phillips R, Chang HM. Split addition of enzymes in enzymatic
hydrolysis at high solids concentration to increase sugar concentration for
bioethanol production. J Ind Eng Chem 2012;18(2):70714.
[25] Dijkerman R, Vervuren MB, Op Den Camp HJ, Van der Drift C. Adsorption
characteristics of cellulolytic enzymes from the anaerobic fungus Piromyces
sp. strain E2 on microcrystalline cellulose. Appl Environ Microbiol 1996;62(1):
205.
[26] Conde-Meja C, Jimnez-Gutirrez A, El-Halwagi M. A comparison of
pretreatment methods for bioethanol production from lignocellulosic
materials. Process Saf Environ 2012;90(3):189202.