Enhanced direct production of sorbitol by cellulose ball-milling

Article abstract of DOI:10.1039/c5gc00039d

The catalytic conversion of lignocellulosic biomass to renewable and valuable chemicals has attracted global interest. Given the abundance of this renewable raw material and its reduced impact on the food chain, it is an attractive source for replacing fossil fuels and obtaining chemicals or fuels in the context of a sustainable economy. In this work, a catalyst (Ru/AC) was developed to perform, in a single step, hydrolysis and hydrogenation of cellulose to sorbitol. An activated carbon supported ruthenium catalyst was examined for the one-pot hydrolytic hydrogenation of cellulose and it has shown to be very active and selective for the conversion of cellulose into sorbitol. When microcrystalline cellulose was used, a conversion of 36% was reached after 5 hours of reaction, with a selectivity to sorbitol of 40%. On the other hand, ball-milled cellulose allowed attaining conversions close to 90%, with a selectivity to sorbitol of 50%. Moreover, if the catalyst was ball-milled together with cellulose, the selectivity to sorbitol could be further increased to almost 80%. The catalyst showed excellent stability after repeated use. In this work we combined hydrolysis and hydrogenation in one-pot (using heterogeneous catalysts instead of homogeneous), in the presence of a Ru/AC catalyst (without any support pre-treatment with acids) and pre-treated cellulose just by ball-milling (instead of using acids). For this reason, the results obtained in this work are one of the best values achieved when using supported metal catalysts to convert cellulose by an environmentally friendly process.

Full text of DOI:10.1039/c5gc00039d

Green Chemistry
View Article Online
View Journal
PAPER
Enhanced direct production of sorbitol by
cellulose ball-milling
Cite this: DOI: 10.1039/c5gc00039d
Lucília S. Ribeiro, José J. M. Órfão and Manuel Fernando R. Pereira*
The catalytic conversion of lignocellulosic biomass to renewable and valuable chemicals has attracted
global interest. Given the abundance of this renewable raw material and its reduced impact on the food
chain, it is an attractive source for replacing fossil fuels and obtaining chemicals or fuels in the context of
a sustainable economy. In this work, a catalyst (Ru/AC) was developed to perform, in a single step, hydro-
lysis and hydrogenation of cellulose to sorbitol. An activated carbon supported ruthenium catalyst was
examined for the one-pot hydrolytic hydrogenation of cellulose and it has shown to be very active and
selective for the conversion of cellulose into sorbitol. When microcrystalline cellulose was used, a conver-
sion of 36% was reached after 5 hours of reaction, with a selectivity to sorbitol of 40%. On the other hand,
ball-milled cellulose allowed attaining conversions close to 90%, with a selectivity to sorbitol of 50%.
Moreover, if the catalyst was ball-milled together with cellulose, the selectivity to sorbitol could be further
increased to almost 80%. The catalyst showed excellent stability after repeated use. In this work we com-
bined hydrolysis and hydrogenation in one-pot (using heterogeneous catalysts instead of homogeneous),
in the presence of a Ru/AC catalyst (without any support pre-treatment with acids) and pre-treated cellu-
lose just by ball-milling (instead of using acids). For this reason, the results obtained in this work are one
of the best values achieved when using supported metal catalysts to convert cellulose by an environmen-
tally friendly process.
Received 7th January 2015,
Accepted 8th March 2015
DOI: 10.1039/c5gc00039d
www.rsc.org/greenchem
Introduction
Steadily increasing greenhouse gas emissions and worldwide
energy demand, together with the limited availability and sus-
tainability of fossil fuels, make it necessary to identify alterna-
tives and develop new environmentally friendly processes on
the basis of renewable feedstocks.1–3 Recently, lignocellulosic
biomass, which does not compete with food sources and is
widely available as a low cost feedstock, has drawn a lot of
attention as one of the most attractive alternatives to replace
fossil resources for the production of fuels and fine
Fig. 1 Catalytic conversion of cellulose into sorbitol.
chemicals.2–4 Cellulose is the most abundant form of metics and textiles.1 In addition, it is one of the twelve build-
biomass.5 Therefore, cellulose is the most promising natural ing block chemicals produced from biomass resources5 and it
resource for the conversion into more valuable chemicals.4 In is also used in the synthesis of surfactants and fine chemicals,
the last few decades, many reaction routes have been develo- such as in the synthesis of vitamin C.6
ped for its use. Among these, the production of polyols,
Due to its robust crystalline structure, the effective degra-
especially sorbitol, via cellulose hydrolysis combined with dation of cellulose is still a challenge.5 Usually, the process for
hydrogenation is an important research field (Fig. 1).6 Sorbitol transformation of cellulose consists of its hydrolysis to
is a promising platform chemical, which can be used as a glucose, followed by the hydrogenation of glucose to polyols.
dispensing agent and a humectant in pharmaceuticals, cos- Homogeneous catalysts, such as sulphuric acid, as well as
cellulase enzymes promote the production of glucose in high
yields from cellulose.5 However, cellulase is expensive and can
hardly be reused;7 on the other hand, the use of strong liquid
Laboratory of Catalysis and Materials (LCM), Associate Laboratory LSRE/LCM,
acids is not sustainable and suffers from serious drawbacks
such as low selectivity, separation of products, corrosion and
Department of Chemical Engineering, Faculty of Engineering, University of Porto,
Portugal. E-mail: fpereira@fe.up.pt
This journal is © The Royal Society of Chemistry 2015
Green Chem.

View Article Online
Paper
Green Chemistry
acid recovery or disposal.7,8 Compared with homogeneous cata-
lysts, heterogeneous solid catalysts have several advantages,3
such as being easily recovered and reused. Recently, instead of
the traditional use of liquid acids, extensive attention has been
given to the direct conversion of cellulose to polyols over
heterogeneous catalysts.7,9–12 Ru has been considered a good
catalyst for hydrogenation with high activity and is relatively
cheap compared to noble metals (such as Pt or Au).4,8 Carbon
materials such as activated carbon and carbon nanotubes are
known as heat- and water-tolerant supports.5 For this reason,
carbon supported metal catalysts have been extensively studied.
In 2006, Fukuoka and Dhepe reported the first hydrolytic
hydrogenation of cellulose using only solid catalysts, in which
Pt/γ-Al2O3 promoted the production of sorbitol and mannitol
at 25% and 6% yields, respectively.13 After that, Luo et al.
reported a cellulose conversion of 86% and a 30% sorbitol
yield over a carbon supported ruthenium catalyst (Ru/C) at
245 °C.14 Since the rate-determining step of the hydrolytic
hydrogenation is the hydrolysis of cellulose to glucose,5 some
pre-treatment of cellulose is required in order to facilitate its
conversion. Usually, this is achieved with chemical
treatments,15–17 where the crystallinity of cellulose is reduced
when it is in contact with acids. For example, Deng et al.
reported that the crystallinity of cellulose decreased by treating
it with phosphoric acid.18 This group reported the conversion
of cellulose with a 69% sorbitol yield,18 which is, to the best of
our knowledge, one of the highest values to date, over a Ru/
CNT catalyst using cellulose pre-treated with phosphoric acid
at a reaction temperature of 185 °C. Geboers et al. used a
mixture of Ru/C and heteropoly acid H4SiW12O40 to enhance
the hydrolysis rate,19 which gave 85% yield of sugar alcohols
from ball-milled cellulose and 36% yield from microcrystalline
cellulose. Though this is one of the best sugar alcohol yields
attained to date, it involved the utilization of an acid, which is
not such an environmentally friendly approach.
Fig. 2 TPR profile of the catalyst.
Fig. 3 N2 adsorption isotherms of the support and catalyst.
Table 1 Textural properties of the support and catalyst
SBET
Smeso
Vmicro
Metal loading dM (nm)
Sample
AC
(m2 g−1
)
(m2 g−1
)
(cm3 g−1
)
(wt%) [ICP]
[TEM]
847
152
151
0.301
0.294
—
—
0.4%Ru/AC 804
0.32 0.02
1.4 0.3
supported ruthenium catalyst are not significantly different
from those of the original activated carbon.
In this work, the pre-treatment of cellulose is done by ball-
milling, avoiding the use of acids, and the problems associated
with their use (such as disposal).
Table 1 also shows the metal loading of the catalyst. The
ruthenium content obtained by ICP analysis (0.32%) is slightly
lower than the nominal value (0.4%), which indicates that a
fraction of ruthenium was not deposited on the support
during the preparation step.
Results and discussion
Characterizations
TEM micrographs (Fig. 4a) were collected in order to get
information about the metal particle size distribution
The catalyst (0.4%Ru/AC) was initially characterised by temp-
erature programmed reduction (TPR). The TPR profile (Fig. 2)
showed a wide reduction peak around 100–250 °C, and so the
catalyst was reduced at 250 °C to assure effective reduction of
the metal.
By analysing the N2 adsorption isotherms (Fig. 3) it is poss-
ible to observe the well-developed microporosity of the support
(AC), which presents a specific area of about 850 m2 g−1. As
expected, no drastic changes in the textural properties of the
catalyst were observed. The BET surface area of Ru/AC
decreased slightly compared to the unloaded carbon (Table 1).
Fig. 4 TEM image (a) and ruthenium particle size distribution on 0.4%
Therefore, it was considered that the textural properties of the Ru/AC (b).
Green Chem.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Green Chemistry
Paper
(Fig. 4b). The average crystallite size of the studied catalyst is the hydrogen bond networks within cellulose. Such changes
presented in Table 1. As observed, the incipient wetness are expected to have great influence on cellulose conversion.
method allows obtaining relatively small metal crystallite sizes
for the catalyst prepared.
The degree of polymerization (DP) of cellulose samples was
determined by the viscosity method. All measurements were
The XRD peak height method is the most widely used performed at 25.0 0.5 °C using a Cannon–Fenske capillary
method to determine the crystallinity index of cellulose, allow- viscometer (size 50) and cupriethylenediamine (Cuene) as the
ing fast comparison of samples.5,20 The substrates used in this solvent.21,22 For each solution (cellulose and solvent), the
study were microcrystalline (sample cellulose_f0_t0) and ball- measurement was repeated three times and the average was
milled cellulose, and the crystallinity index (CrI) was calculated considered. By analysing Table 2, we conclude that there is a
using the following equation:
slight decrease in the DP of cellulose after ball-milling.
Despite the small differences between ball-milled cellulose
samples, a slight decrease in DP with the increase of ball-
milling time and frequency is noticed.
I002 ꢀ IAM
CrI ð%Þ ¼
ꢁ 100
ð1Þ
I002
In order to obtain the particle size distributions of the cel-
lulose samples, the particles were classified according to
different intervals. The average diameters (dav) for each sample
are presented in Table 2, according to the particle size distri-
butions presented in Fig. 6. By analysing Fig. 6a, a decrease in
particle size with the increase in ball-milling frequency can be
observed (Table 2). In Fig. 6b there is a notable difference
between cellulose_f0_t0 and the ball-milled samples. However,
no significant difference is observed between samples ball-
milled for different times.
The elemental analysis of microcrystalline and ball-milled
cellulose samples revealed carbon, hydrogen and oxygen con-
tents of about 44%, 7% and 49%, respectively. Comparing the
results obtained for the different samples, we can ascertain
that the cellulose structure is not altered by ball-milling. Fur-
thermore, the results are compatible with cellulose’s molecular
formula, (C6H10O5)n.
where I002 is the maximum intensity of the (002) lattice diffrac-
tion (2θ = 22.6°) and IAM is the intensity diffraction at 2θ = 18°.
I002 represents both crystalline and amorphous cellulose
whereas IAM represents amorphous cellulose only.
XRD patterns of microcrystalline and ball-milled cellulose
samples are shown in Fig. 5. It is possible to observe the stron-
gest peak at 2θ = 22.6°, which originates from the cellulose
crystalline plane (002). As expected and shown in Table 2, after
ball-milling, cellulose had a reduced crystallinity compared to
the original cellulose. Besides that, a decrease in both crystal-
lite size and crystallinity index with the increase of ball-milling
time and frequency can be noticed. For instance, the degree of
crystallinity of microcrystalline cellulose decreases from 92 to
23% just by ball-milling at a frequency of 20 Hz for 4 h.
Clearly, the transformation of microcrystalline cellulose into
amorphous cellulose suggests that ball-milling has weakened
Fig. 5 XRD patterns of cellulose for different ball-milling frequencies
Fig. 6 Particle size distribution for different ball-milling frequencies (a)
(a) and times (b).
and times (b).
Table 2 Results for the characterization of the different cellulose samples
Ball-milling
Cellulose samples
f (Hz)
t (h)
Crystallite size (nm) [XRD]
CrI (%) [XRD]
DP [Visc.]
dav (nm) [Coulter]
Cellulose_f0_t0
Cellulose_f10_t4
Cellulose_f10_t9
Cellulose_f10_t12
Cellulose_f10_t24
Cellulose_f10_t48
Cellulose_f5_t4
Cellulose_f20_t4
0
10
10
10
10
10
5
0
4
8
12
24
48
4
6.3
5.4
5.0
4.9
3.1
2.0
6.5
2.0
92
87
79
76
48
33
93
23
221
209
207
200
196
191
204
191
68
37
28
32
35
37
58
36
20
4
This journal is © The Royal Society of Chemistry 2015
Green Chem.

View Article Online
Paper
Green Chemistry
and hydrogenation of glucose were also studied separately,
allowing us to conclude that the catalyst influence is mainly in
the hydrogenation step.
Preliminary studies were carried out by varying the stirring
rate and it was found that under the selected conditions (see
the Catalytic experiments section) there were no external diffu-
sional limitations. Blank experiments were performed under
the same conditions; the conversion of microcrystalline cellu-
lose was up to 20% without any catalyst. In the presence of AC,
the cellulose conversion reached 25% with a selectivity to sor-
bitol of 15%.
Fig. 7 TG (in black) and DTG (in blue) curves of cellulose decompo-
sition under nitrogen or air.
The Ru/AC catalyst gave 36% conversion with a selectivity to
sorbitol of 40% after 5 h (Table 3, entry 1).
Due to its robust crystalline structure with a linkage of
β-1,4-glycosidic and hydrogen bonds, it is difficult to hydrolyse
cellulose and transform it into chemicals and fuels. In this
work, in order to overcome this problem, microcrystalline cel-
lulose was pre-treated by ball-milling. The effect of ball-milling
was analysed in the conversion of cellulose, using an activated
carbon supported ruthenium catalyst (0.4% Ru/AC).
The effect of ball-milling on cellulose conversion was
studied for different times and frequencies as depicted in
Fig. 8. While the hydrolytic hydrogenation of the raw sample
(cellulose_f0_t0) only achieves 36% conversion, the ball-milled
sample cellulose_f10_t48 achieves up to 90% conversion
within 5 h of reaction. Moreover, there is also an important
increase in the selectivity to sorbitol (Table 3). An increase in
both catalytic activity and sorbitol selectivity is noticed not
only for increased ball-milling times (entries 2–6), but also for
increased ball-milling frequencies (entries 2, 7 and 8), which is
consistent with the XRD results. Obviously, this must be due
to the physical and structural changes in the cellulose induced
by ball-milling, making glucose chains within cellulose more
accessible.
The mass loss history of microcrystalline cellulose (sample
cellulose_f0_t0) under different atmospheres was tracked as
shown in Fig. 7. A great mass loss between 290 and 360 °C
(almost 90% of total mass) could be found under an inert
atmosphere (N2). Under an oxidative atmosphere (air), a
second mass loss was detected at higher temperatures
(460–560 °C), with an almost total mass loss above 600 °C. The
temperature range corresponding to the first mass loss does
not vary regardless of the surrounding atmosphere. These
results are consistent with those reported in the literature.23–25
Catalytic tests
The successful selective transformation of cellulose to sugar
alcohols requires the development of catalysts with the ability
to promote the coupled hydrolysis/hydrogenation reactions
(leading to sorbitol) and also to minimize secondary reactions,
which lead to undesirable by-products. For better understand-
ing of this coupled reaction effect, both hydrolysis of cellulose
Table 3 Comparison between activities (TOFs), conversions of cellulose (X) and selectivities (S) and yields (Y) of sorbitol obtained after 1 and 5 h of
reactiona
1 h of reaction
5 h of reaction
TOFd
(h−1
X
(%)
Ssorbitol
(%)
Ysorbitol
(%)
X
(%)
Ssorbitol
(%)
Ysorbitol
(%)
Sglucose
(%)
Smannitol
(%)
Entry
Substrate
)
1
2
3
4
5
6
7
Cellulose_f0_t0
Cellulose_f10_t4
Cellulose_f10_t9
Cellulose_f10_t12
Cellulose_f10_t24
Cellulose_f10_t48
Cellulose_f5_t4
Cellulose_f20_t4
(Cellulose + catalyst)_f20_t4
Cellulose_f20_t4 + catalyst_f20_t4
0.39
0.49
0.72
0.84
1.29
1.53
0.32
1.57
2.00
1.60
16.9
21.2
30.6
32.9
51.7
60.1
14.0
61.1
76.4
61.4
59.5
66.3
50.8
56.5
48.8
52.2
69.8
42.8
89.0
80.5
10.0
14.1
15.5
18.6
25.2
31.4
9.8
26.1
68.0
49.4
35.7
50.3
57.2
62.9
76.4
90.8
38.8
87.8
89.4
82.1
40.5
44.5
43.9
49.6
49.5
49.8
38.4
48.0
76.9
79.9
14.5
22.4
25.1
31.2
37.8
45.2
14.9
42.1
68.8
65.6
2.1
15.0
13.1
11.7
8.9
8.5
0.9
8.5
0.0
0.0
0.5
1.7
3.4
2.9
3.9
4.2
0.5
3.4
4.2
4.6
8
9b
10c
a Conditions: cellulose (0.75 g), water (300 mL), 0.4%Ru/AC catalyst (0.3 g), 205 °C, 50 bar H2, 150 rpm. b Cellulose and catalyst were ball-milled
together at 20 Hz for 4 h (sample denoted as (cellulose + catalyst)_f20_t4). c Besides cellulose (Cellulose_f20_t4), the catalyst was ball-milled too
under the same conditions (20 Hz for 4 h) and denoted as catalyst_f20_t4. d TOF (h−1) = (mole of cellulose converted after 1 h)/(reaction time
(= 1 h) × mole of ruthenium active sites); the amount of Ru active sites was calculated from (6 × y × ns × Wcat)/(dM × ρ × N), where y and dM are the
Ru weight fraction (from ICP) and the average particle diameter (from TEM), respectively; ρ and ns are the density of Ru and the number of
Ru atoms per unit of area, respectively; Wcat is the catalyst mass used in the experiments (0.3 g); and N is the Avogadro number.
Green Chem.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Green Chemistry
Paper
perties of the catalyst and the substrate. The hydrolytic hydro-
genation of this mixed sample resulted in 76% conversion
after just 1 h of reaction, with 68% yield of sorbitol (Table 3,
entry 9), whereas the separately ball-milled catalyst (cata-
lyst_f20_t4) and cellulose (cellulose_f20_t4) provided only 49%
yield of sorbitol after 1 h of reaction (entry 10). Also, a great
increase in the initial reaction rate by the procedure men-
tioned above can be observed in Fig. 10. These results indicate
that the mix-milling pre-treatment accelerates the hydrolytic
hydrogenation of cellulose (compare entries 8–10 in Table 3).
Similar results were obtained by Kobayashi et al. for the con-
version of mix-milled activated carbon and cellulose,26 which
allowed concluding that the enhancement of the reaction per-
formance by mix-milling is not due to the mechano-catalytic
hydrolysis during the treatment but due to mixing.
Finally, a very important characteristic for commercial feasi-
bility is the long-term stability of the heterogeneous catalyst.
Thus, to test the catalyst deactivation for cellulose conversion
into sorbitol, four successive tests under the same conditions
were carried out, after sample washing and drying, as shown
in Fig. 11. Due to some losses during the catalyst filtration at
the end of the reaction, a small amount of fresh catalyst (<5 wt
%) was added to the reaction mixture before each run. The
conversion of ball-milled cellulose (sample cellulose_f20_t4)
was maintained around 88% after 5 h in each run, as well as
the selectivity to sorbitol, which was kept almost constant
around 47%. Since the cellulose conversion and sorbitol
selectivity showed no difference up to the 4th cycle, this indi-
Fig. 8 Evolution of cellulose conversion for different ball-milling times
and frequencies.
The results shown in Table 3 also indicate a decrease in sor-
bitol selectivity with reaction time when using as substrate cel-
lulose samples with higher crystallinity degrees (entries 1–4
and 7). If the cellulose crystallinity index is less than 50%
(entries 5, 6 and 8), no significant difference is found in sorbi-
tol selectivity after 1 and 5 h of reaction.
In terms of the effect of ball-milling time, a clear increment
in the activity is observed between microcrystalline cellulose
(cellulose_f0_t0) and cellulose ball-milled for 48 h (0.39 to
1.53 h−1). When comparing microcrystalline cellulose and cel-
lulose ball-milled at a frequency of 20 Hz (cellulose_f20_t4)
there is also an increase of activity from 0.39 to 1.57 h−1
.
These variations are directly related to the decrease in cellulose
crystallinity caused by the increase in ball-milling time and fre-
quency. Indeed, if we plot the activity (TOF) as function of the
crystallinity index (CrI) of cellulose, a strong correlation is
obtained as shown in Fig. 9.
As already mentioned, since the rate-determining step of
the hydrolytic hydrogenation is the hydrolysis of cellulose to
glucose, this step is of great importance. So, according to
Kobayashi et al., who considered that the hydrolysis of cellu-
lose promoted by carbon catalysts also occurs at the solid–
solid interface, the number of collisions is a major obstacle in
this type of reaction.26 Therefore, cellulose and catalyst were
ball-milled together (sample denoted as (cellulose + catalyst)
_f20_t4) as a pre-treatment to improve the contact between the
cellulose and the catalyst, as the mixed state can continue in
the subsequent hydrolysis reaction thanks to the insoluble pro-
Fig. 10 Evolution of the conversion of cellulose ball-milled apart/
together with the catalyst.
Fig. 9 Correlation between the catalytic activity and the crystallinity
Fig. 11 Recycling test of 0.4% Ru/AC catalyst on the conversion of ball-
index of cellulose.
milled cellulose (cellulose_f20_t4) after 5 h of reaction.
This journal is © The Royal Society of Chemistry 2015
Green Chem.

View Article Online
Paper
Green Chemistry
cates that 0.4% Ru/AC exhibits no deactivation and presents Different ball-milled cellulose samples were prepared by
excellent stability. varying the ball-milling frequencies (from 5 to 20 Hz) for 4 h
In summary, according to the presented results, cellulose or by varying the ball-milling times (from 4 to 48 h) at a fre-
conversions around 90% and sorbitol yields near 45% can be quency of 10 Hz. These samples were denoted as cellulose_fre-
reached after 5 h of reaction if cellulose is pre-treated by ball- quency/Hz_time/h, according to the respective ball-milling
milling for 48 h at a frequency of 10 Hz or for 4 h at a fre- frequency and time. Following this notation, microcrystalline
quency of 20 Hz. Moreover, the sorbitol yield can be raised to cellulose was denoted as cellulose_f0_t0, since it was not ball-
almost 70% in just 1 h of reaction if the catalyst is ball-milled milled. The crystallinity index of the prepared cellulose
together with cellulose for 4 h at a frequency of 20 Hz. In samples was determined by XRD.
addition, the catalyst showed excellent stability after repeated
use. Although the comparison between the results obtained in
this work and those reported in the literature is particularly
difficult since the reaction conditions changed from author to
author, to the best of our knowledge, this work reports one of
the highest sorbitol yields ever attained without the use of
non-environmentally friendly approaches.
Preparation of catalysts
A commercial Norit GAC 1240 activated carbon (AC) sample
was used as a support for this study. A Ru-based catalyst was
prepared by incipient wetness impregnation of AC (sieved
between 0.1 and 0.3 mm) with a solution of the corresponding
metallic precursor. The amount of noble metal was calculated
to achieve a metal loading of 0.4 wt%. The support (AC) was
firstly introduced into an ultrasonic bath for 30 minutes. Sec-
ondly, the precursor solution was added dropwise, with a peri-
staltic pump (50 mL h−1), until all the support was wet. Still in
the ultrasonic bath, the maturation and drying occurred for
90 minutes, after which the catalyst was dried overnight in an
oven at 110 °C and then stored for later use.
Experimental
Chemicals and materials
Unless stated otherwise, chemicals were purchased from com-
mercial sources and used as received, without further purifi-
cation. The ultrapure water with a conductivity of 18.2 µS cm−1
was obtained in a Milli-Q Millipore System. Microcrystalline
cellulose and sorbitol (98%) were purchased from Alfa Aesar.
The metal precursor RuCl3 (99.9%, Ru 38%) was also supplied
by Alfa Aesar. Activated carbon GAC 1240 PLUS was provided
by Norit and sulphuric acid (95%) was obtained from VWR.
After heat treatment under nitrogen flow for 3 h (50 cm3
min−1), the catalyst was reduced under hydrogen flow for 3 h
(50 cm3 min−1). The appropriate reduction temperature
(250 °C) was determined by temperature programmed
reduction (TPR). The treatment under nitrogen flow was
carried out at the same temperature used for the reduction.
This sample was denoted as 0.4% Ru/AC.
Cellulose pre-treatment
The crystal structure and hydrogen bonding in cellulose limit
the access to β-1,4-glycosidic bonds by reactants and cata-
lysts.27 So, in order to facilitate the contact with the catalysts,
the crystalline structure of cellulose has to be considered.
Mechanical (milling, grinding, steaming and irradiation) and
chemical (dissolution, alkali, acid, swelling, gas, oxidation and
solvent) treatments decrease the degree of crystallization of cel-
lulose, and this value can be determined by X-ray Diffraction
(XRD), 13C NMR and other methods.17 Milling methods such
as ball-milling are the typical mechanical techniques for dis-
rupting the crystal structure of cellulose, since cellulose’s
hydrogen bonds are cleaved.5 Acid treatments are the chemical
treatments often used for this purpose, but are undesirable
due to reactor corrosion and environmental toxicity. In con-
trast, ball-milling does not suffer from this problem, making it
a great alternative to break the robust crystalline structure of
cellulose and decrease its crystallinity in a sustainable way.
In this work, microcrystalline cellulose was ball-milled in a
ceramic pot with two ZrO2 balls (12 mm of diameter), in order
to reduce its crystallinity before the reaction and study its
effect on the performance of the process. The ball-milled cellu-
Characterization of materials
Both the support and the catalyst were characterized by N2
adsorption at −196 °C using a Quantachrome NOVA Surface
Area and Pore Size analyser, after outgassing the samples at
350 °C for 3 h under vacuum. Surface area calculations were
made using the Brunauer–Emmett–Teller (BET) equation, and
the micropore volumes (Vmicro) and mesopore surface areas
(Smeso) by the t-method. The TPR profile was obtained by using
a fully automated AMI-200 equipment (Altamira Instruments);
the sample was subjected to a 5 °C min−1 heating to 700 °C
under 5% (v/v) H2 flow diluted in He (total flow rate = 30 cm3
min−1). The metal loading of the catalyst was determined in
duplicate by inductively coupled plasma-optical emission spec-
troscopy (ICP/OES), using a PerkinElmer Optima 4300 DV
spectrometer. Electron micrographs of the catalyst were
acquired with a JEOL2010F instrument equipped with an
energy dispersive X-ray spectroscopy detector. Size distri-
butions were determined by the measurement of 200 nanopar-
P
dini
ni
P
ticles and the average diameter was calculated by dM
¼
,
lose was prepared using a laboratory ball mill (Retsch Mixer where ni is the number of particles with diameter di.
Mill MM200) by introducing 1.5 g of commercial microcrystal-
To evaluate the crystallinity index (CrI) of the microcrystal-
line cellulose into the ceramic pot, operating at a frequency of line and ball-milled cellulose samples, XRD patterns were
10 Hz for 4 h. This sample was denoted as cellulose_f10_t4. recorded by using a Philips X’Pert MPD diffractometer (Cu-Kα
Green Chem.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Green Chemistry
Paper
= 0.15406 nm). The diffracted intensity of Cu-Kα radiation was
For determining the conversion of non-soluble cellulose
measured in a 2θ range between 10° and 50°. The degree of into water soluble products, TOC data were obtained with a
polymerization (DP) of cellulose was determined by viscosity Shimadzu TOC 5000-A and the conversion was determined
measured at 25.0 0.5 °C using a Cannon-Fenske capillary using the equation: XTOC = (moles of total organic carbon in
viscometer and cupriethylenediamine (Cuene) as the solvent. the resulting liquid)/(moles of carbon in cellulose charged into
The particle-size distribution of each cellulose sample was the reactor) × 100%. Since the TOC conversion was very close
determined with a Beckman Coulter LS230 Laser Diffraction to the weight conversion (differences smaller than 5%), imply-
Particle Size Analyzer with small volume module plus. Elemen- ing that only a small amount of carbon-containing gas was
tal analysis was performed on an EA1108 CHNS-O elemental generated, and gas products were not considered in our study.
analyser from Carlo Erba Instruments. TG analysis was carried
out under nitrogen and air, with a heating ramp of 10 °C
min−1 from 50 to 800 °C, using an STA 409 PC/4/H Luxx Conclusions
Netzsch thermal analyser.
A Ru catalyst supported on activated carbon was tested in the
one-pot conversion of cellulose. The catalyst showed to be
highly efficient, particularly if cellulose is ball-milled prior to
use in the reaction. Moreover, if the catalyst was ball-milled
together with cellulose, the selectivity to sorbitol could be
further increased. Ball-milling allowed decreasing the crystal-
lite size and crystallinity index of cellulose, without signifi-
cantly changing its polymerization degree. As expected, the
catalytic activity and sorbitol yield increased with the increase
in ball-milling time and frequency, as a consequence of the
decrease in cellulose crystallinity. A strong correlation between
the catalytic activity and cellulose crystallinity was found. In
addition, no obvious deactivation was observed after several
runs. To the best of our knowledge, this article presents one of
the best results ever attained for the direct conversion of cellu-
lose into sorbitol using such a green process.
Catalytic experiments
The hydrolytic hydrogenation of cellulose was performed in a
1000 mL stainless steel reactor (Parr Instruments, USA Mod.
5120) equipped with a gas supply system, a manometer, a
temperature sensor and a sample outlet with a filter, which
prevents the catalyst particles to pass through it. In standard
tests, 300 mL of water, 750 mg of cellulose and 300 mg of cata-
lyst (0.4%Ru/AC) were introduced into the reactor under stir-
ring at 150 rpm. Then the reactor was flushed three times with
N2 to remove air. After heating under this atmosphere to the
desired temperature (205 °C), the reaction was initiated by
switching from an inert gas to H2 (50 bar). Samples (1 mL)
were periodically withdrawn for analysis. After the reaction,
the catalyst and non-hydrolysed cellulose were filtered out. The
conversion of cellulose was calculated as the ratio of the hydro-
lysed cellulose mass to its initial amount; the values were also
verified on the basis of total organic carbon (TOC) data.
Acknowledgements
Product analysis
This work was co-financed by FCT and FEDER under Pro-
gramme COMPETE (Project PEst-C/EQB/LA0020/2013) and by
QREN, ON2 and FEDER (Project NORTE-07-0162-
FEDER-000050). L.S. Ribeiro acknowledges her grant from FCT
(SFRH/BD/86580/2012), financed by project POPH/FSE.
The quantitative analysis of the samples was carried out by
high performance liquid chromatography (HPLC). The chro-
matograph (Elite LaChrom HITACHI) was equipped with a
refractive index (RI) detector and the products were separated
in an ion exclusion column (Alltech OA 1000). The retention
times and calibration curves were determined by comparison
with standard samples. The eluent was a H2SO4 solution
(0.005 M). An injection volume of 30 µL, a flow rate of 0.5 mL
min−1 and a measuring time of 20 min were selected. The
samples were injected into the HPLC directly after the exper-
iments without any pre-treatment other than filtering to
prevent solid particles from entering the column. The selecti-
vity (Si) of each product i at time t was calculated as:
Notes and references
1 A. M. Ruppert, K. Weinberg and R. Palkovits, Angew. Chem.,
Int. Ed., 2012, 51, 2564.
2 D. M. Alonso, S. G. Wettstein and J. A. Dumesic, Chem. Soc.
Rev., 2012, 41, 7965.
3 C.-H. Zhou, X. Xia, C.-X. Lin, D.-S. Tong and J. Beltramini,
Chem. Soc. Rev., 2011, 40, 5588.
4 J. W. Han and H. Lee, Catal. Commun., 2012, 19, 115.
5 M. Yabushita, H. Kobayashi and A. Fukuoka, Appl. Catal.,
B, 2014, 145, 1.
6 J. Xi, Y. Zhang, Q. Xia, X. Liu, J. Ren, G. Lu and Y. Wang,
Appl. Catal., A, 2013, 459, 52.
7 X. Wang, L. Meng, F. Wu, Y. Jiang, L. Wang and X. Mu,
Green Chem., 2012, 14, 758.
8 H. Wang, L. Zhu, S. Peng, F. Peng, H. Yu and J. Yang,
Renewable Sustainable Energy Rev., 2012, 37, 192.
Ci
νi C0 X
Si ¼
ð2Þ
where Ci is the concentration of the product i (mol L−1), C0 is
the initial concentration of cellulose (mol L−1), X is the conver-
sion of cellulose and νi corresponds to the moles of i produced
per mole of cellulose consumed, according to the stoichio-
metry. The yield (Yi) of each product i at time t was determined
as the product between the corresponding selectivity and cellu-
lose conversion.
This journal is © The Royal Society of Chemistry 2015
Green Chem.

View Article Online
Paper
Green Chemistry
9 A. Negoi, K. Triantafyllidis, V. I. Parvulescu and 19 J. Geboers, S. Van de Vyver, K. Carpentier, K. de Bloc-
S. M. Coman, Catal. Today, 2014, 223, 122.
house, P. Jacobs and B. Sels, Chem. Commun., 2010, 46,
10 H. Kobayashi, Y. Ito, T. Komanoya, Y. Hosaka, P. L. Dhepe,
3577.
K. Kasai, K. Hara and A. Fukuoka, Green Chem., 2011, 13, 326. 20 S. Park, J. Baker, M. Himmel, P. Parilla and D. Johnson,
11 P. Dhepe and A. Fukuoka, ChemSusChem, 2008, 1, 969. Biotechnol. Biofuels, 2010, 3, 1.
12 M. Käldström, N. Kumar and D. Y. Murzin, Catal. Today, 21 B. B. Hallac and A. J. Ragauskas, Biofuels, Bioprod. Biorefin.,
2011, 167, 91. 2011, 5, 215.
13 A. Fukuoka and P. Dhepe, Angew. Chem., Int. Ed., 2006, 45, 22 M. L. R. Medina and V. Kumar, Int. J. Pharm., 2007, 337,
5161. 202.
14 C. Luo, S. Wang and H. Liu, Angew. Chem., Int. Ed., 2007, 23 J. J. M. Orfão, F. J. A. Antunes and J. L. Figueiredo, Fuel,
46, 7636. 1999, 78, 349.
15 V. B. Agbor, N. Cicek, R. Sparling, A. Berlin and D. B. Levin, 24 S. Sivasangar, Y. H. Taufiq-Yap, Z. Zainal and K. Kitagawa,
Biotechnol. Adv., 2011, 29, 675. Int. J. Hydrogen Energy, 2013, 38, 16011.
16 P. Kumar, D. M. Barrett, M. J. Delwiche and P. Stroeve, Ind. 25 D. Shen, J. Ye, R. Xiao and H. Zhang, Carbohydr. Polym.,
Eng. Chem. Res., 2009, 48, 3713. 2013, 98, 514.
17 P. Yang, H. Kobayashi and A. Fukuoka, Chin. J. Catal., 26 H. Kobayashi, M. Yabushita, T. Komanoya, K. Hara,
2011, 32, 716. I. Fujita and A. Fukuoka, ACS Catal., 2013, 3, 581.
18 W. Deng, X. Tan, W. Fang, Q. Zhang and Y. Wang, Catal. 27 H. Zhao, J. H. Kwak, Y. Wang, J. A. Franz, J. M. White and
Lett., 2009, 133, 167. J. E. Holladay, Energy Fuels, 2005, 20, 807.
Green Chem.
This journal is © The Royal Society of Chemistry 2015